Packman Build Service PMBS

Changes of Revision 6

x265.changes Changed

x265.spec Changed

arm.patch Changed

@@ -1,18 +1,18 @@
-diff -urN a/source/CMakeLists.txt b/source/CMakeLists.txt
---- a/source/CMakeLists.txt	2014-04-03 00:00:44.000000000 -0600
-+++ b/source/CMakeLists.txt	2014-04-10 07:32:36.022010974 -0600
-@@ -39,10 +39,18 @@
+--- source/CMakeLists.txt.orig	2014-10-04 17:02:08.482815493 +0200
++++ source/CMakeLists.txt	2014-10-04 17:11:52.536865262 +0200
+@@ -46,10 +46,18 @@
          set(X64 1)
          add_definitions(-DX86_64=1)
      endif()
+-elseif(${SYSPROC} STREQUAL "armv6l")
+-    message(STATUS "Detected ARM target processor")
+-    set(ARM 1)
+-    add_definitions(-DX265_ARCH_ARM=1 -DHAVE_ARMV6=1)
 +elseif(${SYSPROC} MATCHES "armv5.*")
 +    message(STATUS "Detected ARMV5 system processor")
 +    set(ARMV5 1)
 +    add_definitions(-DX265_ARCH_ARM=1 -DHAVE_ARMV6=0 -DHAVE_NEON=0)
- elseif(${SYSPROC} STREQUAL "armv6l")
--    message(STATUS "Detected ARM system processor")
--    set(ARM 1)
--    add_definitions(-DX265_ARCH_ARM=1 -DHAVE_ARMV6=1)
++	elseif(${SYSPROC} STREQUAL "armv6l")
 +    message(STATUS "Detected ARMV6 system processor")
 +    set(ARMV6 1)
 +    add_definitions(-DX265_ARCH_ARM=1 -DHAVE_ARMV6=1 -DHAVE_NEON=0)
@@ -23,7 +23,7 @@
  else()
      message(STATUS "CMAKE_SYSTEM_PROCESSOR value `${CMAKE_SYSTEM_PROCESSOR}` is unknown")
      message(STATUS "Please add this value near ${CMAKE_CURRENT_LIST_FILE}:${CMAKE_CURRENT_LIST_LINE}")
-@@ -122,8 +130,8 @@
+@@ -127,8 +135,8 @@
      if(X86 AND NOT X64)
          add_definitions(-march=i686)
      endif()
@@ -32,11 +32,10 @@
 +    if(ARMV7)
 +        add_definitions(-fPIC)
      endif()
- endif(GCC)
- 
-diff -urN a/source/common/cpu.cpp b/source/common/cpu.cpp
---- a/source/common/cpu.cpp	2014-04-03 00:00:44.000000000 -0600
-+++ b/source/common/cpu.cpp	2014-04-10 07:28:51.937129997 -0600
+     check_cxx_compiler_flag(-Wno-narrowing CC_HAS_NO_NARROWING) 
+     check_cxx_compiler_flag(-ffast-math CC_HAS_FAST_MATH) 
+--- source/common/cpu.cpp.orig	2014-10-04 17:13:58.507875996 +0200
++++ source/common/cpu.cpp	2014-10-04 17:21:11.170912864 +0200
 @@ -37,7 +37,7 @@
  #include <machine/cpu.h>
  #endif

x265-hg20140324.tar.bz2/.hg Deleted

x265-hg20140324.tar.bz2/.hg/00changelog.i Deleted

x265-hg20140324.tar.bz2/.hg/branch Deleted

x265-hg20140324.tar.bz2/.hg/cache Deleted

x265-hg20140324.tar.bz2/.hg/cache/branchheads-served Deleted

x265-hg20140324.tar.bz2/.hg/dirstate Deleted

x265-hg20140324.tar.bz2/.hg/hgrc Deleted

x265-hg20140324.tar.bz2/.hg/requires Deleted

x265-hg20140324.tar.bz2/.hg/store Deleted

x265-hg20140324.tar.bz2/.hg/store/00changelog.d Deleted

x265-hg20140324.tar.bz2/.hg/store/00changelog.i Deleted

x265-hg20140324.tar.bz2/.hg/store/00manifest.d Deleted

x265-hg20140324.tar.bz2/.hg/store/00manifest.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data Deleted

x265-hg20140324.tar.bz2/.hg/store/data/_c_o_p_y_i_n_g.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_build_encoder_applications.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_configure_build.bash.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_configure_build.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_configure_solution.bash.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_configure_solution.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_create_regression_package.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_create_regression_package.sh.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_r_e_a_d_m_e.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_regression_tester.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/_run_encoder_applications.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/build_solution.bash.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/config.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/_basketball_drive.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/_four_people.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/_t_app_decoder.exe.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/_t_app_decoder.exe.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/dr__psnr.exe.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/encoder___i__15_p.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/encoder__all___i.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/dr__psnr__script/psnr__script.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl/make-makefile.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl32 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl32/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl32/make-makefile.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl64 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl64/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/icl64/make-makefile.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/linux Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/linux/batch.py.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/linux/make-_makefiles.bash.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/msys Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/msys/make-_makefiles.sh.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/msys/make-x86__64-w64-mingw32-_makefiles.sh.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/msys/toolchain-x86__64-w64-mingw32.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/nmake Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/nmake/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/01build-and-smoke-test.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/02perftest.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/commandlines-example.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/config-example.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/email-csv.py.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/regression/run.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86__64 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86__64/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc10-x86__64/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86__64 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86__64/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc11-x86__64/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86__64 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86__64/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc12-x86__64/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86__64 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86__64/build-all.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/vc9-x86__64/make-solutions.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/xcode Deleted

x265-hg20140324.tar.bz2/.hg/store/data/build/xcode/make-project.sh.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder___i__15_p.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__all___i.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__intra__main.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__intra__main10.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__lowdelay___p__main.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__lowdelay___p__main10.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__lowdelay__main.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__lowdelay__main10.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__randomaccess__main.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/encoder__randomaccess__main10.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_b_q_mall.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_b_q_square.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_b_q_terrace.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_basketball_drill.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_basketball_drill_text.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_basketball_drive.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_basketball_pass.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_blowing_bubbles.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_cactus.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_china_speed.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_four_people.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_johnny.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_kimono.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_kristen_and_sara.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_nebuta_festival__10bit.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_park_scene.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_party_scene.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_people_on_street.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_race_horses.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_race_horses_c.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_slide_editing.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_slide_show.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_steam_locomotive_train__10bit.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_traffic.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_vidyo1.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_vidyo3.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/cfg/per-sequence/_vidyo4.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_doxyfile.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_lookahead_guide.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_lookahead_plan.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_r_e_a_d_m_e__data-structure.ppt.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_r_e_a_d_m_e__data-structure.ppt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/_unit_test_usage.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle/_a_style.exe.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle/_a_style.exe.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle/apply-to-all-source.py.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle/astyle-config.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/astyle/drag-astyle.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/gop-structure-example.pdf.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/_t16._t_x_t.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/_t32._t_x_t.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/_t4._t_x_t.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/_t8._t_x_t.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/intra-16x16.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/intra-32x32.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/intra-4x4.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/intra/intra-8x8.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/mainpage.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/software-manual.pdf.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/software-manual.pdf.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/apply-to-all-source.py.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/codingstyle.cfg.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/drag-uncrustify.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/uncrustify.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/uncrustify.exe.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/doc/uncrustify/uncrustify.exe.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_r_e_a_d_m_e.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/_t_app_dec_cfg.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/_t_app_dec_cfg.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/_t_app_dec_top.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/_t_app_dec_top.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_decoder/decmain.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/_t_app_enc_cfg.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/_t_app_enc_cfg.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/_t_app_enc_top.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/_t_app_enc_top.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_app/_t_app_encoder/encmain.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_c_o_p_y_i_n_g._h_m.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a/ppa.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a/ppa.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a/ppa_api.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_p_p_a/ppa_c_p_u_events.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_r_e_a_d_m_e.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_app_common Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_app_common/program__options__lite.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_app_common/program__options__lite.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_access_unit.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_common_def.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_context_model.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_context_model.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_context_model3_d_buffer.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_context_model3_d_buffer.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_context_tables.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_n_a_l.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_s_e_i.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_s_e_i.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_bit_counter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_bit_stream.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_bit_stream.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_c_a_b_a_c_tables.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_c_a_b_a_c_tables.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_data_c_u.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_data_c_u.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_interpolation_filter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_interpolation_filter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_list.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_loop_filter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_loop_filter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_motion_info.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_motion_info.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_mv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pattern.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pattern.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic_sym.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic_sym.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic_yuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic_yuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_pic_yuv_m_d5.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_prediction.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_prediction.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rd_cost.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rd_cost.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rd_cost_weight_prediction.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rd_cost_weight_prediction.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rom.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_rom.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_sample_adaptive_offset.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_sample_adaptive_offset.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_slice.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_slice.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_tr_quant.cpp.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_tr_quant.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_tr_quant.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_weight_prediction.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_weight_prediction.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_yuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_t_com_yuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_common/_type_def.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_annex_bread.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_annex_bread.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_n_a_lread.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_n_a_lread.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_s_e_iread.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_s_e_iread.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_syntax_element_parser.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_syntax_element_parser.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_bin_coder.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_bin_coder_c_a_b_a_c.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_bin_coder_c_a_b_a_c.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_c_a_v_l_c.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_c_a_v_l_c.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_cu.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_cu.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_entropy.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_entropy.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_gop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_gop.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_sbac.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_sbac.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_slice.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_slice.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_top.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_decoder/_t_dec_top.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_annex_bwrite.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_n_a_lwrite.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_n_a_lwrite.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_s_e_iwrite.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_s_e_iwrite.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_syntax_element_writer.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_syntax_element_writer.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_analyze.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_analyze.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_bin_coder.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_bin_coder_c_a_b_a_c.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_bin_coder_c_a_b_a_c.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_bin_coder_c_a_b_a_c_counter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_bin_coder_c_a_b_a_c_counter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cavlc.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cavlc.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cfg.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cu.cpp.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cu.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_cu.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_entropy.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_entropy.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_g_o_p.cpp.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_g_o_p.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_g_o_p.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_pic.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_pic.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_preanalyzer.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_preanalyzer.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_rate_ctrl.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_rate_ctrl.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_sample_adaptive_offset.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_sample_adaptive_offset.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_sbac.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_sbac.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_search.cpp.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_search.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_search.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_slice.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_slice.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_top.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_t_enc_top.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_weight_pred_analysis.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_encoder/_weight_pred_analysis.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o/_t_video_i_o.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o/_t_video_i_o_y4m.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o/_t_video_i_o_y4m.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o/_t_video_i_o_yuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_lib_video_i_o/_t_video_i_o_yuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/dispatch__example.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/instrset.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/instrset__detect.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special/complexvec.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special/decimal.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special/quaternion.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special/vector3d.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/special/vectormath.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectorclass.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectorf128.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectorf256.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectorf256e.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectori128.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectori256.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/_t_vector_class/vectori256e.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/config.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/config.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/encoder.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/encoder.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/libmd5 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/libmd5/_m_d5.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/libmd5/libmd5.c.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_lib/libmd5/libmd5.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a/ppa.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a/ppa.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a/ppa_api.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_p_p_a/ppa_c_p_u_events.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_test_for_checking___bad_commit.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/_r_e_a_d_m_e.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/dispatch__example.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/instrset.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/instrset__detect.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special/complexvec.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special/decimal.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special/quaternion.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special/vector3d.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/special/vectormath.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectorclass.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectorf128.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectorf256.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectorf256e.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectori128.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectori256.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/_vector_class/vectori256e.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_c_make_a_s_m___y_a_s_m_information.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_c_make_determine_a_s_m___y_a_s_m_compiler.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_c_make_test_a_s_m___y_a_s_m_compiler.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_find_hg.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_find_v_l_d.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/_find_yasm.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/cmake__uninstall.cmake.in.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/mergestaticlibs.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/cmake/version.cmake.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/_intra_pred.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/_t_short_y_u_v.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/_t_short_y_u_v.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/bitstream.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/butterfly.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/common.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/common.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/cpu.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/cpu.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/dct.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/intrapred.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/ipfilter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/lookahead.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/lookahead.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/loopfilter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/lowres.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/lowres.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/macroblock.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/md5.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/md5.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/mv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/param.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/param.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/piclist.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/piclist.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/pixel.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/primitives.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/reference.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/reference.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/shortyuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/shortyuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/threading.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/threading.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/threadpool.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/threadpool.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy-xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/blockcopy.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct-xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/dct.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intra-xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/intrapred.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter-xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter16.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ipfilter8.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/macroblock.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel-xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel16-sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel16.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/pixel8.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/sse.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/utils.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/vec-primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/vecprimitives.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/vec/xop.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/version.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/wavefront.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/wavefront.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/_r_e_a_d_m_e.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/asm-primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/blockcopy8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/blockcopy8.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/const-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/cpu-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/dct8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/dct8.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/intrapred.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/intrapred.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/intrapred16.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/intrapred8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/ipfilter16.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/ipfilter8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/ipfilter8.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/loopfilter.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/loopfilter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/mc-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/mc-a2.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/mc.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel-32.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel-util.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel-util.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel-util8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixel.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/pixeladd8.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/sad-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/sad16-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/ssd-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/x86inc.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/common/x86/x86util.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/getopt Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/getopt/_l_g_p_l.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/getopt/getopt.c.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/getopt/getopt.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/msvc Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/msvc/_l_g_p_l.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/msvc/getopt.c.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/msvc/getopt.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/compat/msvc/stdint.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/dllmain.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_interpolation_filter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_interpolation_filter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_intra_pred.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_t_com_rd_cost___s_s_e.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_t_com_tr_quant___s_s_e.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_t_short_y_u_v.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/_t_short_y_u_v.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/api.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/bitcost.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/bitcost.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/butterfly.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/common.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/common.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/compress.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/compress.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/cpu__detection.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/cpu__detection.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/cturow.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/cturow.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/dpb.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/dpb.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/encoder.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/encoder.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/frameencoder.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/frameencoder.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/framefilter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/framefilter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/instrset.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/instrset__detect.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/interpolation_filter.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/interpolationfilter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/ipfilter.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/macroblock.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/macroblock.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/md5.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/md5.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/motion.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/motion.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/mv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/pixel.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/primitives.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/ratecontrol.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/ratecontrol.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/reference.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/reference.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/slicetype.cpp.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/slicetype.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/slicetype.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/threading.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/threading.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/threadpool.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/threadpool.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/avx.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/avx2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/blockcopy.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/interpolationfilter.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/intrapred.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/ipfilter.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/ipfilter16.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/ipfilter8.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/macroblock.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/pixel.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/pixel16.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/pixel8.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/sse2.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/sse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/sse41.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/sse42.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/ssse3.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/vec-primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/vec/vecprimitives.inc.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/wavefront.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/wavefront.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/weight_prediction.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/weight_prediction.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86 Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/asm-primitives.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/const-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/cpu-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/pixel-32.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/pixel-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/pixel.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/sad-a.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/util.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/x86inc.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/encoder/x86/x86util.asm.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/input.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/input.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/y4m.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/y4m.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/yuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/input/yuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/output.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/output.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/y4m.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/y4m.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/yuv.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/output/yuv.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/_test_bench.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/_unit_test.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/_unit_test.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/butterflyharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/butterflyharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/filterharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/filterharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/intrapredharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/intrapredharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/ipfilterharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/ipfilterharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/mbdstharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/mbdstharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/pixelharness.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/pixelharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/testbench.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/testharness.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/testpool.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/timer.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/unittest._c_p_p.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/unittest.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/test/unittest.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/_h_m decoder Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/_h_m decoder/_t_app_decoder.exe.d Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/_h_m decoder/_t_app_decoder.exe.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/_test_for_checking___bad_commit.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/_c_make_lists.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/_psnr_calculator.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/_psnr_calculator.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/_s_s_i_m_calculator.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/_s_s_i_m_calculator.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/dr__psnr/dr__psnr.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/performance_profiler Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/performance_profiler/_profiler.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/performance_profiler/_readme.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/performance_profiler/config.txt.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/tools/performance_profiler/performance_profiler.bat.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.def.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.def.in.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.pc.in.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265.rc.in.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265__config.h.in.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265cfg.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265cfg.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265enc.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265enc.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265main.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265opts.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265top.cpp.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/source/x265top.h.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/test__new.rtf.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/~2ehgignore.i Deleted

x265-hg20140324.tar.bz2/.hg/store/data/~2ehgtags.i Deleted

x265-hg20140324.tar.bz2/.hg/store/fncache Deleted

@@ -1,634 +0,0 @@
-data/source/encoder/x86/CMakeLists.txt.i
-data/source/compat/getopt/getopt.h.i
-data/source/encoder/framefilter.cpp.i
-data/source/Lib/TLibCommon/TComInterpolationFilter.cpp.i
-data/source/encoder/ipfilter.cpp.i
-data/source/x265top.cpp.i
-data/source/common/vec/avx.cpp.i
-data/source/encoder/primitives.h.i
-data/build/RegressionTester.bat.i
-data/source/encoder/encoder.h.i
-data/doc/intra/T16.TXT.i
-data/source/encoder/vec/ipfilter8.inc.i
-data/source/encoder/dpb.h.i
-data/source/Lib/TLibDecoder/SyntaxElementParser.cpp.i
-data/source/Lib/PPA/ppa.cpp.i
-data/source/encoder/TComRdCost_SSE.cpp.i
-data/source/encoder/InterpolationFilter.cpp.i
-data/cfg/per-sequence/PeopleOnStreet.cfg.i
-data/source/test/filterharness.cpp.i
-data/source/tools/dr_psnr/dr_psnr.cpp.i
-data/source/Lib/TVectorClass/vectori256e.h.i
-data/source/common/vec/ssse3.cpp.i
-data/source/encoder/framefilter.h.i
-data/source/Lib/TLibEncoder/TEncAnalyze.cpp.i
-data/source/tools/dr_psnr/PsnrCalculator.cpp.i
-data/source/encoder/vec/sse3.cpp.i
-data/build/msys/toolchain-x86_64-w64-mingw32.cmake.i
-data/source/common/threadpool.h.i
-data/doc/intra/intra-16x16.txt.i
-data/source/Lib/TLibCommon/TComYuv.h.i
-data/source/Lib/encoder.h.i
-data/source/Lib/TLibCommon/ContextModel3DBuffer.cpp.i
-data/.hgtags.i
-data/source/test/pixelharness.h.i
-data/source/common/vec/dct-ssse3.cpp.i
-data/build/dr_psnr_script/BasketballDrive.cfg.i
-data/source/common/vec/macroblock.inc.i
-data/source/test/CMakeLists.txt.i
-data/source/common/vec/intra-sse3.cpp.i
-data/source/Lib/TLibCommon/TComPattern.cpp.i
-data/source/common/vec/CMakeLists.txt.i
-data/source/tools/TestForChecking_BadCommit.bat.i
-data/source/Lib/TAppCommon/program_options_lite.cpp.i
-data/source/Lib/TLibCommon/TComSlice.h.i
-data/source/Lib/TLibCommon/TComRdCost.cpp.i
-data/source/Lib/COPYING.HM.i
-data/build/xcode/make-project.sh.i
-data/source/App/TAppEncoder/TAppEncTop.cpp.i
-data/source/dllmain.cpp.i
-data/source/common/x86/ipfilter8.asm.i
-data/source/common/vec/sse2.cpp.i
-data/source/x265.def.i
-data/source/common/pixel.cpp.i
-data/source/Lib/TLibEncoder/TEncCavlc.cpp.i
-data/source/common/x86/dct8.asm.i
-data/doc/uncrustify/apply-to-all-source.py.i
-data/source/Lib/TLibDecoder/CMakeLists.txt.i
-data/build/vc12-x86_64/build-all.bat.i
-data/source/common/primitives.h.i
-data/source/input/input.h.i
-data/source/cmake/CMakeASM_YASMInformation.cmake.i
-data/source/Lib/CMakeLists.txt.i
-data/source/Lib/TLibCommon/TComYuv.cpp.i
-data/source/common/threadpool.cpp.i
-data/source/common/x86/ssd-a.asm.i
-data/source/Lib/TLibCommon/TComList.h.i
-data/source/input/CMakeLists.txt.i
-data/source/common/x86/CMakeLists.txt.i
-data/build/dr_psnr_script/TAppDecoder.exe.d
-data/build/vc9-x86/make-solutions.bat.i
-data/build/CreateRegressionPackage.bat.i
-data/source/Lib/TVectorClass/special/quaternion.h.i
-data/build/dr_psnr_script/TAppDecoder.exe.i
-data/source/Lib/TVectorClass/vectori128.h.i
-data/source/test/mbdstharness.cpp.i
-data/source/Lib/TLibCommon/CommonDef.h.i
-data/source/common/x86/sad-a.asm.i
-data/source/PPA/CMakeLists.txt.i
-data/source/Lib/TLibEncoder/TEncBinCoderCABACCounter.h.i
-data/source/VectorClass/instrset.h.i
-data/source/encoder/vec/vecprimitives.inc.i
-data/source/Lib/TLibEncoder/TEncTop.h.i
-data/source/encoder/common.cpp.i
-data/source/tools/performanceProfiler/Profiler.bat.i
-data/source/common/vec/dct-sse41.cpp.i
-data/source/test/intrapredharness.cpp.i
-data/source/Lib/TLibEncoder/TEncSampleAdaptiveOffset.h.i
-data/cfg/per-sequence/BQSquare.cfg.i
-data/source/encoder/reference.cpp.i
-data/source/encoder/dpb.cpp.i
-data/source/Lib/TLibCommon/TComPicYuv.cpp.i
-data/source/common/vec/pixel-avx.cpp.i
-data/source/Lib/TLibCommon/TComPic.cpp.i
-data/source/VectorClass/vectorf256e.h.i
-data/source/common/vec/ipfilter-sse3.cpp.i
-data/doc/intra/intra-4x4.txt.i
-data/source/Lib/TLibDecoder/TDecTop.h.i
-data/source/output/yuv.h.i
-data/source/encoder/bitcost.h.i
-data/source/test/testbench.cpp.i
-data/source/output/CMakeLists.txt.i
-data/source/output/yuv.cpp.i
-data/source/Lib/TLibEncoder/WeightPredAnalysis.h.i
-data/source/Lib/TLibCommon/ContextModel.h.i
-data/source/Lib/TLibEncoder/CMakeLists.txt.i
-data/source/encoder/vec/pixel.inc.i
-data/source/Lib/TLibCommon/TComBitStream.cpp.i
-data/doc/intra/intra-8x8.txt.i
-data/source/common/threading.h.i
-data/source/Lib/libmd5/libmd5.c.i
-data/source/Lib/TLibCommon/TComMotionInfo.cpp.i
-data/source/Lib/TLibEncoder/TEncCavlc.h.i
-data/source/common/vec/dct-xop.cpp.i
-data/source/Lib/TLibEncoder/TEncSearch.cpp.d
-data/build/regression/01build-and-smoke-test.bat.i
-data/source/Lib/TLibEncoder/TEncSearch.cpp.i
-data/doc/astyle/astyle-config.txt.i
-data/cfg/per-sequence/BasketballDrive.cfg.i
-data/build/ConfigureBuild.bat.i
-data/source/Lib/TLibDecoder/TDecBinCoderCABAC.h.i
-data/source/Lib/TVectorClass/vectorf256.h.i
-data/source/Lib/TLibVideoIO/TVideoIOYuv.cpp.i
-data/source/encoder/pixel.cpp.i
-data/source/Lib/TLibCommon/TComSampleAdaptiveOffset.h.i
-data/source/Lib/TLibEncoder/TEncPreanalyzer.cpp.i
-data/build/BuildEncoderApplications.bat.i
-data/source/x265.rc.in.i
-data/doc/intra/T4.TXT.i
-data/source/Lib/TLibVideoIO/TVideoIO.h.i
-data/source/common/TShortYUV.cpp.i
-data/source/cmake/version.cmake.i
-data/source/Lib/TLibEncoder/TEncCfg.h.i
-data/doc/Doxyfile.i
-data/source/common/vec/vecprimitives.inc.i
-data/source/encoder/vec/vec-primitives.cpp.i
-data/doc/astyle/drag-astyle.bat.i
-data/source/Lib/TLibCommon/TComPicYuvMD5.cpp.i
-data/source/test/pixelharness.cpp.i
-data/doc/mainpage.h.i
-data/build/vc11-x86_64/build-all.bat.i
-data/source/encoder/vec/macroblock.inc.i
-data/source/common/x86/mc-a.asm.i
-data/source/x265.h.i
-data/source/Lib/TLibEncoder/TEncTop.cpp.i
-data/source/common/loopfilter.cpp.i
-data/source/App/TAppDecoder/TAppDecCfg.h.i
-data/source/compat/msvc/LGPL.txt.i
-data/source/encoder/mv.h.i
-data/source/Lib/TLibCommon/TComPicSym.cpp.i
-data/source/x265_config.h.in.i
-data/cfg/per-sequence/BQTerrace.cfg.i
-data/doc/astyle/apply-to-all-source.py.i
-data/source/test/UnitTest.cpp.i
-data/source/Lib/TLibCommon/SEI.cpp.i
-data/cfg/per-sequence/NebutaFestival_10bit.cfg.i
-data/source/Lib/TLibEncoder/TEncRateCtrl.cpp.i
-data/source/common/macroblock.cpp.i
-data/source/Lib/TLibEncoder/TEncBinCoder.h.i
-data/source/common/vec/dct-sse3.cpp.i
-data/source/test/TestBench.cpp.i
-data/build/dr_psnr_script/encoder_I_15P.cfg.i
-data/source/Lib/TLibCommon/ContextModel.cpp.i
-data/source/common/vec/intra-xop.cpp.i
-data/cfg/per-sequence/BQMall.cfg.i
-data/source/common/param.cpp.i
-data/source/common/vec/dct-avx2.cpp.i
-data/source/common/vec/intra-avx.cpp.i
-data/source/common/vec/intra-ssse3.cpp.i
-data/build/vc9-x86_64/make-solutions.bat.i
-data/source/x265.def.in.i
-data/source/Lib/TLibDecoder/AnnexBread.h.i
-data/source/common/x86/cpu-a.asm.i
-data/source/cmake/FindHg.cmake.i
-data/source/encoder/x86/asm-primitives.cpp.i
-data/source/common/x86/intrapred16.asm.i
-data/build/nmake/make-solutions.bat.i
-data/source/Lib/TLibDecoder/NALread.h.i
-data/source/test/testharness.h.i
-data/source/Lib/TLibEncoder/TEncCu.cpp.i
-data/source/Lib/TLibEncoder/TEncCu.cpp.d
-data/source/common/vec/ipfilter-avx2.cpp.i
-data/source/encoder/x86/pixel-32.asm.i
-data/source/common/vec/dct.inc.i
-data/source/encoder/vec/avx.cpp.i
-data/doc/UnitTestUsage.txt.i
-data/source/cmake/mergestaticlibs.cmake.i
-data/source/test/unittest.cpp.i
-data/source/PPA/ppa.h.i
-data/source/input/yuv.h.i
-data/source/Lib/TLibDecoder/TDecTop.cpp.i
-data/source/Lib/TVectorClass/vectori256.h.i
-data/source/common/vec/ipfilter-avx.cpp.i
-data/source/Lib/TLibEncoder/TEncAnalyze.h.i
-data/source/Lib/TLibCommon/ContextModel3DBuffer.h.i
-data/source/common/lookahead.cpp.i
-data/source/Lib/PPA/ppaApi.h.i
-data/source/common/shortyuv.h.i
-data/source/Lib/TLibCommon/TComCABACTables.h.i
-data/source/common/vec/pixel-sse41.cpp.i
-data/source/Lib/TLibDecoder/AnnexBread.cpp.i
-data/source/encoder/vec/ssse3.cpp.i
-data/source/encoder/vec/sse42.cpp.i
-data/source/App/TAppEncoder/TAppEncCfg.cpp.i
-data/source/common/vec/vec-primitives.cpp.i
-data/source/App/TAppDecoder/decmain.cpp.i
-data/doc/uncrustify/drag-uncrustify.bat.i
-data/source/tools/dr_psnr/SSIMCalculator.cpp.i
-data/source/Lib/TLibEncoder/TEncSbac.h.i
-data/source/Lib/TLibCommon/TComInterpolationFilter.h.i
-data/source/Lib/TLibCommon/TComMotionInfo.h.i
-data/build/regression/run.bat.i
-data/doc/intra/T8.TXT.i
-data/build/dr_psnr_script/psnr_script.bat.i
-data/source/Lib/TLibCommon/TComSlice.cpp.i
-data/source/encoder/cpu_detection.h.i
-data/source/encoder/vec/intrapred.inc.i
-data/source/App/TAppEncoder/encmain.cpp.i
-data/source/tools/performanceProfiler/performanceProfiler.bat.i
-data/source/encoder/vec/CMakeLists.txt.i
-data/source/common/vec/pixel-sse42.cpp.i
-data/source/Lib/TLibCommon/TComLoopFilter.h.i
-data/source/Lib/TLibCommon/TComRdCostWeightPrediction.h.i
-data/build/vc12-x86_64/make-solutions.bat.i
-data/source/Lib/TLibCommon/NAL.h.i
-data/source/Lib/TVectorClass/dispatch_example.cpp.i
-data/source/common/vec/avx2.cpp.i
-data/source/PPA/ppaCPUEvents.h.i
-data/source/Lib/TVectorClass/instrset.h.i
-data/source/Lib/TLibEncoder/TEncGOP.h.i
-data/source/Lib/TLibCommon/TComPattern.h.i
-data/source/common/reference.cpp.i
-data/source/Lib/TLibDecoder/TDecEntropy.cpp.i
-data/source/common/x86/pixel-util.asm.i
-data/cfg/per-sequence/SlideEditing.cfg.i
-data/source/Lib/TLibCommon/TComCABACTables.cpp.i
-data/source/encoder/api.cpp.i
-data/source/common/x86/x86util.asm.i
-data/source/Lib/TLibDecoder/SEIread.h.i
-data/source/x265top.h.i
-data/source/common/threading.cpp.i
-data/source/encoder/butterfly.h.i
-data/source/encoder/TShortYUV.h.i
-data/source/common/vec/ipfilter-sse42.cpp.i
-data/source/encoder/threading.h.i
-data/source/input/input.cpp.i
-data/source/input/y4m.h.i
-data/build/linux/make-Makefiles.bash.i
-data/cfg/per-sequence/Vidyo1.cfg.i
-data/cfg/per-sequence/SlideShow.cfg.i
-data/source/common/lowres.h.i
-data/source/Lib/TLibCommon/TComRom.h.i
-data/source/common/vec/ipfilter8.inc.i
-data/source/x265cfg.cpp.i
-data/source/common/bitstream.h.i
-data/source/common/vec/pixel-sse2.cpp.i
-data/source/Lib/TLibEncoder/TEncPreanalyzer.h.i
-data/source/Lib/TLibCommon/TComPicYuv.h.i
-data/source/common/md5.cpp.i
-data/source/common/vec/pixel-sse3.cpp.i
-data/source/common/vec/utils.h.i
-data/source/x265main.cpp.i
-data/source/common/butterfly.h.i
-data/source/Lib/TLibEncoder/TEncBinCoderCABACCounter.cpp.i
-data/source/Lib/TLibDecoder/TDecBinCoderCABAC.cpp.i
-data/build/ConfigureSolution.bat.i
-data/source/Lib/TLibEncoder/TEncSlice.cpp.i
-data/source/common/mv.h.i
-data/source/common/vec/blockcopy-avx.cpp.i
-data/source/encoder/cpu_detection.cpp.i
-data/source/common/IntraPred.cpp.i
-data/source/output/y4m.cpp.i
-data/source/encoder/vec/ipfilter16.inc.i
-data/source/encoder/x86/x86inc.asm.i
-data/source/encoder/slicetype.cpp.i
-data/source/encoder/wavefront.h.i
-data/source/encoder/slicetype.cpp.d
-data/source/VectorClass/vectori256.h.i
-data/source/output/output.cpp.i
-data/source/common/vec/intra-sse42.cpp.i
-data/source/Lib/TVectorClass/special/vector3d.h.i
-data/source/App/README.txt.i
-data/source/Lib/TVectorClass/vectorclass.h.i
-data/source/VectorClass/dispatch_example.cpp.i
-data/source/test/intrapredharness.h.i
-data/source/encoder/vec/blockcopy.inc.i
-data/source/common/x86/sad16-a.asm.i
-data/source/Lib/TLibVideoIO/TVideoIOY4m.h.i
-data/source/Lib/TLibCommon/TComBitStream.h.i
-data/build/vc10-x86_64/make-solutions.bat.i
-data/source/Lib/TLibCommon/TComTrQuant.h.i
-data/source/Lib/TLibCommon/SEI.h.i
-data/source/common/vec/blockcopy-ssse3.cpp.i
-data/source/compat/msvc/getopt.h.i
-data/source/common/vec/ipfilter.inc.i
-data/source/Lib/TLibCommon/TComMv.h.i
-data/source/common/common.cpp.i
-data/source/common/vec/pixel16-sse41.cpp.i
-data/build/CreateRegressionPackage.sh.i
-data/build/vc10-x86_64/build-all.bat.i
-data/source/common/x86/intrapred.h.i
-data/source/x265cfg.h.i
-data/source/common/ipfilter.cpp.i
-data/source/encoder/InterpolationFilter.h.i
-data/source/Lib/config.cpp.i
-data/source/common/vec/pixel.inc.i
-data/source/Lib/TVectorClass/special/complexvec.h.i
-data/build/icl64/make-makefile.bat.i
-data/build/vc11-x86/build-all.bat.i
-data/source/Lib/libmd5/libmd5.h.i
-data/source/test/UnitTest.h.i
-data/source/Lib/TLibCommon/TComWeightPrediction.h.i
-data/source/compat/getopt/LGPL.txt.i
-data/source/common/x86/pixeladd8.asm.i
-data/source/output/output.h.i
-data/source/common/vec/ipfilter-xop.cpp.i
-data/source/common/x86/x86inc.asm.i
-data/source/test/filterharness.h.i
-data/source/tools/dr_psnr/PsnrCalculator.h.i
-data/source/VectorClass/special/complexvec.h.i
-data/source/Lib/PPA/CMakeLists.txt.i
-data/source/encoder/motion.h.i
-data/source/encoder/common.h.i
-data/source/encoder/CMakeLists.txt.i
-data/build/regression/commandlines-example.txt.i
-data/source/encoder/reference.h.i
-data/source/encoder/cturow.h.i
-data/source/Lib/TLibCommon/TComPicSym.h.i
-data/source/common/x86/ipfilter8.h.i
-data/source/Lib/TLibEncoder/TEncSearch.h.i
-data/source/common/vec/intra-sse2.cpp.i
-data/source/common/vec/intrapred.inc.i
-data/source/encoder/vec/sse2.cpp.i
-data/source/encoder/vec/pixel8.inc.i
-data/source/VectorClass/vectorf256.h.i
-data/source/output/y4m.h.i
-data/cfg/encoder_lowdelay_P_main10.cfg.i
-data/source/tools/HM decoder/TAppDecoder.exe.i
-data/source/common/vec/blockcopy.inc.i
-data/build/dr_psnr_script/FourPeople.cfg.i
-data/source/common/wavefront.cpp.i
-data/source/Lib/TVectorClass/vectorf256e.h.i
-data/source/compat/getopt/getopt.c.i
-data/source/common/vec/sse41.cpp.i
-data/cfg/per-sequence/BasketballDrill.cfg.i
-data/source/Lib/PPA/ppa.h.i
-data/source/cmake/FindVLD.cmake.i
-data/source/encoder/cturow.cpp.i
-data/source/test/butterflyharness.h.i
-data/source/Lib/TLibEncoder/TEncSbac.cpp.i
-data/source/common/x86/pixel.h.i
-data/source/test/butterflyharness.cpp.i
-data/source/common/x86/pixel-a.asm.i
-data/source/Lib/TLibCommon/CMakeLists.txt.i
-data/source/common/vec/ipfilter-sse41.cpp.i
-data/source/common/vec/intra-avx2.cpp.i
-data/source/encoder/TShortYUV.cpp.i
-data/source/encoder/ratecontrol.cpp.i
-data/cfg/per-sequence/Vidyo3.cfg.i
-data/source/encoder/x86/pixel-a.asm.i
-data/source/Lib/TLibEncoder/TEncPic.h.i
-data/build/msys/make-Makefiles.sh.i
-data/source/VectorClass/CMakeLists.txt.i
-data/doc/intra/T32.TXT.i
-data/source/common/x86/intrapred.asm.i
-data/doc/gop-structure-example.pdf.i
-data/source/cmake/CMakeDetermineASM_YASMCompiler.cmake.i
-data/build/buildSolution.bash.i
-data/source/App/TAppDecoder/CMakeLists.txt.i
-data/source/encoder/instrset_detect.cpp.i
-data/source/common/vec/ipfilter16.inc.i
-data/build/msys/make-x86_64-w64-mingw32-Makefiles.sh.i
-data/source/common/CMakeLists.txt.i
-data/source/Lib/TLibCommon/TComLoopFilter.cpp.i
-data/cfg/encoder_all_I.cfg.i
-data/build/regression/email-csv.py.i
-data/source/Lib/TVectorClass/instrset_detect.cpp.i
-data/build/icl64/build-all.bat.i
-data/source/common/piclist.h.i
-data/source/x265enc.h.i
-data/source/common/x86/pixel-util.h.i
-data/source/common/x86/mc-a2.asm.i
-data/source/VectorClass/vectorf128.h.i
-data/source/encoder/macroblock.h.i
-data/source/Lib/TLibDecoder/TDecBinCoder.h.i
-data/source/Lib/TLibVideoIO/TVideoIOY4m.cpp.i
-data/source/Lib/TLibEncoder/SyntaxElementWriter.cpp.i
-data/source/Lib/TLibDecoder/TDecGop.h.i
-data/source/common/vec/dct-sse2.cpp.i
-data/source/cmake/CMakeTestASM_YASMCompiler.cmake.i
-data/doc/astyle/AStyle.exe.i
-data/source/common/vec/pixel-ssse3.cpp.i
-data/source/Lib/TLibEncoder/NALwrite.cpp.i
-data/doc/astyle/AStyle.exe.d
-data/source/VectorClass/README.txt.i
-data/source/VectorClass/special/decimal.h.i
-data/source/Lib/TLibDecoder/SyntaxElementParser.h.i
-data/source/Lib/TVectorClass/special/vectormath.h.i
-data/source/common/vec/ipfilter-ssse3.cpp.i
-data/source/tools/dr_psnr/CMakeLists.txt.i
-data/build/icl32/build-all.bat.i
-data/cfg/encoder_intra_main10.cfg.i
-data/source/common/x86/blockcopy8.asm.i
-data/source/common/vec/pixel8.inc.i
-data/source/common/lookahead.h.i
-data/cfg/encoder_randomaccess_main.cfg.i
-data/source/Lib/TLibCommon/ContextTables.h.i
-data/build/vc12-x86/make-solutions.bat.i
-data/source/tools/dr_psnr/SSIMCalculator.h.i
-data/build/ConfigureBuild.bash.i
-data/source/compat/msvc/stdint.h.i
-data/build/regression/config-example.txt.i
-data/source/encoder/TComTrQuant_SSE.cpp.i
-data/cfg/encoder_randomaccess_main10.cfg.i
-data/cfg/per-sequence/Kimono.cfg.i
-data/.hgignore.i
-data/source/encoder/vec/avx2.cpp.i
-data/source/encoder/threading.cpp.i
-data/source/Lib/PPA/ppaCPUEvents.h.i
-data/source/App/TAppEncoder/TAppEncTop.h.i
-data/source/input/y4m.cpp.i
-data/cfg/per-sequence/RaceHorses.cfg.i
-data/source/Lib/TVectorClass/CMakeLists.txt.i
-data/source/Lib/TLibDecoder/TDecSlice.cpp.i
-data/cfg/per-sequence/FourPeople.cfg.i
-data/cfg/encoder_intra_main.cfg.i
-data/cfg/per-sequence/KristenAndSara.cfg.i
-data/cfg/encoder_lowdelay_main10.cfg.i
-data/source/Lib/encoder.cpp.i
-data/source/Lib/TLibCommon/TComWeightPrediction.cpp.i
-data/source/Lib/TLibCommon/TComDataCU.h.i
-data/source/encoder/frameencoder.cpp.i
-data/source/encoder/weightPrediction.h.i
-data/source/Lib/TLibCommon/AccessUnit.h.i
-data/build/vc9-x86_64/build-all.bat.i
-data/source/common/version.cpp.i
-data/source/Lib/TLibEncoder/AnnexBwrite.h.i
-data/source/Lib/TLibEncoder/SEIwrite.h.i
-data/source/common/vec/pixel-xop.cpp.i
-data/source/Lib/TLibEncoder/TEncPic.cpp.i
-data/doc/uncrustify/codingstyle.cfg.i
-data/source/Lib/TLibDecoder/TDecEntropy.h.i
-data/source/common/wavefront.h.i
-data/source/Lib/libmd5/MD5.h.i
-data/cfg/encoder_lowdelay_P_main.cfg.i
-data/build/config.txt.i
-data/source/common/vec/blockcopy-avx2.cpp.i
-data/doc/software-manual.pdf.d
-data/cfg/per-sequence/ChinaSpeed.cfg.i
-data/source/Lib/TLibCommon/TComPrediction.h.i
-data/source/common/vec/blockcopy-sse41.cpp.i
-data/source/test/ipfilterharness.h.i
-data/doc/software-manual.pdf.i
-data/source/encoder/x86/pixel.h.i
-data/source/common/vec/blockcopy-sse3.cpp.i
-data/source/common/x86/ipfilter16.asm.i
-data/source/common/x86/pixel-util8.asm.i
-data/source/VectorClass/vectori256e.h.i
-data/source/Lib/TLibDecoder/TDecSbac.cpp.i
-data/source/Lib/TLibCommon/TComSampleAdaptiveOffset.cpp.i
-data/source/Lib/TLibEncoder/TEncCu.h.i
-data/source/encoder/x86/sad-a.asm.i
-data/build/vc10-x86/make-solutions.bat.i
-data/build/regression/02perftest.bat.i
-data/source/VectorClass/special/quaternion.h.i
-data/source/Lib/TVectorClass/special/decimal.h.i
-data/source/Lib/TLibEncoder/TEncBinCoderCABAC.cpp.i
-data/source/common/vec/pixel-avx2.cpp.i
-data/source/Lib/TLibEncoder/TEncBinCoderCABAC.h.i
-data/cfg/per-sequence/ParkScene.cfg.i
-data/doc/intra/intra-32x32.txt.i
-data/build/dr_psnr_script/dr_psnr.exe.i
-data/source/Lib/TLibCommon/TComPic.h.i
-data/source/encoder/threadpool.cpp.i
-data/source/common/vec/dct-sse42.cpp.i
-data/build/ConfigureSolution.bash.i
-data/source/Lib/TLibCommon/TComRdCostWeightPrediction.cpp.i
-data/source/common/vec/xop.cpp.i
-data/source/common/x86/intrapred8.asm.i
-data/source/Lib/TLibCommon/TComPrediction.cpp.i
-data/source/Lib/TLibEncoder/NALwrite.h.i
-data/build/RunEncoderApplications.bat.i
-data/cfg/per-sequence/BasketballPass.cfg.i
-data/source/encoder/interpolationfilter.cpp.i
-data/source/tools/HM decoder/TAppDecoder.exe.d
-data/test_new.rtf.i
-data/source/common/reference.h.i
-data/source/Lib/TLibEncoder/TEncSlice.h.i
-data/source/Lib/TLibDecoder/SEIread.cpp.i
-data/source/common/common.h.i
-data/source/test/testpool.cpp.i
-data/source/common/x86/mc.h.i
-data/source/App/TAppDecoder/TAppDecTop.h.i
-data/source/encoder/ratecontrol.h.i
-data/source/common/x86/loopfilter.asm.i
-data/source/common/x86/dct8.h.i
-data/source/encoder/primitives.cpp.i
-data/build/icl/make-makefile.bat.i
-data/source/encoder/motion.cpp.i
-data/source/test/mbdstharness.h.i
-data/source/Lib/TLibVideoIO/TVideoIOYuv.h.i
-data/source/x265.cpp.i
-data/source/Lib/TLibDecoder/TDecCAVLC.h.i
-data/build/vc11-x86_64/make-solutions.bat.i
-data/source/test/timer.cpp.i
-data/source/Lib/TLibDecoder/TDecSbac.h.i
-data/build/vc11-x86/make-solutions.bat.i
-data/source/common/vec/pixel16.inc.i
-data/source/Lib/TLibEncoder/SEIwrite.cpp.i
-data/source/common/vec/sse3.cpp.i
-data/build/vc12-x86/build-all.bat.i
-data/source/Lib/config.h.i
-data/source/Lib/TLibEncoder/WeightPredAnalysis.cpp.i
-data/source/App/TAppEncoder/TAppEncCfg.h.i
-data/source/encoder/slicetype.h.i
-data/source/Lib/TLibCommon/TComBitCounter.h.i
-data/source/encoder/compress.h.i
-data/doc/LookaheadPlan.txt.i
-data/source/encoder/md5.h.i
-data/source/encoder/md5.cpp.i
-data/source/Lib/TLibEncoder/SyntaxElementWriter.h.i
-data/source/Lib/TLibDecoder/TDecGop.cpp.i
-data/source/compat/msvc/getopt.c.i
-data/source/tools/CMakeLists.txt.i
-data/source/App/TAppDecoder/TAppDecCfg.cpp.i
-data/source/encoder/x86/x86util.asm.i
-data/source/App/TAppDecoder/TAppDecTop.cpp.i
-data/source/tools/performanceProfiler/config.txt.i
-data/source/Lib/TLibEncoder/TEncGOP.cpp.d
-data/source/common/x86/pixel-32.asm.i
-data/source/Lib/TLibEncoder/TEncGOP.cpp.i
-data/source/encoder/encoder.cpp.i
-data/doc/LookaheadGuide.txt.i
-data/source/Lib/TVectorClass/vectorf128.h.i
-data/build/linux/batch.py.i
-data/source/encoder/x86/cpu-a.asm.i
-data/source/test/unittest.CPP.i
-data/source/tools/performanceProfiler/Readme.txt.i
-data/source/Lib/TLibEncoder/TEncRateCtrl.h.i
-data/source/common/vec/blockcopy-sse42.cpp.i
-data/doc/README_data-structure.ppt.d
-data/doc/README_data-structure.ppt.i
-data/source/Lib/TLibEncoder/TEncEntropy.cpp.i
-data/source/common/vec/sse.inc.i
-data/source/encoder/macroblock.cpp.i
-data/source/App/TAppEncoder/CMakeLists.txt.i
-data/source/encoder/IntraPred.cpp.i
-data/source/common/x86/asm-primitives.cpp.i
-data/source/test/unittest.h.i
-data/source/Lib/TLibDecoder/TDecCu.cpp.i
-data/source/common/piclist.cpp.i
-data/source/common/primitives.cpp.i
-data/source/x265.pc.in.i
-data/source/encoder/interpolationFilter.h.i
-data/source/x265enc.cpp.i
-data/cfg/per-sequence/Cactus.cfg.i
-data/source/common/cpu.h.i
-data/source/common/vec/sse42.cpp.i
-data/source/Lib/TLibCommon/TComTrQuant.cpp.i
-data/source/Lib/TAppCommon/program_options_lite.h.i
-data/source/Lib/TLibCommon/TComTrQuant.cpp.d
-data/source/Lib/TLibDecoder/TDecCu.h.i
-data/source/common/intrapred.cpp.i
-data/cfg/per-sequence/SteamLocomotiveTrain_10bit.cfg.i
-data/source/common/vec/ipfilter-sse2.cpp.i
-data/source/encoder/threadpool.h.i
-data/cfg/encoder_I_15P.cfg.i
-data/source/common/x86/README.txt.i
-data/source/common/md5.h.i
-data/source/Lib/TLibDecoder/TDecCAVLC.cpp.i
-data/source/encoder/compress.cpp.i
-data/source/common/dct.cpp.i
-data/source/encoder/x86/const-a.asm.i
-data/cfg/per-sequence/Vidyo4.cfg.i
-data/cfg/per-sequence/Johnny.cfg.i
-data/source/x265opts.h.i
-data/source/common/shortyuv.cpp.i
-data/build/dr_psnr_script/encoder_all_I.cfg.i
-data/source/VectorClass/instrset_detect.cpp.i
-data/source/VectorClass/special/vectormath.h.i
-data/doc/uncrustify/uncrustify.exe.i
-data/source/Lib/TLibCommon/TComDataCU.cpp.i
-data/doc/uncrustify/uncrustify.exe.d
-data/cfg/encoder_lowdelay_main.cfg.i
-data/cfg/per-sequence/BasketballDrillText.cfg.i
-data/source/PPA/ppaApi.h.i
-data/source/Lib/TLibDecoder/NALread.cpp.i
-data/source/input/yuv.cpp.i
-data/source/encoder/vec/sse41.cpp.i
-data/doc/uncrustify/uncrustify.bat.i
-data/source/VectorClass/vectorclass.h.i
-data/source/Lib/TLibEncoder/TEncSampleAdaptiveOffset.cpp.i
-data/cfg/per-sequence/BlowingBubbles.cfg.i
-data/source/encoder/frameencoder.h.i
-data/source/Lib/README.txt.i
-data/source/VectorClass/vectori128.h.i
-data/source/encoder/x86/util.h.i
-data/source/encoder/bitcost.cpp.i
-data/source/VectorClass/special/vector3d.h.i
-data/source/cmake/cmake_uninstall.cmake.in.i
-data/source/common/x86/const-a.asm.i
-data/source/Lib/TLibCommon/TComRdCost.h.i
-data/build/icl/build-all.bat.i
-data/COPYING.i
-data/source/cmake/FindYasm.cmake.i
-data/source/encoder/weightPrediction.cpp.i
-data/source/Lib/TLibDecoder/TDecSlice.h.i
-data/build/vc9-x86/build-all.bat.i
-data/source/encoder/wavefront.cpp.i
-data/source/encoder/vec/pixel16.inc.i
-data/build/vc10-x86/build-all.bat.i
-data/source/common/vec/dct-avx.cpp.i
-data/source/common/vec/blockcopy-xop.cpp.i
-data/cfg/per-sequence/Traffic.cfg.i
-data/source/encoder/vec/interpolationfilter.inc.i
-data/source/TestForChecking_BadCommit.bat.i
-data/source/common/vec/intra-sse41.cpp.i
-data/source/common/TShortYUV.h.i
-data/source/common/x86/loopfilter.h.i
-data/source/Lib/TLibCommon/TComRom.cpp.i
-data/source/encoder/vec/ipfilter.inc.i
-data/source/common/x86/blockcopy8.h.i
-data/cfg/per-sequence/RaceHorsesC.cfg.i
-data/source/common/param.h.i
-data/source/encoder/instrset.h.i
-data/build/README.txt.i
-data/source/Lib/TLibEncoder/TEncEntropy.h.i
-data/source/test/ipfilterharness.cpp.i
-data/build/icl32/make-makefile.bat.i
-data/source/Lib/TLibCommon/TypeDef.h.i
-data/source/CMakeLists.txt.i
-data/cfg/per-sequence/PartyScene.cfg.i
-data/source/common/lowres.cpp.i
-data/source/common/cpu.cpp.i
-data/source/PPA/ppa.cpp.i

x265-hg20140324.tar.bz2/.hg/store/phaseroots Deleted

x265-hg20140324.tar.bz2/.hg/store/undo Deleted

x265-hg20140324.tar.bz2/.hg/store/undo.phaseroots Deleted

x265-hg20140324.tar.bz2/.hg/undo.bookmarks Deleted

x265-hg20140324.tar.bz2/.hg/undo.branch Deleted

x265-hg20140324.tar.bz2/.hg/undo.desc Deleted

x265-hg20140324.tar.bz2/.hg/undo.dirstate Deleted

x265-hg20140324.tar.bz2/.hgignore Deleted

x265-hg20140324.tar.bz2/.hgtags Deleted

x265-hg20140324.tar.bz2/build/regression Deleted

x265-hg20140324.tar.bz2/build/regression/01build-and-smoke-test.bat Deleted

@@ -1,97 +0,0 @@
-@echo off
-
-if not exist %buildconfig% mkdir %buildconfig%
-cd %buildconfig%
-if exist *.txt del *.txt
-
-call:makesolution 8bpp "-D ENABLE_PPA:BOOL=ON"
-call:makesolution 16bpp "-D HIGH_BIT_DEPTH:BOOL=ON"
-exit /B
-
-:makesolution
-set depth=%~1
-if not exist %depth% mkdir %depth%
-cd %depth%
-set name=%generator%-%depth%
-
-:: nuke existing cmake cache
-if exist CMakeFiles rd /s /q CMakeFiles
-if exist CMakeCache.txt del CMakeCache.txt
-if exist cmake_install.cmake del cmake_install.cmake
-
-echo Running cmake for %name%
-if "%buildconfig%" == "msys" (
-
-  echo cd "%CD%" > buildscript.sh
-  echo cmake -D ENABLE_TESTS:BOOL=ON %~2 -G "%generator%" ../../../../source >> buildscript.sh
-  echo make >> buildscript.sh
-
-  %msys% -l "%CD%\buildscript.sh"
-  if exist x265.exe (
-    rem We cannot test MSYS x265 without running in MSYS environment
-    cd ..
-    exit /b 0
-  ) else (
-    echo %name% could not create an x265.exe >> "%LOG%"
-    cd ..
-    exit /b 1
-  )
-
-) else (
-
-  call "%compiler%\..\..\VC\vcvarsall.bat"
-  cmake -D ENABLE_TESTS:BOOL=ON %~2 -G "%generator%" ..\..\..\..\source >> ..\cmake%depth%.txt
-  if not exist x265.sln (
-    echo %name% solution was not created >> "%LOG%"
-    cd ..
-    exit /b 1
-  )
-
-  echo Compiling for release...
-  MSBuild /property:Configuration="Release" x265.sln >> ..\build%depth%_release.txt
-  if %errorlevel% equ 1 (
-    echo Release %name% build failed, refer the build log >> "%LOG%"
-    cd ..
-    exit /b 1
-  )
-  
-  echo Compiling for debug...
-  MSBuild /property:Configuration="Debug" x265.sln >> ..\build%depth%_debug.txt
-  if %errorlevel% equ 1 (
-    echo Debug %name% build failed, refer the build log >> "%LOG%"
-    cd ..
-    exit /b 1
-  )
-)
-
-echo Smoke tests...
-if exist Release\x265.exe (
-  Release\x265.exe %video1% -f %testframes% --wpp --hash 1 -o str1.out -r rec1.yuv --no-progress >> ..\encoder_%depth%.txt 2>&1
-  Release\x265.exe %video2% -f %testframes% --wpp --hash 1 -o str2.out -r rec2.yuv --no-progress >> ..\encoder_%depth%.txt 2>&1
-  Release\x265.exe %video3% -f %testframes% --wpp --hash 1 -o str3.out -r rec3.yuv --no-progress >> ..\encoder_%depth%.txt 2>&1
-
-  %decoder% -b str1.out -o str1.yuv >> ..\decoder_%depth%.txt
-  %decoder% -b str2.out -o str2.yuv >> ..\decoder_%depth%.txt
-  %decoder% -b str3.out -o str3.yuv >> ..\decoder_%depth%.txt
-
-  FC /b rec1.yuv str1.yuv > NUL || echo Reconstructed frame mismatch for %name% %video1% >> ..\..\DiffBin.txt
-  FC /b rec2.yuv str2.yuv > NUL || echo Reconstructed frame mismatch for %name% %video2% >> ..\..\DiffBin.txt
-  FC /b rec3.yuv str3.yuv > NUL || echo Reconstructed frame mismatch for %name% %video3% >> ..\..\DiffBin.txt
-)
-
-echo Leak test...
-if exist Debug\x265.exe (
-:: hopefully you have VLD installed so this will check for leaks
-  Debug\x265.exe %video1% -f %testframes% --wpp -o str4.out -r rec4.yuv --no-progress >> ..\encoder_%depth%.txt 2>&1
-  %decoder% -b str4.out -o str4.yuv >> ..\decoder_%depth%.txt
-  FC /b rec4.yuv str4.yuv > NUL || echo Reconstructed frames mismatch for debug %name% >> ..\..\DiffBin.txt
-)
-
-echo Testbench...
-if exist test\Release\TestBench.exe (
-  test\Release\TestBench.exe >> ..\testbench.txt 2>&1 || echo %name% testbench failed >> "%LOG%"
-)
-if exist test\TestBench.exe (
-  test\TestBench.exe >> ..\testbench.txt 2>&1 || echo %name% testbench failed >> "%LOG%"
-)
-cd ..

x265-hg20140324.tar.bz2/build/regression/02perftest.bat Deleted

x265-hg20140324.tar.bz2/build/regression/commandlines-example.txt Deleted

x265-hg20140324.tar.bz2/build/regression/config-example.txt Deleted

x265-hg20140324.tar.bz2/build/regression/email-csv.py Deleted

@@ -1,49 +0,0 @@
-#!/usr/bin/python
-# this script requires python3
-
-import os
-import smtplib
-import glob
-from email.mime.image import MIMEImage
-from email.mime.multipart import MIMEMultipart
-from email.mime.text import MIMEText
-
-SMTP_SERVER = 'smtp.gmail.com'
-SMTP_PORT = 587
-sender = os.getenv("SMTP_USER", "sender@email.com")
-password = os.getenv("SMTP_PASSWD", "mypassword")
-recipient = 'receiver@email.com'
-subject = 'Performance results from regression test'
-message = 'results attached'
-
-def main():
-    msg = MIMEMultipart()
-    msg['Subject'] = 'Performance Results'
-    msg['To'] = recipient
-    msg['From'] = sender
-
-    for csvfile in glob.glob("*.csv"):
-        if not os.path.isfile(csvfile):
-            continue
-        text = MIMEImage(open(csvfile, 'rb').read(), _subtype="csv")
-        text.add_header('Content-Disposition', 'attachment', filename=csvfile)
-        msg.attach(text)
-
-    part = MIMEText('text', "plain")
-    part.set_payload(message)
-    msg.attach(part)
-    session = smtplib.SMTP(SMTP_SERVER, SMTP_PORT)
-    try:
-        session.ehlo()
-        session.starttls()
-        session.ehlo
-        session.login(sender, password)
-        session.sendmail(sender, recipient, msg.as_string())
-        print('your message has been sent')
-    except Exception as e:
-        print('Unable to send email', e)
-    finally:
-        session.quit()
-
-if __name__ == '__main__':
-    main()

x265-hg20140324.tar.bz2/build/regression/run.bat Deleted

@@ -1,56 +0,0 @@
-@echo off
-
-if not exist config.txt (
-  echo config.txt does not exist.
-  echo Copy config-example to config.txt and edit as necessary
-  exit /b 1
-)
-if not exist commandlines.txt (
-  echo commandlines.txt does not exist.
-  echo Copy commandlines-example to commandlines.txt and edit as necessary
-  exit /b 1
-)
-
-for /f "tokens=1,2,3,4,5,6,7,8 delims==" %%a in (config.txt) do (
-  if %%a==repository set repository=%%b
-  if %%a==testframes set testframes=%%b
-  if %%a==perfframes set perfframes=%%b
-  if %%a==msys set msys=%%b
-  if %%a==decoder set decoder=%%b
-  if %%a==video1 set video1=%%b
-  if %%a==video2 set video2=%%b
-  if %%a==video3 set video3=%%b
-)
-
-set CWD=%CD%
-set LOG="%CWD%"\regression.log
-if exist %LOG% del %LOG%
-
-hg pull -u "%repository%"
-if %errorlevel% equ 1 (
-  echo "Pull failed" >> "%LOG%"
-  exit /b 1
-)
-hg summary >> "%LOG%"
-hg summary
-
-call:buildconfigs "%msys%" msys "MSYS Makefiles"
-call:buildconfigs "%VS110COMNTOOLS%" vc11-x86_64 "Visual Studio 11 Win64"
-call:buildconfigs "%VS110COMNTOOLS%" vc11-x86 "Visual Studio 11"
-call:buildconfigs "%VS100COMNTOOLS%" vc10-x86_64 "Visual Studio 10 Win64"
-call:buildconfigs "%VS100COMNTOOLS%" vc10-x86 "Visual Studio 10"
-call:buildconfigs "%VS90COMNTOOLS%" vc9-x86_64 "Visual Studio 9 2008 Win64"
-call:buildconfigs "%VS90COMNTOOLS%" vc9-x86 "Visual Studio 9 2008 Win64"
-
-call 02perftest.bat
-EXIT /B
-
-:buildconfigs
-set compiler=%~1
-set buildconfig=%2
-set generator=%~3
-
-if exist %compiler% (
-  echo Detected %generator%, building >> "%LOG%"
-  cmd /c 01build-and-smoke-test.bat
-)

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/NAL.h Deleted

@@ -1,79 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#ifndef X265_NAL_H
-#define X265_NAL_H
-
-#include "CommonDef.h"
-#include "x265.h"
-
-namespace x265 {
-// private namespace
-
-/**
- * Represents a single NALunit header and the associated RBSPayload
- */
-struct NALUnit
-{
-    NalUnitType m_nalUnitType;       ///< nal_unit_type
-    uint32_t    m_temporalId;        ///< temporal_id
-    uint32_t    m_reservedZero6Bits; ///< reserved_zero_6bits
-
-    NALUnit() : m_nalUnitType(NAL_UNIT_INVALID), m_temporalId(0), m_reservedZero6Bits(0){}
-
-    NALUnit(NalUnitType nalUnitType)
-    {
-        m_nalUnitType = nalUnitType;
-        m_temporalId = 0;
-        m_reservedZero6Bits = 0;
-    }
-};
-
-/**
- * A single NALunit, with complete payload in EBSP format.
- */
-struct NALUnitEBSP : public NALUnit
-{
-    uint32_t m_packetSize;
-    uint8_t *m_nalUnitData;
-
-    /**
-     * convert the OutputNALUnit #nalu# into EBSP format by writing out
-     * the NALUnit header, then the rbsp_bytes including any
-     * emulation_prevention_three_byte symbols.
-     */
-    void init(const struct OutputNALUnit& nalu);
-};
-}
-
-#endif // ifndef X265_NAL_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/SEI.h Deleted

@@ -1,293 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#ifndef X265_SEI_H
-#define X265_SEI_H
-
-#include "common.h"
-
-namespace x265 {
-// private namespace
-
-//! \ingroup TLibCommon
-//! \{
-class TComSPS;
-
-/**
- * Abstract class representing an SEI message with lightweight RTTI.
- */
-class SEI
-{
-public:
-
-    enum PayloadType
-    {
-        BUFFERING_PERIOD                     = 0,
-        PICTURE_TIMING                       = 1,
-        PAN_SCAN_RECT                        = 2,
-        FILLER_PAYLOAD                       = 3,
-        USER_DATA_REGISTERED_ITU_T_T35       = 4,
-        USER_DATA_UNREGISTERED               = 5,
-        RECOVERY_POINT                       = 6,
-        SCENE_INFO                           = 9,
-        FULL_FRAME_SNAPSHOT                  = 15,
-        PROGRESSIVE_REFINEMENT_SEGMENT_START = 16,
-        PROGRESSIVE_REFINEMENT_SEGMENT_END   = 17,
-        FILM_GRAIN_CHARACTERISTICS           = 19,
-        POST_FILTER_HINT                     = 22,
-        TONE_MAPPING_INFO                    = 23,
-        FRAME_PACKING                        = 45,
-        DISPLAY_ORIENTATION                  = 47,
-        SOP_DESCRIPTION                      = 128,
-        ACTIVE_PARAMETER_SETS                = 129,
-        DECODING_UNIT_INFO                   = 130,
-        TEMPORAL_LEVEL0_INDEX                = 131,
-        DECODED_PICTURE_HASH                 = 132,
-        SCALABLE_NESTING                     = 133,
-        REGION_REFRESH_INFO                  = 134,
-    };
-
-    SEI() {}
-
-    virtual ~SEI() {}
-
-    virtual PayloadType payloadType() const = 0;
-};
-
-class SEIuserDataUnregistered : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return USER_DATA_UNREGISTERED; }
-
-    SEIuserDataUnregistered()
-        : userData(0)
-    {}
-
-    virtual ~SEIuserDataUnregistered()
-    {
-        delete userData;
-    }
-
-    UChar uuid_iso_iec_11578[16];
-    uint32_t userDataLength;
-    UChar *userData;
-};
-
-class SEIDecodedPictureHash : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return DECODED_PICTURE_HASH; }
-
-    SEIDecodedPictureHash() {}
-
-    virtual ~SEIDecodedPictureHash() {}
-
-    enum Method
-    {
-        MD5,
-        CRC,
-        CHECKSUM,
-        RESERVED,
-    } method;
-
-    UChar digest[3][16];
-};
-
-class SEIActiveParameterSets : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return ACTIVE_PARAMETER_SETS; }
-
-    SEIActiveParameterSets()
-        : activeVPSId(0)
-        , m_fullRandomAccessFlag(false)
-        , m_noParamSetUpdateFlag(false)
-        , numSpsIdsMinus1(0)
-    {}
-
-    virtual ~SEIActiveParameterSets() {}
-
-    int activeVPSId;
-    bool m_fullRandomAccessFlag;
-    bool m_noParamSetUpdateFlag;
-    int numSpsIdsMinus1;
-    int activeSeqParamSetId;
-};
-
-class SEIBufferingPeriod : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return BUFFERING_PERIOD; }
-
-    SEIBufferingPeriod()
-        : m_bpSeqParameterSetId(0)
-        , m_rapCpbParamsPresentFlag(false)
-        , m_cpbDelayOffset(0)
-        , m_dpbDelayOffset(0)
-    {
-        ::memset(m_initialCpbRemovalDelay, 0, sizeof(m_initialCpbRemovalDelay));
-        ::memset(m_initialCpbRemovalDelayOffset, 0, sizeof(m_initialCpbRemovalDelayOffset));
-        ::memset(m_initialAltCpbRemovalDelay, 0, sizeof(m_initialAltCpbRemovalDelay));
-        ::memset(m_initialAltCpbRemovalDelayOffset, 0, sizeof(m_initialAltCpbRemovalDelayOffset));
-    }
-
-    virtual ~SEIBufferingPeriod() {}
-
-    uint32_t m_bpSeqParameterSetId;
-    bool m_rapCpbParamsPresentFlag;
-    bool m_cpbDelayOffset;
-    bool m_dpbDelayOffset;
-    uint32_t m_initialCpbRemovalDelay[MAX_CPB_CNT][2];
-    uint32_t m_initialCpbRemovalDelayOffset[MAX_CPB_CNT][2];
-    uint32_t m_initialAltCpbRemovalDelay[MAX_CPB_CNT][2];
-    uint32_t m_initialAltCpbRemovalDelayOffset[MAX_CPB_CNT][2];
-    bool m_concatenationFlag;
-    uint32_t m_auCpbRemovalDelayDelta;
-};
-
-class SEIPictureTiming : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return PICTURE_TIMING; }
-
-    SEIPictureTiming()
-        : m_picStruct(0)
-        , m_sourceScanType(0)
-        , m_duplicateFlag(false)
-        , m_picDpbOutputDuDelay(0)
-        , m_numNalusInDuMinus1(NULL)
-        , m_duCpbRemovalDelayMinus1(NULL)
-    {}
-
-    virtual ~SEIPictureTiming()
-    {
-        delete m_numNalusInDuMinus1;
-        delete m_duCpbRemovalDelayMinus1;
-    }
-
-    uint32_t  m_picStruct;
-    uint32_t  m_sourceScanType;
-    bool  m_duplicateFlag;
-
-    uint32_t  m_auCpbRemovalDelay;
-    uint32_t  m_picDpbOutputDelay;
-    uint32_t  m_picDpbOutputDuDelay;
-    uint32_t  m_numDecodingUnitsMinus1;
-    bool  m_duCommonCpbRemovalDelayFlag;
-    uint32_t  m_duCommonCpbRemovalDelayMinus1;
-    uint32_t* m_numNalusInDuMinus1;
-    uint32_t* m_duCpbRemovalDelayMinus1;
-};
-
-class SEIDecodingUnitInfo : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return DECODING_UNIT_INFO; }
-
-    SEIDecodingUnitInfo()
-        : m_decodingUnitIdx(0)
-        , m_duSptCpbRemovalDelay(0)
-        , m_dpbOutputDuDelayPresentFlag(false)
-        , m_picSptDpbOutputDuDelay(0)
-    {}
-
-    virtual ~SEIDecodingUnitInfo() {}
-
-    int m_decodingUnitIdx;
-    int m_duSptCpbRemovalDelay;
-    bool m_dpbOutputDuDelayPresentFlag;
-    int m_picSptDpbOutputDuDelay;
-};
-
-class SEIRecoveryPoint : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return RECOVERY_POINT; }
-
-    SEIRecoveryPoint() {}
-
-    virtual ~SEIRecoveryPoint() {}
-
-    int  m_recoveryPocCnt;
-    bool m_exactMatchingFlag;
-    bool m_brokenLinkFlag;
-};
-
-class SEIDisplayOrientation : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return DISPLAY_ORIENTATION; }
-
-    SEIDisplayOrientation()
-        : cancelFlag(true)
-        , persistenceFlag(0)
-        , extensionFlag(false)
-    {}
-
-    virtual ~SEIDisplayOrientation() {}
-
-    bool cancelFlag;
-    bool horFlip;
-    bool verFlip;
-
-    uint32_t anticlockwiseRotation;
-    bool persistenceFlag;
-    bool extensionFlag;
-};
-
-class SEIGradualDecodingRefreshInfo : public SEI
-{
-public:
-
-    PayloadType payloadType() const { return REGION_REFRESH_INFO; }
-
-    SEIGradualDecodingRefreshInfo()
-        : m_gdrForegroundFlag(0)
-    {}
-
-    virtual ~SEIGradualDecodingRefreshInfo() {}
-
-    bool m_gdrForegroundFlag;
-};
-}
-//! \}
-
-#endif // ifndef X265_SEI_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComBitCounter.h Deleted

@@ -1,77 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComBitCounter.h
-    \brief    Class for counting bits (header)
-*/
-
-#ifndef X265_COMBITCOUNTER_H
-#define X265_COMBITCOUNTER_H
-
-#include "TComBitStream.h"
-
-namespace x265 {
-// private namespace
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// class for counting bits
-class TComBitCounter : public TComBitIf
-{
-protected:
-
-    uint32_t  m_uiBitCounter;
-
-public:
-
-    TComBitCounter()            {}
-
-    virtual ~TComBitCounter()   {}
-
-    void        write(uint32_t /*uiBits*/, uint32_t uiNumberOfBits)  { m_uiBitCounter += uiNumberOfBits; }
-
-    void        writeByte(uint32_t /*val*/)                      { m_uiBitCounter += 8; }
-
-    void        resetBits()                                  { m_uiBitCounter = 0; }
-
-    uint32_t getNumberOfWrittenBits() const { return m_uiBitCounter; }
-};
-}
-//! \}
-
-#endif // ifndef X265_COMBITCOUNTER_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComBitStream.cpp Deleted

@@ -1,221 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComBitStream.cpp
-    \brief    class for handling bitstream
-*/
-
-#include "TComBitStream.h"
-#include "common.h"
-
-using namespace x265;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
-TComOutputBitstream::TComOutputBitstream()
-{
-    m_fifo = X265_MALLOC(uint8_t, MIN_FIFO_SIZE);
-    m_buffsize = MIN_FIFO_SIZE;
-    clear();
-}
-
-TComOutputBitstream::~TComOutputBitstream()
-{
-    X265_FREE(m_fifo);
-}
-
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
-
-char* TComOutputBitstream::getByteStream() const
-{
-    return (char*)m_fifo;
-}
-
-uint32_t TComOutputBitstream::getByteStreamLength() const
-{
-    return m_fsize;
-}
-
-void TComOutputBitstream::clear()
-{
-    m_held_bits = 0;
-    m_num_held_bits = 0;
-    m_fsize = 0;
-}
-
-void TComOutputBitstream::write(uint32_t bits, uint32_t numBits)
-{
-    assert(numBits <= 32);
-    assert(numBits == 32 || (bits & (~0 << numBits)) == 0);
-
-    /* any modulo 8 remainder of num_total_bits cannot be written this time,
-     * and will be held until next time. */
-    uint32_t num_total_bits = numBits + m_num_held_bits;
-    uint32_t next_num_held_bits = num_total_bits % 8;
-
-    /* form a byte aligned word (write_bits), by concatenating any held bits
-     * with the new bits, discarding the bits that will form the next_held_bits.
-     * eg: H = held bits, V = n new bits        /---- next_held_bits
-     * len(H)=7, len(V)=1: ... ---- HHHH HHHV . 0000 0000, next_num_held_bits=0
-     * len(H)=7, len(V)=2: ... ---- HHHH HHHV . V000 0000, next_num_held_bits=1
-     * if total_bits < 8, the value of v_ is not used */
-    UChar next_held_bits = bits << (8 - next_num_held_bits);
-
-    if (!(num_total_bits >> 3))
-    {
-        /* insufficient bits accumulated to write out, append new_held_bits to
-         * current held_bits */
-        /* NB, this requires that v only contains 0 in bit positions {31..n} */
-        m_held_bits |= next_held_bits;
-        m_num_held_bits = next_num_held_bits;
-        return;
-    }
-
-    /* topword serves to justify held_bits to align with the msb of uiBits */
-    uint32_t topword = (numBits - next_num_held_bits) & ~((1 << 3) - 1);
-    uint32_t write_bits = (m_held_bits << topword) | (bits >> next_num_held_bits);
-
-    switch (num_total_bits >> 3)
-    {
-    case 4: push_back(write_bits >> 24);
-    case 3: push_back(write_bits >> 16);
-    case 2: push_back(write_bits >> 8);
-    case 1: push_back(write_bits);
-    }
-
-    m_held_bits = next_held_bits;
-    m_num_held_bits = next_num_held_bits;
-}
-
-void TComOutputBitstream::writeByte(uint32_t val)
-{
-    // NOTE: we are here only in Cabac
-    assert(m_num_held_bits == 0);
-
-    push_back(val);
-}
-
-void TComOutputBitstream::writeAlignOne()
-{
-    uint32_t numBits = getNumBitsUntilByteAligned();
-
-    write((1 << numBits) - 1, numBits);
-}
-
-void TComOutputBitstream::writeAlignZero()
-{
-    if (!m_num_held_bits)
-        return;
-
-    push_back(m_held_bits);
-    m_held_bits = 0;
-    m_num_held_bits = 0;
-}
-
-/**
- * add substream to the end of the current bitstream
- */
-void TComOutputBitstream::addSubstream(TComOutputBitstream* substream)
-{
-    uint32_t numBits = substream->getNumberOfWrittenBits();
-
-    const uint8_t* rbsp = substream->getFIFO();
-
-    for (uint32_t count = 0; count < substream->m_fsize; count++)
-    {
-        write(rbsp[count], 8);
-    }
-
-    if (numBits & 0x7)
-    {
-        write(substream->getHeldBits() >> (8 - (numBits & 0x7)), numBits & 0x7);
-    }
-}
-
-void TComOutputBitstream::writeByteAlignment()
-{
-    write(1, 1);
-    writeAlignZero();
-}
-
-int TComOutputBitstream::countStartCodeEmulations()
-{
-    int numStartCodes = 0;
-    uint8_t *rbsp = getFIFO();
-    uint32_t fsize = getByteStreamLength();
-
-    for (uint32_t i = 0; i + 2 < fsize; i++)
-    {
-        if (!rbsp[i] && !rbsp[i + 1] && rbsp[i + 2] <= 3)
-        {
-            numStartCodes++;
-            i++;
-        }
-    }
-
-    return numStartCodes;
-}
-
-void TComOutputBitstream::push_back(uint8_t val)
-{
-    if (!m_fifo)
-        return;
-
-    if (m_fsize >= m_buffsize)
-    {
-        /** reallocate buffer with doubled size */
-        uint8_t *temp = X265_MALLOC(uint8_t, m_buffsize * 2);
-        if (temp)
-        {
-            ::memcpy(temp, m_fifo, m_fsize);
-            X265_FREE(m_fifo);
-            m_fifo = temp;
-            m_buffsize *= 2;
-        }
-        else
-        {
-            x265_log(NULL, X265_LOG_ERROR, "Unable to realloc bitstream buffer");
-            return;
-        }
-    }
-    m_fifo[m_fsize++] = val;
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComBitStream.h Deleted

@@ -1,165 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComBitStream.h
-    \brief    class for handling bitstream (header)
-*/
-
-#ifndef X265_COMBITSTREAM_H
-#define X265_COMBITSTREAM_H
-
-#include "CommonDef.h"
-
-//! \ingroup TLibCommon
-//! \{
-
-namespace x265 {
-// private namespace
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// pure virtual class for basic bit handling
-class TComBitIf
-{
-public:
-
-    virtual void        writeAlignOne() {}
-
-    virtual void        writeAlignZero() {}
-
-    virtual void        write(uint32_t uiBits, uint32_t uiNumberOfBits)  = 0;
-    virtual void        writeByte(uint32_t val)                      = 0;
-    virtual void        resetBits()                              = 0;
-    virtual uint32_t getNumberOfWrittenBits() const = 0;
-    virtual ~TComBitIf() {}
-};
-
-/**
- * Model of a writable bitstream that accumulates bits to produce a
- * bytestream.
- */
-class TComOutputBitstream : public TComBitIf
-{
-    /**
-     * FIFO for storage of bytes.  Use:
-     *  - fifo.push_back(x) to append words
-     *  - fifo.clear() to empty the FIFO
-     *  - &fifo.front() to get a pointer to the data array.
-     *    NB, this pointer is only valid until the next push_back()/clear()
-     */
-    uint8_t *m_fifo;
-    uint32_t m_fsize;
-    uint32_t m_buffsize;
-
-    uint32_t m_num_held_bits; /// number of bits not flushed to bytestream.
-    UChar m_held_bits; /// the bits held and not flushed to bytestream.
-    /// this value is always msb-aligned, bigendian.
-
-public:
-
-    // create / destroy
-    TComOutputBitstream();
-    ~TComOutputBitstream();
-
-    // interface for encoding
-
-    /**
-     * append uiNumberOfBits least significant bits of uiBits to
-     * the current bitstream
-     */
-    void        write(uint32_t uiBits, uint32_t uiNumberOfBits);
-    void        writeByte(uint32_t val);
-
-    /** insert one bits until the bitstream is byte-aligned */
-    void        writeAlignOne();
-
-    /** insert zero bits until the bitstream is byte-aligned */
-    void        writeAlignZero();
-
-    /** this function should never be called */
-    void resetBits() { assert(0); }
-
-    // utility functions
-
-    /**
-     * Return a pointer to the start of the byte-stream buffer.
-     * Pointer is valid until the next write/flush/reset call.
-     * NB, data is arranged such that subsequent bytes in the
-     * bytestream are stored in ascending addresses.
-     */
-    char* getByteStream() const;
-
-    /**
-     * Return the number of valid bytes available from  getByteStream()
-     */
-    uint32_t getByteStreamLength() const;
-
-    /**
-     * Reset all internal state.
-     */
-    void clear();
-
-    /**
-     * returns the number of bits that need to be written to
-     * achieve byte alignment.
-     */
-    int getNumBitsUntilByteAligned() { return (8 - m_num_held_bits) & 0x7; }
-
-    /**
-     * Return the number of bits that have been written since the last clear()
-     */
-    uint32_t getNumberOfWrittenBits() const { return m_fsize * 8 + m_num_held_bits; }
-
-    /**
-     * Return a reference to the internal fifo
-     */
-    uint8_t* getFIFO() { return m_fifo; }
-
-    UChar getHeldBits()          { return m_held_bits; }
-
-    /** Return a reference to the internal fifo */
-    uint8_t* getFIFO() const { return m_fifo; }
-
-    void          addSubstream(TComOutputBitstream* pcSubstream);
-    void writeByteAlignment();
-
-    //! returns the number of start code emulations contained in the current buffer
-    int countStartCodeEmulations();
-    void push_back(uint8_t val);
-};
-}
-//! \}
-
-#endif // ifndef X265_COMBITSTREAM_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComLoopFilter.cpp Deleted

@@ -1,879 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComLoopFilter.cpp
-    \brief    deblocking filter
-*/
-
-#include "TComLoopFilter.h"
-#include "TComSlice.h"
-#include "mv.h"
-
-using namespace x265;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-#define   QpUV(iQpY, chFmt)  (((iQpY) < 0) ? (iQpY) : (((iQpY) > 57) ? ((iQpY) - 6) : g_chromaScale[chFmt][(iQpY)]))
-#define DEFAULT_INTRA_TC_OFFSET 2 ///< Default intra TC offset
-
-// ====================================================================================================================
-// Tables
-// ====================================================================================================================
-
-const UChar TComLoopFilter::sm_tcTable[54] =
-{
-    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 5, 5, 6, 6, 7, 8, 9, 10, 11, 13, 14, 16, 18, 20, 22, 24
-};
-
-const UChar TComLoopFilter::sm_betaTable[52] =
-{
-    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62, 64
-};
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
-TComLoopFilter::TComLoopFilter()
-    : m_numPartitions(0)
-    , m_bLFCrossTileBoundary(true)
-{
-    for (uint32_t dir = 0; dir < 2; dir++)
-    {
-        m_blockingStrength[dir] = NULL;
-        m_bEdgeFilter[dir] = NULL;
-    }
-}
-
-TComLoopFilter::~TComLoopFilter()
-{}
-
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
-void TComLoopFilter::setCfg(bool bLFCrossTileBoundary)
-{
-    m_bLFCrossTileBoundary = bLFCrossTileBoundary;
-}
-
-void TComLoopFilter::create(uint32_t maxCuDepth)
-{
-    destroy();
-    m_numPartitions = 1 << (maxCuDepth << 1);
-    for (uint32_t dir = 0; dir < 2; dir++)
-    {
-        m_blockingStrength[dir] = new UChar[m_numPartitions];
-        m_bEdgeFilter[dir] = new bool[m_numPartitions];
-    }
-}
-
-void TComLoopFilter::destroy()
-{
-    for (uint32_t dir = 0; dir < 2; dir++)
-    {
-        delete [] m_blockingStrength[dir];
-        m_blockingStrength[dir] = NULL;
-        delete [] m_bEdgeFilter[dir];
-        m_bEdgeFilter[dir] = NULL;
-    }
-}
-
-/**
- - call deblocking function for every CU
- .
- \param  pic   picture class (TComPic) pointer
- */
-void TComLoopFilter::loopFilterPic(TComPic* pic)
-{
-    // TODO: Min, thread parallelism later
-    // Horizontal filtering
-    for (uint32_t cuAddr = 0; cuAddr < pic->getNumCUsInFrame(); cuAddr++)
-    {
-        TComDataCU* cu = pic->getCU(cuAddr);
-
-        ::memset(m_blockingStrength[EDGE_VER], 0, sizeof(UChar) * m_numPartitions);
-        ::memset(m_bEdgeFilter[EDGE_VER], 0, sizeof(bool) * m_numPartitions);
-
-        // CU-based deblocking
-        xDeblockCU(cu, 0, 0, EDGE_VER);
-
-        // Vertical filtering
-        // NOTE: delay one CU to avoid conflict between V and H
-        if (cuAddr > 0)
-        {
-            cu = pic->getCU(cuAddr - 1);
-            ::memset(m_blockingStrength[EDGE_HOR], 0, sizeof(UChar) * m_numPartitions);
-            ::memset(m_bEdgeFilter[EDGE_HOR], 0, sizeof(bool) * m_numPartitions);
-
-            xDeblockCU(cu, 0, 0, EDGE_HOR);
-        }
-    }
-
-    // Last H-Filter
-    {
-        TComDataCU* cu = pic->getCU(pic->getNumCUsInFrame() - 1);
-        ::memset(m_blockingStrength[EDGE_HOR], 0, sizeof(UChar) * m_numPartitions);
-        ::memset(m_bEdgeFilter[EDGE_HOR], 0, sizeof(bool) * m_numPartitions);
-
-        xDeblockCU(cu, 0, 0, EDGE_HOR);
-    }
-}
-
-void TComLoopFilter::loopFilterCU(TComDataCU* cu, int dir)
-{
-    ::memset(m_blockingStrength[dir], 0, sizeof(UChar) * m_numPartitions);
-    ::memset(m_bEdgeFilter[dir], 0, sizeof(bool) * m_numPartitions);
-
-    // CU-based deblocking
-    xDeblockCU(cu, 0, 0, dir);
-}
-
-// ====================================================================================================================
-// Protected member functions
-// ====================================================================================================================
-
-/**
- - Deblocking filter process in CU-based (the same function as conventional's)
- .
- \param Edge          the direction of the edge in block boundary (horizonta/vertical), which is added newly
-*/
-void TComLoopFilter::xDeblockCU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int edge)
-{
-    if (cu->getPic() == 0 || cu->getPartitionSize(absZOrderIdx) == SIZE_NONE)
-    {
-        return;
-    }
-    TComPic* pic     = cu->getPic();
-    uint32_t curNumParts = pic->getNumPartInCU() >> (depth << 1);
-    uint32_t qNumParts   = curNumParts >> 2;
-
-    if (cu->getDepth(absZOrderIdx) > depth)
-    {
-        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
-        {
-            uint32_t lpelx   = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absZOrderIdx]];
-            uint32_t tpely   = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absZOrderIdx]];
-            if ((lpelx < cu->getSlice()->getSPS()->getPicWidthInLumaSamples()) && (tpely < cu->getSlice()->getSPS()->getPicHeightInLumaSamples()))
-            {
-                xDeblockCU(cu, absZOrderIdx, depth + 1, edge);
-            }
-        }
-
-        return;
-    }
-
-    xSetLoopfilterParam(cu, absZOrderIdx);
-
-    xSetEdgefilterTU(cu, absZOrderIdx, absZOrderIdx, depth);
-    xSetEdgefilterPU(cu, absZOrderIdx);
-
-    int dir = edge;
-    for (uint32_t partIdx = absZOrderIdx; partIdx < absZOrderIdx + curNumParts; partIdx++)
-    {
-        uint32_t bsCheck;
-        if ((g_maxCUSize >> g_maxCUDepth) == 4)
-        {
-            bsCheck = (dir == EDGE_VER && partIdx % 2 == 0) || (dir == EDGE_HOR && (partIdx - ((partIdx >> 2) << 2)) / 2 == 0);
-        }
-        else
-        {
-            bsCheck = 1;
-        }
-
-        if (m_bEdgeFilter[dir][partIdx] && bsCheck)
-        {
-            xGetBoundaryStrengthSingle(cu, dir, partIdx);
-        }
-    }
-
-    uint32_t pelsInPart = g_maxCUSize >> g_maxCUDepth;
-    uint32_t partIdxIncr = DEBLOCK_SMALLEST_BLOCK / pelsInPart ? DEBLOCK_SMALLEST_BLOCK / pelsInPart : 1;
-
-    uint32_t sizeInPU = pic->getNumPartInWidth() >> (depth);
-    const bool bAlwaysDoChroma = (cu->getChromaFormat() == CHROMA_444);
-    for (uint32_t e = 0; e < sizeInPU; e += partIdxIncr)
-    {
-        xEdgeFilterLuma(cu, absZOrderIdx, depth, dir, e);
-        if (bAlwaysDoChroma || (pelsInPart > DEBLOCK_SMALLEST_BLOCK) || (e % ((DEBLOCK_SMALLEST_BLOCK << 1) / pelsInPart)) == 0)
-        {
-            xEdgeFilterChroma(cu, absZOrderIdx, depth, dir, e);
-        }
-    }
-}
-
-void TComLoopFilter::xSetEdgefilterMultiple(TComDataCU* cu, uint32_t scanIdx, uint32_t depth, int dir, int edgeIdx, bool bValue, uint32_t widthInBaseUnits, uint32_t heightInBaseUnits)
-{
-    if (widthInBaseUnits == 0)
-    {
-        widthInBaseUnits  = cu->getPic()->getNumPartInWidth() >> depth;
-    }
-    if (heightInBaseUnits == 0)
-    {
-        heightInBaseUnits = cu->getPic()->getNumPartInHeight() >> depth;
-    }
-    const uint32_t numElem = dir == 0 ? heightInBaseUnits : widthInBaseUnits;
-    assert(numElem > 0);
-    assert(widthInBaseUnits > 0);
-    assert(heightInBaseUnits > 0);
-    for (uint32_t i = 0; i < numElem; i++)
-    {
-        const uint32_t bsidx = xCalcBsIdx(cu, scanIdx, dir, edgeIdx, i);
-        m_bEdgeFilter[dir][bsidx] = bValue;
-        if (edgeIdx == 0)
-        {
-            m_blockingStrength[dir][bsidx] = bValue;
-        }
-    }
-}
-
-void TComLoopFilter::xSetEdgefilterTU(TComDataCU* cu, uint32_t absTUPartIdx, uint32_t absZOrderIdx, uint32_t depth)
-{
-    if (cu->getTransformIdx(absZOrderIdx) + cu->getDepth(absZOrderIdx) > depth)
-    {
-        const uint32_t curNumParts = cu->getPic()->getNumPartInCU() >> (depth << 1);
-        const uint32_t qNumParts   = curNumParts >> 2;
-        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
-        {
-            uint32_t nsAddr = absZOrderIdx;
-            xSetEdgefilterTU(cu, nsAddr, absZOrderIdx, depth + 1);
-        }
-
-        return;
-    }
-
-    int trWidth  = cu->getCUSize(absZOrderIdx) >> cu->getTransformIdx(absZOrderIdx);
-    int trHeight = cu->getCUSize(absZOrderIdx) >> cu->getTransformIdx(absZOrderIdx);
-
-    uint32_t widthInBaseUnits  = trWidth / (g_maxCUSize >> g_maxCUDepth);
-    uint32_t heightInBaseUnits = trHeight / (g_maxCUSize >> g_maxCUDepth);
-
-    xSetEdgefilterMultiple(cu, absTUPartIdx, depth, EDGE_VER, 0, true, widthInBaseUnits, heightInBaseUnits);
-    xSetEdgefilterMultiple(cu, absTUPartIdx, depth, EDGE_HOR, 0, true, widthInBaseUnits, heightInBaseUnits);
-}
-
-void TComLoopFilter::xSetEdgefilterPU(TComDataCU* cu, uint32_t absZOrderIdx)
-{
-    const uint32_t depth = cu->getDepth(absZOrderIdx);
-    const uint32_t widthInBaseUnits  = cu->getPic()->getNumPartInWidth() >> depth;
-    const uint32_t heightInBaseUnits = cu->getPic()->getNumPartInHeight() >> depth;
-    const uint32_t hWidthInBaseUnits  = widthInBaseUnits  >> 1;
-    const uint32_t hHeightInBaseUnits = heightInBaseUnits >> 1;
-    const uint32_t qWidthInBaseUnits  = widthInBaseUnits  >> 2;
-    const uint32_t qHeightInBaseUnits = heightInBaseUnits >> 2;
-
-    xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_VER, 0, m_lfcuParam.bLeftEdge);
-    xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_HOR, 0, m_lfcuParam.bTopEdge);
-
-    switch (cu->getPartitionSize(absZOrderIdx))
-    {
-    case SIZE_2Nx2N:
-    {
-        break;
-    }
-    case SIZE_2NxN:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_HOR, hHeightInBaseUnits, true);
-        break;
-    }
-    case SIZE_Nx2N:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_VER, hWidthInBaseUnits, true);
-        break;
-    }
-    case SIZE_NxN:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_VER, hWidthInBaseUnits, true);
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_HOR, hHeightInBaseUnits, true);
-        break;
-    }
-    case SIZE_2NxnU:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_HOR, qHeightInBaseUnits, true);
-        break;
-    }
-    case SIZE_2NxnD:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_HOR, heightInBaseUnits - qHeightInBaseUnits, true);
-        break;
-    }
-    case SIZE_nLx2N:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_VER, qWidthInBaseUnits, true);
-        break;
-    }
-    case SIZE_nRx2N:
-    {
-        xSetEdgefilterMultiple(cu, absZOrderIdx, depth, EDGE_VER, widthInBaseUnits - qWidthInBaseUnits, true);
-        break;
-    }
-    default:
-    {
-        break;
-    }
-    }
-}
-
-void TComLoopFilter::xSetLoopfilterParam(TComDataCU* cu, uint32_t absZOrderIdx)
-{
-    uint32_t x = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absZOrderIdx]];
-    uint32_t y = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absZOrderIdx]];
-
-    TComDataCU* tempCU;
-    uint32_t        tempPartIdx;
-
-    // We can't here when DeblockingDisable flag is true
-    assert(!cu->getSlice()->getDeblockingFilterDisable());
-
-    if (x == 0)
-    {
-        m_lfcuParam.bLeftEdge = false;
-    }
-    else
-    {
-        tempCU = cu->getPULeft(tempPartIdx, absZOrderIdx, !true, !m_bLFCrossTileBoundary);
-        if (tempCU)
-        {
-            m_lfcuParam.bLeftEdge = true;
-        }
-        else
-        {
-            m_lfcuParam.bLeftEdge = false;
-        }
-    }
-
-    if (y == 0)
-    {
-        m_lfcuParam.bTopEdge = false;
-    }
-    else
-    {
-        tempCU = cu->getPUAbove(tempPartIdx, absZOrderIdx, !true, false, !m_bLFCrossTileBoundary);
-        if (tempCU)
-        {
-            m_lfcuParam.bTopEdge = true;
-        }
-        else
-        {
-            m_lfcuParam.bTopEdge = false;
-        }
-    }
-}
-
-void TComLoopFilter::xGetBoundaryStrengthSingle(TComDataCU* cu, int dir, uint32_t absPartIdx)
-{
-    TComSlice* const slice = cu->getSlice();
-
-    const uint32_t partQ = absPartIdx;
-    TComDataCU* const cuQ = cu;
-
-    uint32_t partP;
-    TComDataCU* cuP;
-    uint32_t bs = 0;
-
-    //-- Calculate Block Index
-    if (dir == EDGE_VER)
-    {
-        cuP = cuQ->getPULeft(partP, partQ, !true, !m_bLFCrossTileBoundary);
-    }
-    else // (dir == EDGE_HOR)
-    {
-        cuP = cuQ->getPUAbove(partP, partQ, !true, false, !m_bLFCrossTileBoundary);
-    }
-
-    //-- Set BS for Intra MB : BS = 4 or 3
-    if (cuP->isIntra(partP) || cuQ->isIntra(partQ))
-    {
-        bs = 2;
-    }
-
-    //-- Set BS for not Intra MB : BS = 2 or 1 or 0
-    if (!cuP->isIntra(partP) && !cuQ->isIntra(partQ))
-    {
-        uint32_t nsPartQ = partQ;
-        uint32_t nsPartP = partP;
-
-        if (m_blockingStrength[dir][absPartIdx] && (cuQ->getCbf(nsPartQ, TEXT_LUMA, cuQ->getTransformIdx(nsPartQ)) != 0 || cuP->getCbf(nsPartP, TEXT_LUMA, cuP->getTransformIdx(nsPartP)) != 0))
-        {
-            bs = 1;
-        }
-        else
-        {
-            if (dir == EDGE_HOR)
-            {
-                cuP = cuQ->getPUAbove(partP, partQ, !true, false, !m_bLFCrossTileBoundary);
-            }
-            if (slice->isInterB() || cuP->getSlice()->isInterB())
-            {
-                int refIdx;
-                TComPic *refP0, *refP1, *refQ0, *refQ1;
-                refIdx = cuP->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partP);
-                refP0 = (refIdx < 0) ? NULL : cuP->getSlice()->getRefPic(REF_PIC_LIST_0, refIdx);
-                refIdx = cuP->getCUMvField(REF_PIC_LIST_1)->getRefIdx(partP);
-                refP1 = (refIdx < 0) ? NULL : cuP->getSlice()->getRefPic(REF_PIC_LIST_1, refIdx);
-                refIdx = cuQ->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partQ);
-                refQ0 = (refIdx < 0) ? NULL : slice->getRefPic(REF_PIC_LIST_0, refIdx);
-                refIdx = cuQ->getCUMvField(REF_PIC_LIST_1)->getRefIdx(partQ);
-                refQ1 = (refIdx < 0) ? NULL : slice->getRefPic(REF_PIC_LIST_1, refIdx);
-
-                MV mvp0 = cuP->getCUMvField(REF_PIC_LIST_0)->getMv(partP);
-                MV mvp1 = cuP->getCUMvField(REF_PIC_LIST_1)->getMv(partP);
-                MV mvq0 = cuQ->getCUMvField(REF_PIC_LIST_0)->getMv(partQ);
-                MV mvq1 = cuQ->getCUMvField(REF_PIC_LIST_1)->getMv(partQ);
-
-                if (refP0 == NULL) mvp0 = 0;
-                if (refP1 == NULL) mvp1 = 0;
-                if (refQ0 == NULL) mvq0 = 0;
-                if (refQ1 == NULL) mvq1 = 0;
-
-                if (((refP0 == refQ0) && (refP1 == refQ1)) || ((refP0 == refQ1) && (refP1 == refQ0)))
-                {
-                    if (refP0 != refP1) // Different L0 & L1
-                    {
-                        if (refP0 == refQ0)
-                        {
-                            bs  = ((abs(mvq0.x - mvp0.x) >= 4) ||
-                                   (abs(mvq0.y - mvp0.y) >= 4) ||
-                                   (abs(mvq1.x - mvp1.x) >= 4) ||
-                                   (abs(mvq1.y - mvp1.y) >= 4)) ? 1 : 0;
-                        }
-                        else
-                        {
-                            bs  = ((abs(mvq1.x - mvp0.x) >= 4) ||
-                                   (abs(mvq1.y - mvp0.y) >= 4) ||
-                                   (abs(mvq0.x - mvp1.x) >= 4) ||
-                                   (abs(mvq0.y - mvp1.y) >= 4)) ? 1 : 0;
-                        }
-                    }
-                    else // Same L0 & L1
-                    {
-                        bs  = ((abs(mvq0.x - mvp0.x) >= 4) ||
-                               (abs(mvq0.y - mvp0.y) >= 4) ||
-                               (abs(mvq1.x - mvp1.x) >= 4) ||
-                               (abs(mvq1.y - mvp1.y) >= 4)) &&
-                            ((abs(mvq1.x - mvp0.x) >= 4) ||
-                             (abs(mvq1.y - mvp0.y) >= 4) ||
-                             (abs(mvq0.x - mvp1.x) >= 4) ||
-                             (abs(mvq0.y - mvp1.y) >= 4)) ? 1 : 0;
-                    }
-                }
-                else // for all different Ref_Idx
-                {
-                    bs = 1;
-                }
-            }
-            else // slice->isInterP()
-            {
-                int refIdx;
-                TComPic *refp0, *refq0;
-                refIdx = cuP->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partP);
-                refp0 = (refIdx < 0) ? NULL : cuP->getSlice()->getRefPic(REF_PIC_LIST_0, refIdx);
-                refIdx = cuQ->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partQ);
-                refq0 = (refIdx < 0) ? NULL : slice->getRefPic(REF_PIC_LIST_0, refIdx);
-                MV mvp0 = cuP->getCUMvField(REF_PIC_LIST_0)->getMv(partP);
-                MV mvq0 = cuQ->getCUMvField(REF_PIC_LIST_0)->getMv(partQ);
-
-                if (refp0 == NULL) mvp0 = 0;
-                if (refq0 == NULL) mvq0 = 0;
-
-                bs = ((refp0 != refq0) ||
-                      (abs(mvq0.x - mvp0.x) >= 4) ||
-                      (abs(mvq0.y - mvp0.y) >= 4)) ? 1 : 0;
-            }
-        } // enf of "if( one of BCBP == 0 )"
-    } // enf of "if( not Intra )"
-
-    m_blockingStrength[dir][absPartIdx] = bs;
-}
-
-void TComLoopFilter::xEdgeFilterLuma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int dir, int edge)
-{
-    TComPicYuv* reconYuv = cu->getPic()->getPicYuvRec();
-    Pel* src = reconYuv->getLumaAddr(cu->getAddr(), absZOrderIdx);
-    Pel* tmpsrc = src;
-
-    int stride = reconYuv->getStride();
-    int qp = 0;
-    int qpP = 0;
-    int qpQ = 0;
-    uint32_t numParts = cu->getPic()->getNumPartInWidth() >> depth;
-
-    uint32_t pelsInPart = g_maxCUSize >> g_maxCUDepth;
-    uint32_t bsAbsIdx = 0, bs = 0;
-    int  offset, srcStep;
-
-    bool  bPCMFilter = (cu->getSlice()->getSPS()->getUsePCM() && cu->getSlice()->getSPS()->getPCMFilterDisableFlag()) ? true : false;
-    bool  bPartPNoFilter = false;
-    bool  bPartQNoFilter = false;
-    uint32_t  partP = 0;
-    uint32_t  partQ = 0;
-    TComDataCU* cuP = cu;
-    TComDataCU* cuQ = cu;
-    int  betaOffsetDiv2 = cuQ->getSlice()->getDeblockingFilterBetaOffsetDiv2();
-    int  tcOffsetDiv2 = cuQ->getSlice()->getDeblockingFilterTcOffsetDiv2();
-
-    if (dir == EDGE_VER)
-    {
-        offset = 1;
-        srcStep = stride;
-        tmpsrc += edge * pelsInPart;
-    }
-    else // (dir == EDGE_HOR)
-    {
-        offset = stride;
-        srcStep = 1;
-        tmpsrc += edge * pelsInPart * stride;
-    }
-
-    for (uint32_t idx = 0; idx < numParts; idx++)
-    {
-        bsAbsIdx = xCalcBsIdx(cu, absZOrderIdx, dir, edge, idx);
-        bs = m_blockingStrength[dir][bsAbsIdx];
-        if (bs)
-        {
-            qpQ = cu->getQP(bsAbsIdx);
-            partQ = bsAbsIdx;
-            // Derive neighboring PU index
-            if (dir == EDGE_VER)
-            {
-                cuP = cuQ->getPULeft(partP, partQ, !true, !m_bLFCrossTileBoundary);
-            }
-            else // (dir == EDGE_HOR)
-            {
-                cuP = cuQ->getPUAbove(partP, partQ, !true, false, !m_bLFCrossTileBoundary);
-            }
-
-            qpP = cuP->getQP(partP);
-            qp = (qpP + qpQ + 1) >> 1;
-            int bitdepthScale = 1 << (X265_DEPTH - 8);
-
-            int indexTC = Clip3(0, MAX_QP + DEFAULT_INTRA_TC_OFFSET, int(qp + DEFAULT_INTRA_TC_OFFSET * (bs - 1) + (tcOffsetDiv2 << 1)));
-            int indexB = Clip3(0, MAX_QP, qp + (betaOffsetDiv2 << 1));
-
-            int tc =  sm_tcTable[indexTC] * bitdepthScale;
-            int beta = sm_betaTable[indexB] * bitdepthScale;
-            int sideThreshold = (beta + (beta >> 1)) >> 3;
-            int thrCut = tc * 10;
-
-            uint32_t blocksInPart = pelsInPart / 4 ? pelsInPart / 4 : 1;
-            for (uint32_t blkIdx = 0; blkIdx < blocksInPart; blkIdx++)
-            {
-                int dp0 = xCalcDP(tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 0), offset);
-                int dq0 = xCalcDQ(tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 0), offset);
-                int dp3 = xCalcDP(tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 3), offset);
-                int dq3 = xCalcDQ(tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 3), offset);
-                int d0 = dp0 + dq0;
-                int d3 = dp3 + dq3;
-
-                int dp = dp0 + dp3;
-                int dq = dq0 + dq3;
-                int d =  d0 + d3;
-
-                if (bPCMFilter || cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-                {
-                    // Check if each of PUs is I_PCM with LF disabling
-                    bPartPNoFilter = (bPCMFilter && cuP->getIPCMFlag(partP));
-                    bPartQNoFilter = (bPCMFilter && cuQ->getIPCMFlag(partQ));
-
-                    // check if each of PUs is lossless coded
-                    bPartPNoFilter = bPartPNoFilter || (cuP->isLosslessCoded(partP));
-                    bPartQNoFilter = bPartQNoFilter || (cuQ->isLosslessCoded(partQ));
-                }
-
-                if (d < beta)
-                {
-                    bool bFilterP = (dp < sideThreshold);
-                    bool bFilterQ = (dq < sideThreshold);
-
-                    bool sw =  xUseStrongFiltering(offset, 2 * d0, beta, tc, tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 0))
-                        && xUseStrongFiltering(offset, 2 * d3, beta, tc, tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + 3));
-
-                    for (int i = 0; i < DEBLOCK_SMALLEST_BLOCK / 2; i++)
-                    {
-                        xPelFilterLuma(tmpsrc + srcStep * (idx * pelsInPart + blkIdx * 4 + i), offset, tc, sw, bPartPNoFilter, bPartQNoFilter, thrCut, bFilterP, bFilterQ);
-                    }
-                }
-            }
-        }
-    }
-}
-
-void TComLoopFilter::xEdgeFilterChroma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int dir, int edge)
-{
-    TComPicYuv* reconYuv = cu->getPic()->getPicYuvRec();
-    int stride = reconYuv->getCStride();
-    Pel* srcCb = reconYuv->getCbAddr(cu->getAddr(), absZOrderIdx);
-    Pel* srcCr = reconYuv->getCrAddr(cu->getAddr(), absZOrderIdx);
-    int qp = 0;
-    int qpP = 0;
-    int qpQ = 0;
-    uint32_t  pelsInPartChroma = g_maxCUSize >> (g_maxCUDepth + cu->getHorzChromaShift());
-    int   offset, srcStep;
-
-    const uint32_t lcuWidthInBaseUnits = cu->getPic()->getNumPartInWidth();
-
-    bool  bPCMFilter = (cu->getSlice()->getSPS()->getUsePCM() && cu->getSlice()->getSPS()->getPCMFilterDisableFlag()) ? true : false;
-    bool  bPartPNoFilter = false;
-    bool  bPartQNoFilter = false;
-    uint32_t  partP;
-    uint32_t  partQ;
-    TComDataCU* cuP;
-    TComDataCU* cuQ = cu;
-    int tcOffsetDiv2 = cu->getSlice()->getDeblockingFilterTcOffsetDiv2();
-
-    // Vertical Position
-    uint32_t edgeNumInLCUVert = g_zscanToRaster[absZOrderIdx] % lcuWidthInBaseUnits + edge;
-    uint32_t edgeNumInLCUHor = g_zscanToRaster[absZOrderIdx] / lcuWidthInBaseUnits + edge;
-
-    if ((pelsInPartChroma < DEBLOCK_SMALLEST_BLOCK) &&
-        (((edgeNumInLCUVert % (DEBLOCK_SMALLEST_BLOCK / pelsInPartChroma)) && (dir == 0)) ||
-         ((edgeNumInLCUHor % (DEBLOCK_SMALLEST_BLOCK / pelsInPartChroma)) && dir)))
-    {
-        return;
-    }
-
-    uint32_t numParts = cu->getPic()->getNumPartInWidth() >> depth;
-
-    uint32_t bsAbsIdx;
-    UChar bs;
-
-    Pel* tmpSrcCb = srcCb;
-    Pel* tmpSrcCr = srcCr;
-
-    if (dir == EDGE_VER)
-    {
-        offset   = 1;
-        srcStep  = stride;
-        tmpSrcCb += edge * pelsInPartChroma;
-        tmpSrcCr += edge * pelsInPartChroma;
-    }
-    else // (dir == EDGE_HOR)
-    {
-        offset   = stride;
-        srcStep  = 1;
-        tmpSrcCb += edge * stride * pelsInPartChroma;
-        tmpSrcCr += edge * stride * pelsInPartChroma;
-    }
-
-    for (uint32_t idx = 0; idx < numParts; idx++)
-    {
-        bsAbsIdx = xCalcBsIdx(cu, absZOrderIdx, dir, edge, idx);
-        bs = m_blockingStrength[dir][bsAbsIdx];
-
-        if (bs > 1)
-        {
-            qpQ = cu->getQP(bsAbsIdx);
-            partQ = bsAbsIdx;
-            // Derive neighboring PU index
-            if (dir == EDGE_VER)
-            {
-                cuP = cuQ->getPULeft(partP, partQ, !true, !m_bLFCrossTileBoundary);
-            }
-            else // (dir == EDGE_HOR)
-            {
-                cuP = cuQ->getPUAbove(partP, partQ, !true, false, !m_bLFCrossTileBoundary);
-            }
-
-            qpP = cuP->getQP(partP);
-
-            if (bPCMFilter || cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-            {
-                // Check if each of PUs is I_PCM with LF disabling
-                bPartPNoFilter = (bPCMFilter && cuP->getIPCMFlag(partP));
-                bPartQNoFilter = (bPCMFilter && cuQ->getIPCMFlag(partQ));
-
-                // check if each of PUs is lossless coded
-                bPartPNoFilter = bPartPNoFilter || (cuP->isLosslessCoded(partP));
-                bPartQNoFilter = bPartQNoFilter || (cuQ->isLosslessCoded(partQ));
-            }
-
-            for (uint32_t chromaIdx = 0; chromaIdx < 2; chromaIdx++)
-            {
-                int chromaQPOffset  = (chromaIdx == 0) ? cu->getSlice()->getPPS()->getChromaCbQpOffset() : cu->getSlice()->getPPS()->getChromaCrQpOffset();
-                Pel* piTmpSrcChroma = (chromaIdx == 0) ? tmpSrcCb : tmpSrcCr;
-                qp = QpUV((((qpP + qpQ + 1) >> 1) + chromaQPOffset), cu->getChromaFormat());
-                int iBitdepthScale = 1 << (X265_DEPTH - 8);
-
-                int iIndexTC = Clip3(0, MAX_QP + DEFAULT_INTRA_TC_OFFSET, qp + DEFAULT_INTRA_TC_OFFSET * (bs - 1) + (tcOffsetDiv2 << 1));
-                int iTc =  sm_tcTable[iIndexTC] * iBitdepthScale;
-
-                for (uint32_t uiStep = 0; uiStep < pelsInPartChroma; uiStep++)
-                {
-                    xPelFilterChroma(piTmpSrcChroma + srcStep * (uiStep + idx * pelsInPartChroma), offset, iTc, bPartPNoFilter, bPartQNoFilter);
-                }
-            }
-        }
-    }
-}
-
-/**
- - Deblocking for the luminance component with strong or weak filter
- .
- \param src           pointer to picture data
- \param offset         offset value for picture data
- \param tc              tc value
- \param sw              decision strong/weak filter
- \param bPartPNoFilter  indicator to disable filtering on partP
- \param bPartQNoFilter  indicator to disable filtering on partQ
- \param iThrCut         threshold value for weak filter decision
- \param bFilterSecondP  decision weak filter/no filter for partP
- \param bFilterSecondQ  decision weak filter/no filter for partQ
-*/
-inline void TComLoopFilter::xPelFilterLuma(Pel* src, int offset, int tc, bool sw, bool bPartPNoFilter, bool bPartQNoFilter, int thrCut, bool bFilterSecondP, bool bFilterSecondQ)
-{
-    int delta;
-
-    int16_t m4  = (int16_t)src[0];
-    int16_t m3  = (int16_t)src[-offset];
-    int16_t m5  = (int16_t)src[offset];
-    int16_t m2  = (int16_t)src[-offset * 2];
-    int16_t m6  = (int16_t)src[offset * 2];
-    int16_t m1  = (int16_t)src[-offset * 3];
-    int16_t m7  = (int16_t)src[offset * 3];
-    int16_t m0  = (int16_t)src[-offset * 4];
-
-    if (sw)
-    {
-        src[-offset]     = (Pel)Clip3(m3 - 2 * tc, m3 + 2 * tc, ((m1 + 2 * m2 + 2 * m3 + 2 * m4 + m5 + 4) >> 3));
-        src[0]           = (Pel)Clip3(m4 - 2 * tc, m4 + 2 * tc, ((m2 + 2 * m3 + 2 * m4 + 2 * m5 + m6 + 4) >> 3));
-        src[-offset * 2] = (Pel)Clip3(m2 - 2 * tc, m2 + 2 * tc, ((m1 + m2 + m3 + m4 + 2) >> 2));
-        src[offset]      = (Pel)Clip3(m5 - 2 * tc, m5 + 2 * tc, ((m3 + m4 + m5 + m6 + 2) >> 2));
-        src[-offset * 3] = (Pel)Clip3(m1 - 2 * tc, m1 + 2 * tc, ((2 * m0 + 3 * m1 + m2 + m3 + m4 + 4) >> 3));
-        src[offset * 2]  = (Pel)Clip3(m6 - 2 * tc, m6 + 2 * tc, ((m3 + m4 + m5 + 3 * m6 + 2 * m7 + 4) >> 3));
-    }
-    else
-    {
-        /* Weak filter */
-        delta = (9 * (m4 - m3) - 3 * (m5 - m2) + 8) >> 4;
-
-        if (abs(delta) < thrCut)
-        {
-            delta = Clip3(-tc, tc, delta);
-            src[-offset] = (Pel)ClipY((m3 + delta));
-            src[0] = (Pel)ClipY((m4 - delta));
-
-            int tc2 = tc >> 1;
-            if (bFilterSecondP)
-            {
-                int delta1 = Clip3(-tc2, tc2, ((((m1 + m3 + 1) >> 1) - m2 + delta) >> 1));
-                src[-offset * 2] = (Pel)ClipY((m2 + delta1));
-            }
-            if (bFilterSecondQ)
-            {
-                int delta2 = Clip3(-tc2, tc2, ((((m6 + m4 + 1) >> 1) - m5 - delta) >> 1));
-                src[offset] = (Pel)ClipY((m5 + delta2));
-            }
-        }
-    }
-
-    if (bPartPNoFilter)
-    {
-        src[-offset] = (Pel)m3;
-        src[-offset * 2] = (Pel)m2;
-        src[-offset * 3] = (Pel)m1;
-    }
-    if (bPartQNoFilter)
-    {
-        src[0] = (Pel)m4;
-        src[offset] = (Pel)m5;
-        src[offset * 2] = (Pel)m6;
-    }
-}
-
-/**
- - Deblocking of one line/column for the chrominance component
- .
- \param src           pointer to picture data
- \param offset         offset value for picture data
- \param tc              tc value
- \param bPartPNoFilter  indicator to disable filtering on partP
- \param bPartQNoFilter  indicator to disable filtering on partQ
- */
-inline void TComLoopFilter::xPelFilterChroma(Pel* src, int offset, int tc, bool bPartPNoFilter, bool bPartQNoFilter)
-{
-    int delta;
-
-    int16_t m4  = (int16_t)src[0];
-    int16_t m3  = (int16_t)src[-offset];
-    int16_t m5  = (int16_t)src[offset];
-    int16_t m2  = (int16_t)src[-offset * 2];
-
-    delta = Clip3(-tc, tc, ((((m4 - m3) << 2) + m2 - m5 + 4) >> 3));
-    src[-offset] = (Pel)ClipC(m3 + delta);
-    src[0] = (Pel)ClipC(m4 - delta);
-
-    if (bPartPNoFilter)
-    {
-        src[-offset] = (Pel)m3;
-    }
-    if (bPartQNoFilter)
-    {
-        src[0] = (Pel)m4;
-    }
-}
-
-/**
- - Decision between strong and weak filter
- .
- \param offset         offset value for picture data
- \param d               d value
- \param beta            beta value
- \param tc              tc value
- \param src           pointer to picture data
- */
-inline bool TComLoopFilter::xUseStrongFiltering(int offset, int d, int beta, int tc, Pel* src)
-{
-    int16_t m4  = (int16_t)src[0];
-    int16_t m3  = (int16_t)src[-offset];
-    int16_t m7  = (int16_t)src[offset * 3];
-    int16_t m0  = (int16_t)src[-offset * 4];
-
-    int d_strong = abs(m0 - m3) + abs(m7 - m4);
-
-    return (d_strong < (beta >> 3)) && (d < (beta >> 2)) && (abs(m3 - m4) < ((tc * 5 + 1) >> 1));
-}
-
-inline int TComLoopFilter::xCalcDP(Pel* src, int offset)
-{
-    return abs(static_cast<int>(src[-offset * 3]) - 2 * src[-offset * 2] + src[-offset]);
-}
-
-inline int TComLoopFilter::xCalcDQ(Pel* src, int offset)
-{
-    return abs(static_cast<int>(src[0]) - 2 * src[offset] + src[offset * 2]);
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComLoopFilter.h Deleted

@@ -1,144 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComLoopFilter.h
-    \brief    deblocking filter (header)
-*/
-
-#ifndef X265_TCOMLOOPFILTER_H
-#define X265_TCOMLOOPFILTER_H
-
-#include "CommonDef.h"
-#include "TComPic.h"
-
-//! \ingroup TLibCommon
-//! \{
-
-namespace x265 {
-// private namespace
-
-#define DEBLOCK_SMALLEST_BLOCK  8
-#define EDGE_VER                0
-#define EDGE_HOR                1
-
-/// parameters for deblocking filter
-typedef struct _LFCUParam
-{
-    bool bLeftEdge;                       ///< indicates left edge
-    bool bTopEdge;                        ///< indicates top edge
-} LFCUParam;
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// deblocking filter class
-class TComLoopFilter
-{
-private:
-
-    uint32_t      m_numPartitions;
-    UChar*    m_blockingStrength[2];            ///< Bs for [Ver/Hor][Y/U/V][Blk_Idx]
-    bool*     m_bEdgeFilter[2];
-    LFCUParam m_lfcuParam;                ///< status structure
-
-    bool      m_bLFCrossTileBoundary;
-
-protected:
-
-    /// CU-level deblocking function
-    void xDeblockCU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int Edge);
-
-    // set / get functions
-    void xSetLoopfilterParam(TComDataCU* cu, uint32_t absZOrderIdx);
-    // filtering functions
-    void xSetEdgefilterTU(TComDataCU* cu, uint32_t absTUPartIdx, uint32_t absZOrderIdx, uint32_t depth);
-    void xSetEdgefilterPU(TComDataCU* cu, uint32_t absZOrderIdx);
-    void xGetBoundaryStrengthSingle(TComDataCU* cu, int dir, uint32_t partIdx);
-    uint32_t xCalcBsIdx(TComDataCU* cu, uint32_t absZOrderIdx, int dir, int edgeIdx, int baseUnitIdx)
-    {
-        TComPic* const pic = cu->getPic();
-        const uint32_t lcuWidthInBaseUnits = pic->getNumPartInWidth();
-
-        if (dir == 0)
-        {
-            return g_rasterToZscan[g_zscanToRaster[absZOrderIdx] + baseUnitIdx * lcuWidthInBaseUnits + edgeIdx];
-        }
-        else
-        {
-            return g_rasterToZscan[g_zscanToRaster[absZOrderIdx] + edgeIdx * lcuWidthInBaseUnits + baseUnitIdx];
-        }
-    }
-
-    void xSetEdgefilterMultiple(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int dir, int edgeIdx, bool bValue, uint32_t widthInBaseUnits = 0, uint32_t heightInBaseUnits = 0);
-
-    void xEdgeFilterLuma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int dir, int edge);
-    void xEdgeFilterChroma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int dir, int edge);
-
-    inline void xPelFilterLuma(Pel* src, int offset, int tc, bool sw, bool bPartPNoFilter, bool bPartQNoFilter, int iThrCut, bool bFilterSecondP, bool bFilterSecondQ);
-    inline void xPelFilterChroma(Pel* src, int offset, int tc, bool bPartPNoFilter, bool bPartQNoFilter);
-
-    inline bool xUseStrongFiltering(int offset, int d, int beta, int tc, Pel* src);
-    inline int xCalcDP(Pel* src, int offset);
-    inline int xCalcDQ(Pel* src, int offset);
-
-    static const UChar sm_tcTable[54];
-    static const UChar sm_betaTable[52];
-
-public:
-
-    TComLoopFilter();
-    virtual ~TComLoopFilter();
-
-    void  create(uint32_t maxCUDepth);
-    void  destroy();
-
-    /// set configuration
-    void setCfg(bool bLFCrossTileBoundary);
-
-    /// picture-level deblocking filter
-    void loopFilterPic(TComPic* pic);
-
-    void loopFilterCU(TComDataCU* cu, int dir);
-
-    static int getBeta(int qp)
-    {
-        int indexB = Clip3(0, MAX_QP, qp);
-
-        return sm_betaTable[indexB];
-    }
-};
-}
-//! \}
-
-#endif // ifndef X265_TCOMLOOPFILTER_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPic.cpp Deleted

@@ -1,200 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComPic.cpp
-    \brief    picture class
-*/
-
-#include "TComPic.h"
-#include "SEI.h"
-#include "mv.h"
-#include "encoder.h"
-
-using namespace x265;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
-TComPic::TComPic()
-    : m_picSym(NULL)
-    , m_origPicYuv(NULL)
-    , m_reconPicYuv(NULL)
-    , m_bUsedByCurr(false)
-    , m_bIsLongTerm(false)
-    , m_bCheckLTMSB(false)
-    , m_rowDiagQp(NULL)
-    , m_rowDiagQScale(NULL)
-    , m_rowDiagSatd(NULL)
-    , m_rowDiagIntraSatd(NULL)
-    , m_rowEncodedBits(NULL)
-    , m_numEncodedCusPerRow(NULL)
-    , m_rowSatdForVbv(NULL)
-    , m_cuCostsForVbv(NULL)
-    , m_intraCuCostsForVbv(NULL)
-{
-    m_reconRowCount.set(0);
-    m_countRefEncoders = 0;
-    memset(&m_lowres, 0, sizeof(m_lowres));
-    m_next = NULL;
-    m_prev = NULL;
-    m_SSDY = 0;
-    m_SSDU = 0;
-    m_SSDV = 0;
-    m_ssim = 0;
-    m_ssimCnt = 0;
-    m_frameTime = 0.0;
-    m_elapsedCompressTime = 0.0;
-    m_qpaAq = 0;
-    m_qpaRc = 0;
-    m_avgQpRc = 0;
-    m_avgQpAq = 0;
-    m_bChromaPlanesExtended = false;
-}
-
-TComPic::~TComPic()
-{}
-
-bool TComPic::create(Encoder* cfg)
-{
-    /* store conformance window parameters with picture */
-    m_conformanceWindow = cfg->m_conformanceWindow;
-
-    /* store display window parameters with picture */
-    m_defaultDisplayWindow = cfg->m_defaultDisplayWindow;
-
-    m_picSym = new TComPicSym;
-    m_origPicYuv = new TComPicYuv;
-    m_reconPicYuv = new TComPicYuv;
-    if (!m_picSym || !m_origPicYuv || !m_reconPicYuv)
-        return false;
-
-    bool ok = true;
-    ok &= m_picSym->create(cfg->param->sourceWidth, cfg->param->sourceHeight, cfg->param->internalCsp, g_maxCUSize, g_maxCUDepth);
-    ok &= m_origPicYuv->create(cfg->param->sourceWidth, cfg->param->sourceHeight, cfg->param->internalCsp, g_maxCUSize, g_maxCUSize, g_maxCUDepth);
-    ok &= m_reconPicYuv->create(cfg->param->sourceWidth, cfg->param->sourceHeight, cfg->param->internalCsp, g_maxCUSize, g_maxCUSize, g_maxCUDepth);
-    ok &= m_lowres.create(m_origPicYuv, cfg->param->bframes, !!cfg->param->rc.aqMode);
-
-    bool isVbv = cfg->param->rc.vbvBufferSize > 0 && cfg->param->rc.vbvMaxBitrate > 0;
-    if (ok && (isVbv || cfg->param->rc.aqMode))
-    {
-        int numRows = m_picSym->getFrameHeightInCU();
-        int numCols = m_picSym->getFrameWidthInCU();
-
-        if (cfg->param->rc.aqMode)
-            CHECKED_MALLOC(m_qpaAq, double, numRows);
-        if (isVbv)
-        {
-            CHECKED_MALLOC(m_rowDiagQp, double, numRows);
-            CHECKED_MALLOC(m_rowDiagQScale, double, numRows);
-            CHECKED_MALLOC(m_rowDiagSatd, uint32_t, numRows);
-            CHECKED_MALLOC(m_rowDiagIntraSatd, uint32_t, numRows);
-            CHECKED_MALLOC(m_rowEncodedBits, uint32_t, numRows);
-            CHECKED_MALLOC(m_numEncodedCusPerRow, uint32_t, numRows);
-            CHECKED_MALLOC(m_rowSatdForVbv, uint32_t, numRows);
-            CHECKED_MALLOC(m_cuCostsForVbv, uint32_t, numRows * numCols);
-            CHECKED_MALLOC(m_intraCuCostsForVbv, uint32_t, numRows * numCols);
-            CHECKED_MALLOC(m_qpaRc, double, numRows);
-        }
-        reInit(cfg);
-    }
-
-    return ok;
-
-fail:
-    ok = false;
-    return ok;
-}
-
-void TComPic::reInit(Encoder* cfg)
-{
-    if (cfg->param->rc.vbvBufferSize > 0 && cfg->param->rc.vbvMaxBitrate > 0)
-    {
-        int numRows = m_picSym->getFrameHeightInCU();
-        int numCols = m_picSym->getFrameWidthInCU();
-        memset(m_rowDiagQp, 0, numRows * sizeof(double));
-        memset(m_rowDiagQScale, 0, numRows * sizeof(double));
-        memset(m_rowDiagSatd, 0, numRows * sizeof(uint32_t));
-        memset(m_rowDiagIntraSatd, 0, numRows * sizeof(uint32_t));
-        memset(m_rowEncodedBits, 0, numRows * sizeof(uint32_t));
-        memset(m_numEncodedCusPerRow, 0, numRows * sizeof(uint32_t));
-        memset(m_rowSatdForVbv, 0, numRows * sizeof(uint32_t));
-        memset(m_cuCostsForVbv, 0,  numRows * numCols * sizeof(uint32_t));
-        memset(m_intraCuCostsForVbv, 0, numRows * numCols * sizeof(uint32_t));
-        memset(m_qpaRc, 0, numRows * sizeof(double));
-    }
-    if (cfg->param->rc.aqMode)
-        memset(m_qpaAq, 0,  m_picSym->getFrameHeightInCU() * sizeof(double));
-}
-
-void TComPic::destroy()
-{
-    if (m_picSym)
-    {
-        m_picSym->destroy();
-        delete m_picSym;
-        m_picSym = NULL;
-    }
-
-    if (m_origPicYuv)
-    {
-        m_origPicYuv->destroy();
-        delete m_origPicYuv;
-        m_origPicYuv = NULL;
-    }
-
-    if (m_reconPicYuv)
-    {
-        m_reconPicYuv->destroy();
-        delete m_reconPicYuv;
-        m_reconPicYuv = NULL;
-    }
-    m_lowres.destroy();
-
-    X265_FREE(m_rowDiagQp);
-    X265_FREE(m_rowDiagQScale);
-    X265_FREE(m_rowDiagSatd);
-    X265_FREE(m_rowDiagIntraSatd);
-    X265_FREE(m_rowEncodedBits);
-    X265_FREE(m_numEncodedCusPerRow);
-    X265_FREE(m_rowSatdForVbv);
-    X265_FREE(m_cuCostsForVbv);
-    X265_FREE(m_intraCuCostsForVbv);
-    X265_FREE(m_qpaAq);
-    X265_FREE(m_qpaRc);
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPic.h Deleted

@@ -1,180 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComPic.h
-    \brief    picture class (header)
-*/
-
-#ifndef X265_TCOMPIC_H
-#define X265_TCOMPIC_H
-
-// Include files
-#include "CommonDef.h"
-#include "TComPicSym.h"
-#include "TComPicYuv.h"
-#include "lowres.h"
-#include "threading.h"
-#include "md5.h"
-
-namespace x265 {
-// private namespace
-
-class Encoder;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// picture class (symbol + YUV buffers)
-class TComPic
-{
-private:
-
-    TComPicSym*           m_picSym;
-    TComPicYuv*           m_origPicYuv;
-    TComPicYuv*           m_reconPicYuv;
-
-    Window                m_conformanceWindow;
-    Window                m_defaultDisplayWindow;
-
-    bool                  m_bUsedByCurr;          // Used by current picture
-    bool                  m_bIsLongTerm;          // IS long term picture
-    bool                  m_bCheckLTMSB;
-
-public:
-
-    //** Frame Parallelism - notification between FrameEncoders of available motion reference rows **
-    ThreadSafeInteger     m_reconRowCount;      // count of CTU rows completely reconstructed and extended for motion reference
-    volatile uint32_t     m_countRefEncoders;   // count of FrameEncoder threads monitoring m_reconRowCount
-    void*                 m_userData;           // user provided pointer passed in with this picture
-
-    int64_t               m_pts;                // user provided presentation time stamp
-    int64_t               m_reorderedPts;
-    int64_t               m_dts;
-
-    Lowres                m_lowres;
-
-    TComPic*              m_next;               // PicList doubly linked list pointers
-    TComPic*              m_prev;
-
-    uint64_t              m_SSDY;
-    uint64_t              m_SSDU;
-    uint64_t              m_SSDV;
-    double                m_ssim;
-    int                   m_ssimCnt;
-
-    double                m_elapsedCompressTime; // elapsed time spent in worker threads
-    double                m_frameTime;           // wall time from frame start to finish
-
-    MD5Context            m_state[3];
-    uint32_t              m_crc[3];
-    uint32_t              m_checksum[3];
-    bool                  m_bChromaPlanesExtended; // orig chroma planes motion extended for weightp analysis
-    double*               m_rowDiagQp;
-    double*               m_rowDiagQScale;
-    uint32_t*             m_rowDiagSatd;
-    uint32_t*             m_rowDiagIntraSatd;
-    uint32_t*             m_rowEncodedBits;
-    uint32_t*             m_numEncodedCusPerRow;
-    uint32_t*             m_rowSatdForVbv;
-    uint32_t*             m_cuCostsForVbv;
-    uint32_t*             m_intraCuCostsForVbv;
-    double*               m_qpaAq;
-    double*               m_qpaRc;
-    double                m_avgQpRc; //avg QP as decided by ratecontrol
-    double                m_avgQpAq; //avg QP as decided by AQ in addition to ratecontrol
-
-    TComPic();
-    virtual ~TComPic();
-
-    bool          create(Encoder* cfg);
-    virtual void  destroy();
-    void          reInit(Encoder* cfg);
-
-    bool          getUsedByCurr()           { return m_bUsedByCurr; }
-
-    void          setUsedByCurr(bool bUsed) { m_bUsedByCurr = bUsed; }
-
-    bool          getIsLongTerm()           { return m_bIsLongTerm; }
-
-    void          setIsLongTerm(bool lt)    { m_bIsLongTerm = lt; }
-
-    void          setCheckLTMSBPresent(bool b) { m_bCheckLTMSB = b; }
-
-    bool          getCheckLTMSBPresent()  { return m_bCheckLTMSB; }
-
-    TComPicSym*   getPicSym()             { return m_picSym; }
-
-    TComSlice*    getSlice()              { return m_picSym->getSlice(); }
-
-    int           getPOC()                { return m_picSym->getSlice()->getPOC(); }
-
-    TComDataCU*   getCU(uint32_t cuAddr)  { return m_picSym->getCU(cuAddr); }
-
-    TComPicYuv*   getPicYuvOrg()          { return m_origPicYuv; }
-
-    TComPicYuv*   getPicYuvRec()          { return m_reconPicYuv; }
-
-    uint32_t      getNumCUsInFrame()      { return m_picSym->getNumberOfCUsInFrame(); }
-
-    uint32_t      getNumPartInWidth()     { return m_picSym->getNumPartInWidth(); }
-
-    uint32_t      getNumPartInHeight()    { return m_picSym->getNumPartInHeight(); }
-
-    uint32_t      getNumPartInCU()        { return m_picSym->getNumPartition(); }
-
-    uint32_t      getFrameWidthInCU()     { return m_picSym->getFrameWidthInCU(); }
-
-    uint32_t      getFrameHeightInCU()    { return m_picSym->getFrameHeightInCU(); }
-
-    uint32_t      getMinCUSize()          { return m_picSym->getMinCUSize(); }
-
-    uint32_t      getParPelX(UChar partIdx) { return getParPelX(partIdx); }
-
-    uint32_t      getParPelY(UChar partIdx) { return getParPelX(partIdx); }
-
-    int           getStride()             { return m_reconPicYuv->getStride(); }
-
-    int           getCStride()            { return m_reconPicYuv->getCStride(); }
-
-    Window&       getConformanceWindow()  { return m_conformanceWindow; }
-
-    Window&       getDefDisplayWindow()   { return m_defaultDisplayWindow; }
-}; // END CLASS DEFINITION TComPic
-}
-//! \}
-
-#endif // ifndef X265_TCOMPIC_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPrediction.cpp Deleted

@@ -1,607 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComPrediction.cpp
-    \brief    prediction class
-*/
-
-#include "TComPrediction.h"
-#include "primitives.h"
-#include "common.h"
-
-using namespace x265;
-
-//! \ingroup TLibCommon
-//! \{
-
-static const UChar intraFilterThreshold[5] =
-{
-    10, //4x4
-    7,  //8x8
-    1,  //16x16
-    0,  //32x32
-    10, //64x64
-};
-
-// ====================================================================================================================
-// Constructor / destructor / initialize
-// ====================================================================================================================
-
-TComPrediction::TComPrediction()
-{
-    m_predBuf = NULL;
-    m_refAbove = NULL;
-    m_refAboveFlt = NULL;
-    m_refLeft = NULL;
-    m_refLeftFlt = NULL;
-    m_immedVals = NULL;
-}
-
-TComPrediction::~TComPrediction()
-{
-    X265_FREE(m_predBuf);
-    X265_FREE(m_refAbove);
-    X265_FREE(m_refAboveFlt);
-    X265_FREE(m_refLeft);
-    X265_FREE(m_refLeftFlt);
-    X265_FREE(m_immedVals);
-
-    m_predYuv[0].destroy();
-    m_predYuv[1].destroy();
-    m_predShortYuv[0].destroy();
-    m_predShortYuv[1].destroy();
-    m_predTempYuv.destroy();
-}
-
-void TComPrediction::initTempBuff(int csp)
-{
-    m_hChromaShift = CHROMA_H_SHIFT(csp);
-    m_vChromaShift = CHROMA_V_SHIFT(csp);
-
-    if (m_predBuf == NULL)
-    {
-        m_predBufHeight = ((MAX_CU_SIZE + 2) << 4);
-        m_predBufStride = ((MAX_CU_SIZE + 8) << 4);
-        m_predBuf = X265_MALLOC(pixel, m_predBufStride * m_predBufHeight);
-
-        m_refAbove = X265_MALLOC(pixel, 3 * MAX_CU_SIZE);
-        m_refAboveFlt = X265_MALLOC(pixel, 3 * MAX_CU_SIZE);
-        m_refLeft = X265_MALLOC(pixel, 3 * MAX_CU_SIZE);
-        m_refLeftFlt = X265_MALLOC(pixel, 3 * MAX_CU_SIZE);
-
-        m_predYuv[0].create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
-        m_predYuv[1].create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
-        m_predShortYuv[0].create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
-        m_predShortYuv[1].create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
-        m_predTempYuv.create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
-
-        m_immedVals = X265_MALLOC(int16_t, 64 * (64 + NTAPS_LUMA - 1));
-    }
-}
-
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
-
-bool TComPrediction::filteringIntraReferenceSamples(uint32_t dirMode, uint32_t width)
-{
-    bool bFilter;
-
-    if (dirMode == DC_IDX)
-    {
-        bFilter = false; // no smoothing for DC or LM chroma
-    }
-    else
-    {
-        int diff = std::min<int>(abs((int)dirMode - HOR_IDX), abs((int)dirMode - VER_IDX));
-        uint32_t sizeIndex = g_convertToBit[width];
-        bFilter = diff > intraFilterThreshold[sizeIndex];
-    }
-
-    return bFilter;
-}
-
-void TComPrediction::predIntraLumaAng(uint32_t dirMode, pixel* dst, intptr_t stride, int width)
-{
-    assert(width >= 4 && width <= 64);
-    int log2BlkSize = g_convertToBit[width];
-    bool bUseFilteredPredictions = TComPrediction::filteringIntraReferenceSamples(dirMode, width);
-
-    pixel *refLft, *refAbv;
-    refLft = m_refLeft + width - 1;
-    refAbv = m_refAbove + width - 1;
-
-    pixel *src = m_predBuf;
-    if (bUseFilteredPredictions)
-    {
-        src += ADI_BUF_STRIDE * (2 * width + 1);
-        refLft = m_refLeftFlt + width - 1;
-        refAbv = m_refAboveFlt + width - 1;
-    }
-
-    bool bFilter = width <= 16 && dirMode != PLANAR_IDX;
-    primitives.intra_pred[log2BlkSize][dirMode](dst, stride, refLft, refAbv, dirMode, bFilter);
-}
-
-// Angular chroma
-void TComPrediction::predIntraChromaAng(pixel* src, uint32_t dirMode, pixel* dst, intptr_t stride, int width, int height, int chFmt)
-{
-    int log2BlkSize = g_convertToBit[width];
-
-    // Create the prediction
-    pixel refAbv[3 * MAX_CU_SIZE];
-    pixel refLft[3 * MAX_CU_SIZE];
-
-    bool bUseFilteredPredictions = true;
-
-    if (chFmt != CHROMA_444)
-    {
-        bUseFilteredPredictions = false;
-    }
-    else
-    {
-        assert(width >= 4 && height >= 4 && width < 128 && height < 128);
-        bUseFilteredPredictions = TComPrediction::filteringIntraReferenceSamples(dirMode, width);
-    }
-
-    if (bUseFilteredPredictions)
-    {
-        uint32_t cuWidth2  = width << 1;
-        uint32_t cuHeight2 = height << 1;
-        // generate filtered intra prediction samples
-        // left and left above border + above and above right border + top left corner = length of 3. filter buffer
-        int bufSize = cuHeight2 + cuWidth2 + 1;
-        uint32_t wh = ADI_BUF_STRIDE * height;         // number of elements in one buffer
-
-        pixel* filteredBuf1 = src + wh;             // 1. filter buffer
-        pixel* filteredBuf2 = filteredBuf1 + wh;    // 2. filter buffer
-        pixel* filterBuf    = filteredBuf2 + wh;    // buffer for 2. filtering (sequential)
-        pixel* filterBufN   = filterBuf + bufSize;  // buffer for 1. filtering (sequential)
-
-        int l = 0;
-        // left border from bottom to top
-        for (int i = 0; i < cuHeight2; i++)
-        {
-            filterBuf[l++] = src[ADI_BUF_STRIDE * (cuHeight2 - i)];
-        }
-
-        // top left corner
-        filterBuf[l++] = src[0];
-
-        // above border from left to right
-        memcpy(&filterBuf[l], &src[1], cuWidth2 * sizeof(*filterBuf));
-
-        // 1. filtering with [1 2 1]
-        filterBufN[0] = filterBuf[0];
-        filterBufN[bufSize - 1] = filterBuf[bufSize - 1];
-        for (int i = 1; i < bufSize - 1; i++)
-        {
-            filterBufN[i] = (filterBuf[i - 1] + 2 * filterBuf[i] + filterBuf[i + 1] + 2) >> 2;
-        }
-
-        // fill 1. filter buffer with filtered values
-        l = 0;
-        for (int i = 0; i < cuHeight2; i++)
-        {
-            filteredBuf1[ADI_BUF_STRIDE * (cuHeight2 - i)] = filterBufN[l++];
-        }
-
-        filteredBuf1[0] = filterBufN[l++];
-        memcpy(&filteredBuf1[1], &filterBufN[l], cuWidth2 * sizeof(*filteredBuf1));
-
-        int limit = (2 * width + 1);
-        src += wh;
-        memcpy(refAbv + width - 1, src, (limit) * sizeof(pixel));
-        for (int k = 0; k < limit; k++)
-        {
-            refLft[k + width - 1] = src[k * ADI_BUF_STRIDE];
-        }
-    }
-    else
-    {
-        int limit = (dirMode <= 25 && dirMode >= 11) ? (width + 1 + 1) : (2 * width + 1);
-        memcpy(refAbv + width - 1, src, (limit) * sizeof(pixel));
-        for (int k = 0; k < limit; k++)
-        {
-            refLft[k + width - 1] = src[k * ADI_BUF_STRIDE];
-        }
-    }
-
-    primitives.intra_pred[log2BlkSize][dirMode](dst, stride, refLft + width - 1, refAbv + width - 1, dirMode, 0);
-}
-
-/** Function for checking identical motion.
- * \param TComDataCU* cu
- * \param uint32_t PartAddr
- */
-bool TComPrediction::xCheckIdenticalMotion(TComDataCU* cu, uint32_t partAddr)
-{
-    assert(cu->getSlice()->isInterB());
-    if (!cu->getSlice()->getPPS()->getWPBiPred())
-    {
-        int refIdxL0 = cu->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partAddr);
-        int refIdxL1 = cu->getCUMvField(REF_PIC_LIST_1)->getRefIdx(partAddr);
-        if (refIdxL0 >= 0 && refIdxL1 >= 0)
-        {
-            int refPOCL0 = cu->getSlice()->getRefPic(REF_PIC_LIST_0, refIdxL0)->getPOC();
-            int refPOCL1 = cu->getSlice()->getRefPic(REF_PIC_LIST_1, refIdxL1)->getPOC();
-            if (refPOCL0 == refPOCL1 && cu->getCUMvField(REF_PIC_LIST_0)->getMv(partAddr) == cu->getCUMvField(REF_PIC_LIST_1)->getMv(partAddr))
-            {
-                return true;
-            }
-        }
-    }
-    return false;
-}
-
-void TComPrediction::motionCompensation(TComDataCU* cu, TComYuv* predYuv, int list, int partIdx, bool bLuma, bool bChroma)
-{
-    int  width;
-    int  height;
-    uint32_t partAddr;
-
-    assert(partIdx >= 0);
-    {
-        cu->getPartIndexAndSize(partIdx, partAddr, width, height);
-        if (cu->getSlice()->isInterP())
-            list = REF_PIC_LIST_0;
-        if (list != REF_PIC_LIST_X)
-        {
-            if (cu->getSlice()->getPPS()->getUseWP())
-            {
-                ShortYuv* shortYuv = &m_predShortYuv[0];
-                int refId = cu->getCUMvField(list)->getRefIdx(partAddr);
-                assert(refId >= 0);
-
-                MV mv = cu->getCUMvField(list)->getMv(partAddr);
-                cu->clipMv(mv);
-                if (bLuma)
-                    xPredInterLumaBlk(cu, cu->getSlice()->getRefPic(list, refId)->getPicYuvRec(), partAddr, &mv, width, height, shortYuv);
-                if (bChroma)
-                    xPredInterChromaBlk(cu, cu->getSlice()->getRefPic(list, refId)->getPicYuvRec(), partAddr, &mv, width, height, shortYuv);
-
-                xWeightedPredictionUni(cu, shortYuv, partAddr, width, height, list, predYuv, -1, bLuma, bChroma);
-            }
-            else
-            {
-                xPredInterUni(cu, partAddr, width, height, list, predYuv, bLuma, bChroma);
-            }
-        }
-        else
-        {
-            if (xCheckIdenticalMotion(cu, partAddr))
-            {
-                xPredInterUni(cu, partAddr, width, height, REF_PIC_LIST_0, predYuv, bLuma, bChroma);
-            }
-            else
-            {
-                xPredInterBi(cu, partAddr, width, height, predYuv, bLuma, bChroma);
-            }
-        }
-    }
-}
-
-void TComPrediction::xPredInterUni(TComDataCU* cu, uint32_t partAddr, int width, int height, int list, TComYuv* outPredYuv, bool bLuma, bool bChroma)
-{
-    int refIdx = cu->getCUMvField(list)->getRefIdx(partAddr);
-
-    assert(refIdx >= 0);
-
-    MV mv = cu->getCUMvField(list)->getMv(partAddr);
-    cu->clipMv(mv);
-
-    if (bLuma)
-        xPredInterLumaBlk(cu, cu->getSlice()->getRefPic(list, refIdx)->getPicYuvRec(), partAddr, &mv, width, height, outPredYuv);
-
-    if (bChroma)
-        xPredInterChromaBlk(cu, cu->getSlice()->getRefPic(list, refIdx)->getPicYuvRec(), partAddr, &mv, width, height, outPredYuv);
-}
-
-void TComPrediction::xPredInterUni(TComDataCU* cu, uint32_t partAddr, int width, int height, int list, ShortYuv* outPredYuv, bool bLuma, bool bChroma)
-{
-    int refIdx = cu->getCUMvField(list)->getRefIdx(partAddr);
-
-    assert(refIdx >= 0);
-
-    MV mv = cu->getCUMvField(list)->getMv(partAddr);
-    cu->clipMv(mv);
-
-    if (bLuma)
-        xPredInterLumaBlk(cu, cu->getSlice()->getRefPic(list, refIdx)->getPicYuvRec(), partAddr, &mv, width, height, outPredYuv);
-    if (bChroma)
-        xPredInterChromaBlk(cu, cu->getSlice()->getRefPic(list, refIdx)->getPicYuvRec(), partAddr, &mv, width, height, outPredYuv);
-}
-
-void TComPrediction::xPredInterBi(TComDataCU* cu, uint32_t partAddr, int width, int height, TComYuv* outPredYuv, bool bLuma, bool bChroma)
-{
-    assert(cu->getSlice()->isInterB());
-
-    int refIdx[2];
-    refIdx[0] = cu->getCUMvField(REF_PIC_LIST_0)->getRefIdx(partAddr);
-    refIdx[1] = cu->getCUMvField(REF_PIC_LIST_1)->getRefIdx(partAddr);
-
-    if (refIdx[0] >= 0 && refIdx[1] >= 0)
-    {
-        for (int list = 0; list < 2; list++)
-        {
-            assert(refIdx[list] < cu->getSlice()->getNumRefIdx(list));
-
-            xPredInterUni(cu, partAddr, width, height, list, &m_predShortYuv[list], bLuma, bChroma);
-        }
-
-        if (cu->getSlice()->getPPS()->getWPBiPred())
-        {
-            xWeightedPredictionBi(cu, &m_predShortYuv[0], &m_predShortYuv[1], refIdx[0], refIdx[1], partAddr, width, height, outPredYuv, bLuma, bChroma);
-        }
-        else
-        {
-            outPredYuv->addAvg(&m_predShortYuv[0], &m_predShortYuv[1], partAddr, width, height, bLuma, bChroma);
-        }
-    }
-    else if (cu->getSlice()->getPPS()->getWPBiPred())
-    {
-        for (int list = 0; list < 2; list++)
-        {
-            if (refIdx[list] < 0) continue;
-
-            assert(refIdx[list] < cu->getSlice()->getNumRefIdx(list));
-
-            xPredInterUni(cu, partAddr, width, height, list, &m_predShortYuv[list], bLuma, bChroma);
-        }
-
-        xWeightedPredictionBi(cu, &m_predShortYuv[0], &m_predShortYuv[1], refIdx[0], refIdx[1], partAddr, width, height, outPredYuv, bLuma, bChroma);
-    }
-    else if (refIdx[0] >= 0)
-    {
-        const int list = 0;
-
-        assert(refIdx[list] < cu->getSlice()->getNumRefIdx(list));
-
-        xPredInterUni(cu, partAddr, width, height, list, outPredYuv, bLuma, bChroma);
-    }
-    else
-    {
-        assert(refIdx[1] >= 0);
-
-        const int list = 1;
-
-        assert(refIdx[list] < cu->getSlice()->getNumRefIdx(list));
-
-        xPredInterUni(cu, partAddr, width, height, list, outPredYuv, bLuma, bChroma);
-    }
-}
-
-/**
- * \brief Generate motion-compensated luma block
- *
- * \param cu       Pointer to current CU
- * \param refPic   Pointer to reference picture
- * \param partAddr Address of block within CU
- * \param mv       Motion vector
- * \param width    Width of block
- * \param height   Height of block
- * \param dstPic   Pointer to destination picture
- */
-void TComPrediction::xPredInterLumaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, TComYuv *dstPic)
-{
-    int dstStride = dstPic->getStride();
-    pixel *dst      = dstPic->getLumaAddr(partAddr);
-
-    int srcStride = refPic->getStride();
-    int srcOffset = (mv->x >> 2) + (mv->y >> 2) * srcStride;
-    int partEnum = partitionFromSizes(width, height);
-    pixel* src = refPic->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + srcOffset;
-
-    int xFrac = mv->x & 0x3;
-    int yFrac = mv->y & 0x3;
-
-    if ((yFrac | xFrac) == 0)
-    {
-        primitives.luma_copy_pp[partEnum](dst, dstStride, src, srcStride);
-    }
-    else if (yFrac == 0)
-    {
-        primitives.luma_hpp[partEnum](src, srcStride, dst, dstStride, xFrac);
-    }
-    else if (xFrac == 0)
-    {
-        primitives.luma_vpp[partEnum](src, srcStride, dst, dstStride, yFrac);
-    }
-    else
-    {
-        int tmpStride = width;
-        int filterSize = NTAPS_LUMA;
-        int halfFilterSize = (filterSize >> 1);
-        primitives.luma_hps[partEnum](src, srcStride, m_immedVals, tmpStride, xFrac, 1);
-        primitives.luma_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * tmpStride, tmpStride, dst, dstStride, yFrac);
-    }
-}
-
-//Motion compensated block for biprediction
-void TComPrediction::xPredInterLumaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, ShortYuv *dstPic)
-{
-    int refStride = refPic->getStride();
-    int refOffset = (mv->x >> 2) + (mv->y >> 2) * refStride;
-    pixel *ref    = refPic->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + refOffset;
-
-    int dstStride = dstPic->m_width;
-    int16_t *dst  = dstPic->getLumaAddr(partAddr);
-
-    int xFrac = mv->x & 0x3;
-    int yFrac = mv->y & 0x3;
-
-    int partEnum = partitionFromSizes(width, height);
-
-    assert((width % 4) + (height % 4) == 0);
-    assert(dstStride == MAX_CU_SIZE);
-
-    if ((yFrac | xFrac) == 0)
-    {
-        primitives.luma_p2s(ref, refStride, dst, width, height);
-    }
-    else if (yFrac == 0)
-    {
-        primitives.luma_hps[partEnum](ref, refStride, dst, dstStride, xFrac, 0);
-    }
-    else if (xFrac == 0)
-    {
-        primitives.luma_vps[partEnum](ref, refStride, dst, dstStride, yFrac);
-    }
-    else
-    {
-        int tmpStride = width;
-        int filterSize = NTAPS_LUMA;
-        int halfFilterSize = (filterSize >> 1);
-        primitives.luma_hps[partEnum](ref, refStride, m_immedVals, tmpStride, xFrac, 1);
-        primitives.luma_vss[partEnum](m_immedVals + (halfFilterSize - 1) * tmpStride, tmpStride, dst, dstStride, yFrac);
-    }
-}
-
-/**
- * \brief Generate motion-compensated chroma block
- *
- * \param cu       Pointer to current CU
- * \param refPic   Pointer to reference picture
- * \param partAddr Address of block within CU
- * \param mv       Motion vector
- * \param width    Width of block
- * \param height   Height of block
- * \param dstPic   Pointer to destination picture
- */
-void TComPrediction::xPredInterChromaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, TComYuv *dstPic)
-{
-    int refStride = refPic->getCStride();
-    int dstStride = dstPic->getCStride();
-
-    int shiftHor = (2 + cu->getHorzChromaShift());
-    int shiftVer = (2 + cu->getVertChromaShift());
-
-    int refOffset = (mv->x >> shiftHor) + (mv->y >> shiftVer) * refStride;
-
-    pixel* refCb = refPic->getCbAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + refOffset;
-    pixel* refCr = refPic->getCrAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + refOffset;
-
-    pixel* dstCb = dstPic->getCbAddr(partAddr);
-    pixel* dstCr = dstPic->getCrAddr(partAddr);
-
-    int xFrac = mv->x & ((1 << shiftHor) - 1);
-    int yFrac = mv->y & ((1 << shiftVer) - 1);
-
-    int partEnum = partitionFromSizes(width, height);
-    int csp = cu->getChromaFormat();
-
-    if ((yFrac | xFrac) == 0)
-    {
-        primitives.chroma[csp].copy_pp[partEnum](dstCb, dstStride, refCb, refStride);
-        primitives.chroma[csp].copy_pp[partEnum](dstCr, dstStride, refCr, refStride);
-    }
-    else if (yFrac == 0)
-    {
-        primitives.chroma[csp].filter_hpp[partEnum](refCb, refStride, dstCb, dstStride, xFrac << (1 - cu->getHorzChromaShift()));
-        primitives.chroma[csp].filter_hpp[partEnum](refCr, refStride, dstCr, dstStride, xFrac << (1 - cu->getHorzChromaShift()));
-    }
-    else if (xFrac == 0)
-    {
-        primitives.chroma[csp].filter_vpp[partEnum](refCb, refStride, dstCb, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-        primitives.chroma[csp].filter_vpp[partEnum](refCr, refStride, dstCr, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-    }
-    else
-    {
-        int extStride = width >> m_hChromaShift;
-        int filterSize = NTAPS_CHROMA;
-        int halfFilterSize = (filterSize >> 1);
-
-        primitives.chroma[csp].filter_hps[partEnum](refCb, refStride, m_immedVals, extStride, xFrac << (1 - cu->getHorzChromaShift()), 1);
-        primitives.chroma[csp].filter_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCb, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-
-        primitives.chroma[csp].filter_hps[partEnum](refCr, refStride, m_immedVals, extStride, xFrac << (1 - cu->getHorzChromaShift()), 1);
-        primitives.chroma[csp].filter_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCr, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-    }
-}
-
-// Generate motion compensated block when biprediction
-void TComPrediction::xPredInterChromaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, ShortYuv *dstPic)
-{
-    int refStride = refPic->getCStride();
-    int dstStride = dstPic->m_cwidth;
-
-    int shiftHor = (2 + cu->getHorzChromaShift());
-    int shiftVer = (2 + cu->getVertChromaShift());
-
-    int refOffset = (mv->x >> shiftHor) + (mv->y >> shiftVer) * refStride;
-
-    pixel* refCb = refPic->getCbAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + refOffset;
-    pixel* refCr = refPic->getCrAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr) + refOffset;
-
-    int16_t* dstCb = dstPic->getCbAddr(partAddr);
-    int16_t* dstCr = dstPic->getCrAddr(partAddr);
-
-    int xFrac = mv->x & ((1 << shiftHor) - 1);
-    int yFrac = mv->y & ((1 << shiftVer) - 1);
-
-    int partEnum = partitionFromSizes(width, height);
-    int csp = cu->getChromaFormat();
-
-    uint32_t cxWidth = width   >> m_hChromaShift;
-    uint32_t cxHeight = height >> m_vChromaShift;
-
-    assert(((cxWidth | cxHeight) % 2) == 0);
-
-    if ((yFrac | xFrac) == 0)
-    {
-        primitives.chroma_p2s[csp](refCb, refStride, dstCb, cxWidth, cxHeight);
-        primitives.chroma_p2s[csp](refCr, refStride, dstCr, cxWidth, cxHeight);
-    }
-    else if (yFrac == 0)
-    {
-        primitives.chroma[csp].filter_hps[partEnum](refCb, refStride, dstCb, dstStride, xFrac << (1 - cu->getHorzChromaShift()), 0);
-        primitives.chroma[csp].filter_hps[partEnum](refCr, refStride, dstCr, dstStride, xFrac << (1 - cu->getHorzChromaShift()), 0);
-    }
-    else if (xFrac == 0)
-    {
-        primitives.chroma[csp].filter_vps[partEnum](refCb, refStride, dstCb, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-        primitives.chroma[csp].filter_vps[partEnum](refCr, refStride, dstCr, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-    }
-    else
-    {
-        int extStride = cxWidth;
-        int filterSize = NTAPS_CHROMA;
-        int halfFilterSize = (filterSize >> 1);
-        primitives.chroma[csp].filter_hps[partEnum](refCb, refStride, m_immedVals, extStride, xFrac << (1 - cu->getHorzChromaShift()), 1);
-        primitives.chroma[csp].filter_vss[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCb, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-        primitives.chroma[csp].filter_hps[partEnum](refCr, refStride, m_immedVals, extStride, xFrac << (1 - cu->getHorzChromaShift()), 1);
-        primitives.chroma[csp].filter_vss[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCr, dstStride, yFrac << (1 - cu->getVertChromaShift()));
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPrediction.h Deleted

@@ -1,115 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComPrediction.h
-    \brief    prediction class (header)
-*/
-
-#ifndef X265_TCOMPREDICTION_H
-#define X265_TCOMPREDICTION_H
-
-// Include files
-#include "TComPic.h"
-#include "TComMotionInfo.h"
-#include "TComPattern.h"
-#include "TComTrQuant.h"
-#include "TComWeightPrediction.h"
-#include "shortyuv.h"
-
-namespace x265 {
-// private namespace
-
-struct ReferencePlanes;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// prediction class
-class TComPrediction : public TComWeightPrediction
-{
-protected:
-
-    // references sample for IntraPrediction
-    TComYuv   m_predYuv[2];
-    ShortYuv  m_predShortYuv[2];
-    TComYuv   m_predTempYuv;
-
-    int16_t*  m_immedVals;
-    int       m_hChromaShift;
-    int       m_vChromaShift;
-
-    // motion compensation functions
-    void xPredInterUni(TComDataCU* cu, uint32_t partAddr, int width, int height, int picList, TComYuv* outPredYuv, bool bLuma, bool bChroma);
-    void xPredInterUni(TComDataCU* cu, uint32_t partAddr, int width, int height, int picList, ShortYuv* outPredYuv, bool bLuma, bool bChroma);
-    void xPredInterLumaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, TComYuv *dstPic);
-    void xPredInterLumaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, ShortYuv *dstPic);
-    void xPredInterChromaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, TComYuv *dstPic);
-    void xPredInterChromaBlk(TComDataCU *cu, TComPicYuv *refPic, uint32_t partAddr, MV *mv, int width, int height, ShortYuv *dstPic);
-
-    void xPredInterBi(TComDataCU* cu, uint32_t partAddr, int width, int height, TComYuv* outPredYuv, bool bLuma, bool bChroma);
-
-    void xGetLLSPrediction(TComPattern* pcPattern, int* src0, int srcstride, pixel* dst0, int dststride, uint32_t width, uint32_t height, uint32_t ext0);
-
-    bool xCheckIdenticalMotion(TComDataCU* cu, uint32_t PartAddr);
-
-public:
-
-    pixel*    m_predBuf;
-    pixel*    m_refAbove;
-    pixel*    m_refAboveFlt;
-    pixel*    m_refLeft;
-    pixel*    m_refLeftFlt;
-    int       m_predBufStride;
-    int       m_predBufHeight;
-
-    TComPrediction();
-    virtual ~TComPrediction();
-
-    void initTempBuff(int csp);
-
-    // inter
-    void motionCompensation(TComDataCU* cu, TComYuv* predYuv, int picList = REF_PIC_LIST_X, int partIdx = -1, bool bLuma = true, bool bChroma = true);
-
-    // Angular Intra
-    void predIntraLumaAng(uint32_t dirMode, pixel* pred, intptr_t stride, int width);
-    void predIntraChromaAng(pixel* src, uint32_t dirMode, pixel* pred, intptr_t stride, int width, int height, int chFmt);
-    bool filteringIntraReferenceSamples(uint32_t dirMode, uint32_t width);
-};
-}
-//! \}
-
-#endif // ifndef X265_TCOMPREDICTION_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComRdCost.h Deleted

@@ -1,104 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComRdCost.h
-    \brief    RD cost computation classes (header)
-*/
-
-#ifndef X265_TCOMRDCOST_H
-#define X265_TCOMRDCOST_H
-
-#include "CommonDef.h"
-#include "common.h"
-
-//! \ingroup TLibCommon
-//! \{
-
-namespace x265 {
-// private namespace
-
-class TComPattern;
-
-/// RD cost computation class
-class TComRdCost
-{
-private:
-
-    double    m_lambda2;
-
-    double    m_lambda;
-
-    uint64_t  m_lambdaMotionSSE;  // m_lambda2 w/ 16 bits of fraction
-
-    uint64_t  m_lambdaMotionSAD;  // m_lambda w/ 16 bits of fraction
-
-    uint32_t  m_cbDistortionWeight;
-
-    uint32_t  m_crDistortionWeight;
-
-public:
-
-    void setLambda(double lambda)
-    {
-        m_lambda2         = lambda;
-        m_lambda          = sqrt(m_lambda2);
-        m_lambdaMotionSAD = (uint64_t)floor(65536.0 * m_lambda);
-        m_lambdaMotionSSE = (uint64_t)floor(65536.0 * m_lambda2);
-    }
-
-    void setCbDistortionWeight(double cbDistortionWeight)
-    {
-        m_cbDistortionWeight = (uint32_t)floor(256.0 * cbDistortionWeight);
-    }
-
-    void setCrDistortionWeight(double crDistortionWeight)
-    {
-        m_crDistortionWeight = (uint32_t)floor(256.0 * crDistortionWeight);
-    }
-
-    inline uint64_t calcRdCost(uint32_t distortion, uint32_t bits) { return distortion + ((bits * m_lambdaMotionSSE + 32768) >> 16); }
-
-    inline uint64_t calcRdSADCost(uint32_t sadCost, uint32_t bits) { return sadCost + ((bits * m_lambdaMotionSAD + 32768) >> 16); }
-
-    inline uint32_t getCost(uint32_t bits)                     { return (uint32_t)((bits * m_lambdaMotionSAD + 32768) >> 16); }
-
-    inline uint32_t scaleChromaDistCb(uint32_t dist)           { return ((dist * m_cbDistortionWeight) + 128) >> 8; }
-
-    inline uint32_t scaleChromaDistCr(uint32_t dist)           { return ((dist * m_crDistortionWeight) + 128) >> 8; }
-
-    inline double   getSADLambda() const                       { return m_lambda; }
-};
-}
-//! \}
-
-#endif // ifndef X265_TCOMRDCOST_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComSampleAdaptiveOffset.cpp Deleted

@@ -1,1430 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComSampleAdaptiveOffset.cpp
-    \brief    sample adaptive offset class
-*/
-
-#include "TComSampleAdaptiveOffset.h"
-#include "common.h"
-
-namespace x265 {
-//! \ingroup TLibCommon
-//! \{
-SAOParam::~SAOParam()
-{
-    for (int i = 0; i < 3; i++)
-    {
-        delete [] saoPart[i];
-    }
-}
-
-// ====================================================================================================================
-// Tables
-// ====================================================================================================================
-
-TComSampleAdaptiveOffset::TComSampleAdaptiveOffset()
-{
-    m_clipTable = NULL;
-    m_clipTableBase = NULL;
-    m_chromaClipTable = NULL;
-    m_chromaClipTableBase = NULL;
-    m_offsetBo = NULL;
-    m_chromaOffsetBo = NULL;
-    m_lumaTableBo = NULL;
-    m_chromaTableBo = NULL;
-    m_upBuff1 = NULL;
-    m_upBuff2 = NULL;
-    m_upBufft = NULL;
-
-    m_tmpU1[0] = NULL;
-    m_tmpU1[1] = NULL;
-    m_tmpU1[2] = NULL;
-    m_tmpU2[0] = NULL;
-    m_tmpU2[0] = NULL;
-    m_tmpU2[0] = NULL;
-    m_tmpL1 = NULL;
-    m_tmpL2 = NULL;
-}
-
-TComSampleAdaptiveOffset::~TComSampleAdaptiveOffset()
-{}
-
-const int TComSampleAdaptiveOffset::m_numCulPartsLevel[5] =
-{
-    1, //level 0
-    5, //level 1
-    21, //level 2
-    85, //level 3
-    341, //level 4
-};
-
-const uint32_t TComSampleAdaptiveOffset::m_eoTable[9] =
-{
-    1, //0
-    2, //1
-    0, //2
-    3, //3
-    4, //4
-    0, //5
-    0, //6
-    0, //7
-    0
-};
-
-const int TComSampleAdaptiveOffset::m_numClass[MAX_NUM_SAO_TYPE] =
-{
-    SAO_EO_LEN,
-    SAO_EO_LEN,
-    SAO_EO_LEN,
-    SAO_EO_LEN,
-    SAO_BO_LEN
-};
-
-const uint32_t TComSampleAdaptiveOffset::m_maxDepth = SAO_MAX_DEPTH;
-
-/** convert Level Row Col to Idx
- * \param   level,  row,  col
- */
-int  TComSampleAdaptiveOffset::convertLevelRowCol2Idx(int level, int row, int col) const
-{
-    int idx;
-
-    if (level == 0)
-    {
-        idx = 0;
-    }
-    else if (level == 1)
-    {
-        idx = 1 + row * 2 + col;
-    }
-    else if (level == 2)
-    {
-        idx = 5 + row * 4 + col;
-    }
-    else if (level == 3)
-    {
-        idx = 21 + row * 8 + col;
-    }
-    else // (level == 4)
-    {
-        idx = 85 + row * 16 + col;
-    }
-    return idx;
-}
-
-/** create SampleAdaptiveOffset memory.
- * \param
- */
-void TComSampleAdaptiveOffset::create(uint32_t sourceWidth, uint32_t sourceHeight, uint32_t maxCUWidth, uint32_t maxCUHeight, int csp)
-{
-    m_hChromaShift = CHROMA_H_SHIFT(csp);
-    m_vChromaShift = CHROMA_V_SHIFT(csp);
-
-    m_picWidth  = sourceWidth;
-    m_picHeight = sourceHeight;
-
-    m_maxCUWidth  = maxCUWidth;
-    m_maxCUHeight = maxCUHeight;
-
-    m_numCuInWidth  = m_picWidth / m_maxCUWidth;
-    m_numCuInWidth += (m_picWidth % m_maxCUWidth) ? 1 : 0;
-
-    m_numCuInHeight  = m_picHeight / m_maxCUHeight;
-    m_numCuInHeight += (m_picHeight % m_maxCUHeight) ? 1 : 0;
-
-    int maxSplitLevelHeight = (int)(logf((float)m_numCuInHeight) / logf(2.0));
-    int maxSplitLevelWidth  = (int)(logf((float)m_numCuInWidth) / logf(2.0));
-
-    m_maxSplitLevel = (maxSplitLevelHeight < maxSplitLevelWidth) ? (maxSplitLevelHeight) : (maxSplitLevelWidth);
-    m_maxSplitLevel = (m_maxSplitLevel < m_maxDepth) ? (m_maxSplitLevel) : (m_maxDepth);
-
-    /* various structures are overloaded to store per component data.
-     * m_iNumTotalParts must allow for sufficient storage in any allocated arrays */
-    m_numTotalParts  = X265_MAX(3, m_numCulPartsLevel[m_maxSplitLevel]);
-
-    uint32_t pixelRangeY = 1 << X265_DEPTH;
-    uint32_t boRangeShiftY = X265_DEPTH - SAO_BO_BITS;
-
-    m_lumaTableBo = X265_MALLOC(pixel, pixelRangeY);
-    for (int k2 = 0; k2 < pixelRangeY; k2++)
-    {
-        m_lumaTableBo[k2] = 1 + (k2 >> boRangeShiftY);
-    }
-
-    uint32_t pixelRangeC = 1 << X265_DEPTH;
-    uint32_t boRangeShiftC = X265_DEPTH - SAO_BO_BITS;
-
-    m_chromaTableBo = X265_MALLOC(pixel, pixelRangeC);
-    for (int k2 = 0; k2 < pixelRangeC; k2++)
-    {
-        m_chromaTableBo[k2] = 1 + (k2 >> boRangeShiftC);
-    }
-
-    m_upBuff1 = X265_MALLOC(int, m_picWidth + 2);
-    m_upBuff2 = X265_MALLOC(int, m_picWidth + 2);
-    m_upBufft = X265_MALLOC(int, m_picWidth + 2);
-
-    m_upBuff1++;
-    m_upBuff2++;
-    m_upBufft++;
-    int16_t i;
-
-    uint32_t maxY  = (1 << X265_DEPTH) - 1;
-    uint32_t minY  = 0;
-
-    int rangeExt = maxY >> 1;
-
-    m_clipTableBase = X265_MALLOC(pixel, maxY + 2 * rangeExt);
-    m_offsetBo      = X265_MALLOC(int, maxY + 2 * rangeExt);
-
-    for (i = 0; i < (minY + rangeExt); i++)
-    {
-        m_clipTableBase[i] = minY;
-    }
-
-    for (i = minY + rangeExt; i < (maxY +  rangeExt); i++)
-    {
-        m_clipTableBase[i] = i - rangeExt;
-    }
-
-    for (i = maxY + rangeExt; i < (maxY + 2 * rangeExt); i++)
-    {
-        m_clipTableBase[i] = maxY;
-    }
-
-    m_clipTable = &(m_clipTableBase[rangeExt]);
-
-    uint32_t maxC = (1 << X265_DEPTH) - 1;
-    uint32_t minC = 0;
-
-    int rangeExtC = maxC >> 1;
-
-    m_chromaClipTableBase = X265_MALLOC(pixel, maxC + 2 * rangeExtC);
-    m_chromaOffsetBo      = X265_MALLOC(int, maxC + 2 * rangeExtC);
-
-    for (i = 0; i < (minC + rangeExtC); i++)
-    {
-        m_chromaClipTableBase[i] = minC;
-    }
-
-    for (i = minC + rangeExtC; i < (maxC +  rangeExtC); i++)
-    {
-        m_chromaClipTableBase[i] = i - rangeExtC;
-    }
-
-    for (i = maxC + rangeExtC; i < (maxC + 2 * rangeExtC); i++)
-    {
-        m_chromaClipTableBase[i] = maxC;
-    }
-
-    m_chromaClipTable = &(m_chromaClipTableBase[rangeExtC]);
-
-    m_tmpL1 = X265_MALLOC(pixel, m_maxCUHeight + 1);
-    m_tmpL2 = X265_MALLOC(pixel, m_maxCUHeight + 1);
-
-    m_tmpU1[0] = X265_MALLOC(pixel, m_picWidth);
-    m_tmpU1[1] = X265_MALLOC(pixel, m_picWidth);
-    m_tmpU1[2] = X265_MALLOC(pixel, m_picWidth);
-
-    m_tmpU2[0] = X265_MALLOC(pixel, m_picWidth);
-    m_tmpU2[1] = X265_MALLOC(pixel, m_picWidth);
-    m_tmpU2[2] = X265_MALLOC(pixel, m_picWidth);
-}
-
-/** destroy SampleAdaptiveOffset memory.
- * \param
- */
-void TComSampleAdaptiveOffset::destroy()
-{
-    X265_FREE(m_clipTableBase);
-    m_clipTableBase = NULL;
-    X265_FREE(m_offsetBo);
-    m_offsetBo = NULL;
-    X265_FREE(m_lumaTableBo);
-    m_lumaTableBo = NULL;
-    X265_FREE(m_chromaClipTableBase);
-    m_chromaClipTableBase = NULL;
-    X265_FREE(m_chromaOffsetBo);
-    m_chromaOffsetBo = NULL;
-    X265_FREE(m_chromaTableBo);
-    m_chromaTableBo = NULL;
-
-    m_upBuff1--;
-    X265_FREE(m_upBuff1);
-    m_upBuff1 = NULL;
-
-    m_upBuff2--;
-    X265_FREE(m_upBuff2);
-    m_upBuff2 = NULL;
-
-    m_upBufft--;
-    X265_FREE(m_upBufft);
-    m_upBufft = NULL;
-
-    X265_FREE(m_tmpL1);
-    m_tmpL1 = NULL;
-    X265_FREE(m_tmpL2);
-    m_tmpL2 = NULL;
-    X265_FREE(m_tmpU1[0]);
-
-    m_tmpU1[0] = NULL;
-    X265_FREE(m_tmpU1[1]);
-    m_tmpU1[1] = NULL;
-    X265_FREE(m_tmpU1[2]);
-    m_tmpU1[2] = NULL;
-
-    X265_FREE(m_tmpU2[0]);
-    m_tmpU2[0] = NULL;
-    X265_FREE(m_tmpU2[1]);
-    m_tmpU2[1] = NULL;
-    X265_FREE(m_tmpU2[2]);
-    m_tmpU2[2] = NULL;
-
-}
-
-/** allocate memory for SAO parameters
- * \param    *saoParam
- */
-void TComSampleAdaptiveOffset::allocSaoParam(SAOParam *saoParam) const
-{
-    saoParam->maxSplitLevel = m_maxSplitLevel;
-    saoParam->saoPart[0] = new SAOQTPart[m_numCulPartsLevel[saoParam->maxSplitLevel]];
-    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 0);
-    saoParam->saoPart[1] = new SAOQTPart[m_numCulPartsLevel[saoParam->maxSplitLevel]];
-    saoParam->saoPart[2] = new SAOQTPart[m_numCulPartsLevel[saoParam->maxSplitLevel]];
-    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 1);
-    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 2);
-    saoParam->numCuInWidth  = m_numCuInWidth;
-    saoParam->numCuInHeight = m_numCuInHeight;
-    saoParam->saoLcuParam[0] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
-    saoParam->saoLcuParam[1] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
-    saoParam->saoLcuParam[2] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
-}
-
-/** initialize SAO parameters
- * \param    *saoParam,  iPartLevel,  iPartRow,  iPartCol,  iParentPartIdx,  StartCUX,  EndCUX,  StartCUY,  EndCUY,  yCbCr
- */
-void TComSampleAdaptiveOffset::initSAOParam(SAOParam *saoParam, int partLevel, int partRow, int partCol, int parentPartIdx, int startCUX, int endCUX, int startCUY, int endCUY, int yCbCr) const
-{
-    int j;
-    int partIdx = convertLevelRowCol2Idx(partLevel, partRow, partCol);
-
-    SAOQTPart* saoPart;
-
-    saoPart = &(saoParam->saoPart[yCbCr][partIdx]);
-
-    saoPart->partIdx   = partIdx;
-    saoPart->partLevel = partLevel;
-    saoPart->partRow   = partRow;
-    saoPart->partCol   = partCol;
-
-    saoPart->startCUX  = startCUX;
-    saoPart->endCUX    = endCUX;
-    saoPart->startCUY  = startCUY;
-    saoPart->endCUY    = endCUY;
-
-    saoPart->upPartIdx = parentPartIdx;
-    saoPart->bestType  = -1;
-    saoPart->length    =  0;
-
-    saoPart->subTypeIdx = 0;
-
-    for (j = 0; j < MAX_NUM_SAO_OFFSETS; j++)
-    {
-        saoPart->offset[j] = 0;
-    }
-
-    if (saoPart->partLevel != m_maxSplitLevel)
-    {
-        int downLevel    = (partLevel + 1);
-        int downRowStart = (partRow << 1);
-        int downColStart = (partCol << 1);
-
-        int downRowIdx, downColIdx;
-        int numCUWidth,  numCUHeight;
-        int numCULeft;
-        int numCUTop;
-
-        int downStartCUX, downStartCUY;
-        int downEndCUX, downEndCUY;
-
-        numCUWidth  = endCUX - startCUX + 1;
-        numCUHeight = endCUY - startCUY + 1;
-        numCULeft   = (numCUWidth  >> 1);
-        numCUTop    = (numCUHeight >> 1);
-
-        downStartCUX = startCUX;
-        downEndCUX  = downStartCUX + numCULeft - 1;
-        downStartCUY = startCUY;
-        downEndCUY  = downStartCUY + numCUTop  - 1;
-        downRowIdx = downRowStart + 0;
-        downColIdx = downColStart + 0;
-
-        saoPart->downPartsIdx[0] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
-
-        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, yCbCr);
-
-        downStartCUX = startCUX + numCULeft;
-        downEndCUX   = endCUX;
-        downStartCUY = startCUY;
-        downEndCUY   = downStartCUY + numCUTop - 1;
-        downRowIdx  = downRowStart + 0;
-        downColIdx  = downColStart + 1;
-
-        saoPart->downPartsIdx[1] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
-
-        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx,  downStartCUX, downEndCUX, downStartCUY, downEndCUY, yCbCr);
-
-        downStartCUX = startCUX;
-        downEndCUX   = downStartCUX + numCULeft - 1;
-        downStartCUY = startCUY + numCUTop;
-        downEndCUY   = endCUY;
-        downRowIdx  = downRowStart + 1;
-        downColIdx  = downColStart + 0;
-
-        saoPart->downPartsIdx[2] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
-
-        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, yCbCr);
-
-        downStartCUX = startCUX + numCULeft;
-        downEndCUX   = endCUX;
-        downStartCUY = startCUY + numCUTop;
-        downEndCUY   = endCUY;
-        downRowIdx  = downRowStart + 1;
-        downColIdx  = downColStart + 1;
-
-        saoPart->downPartsIdx[3] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
-
-        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, yCbCr);
-    }
-    else
-    {
-        saoPart->downPartsIdx[0] = saoPart->downPartsIdx[1] = saoPart->downPartsIdx[2] = saoPart->downPartsIdx[3] = -1;
-    }
-}
-
-/** free memory of SAO parameters
- * \param   saoParam
- */
-void TComSampleAdaptiveOffset::freeSaoParam(SAOParam *saoParam)
-{
-    delete [] saoParam->saoPart[0];
-    delete [] saoParam->saoPart[1];
-    delete [] saoParam->saoPart[2];
-    saoParam->saoPart[0] = 0;
-    saoParam->saoPart[1] = 0;
-    saoParam->saoPart[2] = 0;
-    delete [] saoParam->saoLcuParam[0];
-    saoParam->saoLcuParam[0] = NULL;
-    delete [] saoParam->saoLcuParam[1];
-    saoParam->saoLcuParam[1] = NULL;
-    delete [] saoParam->saoLcuParam[2];
-    saoParam->saoLcuParam[2] = NULL;
-}
-
-/** reset SAO parameters
- * \param   saoParam
- */
-void TComSampleAdaptiveOffset::resetSAOParam(SAOParam *saoParam)
-{
-    int numComponet = 3;
-
-    for (int c = 0; c < numComponet; c++)
-    {
-        if (c < 2)
-        {
-            saoParam->bSaoFlag[c] = 0;
-        }
-        for (int i = 0; i < m_numCulPartsLevel[m_maxSplitLevel]; i++)
-        {
-            saoParam->saoPart[c][i].bestType     = -1;
-            saoParam->saoPart[c][i].length       =  0;
-            saoParam->saoPart[c][i].bSplit       = false;
-            saoParam->saoPart[c][i].bProcessed   = false;
-            saoParam->saoPart[c][i].minCost      = MAX_DOUBLE;
-            saoParam->saoPart[c][i].minDist      = MAX_INT;
-            saoParam->saoPart[c][i].minRate      = MAX_INT;
-            saoParam->saoPart[c][i].subTypeIdx   = 0;
-            for (int j = 0; j < MAX_NUM_SAO_OFFSETS; j++)
-            {
-                saoParam->saoPart[c][i].offset[j] = 0;
-                saoParam->saoPart[c][i].offset[j] = 0;
-                saoParam->saoPart[c][i].offset[j] = 0;
-            }
-        }
-
-        saoParam->oneUnitFlag[0] = 0;
-        saoParam->oneUnitFlag[1] = 0;
-        saoParam->oneUnitFlag[2] = 0;
-        resetLcuPart(saoParam->saoLcuParam[0]);
-        resetLcuPart(saoParam->saoLcuParam[1]);
-        resetLcuPart(saoParam->saoLcuParam[2]);
-    }
-}
-
-/** get the sign of input variable
- * \param   x
- */
-inline int xSign(int x)
-{
-    return (x >> 31) | ((int)((((uint32_t)-x)) >> 31));
-}
-
-/** initialize variables for SAO process
- * \param  pic picture data pointer
- */
-void TComSampleAdaptiveOffset::createPicSaoInfo(TComPic* pic)
-{
-    m_pic = pic;
-}
-
-/** sample adaptive offset process for one LCU crossing LCU boundary
- * \param   addr, iSaoType, yCbCr
- */
-void TComSampleAdaptiveOffset::processSaoCu(int addr, int saoType, int yCbCr)
-{
-    int x, y;
-    TComDataCU *tmpCu = m_pic->getCU(addr);
-    pixel* rec;
-    int  stride;
-    int  lcuWidth  = m_maxCUWidth;
-    int  lcuHeight = m_maxCUHeight;
-    uint32_t lpelx     = tmpCu->getCUPelX();
-    uint32_t tpely     = tmpCu->getCUPelY();
-    uint32_t rpelx;
-    uint32_t bpely;
-    int  edgeType;
-    int  signDown;
-    int  signDown1;
-    int  signDown2;
-    int picWidthTmp;
-    int picHeightTmp;
-    int startX;
-    int startY;
-    int endX;
-    int endY;
-    int isChroma = (yCbCr != 0) ? 1 : 0;
-    int shift;
-    int cuHeightTmp;
-    pixel* tmpLSwap;
-    pixel* tmpL;
-    pixel* tmpU;
-    pixel* clipTbl = NULL;
-    int32_t *offsetBo = NULL;
-    int32_t *tmp_swap;
-
-    picWidthTmp  = (isChroma == 0) ? m_picWidth  : m_picWidth  >> m_hChromaShift;
-    picHeightTmp = (isChroma == 0) ? m_picHeight : m_picHeight >> m_vChromaShift;
-    lcuWidth     = (isChroma == 0) ? lcuWidth    : lcuWidth    >> m_hChromaShift;
-    lcuHeight    = (isChroma == 0) ? lcuHeight   : lcuHeight   >> m_vChromaShift;
-    lpelx        = (isChroma == 0) ? lpelx       : lpelx       >> m_hChromaShift;
-    tpely        = (isChroma == 0) ? tpely       : tpely       >> m_vChromaShift;
-
-    rpelx        = lpelx + lcuWidth;
-    bpely        = tpely + lcuHeight;
-    rpelx        = rpelx > picWidthTmp  ? picWidthTmp  : rpelx;
-    bpely        = bpely > picHeightTmp ? picHeightTmp : bpely;
-    lcuWidth     = rpelx - lpelx;
-    lcuHeight    = bpely - tpely;
-
-    if (tmpCu->getPic() == 0)
-    {
-        return;
-    }
-    if (yCbCr == 0)
-    {
-        rec    = m_pic->getPicYuvRec()->getLumaAddr(addr);
-        stride = m_pic->getStride();
-    }
-    else if (yCbCr == 1)
-    {
-        rec    = m_pic->getPicYuvRec()->getCbAddr(addr);
-        stride = m_pic->getCStride();
-    }
-    else
-    {
-        rec    = m_pic->getPicYuvRec()->getCrAddr(addr);
-        stride = m_pic->getCStride();
-    }
-
-//   if (iSaoType!=SAO_BO_0 || iSaoType!=SAO_BO_1)
-    {
-        cuHeightTmp  = (isChroma == 0) ? m_maxCUHeight  : (m_maxCUHeight  >> m_vChromaShift);
-        shift = (isChroma == 0) ? (m_maxCUWidth - 1) : ((m_maxCUWidth >> m_hChromaShift) - 1);
-        for (int i = 0; i < cuHeightTmp + 1; i++)
-        {
-            m_tmpL2[i] = rec[shift];
-            rec += stride;
-        }
-
-        rec -= (stride * (cuHeightTmp + 1));
-
-        tmpL = m_tmpL1;
-        tmpU = &(m_tmpU1[yCbCr][lpelx]);
-    }
-
-    clipTbl = (yCbCr == 0) ? m_clipTable : m_chromaClipTable;
-    offsetBo = (yCbCr == 0) ? m_offsetBo : m_chromaOffsetBo;
-
-    switch (saoType)
-    {
-    case SAO_EO_0: // dir: -
-    {
-      pixel firstPxl = 0, lastPxl = 0;
-      startX = (lpelx == 0) ? 1 : 0;
-      endX   = (rpelx == picWidthTmp) ? lcuWidth-1 : lcuWidth;
-      if (lcuWidth % 16)
-      {
-          int8_t signRight;
-          for (y = 0; y < lcuHeight; y++)
-          {
-              int8_t signLeft = xSign(rec[startX] - tmpL[y]);
-              for (x = startX; x < endX; x++)
-              {
-                  signRight = xSign(rec[x] - rec[x+1]);
-                  edgeType = signRight + signLeft + 2;
-                  signLeft  = -signRight;
-
-                  rec[x] =  Clip3(0, (1 << X265_DEPTH) - 1, rec[x] + m_offsetEo[edgeType]);
-              }
-              rec += stride;
-          }
-      }
-      else
-      {
-          for (y = 0; y < lcuHeight; y++)
-          {
-              int8_t signLeft = xSign(rec[startX] - tmpL[y]);
-
-              if (lpelx == 0)
-              {
-                  firstPxl = rec[0];
-              }
-
-              if (rpelx == picWidthTmp)
-              {
-                  lastPxl = rec[lcuWidth - 1];
-              }
-
-              primitives.saoCuOrgE0(rec, m_offsetEo, lcuWidth, signLeft);
-
-              if (lpelx == 0)
-              {
-                  rec[0] = firstPxl;
-              }
-
-              if (rpelx == picWidthTmp)
-              {
-                  rec[lcuWidth - 1] = lastPxl;
-              }
-              rec += stride;
-          }
-      }
-        break;
-    }
-    case SAO_EO_1: // dir: |
-    {
-        startY = (tpely == 0) ? 1 : 0;
-        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-        if (tpely == 0)
-        {
-            rec += stride;
-        }
-        for (x = 0; x < lcuWidth; x++)
-        {
-            m_upBuff1[x] = xSign(rec[x] - tmpU[x]);
-        }
-
-        for (y = startY; y < endY; y++)
-        {
-            for (x = 0; x < lcuWidth; x++)
-            {
-                signDown  = xSign(rec[x] - rec[x + stride]);
-                edgeType = signDown + m_upBuff1[x] + 2;
-                m_upBuff1[x] = -signDown;
-
-                rec[x] = clipTbl[rec[x] + m_offsetEo[edgeType]];
-            }
-
-            rec += stride;
-        }
-
-        break;
-    }
-    case SAO_EO_2: // dir: 135
-    {
-        startX = (lpelx == 0)            ? 1 : 0;
-        endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
-
-        startY = (tpely == 0) ?             1 : 0;
-        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-
-        if (tpely == 0)
-        {
-            rec += stride;
-        }
-
-        for (x = startX; x < endX; x++)
-        {
-            m_upBuff1[x] = xSign(rec[x] - tmpU[x - 1]);
-        }
-
-        for (y = startY; y < endY; y++)
-        {
-            signDown2 = xSign(rec[stride + startX] - tmpL[y]);
-            for (x = startX; x < endX; x++)
-            {
-                signDown1      =  xSign(rec[x] - rec[x + stride + 1]);
-                edgeType      =  signDown1 + m_upBuff1[x] + 2;
-                m_upBufft[x + 1] = -signDown1;
-                rec[x] = clipTbl[rec[x] + m_offsetEo[edgeType]];
-            }
-
-            m_upBufft[startX] = signDown2;
-
-            tmp_swap  = m_upBuff1;
-            m_upBuff1 = m_upBufft;
-            m_upBufft = tmp_swap;
-
-            rec += stride;
-        }
-
-        break;
-    }
-    case SAO_EO_3: // dir: 45
-    {
-        startX = (lpelx == 0) ? 1 : 0;
-        endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
-
-        startY = (tpely == 0) ? 1 : 0;
-        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-
-        if (startY == 1)
-        {
-            rec += stride;
-        }
-
-        for (x = startX - 1; x < endX; x++)
-        {
-            m_upBuff1[x] = xSign(rec[x] - tmpU[x + 1]);
-        }
-
-        for (y = startY; y < endY; y++)
-        {
-            x = startX;
-            signDown1      =  xSign(rec[x] - tmpL[y + 1]);
-            edgeType      =  signDown1 + m_upBuff1[x] + 2;
-            m_upBuff1[x - 1] = -signDown1;
-            rec[x] = clipTbl[rec[x] + m_offsetEo[edgeType]];
-            for (x = startX + 1; x < endX; x++)
-            {
-                signDown1      =  xSign(rec[x] - rec[x + stride - 1]);
-                edgeType      =  signDown1 + m_upBuff1[x] + 2;
-                m_upBuff1[x - 1] = -signDown1;
-                rec[x] = clipTbl[rec[x] + m_offsetEo[edgeType]];
-            }
-
-            m_upBuff1[endX - 1] = xSign(rec[endX - 1 + stride] - rec[endX]);
-
-            rec += stride;
-        }
-
-        break;
-    }
-    case SAO_BO:
-    {
-        for (y = 0; y < lcuHeight; y++)
-        {
-            for (x = 0; x < lcuWidth; x++)
-            {
-                rec[x] = offsetBo[rec[x]];
-            }
-
-            rec += stride;
-        }
-
-        break;
-    }
-    default: break;
-    }
-
-//   if (iSaoType!=SAO_BO_0 || iSaoType!=SAO_BO_1)
-    {
-        tmpLSwap = m_tmpL1;
-        m_tmpL1  = m_tmpL2;
-        m_tmpL2  = tmpLSwap;
-    }
-}
-
-/** Sample adaptive offset process
- * \param pic, saoParam
- */
-void TComSampleAdaptiveOffset::SAOProcess(SAOParam* saoParam)
-{
-    {
-        m_saoBitIncreaseY = X265_MAX(X265_DEPTH - 10, 0);
-        m_saoBitIncreaseC = X265_MAX(X265_DEPTH - 10, 0);
-
-        if (m_saoLcuBasedOptimization)
-        {
-            saoParam->oneUnitFlag[0] = 0;
-            saoParam->oneUnitFlag[1] = 0;
-            saoParam->oneUnitFlag[2] = 0;
-        }
-        int iY  = 0;
-        {
-            processSaoUnitAll(saoParam->saoLcuParam[iY], saoParam->oneUnitFlag[iY], iY);
-        }
-        {
-            processSaoUnitAll(saoParam->saoLcuParam[1], saoParam->oneUnitFlag[1], 1); //Cb
-            processSaoUnitAll(saoParam->saoLcuParam[2], saoParam->oneUnitFlag[2], 2); //Cr
-        }
-        m_pic = NULL;
-    }
-}
-
-pixel* TComSampleAdaptiveOffset::getPicYuvAddr(TComPicYuv* picYuv, int yCbCr, int addr)
-{
-    switch (yCbCr)
-    {
-    case 0:
-        return picYuv->getLumaAddr(addr);
-        break;
-    case 1:
-        return picYuv->getCbAddr(addr);
-        break;
-    case 2:
-        return picYuv->getCrAddr(addr);
-        break;
-    default:
-        return NULL;
-        break;
-    }
-}
-
-/** Process SAO all units
- * \param saoLcuParam SAO LCU parameters
- * \param oneUnitFlag one unit flag
- * \param yCbCr color componet index
- */
-void TComSampleAdaptiveOffset::processSaoUnitAll(SaoLcuParam* saoLcuParam, bool oneUnitFlag, int yCbCr)
-{
-    pixel *rec;
-    int picWidthTmp;
-
-    if (yCbCr == 0)
-    {
-        rec        = m_pic->getPicYuvRec()->getLumaAddr();
-        picWidthTmp = m_picWidth;
-    }
-    else if (yCbCr == 1)
-    {
-        rec        = m_pic->getPicYuvRec()->getCbAddr();
-        picWidthTmp = m_picWidth >> m_hChromaShift;
-    }
-    else
-    {
-        rec        = m_pic->getPicYuvRec()->getCrAddr();
-        picWidthTmp = m_picWidth >> m_hChromaShift;
-    }
-
-    memcpy(m_tmpU1[yCbCr], rec, sizeof(pixel) * picWidthTmp);
-
-    int  i;
-    uint32_t edgeType;
-    pixel* lumaTable = NULL;
-    pixel* clipTable = NULL;
-    int32_t* offsetBo = NULL;
-    int  typeIdx;
-
-    int offset[LUMA_GROUP_NUM + 1];
-    int idxX;
-    int idxY;
-    int addr;
-    int frameWidthInCU = m_pic->getFrameWidthInCU();
-    int frameHeightInCU = m_pic->getFrameHeightInCU();
-    int stride;
-    pixel *tmpUSwap;
-    int sChroma = (yCbCr == 0) ? 0 : 1;
-    bool mergeLeftFlag;
-    int saoBitIncrease = (yCbCr == 0) ? m_saoBitIncreaseY : m_saoBitIncreaseC;
-
-    offsetBo = (yCbCr == 0) ? m_offsetBo : m_chromaOffsetBo;
-
-    offset[0] = 0;
-    for (idxY = 0; idxY < frameHeightInCU; idxY++)
-    {
-        addr = idxY * frameWidthInCU;
-        if (yCbCr == 0)
-        {
-            rec  = m_pic->getPicYuvRec()->getLumaAddr(addr);
-            stride = m_pic->getStride();
-            picWidthTmp = m_picWidth;
-        }
-        else if (yCbCr == 1)
-        {
-            rec  = m_pic->getPicYuvRec()->getCbAddr(addr);
-            stride = m_pic->getCStride();
-            picWidthTmp = m_picWidth >> m_hChromaShift;
-        }
-        else
-        {
-            rec  = m_pic->getPicYuvRec()->getCrAddr(addr);
-            stride = m_pic->getCStride();
-            picWidthTmp = m_picWidth >> m_hChromaShift;
-        }
-        uint32_t cuHeightTmp  = (sChroma == 0) ? m_maxCUHeight  : (m_maxCUHeight  >> m_vChromaShift);
-        for (i = 0; i < cuHeightTmp + 1; i++)
-        {
-            m_tmpL1[i] = rec[0];
-            rec += stride;
-        }
-
-        rec -= (stride << 1);
-
-        memcpy(m_tmpU2[yCbCr], rec, sizeof(pixel) * picWidthTmp);
-
-        for (idxX = 0; idxX < frameWidthInCU; idxX++)
-        {
-            addr = idxY * frameWidthInCU + idxX;
-
-            if (oneUnitFlag)
-            {
-                typeIdx = saoLcuParam[0].typeIdx;
-                mergeLeftFlag = (addr == 0) ? 0 : 1;
-            }
-            else
-            {
-                typeIdx = saoLcuParam[addr].typeIdx;
-                mergeLeftFlag = saoLcuParam[addr].mergeLeftFlag;
-            }
-            if (typeIdx >= 0)
-            {
-                if (!mergeLeftFlag)
-                {
-                    if (typeIdx == SAO_BO)
-                    {
-                        for (i = 0; i < SAO_MAX_BO_CLASSES + 1; i++)
-                        {
-                            offset[i] = 0;
-                        }
-
-                        for (i = 0; i < saoLcuParam[addr].length; i++)
-                        {
-                            offset[(saoLcuParam[addr].subTypeIdx + i) % SAO_MAX_BO_CLASSES  + 1] = saoLcuParam[addr].offset[i] << saoBitIncrease;
-                        }
-
-                        lumaTable = (yCbCr == 0) ? m_lumaTableBo : m_chromaTableBo;
-                        clipTable = (yCbCr == 0) ? m_clipTable : m_chromaClipTable;
-
-                        for (i = 0; i < (1 << X265_DEPTH); i++)
-                        {
-                            offsetBo[i] = clipTable[i + offset[lumaTable[i]]];
-                        }
-                    }
-                    if (typeIdx == SAO_EO_0 || typeIdx == SAO_EO_1 || typeIdx == SAO_EO_2 || typeIdx == SAO_EO_3)
-                    {
-                        for (i = 0; i < saoLcuParam[addr].length; i++)
-                        {
-                            offset[i + 1] = saoLcuParam[addr].offset[i] << saoBitIncrease;
-                        }
-
-                        for (edgeType = 0; edgeType < 6; edgeType++)
-                        {
-                            m_offsetEo[edgeType] = offset[m_eoTable[edgeType]];
-                        }
-                    }
-                }
-                processSaoCu(addr, typeIdx, yCbCr);
-            }
-            else
-            {
-                if (idxX != (frameWidthInCU - 1))
-                {
-                    if (yCbCr == 0)
-                    {
-                        rec  = m_pic->getPicYuvRec()->getLumaAddr(addr);
-                        stride = m_pic->getStride();
-                    }
-                    else if (yCbCr == 1)
-                    {
-                        rec  = m_pic->getPicYuvRec()->getCbAddr(addr);
-                        stride = m_pic->getCStride();
-                    }
-                    else
-                    {
-                        rec  = m_pic->getPicYuvRec()->getCrAddr(addr);
-                        stride = m_pic->getCStride();
-                    }
-
-                    int widthShift = (sChroma == 0) ? m_maxCUWidth : (m_maxCUWidth >> m_hChromaShift);
-                    for (i = 0; i < cuHeightTmp + 1; i++)
-                    {
-                        m_tmpL1[i] = rec[widthShift - 1];
-                        rec += stride;
-                    }
-                }
-            }
-        }
-
-        tmpUSwap       = m_tmpU1[yCbCr];
-        m_tmpU1[yCbCr] = m_tmpU2[yCbCr];
-        m_tmpU2[yCbCr] = tmpUSwap;
-    }
-}
-
-/** Process SAO all units
- * \param saoLcuParam SAO LCU parameters
- * \param oneUnitFlag one unit flag
- * \param yCbCr color componet index
- */
-void TComSampleAdaptiveOffset::processSaoUnitRow(SaoLcuParam* saoLcuParam, int idxY, int yCbCr)
-{
-    pixel *rec;
-    int picWidthTmp;
-
-    if (yCbCr == 0)
-    {
-        rec        = m_pic->getPicYuvRec()->getLumaAddr();
-        picWidthTmp = m_picWidth;
-    }
-    else if (yCbCr == 1)
-    {
-        rec        = m_pic->getPicYuvRec()->getCbAddr();
-        picWidthTmp = m_picWidth >> m_hChromaShift;
-    }
-    else
-    {
-        rec        = m_pic->getPicYuvRec()->getCrAddr();
-        picWidthTmp = m_picWidth >> m_hChromaShift;
-    }
-
-    if (idxY == 0)
-        memcpy(m_tmpU1[yCbCr], rec, sizeof(Pel) * picWidthTmp);
-
-    int  i;
-    uint32_t edgeType;
-    pixel* lumaTable = NULL;
-    pixel* clipTable = NULL;
-    int32_t* offsetBo = NULL;
-    int  typeIdx;
-
-    int offset[LUMA_GROUP_NUM + 1];
-    int idxX;
-    int addr;
-    int frameWidthInCU = m_pic->getFrameWidthInCU();
-    int stride;
-    pixel *tmpUSwap;
-    int sChroma = (yCbCr == 0) ? 0 : 1;
-    bool mergeLeftFlag;
-    int saoBitIncrease = (yCbCr == 0) ? m_saoBitIncreaseY : m_saoBitIncreaseC;
-
-    offsetBo = (yCbCr == 0) ? m_offsetBo : m_chromaOffsetBo;
-
-    offset[0] = 0;
-    {
-        addr = idxY * frameWidthInCU;
-        if (yCbCr == 0)
-        {
-            rec  = m_pic->getPicYuvRec()->getLumaAddr(addr);
-            stride = m_pic->getStride();
-            picWidthTmp = m_picWidth;
-        }
-        else if (yCbCr == 1)
-        {
-            rec  = m_pic->getPicYuvRec()->getCbAddr(addr);
-            stride = m_pic->getCStride();
-            picWidthTmp = m_picWidth >> m_hChromaShift;
-        }
-        else
-        {
-            rec  = m_pic->getPicYuvRec()->getCrAddr(addr);
-            stride = m_pic->getCStride();
-            picWidthTmp = m_picWidth >> m_hChromaShift;
-        }
-        uint32_t maxCUHeight  = (sChroma == 0) ? m_maxCUHeight : (m_maxCUHeight >> m_vChromaShift);
-        for (i = 0; i < maxCUHeight + 1; i++)
-        {
-            m_tmpL1[i] = rec[0];
-            rec += stride;
-        }
-
-        rec -= (stride << 1);
-
-        memcpy(m_tmpU2[yCbCr], rec, sizeof(pixel) * picWidthTmp);
-
-        for (idxX = 0; idxX < frameWidthInCU; idxX++)
-        {
-            addr = idxY * frameWidthInCU + idxX;
-
-            typeIdx = saoLcuParam[addr].typeIdx;
-            mergeLeftFlag = saoLcuParam[addr].mergeLeftFlag;
-
-            if (typeIdx >= 0)
-            {
-                if (!mergeLeftFlag)
-                {
-                    if (typeIdx == SAO_BO)
-                    {
-                        for (i = 0; i < SAO_MAX_BO_CLASSES + 1; i++)
-                        {
-                            offset[i] = 0;
-                        }
-
-                        for (i = 0; i < saoLcuParam[addr].length; i++)
-                        {
-                            offset[(saoLcuParam[addr].subTypeIdx + i) % SAO_MAX_BO_CLASSES  + 1] = saoLcuParam[addr].offset[i] << saoBitIncrease;
-                        }
-
-                        lumaTable = (yCbCr == 0) ? m_lumaTableBo : m_chromaTableBo;
-                        clipTable = (yCbCr == 0) ? m_clipTable : m_chromaClipTable;
-
-                        for (i = 0; i < (1 << X265_DEPTH); i++)
-                        {
-                            offsetBo[i] = clipTable[i + offset[lumaTable[i]]];
-                        }
-                    }
-                    if (typeIdx == SAO_EO_0 || typeIdx == SAO_EO_1 || typeIdx == SAO_EO_2 || typeIdx == SAO_EO_3)
-                    {
-                        for (i = 0; i < saoLcuParam[addr].length; i++)
-                        {
-                            offset[i + 1] = saoLcuParam[addr].offset[i] << saoBitIncrease;
-                        }
-
-                        for (edgeType = 0; edgeType < 6; edgeType++)
-                        {
-                            m_offsetEo[edgeType] = offset[m_eoTable[edgeType]];
-                        }
-                    }
-                }
-                processSaoCu(addr, typeIdx, yCbCr);
-            }
-            else
-            {
-                if (idxX != (frameWidthInCU - 1))
-                {
-                    if (yCbCr == 0)
-                    {
-                        rec  = m_pic->getPicYuvRec()->getLumaAddr(addr);
-                        stride = m_pic->getStride();
-                    }
-                    else if (yCbCr == 1)
-                    {
-                        rec  = m_pic->getPicYuvRec()->getCbAddr(addr);
-                        stride = m_pic->getCStride();
-                    }
-                    else
-                    {
-                        rec  = m_pic->getPicYuvRec()->getCrAddr(addr);
-                        stride = m_pic->getCStride();
-                    }
-
-                    int widthShift = (sChroma == 0) ? m_maxCUWidth : (m_maxCUWidth >> m_hChromaShift);
-                    for (i = 0; i < maxCUHeight + 1; i++)
-                    {
-                        m_tmpL1[i] = rec[widthShift - 1];
-                        rec += stride;
-                    }
-                }
-            }
-        }
-
-        tmpUSwap       = m_tmpU1[yCbCr];
-        m_tmpU1[yCbCr] = m_tmpU2[yCbCr];
-        m_tmpU2[yCbCr] = tmpUSwap;
-    }
-}
-
-void TComSampleAdaptiveOffset::resetLcuPart(SaoLcuParam* saoLcuParam)
-{
-    int i, j;
-
-    for (i = 0; i < m_numCuInWidth * m_numCuInHeight; i++)
-    {
-        saoLcuParam[i].mergeUpFlag   =  1;
-        saoLcuParam[i].mergeLeftFlag =  0;
-        saoLcuParam[i].partIdx       =  0;
-        saoLcuParam[i].typeIdx       = -1;
-        for (j = 0; j < MAX_NUM_SAO_OFFSETS; j++)
-        {
-            saoLcuParam[i].offset[j] = 0;
-        }
-
-        saoLcuParam[i].subTypeIdx = 0;
-    }
-}
-
-/** convert QP part to SAO unit
-* \param saoParam SAO parameter
-* \param partIdx SAO part index
-* \param yCbCr color component index
- */
-void TComSampleAdaptiveOffset::convertQT2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int yCbCr)
-{
-    SAOQTPart*  saoPart = &(saoParam->saoPart[yCbCr][partIdx]);
-
-    if (!saoPart->bSplit)
-    {
-        convertOnePart2SaoUnit(saoParam, partIdx, yCbCr);
-        return;
-    }
-
-    if (saoPart->partLevel < m_maxSplitLevel)
-    {
-        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[0], yCbCr);
-        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[1], yCbCr);
-        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[2], yCbCr);
-        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[3], yCbCr);
-    }
-}
-
-/** convert one SAO part to SAO unit
-* \param saoParam SAO parameter
-* \param partIdx SAO part index
-* \param yCbCr color component index
- */
-void TComSampleAdaptiveOffset::convertOnePart2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int yCbCr)
-{
-    int j;
-    int idxX;
-    int idxY;
-    int addr;
-    int frameWidthInCU = m_pic->getFrameWidthInCU();
-    SAOQTPart* saoQTPart = saoParam->saoPart[yCbCr];
-    SaoLcuParam* saoLcuParam = saoParam->saoLcuParam[yCbCr];
-
-    for (idxY = saoQTPart[partIdx].startCUY; idxY <= saoQTPart[partIdx].endCUY; idxY++)
-    {
-        for (idxX = saoQTPart[partIdx].startCUX; idxX <= saoQTPart[partIdx].endCUX; idxX++)
-        {
-            addr = idxY * frameWidthInCU + idxX;
-            saoLcuParam[addr].partIdxTmp = (int)partIdx;
-            saoLcuParam[addr].typeIdx    = saoQTPart[partIdx].bestType;
-            saoLcuParam[addr].subTypeIdx = saoQTPart[partIdx].subTypeIdx;
-            if (saoLcuParam[addr].typeIdx != -1)
-            {
-                saoLcuParam[addr].length    = saoQTPart[partIdx].length;
-                for (j = 0; j < MAX_NUM_SAO_OFFSETS; j++)
-                {
-                    saoLcuParam[addr].offset[j] = saoQTPart[partIdx].offset[j];
-                }
-            }
-            else
-            {
-                saoLcuParam[addr].length    = 0;
-                saoLcuParam[addr].subTypeIdx = saoQTPart[partIdx].subTypeIdx;
-                for (j = 0; j < MAX_NUM_SAO_OFFSETS; j++)
-                {
-                    saoLcuParam[addr].offset[j] = 0;
-                }
-            }
-        }
-    }
-}
-
-void TComSampleAdaptiveOffset::resetSaoUnit(SaoLcuParam* saoUnit)
-{
-    saoUnit->partIdx       = 0;
-    saoUnit->partIdxTmp    = 0;
-    saoUnit->mergeLeftFlag = 0;
-    saoUnit->mergeUpFlag   = 0;
-    saoUnit->typeIdx       = -1;
-    saoUnit->length        = 0;
-    saoUnit->subTypeIdx    = 0;
-
-    for (int i = 0; i < 4; i++)
-    {
-        saoUnit->offset[i] = 0;
-    }
-}
-
-void TComSampleAdaptiveOffset::copySaoUnit(SaoLcuParam* saoUnitDst, SaoLcuParam* saoUnitSrc)
-{
-    saoUnitDst->mergeLeftFlag = saoUnitSrc->mergeLeftFlag;
-    saoUnitDst->mergeUpFlag   = saoUnitSrc->mergeUpFlag;
-    saoUnitDst->typeIdx       = saoUnitSrc->typeIdx;
-    saoUnitDst->length        = saoUnitSrc->length;
-
-    saoUnitDst->subTypeIdx  = saoUnitSrc->subTypeIdx;
-    for (int i = 0; i < 4; i++)
-    {
-        saoUnitDst->offset[i] = saoUnitSrc->offset[i];
-    }
-}
-
-static void xPCMRestoration(TComPic* pic);
-static void xPCMSampleRestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, TextType ttText);
-
-/** PCM LF disable process.
- * \param pic picture (TComPic) pointer
- * \returns void
- *
- * \note Replace filtered sample values of PCM mode blocks with the transmitted and reconstructed ones.
- */
-void PCMLFDisableProcess(TComPic* pic)
-{
-    xPCMRestoration(pic);
-}
-
-/** Picture-level PCM restoration.
- * \param pic picture (TComPic) pointer
- * \returns void
- */
-static void xPCMRestoration(TComPic* pic)
-{
-    bool  bPCMFilter = (pic->getSlice()->getSPS()->getUsePCM() && pic->getSlice()->getSPS()->getPCMFilterDisableFlag()) ? true : false;
-
-    if (bPCMFilter || pic->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        for (uint32_t cuAddr = 0; cuAddr < pic->getNumCUsInFrame(); cuAddr++)
-        {
-            TComDataCU* cu = pic->getCU(cuAddr);
-
-            xPCMCURestoration(cu, 0, 0);
-        }
-    }
-}
-
-/** PCM CU restoration.
- * \param cu pointer to current CU
- * \param absPartIdx part index
- * \param depth CU depth
- * \returns void
- */
-void xPCMCURestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth)
-{
-    TComPic* pic     = cu->getPic();
-    uint32_t curNumParts = pic->getNumPartInCU() >> (depth << 1);
-    uint32_t qNumParts   = curNumParts >> 2;
-
-    // go to sub-CU
-    if (cu->getDepth(absZOrderIdx) > depth)
-    {
-        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
-        {
-            uint32_t lpelx   = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absZOrderIdx]];
-            uint32_t tpely   = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absZOrderIdx]];
-            if ((lpelx < cu->getSlice()->getSPS()->getPicWidthInLumaSamples()) && (tpely < cu->getSlice()->getSPS()->getPicHeightInLumaSamples()))
-                xPCMCURestoration(cu, absZOrderIdx, depth + 1);
-        }
-
-        return;
-    }
-
-    // restore PCM samples
-    if ((cu->getIPCMFlag(absZOrderIdx) && pic->getSlice()->getSPS()->getPCMFilterDisableFlag()) || cu->isLosslessCoded(absZOrderIdx))
-    {
-        xPCMSampleRestoration(cu, absZOrderIdx, depth, TEXT_LUMA);
-        xPCMSampleRestoration(cu, absZOrderIdx, depth, TEXT_CHROMA_U);
-        xPCMSampleRestoration(cu, absZOrderIdx, depth, TEXT_CHROMA_V);
-    }
-}
-
-/** PCM sample restoration.
- * \param cu pointer to current CU
- * \param absPartIdx part index
- * \param depth CU depth
- * \param ttText texture component type
- * \returns void
- */
-static void xPCMSampleRestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, TextType ttText)
-{
-    TComPicYuv* pcPicYuvRec = cu->getPic()->getPicYuvRec();
-    pixel* src;
-    pixel* pcm;
-    uint32_t stride;
-    uint32_t width;
-    uint32_t height;
-    uint32_t pcmLeftShiftBit;
-    uint32_t x, y;
-    uint32_t minCoeffSize = cu->getPic()->getMinCUSize() * cu->getPic()->getMinCUSize();
-    uint32_t lumaOffset   = minCoeffSize * absZOrderIdx;
-    uint32_t chromaOffset = lumaOffset >> 2;
-
-    if (ttText == TEXT_LUMA)
-    {
-        src = pcPicYuvRec->getLumaAddr(cu->getAddr(), absZOrderIdx);
-        pcm = cu->getPCMSampleY() + lumaOffset;
-        stride  = pcPicYuvRec->getStride();
-        width  = (g_maxCUSize >> depth);
-        height = (g_maxCUSize >> depth);
-        if (cu->isLosslessCoded(absZOrderIdx) && !cu->getIPCMFlag(absZOrderIdx))
-        {
-            pcmLeftShiftBit = 0;
-        }
-        else
-        {
-            pcmLeftShiftBit = X265_DEPTH - cu->getSlice()->getSPS()->getPCMBitDepthLuma();
-        }
-    }
-    else
-    {
-        if (ttText == TEXT_CHROMA_U)
-        {
-            src = pcPicYuvRec->getCbAddr(cu->getAddr(), absZOrderIdx);
-            pcm = cu->getPCMSampleCb() + chromaOffset;
-        }
-        else
-        {
-            src = pcPicYuvRec->getCrAddr(cu->getAddr(), absZOrderIdx);
-            pcm = cu->getPCMSampleCr() + chromaOffset;
-        }
-
-        stride = pcPicYuvRec->getCStride();
-        width  = ((g_maxCUSize >> depth) / 2);
-        height = ((g_maxCUSize >> depth) / 2);
-        if (cu->isLosslessCoded(absZOrderIdx) && !cu->getIPCMFlag(absZOrderIdx))
-        {
-            pcmLeftShiftBit = 0;
-        }
-        else
-        {
-            pcmLeftShiftBit = X265_DEPTH - cu->getSlice()->getSPS()->getPCMBitDepthChroma();
-        }
-    }
-
-    //TODO Optimized Primitives 
-    for (y = 0; y < height; y++)
-    {
-        for (x = 0; x < width; x++)
-        {
-            src[x] = (pcm[x] << pcmLeftShiftBit);
-        }
-
-        pcm += width;
-        src += stride;
-    }
-}
-}
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComSampleAdaptiveOffset.h Deleted

@@ -1,222 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComSampleAdaptiveOffset.h
-    \brief    sample adaptive offset class (header)
-*/
-
-#ifndef X265_TCOMSAMPLEADAPTIVEOFFSET_H
-#define X265_TCOMSAMPLEADAPTIVEOFFSET_H
-
-#include "CommonDef.h"
-#include "TComPic.h"
-
-//! \ingroup TLibCommon
-//! \{
-
-namespace x265 {
-// private namespace
-
-#define NUM_DOWN_PART 4
-
-enum SAOTypeLen
-{
-    SAO_EO_LEN    = 4,
-    SAO_BO_LEN    = 4,
-    SAO_MAX_BO_CLASSES = 32
-};
-
-enum SAOType
-{
-    SAO_EO_0 = 0,
-    SAO_EO_1,
-    SAO_EO_2,
-    SAO_EO_3,
-    SAO_BO,
-    MAX_NUM_SAO_TYPE
-};
-
-typedef struct _SaoQTPart
-{
-    int         bestType;
-    int         length;
-    int         subTypeIdx;                ///< indicates EO class or BO band position
-    int         offset[4];
-    int         startCUX;
-    int         startCUY;
-    int         endCUX;
-    int         endCUY;
-
-    int         partIdx;
-    int         partLevel;
-    int         partCol;
-    int         partRow;
-
-    int         downPartsIdx[NUM_DOWN_PART];
-    int         upPartIdx;
-
-    bool        bSplit;
-
-    //---- encoder only start -----//
-    bool        bProcessed;
-    double      minCost;
-    int64_t     minDist;
-    int         minRate;
-    //---- encoder only end -----//
-} SAOQTPart;
-
-typedef struct _SaoLcuParam
-{
-    bool       mergeUpFlag;
-    bool       mergeLeftFlag;
-    int        typeIdx;
-    int        subTypeIdx;                ///< indicates EO class or BO band position
-    int        offset[4];
-    int        partIdx;
-    int        partIdxTmp;
-    int        length;
-} SaoLcuParam;
-
-struct SAOParam
-{
-    bool       bSaoFlag[2];
-    SAOQTPart* saoPart[3];
-    int        maxSplitLevel;
-    bool         oneUnitFlag[3];
-    SaoLcuParam* saoLcuParam[3];
-    int          numCuInHeight;
-    int          numCuInWidth;
-    ~SAOParam();
-};
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-
-#define SAO_MAX_DEPTH                 4
-#define SAO_BO_BITS                   5
-#define LUMA_GROUP_NUM                (1 << SAO_BO_BITS)
-#define MAX_NUM_SAO_OFFSETS           4
-#define MAX_NUM_SAO_CLASS             33
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// Sample Adaptive Offset class
-class TComSampleAdaptiveOffset
-{
-protected:
-
-    TComPic* m_pic;
-
-    static const uint32_t m_maxDepth;
-    static const int m_numCulPartsLevel[5];
-    static const uint32_t m_eoTable[9];
-    static const int m_numClass[MAX_NUM_SAO_TYPE];
-    int32_t *m_offsetBo;
-    int32_t *m_chromaOffsetBo;
-    int8_t m_offsetEo[LUMA_GROUP_NUM];
-    int  m_picWidth;
-    int  m_picHeight;
-    uint32_t m_maxSplitLevel;
-    uint32_t m_maxCUWidth;
-    uint32_t m_maxCUHeight;
-    int  m_numCuInWidth;
-    int  m_numCuInHeight;
-    int  m_numTotalParts;
-    int m_hChromaShift;
-    int m_vChromaShift;
-
-    uint32_t m_saoBitIncreaseY;
-    uint32_t m_saoBitIncreaseC; //for chroma
-    uint32_t m_qp;
-
-    pixel* m_clipTable;
-    pixel* m_clipTableBase;
-    pixel* m_lumaTableBo;
-    pixel* m_chromaClipTable;
-    pixel* m_chromaClipTableBase;
-    pixel* m_chromaTableBo;
-    int32_t    *m_upBuff1;
-    int32_t    *m_upBuff2;
-    int32_t    *m_upBufft;
-    TComPicYuv* m_tmpYuv;  //!< temporary picture buffer pointer when non-across slice/tile boundary SAO is enabled
-
-    pixel* m_tmpU1[3];
-    pixel* m_tmpU2[3];
-    pixel* m_tmpL1;
-    pixel* m_tmpL2;
-    int     m_maxNumOffsetsPerPic;
-    bool    m_saoLcuBoundary;
-    bool    m_saoLcuBasedOptimization;
-
-public:
-
-    TComSampleAdaptiveOffset();
-    virtual ~TComSampleAdaptiveOffset();
-    void create(uint32_t sourceWidth, uint32_t sourceHeight, uint32_t maxCUWidth, uint32_t maxCUHeight, int csp);
-    void destroy();
-
-    int  convertLevelRowCol2Idx(int level, int row, int col) const;
-
-    void initSAOParam(SAOParam* saoParam, int partLevel, int partRow, int partCol, int parentPartIdx, int startCUX, int endCUX, int startCUY, int endCUY, int yCbCr) const;
-    void allocSaoParam(SAOParam* saoParam) const;
-    void resetSAOParam(SAOParam* saoParam);
-    static void freeSaoParam(SAOParam* saoParam);
-
-    void SAOProcess(SAOParam* saoParam);
-    pixel* getPicYuvAddr(TComPicYuv* picYuv, int yCbCr, int addr = 0);
-
-    void processSaoCu(int addr, int partIdx, int yCbCr); //!< LCU-basd SAO process without slice granularity
-    void createPicSaoInfo(TComPic* pic);
-
-    void resetLcuPart(SaoLcuParam* saoLcuParam);
-    void convertQT2SaoUnit(SAOParam* saoParam, uint32_t partIdx, int yCbCr);
-    void convertOnePart2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int yCbCr);
-    void processSaoUnitAll(SaoLcuParam* saoLcuParam, bool oneUnitFlag, int yCbCr);
-    void processSaoUnitRow(SaoLcuParam* saoLcuParam, int idxY, int yCbCr);
-    void setSaoLcuBasedOptimization(int bVal)  { m_saoLcuBasedOptimization = bVal != 0; }
-    void setSaoLcuBoundary(int bVal)  { m_saoLcuBoundary = bVal != 0; }
-
-    bool getSaoLcuBasedOptimization()           { return m_saoLcuBasedOptimization; }
-
-    void resetSaoUnit(SaoLcuParam* saoUnit);
-    void copySaoUnit(SaoLcuParam* saoUnitDst, SaoLcuParam* saoUnitSrc);
-};
-
-void PCMLFDisableProcess(TComPic* pic);
-void xPCMCURestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth);
-}
-
-//! \}
-#endif // ifndef X265_TCOMSAMPLEADAPTIVEOFFSET_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComSlice.cpp Deleted

@@ -1,1047 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComSlice.cpp
-    \brief    slice header and SPS class
-*/
-
-#include "CommonDef.h"
-#include "TComSlice.h"
-#include "TComPic.h"
-#include "TLibEncoder/TEncSbac.h"
-#include "threadpool.h"
-
-using namespace x265;
-
-//! \ingroup TLibCommon
-//! \{
-
-TComSlice::TComSlice()
-    : m_ppsId(-1)
-    , m_poc(0)
-    , m_lastIDR(0)
-    , m_nalUnitType(NAL_UNIT_CODED_SLICE_IDR_W_RADL)
-    , m_sliceType(I_SLICE)
-    , m_sliceQp(0)
-    , m_dependentSliceSegmentFlag(false)
-    , m_deblockingFilterDisable(false)
-    , m_deblockingFilterOverrideFlag(false)
-    , m_deblockingFilterBetaOffsetDiv2(0)
-    , m_deblockingFilterTcOffsetDiv2(0)
-    , m_bCheckLDC(false)
-    , m_sliceQpDelta(0)
-    , m_sliceQpDeltaCb(0)
-    , m_sliceQpDeltaCr(0)
-    , m_bReferenced(false)
-    , m_sps(NULL)
-    , m_pps(NULL)
-    , m_pic(NULL)
-    , m_colFromL0Flag(1)
-    , m_colRefIdx(0)
-    , m_sliceCurEndCUAddr(0)
-    , m_nextSlice(false)
-    , m_sliceBits(0)
-    , m_sliceSegmentBits(0)
-    , m_bFinalized(false)
-    , m_tileOffstForMultES(0)
-    , m_substreamSizes(NULL)
-    , m_cabacInitFlag(false)
-    , m_bLMvdL1Zero(false)
-    , m_numEntryPointOffsets(0)
-    , m_temporalLayerNonReferenceFlag(false)
-    , m_enableTMVPFlag(true)
-{
-    m_numRefIdx[0] = m_numRefIdx[1] = 0;
-
-    for (int numCount = 0; numCount < MAX_NUM_REF; numCount++)
-    {
-        m_refPicList[0][numCount] = NULL;
-        m_refPicList[1][numCount] = NULL;
-        m_refPOCList[0][numCount] = 0;
-        m_refPOCList[1][numCount] = 0;
-    }
-
-    resetWpScaling();
-    m_saoEnabledFlag = false;
-}
-
-TComSlice::~TComSlice()
-{
-    delete[] m_substreamSizes;
-}
-
-void TComSlice::initSlice()
-{
-    m_numRefIdx[0] = 0;
-    m_numRefIdx[1] = 0;
-
-    m_colFromL0Flag = 1;
-
-    m_colRefIdx = 0;
-    m_bCheckLDC = false;
-    m_sliceQpDeltaCb = 0;
-    m_sliceQpDeltaCr = 0;
-    m_maxNumMergeCand = MRG_MAX_NUM_CANDS;
-
-    m_bFinalized = false;
-
-    m_cabacInitFlag = false;
-    m_numEntryPointOffsets = 0;
-    m_enableTMVPFlag = true;
-}
-
-bool TComSlice::getRapPicFlag()
-{
-    return getNalUnitType() == NAL_UNIT_CODED_SLICE_IDR_W_RADL
-           || getNalUnitType() == NAL_UNIT_CODED_SLICE_IDR_N_LP
-           || getNalUnitType() == NAL_UNIT_CODED_SLICE_BLA_N_LP
-           || getNalUnitType() == NAL_UNIT_CODED_SLICE_BLA_W_RADL
-           || getNalUnitType() == NAL_UNIT_CODED_SLICE_BLA_W_LP
-           || getNalUnitType() == NAL_UNIT_CODED_SLICE_CRA;
-}
-
-/**
- - allocate table to contain substream sizes to be written to the slice header.
- .
- \param uiNumSubstreams Number of substreams -- the allocation will be this value - 1.
- */
-void  TComSlice::allocSubstreamSizes(uint32_t numSubstreams)
-{
-    delete[] m_substreamSizes;
-    m_substreamSizes = new uint32_t[numSubstreams > 0 ? numSubstreams - 1 : 0];
-}
-
-TComPic* TComSlice::xGetRefPic(PicList& picList, int poc)
-{
-    TComPic *iterPic = picList.first();
-    TComPic* pic = NULL;
-
-    while (iterPic)
-    {
-        pic = iterPic;
-        if (pic->getPOC() == poc)
-        {
-            break;
-        }
-        iterPic = iterPic->m_next;
-    }
-
-    return pic;
-}
-
-TComPic* TComSlice::xGetLongTermRefPic(PicList& picList, int poc, bool pocHasMsb)
-{
-    TComPic* iterPic = picList.first();
-    TComPic* pic = iterPic;
-    TComPic* stPic = pic;
-
-    int pocCycle = 1 << getSPS()->getBitsForPOC();
-
-    if (!pocHasMsb)
-    {
-        poc = poc % pocCycle;
-    }
-
-    while (iterPic)
-    {
-        pic = iterPic;
-        if (pic && pic->getPOC() != this->getPOC() && pic->getSlice()->isReferenced())
-        {
-            int picPoc = pic->getPOC();
-            if (!pocHasMsb)
-            {
-                picPoc = picPoc % pocCycle;
-            }
-
-            if (poc == picPoc)
-            {
-                if (pic->getIsLongTerm())
-                {
-                    return pic;
-                }
-                else
-                {
-                    stPic = pic;
-                }
-                break;
-            }
-        }
-
-        iterPic = iterPic->m_next;
-    }
-
-    return stPic;
-}
-
-void TComSlice::setRefPOCList()
-{
-    for (int dir = 0; dir < 2; dir++)
-    {
-        for (int numRefIdx = 0; numRefIdx < m_numRefIdx[dir]; numRefIdx++)
-        {
-            m_refPOCList[dir][numRefIdx] = m_refPicList[dir][numRefIdx]->getPOC();
-        }
-    }
-}
-
-void TComSlice::setRefPicList(PicList& picList)
-{
-    if (m_sliceType == I_SLICE)
-    {
-        ::memset(m_refPicList, 0, sizeof(m_refPicList));
-        ::memset(m_numRefIdx,  0, sizeof(m_numRefIdx));
-
-        return;
-    }
-
-    m_numRefIdx[0] = getNumRefIdx(REF_PIC_LIST_0);
-    m_numRefIdx[1] = getNumRefIdx(REF_PIC_LIST_1);
-
-    TComPic* refPic = NULL;
-    TComPic* refPicSetStCurr0[MAX_NUM_REF];
-    TComPic* refPicSetStCurr1[MAX_NUM_REF];
-    TComPic* refPicSetLtCurr[MAX_NUM_REF];
-    uint32_t numPocStCurr0 = 0;
-    uint32_t numPocStCurr1 = 0;
-    uint32_t numPocLtCurr = 0;
-    int i;
-
-    for (i = 0; i < m_rps->getNumberOfNegativePictures(); i++)
-    {
-        if (m_rps->getUsed(i))
-        {
-            refPic = xGetRefPic(picList, getPOC() + m_rps->getDeltaPOC(i));
-            refPic->setIsLongTerm(0);
-            refPicSetStCurr0[numPocStCurr0] = refPic;
-            numPocStCurr0++;
-            refPic->setCheckLTMSBPresent(false);
-        }
-    }
-
-    for (; i < m_rps->getNumberOfNegativePictures() + m_rps->getNumberOfPositivePictures(); i++)
-    {
-        if (m_rps->getUsed(i))
-        {
-            refPic = xGetRefPic(picList, getPOC() + m_rps->getDeltaPOC(i));
-            refPic->setIsLongTerm(0);
-            refPicSetStCurr1[numPocStCurr1] = refPic;
-            numPocStCurr1++;
-            refPic->setCheckLTMSBPresent(false);
-        }
-    }
-
-    for (i = m_rps->getNumberOfNegativePictures() + m_rps->getNumberOfPositivePictures() + m_rps->getNumberOfLongtermPictures() - 1;
-         i > m_rps->getNumberOfNegativePictures() + m_rps->getNumberOfPositivePictures() - 1; i--)
-    {
-        if (m_rps->getUsed(i))
-        {
-            refPic = xGetLongTermRefPic(picList, m_rps->getPOC(i), m_rps->getCheckLTMSBPresent(i));
-            refPic->setIsLongTerm(1);
-            refPicSetLtCurr[numPocLtCurr] = refPic;
-            numPocLtCurr++;
-        }
-        if (refPic == NULL)
-        {
-            refPic = xGetLongTermRefPic(picList, m_rps->getPOC(i), m_rps->getCheckLTMSBPresent(i));
-        }
-        refPic->setCheckLTMSBPresent(m_rps->getCheckLTMSBPresent(i));
-    }
-
-    // ref_pic_list_init
-    TComPic* rpsCurrList0[MAX_NUM_REF + 1];
-    TComPic* rpsCurrList1[MAX_NUM_REF + 1];
-    int numPocTotalCurr = numPocStCurr0 + numPocStCurr1 + numPocLtCurr;
-
-    int cIdx = 0;
-    for (i = 0; i < numPocStCurr0; i++, cIdx++)
-    {
-        rpsCurrList0[cIdx] = refPicSetStCurr0[i];
-    }
-
-    for (i = 0; i < numPocStCurr1; i++, cIdx++)
-    {
-        rpsCurrList0[cIdx] = refPicSetStCurr1[i];
-    }
-
-    for (i = 0; i < numPocLtCurr; i++, cIdx++)
-    {
-        rpsCurrList0[cIdx] = refPicSetLtCurr[i];
-    }
-
-    assert(cIdx == numPocTotalCurr);
-
-    if (m_sliceType == B_SLICE)
-    {
-        cIdx = 0;
-        for (i = 0; i < numPocStCurr1; i++, cIdx++)
-        {
-            rpsCurrList1[cIdx] = refPicSetStCurr1[i];
-        }
-
-        for (i = 0; i < numPocStCurr0; i++, cIdx++)
-        {
-            rpsCurrList1[cIdx] = refPicSetStCurr0[i];
-        }
-
-        for (i = 0; i < numPocLtCurr; i++, cIdx++)
-        {
-            rpsCurrList1[cIdx] = refPicSetLtCurr[i];
-        }
-
-        assert(cIdx == numPocTotalCurr);
-    }
-
-    ::memset(m_bIsUsedAsLongTerm, 0, sizeof(m_bIsUsedAsLongTerm));
-
-    for (int rIdx = 0; rIdx < m_numRefIdx[0]; rIdx++)
-    {
-        cIdx = rIdx % numPocTotalCurr;
-        assert(cIdx >= 0 && cIdx < numPocTotalCurr);
-        m_refPicList[0][rIdx] = rpsCurrList0[cIdx];
-        m_bIsUsedAsLongTerm[0][rIdx] = (cIdx >= numPocStCurr0 + numPocStCurr1);
-    }
-
-    if (m_sliceType != B_SLICE)
-    {
-        m_numRefIdx[1] = 0;
-        ::memset(m_refPicList[1], 0, sizeof(m_refPicList[1]));
-    }
-    else
-    {
-        for (int rIdx = 0; rIdx < m_numRefIdx[1]; rIdx++)
-        {
-            cIdx = rIdx % numPocTotalCurr;
-            assert(cIdx >= 0 && cIdx < numPocTotalCurr);
-            m_refPicList[1][rIdx] = rpsCurrList1[cIdx];
-            m_bIsUsedAsLongTerm[1][rIdx] = (cIdx >= numPocStCurr0 + numPocStCurr1);
-        }
-    }
-}
-
-int TComSlice::getNumRpsCurrTempList()
-{
-    int numRpsCurrTempList = 0;
-
-    if (m_sliceType == I_SLICE)
-    {
-        return 0;
-    }
-    for (uint32_t i = 0; i < m_rps->getNumberOfNegativePictures() + m_rps->getNumberOfPositivePictures() + m_rps->getNumberOfLongtermPictures(); i++)
-    {
-        if (m_rps->getUsed(i))
-        {
-            numRpsCurrTempList++;
-        }
-    }
-
-    return numRpsCurrTempList;
-}
-
-/** get WP tables for weighted pred
- * \param int
- * \param refIdx
- * \param *&wpScalingParam
- * \returns void
- */
-void TComSlice::getWpScaling(int l, int refIdx, wpScalingParam *&wp)
-{
-    wp = m_weightPredTable[l][refIdx];
-}
-
-/** reset Default WP tables settings : no weight.
- * \param wpScalingParam
- * \returns void
- */
-void  TComSlice::resetWpScaling()
-{
-    for (int e = 0; e < 2; e++)
-    {
-        for (int i = 0; i < MAX_NUM_REF; i++)
-        {
-            for (int yuv = 0; yuv < 3; yuv++)
-            {
-                wpScalingParam  *pwp = &(m_weightPredTable[e][i][yuv]);
-                pwp->bPresentFlag    = false;
-                pwp->log2WeightDenom = 0;
-                pwp->inputWeight     = 1;
-                pwp->inputOffset     = 0;
-            }
-        }
-    }
-}
-
-/** init WP table
- * \returns void
- */
-void  TComSlice::initWpScaling()
-{
-    for (int e = 0; e < 2; e++)
-    {
-        for (int i = 0; i < MAX_NUM_REF; i++)
-        {
-            for (int yuv = 0; yuv < 3; yuv++)
-            {
-                wpScalingParam  *pwp = &(m_weightPredTable[e][i][yuv]);
-                if (!pwp->bPresentFlag)
-                {
-                    // Inferring values not present :
-                    pwp->inputWeight = (1 << pwp->log2WeightDenom);
-                    pwp->inputOffset = 0;
-                }
-
-                pwp->w      = pwp->inputWeight;
-                pwp->o      = pwp->inputOffset << (X265_DEPTH - 8);
-                pwp->shift  = pwp->log2WeightDenom;
-                pwp->round  = (pwp->log2WeightDenom >= 1) ? (1 << (pwp->log2WeightDenom - 1)) : (0);
-            }
-        }
-    }
-}
-
-// ------------------------------------------------------------------------------------------------
-// Video parameter set (VPS)
-// ------------------------------------------------------------------------------------------------
-TComVPS::TComVPS()
-    : m_VPSId(0)
-    , m_maxTLayers(1)
-    , m_maxLayers(1)
-    , m_bTemporalIdNestingFlag(false)
-    , m_numHrdParameters(0)
-    , m_maxNuhReservedZeroLayerId(0)
-    , m_hrdParameters(NULL)
-    , m_hrdOpSetIdx(NULL)
-    , m_cprmsPresentFlag(NULL)
-{
-    for (int i = 0; i < MAX_TLAYER; i++)
-    {
-        m_numReorderPics[i] = 0;
-        m_maxDecPicBuffering[i] = 1;
-        m_maxLatencyIncrease[i] = 0;
-    }
-}
-
-TComVPS::~TComVPS()
-{
-    delete[] m_hrdParameters;
-    delete[] m_hrdOpSetIdx;
-    delete[] m_cprmsPresentFlag;
-}
-
-// ------------------------------------------------------------------------------------------------
-// Sequence parameter set (SPS)
-// ------------------------------------------------------------------------------------------------
-
-TComSPS::TComSPS()
-    : m_SPSId(0)
-    , m_VPSId(0)
-    , m_chromaFormatIdc(X265_CSP_I420)
-    , m_maxTLayers(1)
-// Structure
-    , m_picWidthInLumaSamples(352)
-    , m_picHeightInLumaSamples(288)
-    , m_log2MinCodingBlockSize(0)
-    , m_log2DiffMaxMinCodingBlockSize(0)
-    , m_maxCUSize(32)
-    , m_maxCUDepth(3)
-    , m_bLongTermRefsPresent(false)
-    , m_quadtreeTULog2MaxSize(0)
-    , m_quadtreeTULog2MinSize(0)
-    , m_quadtreeTUMaxDepthInter(0)
-    , m_quadtreeTUMaxDepthIntra(0)
-// Tool list
-    , m_usePCM(false)
-    , m_pcmLog2MaxSize(5)
-    , m_pcmLog2MinSize(7)
-    , m_bitDepthY(8)
-    , m_bitDepthC(8)
-    , m_qpBDOffsetY(0)
-    , m_qpBDOffsetC(0)
-    , m_useLossless(false)
-    , m_pcmBitDepthLuma(8)
-    , m_pcmBitDepthChroma(8)
-    , m_bPCMFilterDisableFlag(false)
-    , m_bitsForPOC(8)
-    , m_numLongTermRefPicSPS(0)
-    , m_maxTrSize(32)
-    , m_bUseSAO(false)
-    , m_bTemporalIdNestingFlag(false)
-    , m_scalingListEnabledFlag(false)
-    , m_useStrongIntraSmoothing(false)
-    , m_vuiParametersPresentFlag(false)
-    , m_vuiParameters()
-{
-    for (int i = 0; i < MAX_TLAYER; i++)
-    {
-        m_maxLatencyIncrease[i] = 0;
-        m_maxDecPicBuffering[i] = 1;
-        m_numReorderPics[i]       = 0;
-    }
-
-    m_scalingList = new TComScalingList;
-    ::memset(m_ltRefPicPocLsbSps, 0, sizeof(m_ltRefPicPocLsbSps));
-    ::memset(m_usedByCurrPicLtSPSFlag, 0, sizeof(m_usedByCurrPicLtSPSFlag));
-}
-
-TComSPS::~TComSPS()
-{
-    delete m_scalingList;
-    m_RPSList.destroy();
-}
-
-void  TComSPS::createRPSList(int numRPS)
-{
-    m_RPSList.destroy();
-    m_RPSList.create(numRPS);
-}
-
-void TComSPS::setHrdParameters(uint32_t fpsNum, uint32_t fpsDenom, uint32_t numDU, uint32_t bitRate, bool randomAccess)
-{
-    if (!getVuiParametersPresentFlag())
-    {
-        return;
-    }
-
-    TComVUI *vui = getVuiParameters();
-    TComHRD *hrd = vui->getHrdParameters();
-
-    TimingInfo *timingInfo = vui->getTimingInfo();
-    timingInfo->setTimingInfoPresentFlag(true);
-    timingInfo->setNumUnitsInTick(fpsDenom);
-    timingInfo->setTimeScale(fpsNum);
-
-    bool rateCnt = (bitRate > 0);
-    hrd->setVclHrdParametersPresentFlag(rateCnt);
-
-    if (hrd->getSubPicHrdParamsPresentFlag())
-    {
-        hrd->setTickDivisorMinus2(100 - 2);
-        hrd->setDuCpbRemovalDelayLengthMinus1(7);                  // 8-bit precision ( plus 1 for last DU in AU )
-        hrd->setSubPicCpbParamsInPicTimingSEIFlag(true);
-        hrd->setDpbOutputDelayDuLengthMinus1(5 + 7);               // With sub-clock tick factor of 100, at least 7 bits to have the same value as AU dpb delay
-    }
-    else
-    {
-        hrd->setSubPicCpbParamsInPicTimingSEIFlag(false);
-    }
-
-    hrd->setBitRateScale(4);                                       // in units of 2~( 6 + 4 ) = 1,024 bps
-    hrd->setCpbSizeScale(6);                                       // in units of 2~( 4 + 4 ) = 1,024 bit
-    hrd->setDuCpbSizeScale(6);                                     // in units of 2~( 4 + 4 ) = 1,024 bit
-
-    hrd->setInitialCpbRemovalDelayLengthMinus1(15);                // assuming 0.5 sec, log2( 90,000 * 0.5 ) = 16-bit
-    if (randomAccess)
-    {
-        hrd->setCpbRemovalDelayLengthMinus1(5);                    // 32 = 2^5 (plus 1)
-        hrd->setDpbOutputDelayLengthMinus1(5);                     // 32 + 3 = 2^6
-    }
-    else
-    {
-        hrd->setCpbRemovalDelayLengthMinus1(9);                    // max. 2^10
-        hrd->setDpbOutputDelayLengthMinus1(9);                     // max. 2^10
-    }
-
-/*
-   Note: only the case of "vps_max_temporal_layers_minus1 = 0" is supported.
-*/
-    int i, j;
-    uint32_t birateValue, cpbSizeValue;
-    uint32_t ducpbSizeValue;
-    uint32_t duBitRateValue = 0;
-
-    for (i = 0; i < MAX_TLAYER; i++)
-    {
-        hrd->setFixedPicRateFlag(i, 1);
-        hrd->setPicDurationInTcMinus1(i, 0);
-        hrd->setLowDelayHrdFlag(i, 0);
-        hrd->setCpbCntMinus1(i, 0);
-
-        birateValue  = bitRate;
-        cpbSizeValue = bitRate;                                 // 1 second
-        ducpbSizeValue = bitRate / numDU;
-        duBitRateValue = bitRate;
-        for (j = 0; j < (hrd->getCpbCntMinus1(i) + 1); j++)
-        {
-            hrd->setBitRateValueMinus1(i, j, 0, (birateValue  - 1));
-            hrd->setCpbSizeValueMinus1(i, j, 0, (cpbSizeValue - 1));
-            hrd->setDuCpbSizeValueMinus1(i, j, 0, (ducpbSizeValue - 1));
-            hrd->setCbrFlag(i, j, 0, (j == 0));
-
-            hrd->setBitRateValueMinus1(i, j, 1, (birateValue  - 1));
-            hrd->setCpbSizeValueMinus1(i, j, 1, (cpbSizeValue - 1));
-            hrd->setDuCpbSizeValueMinus1(i, j, 1, (ducpbSizeValue - 1));
-            hrd->setDuBitRateValueMinus1(i, j, 1, (duBitRateValue - 1));
-            hrd->setCbrFlag(i, j, 1, (j == 0));
-        }
-    }
-}
-
-TComPPS::TComPPS()
-    : m_PPSId(0)
-    , m_SPSId(0)
-    , m_picInitQPMinus26(0)
-    , m_useDQP(false)
-    , m_bConstrainedIntraPred(false)
-    , m_bSliceChromaQpFlag(false)
-    , m_sps(NULL)
-    , m_maxCuDQPDepth(0)
-    , m_minCuDQPSize(0)
-    , m_chromaCbQpOffset(0)
-    , m_chromaCrQpOffset(0)
-    , m_numRefIdxL0DefaultActive(1)
-    , m_numRefIdxL1DefaultActive(1)
-    , m_transquantBypassEnableFlag(false)
-    , m_useTransformSkip(false)
-    , m_entropyCodingSyncEnabledFlag(false)
-    , m_loopFilterAcrossTilesEnabledFlag(true)
-    , m_signHideFlag(0)
-    , m_cabacInitPresentFlag(false)
-    , m_encCABACTableIdx(I_SLICE)
-    , m_sliceHeaderExtensionPresentFlag(false)
-    , m_listsModificationPresentFlag(0)
-    , m_numExtraSliceHeaderBits(0)
-{
-    m_scalingList = new TComScalingList;
-}
-
-TComPPS::~TComPPS()
-{
-    delete m_scalingList;
-}
-
-TComReferencePictureSet::TComReferencePictureSet()
-    : m_deltaRIdxMinus1(0)
-    , m_deltaRPS(0)
-    , m_numRefIdc(0)
-    , m_numberOfPictures(0)
-    , m_numberOfNegativePictures(0)
-    , m_numberOfPositivePictures(0)
-    , m_interRPSPrediction(0)
-    , m_numberOfLongtermPictures(0)
-{
-    ::memset(m_deltaPOC, 0, sizeof(m_deltaPOC));
-    ::memset(m_POC, 0, sizeof(m_POC));
-    ::memset(m_used, 0, sizeof(m_used));
-    ::memset(m_refIdc, 0, sizeof(m_refIdc));
-}
-
-TComReferencePictureSet::~TComReferencePictureSet()
-{}
-
-void TComReferencePictureSet::setUsed(int bufferNum, bool used)
-{
-    m_used[bufferNum] = used;
-}
-
-void TComReferencePictureSet::setDeltaPOC(int bufferNum, int deltaPOC)
-{
-    m_deltaPOC[bufferNum] = deltaPOC;
-}
-
-int TComReferencePictureSet::getUsed(int bufferNum) const
-{
-    return m_used[bufferNum];
-}
-
-int TComReferencePictureSet::getDeltaPOC(int bufferNum) const
-{
-    return m_deltaPOC[bufferNum];
-}
-
-int TComReferencePictureSet::getNumberOfPictures() const
-{
-    return m_numberOfPictures;
-}
-
-int TComReferencePictureSet::getPOC(int bufferNum) const
-{
-    return m_POC[bufferNum];
-}
-
-void TComReferencePictureSet::setPOC(int bufferNum, int POC)
-{
-    m_POC[bufferNum] = POC;
-}
-
-bool TComReferencePictureSet::getCheckLTMSBPresent(int bufferNum)
-{
-    return m_bCheckLTMSB[bufferNum];
-}
-
-void TComReferencePictureSet::setCheckLTMSBPresent(int bufferNum, bool b)
-{
-    m_bCheckLTMSB[bufferNum] = b;
-}
-
-/** get the reference idc value at uiBufferNum
- * \param uiBufferNum
- * \returns int
- */
-int  TComReferencePictureSet::getRefIdc(int bufferNum) const
-{
-    return m_refIdc[bufferNum];
-}
-
-/** Sorts the deltaPOC and Used by current values in the RPS based on the deltaPOC values.
- *  deltaPOC values are sorted with -ve values before the +ve values.  -ve values are in decreasing order.
- *  +ve values are in increasing order.
- * \returns void
- */
-void TComReferencePictureSet::sortDeltaPOC()
-{
-    // sort in increasing order (smallest first)
-    for (int j = 1; j < getNumberOfPictures(); j++)
-    {
-        int deltaPOC = getDeltaPOC(j);
-        bool used = getUsed(j);
-        for (int k = j - 1; k >= 0; k--)
-        {
-            int temp = getDeltaPOC(k);
-            if (deltaPOC < temp)
-            {
-                setDeltaPOC(k + 1, temp);
-                setUsed(k + 1, getUsed(k));
-                setDeltaPOC(k, deltaPOC);
-                setUsed(k, used);
-            }
-        }
-    }
-
-    // flip the negative values to largest first
-    int numNegPics = getNumberOfNegativePictures();
-    for (int j = 0, k = numNegPics - 1; j < numNegPics >> 1; j++, k--)
-    {
-        int deltaPOC = getDeltaPOC(j);
-        bool used = getUsed(j);
-        setDeltaPOC(j, getDeltaPOC(k));
-        setUsed(j, getUsed(k));
-        setDeltaPOC(k, deltaPOC);
-        setUsed(k, used);
-    }
-}
-
-/** Prints the deltaPOC and RefIdc (if available) values in the RPS.
- *  A "*" is added to the deltaPOC value if it is Used bu current.
- * \returns void
- */
-void TComReferencePictureSet::printDeltaPOC()
-{
-    printf("DeltaPOC = { ");
-    for (int j = 0; j < getNumberOfPictures(); j++)
-    {
-        printf("%d%s ", getDeltaPOC(j), (getUsed(j) == 1) ? "*" : "");
-    }
-
-    if (getInterRPSPrediction())
-    {
-        printf("}, RefIdc = { ");
-        for (int j = 0; j < getNumRefIdc(); j++)
-        {
-            printf("%d ", getRefIdc(j));
-        }
-    }
-    printf("}\n");
-}
-
-TComRPSList::TComRPSList()
-    : m_numberOfReferencePictureSets(0)
-    , m_referencePictureSets(NULL)
-{}
-
-TComRPSList::~TComRPSList()
-{}
-
-void TComRPSList::create(int numberOfReferencePictureSets)
-{
-    m_numberOfReferencePictureSets = numberOfReferencePictureSets;
-    m_referencePictureSets = new TComReferencePictureSet[numberOfReferencePictureSets];
-}
-
-void TComRPSList::destroy()
-{
-    delete [] m_referencePictureSets;
-    m_numberOfReferencePictureSets = 0;
-    m_referencePictureSets = NULL;
-}
-
-TComReferencePictureSet* TComRPSList::getReferencePictureSet(int referencePictureSetNum)
-{
-    return &m_referencePictureSets[referencePictureSetNum];
-}
-
-int TComRPSList::getNumberOfReferencePictureSets() const
-{
-    return m_numberOfReferencePictureSets;
-}
-
-TComScalingList::TComScalingList()
-{
-    m_useTransformSkip = false;
-    init();
-}
-
-TComScalingList::~TComScalingList()
-{
-    destroy();
-}
-
-/** set default quantization matrix to array
-*/
-void TComSlice::setDefaultScalingList()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            getScalingList()->processDefaultMarix(sizeId, listId);
-        }
-    }
-}
-
-/** check if use default quantization matrix
- * \returns true if use default quantization matrix in all size
-*/
-bool TComSlice::checkDefaultScalingList()
-{
-    uint32_t defaultCounter = 0;
-
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            if (!memcmp(getScalingList()->getScalingListAddress(sizeId, listId), getScalingList()->getScalingListDefaultAddress(sizeId, listId), sizeof(int) * X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId])) // check value of matrix
-                && ((sizeId < SCALING_LIST_16x16) || (getScalingList()->getScalingListDC(sizeId, listId) == 16))) // check DC value
-            {
-                defaultCounter++;
-            }
-        }
-    }
-
-    return (defaultCounter == (SCALING_LIST_NUM * SCALING_LIST_SIZE_NUM - 4)) ? false : true; // -4 for 32x32
-}
-
-/** get scaling matrix from RefMatrixID
- * \param sizeId size index
- * \param Index of input matrix
- * \param Index of reference matrix
- */
-void TComScalingList::processRefMatrix(uint32_t sizeId, uint32_t listId, uint32_t refListId)
-{
-    ::memcpy(getScalingListAddress(sizeId, listId), ((listId == refListId) ? getScalingListDefaultAddress(sizeId, refListId) : getScalingListAddress(sizeId, refListId)), sizeof(int) * X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId]));
-}
-
-/** parse syntax information
- *  \param pchFile syntax information
- *  \returns false if successful
- */
-bool TComScalingList::xParseScalingList(char* pchFile)
-{
-    FILE *fp;
-    char line[1024];
-    uint32_t sizeIdc, listIdc;
-    uint32_t i, size = 0;
-    int32_t *src = 0, data;
-    char *ret;
-    uint32_t  retval;
-
-    if ((fp = fopen(pchFile, "r")) == (FILE*)NULL)
-    {
-        printf("can't open file %s :: set Default Matrix\n", pchFile);
-        return true;
-    }
-
-    for (sizeIdc = 0; sizeIdc < SCALING_LIST_SIZE_NUM; sizeIdc++)
-    {
-        size = X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeIdc]);
-        for (listIdc = 0; listIdc < g_scalingListNum[sizeIdc]; listIdc++)
-        {
-            src = getScalingListAddress(sizeIdc, listIdc);
-
-            fseek(fp, 0, 0);
-            do
-            {
-                ret = fgets(line, 1024, fp);
-                if ((ret == NULL) || (strstr(line, MatrixType[sizeIdc][listIdc]) == NULL && feof(fp)))
-                {
-                    printf("Error: can't read Matrix :: set Default Matrix\n");
-                    return true;
-                }
-            }
-            while (strstr(line, MatrixType[sizeIdc][listIdc]) == NULL);
-            for (i = 0; i < size; i++)
-            {
-                retval = fscanf(fp, "%d,", &data);
-                if (retval != 1)
-                {
-                    printf("Error: can't read Matrix :: set Default Matrix\n");
-                    return true;
-                }
-                src[i] = data;
-            }
-
-            //set DC value for default matrix check
-            setScalingListDC(sizeIdc, listIdc, src[0]);
-
-            if (sizeIdc > SCALING_LIST_8x8)
-            {
-                fseek(fp, 0, 0);
-                do
-                {
-                    ret = fgets(line, 1024, fp);
-                    if ((ret == NULL) || (strstr(line, MatrixType_DC[sizeIdc][listIdc]) == NULL && feof(fp)))
-                    {
-                        printf("Error: can't read DC :: set Default Matrix\n");
-                        return true;
-                    }
-                }
-                while (strstr(line, MatrixType_DC[sizeIdc][listIdc]) == NULL);
-                retval = fscanf(fp, "%d,", &data);
-                if (retval != 1)
-                {
-                    printf("Error: can't read Matrix :: set Default Matrix\n");
-                    return true;
-                }
-                //overwrite DC value when size of matrix is larger than 16x16
-                setScalingListDC(sizeIdc, listIdc, data);
-            }
-        }
-    }
-
-    fclose(fp);
-    return false;
-}
-
-/** initialization process of quantization matrix array
- */
-void TComScalingList::init()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            m_scalingListCoef[sizeId][listId] = new int[X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId])];
-        }
-    }
-}
-
-/** destroy quantization matrix array
- */
-void TComScalingList::destroy()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            if (m_scalingListCoef[sizeId][listId]) delete [] m_scalingListCoef[sizeId][listId];
-        }
-    }
-}
-
-/** get default address of quantization matrix
- * \param sizeId size index
- * \param listId list index
- * \returns pointer of quantization matrix
- */
-int32_t* TComScalingList::getScalingListDefaultAddress(uint32_t sizeId, uint32_t listId)
-{
-    int32_t *src = 0;
-
-    switch (sizeId)
-    {
-    case SCALING_LIST_4x4:
-        src = g_quantTSDefault4x4;
-        break;
-    case SCALING_LIST_8x8:
-        src = (listId < 3) ? g_quantIntraDefault8x8 : g_quantInterDefault8x8;
-        break;
-    case SCALING_LIST_16x16:
-        src = (listId < 3) ? g_quantIntraDefault8x8 : g_quantInterDefault8x8;
-        break;
-    case SCALING_LIST_32x32:
-        src = (listId < 1) ? g_quantIntraDefault8x8 : g_quantInterDefault8x8;
-        break;
-    default:
-        assert(0);
-        src = NULL;
-        break;
-    }
-
-    return src;
-}
-
-/** process of default matrix
- * \param sizeId size index
- * \param Index of input matrix
- */
-void TComScalingList::processDefaultMarix(uint32_t sizeId, uint32_t listId)
-{
-    ::memcpy(getScalingListAddress(sizeId, listId), getScalingListDefaultAddress(sizeId, listId), sizeof(int) * X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId]));
-    setScalingListDC(sizeId, listId, SCALING_LIST_DC);
-}
-
-/** check DC value of matrix for default matrix signaling
- */
-void TComScalingList::checkDcOfMatrix()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            //check default matrix?
-            if (getScalingListDC(sizeId, listId) == 0)
-            {
-                processDefaultMarix(sizeId, listId);
-            }
-        }
-    }
-}
-
-ProfileTierLevel::ProfileTierLevel()
-    : m_profileSpace(0)
-    , m_tierFlag(false)
-    , m_profileIdc(0)
-    , m_levelIdc(0)
-    , m_progressiveSourceFlag(false)
-    , m_interlacedSourceFlag(false)
-    , m_nonPackedConstraintFlag(false)
-    , m_frameOnlyConstraintFlag(false)
-{
-    ::memset(m_profileCompatibilityFlag, 0, sizeof(m_profileCompatibilityFlag));
-}
-
-TComPTL::TComPTL()
-{
-    ::memset(m_subLayerProfilePresentFlag, 0, sizeof(m_subLayerProfilePresentFlag));
-    ::memset(m_subLayerLevelPresentFlag,   0, sizeof(m_subLayerLevelPresentFlag));
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComSlice.h Deleted

@@ -1,1576 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComSlice.h
-    \brief    slice header and SPS class (header)
-*/
-
-#ifndef X265_TCOMSLICE_H
-#define X265_TCOMSLICE_H
-
-#include "CommonDef.h"
-#include "TComRom.h"
-#include "x265.h"  // NAL type enums
-#include "piclist.h"
-#include "common.h"
-
-//! \ingroup TLibCommon
-//! \{
-
-namespace x265 {
-// private namespace
-
-class TComPic;
-class TComTrQuant;
-class MotionReference;
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// Reference Picture Set class
-class TComReferencePictureSet
-{
-private:
-
-    // Parameters for inter RPS prediction
-    int  m_deltaRIdxMinus1;
-    int  m_deltaRPS;
-    int  m_numRefIdc;
-    int  m_refIdc[MAX_NUM_REF_PICS + 1];
-
-    // Parameters for long term references
-    bool m_bCheckLTMSB[MAX_NUM_REF_PICS];
-    int  m_pocLSBLT[MAX_NUM_REF_PICS];
-    int  m_deltaPOCMSBCycleLT[MAX_NUM_REF_PICS];
-    bool m_deltaPocMSBPresentFlag[MAX_NUM_REF_PICS];
-
-public:
-
-    int  m_numberOfPictures;
-    int  m_numberOfNegativePictures;
-    int  m_numberOfPositivePictures;
-    int  m_deltaPOC[MAX_NUM_REF_PICS];
-    bool m_used[MAX_NUM_REF_PICS];
-    int  m_POC[MAX_NUM_REF_PICS];
-
-    bool m_interRPSPrediction;
-    int  m_numberOfLongtermPictures;          // Zero when disabled
-
-    TComReferencePictureSet();
-    virtual ~TComReferencePictureSet();
-    int   getPocLSBLT(int i)                       { return m_pocLSBLT[i]; }
-
-    void  setPocLSBLT(int i, int x)                { m_pocLSBLT[i] = x; }
-
-    int   getDeltaPocMSBCycleLT(int i)             { return m_deltaPOCMSBCycleLT[i]; }
-
-    void  setDeltaPocMSBCycleLT(int i, int x)      { m_deltaPOCMSBCycleLT[i] = x; }
-
-    bool  getDeltaPocMSBPresentFlag(int i)         { return m_deltaPocMSBPresentFlag[i]; }
-
-    void  setDeltaPocMSBPresentFlag(int i, bool x) { m_deltaPocMSBPresentFlag[i] = x; }
-
-    void setUsed(int bufferNum, bool used);
-    void setDeltaPOC(int bufferNum, int deltaPOC);
-    void setPOC(int bufferNum, int deltaPOC);
-    void setCheckLTMSBPresent(int bufferNum, bool b);
-    bool getCheckLTMSBPresent(int bufferNum);
-
-    int  getUsed(int bufferNum) const;
-    int  getDeltaPOC(int bufferNum) const;
-    int  getPOC(int bufferNum) const;
-    int  getNumberOfPictures() const;
-
-    int  getNumberOfNegativePictures() const      { return m_numberOfNegativePictures; }
-
-    int  getNumberOfPositivePictures() const      { return m_numberOfPositivePictures; }
-
-    int  getNumberOfLongtermPictures() const      { return m_numberOfLongtermPictures; }
-
-    bool getInterRPSPrediction() const            { return m_interRPSPrediction; }
-
-    int  getDeltaRIdxMinus1() const               { return m_deltaRIdxMinus1; }
-
-    int  getDeltaRPS() const                      { return m_deltaRPS; }
-
-    int  getNumRefIdc() const                     { return m_numRefIdc; }
-
-    int  getRefIdc(int bufferNum) const;
-
-    void sortDeltaPOC();
-    void printDeltaPOC();
-};
-
-/// Reference Picture Set set class
-class TComRPSList
-{
-private:
-
-    int  m_numberOfReferencePictureSets;
-    TComReferencePictureSet* m_referencePictureSets;
-
-public:
-
-    TComRPSList();
-    virtual ~TComRPSList();
-
-    void  create(int numberOfEntries);
-    void  destroy();
-
-    TComReferencePictureSet* getReferencePictureSet(int referencePictureSetNum);
-    int getNumberOfReferencePictureSets() const;
-};
-
-/// SCALING_LIST class
-class TComScalingList
-{
-public:
-
-    TComScalingList();
-    virtual ~TComScalingList();
-    void     setScalingListPresentFlag(bool b)                  { m_scalingListPresentFlag = b; }
-
-    bool     getScalingListPresentFlag()                        { return m_scalingListPresentFlag; }
-
-    bool     getUseTransformSkip()                              { return m_useTransformSkip; }
-
-    void     setUseTransformSkip(bool b)                        { m_useTransformSkip = b; }
-
-    int32_t*     getScalingListAddress(uint32_t sizeId, uint32_t listId)    { return m_scalingListCoef[sizeId][listId]; }          //!< get matrix coefficient
-
-    bool     checkPredMode(uint32_t sizeId, uint32_t listId);
-    void     setRefMatrixId(uint32_t sizeId, uint32_t listId, uint32_t u)   { m_refMatrixId[sizeId][listId] = u;    }                    //!< set reference matrix ID
-
-    uint32_t     getRefMatrixId(uint32_t sizeId, uint32_t listId)           { return m_refMatrixId[sizeId][listId]; }                    //!< get reference matrix ID
-
-    int32_t*     getScalingListDefaultAddress(uint32_t sizeId, uint32_t listId);                                                         //!< get default matrix coefficient
-    void     processDefaultMarix(uint32_t sizeId, uint32_t listId);
-    void     setScalingListDC(uint32_t sizeId, uint32_t listId, uint32_t u) { m_scalingListDC[sizeId][listId] = u; }                   //!< set DC value
-
-    int      getScalingListDC(uint32_t sizeId, uint32_t listId)         { return m_scalingListDC[sizeId][listId]; }                //!< get DC value
-
-    void     checkDcOfMatrix();
-    void     processRefMatrix(uint32_t sizeId, uint32_t listId, uint32_t refListId);
-    bool     xParseScalingList(char* pchFile);
-
-private:
-
-    void     init();
-    void     destroy();
-    int      m_scalingListDC[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM];              //!< the DC value of the matrix coefficient for 16x16
-    bool     m_useDefaultScalingMatrixFlag[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM]; //!< UseDefaultScalingMatrixFlag
-    uint32_t m_refMatrixId[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM];                //!< RefMatrixID
-    bool     m_scalingListPresentFlag;                                              //!< flag for using default matrix
-    uint32_t m_predMatrixId[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM];               //!< reference list index
-    int      *m_scalingListCoef[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM];           //!< quantization matrix
-    bool     m_useTransformSkip;                                                    //!< transform skipping flag for setting default scaling matrix for 4x4
-};
-
-class ProfileTierLevel
-{
-    int     m_profileSpace;
-    bool    m_tierFlag;
-    int     m_profileIdc;
-    bool    m_profileCompatibilityFlag[32];
-    int     m_levelIdc;
-    bool    m_progressiveSourceFlag;
-    bool    m_interlacedSourceFlag;
-    bool    m_nonPackedConstraintFlag;
-    bool    m_frameOnlyConstraintFlag;
-
-public:
-
-    ProfileTierLevel();
-
-    int getProfileSpace() const { return m_profileSpace; }
-
-    bool getTierFlag() const    { return m_tierFlag; }
-
-    void setTierFlag(bool x)    { m_tierFlag = x; }
-
-    int getProfileIdc() const   { return m_profileIdc; }
-
-    void setProfileIdc(int x)   { m_profileIdc = x; }
-
-    bool getProfileCompatibilityFlag(int i) const   { return m_profileCompatibilityFlag[i]; }
-
-    void setProfileCompatibilityFlag(int i, bool x) { m_profileCompatibilityFlag[i] = x; }
-
-    int getLevelIdc() const                 { return m_levelIdc; }
-
-    void setLevelIdc(int x)                 { m_levelIdc = x; }
-
-    bool getProgressiveSourceFlag() const   { return m_progressiveSourceFlag; }
-
-    void setProgressiveSourceFlag(bool b)   { m_progressiveSourceFlag = b; }
-
-    bool getInterlacedSourceFlag() const    { return m_interlacedSourceFlag; }
-
-    void setInterlacedSourceFlag(bool b)    { m_interlacedSourceFlag = b; }
-
-    bool getNonPackedConstraintFlag() const { return m_nonPackedConstraintFlag; }
-
-    void setNonPackedConstraintFlag(bool b) { m_nonPackedConstraintFlag = b; }
-
-    bool getFrameOnlyConstraintFlag() const { return m_frameOnlyConstraintFlag; }
-
-    void setFrameOnlyConstraintFlag(bool b) { m_frameOnlyConstraintFlag = b; }
-};
-
-class TComPTL
-{
-    ProfileTierLevel m_generalPTL;
-    ProfileTierLevel m_subLayerPTL[6];    // max. value of max_sub_layers_minus1 is 6
-    bool m_subLayerProfilePresentFlag[6];
-    bool m_subLayerLevelPresentFlag[6];
-
-public:
-
-    TComPTL();
-    bool getSubLayerProfilePresentFlag(int i) const { return m_subLayerProfilePresentFlag[i]; }
-
-    void setSubLayerProfilePresentFlag(int i, bool x) { m_subLayerProfilePresentFlag[i] = x; }
-
-    bool getSubLayerLevelPresentFlag(int i) const { return m_subLayerLevelPresentFlag[i]; }
-
-    ProfileTierLevel* getGeneralPTL() { return &m_generalPTL; }
-
-    ProfileTierLevel* getSubLayerPTL(int i) { return &m_subLayerPTL[i]; }
-};
-
-/// VPS class
-
-struct HrdSubLayerInfo
-{
-    bool fixedPicRateFlag;
-    bool fixedPicRateWithinCvsFlag;
-    uint32_t picDurationInTcMinus1;
-    bool lowDelayHrdFlag;
-    uint32_t cpbCntMinus1;
-    uint32_t bitRateValueMinus1[MAX_CPB_CNT][2];
-    uint32_t cpbSizeValue[MAX_CPB_CNT][2];
-    uint32_t ducpbSizeValue[MAX_CPB_CNT][2];
-    bool cbrFlag[MAX_CPB_CNT][2];
-    uint32_t duBitRateValue[MAX_CPB_CNT][2];
-};
-
-class TComHRD
-{
-private:
-
-    bool m_nalHrdParametersPresentFlag;
-    bool m_vclHrdParametersPresentFlag;
-    bool m_subPicHrdParamsPresentFlag;
-    uint32_t m_tickDivisorMinus2;
-    uint32_t m_duCpbRemovalDelayLengthMinus1;
-    bool m_subPicCpbParamsInPicTimingSEIFlag;
-    uint32_t m_dpbOutputDelayDuLengthMinus1;
-    uint32_t m_bitRateScale;
-    uint32_t m_cpbSizeScale;
-    uint32_t m_ducpbSizeScale;
-    uint32_t m_initialCpbRemovalDelayLengthMinus1;
-    uint32_t m_cpbRemovalDelayLengthMinus1;
-    uint32_t m_dpbOutputDelayLengthMinus1;
-    HrdSubLayerInfo m_HRD[MAX_TLAYER];
-
-public:
-
-    TComHRD()
-        : m_nalHrdParametersPresentFlag(0)
-        , m_vclHrdParametersPresentFlag(0)
-        , m_subPicHrdParamsPresentFlag(false)
-        , m_tickDivisorMinus2(0)
-        , m_duCpbRemovalDelayLengthMinus1(0)
-        , m_subPicCpbParamsInPicTimingSEIFlag(false)
-        , m_dpbOutputDelayDuLengthMinus1(0)
-        , m_bitRateScale(0)
-        , m_cpbSizeScale(0)
-        , m_initialCpbRemovalDelayLengthMinus1(0)
-        , m_cpbRemovalDelayLengthMinus1(0)
-        , m_dpbOutputDelayLengthMinus1(0)
-    {}
-
-    virtual ~TComHRD() {}
-
-    void setNalHrdParametersPresentFlag(bool flag) { m_nalHrdParametersPresentFlag = flag; }
-
-    bool getNalHrdParametersPresentFlag() { return m_nalHrdParametersPresentFlag; }
-
-    void setVclHrdParametersPresentFlag(bool flag) { m_vclHrdParametersPresentFlag = flag; }
-
-    bool getVclHrdParametersPresentFlag() { return m_vclHrdParametersPresentFlag; }
-
-    void setSubPicHrdParamsPresentFlag(bool flag) { m_subPicHrdParamsPresentFlag = flag; }
-
-    bool getSubPicHrdParamsPresentFlag() { return m_subPicHrdParamsPresentFlag; }
-
-    void setTickDivisorMinus2(uint32_t value) { m_tickDivisorMinus2 = value; }
-
-    uint32_t getTickDivisorMinus2() { return m_tickDivisorMinus2; }
-
-    void setDuCpbRemovalDelayLengthMinus1(uint32_t value) { m_duCpbRemovalDelayLengthMinus1 = value; }
-
-    uint32_t getDuCpbRemovalDelayLengthMinus1() { return m_duCpbRemovalDelayLengthMinus1; }
-
-    void setSubPicCpbParamsInPicTimingSEIFlag(bool flag) { m_subPicCpbParamsInPicTimingSEIFlag = flag; }
-
-    bool getSubPicCpbParamsInPicTimingSEIFlag() { return m_subPicCpbParamsInPicTimingSEIFlag; }
-
-    void setDpbOutputDelayDuLengthMinus1(uint32_t value) { m_dpbOutputDelayDuLengthMinus1 = value; }
-
-    uint32_t getDpbOutputDelayDuLengthMinus1() { return m_dpbOutputDelayDuLengthMinus1; }
-
-    void setBitRateScale(uint32_t value) { m_bitRateScale = value; }
-
-    uint32_t getBitRateScale() { return m_bitRateScale; }
-
-    void setCpbSizeScale(uint32_t value) { m_cpbSizeScale = value; }
-
-    uint32_t getCpbSizeScale() { return m_cpbSizeScale; }
-
-    void setDuCpbSizeScale(uint32_t value) { m_ducpbSizeScale = value; }
-
-    uint32_t getDuCpbSizeScale() { return m_ducpbSizeScale; }
-
-    void setInitialCpbRemovalDelayLengthMinus1(uint32_t value) { m_initialCpbRemovalDelayLengthMinus1 = value; }
-
-    uint32_t getInitialCpbRemovalDelayLengthMinus1() { return m_initialCpbRemovalDelayLengthMinus1; }
-
-    void setCpbRemovalDelayLengthMinus1(uint32_t value) { m_cpbRemovalDelayLengthMinus1 = value; }
-
-    uint32_t getCpbRemovalDelayLengthMinus1() { return m_cpbRemovalDelayLengthMinus1; }
-
-    void setDpbOutputDelayLengthMinus1(uint32_t value) { m_dpbOutputDelayLengthMinus1 = value; }
-
-    uint32_t getDpbOutputDelayLengthMinus1() { return m_dpbOutputDelayLengthMinus1; }
-
-    void setFixedPicRateFlag(int layer, bool flag) { m_HRD[layer].fixedPicRateFlag = flag; }
-
-    bool getFixedPicRateFlag(int layer) { return m_HRD[layer].fixedPicRateFlag; }
-
-    void setFixedPicRateWithinCvsFlag(int layer, bool flag) { m_HRD[layer].fixedPicRateWithinCvsFlag = flag; }
-
-    bool getFixedPicRateWithinCvsFlag(int layer) { return m_HRD[layer].fixedPicRateWithinCvsFlag; }
-
-    void setPicDurationInTcMinus1(int layer, uint32_t value) { m_HRD[layer].picDurationInTcMinus1 = value; }
-
-    uint32_t getPicDurationInTcMinus1(int layer) { return m_HRD[layer].picDurationInTcMinus1; }
-
-    void setLowDelayHrdFlag(int layer, bool flag) { m_HRD[layer].lowDelayHrdFlag = flag; }
-
-    bool getLowDelayHrdFlag(int layer) { return m_HRD[layer].lowDelayHrdFlag; }
-
-    void setCpbCntMinus1(int layer, uint32_t value) { m_HRD[layer].cpbCntMinus1 = value; }
-
-    uint32_t getCpbCntMinus1(int layer) { return m_HRD[layer].cpbCntMinus1; }
-
-    void setBitRateValueMinus1(int layer, int cpbcnt, int nalOrVcl, uint32_t value) { m_HRD[layer].bitRateValueMinus1[cpbcnt][nalOrVcl] = value; }
-
-    uint32_t getBitRateValueMinus1(int layer, int cpbcnt, int nalOrVcl) { return m_HRD[layer].bitRateValueMinus1[cpbcnt][nalOrVcl]; }
-
-    void setCpbSizeValueMinus1(int layer, int cpbcnt, int nalOrVcl, uint32_t value) { m_HRD[layer].cpbSizeValue[cpbcnt][nalOrVcl] = value; }
-
-    uint32_t getCpbSizeValueMinus1(int layer, int cpbcnt, int nalOrVcl)  { return m_HRD[layer].cpbSizeValue[cpbcnt][nalOrVcl]; }
-
-    void setDuCpbSizeValueMinus1(int layer, int cpbcnt, int nalOrVcl, uint32_t value) { m_HRD[layer].ducpbSizeValue[cpbcnt][nalOrVcl] = value; }
-
-    uint32_t getDuCpbSizeValueMinus1(int layer, int cpbcnt, int nalOrVcl)  { return m_HRD[layer].ducpbSizeValue[cpbcnt][nalOrVcl]; }
-
-    void setDuBitRateValueMinus1(int layer, int cpbcnt, int nalOrVcl, uint32_t value) { m_HRD[layer].duBitRateValue[cpbcnt][nalOrVcl] = value; }
-
-    uint32_t getDuBitRateValueMinus1(int layer, int cpbcnt, int nalOrVcl) { return m_HRD[layer].duBitRateValue[cpbcnt][nalOrVcl]; }
-
-    void setCbrFlag(int layer, int cpbcnt, int nalOrVcl, bool value) { m_HRD[layer].cbrFlag[cpbcnt][nalOrVcl] = value; }
-
-    bool getCbrFlag(int layer, int cpbcnt, int nalOrVcl) { return m_HRD[layer].cbrFlag[cpbcnt][nalOrVcl]; }
-
-    bool getCpbDpbDelaysPresentFlag() { return getNalHrdParametersPresentFlag() || getVclHrdParametersPresentFlag(); }
-};
-
-class TimingInfo
-{
-    bool m_timingInfoPresentFlag;
-    uint32_t m_numUnitsInTick;
-    uint32_t m_timeScale;
-    bool m_pocProportionalToTimingFlag;
-    int  m_numTicksPocDiffOneMinus1;
-
-public:
-
-    TimingInfo()
-        : m_timingInfoPresentFlag(false)
-        , m_numUnitsInTick(1001)
-        , m_timeScale(60000)
-        , m_pocProportionalToTimingFlag(false)
-        , m_numTicksPocDiffOneMinus1(0) {}
-
-    void setTimingInfoPresentFlag(bool flag)    { m_timingInfoPresentFlag = flag; }
-
-    bool getTimingInfoPresentFlag()             { return m_timingInfoPresentFlag; }
-
-    void setNumUnitsInTick(uint32_t value)      { m_numUnitsInTick = value; }
-
-    uint32_t getNumUnitsInTick()                { return m_numUnitsInTick; }
-
-    void setTimeScale(uint32_t value)           { m_timeScale = value; }
-
-    uint32_t getTimeScale()                     { return m_timeScale; }
-
-    bool getPocProportionalToTimingFlag()       { return m_pocProportionalToTimingFlag; }
-
-    void setPocProportionalToTimingFlag(bool x) { m_pocProportionalToTimingFlag = x; }
-
-    int  getNumTicksPocDiffOneMinus1()          { return m_numTicksPocDiffOneMinus1; }
-
-    void setNumTicksPocDiffOneMinus1(int x)     { m_numTicksPocDiffOneMinus1 = x; }
-};
-
-class TComVPS
-{
-private:
-
-    int         m_VPSId;
-    uint32_t    m_maxTLayers;
-    uint32_t    m_maxLayers;
-    bool        m_bTemporalIdNestingFlag;
-
-    uint32_t    m_numReorderPics[MAX_TLAYER];
-    uint32_t    m_maxDecPicBuffering[MAX_TLAYER];
-    uint32_t    m_maxLatencyIncrease[MAX_TLAYER];  // Really max latency increase plus 1 (value 0 expresses no limit)
-
-    uint32_t    m_numHrdParameters;
-    uint32_t    m_maxNuhReservedZeroLayerId;
-    TComHRD*    m_hrdParameters;
-    uint32_t*   m_hrdOpSetIdx;
-    bool*       m_cprmsPresentFlag;
-    uint32_t    m_numOpSets;
-    bool        m_layerIdIncludedFlag[MAX_VPS_OP_SETS_PLUS1][MAX_VPS_NUH_RESERVED_ZERO_LAYER_ID_PLUS1];
-
-    TComPTL     m_ptl;
-    TimingInfo  m_timingInfo;
-
-public:
-
-    TComVPS();
-    virtual ~TComVPS();
-
-    void    createHrdParamBuffer()
-    {
-        m_hrdParameters    = new TComHRD[getNumHrdParameters()];
-        m_hrdOpSetIdx      = new uint32_t[getNumHrdParameters()];
-        m_cprmsPresentFlag = new bool[getNumHrdParameters()];
-    }
-
-    TComHRD* getHrdParameters(uint32_t i)             { return &m_hrdParameters[i]; }
-
-    uint32_t    getHrdOpSetIdx(uint32_t i)                { return m_hrdOpSetIdx[i]; }
-
-    void    setHrdOpSetIdx(uint32_t val, uint32_t i)      { m_hrdOpSetIdx[i] = val; }
-
-    bool    getCprmsPresentFlag(uint32_t i)           { return m_cprmsPresentFlag[i]; }
-
-    void    setCprmsPresentFlag(bool val, uint32_t i) { m_cprmsPresentFlag[i] = val; }
-
-    int     getVPSId()                            { return m_VPSId; }
-
-    void    setVPSId(int i)                       { m_VPSId = i; }
-
-    uint32_t    getMaxTLayers()                       { return m_maxTLayers; }
-
-    void    setMaxTLayers(uint32_t t)                 { m_maxTLayers = t; }
-
-    uint32_t    getMaxLayers()                        { return m_maxLayers; }
-
-    void    setMaxLayers(uint32_t l)                  { m_maxLayers = l; }
-
-    bool    getTemporalNestingFlag()              { return m_bTemporalIdNestingFlag; }
-
-    void    setTemporalNestingFlag(bool t)        { m_bTemporalIdNestingFlag = t; }
-
-    void    setNumReorderPics(uint32_t v, uint32_t tLayer)     { m_numReorderPics[tLayer] = v; }
-
-    uint32_t    getNumReorderPics(uint32_t tLayer)             { return m_numReorderPics[tLayer]; }
-
-    void    setMaxDecPicBuffering(uint32_t v, uint32_t tLayer) { m_maxDecPicBuffering[tLayer] = v; }
-
-    uint32_t    getMaxDecPicBuffering(uint32_t tLayer)         { return m_maxDecPicBuffering[tLayer]; }
-
-    void    setMaxLatencyIncrease(uint32_t v, uint32_t tLayer) { m_maxLatencyIncrease[tLayer] = v; }
-
-    uint32_t    getMaxLatencyIncrease(uint32_t tLayer)         { return m_maxLatencyIncrease[tLayer]; }
-
-    uint32_t    getNumHrdParameters()                      { return m_numHrdParameters; }
-
-    void    setNumHrdParameters(uint32_t v)                { m_numHrdParameters = v; }
-
-    uint32_t    getMaxNuhReservedZeroLayerId()             { return m_maxNuhReservedZeroLayerId; }
-
-    void    setMaxNuhReservedZeroLayerId(uint32_t v)       { m_maxNuhReservedZeroLayerId = v; }
-
-    uint32_t    getMaxOpSets()                             { return m_numOpSets; }
-
-    void    setMaxOpSets(uint32_t v)                       { m_numOpSets = v; }
-
-    bool    getLayerIdIncludedFlag(uint32_t opsIdx, uint32_t id)        { return m_layerIdIncludedFlag[opsIdx][id]; }
-
-    void    setLayerIdIncludedFlag(bool v, uint32_t opsIdx, uint32_t id) { m_layerIdIncludedFlag[opsIdx][id] = v; }
-
-    TComPTL* getPTL()           { return &m_ptl; }
-
-    TimingInfo* getTimingInfo() { return &m_timingInfo; }
-};
-
-class Window
-{
-public:
-
-    bool          m_enabledFlag;
-    int           m_winLeftOffset;
-    int           m_winRightOffset;
-    int           m_winTopOffset;
-    int           m_winBottomOffset;
-
-    Window()
-        : m_enabledFlag(false)
-        , m_winLeftOffset(0)
-        , m_winRightOffset(0)
-        , m_winTopOffset(0)
-        , m_winBottomOffset(0)
-    {}
-};
-
-class TComVUI
-{
-private:
-
-    bool m_aspectRatioInfoPresentFlag;
-    int  m_aspectRatioIdc;
-    int  m_sarWidth;
-    int  m_sarHeight;
-    bool m_overscanInfoPresentFlag;
-    bool m_overscanAppropriateFlag;
-    bool m_videoSignalTypePresentFlag;
-    int  m_videoFormat;
-    bool m_videoFullRangeFlag;
-    bool m_colourDescriptionPresentFlag;
-    int  m_colourPrimaries;
-    int  m_transferCharacteristics;
-    int  m_matrixCoefficients;
-    bool m_chromaLocInfoPresentFlag;
-    int  m_chromaSampleLocTypeTopField;
-    int  m_chromaSampleLocTypeBottomField;
-    bool m_neutralChromaIndicationFlag;
-    bool m_fieldSeqFlag;
-
-    Window m_defaultDisplayWindow;
-    bool m_frameFieldInfoPresentFlag;
-    bool m_hrdParametersPresentFlag;
-    bool m_bitstreamRestrictionFlag;
-    bool m_tilesFixedStructureFlag;
-    bool m_motionVectorsOverPicBoundariesFlag;
-    bool m_restrictedRefPicListsFlag;
-    int  m_minSpatialSegmentationIdc;
-    int  m_maxBytesPerPicDenom;
-    int  m_maxBitsPerMinCuDenom;
-    int  m_log2MaxMvLengthHorizontal;
-    int  m_log2MaxMvLengthVertical;
-    TComHRD m_hrdParameters;
-    TimingInfo m_timingInfo;
-
-public:
-
-    TComVUI()
-        : m_aspectRatioInfoPresentFlag(false)
-        , m_aspectRatioIdc(0)
-        , m_sarWidth(0)
-        , m_sarHeight(0)
-        , m_overscanInfoPresentFlag(false)
-        , m_overscanAppropriateFlag(false)
-        , m_videoSignalTypePresentFlag(false)
-        , m_videoFormat(5)
-        , m_videoFullRangeFlag(false)
-        , m_colourDescriptionPresentFlag(false)
-        , m_colourPrimaries(2)
-        , m_transferCharacteristics(2)
-        , m_matrixCoefficients(2)
-        , m_chromaLocInfoPresentFlag(false)
-        , m_chromaSampleLocTypeTopField(0)
-        , m_chromaSampleLocTypeBottomField(0)
-        , m_neutralChromaIndicationFlag(false)
-        , m_fieldSeqFlag(false)
-        , m_frameFieldInfoPresentFlag(false)
-        , m_hrdParametersPresentFlag(false)
-        , m_bitstreamRestrictionFlag(false)
-        , m_tilesFixedStructureFlag(false)
-        , m_motionVectorsOverPicBoundariesFlag(true)
-        , m_restrictedRefPicListsFlag(1)
-        , m_minSpatialSegmentationIdc(0)
-        , m_maxBytesPerPicDenom(2)
-        , m_maxBitsPerMinCuDenom(1)
-        , m_log2MaxMvLengthHorizontal(15)
-        , m_log2MaxMvLengthVertical(15)
-    {}
-
-    virtual ~TComVUI() {}
-
-    bool getAspectRatioInfoPresentFlag() { return m_aspectRatioInfoPresentFlag; }
-
-    void setAspectRatioInfoPresentFlag(bool i) { m_aspectRatioInfoPresentFlag = i; }
-
-    int getAspectRatioIdc() { return m_aspectRatioIdc; }
-
-    void setAspectRatioIdc(int i) { m_aspectRatioIdc = i; }
-
-    int getSarWidth() { return m_sarWidth; }
-
-    void setSarWidth(int i) { m_sarWidth = i; }
-
-    int getSarHeight() { return m_sarHeight; }
-
-    void setSarHeight(int i) { m_sarHeight = i; }
-
-    bool getOverscanInfoPresentFlag() { return m_overscanInfoPresentFlag; }
-
-    void setOverscanInfoPresentFlag(bool i) { m_overscanInfoPresentFlag = i; }
-
-    bool getOverscanAppropriateFlag() { return m_overscanAppropriateFlag; }
-
-    void setOverscanAppropriateFlag(bool i) { m_overscanAppropriateFlag = i; }
-
-    bool getVideoSignalTypePresentFlag() { return m_videoSignalTypePresentFlag; }
-
-    void setVideoSignalTypePresentFlag(bool i) { m_videoSignalTypePresentFlag = i; }
-
-    int getVideoFormat() { return m_videoFormat; }
-
-    void setVideoFormat(int i) { m_videoFormat = i; }
-
-    bool getVideoFullRangeFlag() { return m_videoFullRangeFlag; }
-
-    void setVideoFullRangeFlag(bool i) { m_videoFullRangeFlag = i; }
-
-    bool getColourDescriptionPresentFlag() { return m_colourDescriptionPresentFlag; }
-
-    void setColourDescriptionPresentFlag(bool i) { m_colourDescriptionPresentFlag = i; }
-
-    int getColourPrimaries() { return m_colourPrimaries; }
-
-    void setColourPrimaries(int i) { m_colourPrimaries = i; }
-
-    int getTransferCharacteristics() { return m_transferCharacteristics; }
-
-    void setTransferCharacteristics(int i) { m_transferCharacteristics = i; }
-
-    int getMatrixCoefficients() { return m_matrixCoefficients; }
-
-    void setMatrixCoefficients(int i) { m_matrixCoefficients = i; }
-
-    bool getChromaLocInfoPresentFlag() { return m_chromaLocInfoPresentFlag; }
-
-    void setChromaLocInfoPresentFlag(bool i) { m_chromaLocInfoPresentFlag = i; }
-
-    int getChromaSampleLocTypeTopField() { return m_chromaSampleLocTypeTopField; }
-
-    void setChromaSampleLocTypeTopField(int i) { m_chromaSampleLocTypeTopField = i; }
-
-    int getChromaSampleLocTypeBottomField() { return m_chromaSampleLocTypeBottomField; }
-
-    void setChromaSampleLocTypeBottomField(int i) { m_chromaSampleLocTypeBottomField = i; }
-
-    bool getNeutralChromaIndicationFlag() { return m_neutralChromaIndicationFlag; }
-
-    void setNeutralChromaIndicationFlag(bool i) { m_neutralChromaIndicationFlag = i; }
-
-    bool getFieldSeqFlag() { return m_fieldSeqFlag; }
-
-    void setFieldSeqFlag(bool i) { m_fieldSeqFlag = i; }
-
-    bool getFrameFieldInfoPresentFlag() { return m_frameFieldInfoPresentFlag; }
-
-    void setFrameFieldInfoPresentFlag(bool i) { m_frameFieldInfoPresentFlag = i; }
-
-    Window& getDefaultDisplayWindow() { return m_defaultDisplayWindow; }
-
-    void setDefaultDisplayWindow(Window& defaultDisplayWindow) { m_defaultDisplayWindow = defaultDisplayWindow; }
-
-    bool getHrdParametersPresentFlag() { return m_hrdParametersPresentFlag; }
-
-    void setHrdParametersPresentFlag(bool i) { m_hrdParametersPresentFlag = i; }
-
-    bool getBitstreamRestrictionFlag() { return m_bitstreamRestrictionFlag; }
-
-    void setBitstreamRestrictionFlag(bool i) { m_bitstreamRestrictionFlag = i; }
-
-    bool getTilesFixedStructureFlag() { return m_tilesFixedStructureFlag; }
-
-    void setTilesFixedStructureFlag(bool i) { m_tilesFixedStructureFlag = i; }
-
-    bool getMotionVectorsOverPicBoundariesFlag() { return m_motionVectorsOverPicBoundariesFlag; }
-
-    void setMotionVectorsOverPicBoundariesFlag(bool i) { m_motionVectorsOverPicBoundariesFlag = i; }
-
-    bool getRestrictedRefPicListsFlag() { return m_restrictedRefPicListsFlag; }
-
-    void setRestrictedRefPicListsFlag(bool b) { m_restrictedRefPicListsFlag = b; }
-
-    int getMinSpatialSegmentationIdc() { return m_minSpatialSegmentationIdc; }
-
-    void setMinSpatialSegmentationIdc(int i) { m_minSpatialSegmentationIdc = i; }
-
-    int getMaxBytesPerPicDenom() { return m_maxBytesPerPicDenom; }
-
-    void setMaxBytesPerPicDenom(int i) { m_maxBytesPerPicDenom = i; }
-
-    int getMaxBitsPerMinCuDenom() { return m_maxBitsPerMinCuDenom; }
-
-    void setMaxBitsPerMinCuDenom(int i) { m_maxBitsPerMinCuDenom = i; }
-
-    int getLog2MaxMvLengthHorizontal() { return m_log2MaxMvLengthHorizontal; }
-
-    void setLog2MaxMvLengthHorizontal(int i) { m_log2MaxMvLengthHorizontal = i; }
-
-    int getLog2MaxMvLengthVertical() { return m_log2MaxMvLengthVertical; }
-
-    void setLog2MaxMvLengthVertical(int i) { m_log2MaxMvLengthVertical = i; }
-
-    TComHRD* getHrdParameters() { return &m_hrdParameters; }
-
-    TimingInfo* getTimingInfo() { return &m_timingInfo; }
-};
-
-/// SPS class
-class TComSPS
-{
-private:
-
-    int         m_SPSId;
-    int         m_VPSId;
-    int         m_chromaFormatIdc;
-    bool        m_colorPlaneFlag;
-    uint32_t    m_maxTLayers;         // maximum number of temporal layers
-
-    // Structure
-    uint32_t    m_picWidthInLumaSamples;
-    uint32_t    m_picHeightInLumaSamples;
-
-    int         m_log2MinCodingBlockSize;
-    int         m_log2DiffMaxMinCodingBlockSize;
-    uint32_t    m_maxCUSize;
-    uint32_t    m_maxCUDepth;
-
-    Window      m_conformanceWindow;
-
-    TComRPSList m_RPSList;
-    bool        m_bLongTermRefsPresent;
-    bool        m_TMVPFlagsPresent;
-    int         m_numReorderPics[MAX_TLAYER];
-
-    // Tool list
-    uint32_t    m_quadtreeTULog2MaxSize;
-    uint32_t    m_quadtreeTULog2MinSize;
-    uint32_t    m_quadtreeTUMaxDepthInter;
-    uint32_t    m_quadtreeTUMaxDepthIntra;
-    bool        m_usePCM;
-    uint32_t    m_pcmLog2MaxSize;
-    uint32_t    m_pcmLog2MinSize;
-    bool        m_useAMP;
-
-    // Parameter
-    int         m_bitDepthY;
-    int         m_bitDepthC;
-    int         m_qpBDOffsetY;
-    int         m_qpBDOffsetC;
-
-    bool        m_useLossless;
-
-    uint32_t    m_pcmBitDepthLuma;
-    uint32_t    m_pcmBitDepthChroma;
-    bool        m_bPCMFilterDisableFlag;
-
-    uint32_t    m_bitsForPOC;
-    uint32_t    m_numLongTermRefPicSPS;
-    uint32_t    m_ltRefPicPocLsbSps[33];
-    bool        m_usedByCurrPicLtSPSFlag[33];
-
-    // Max physical transform size
-    uint32_t    m_maxTrSize;
-
-    int m_iAMPAcc[MAX_CU_DEPTH];
-    bool        m_bUseSAO;
-
-    bool        m_bTemporalIdNestingFlag; // temporal_id_nesting_flag
-
-    bool        m_scalingListEnabledFlag;
-    bool        m_scalingListPresentFlag;
-    TComScalingList* m_scalingList; //!< ScalingList class pointer
-    uint32_t    m_maxDecPicBuffering[MAX_TLAYER];
-    uint32_t   m_maxLatencyIncrease[MAX_TLAYER]; // Really max latency increase plus 1 (value 0 expresses no limit)
-
-    bool        m_useDF;
-    bool        m_useStrongIntraSmoothing;
-
-    bool        m_vuiParametersPresentFlag;
-    TComVUI     m_vuiParameters;
-    TComPTL     m_ptl;
-
-public:
-
-    TComSPS();
-    virtual ~TComSPS();
-
-    int  getVPSId()         { return m_VPSId; }
-
-    void setVPSId(int i)    { m_VPSId = i; }
-
-    int  getSPSId()         { return m_SPSId; }
-
-    void setSPSId(int i)    { m_SPSId = i; }
-
-    int  getChromaFormatIdc()         { return m_chromaFormatIdc; }
-
-    void setChromaFormatIdc(int i)    { m_chromaFormatIdc = i; }
-
-    void setSeparateColorPlaneFlag(bool c)      { m_colorPlaneFlag = c; }
-
-    bool getSeparateColorPlaneFlag()      { return m_colorPlaneFlag; }
-
-    static int getWinUnitX(int chromaFormatIdc) { assert(chromaFormatIdc > 0 && chromaFormatIdc <= MAX_CHROMA_FORMAT_IDC); return g_winUnitX[chromaFormatIdc]; }
-
-    static int getWinUnitY(int chromaFormatIdc) { assert(chromaFormatIdc > 0 && chromaFormatIdc <= MAX_CHROMA_FORMAT_IDC); return g_winUnitY[chromaFormatIdc]; }
-
-    // structure
-    void setPicWidthInLumaSamples(uint32_t u) { m_picWidthInLumaSamples = u; }
-
-    uint32_t getPicWidthInLumaSamples() const { return m_picWidthInLumaSamples; }
-
-    void setPicHeightInLumaSamples(uint32_t u) { m_picHeightInLumaSamples = u; }
-
-    uint32_t getPicHeightInLumaSamples() const { return m_picHeightInLumaSamples; }
-
-    Window& getConformanceWindow() { return m_conformanceWindow; }
-
-    void    setConformanceWindow(Window& conformanceWindow) { m_conformanceWindow = conformanceWindow; }
-
-    uint32_t  getNumLongTermRefPicSPS() const { return m_numLongTermRefPicSPS; }
-
-    void  setNumLongTermRefPicSPS(uint32_t val) { m_numLongTermRefPicSPS = val; }
-
-    uint32_t  getLtRefPicPocLsbSps(uint32_t index) const { return m_ltRefPicPocLsbSps[index]; }
-
-    void  setLtRefPicPocLsbSps(uint32_t index, uint32_t val) { m_ltRefPicPocLsbSps[index] = val; }
-
-    bool getUsedByCurrPicLtSPSFlag(int i) const { return m_usedByCurrPicLtSPSFlag[i]; }
-
-    void setUsedByCurrPicLtSPSFlag(int i, bool x) { m_usedByCurrPicLtSPSFlag[i] = x; }
-
-    int  getLog2MinCodingBlockSize() const { return m_log2MinCodingBlockSize; }
-
-    void setLog2MinCodingBlockSize(int val) { m_log2MinCodingBlockSize = val; }
-
-    int  getLog2DiffMaxMinCodingBlockSize() const { return m_log2DiffMaxMinCodingBlockSize; }
-
-    void setLog2DiffMaxMinCodingBlockSize(int val) { m_log2DiffMaxMinCodingBlockSize = val; }
-
-    void setMaxCUSize(uint32_t u) { m_maxCUSize = u; }
-
-    uint32_t getMaxCUSize() const  { return m_maxCUSize; }
-
-    void setMaxCUDepth(uint32_t u) { m_maxCUDepth = u; }
-
-    uint32_t getMaxCUDepth() const { return m_maxCUDepth; }
-
-    void setUsePCM(bool b)   { m_usePCM = b; }
-
-    bool getUsePCM() const   { return m_usePCM; }
-
-    void setPCMLog2MaxSize(uint32_t u) { m_pcmLog2MaxSize = u; }
-
-    uint32_t getPCMLog2MaxSize() const { return m_pcmLog2MaxSize; }
-
-    void setPCMLog2MinSize(uint32_t u) { m_pcmLog2MinSize = u; }
-
-    uint32_t getPCMLog2MinSize() const { return m_pcmLog2MinSize; }
-
-    void setBitsForPOC(uint32_t u) { m_bitsForPOC = u; }
-
-    uint32_t getBitsForPOC() const { return m_bitsForPOC; }
-
-    bool getUseAMP() { return m_useAMP; }
-
-    void setUseAMP(bool b) { m_useAMP = b; }
-
-    void setQuadtreeTULog2MaxSize(uint32_t u)   { m_quadtreeTULog2MaxSize = u; }
-
-    uint32_t getQuadtreeTULog2MaxSize() const   { return m_quadtreeTULog2MaxSize; }
-
-    void setQuadtreeTULog2MinSize(uint32_t u)   { m_quadtreeTULog2MinSize = u; }
-
-    uint32_t getQuadtreeTULog2MinSize() const   { return m_quadtreeTULog2MinSize; }
-
-    void setQuadtreeTUMaxDepthInter(uint32_t u) { m_quadtreeTUMaxDepthInter = u; }
-
-    void setQuadtreeTUMaxDepthIntra(uint32_t u) { m_quadtreeTUMaxDepthIntra = u; }
-
-    uint32_t getQuadtreeTUMaxDepthInter() const   { return m_quadtreeTUMaxDepthInter; }
-
-    uint32_t getQuadtreeTUMaxDepthIntra() const    { return m_quadtreeTUMaxDepthIntra; }
-
-    void setNumReorderPics(int i, uint32_t tlayer) { m_numReorderPics[tlayer] = i; }
-
-    int  getNumReorderPics(uint32_t tlayer) const  { return m_numReorderPics[tlayer]; }
-
-    void createRPSList(int numRPS);
-
-    TComRPSList* getRPSList()                  { return &m_RPSList; }
-
-    bool      getLongTermRefsPresent() const   { return m_bLongTermRefsPresent; }
-
-    void      setLongTermRefsPresent(bool b)   { m_bLongTermRefsPresent = b; }
-
-    bool      getTMVPFlagsPresent() const   { return m_TMVPFlagsPresent; }
-
-    void      setTMVPFlagsPresent(bool b)   { m_TMVPFlagsPresent = b; }
-
-    // physical transform
-    void setMaxTrSize(uint32_t u)   { m_maxTrSize = u; }
-
-    uint32_t getMaxTrSize() const   { return m_maxTrSize; }
-
-    // Tool list
-    bool getUseLossless() const { return m_useLossless; }
-
-    void setUseLossless(bool b) { m_useLossless  = b; }
-
-    // AMP accuracy
-    int       getAMPAcc(uint32_t depth) const { return m_iAMPAcc[depth]; }
-
-    void      setAMPAcc(uint32_t depth, int iAccu) { assert(depth < g_maxCUDepth);  m_iAMPAcc[depth] = iAccu; }
-
-    // Bit-depth
-    int      getBitDepthY() const { return m_bitDepthY; }
-
-    void     setBitDepthY(int u) { m_bitDepthY = u; }
-
-    int      getBitDepthC() const { return m_bitDepthC; }
-
-    void     setBitDepthC(int u) { m_bitDepthC = u; }
-
-    int       getQpBDOffsetY() const { return m_qpBDOffsetY; }
-
-    void      setQpBDOffsetY(int value) { m_qpBDOffsetY = value; }
-
-    int       getQpBDOffsetC() const { return m_qpBDOffsetC; }
-
-    void      setQpBDOffsetC(int value) { m_qpBDOffsetC = value; }
-
-    void setUseSAO(bool bVal)  { m_bUseSAO = bVal; }
-
-    bool getUseSAO() const { return m_bUseSAO; }
-
-    uint32_t      getMaxTLayers() const                   { return m_maxTLayers; }
-
-    void      setMaxTLayers(uint32_t maxTLayers)          { assert(maxTLayers <= MAX_TLAYER); m_maxTLayers = maxTLayers; }
-
-    bool      getTemporalIdNestingFlag() const        { return m_bTemporalIdNestingFlag; }
-
-    void      setTemporalIdNestingFlag(bool bValue)   { m_bTemporalIdNestingFlag = bValue; }
-
-    uint32_t      getPCMBitDepthLuma() const { return m_pcmBitDepthLuma; }
-
-    void      setPCMBitDepthLuma(uint32_t u) { m_pcmBitDepthLuma = u; }
-
-    uint32_t      getPCMBitDepthChroma() const { return m_pcmBitDepthChroma; }
-
-    void      setPCMBitDepthChroma(uint32_t u) { m_pcmBitDepthChroma = u; }
-
-    void      setPCMFilterDisableFlag(bool bValue)    { m_bPCMFilterDisableFlag = bValue; }
-
-    bool      getPCMFilterDisableFlag() const         { return m_bPCMFilterDisableFlag; }
-
-    bool getScalingListFlag() const { return m_scalingListEnabledFlag; }
-
-    void setScalingListFlag(bool b) { m_scalingListEnabledFlag = b; }
-
-    bool getScalingListPresentFlag() const { return m_scalingListPresentFlag; }
-
-    void setScalingListPresentFlag(bool b) { m_scalingListPresentFlag = b; }
-
-    void setScalingList(TComScalingList *scalingList);
-    TComScalingList* getScalingList() { return m_scalingList; } //!< get ScalingList class pointer in SPS
-
-    uint32_t getMaxDecPicBuffering(uint32_t tlayer) { return m_maxDecPicBuffering[tlayer]; }
-
-    void setMaxDecPicBuffering(uint32_t ui, uint32_t tlayer) { m_maxDecPicBuffering[tlayer] = ui; }
-
-    uint32_t getMaxLatencyIncrease(uint32_t tlayer) { return m_maxLatencyIncrease[tlayer]; }
-
-    void setMaxLatencyIncrease(uint32_t ui, uint32_t tlayer) { m_maxLatencyIncrease[tlayer] = ui; }
-
-    void setUseStrongIntraSmoothing(bool bVal) { m_useStrongIntraSmoothing = bVal; }
-
-    bool getUseStrongIntraSmoothing() const { return m_useStrongIntraSmoothing; }
-
-    bool getVuiParametersPresentFlag() { return m_vuiParametersPresentFlag; }
-
-    void setVuiParametersPresentFlag(bool b) { m_vuiParametersPresentFlag = b; }
-
-    TComVUI* getVuiParameters() { return &m_vuiParameters; }
-
-    void setHrdParameters(uint32_t fpsNum, uint32_t fpsDenom, uint32_t numDU, uint32_t bitRate, bool randomAccess);
-
-    TComPTL* getPTL() { return &m_ptl; }
-};
-
-/// PPS class
-class TComPPS
-{
-private:
-
-    int      m_PPSId;                  // pic_parameter_set_id
-    int      m_SPSId;                  // seq_parameter_set_id
-    int      m_picInitQPMinus26;
-    bool     m_useDQP;
-    bool     m_bConstrainedIntraPred;  // constrained_intra_pred_flag
-    bool     m_bSliceChromaQpFlag;     // slicelevel_chroma_qp_flag
-
-    // access channel
-    TComSPS* m_sps;
-    uint32_t m_maxCuDQPDepth;
-    uint32_t m_minCuDQPSize;
-
-    int      m_chromaCbQpOffset;
-    int      m_chromaCrQpOffset;
-
-    uint32_t m_numRefIdxL0DefaultActive;
-    uint32_t m_numRefIdxL1DefaultActive;
-
-    bool     m_bUseWeightPred;         // Use of Weighting Prediction (P_SLICE)
-    bool     m_useWeightedBiPred;      // Use of Weighting Bi-Prediction (B_SLICE)
-    bool     m_outputFlagPresentFlag; // Indicates the presence of output_flag in slice header
-
-    bool     m_transquantBypassEnableFlag; // Indicates presence of cu_transquant_bypass_flag in CUs.
-    bool     m_useTransformSkip;
-    bool     m_entropyCodingSyncEnabledFlag; //!< Indicates the presence of wavefronts
-
-    bool     m_loopFilterAcrossTilesEnabledFlag;
-
-    int      m_signHideFlag;
-
-    bool     m_cabacInitPresentFlag;
-    uint32_t m_encCABACTableIdx;         // Used to transmit table selection across slices
-
-    bool     m_sliceHeaderExtensionPresentFlag;
-    bool     m_deblockingFilterControlPresentFlag;
-    bool     m_deblockingFilterOverrideEnabledFlag;
-    bool     m_picDisableDeblockingFilterFlag;
-    int      m_deblockingFilterBetaOffsetDiv2;  //< beta offset for deblocking filter
-    int      m_deblockingFilterTcOffsetDiv2;    //< tc offset for deblocking filter
-    bool     m_scalingListPresentFlag;
-    TComScalingList* m_scalingList; //!< ScalingList class pointer
-
-    bool     m_listsModificationPresentFlag;
-    uint32_t m_log2ParallelMergeLevelMinus2;
-    int      m_numExtraSliceHeaderBits;
-
-public:
-
-    TComPPS();
-    virtual ~TComPPS();
-
-    int       getPPSId() const { return m_PPSId; }
-
-    void      setPPSId(int i) { m_PPSId = i; }
-
-    int       getSPSId() const{ return m_SPSId; }
-
-    void      setSPSId(int i) { m_SPSId = i; }
-
-    int       getPicInitQPMinus26() const { return m_picInitQPMinus26; }
-
-    void      setPicInitQPMinus26(int i)  { m_picInitQPMinus26 = i; }
-
-    bool      getUseDQP() const           { return m_useDQP; }
-
-    void      setUseDQP(bool b)           { m_useDQP = b; }
-
-    bool      getConstrainedIntraPred() const { return m_bConstrainedIntraPred; }
-
-    void      setConstrainedIntraPred(bool b) { m_bConstrainedIntraPred = b; }
-
-    bool      getSliceChromaQpFlag() const { return m_bSliceChromaQpFlag; }
-
-    void      setSliceChromaQpFlag(bool b) { m_bSliceChromaQpFlag = b; }
-
-    void      setSPS(TComSPS* sps) { m_sps = sps; }
-
-    TComSPS*  getSPS() { return m_sps; }
-
-    void      setMaxCuDQPDepth(uint32_t u) { m_maxCuDQPDepth = u; }
-
-    uint32_t  getMaxCuDQPDepth() const { return m_maxCuDQPDepth; }
-
-    void      setMinCuDQPSize(uint32_t u) { m_minCuDQPSize = u; }
-
-    uint32_t  getMinCuDQPSize() const { return m_minCuDQPSize; }
-
-    void      setChromaCbQpOffset(int i) { m_chromaCbQpOffset = i; }
-
-    int       getChromaCbQpOffset() const { return m_chromaCbQpOffset; }
-
-    void      setChromaCrQpOffset(int i) { m_chromaCrQpOffset = i; }
-
-    int       getChromaCrQpOffset() const { return m_chromaCrQpOffset; }
-
-    void      setNumRefIdxL0DefaultActive(uint32_t i) { m_numRefIdxL0DefaultActive = i; }
-
-    uint32_t  getNumRefIdxL0DefaultActive() const     { return m_numRefIdxL0DefaultActive; }
-
-    void      setNumRefIdxL1DefaultActive(uint32_t i) { m_numRefIdxL1DefaultActive = i; }
-
-    uint32_t  getNumRefIdxL1DefaultActive() const     { return m_numRefIdxL1DefaultActive; }
-
-    bool getUseWP() const    { return m_bUseWeightPred; }
-
-    bool getWPBiPred() const { return m_useWeightedBiPred; }
-
-    void setUseWP(bool b)    { m_bUseWeightPred = b; }
-
-    void setWPBiPred(bool b) { m_useWeightedBiPred = b; }
-
-    void      setOutputFlagPresentFlag(bool b)  { m_outputFlagPresentFlag = b; }
-
-    bool      getOutputFlagPresentFlag() const  { return m_outputFlagPresentFlag; }
-
-    void      setTransquantBypassEnableFlag(bool b) { m_transquantBypassEnableFlag = b; }
-
-    bool      getTransquantBypassEnableFlag() const { return m_transquantBypassEnableFlag; }
-
-    bool      getUseTransformSkip() const { return m_useTransformSkip; }
-
-    void      setUseTransformSkip(bool b) { m_useTransformSkip = b; }
-
-    void    setLoopFilterAcrossTilesEnabledFlag(bool b) { m_loopFilterAcrossTilesEnabledFlag = b; }
-
-    bool    getLoopFilterAcrossTilesEnabledFlag()      { return m_loopFilterAcrossTilesEnabledFlag; }
-
-    bool    getEntropyCodingSyncEnabledFlag() const    { return m_entropyCodingSyncEnabledFlag; }
-
-    void    setEntropyCodingSyncEnabledFlag(bool val)  { m_entropyCodingSyncEnabledFlag = val; }
-
-    void    setSignHideFlag(int signHideFlag)       { m_signHideFlag = signHideFlag; }
-
-    int     getSignHideFlag() const                 { return m_signHideFlag; }
-
-    void     setCabacInitPresentFlag(bool flag)     { m_cabacInitPresentFlag = flag; }
-
-    void     setEncCABACTableIdx(int idx)           { m_encCABACTableIdx = idx; }
-
-    bool     getCabacInitPresentFlag() const        { return m_cabacInitPresentFlag; }
-
-    uint32_t getEncCABACTableIdx() const            { return m_encCABACTableIdx; }
-
-    void     setDeblockingFilterControlPresentFlag(bool val)  { m_deblockingFilterControlPresentFlag = val; }
-
-    bool     getDeblockingFilterControlPresentFlag() const    { return m_deblockingFilterControlPresentFlag; }
-
-    void     setDeblockingFilterOverrideEnabledFlag(bool val) { m_deblockingFilterOverrideEnabledFlag = val; }
-
-    bool     getDeblockingFilterOverrideEnabledFlag() const   { return m_deblockingFilterOverrideEnabledFlag; }
-
-    void     setPicDisableDeblockingFilterFlag(bool val)      { m_picDisableDeblockingFilterFlag = val; }     //!< set offset for deblocking filter disabled
-
-    bool     getPicDisableDeblockingFilterFlag() const        { return m_picDisableDeblockingFilterFlag; }    //!< get offset for deblocking filter disabled
-
-    void     setDeblockingFilterBetaOffsetDiv2(int val)       { m_deblockingFilterBetaOffsetDiv2 = val; }     //!< set beta offset for deblocking filter
-
-    int      getDeblockingFilterBetaOffsetDiv2() const        { return m_deblockingFilterBetaOffsetDiv2; }    //!< get beta offset for deblocking filter
-
-    void     setDeblockingFilterTcOffsetDiv2(int val)         { m_deblockingFilterTcOffsetDiv2 = val; }       //!< set tc offset for deblocking filter
-
-    int      getDeblockingFilterTcOffsetDiv2()                { return m_deblockingFilterTcOffsetDiv2; }      //!< get tc offset for deblocking filter
-
-    bool     getScalingListPresentFlag() const { return m_scalingListPresentFlag; }
-
-    void     setScalingListPresentFlag(bool b) { m_scalingListPresentFlag = b; }
-
-    void     setScalingList(TComScalingList *scalingList);
-
-    const TComScalingList* getScalingList() const { return m_scalingList; } //!< get ScalingList class pointer in PPS
-
-    bool getListsModificationPresentFlag() const { return m_listsModificationPresentFlag; }
-
-    void setListsModificationPresentFlag(bool b) { m_listsModificationPresentFlag = b; }
-
-    uint32_t getLog2ParallelMergeLevelMinus2() const { return m_log2ParallelMergeLevelMinus2; }
-
-    void setLog2ParallelMergeLevelMinus2(uint32_t mrgLevel) { m_log2ParallelMergeLevelMinus2 = mrgLevel; }
-
-    int getNumExtraSliceHeaderBits() const { return m_numExtraSliceHeaderBits; }
-
-    void setNumExtraSliceHeaderBits(int i) { m_numExtraSliceHeaderBits = i; }
-
-    bool getSliceHeaderExtensionPresentFlag() const    { return m_sliceHeaderExtensionPresentFlag; }
-
-    void setSliceHeaderExtensionPresentFlag(bool val)  { m_sliceHeaderExtensionPresentFlag = val; }
-};
-
-typedef struct wpScalingParam
-{
-    // Explicit weighted prediction parameters parsed in slice header,
-    // or Implicit weighted prediction parameters (8 bits depth values).
-    bool        bPresentFlag;
-    uint32_t    log2WeightDenom;
-    int         inputWeight;
-    int         inputOffset;
-
-    // Weighted prediction scaling values built from above parameters (bitdepth scaled):
-    int         w, o, offset, shift, round;
-
-    /* makes a non-h265 weight (i.e. fix7), into an h265 weight */
-    void setFromWeightAndOffset(int weight, int _offset, int denom = 7)
-    {
-        inputOffset = _offset;
-        log2WeightDenom = denom;
-        inputWeight = weight;
-        while (log2WeightDenom > 0 && (inputWeight > 127))
-        {
-            log2WeightDenom--;
-            inputWeight >>= 1;
-        }
-
-        inputWeight = X265_MIN(inputWeight, 127);
-    }
-} wpScalingParam;
-
-/// slice header class
-class TComSlice
-{
-private:
-
-    //  Bitstream writing
-    bool        m_saoEnabledFlag;
-    bool        m_saoEnabledFlagChroma; ///< SAO Cb&Cr enabled flag
-    int         m_ppsId;                ///< picture parameter set ID
-    bool        m_picOutputFlag;        ///< pic_output_flag
-    int         m_poc;
-    int         m_lastIDR;
-
-    TComReferencePictureSet *m_rps;
-    TComReferencePictureSet m_localRPS;
-    int         m_bdIdx;
-    NalUnitType m_nalUnitType;       ///< Nal unit type for the slice
-    SliceType   m_sliceType;
-    int         m_sliceQp;
-    bool        m_dependentSliceSegmentFlag;
-    bool        m_deblockingFilterDisable;
-    bool        m_deblockingFilterOverrideFlag;    //< offsets for deblocking filter inherit from PPS
-    int         m_deblockingFilterBetaOffsetDiv2;  //< beta offset for deblocking filter
-    int         m_deblockingFilterTcOffsetDiv2;    //< tc offset for deblocking filter
-    int         m_numRefIdx[2];     //  for multiple reference of current slice
-
-    bool        m_bCheckLDC;
-
-    //  Data
-    int         m_sliceQpDelta;
-    int         m_sliceQpDeltaCb;
-    int         m_sliceQpDeltaCr;
-    TComPic*    m_refPicList[2][MAX_NUM_REF + 1];
-    int         m_refPOCList[2][MAX_NUM_REF + 1];
-    bool        m_bIsUsedAsLongTerm[2][MAX_NUM_REF + 1];
-
-    // referenced slice?
-    bool        m_bReferenced;
-
-    // access channel
-    TComSPS*    m_sps;
-    TComPPS*    m_pps;
-    TComVPS*    m_vps;
-    TComPic*    m_pic;
-    uint32_t    m_colFromL0Flag; // collocated picture from List0 flag
-
-    uint32_t    m_colRefIdx;
-    uint32_t    m_maxNumMergeCand;
-
-    uint32_t    m_sliceCurEndCUAddr;
-    bool        m_nextSlice;
-    uint32_t    m_sliceBits;
-    uint32_t    m_sliceSegmentBits;
-    bool        m_bFinalized;
-
-    uint32_t    m_tileOffstForMultES;
-
-    uint32_t*   m_substreamSizes;
-    TComScalingList* m_scalingList; //!< pointer of quantization matrix
-    bool        m_cabacInitFlag;
-
-    bool       m_bLMvdL1Zero;
-    int        m_numEntryPointOffsets;
-    bool       m_temporalLayerNonReferenceFlag;
-
-    bool       m_enableTMVPFlag;
-
-public:
-
-    wpScalingParam  m_weightPredTable[2][MAX_NUM_REF][3]; // [REF_PIC_LIST_0 or REF_PIC_LIST_1][refIdx][0:Y, 1:U, 2:V]
-
-    TComSlice();
-    virtual ~TComSlice();
-    void      initSlice();
-
-    void      setVPS(TComVPS* vps)            { m_vps = vps; }
-
-    TComVPS*  getVPS()                        { return m_vps; }
-
-    void      setSPS(TComSPS* sps)            { m_sps = sps; }
-
-    TComSPS*  getSPS()                        { return m_sps; }
-
-    void      setPPS(TComPPS* pps)            { assert(pps != NULL); m_pps = pps; m_ppsId = pps->getPPSId(); }
-
-    TComPPS*  getPPS()                        { return m_pps; }
-
-    void      setPPSId(int ppsid)             { m_ppsId = ppsid; }
-
-    int       getPPSId()                      { return m_ppsId; }
-
-    void      setPicOutputFlag(bool b)        { m_picOutputFlag = b; }
-
-    bool      getPicOutputFlag()              { return m_picOutputFlag; }
-
-    void      setSaoEnabledFlag(bool s)       { m_saoEnabledFlag = s; }
-
-    bool      getSaoEnabledFlag()             { return m_saoEnabledFlag; }
-
-    void      setSaoEnabledFlagChroma(bool s) { m_saoEnabledFlagChroma = s; }   //!< set SAO Cb&Cr enabled flag
-
-    bool      getSaoEnabledFlagChroma()       { return m_saoEnabledFlagChroma; }      //!< get SAO Cb&Cr enabled flag
-
-    void      setRPS(TComReferencePictureSet *rps) { m_rps = rps; }
-
-    TComReferencePictureSet*  getRPS()            { return m_rps; }
-
-    TComReferencePictureSet*  getLocalRPS()       { return &m_localRPS; }
-
-    void      setRPSidx(int bdidx)                { m_bdIdx = bdidx; }
-
-    int       getRPSidx()                         { return m_bdIdx; }
-
-    void      setLastIDR(int idrPoc)              { m_lastIDR = idrPoc; }
-
-    int       getLastIDR()                        { return m_lastIDR; }
-
-    SliceType getSliceType()                      { return m_sliceType; }
-
-    int       getPOC()                            { return m_poc; }
-
-    int       getSliceQp()                        { return m_sliceQp; }
-
-    bool      getDependentSliceSegmentFlag() const   { return m_dependentSliceSegmentFlag; }
-
-    void      setDependentSliceSegmentFlag(bool val) { m_dependentSliceSegmentFlag = val; }
-
-    int       getSliceQpDelta()                   { return m_sliceQpDelta; }
-
-    int       getSliceQpDeltaCb()                 { return m_sliceQpDeltaCb; }
-
-    int       getSliceQpDeltaCr()                 { return m_sliceQpDeltaCr; }
-
-    bool      getDeblockingFilterDisable()        { return m_deblockingFilterDisable; }
-
-    bool      getDeblockingFilterOverrideFlag()   { return m_deblockingFilterOverrideFlag; }
-
-    int       getDeblockingFilterBetaOffsetDiv2() { return m_deblockingFilterBetaOffsetDiv2; }
-
-    int       getDeblockingFilterTcOffsetDiv2()   { return m_deblockingFilterTcOffsetDiv2; }
-
-    int       getNumRefIdx(int e)                 { return m_numRefIdx[e]; }
-
-    TComPic*  getPic()                            { return m_pic; }
-
-    TComPic*  getRefPic(int e, int refIdx) { return m_refPicList[e][refIdx]; }
-
-    int       getRefPOC(int e, int refIdx) { return m_refPOCList[e][refIdx]; }
-
-    uint32_t  getColFromL0Flag()                  { return m_colFromL0Flag; }
-
-    uint32_t  getColRefIdx()                      { return m_colRefIdx; }
-
-    bool      getIsUsedAsLongTerm(int i, int j)   { return m_bIsUsedAsLongTerm[i][j]; }
-
-    bool      getCheckLDC()                       { return m_bCheckLDC; }
-
-    bool      getMvdL1ZeroFlag()                  { return m_bLMvdL1Zero; }
-
-    int       getNumRpsCurrTempList();
-
-    void      setReferenced(bool b)            { m_bReferenced = b; }
-
-    bool      isReferenced()                   { return m_bReferenced; }
-
-    void      setPOC(int i)                    { m_poc = i; }
-
-    void      setNalUnitType(NalUnitType e)    { m_nalUnitType = e; }
-
-    NalUnitType getNalUnitType() const         { return m_nalUnitType; }
-
-    bool      getRapPicFlag();
-    bool      getIdrPicFlag()
-    {
-        return getNalUnitType() == NAL_UNIT_CODED_SLICE_IDR_W_RADL ||
-               getNalUnitType() == NAL_UNIT_CODED_SLICE_IDR_N_LP;
-    }
-
-    bool      isIRAP() const                   { return (getNalUnitType() >= 16) && (getNalUnitType() <= 23); }
-
-    void      setSliceType(SliceType e)               { m_sliceType = e; }
-
-    void      setSliceQp(int i)                       { m_sliceQp = i; }
-
-    void      setSliceQpDelta(int i)                  { m_sliceQpDelta = i; }
-
-    void      setSliceQpDeltaCb(int i)                { m_sliceQpDeltaCb = i; }
-
-    void      setSliceQpDeltaCr(int i)                { m_sliceQpDeltaCr = i; }
-
-    void      setDeblockingFilterDisable(bool b)      { m_deblockingFilterDisable = b; }
-
-    void      setDeblockingFilterOverrideFlag(bool b) { m_deblockingFilterOverrideFlag = b; }
-
-    void      setDeblockingFilterBetaOffsetDiv2(int i) { m_deblockingFilterBetaOffsetDiv2 = i; }
-
-    void      setDeblockingFilterTcOffsetDiv2(int i)   { m_deblockingFilterTcOffsetDiv2 = i; }
-
-    void      setRefPic(TComPic* p, int e, int refIdx) { m_refPicList[e][refIdx] = p; }
-
-    void      setRefPOC(int i, int e, int refIdx) { m_refPOCList[e][refIdx] = i; }
-
-    void      setNumRefIdx(int e, int i)   { m_numRefIdx[e] = i; }
-
-    void      setPic(TComPic* p)                  { m_pic = p; }
-
-    void      setRefPicList(PicList& picList);
-
-    void      setRefPOCList();
-
-    void      setColFromL0Flag(uint32_t colFromL0) { m_colFromL0Flag = colFromL0; }
-
-    void      setColRefIdx(uint32_t refIdx)     { m_colRefIdx = refIdx; }
-
-    void      setCheckLDC(bool b)           { m_bCheckLDC = b; }
-
-    void      setMvdL1ZeroFlag(bool b)      { m_bLMvdL1Zero = b; }
-
-    bool      isIntra()                     { return m_sliceType == I_SLICE; }
-
-    bool      isInterB()                    { return m_sliceType == B_SLICE; }
-
-    bool      isInterP()                    { return m_sliceType == P_SLICE; }
-
-    void setTLayerInfo(uint32_t tlayer);
-
-    void setMaxNumMergeCand(uint32_t val)          { m_maxNumMergeCand = val; }
-
-    uint32_t getMaxNumMergeCand()                  { return m_maxNumMergeCand; }
-
-    void setSliceCurEndCUAddr(uint32_t uiAddr)     { m_sliceCurEndCUAddr = uiAddr; }
-
-    uint32_t getSliceCurEndCUAddr()                { return m_sliceCurEndCUAddr; }
-
-    void setNextSlice(bool b)                  { m_nextSlice = b; }
-
-    bool isNextSlice()                         { return m_nextSlice; }
-
-    void setSliceBits(uint32_t val)            { m_sliceBits = val; }
-
-    uint32_t getSliceBits()                    { return m_sliceBits; }
-
-    void setSliceSegmentBits(uint32_t val)     { m_sliceSegmentBits = val; }
-
-    uint32_t getSliceSegmentBits()             { return m_sliceSegmentBits; }
-
-    void setFinalized(bool val)                { m_bFinalized = val; }
-
-    bool getFinalized()                        { return m_bFinalized; }
-
-    void  setWpScaling(wpScalingParam wp[2][MAX_NUM_REF][3]) { memcpy(m_weightPredTable, wp, sizeof(wpScalingParam) * 2 * MAX_NUM_REF * 3); }
-
-    void  getWpScaling(int e, int refIdx, wpScalingParam *&wp);
-
-    void  resetWpScaling();
-    void  initWpScaling();
-    inline bool applyWP() { return (m_sliceType == P_SLICE && m_pps->getUseWP()) || (m_sliceType == B_SLICE && m_pps->getWPBiPred()); }
-
-    void setTileOffstForMultES(uint32_t offset) { m_tileOffstForMultES = offset; }
-
-    uint32_t getTileOffstForMultES()           { return m_tileOffstForMultES; }
-
-    void allocSubstreamSizes(uint32_t uiNumSubstreams);
-    uint32_t* getSubstreamSizes()              { return m_substreamSizes; }
-
-    void  setScalingList(TComScalingList* scalingList) { m_scalingList = scalingList; }
-
-    TComScalingList*   getScalingList()        { return m_scalingList; }
-
-    void  setDefaultScalingList();
-    bool  checkDefaultScalingList();
-    void      setCabacInitFlag(bool val)   { m_cabacInitFlag = val; }   //!< set CABAC initial flag
-
-    bool      getCabacInitFlag()           { return m_cabacInitFlag; }  //!< get CABAC initial flag
-
-    void      setNumEntryPointOffsets(int val)  { m_numEntryPointOffsets = val; }
-
-    int       getNumEntryPointOffsets()         { return m_numEntryPointOffsets; }
-
-    bool      getTemporalLayerNonReferenceFlag()       { return m_temporalLayerNonReferenceFlag; }
-
-    void      setTemporalLayerNonReferenceFlag(bool x) { m_temporalLayerNonReferenceFlag = x; }
-
-    void      setEnableTMVPFlag(bool b)    { m_enableTMVPFlag = b; }
-
-    bool      getEnableTMVPFlag()          { return m_enableTMVPFlag; }
-
-protected:
-
-    TComPic*  xGetRefPic(PicList& picList, int poc);
-
-    TComPic*  xGetLongTermRefPic(PicList& picList, int poc, bool pocHasMsb);
-}; // END CLASS DEFINITION TComSlice
-}
-//! \}
-
-#endif // ifndef X265_TCOMSLICE_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComTrQuant.cpp Deleted

@@ -1,1582 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComTrQuant.cpp
-    \brief    transform and quantization class
-*/
-
-#include "TComTrQuant.h"
-#include "TComPic.h"
-#include "ContextTables.h"
-#include "primitives.h"
-
-using namespace x265;
-
-typedef struct
-{
-    int    nnzBeforePos0;
-    double codedLevelAndDist; // distortion and level cost only
-    double uncodedDist;  // all zero coded block distortion
-    double sigCost;
-    double sigCost0;
-} coeffGroupRDStats;
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-
-#define RDOQ_CHROMA 1  ///< use of RDOQ in chroma
-
-inline static int x265_min_fast(int x, int y)
-{
-    return y + ((x - y) & ((x - y) >> (sizeof(int) * CHAR_BIT - 1))); // min(x, y)
-}
-
-// ====================================================================================================================
-// TComTrQuant class member functions
-// ====================================================================================================================
-
-TComTrQuant::TComTrQuant()
-{
-    m_qpParam.clear();
-
-    // allocate temporary buffers
-    // OPT_ME: I may reduce this to short and output matched, but I am not sure it is right.
-    m_tmpCoeff = X265_MALLOC(int32_t, MAX_CU_SIZE * MAX_CU_SIZE);
-
-    // allocate bit estimation class (for RDOQ)
-    m_estBitsSbac = new estBitsSbacStruct;
-    initScalingList();
-}
-
-TComTrQuant::~TComTrQuant()
-{
-    // delete temporary buffers
-    if (m_tmpCoeff)
-    {
-        X265_FREE(m_tmpCoeff);
-    }
-
-    // delete bit estimation class
-    delete m_estBitsSbac;
-    destroyScalingList();
-}
-
-/** Set qP for Quantization.
- * \param qpy QPy
- * \param bLowpass
- * \param sliceType
- * \param ttype
- * \param qpBdOffset
- * \param chromaQPOffset
- *
- * return void
- */
-void TComTrQuant::setQPforQuant(int qpy, TextType ttype, int qpBdOffset, int chromaQPOffset, int chFmt)
-{
-    int qpScaled;
-
-    if (ttype == TEXT_LUMA)
-    {
-        qpScaled = qpy + qpBdOffset;
-    }
-    else
-    {
-        qpScaled = Clip3(-qpBdOffset, 57, qpy + chromaQPOffset);
-
-        if (qpScaled < 0)
-        {
-            qpScaled = qpScaled + qpBdOffset;
-        }
-        else
-        {
-            qpScaled = g_chromaScale[chFmt][qpScaled] + qpBdOffset;
-        }
-    }
-    m_qpParam.setQpParam(qpScaled);
-}
-
-// To minimize the distortion only. No rate is considered.
-void TComTrQuant::signBitHidingHDQ(TCoeff* qCoef, TCoeff* coef, int32_t* deltaU, const TUEntropyCodingParameters &codingParameters)
-{
-    const uint32_t log2TrSizeCG = codingParameters.log2TrSizeCG;
-
-    int lastCG = -1;
-    int absSum = 0;
-    int n;
-
-    for (int subSet = (1 << log2TrSizeCG * 2) - 1; subSet >= 0; subSet--)
-    {
-        int  subPos = subSet << LOG2_SCAN_SET_SIZE;
-        int  firstNZPosInCG = SCAN_SET_SIZE, lastNZPosInCG = -1;
-        absSum = 0;
-
-        for (n = SCAN_SET_SIZE - 1; n >= 0; --n)
-        {
-            if (qCoef[codingParameters.scan[n + subPos]])
-            {
-                lastNZPosInCG = n;
-                break;
-            }
-        }
-
-        for (n = 0; n < SCAN_SET_SIZE; n++)
-        {
-            if (qCoef[codingParameters.scan[n + subPos]])
-            {
-                firstNZPosInCG = n;
-                break;
-            }
-        }
-
-        for (n = firstNZPosInCG; n <= lastNZPosInCG; n++)
-        {
-            absSum += qCoef[codingParameters.scan[n + subPos]];
-        }
-
-        if (lastNZPosInCG >= 0 && lastCG == -1)
-        {
-            lastCG = 1;
-        }
-
-        if (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD)
-        {
-            uint32_t signbit = (qCoef[codingParameters.scan[subPos + firstNZPosInCG]] > 0 ? 0 : 1);
-            if (signbit != (absSum & 0x1)) //compare signbit with sum_parity
-            {
-                int minCostInc = MAX_INT,  minPos = -1, finalChange = 0, curCost = MAX_INT, curChange = 0;
-
-                for (n = (lastCG == 1 ? lastNZPosInCG : SCAN_SET_SIZE - 1); n >= 0; --n)
-                {
-                    uint32_t blkPos   = codingParameters.scan[n + subPos];
-                    if (qCoef[blkPos] != 0)
-                    {
-                        if (deltaU[blkPos] > 0)
-                        {
-                            curCost = -deltaU[blkPos];
-                            curChange = 1;
-                        }
-                        else
-                        {
-                            //curChange =-1;
-                            if (n == firstNZPosInCG && abs(qCoef[blkPos]) == 1)
-                            {
-                                curCost = MAX_INT;
-                            }
-                            else
-                            {
-                                curCost = deltaU[blkPos];
-                                curChange = -1;
-                            }
-                        }
-                    }
-                    else
-                    {
-                        if (n < firstNZPosInCG)
-                        {
-                            uint32_t thisSignBit = (coef[blkPos] >= 0 ? 0 : 1);
-                            if (thisSignBit != signbit)
-                            {
-                                curCost = MAX_INT;
-                            }
-                            else
-                            {
-                                curCost = -(deltaU[blkPos]);
-                                curChange = 1;
-                            }
-                        }
-                        else
-                        {
-                            curCost = -(deltaU[blkPos]);
-                            curChange = 1;
-                        }
-                    }
-
-                    if (curCost < minCostInc)
-                    {
-                        minCostInc = curCost;
-                        finalChange = curChange;
-                        minPos = blkPos;
-                    }
-                } //CG loop
-
-                if (qCoef[minPos] == 32767 || qCoef[minPos] == -32768)
-                {
-                    finalChange = -1;
-                }
-
-                if (coef[minPos] >= 0)
-                {
-                    qCoef[minPos] += finalChange;
-                }
-                else
-                {
-                    qCoef[minPos] -= finalChange;
-                }
-            } // Hide
-        }
-        if (lastCG == 1)
-        {
-            lastCG = 0;
-        }
-    } // TU loop
-}
-
-uint32_t TComTrQuant::xQuant(TComDataCU* cu, int32_t* coef, TCoeff* qCoef, int trSize,
-                             TextType ttype, uint32_t absPartIdx, int32_t *lastPos, bool curUseRDOQ)
-{
-    uint32_t acSum = 0;
-    int add = 0;
-    bool useRDOQ = (cu->getTransformSkip(absPartIdx, ttype) ? m_useRDOQTS : m_useRDOQ) && curUseRDOQ;
-
-#if _MSC_VER
-#pragma warning(disable: 4127) // conditional expression is constant
-#endif
-    if (useRDOQ && (ttype == TEXT_LUMA || RDOQ_CHROMA))
-    {
-        acSum = xRateDistOptQuant(cu, coef, qCoef, trSize, ttype, absPartIdx, lastPos);
-    }
-    else
-    {
-        const uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-        TUEntropyCodingParameters codingParameters;
-        getTUEntropyCodingParameters(cu, codingParameters, absPartIdx, log2TrSize, ttype);
-        int deltaU[32 * 32];
-
-        int scalingListType = (cu->isIntra(absPartIdx) ? 0 : 3) + ttype;
-        assert(scalingListType < 6);
-        int32_t *quantCoeff = 0;
-        quantCoeff = getQuantCoeff(scalingListType, m_qpParam.m_rem, log2TrSize - 2);
-
-        int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize; // Represents scaling through forward transform
-
-        int qbits = QUANT_SHIFT + m_qpParam.m_per + transformShift;
-        add = (cu->getSlice()->getSliceType() == I_SLICE ? 171 : 85) << (qbits - 9);
-
-        int numCoeff = 1 << log2TrSize * 2;
-        acSum += primitives.quant(coef, quantCoeff, deltaU, qCoef, qbits, add, numCoeff, lastPos);
-
-        if (cu->getSlice()->getPPS()->getSignHideFlag() && acSum >= 2)
-        {
-            signBitHidingHDQ(qCoef, coef, deltaU, codingParameters);
-        }
-    }
-    return acSum;
-}
-
-void TComTrQuant::init(uint32_t maxTrSize, int useRDOQ, int useRDOQTS, int useTransformSkipFast)
-{
-    m_maxTrSize            = maxTrSize;
-    m_useRDOQ              = useRDOQ;
-    m_useRDOQTS            = useRDOQTS;
-    m_useTransformSkipFast = useTransformSkipFast;
-}
-
-uint32_t TComTrQuant::transformNxN(TComDataCU* cu,
-                                   int16_t*    residual,
-                                   uint32_t    stride,
-                                   TCoeff*     coeff,
-                                   uint32_t    trSize,
-                                   TextType    ttype,
-                                   uint32_t    absPartIdx,
-                                   int32_t*    lastPos,
-                                   bool        useTransformSkip,
-                                   bool        curUseRDOQ)
-{
-    if (cu->getCUTransquantBypass(absPartIdx))
-    {
-        uint32_t absSum = 0;
-        for (uint32_t k = 0; k < trSize; k++)
-        {
-            for (uint32_t j = 0; j < trSize; j++)
-            {
-                coeff[k * trSize + j] = ((int16_t)residual[k * stride + j]);
-                absSum += abs(residual[k * stride + j]);
-            }
-        }
-
-        return absSum;
-    }
-
-    uint32_t mode; //luma intra pred
-    if (ttype == TEXT_LUMA && cu->getPredictionMode(absPartIdx) == MODE_INTRA)
-    {
-        mode = cu->getLumaIntraDir(absPartIdx);
-    }
-    else
-    {
-        mode = REG_DCT;
-    }
-
-    assert((cu->getSlice()->getSPS()->getMaxTrSize() >= trSize));
-    if (useTransformSkip)
-    {
-        xTransformSkip(residual, stride, m_tmpCoeff, trSize);
-    }
-    else
-    {
-        // TODO: this may need larger data types for X265_DEPTH > 8
-        const uint32_t log2BlockSize = g_convertToBit[trSize];
-        primitives.dct[DCT_4x4 + log2BlockSize - ((trSize == 4) && (mode != REG_DCT))](residual, m_tmpCoeff, stride);
-    }
-    return xQuant(cu, m_tmpCoeff, coeff, trSize, ttype, absPartIdx, lastPos, curUseRDOQ);
-}
-
-void TComTrQuant::invtransformNxN(bool transQuantBypass, uint32_t mode, int16_t* residual, uint32_t stride, TCoeff* coeff, uint32_t trSize, int scalingListType, bool useTransformSkip, int lastPos)
-{
-    if (transQuantBypass)
-    {
-        for (uint32_t k = 0; k < trSize; k++)
-        {
-            for (uint32_t j = 0; j < trSize; j++)
-            {
-                residual[k * stride + j] = (int16_t)(coeff[k * trSize + j]);
-            }
-        }
-
-        return;
-    }
-
-    // Values need to pass as input parameter in dequant
-    int per = m_qpParam.m_per;
-    int rem = m_qpParam.m_rem;
-    bool useScalingList = getUseScalingList();
-    const uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
-    int shift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - transformShift;
-    int32_t *dequantCoef = getDequantCoeff(scalingListType, m_qpParam.m_rem, log2TrSize - 2);
-
-    if (!useScalingList)
-    {
-        static const int invQuantScales[6] = { 40, 45, 51, 57, 64, 72 };
-        int scale = invQuantScales[rem] << per;
-        primitives.dequant_normal(coeff, m_tmpCoeff, trSize * trSize, scale, shift);
-    }
-    else
-    {
-        // CHECK_ME: the code is not verify since this is DEAD path
-        primitives.dequant_scaling(coeff, dequantCoef, m_tmpCoeff, trSize * trSize, per, shift);
-    }
-
-    if (useTransformSkip == true)
-    {
-        xITransformSkip(m_tmpCoeff, residual, stride, trSize);
-    }
-    else
-    {
-        // CHECK_ME: we can't here when no any coeff
-        assert(lastPos >= 0);
-
-        const uint32_t log2BlockSize = log2TrSize - 2;
-
-        // DC only
-        if (lastPos == 0 && !((trSize == 4) && (mode != REG_DCT)))
-        {
-            const int shift_1st = 7;
-            const int add_1st = 1 << (shift_1st - 1);
-            const int shift_2nd = 12 - (X265_DEPTH - 8);
-            const int add_2nd = 1 << (shift_2nd - 1);
-
-            int dc_val = (((m_tmpCoeff[0] * 64 + add_1st) >> shift_1st) * 64 + add_2nd) >> shift_2nd;
-            primitives.blockfill_s[log2BlockSize](residual, stride, dc_val);
-
-            return;
-        }
-
-        // TODO: this may need larger data types for X265_DEPTH > 8
-        primitives.idct[IDCT_4x4 + log2BlockSize - ((trSize == 4) && (mode != REG_DCT))](m_tmpCoeff, residual, stride);
-    }
-}
-
-// ------------------------------------------------------------------------------------------------
-// Logical transform
-// ------------------------------------------------------------------------------------------------
-
-/** Wrapper function between HM interface and core 4x4 transform skipping
- *  \param resiBlock input data (residual)
- *  \param psCoeff output data (transform coefficients)
- *  \param stride stride of input residual data
- *  \param size transform size (size x size)
- */
-void TComTrQuant::xTransformSkip(int16_t* resiBlock, uint32_t stride, int32_t* coeff, int trSize)
-{
-    uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-    int  shift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
-    uint32_t transformSkipShift;
-    int  j, k;
-    if (shift >= 0)
-    {
-        primitives.cvt16to32_shl(coeff, resiBlock, stride, shift, trSize);
-    }
-    else
-    {
-        //The case when X265_DEPTH > 13
-        int offset;
-        transformSkipShift = -shift;
-        offset = (1 << (transformSkipShift - 1));
-        for (j = 0; j < trSize; j++)
-        {
-            for (k = 0; k < trSize; k++)
-            {
-                coeff[j * trSize + k] = (resiBlock[j * stride + k] + offset) >> transformSkipShift;
-            }
-        }
-    }
-}
-
-/** Wrapper function between HM interface and core NxN transform skipping
- *  \param plCoef input data (coefficients)
- *  \param pResidual output data (residual)
- *  \param stride stride of input residual data
- *  \param size transform size (size x size)
- */
-void TComTrQuant::xITransformSkip(int32_t* coef, int16_t* residual, uint32_t stride, int trSize)
-{
-    uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-    int  shift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
-    int  j, k;
-    if (shift > 0)
-    {
-        primitives.cvt32to16_shr(residual, coef, stride, shift, trSize);
-    }
-    else
-    {
-        //The case when X265_DEPTH >= 13
-        uint32_t transformSkipShift = -shift;
-        for (j = 0; j < trSize; j++)
-        {
-            for (k = 0; k < trSize; k++)
-            {
-                residual[j * stride + k] =  coef[j * trSize + k] << transformSkipShift;
-            }
-        }
-    }
-}
-
-void TComTrQuant::getTUEntropyCodingParameters(TComDataCU*                cu,
-                                               TUEntropyCodingParameters &result,
-                                               uint32_t                   absPartIdx,
-                                               uint32_t                   log2TrSize,
-                                               TextType                   ttype)
-{
-    //set the group layout
-    const uint32_t log2TrSizeCG = log2TrSize - MLS_CG_LOG2_SIZE;
-    result.log2TrSize   = log2TrSize;
-    result.log2TrSizeCG = log2TrSizeCG;
-
-    //set the scan orders
-    result.scanType = COEFF_SCAN_TYPE(cu->getCoefScanIdx(absPartIdx, log2TrSize, ttype == TEXT_LUMA, cu->isIntra(absPartIdx)));
-    result.scan   = g_scanOrder[SCAN_GROUPED_4x4][result.scanType][log2TrSize];
-    result.scanCG = g_scanOrder[SCAN_UNGROUPED][result.scanType][log2TrSizeCG];
-
-    //set the significance map context selection parameters
-    TextType ctype = ttype == TEXT_LUMA ? TEXT_LUMA : TEXT_CHROMA;
-    result.ctype = ctype;
-    if (log2TrSize == 2)
-    {
-        result.firstSignificanceMapContext = significanceMapContextSetStart[ctype][CONTEXT_TYPE_4x4];
-    }
-    else if (log2TrSize == 3)
-    {
-        result.firstSignificanceMapContext = significanceMapContextSetStart[ctype][CONTEXT_TYPE_8x8];
-        if (result.scanType != SCAN_DIAG)
-            result.firstSignificanceMapContext += nonDiagonalScan8x8ContextOffset[ctype];
-    }
-    else
-    {
-        result.firstSignificanceMapContext = significanceMapContextSetStart[ctype][CONTEXT_TYPE_NxN];
-    }
-}
-
-/** RDOQ with CABAC
- * \param cu pointer to coding unit structure
- * \param plSrcCoeff pointer to input buffer
- * \param piDstCoeff reference to pointer to output buffer
- * \param width block width
- * \param height block height
- * \param uiAbsSum reference to absolute sum of quantized transform coefficient
- * \param ttype plane type / luminance or chrominance
- * \param absPartIdx absolute partition index
- * \returns void
- * Rate distortion optimized quantization for entropy
- * coding engines using probability models like CABAC
- */
-uint32_t TComTrQuant::xRateDistOptQuant(TComDataCU* cu, int32_t* srcCoeff, TCoeff* dstCoeff, uint32_t trSize,
-                                        TextType ttype, uint32_t absPartIdx, int32_t *lastPos)
-{
-    const uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-    uint32_t absSum = 0;
-    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize; // Represents scaling through forward transform
-    uint32_t       goRiceParam      = 0;
-    double     blockUncodedCost = 0;
-    int scalingListType = (cu->isIntra(absPartIdx) ? 0 : 3) + ttype;
-
-    assert(scalingListType < 6);
-
-    int qbits = QUANT_SHIFT + m_qpParam.m_per + transformShift; // Right shift of non-RDOQ quantizer;  level = (coeff*Q + offset)>>q_bits
-    double *errScaleOrg = getErrScaleCoeff(scalingListType, log2TrSize - 2, m_qpParam.m_rem);
-    int32_t *qCoefOrg = getQuantCoeff(scalingListType, m_qpParam.m_rem, log2TrSize - 2);
-    int32_t *qCoef = qCoefOrg;
-    double *errScale = errScaleOrg;
-
-    double costCoeff[32 * 32];
-    double costSig[32 * 32];
-    double costCoeff0[32 * 32];
-
-    int rateIncUp[32 * 32];
-    int rateIncDown[32 * 32];
-    int sigRateDelta[32 * 32];
-    int deltaU[32 * 32];
-    TUEntropyCodingParameters codingParameters;
-    getTUEntropyCodingParameters(cu, codingParameters, absPartIdx, log2TrSize, ttype);
-
-    const uint32_t cgSize = (1 << MLS_CG_SIZE); // 16
-    double costCoeffGroupSig[MLS_GRP_NUM];
-    uint64_t sigCoeffGroupFlag64 = 0;
-    uint32_t   ctxSet    = 0;
-    int    c1            = 1;
-    int    c2            = 0;
-    double baseCost      = 0;
-    int    lastScanPos   = -1;
-    uint32_t   c1Idx     = 0;
-    uint32_t   c2Idx     = 0;
-    int    cgLastScanPos = -1;
-    uint32_t cgNum = 1 << codingParameters.log2TrSizeCG * 2;
-
-    int scanPos;
-    coeffGroupRDStats rdStats;
-
-    for (int cgScanPos = cgNum - 1; cgScanPos >= 0; cgScanPos--)
-    {
-        const uint32_t cgBlkPos = codingParameters.scanCG[cgScanPos];
-        const uint32_t cgPosY   = cgBlkPos >> codingParameters.log2TrSizeCG;
-        const uint32_t cgPosX   = cgBlkPos - (cgPosY << codingParameters.log2TrSizeCG);
-        const uint64_t cgBlkPosMask = ((uint64_t)1 << cgBlkPos);
-        memset(&rdStats, 0, sizeof(coeffGroupRDStats));
-        assert((trSize >> 2) == (1 << codingParameters.log2TrSizeCG));
-        const int patternSigCtx = TComTrQuant::calcPatternSigCtx(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
-        for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
-        {
-            scanPos = cgScanPos * cgSize + scanPosinCG;
-            //===== quantization =====
-            uint32_t blkPos = codingParameters.scan[scanPos];
-            // set coeff
-            int Q = qCoef[blkPos];
-            double scaleFactor = errScale[blkPos];
-            int levelDouble    = srcCoeff[blkPos];
-            levelDouble        = (int)std::min<int64_t>((int64_t)abs((int)levelDouble) * Q, MAX_INT - (1 << (qbits - 1)));
-
-            uint32_t maxAbsLevel = (levelDouble + (1 << (qbits - 1))) >> qbits;
-
-            costCoeff0[scanPos] = ((uint64_t)levelDouble * levelDouble) * scaleFactor;
-            blockUncodedCost   += costCoeff0[scanPos];
-            dstCoeff[blkPos]    = maxAbsLevel;
-
-            if (maxAbsLevel > 0 && lastScanPos < 0)
-            {
-                lastScanPos   = scanPos;
-                ctxSet        = (scanPos < SCAN_SET_SIZE || ttype != TEXT_LUMA) ? 0 : 2;
-                cgLastScanPos = cgScanPos;
-            }
-
-            if (lastScanPos >= 0)
-            {
-                const uint32_t c1c2Idx = ((c1Idx - 8) >> (sizeof(int) * CHAR_BIT - 1)) + (((-(int)c2Idx) >> (sizeof(int) * CHAR_BIT - 1)) + 1) * 2;
-                const uint32_t baseLevel = ((uint32_t)0xD9 >> (c1c2Idx * 2)) & 3;  // {1, 2, 1, 3}
-                assert(C2FLAG_NUMBER == 1);
-                assert(!!(c1Idx < C1FLAG_NUMBER) == ((c1Idx - 8) >> (sizeof(int) * CHAR_BIT - 1)));
-                assert(!!(c2Idx == 0) == ((-(int)c2Idx) >> (sizeof(int) * CHAR_BIT - 1)) + 1);
-                assert(baseLevel == ((c1Idx < C1FLAG_NUMBER) ? (2 + (c2Idx == 0)) : 1));
-
-                rateIncUp[blkPos] = 0;
-                rateIncDown[blkPos] = 0;
-                deltaU[blkPos] = 0;
-                sigRateDelta[blkPos] = 0;
-
-                //===== coefficient level estimation =====
-                uint32_t level;
-                const uint32_t oneCtx = 4 * ctxSet + c1;
-                const uint32_t absCtx = ctxSet + c2;
-                const int *greaterOneBits = m_estBitsSbac->greaterOneBits[oneCtx];
-                const int *levelAbsBits = m_estBitsSbac->levelAbsBits[absCtx];
-                double curCostSig = 0;
-
-                costCoeff[scanPos] = MAX_DOUBLE;
-                if (scanPos == lastScanPos)
-                {
-                    level = xGetCodedLevel(costCoeff[scanPos], curCostSig, costSig[scanPos],
-                                           levelDouble, maxAbsLevel, baseLevel, greaterOneBits, levelAbsBits, goRiceParam,
-                                           c1c2Idx, qbits, scaleFactor, 1);
-                }
-                else
-                {
-                    const uint32_t ctxSig = getSigCtxInc(patternSigCtx, log2TrSize, trSize, blkPos, codingParameters);
-                    if (maxAbsLevel < 3)
-                    {
-                        costSig[scanPos] = xGetRateSigCoef(0, ctxSig);
-                        costCoeff[scanPos] = costCoeff0[scanPos] + costSig[scanPos];
-                    }
-                    if (maxAbsLevel != 0)
-                    {
-                        curCostSig = xGetRateSigCoef(1, ctxSig);
-                        level = xGetCodedLevel(costCoeff[scanPos], curCostSig, costSig[scanPos],
-                                               levelDouble, maxAbsLevel, baseLevel, greaterOneBits, levelAbsBits, goRiceParam,
-                                               c1c2Idx, qbits, scaleFactor, 0);
-                    }
-                    else
-                    {
-                        level = 0;
-                    }
-                    sigRateDelta[blkPos] = m_estBitsSbac->significantBits[ctxSig][1] - m_estBitsSbac->significantBits[ctxSig][0];
-                }
-                deltaU[blkPos] = (levelDouble - ((int)level << qbits)) >> (qbits - 8);
-                if (level > 0)
-                {
-                    int rateNow = xGetICRate(level, level - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx);
-                    rateIncUp[blkPos] = xGetICRate(level + 1, level + 1 - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx) - rateNow;
-                    rateIncDown[blkPos] = xGetICRate(level - 1, level - 1 - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx) - rateNow;
-                }
-                else // level == 0
-                {
-                    rateIncUp[blkPos] = greaterOneBits[0];
-                }
-                dstCoeff[blkPos] = level;
-                baseCost           += costCoeff[scanPos];
-
-                if (level >= baseLevel)
-                {
-                    if (goRiceParam < 4 && level > (3 << goRiceParam))
-                    {
-                        goRiceParam++;
-                    }
-                }
-                c1Idx -= (-(int32_t)level) >> 31;
-
-                //===== update bin model =====
-                if (level > 1)
-                {
-                    c1 = 0;
-                    c2 += (uint32_t)(c2 - 2) >> 31;
-                    c2Idx++;
-                }
-                else if ((c1 < 3) && (c1 > 0) && level)
-                {
-                    c1++;
-                }
-
-                //===== context set update =====
-                if ((scanPos % SCAN_SET_SIZE == 0) && (scanPos > 0))
-                {
-                    c2                = 0;
-                    goRiceParam     = 0;
-
-                    c1Idx   = 0;
-                    c2Idx   = 0;
-                    ctxSet = (scanPos == SCAN_SET_SIZE || ttype != TEXT_LUMA) ? 0 : 2;
-                    assert(c1 >= 0);
-                    ctxSet -= ((int32_t)(c1 - 1) >> 31);
-                    c1 = 1;
-                }
-            }
-            else
-            {
-                costCoeff[scanPos] = 0;
-                baseCost += costCoeff0[scanPos];
-            }
-            rdStats.sigCost += costSig[scanPos];
-            if (scanPosinCG == 0)
-            {
-                rdStats.sigCost0 = costSig[scanPos];
-            }
-            if (dstCoeff[blkPos])
-            {
-                sigCoeffGroupFlag64 |= cgBlkPosMask;
-                rdStats.codedLevelAndDist += costCoeff[scanPos] - costSig[scanPos];
-                rdStats.uncodedDist += costCoeff0[scanPos];
-                if (scanPosinCG != 0)
-                {
-                    rdStats.nnzBeforePos0++;
-                }
-            }
-        } //end for (scanPosinCG)
-
-        if (cgLastScanPos >= 0)
-        {
-            costCoeffGroupSig[cgScanPos] = 0;
-            if (cgScanPos)
-            {
-                if ((sigCoeffGroupFlag64 & cgBlkPosMask) == 0)
-                {
-                    uint32_t ctxSig = getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
-                    baseCost += xGetRateSigCoeffGroup(0, ctxSig) - rdStats.sigCost;
-                    costCoeffGroupSig[cgScanPos] = xGetRateSigCoeffGroup(0, ctxSig);
-                }
-                else
-                {
-                    if (cgScanPos < cgLastScanPos) //skip the last coefficient group, which will be handled together with last position below.
-                    {
-                        if (rdStats.nnzBeforePos0 == 0)
-                        {
-                            baseCost -= rdStats.sigCost0;
-                            rdStats.sigCost -= rdStats.sigCost0;
-                        }
-                        // rd-cost if SigCoeffGroupFlag = 0, initialization
-                        double costZeroCG = baseCost;
-
-                        // add SigCoeffGroupFlag cost to total cost
-                        uint32_t ctxSig = getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
-                        if (cgScanPos < cgLastScanPos)
-                        {
-                            baseCost  += xGetRateSigCoeffGroup(1, ctxSig);
-                            costZeroCG += xGetRateSigCoeffGroup(0, ctxSig);
-                            costCoeffGroupSig[cgScanPos] = xGetRateSigCoeffGroup(1, ctxSig);
-                        }
-
-                        // try to convert the current coeff group from non-zero to all-zero
-                        costZeroCG += rdStats.uncodedDist; // distortion for resetting non-zero levels to zero levels
-                        costZeroCG -= rdStats.codedLevelAndDist; // distortion and level cost for keeping all non-zero levels
-                        costZeroCG -= rdStats.sigCost; // sig cost for all coeffs, including zero levels and non-zerl levels
-
-                        // if we can save cost, change this block to all-zero block
-                        if (costZeroCG < baseCost)
-                        {
-                            sigCoeffGroupFlag64 &= ~cgBlkPosMask;
-                            baseCost = costZeroCG;
-                            if (cgScanPos < cgLastScanPos)
-                            {
-                                costCoeffGroupSig[cgScanPos] = xGetRateSigCoeffGroup(0, ctxSig);
-                            }
-                            // reset coeffs to 0 in this block
-                            for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
-                            {
-                                scanPos      = cgScanPos * cgSize + scanPosinCG;
-                                uint32_t blkPos = codingParameters.scan[scanPos];
-                                if (dstCoeff[blkPos])
-                                {
-                                    costCoeff[scanPos] = costCoeff0[scanPos];
-                                    costSig[scanPos] = 0;
-                                }
-                                dstCoeff[blkPos] = 0;
-                            }
-                        } // end if ( d64CostAllZeros < baseCost )
-                    }
-                } // end if if (sigCoeffGroupFlag[ cgBlkPos ] == 0)
-            }
-            else
-            {
-                sigCoeffGroupFlag64 |= cgBlkPosMask;
-            }
-        }
-    } //end for (cgScanPos)
-
-    //===== estimate last position =====
-    if (lastScanPos < 0)
-    {
-        return absSum;
-    }
-
-    double bestCost = 0;
-    int    ctxCbf = 0;
-    int    bestLastIdxp1 = 0;
-    if (!cu->isIntra(absPartIdx) && ttype == TEXT_LUMA && cu->getTransformIdx(absPartIdx) == 0)
-    {
-        ctxCbf    = 0;
-        bestCost  = blockUncodedCost + xGetICost(m_estBitsSbac->blockRootCbpBits[ctxCbf][0]);
-        baseCost += xGetICost(m_estBitsSbac->blockRootCbpBits[ctxCbf][1]);
-    }
-    else
-    {
-        ctxCbf    = cu->getCtxQtCbf(ttype, cu->getTransformIdx(absPartIdx));
-        bestCost  = blockUncodedCost + xGetICost(m_estBitsSbac->blockCbpBits[ctxCbf][0]);
-        baseCost += xGetICost(m_estBitsSbac->blockCbpBits[ctxCbf][1]);
-    }
-
-    bool foundLast = false;
-    for (int cgScanPos = cgLastScanPos; cgScanPos >= 0; cgScanPos--)
-    {
-        uint32_t cgBlkPos = codingParameters.scanCG[cgScanPos];
-        baseCost -= costCoeffGroupSig[cgScanPos];
-        if (sigCoeffGroupFlag64 & ((uint64_t)1 << cgBlkPos))
-        {
-            for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
-            {
-                scanPos = cgScanPos * cgSize + scanPosinCG;
-                if (scanPos > lastScanPos) continue;
-                uint32_t blkPos = codingParameters.scan[scanPos];
-                if (dstCoeff[blkPos])
-                {
-                    uint32_t posY = blkPos >> log2TrSize;
-                    uint32_t posX = blkPos - (posY << log2TrSize);
-                    double costLast = codingParameters.scanType == SCAN_VER ? xGetRateLast(posY, posX) : xGetRateLast(posX, posY);
-                    double totalCost = baseCost + costLast - costSig[scanPos];
-
-                    if (totalCost < bestCost)
-                    {
-                        bestLastIdxp1  = scanPos + 1;
-                        bestCost     = totalCost;
-                    }
-                    if (dstCoeff[blkPos] > 1)
-                    {
-                        foundLast = true;
-                        break;
-                    }
-                    baseCost -= costCoeff[scanPos];
-                    baseCost += costCoeff0[scanPos];
-                }
-                else
-                {
-                    baseCost -= costSig[scanPos];
-                }
-            } //end for
-
-            if (foundLast)
-            {
-                break;
-            }
-        } // end if (sigCoeffGroupFlag[ cgBlkPos ])
-    } // end for
-
-    for (int pos = 0; pos < bestLastIdxp1; pos++)
-    {
-        int blkPos = codingParameters.scan[pos];
-        int level  = dstCoeff[blkPos];
-        absSum += level;
-        if (level)
-            *lastPos = blkPos;
-        dstCoeff[blkPos] = (srcCoeff[blkPos] < 0) ? -level : level;
-    }
-
-    //===== clean uncoded coefficients =====
-    for (int pos = bestLastIdxp1; pos <= lastScanPos; pos++)
-    {
-        dstCoeff[codingParameters.scan[pos]] = 0;
-    }
-
-    if (cu->getSlice()->getPPS()->getSignHideFlag() && absSum >= 2)
-    {
-        int64_t rdFactor = (int64_t)(
-                g_invQuantScales[m_qpParam.rem()] * g_invQuantScales[m_qpParam.rem()] * (1 << (2 * m_qpParam.m_per))
-                / m_lambda / 16 / (1 << DISTORTION_PRECISION_ADJUSTMENT(2 * (X265_DEPTH - 8)))
-                + 0.5);
-        int lastCG = -1;
-        int tmpSum = 0;
-        int n;
-
-        for (int subSet = (trSize * trSize - 1) >> LOG2_SCAN_SET_SIZE; subSet >= 0; subSet--)
-        {
-            int subPos = subSet << LOG2_SCAN_SET_SIZE;
-            int firstNZPosInCG = SCAN_SET_SIZE, lastNZPosInCG = -1;
-            tmpSum = 0;
-
-            for (n = SCAN_SET_SIZE - 1; n >= 0; --n)
-            {
-                if (dstCoeff[codingParameters.scan[n + subPos]])
-                {
-                    lastNZPosInCG = n;
-                    break;
-                }
-            }
-
-            for (n = 0; n < SCAN_SET_SIZE; n++)
-            {
-                if (dstCoeff[codingParameters.scan[n + subPos]])
-                {
-                    firstNZPosInCG = n;
-                    break;
-                }
-            }
-
-            for (n = firstNZPosInCG; n <= lastNZPosInCG; n++)
-            {
-                tmpSum += dstCoeff[codingParameters.scan[n + subPos]];
-            }
-
-            if (lastNZPosInCG >= 0 && lastCG == -1)
-            {
-                lastCG = 1;
-            }
-
-            if (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD)
-            {
-                uint32_t signbit = (dstCoeff[codingParameters.scan[subPos + firstNZPosInCG]] > 0 ? 0 : 1);
-                if (signbit != (tmpSum & 0x1)) // hide but need tune
-                {
-                    // calculate the cost
-                    int64_t minCostInc = MAX_INT64, curCost = MAX_INT64;
-                    int minPos = -1, finalChange = 0, curChange = 0;
-
-                    for (n = (lastCG == 1 ? lastNZPosInCG : SCAN_SET_SIZE - 1); n >= 0; --n)
-                    {
-                        uint32_t blkPos   = codingParameters.scan[n + subPos];
-                        if (dstCoeff[blkPos] != 0)
-                        {
-                            int64_t costUp   = rdFactor * (-deltaU[blkPos]) + rateIncUp[blkPos];
-                            int64_t costDown = rdFactor * (deltaU[blkPos]) + rateIncDown[blkPos] -
-                                (abs(dstCoeff[blkPos]) == 1 ? ((1 << 15) + sigRateDelta[blkPos]) : 0);
-
-                            if (lastCG == 1 && lastNZPosInCG == n && abs(dstCoeff[blkPos]) == 1)
-                            {
-                                costDown -= (4 << 15);
-                            }
-
-                            if (costUp < costDown)
-                            {
-                                curCost = costUp;
-                                curChange =  1;
-                            }
-                            else
-                            {
-                                curChange = -1;
-                                if (n == firstNZPosInCG && abs(dstCoeff[blkPos]) == 1)
-                                {
-                                    curCost = MAX_INT64;
-                                }
-                                else
-                                {
-                                    curCost = costDown;
-                                }
-                            }
-                        }
-                        else
-                        {
-                            curCost = rdFactor * (-(abs(deltaU[blkPos]))) + (1 << 15) + rateIncUp[blkPos] + sigRateDelta[blkPos];
-                            curChange = 1;
-
-                            if (n < firstNZPosInCG)
-                            {
-                                uint32_t thissignbit = (srcCoeff[blkPos] >= 0 ? 0 : 1);
-                                if (thissignbit != signbit)
-                                {
-                                    curCost = MAX_INT64;
-                                }
-                            }
-                        }
-
-                        if (curCost < minCostInc)
-                        {
-                            minCostInc = curCost;
-                            finalChange = curChange;
-                            minPos = blkPos;
-                        }
-                    }
-
-                    if (dstCoeff[minPos] == 32767 || dstCoeff[minPos] == -32768)
-                    {
-                        finalChange = -1;
-                    }
-
-                    if (srcCoeff[minPos] >= 0)
-                    {
-                        dstCoeff[minPos] += finalChange;
-                    }
-                    else
-                    {
-                        dstCoeff[minPos] -= finalChange;
-                    }
-                }
-            }
-
-            if (lastCG == 1)
-            {
-                lastCG = 0;
-            }
-        }
-    }
-
-    return absSum;
-}
-
-/** Pattern decision for context derivation process of significant_coeff_flag
- * \param sigCoeffGroupFlag pointer to prior coded significant coeff group
- * \param cgPosX column of current coefficient group
- * \param cgPosY row of current coefficient group
- * \param width width of the block
- * \param height height of the block
- * \returns pattern for current coefficient group
- */
-uint32_t TComTrQuant::calcPatternSigCtx(const uint64_t sigCoeffGroupFlag64, const uint32_t cgPosX, const uint32_t cgPosY, const uint32_t log2TrSizeCG)
-{
-    if (log2TrSizeCG == 0) return 0;
-
-    const uint32_t trSizeCG = 1 << log2TrSizeCG;
-    assert(trSizeCG <= 32);
-    const uint32_t sigPos = sigCoeffGroupFlag64 >> (1 + (cgPosY << log2TrSizeCG) + cgPosX);
-    const uint32_t sigRight = ((int32_t)(cgPosX - (trSizeCG - 1)) >> 31) & (sigPos & 1);
-    const uint32_t sigLower = ((int32_t)(cgPosY - (trSizeCG - 1)) >> 31) & (sigPos >> (trSizeCG - 2)) & 2;
-
-    return sigRight + sigLower;
-}
-
-/** Context derivation process of coeff_abs_significant_flag
- * \param patternSigCtx pattern for current coefficient group
- * \param posX column of current scan position
- * \param posY row of current scan position
- * \param log2BlockSize log2 value of block size (square block)
- * \param width width of the block
- * \param height height of the block
- * \param textureType texture type (TEXT_LUMA...)
- * \returns ctxInc for current scan position
- */
-uint32_t TComTrQuant::getSigCtxInc(const uint32_t                   patternSigCtx,
-                                   const uint32_t                   log2TrSize,
-                                   const uint32_t                   trSize,
-                                   const uint32_t                   blkPos,
-                                   const TUEntropyCodingParameters &codingParameters)
-{
-    static const uint8_t ctxIndMap[16] =
-    {
-        0, 1, 4, 5,
-        2, 3, 4, 5,
-        6, 6, 8, 8,
-        7, 7, 8, 8
-    };
-
-    if (blkPos == 0) return 0; //special case for the DC context variable
-
-    if (log2TrSize == 2) //4x4
-    {
-        return ctxIndMap[blkPos];
-    }
-
-    const uint32_t posY           = blkPos >> log2TrSize;
-    const uint32_t posX           = blkPos & (trSize - 1);
-    assert((blkPos - (posY << log2TrSize)) == posX);
-
-    int posXinSubset = blkPos & 3;
-    assert((posX & 3) == (blkPos & 3));
-    int posYinSubset = posY & 3;
-
-    // NOTE: [patternSigCtx][posXinSubset][posYinSubset]
-    static const uint8_t table_cnt[4][4][4] =
-    {
-        // patternSigCtx = 0
-        {
-            { 2, 1, 1, 0 },
-            { 1, 1, 0, 0 },
-            { 1, 0, 0, 0 },
-            { 0, 0, 0, 0 },
-        },
-        // patternSigCtx = 1
-        {
-            { 2, 1, 0, 0 },
-            { 2, 1, 0, 0 },
-            { 2, 1, 0, 0 },
-            { 2, 1, 0, 0 },
-        },
-        // patternSigCtx = 2
-        {
-            { 2, 2, 2, 2 },
-            { 1, 1, 1, 1 },
-            { 0, 0, 0, 0 },
-            { 0, 0, 0, 0 },
-        },
-        // patternSigCtx = 3
-        {
-            { 2, 2, 2, 2 },
-            { 2, 2, 2, 2 },
-            { 2, 2, 2, 2 },
-            { 2, 2, 2, 2 },
-        }
-    };
-
-    int cnt = table_cnt[patternSigCtx][posXinSubset][posYinSubset];
-    int offset = codingParameters.firstSignificanceMapContext;
-
-    offset += cnt;
-
-    return (codingParameters.ctype == TEXT_LUMA && (posX | posY) >= 4) ? 3 + offset : offset;
-}
-
-/** Get the best level in RD sense
- * \param codedCost reference to coded cost
- * \param codedCost0 reference to cost when coefficient is 0
- * \param codedCostSig reference to cost of significant coefficient
- * \param levelDouble reference to unscaled quantized level
- * \param maxAbsLevel scaled quantized level
- * \param ctxNumSig current ctxInc for coeff_abs_significant_flag
- * \param ctxNumOne current ctxInc for coeff_abs_level_greater1 (1st bin of coeff_abs_level_minus1 in AVC)
- * \param ctxNumAbs current ctxInc for coeff_abs_level_greater2 (remaining bins of coeff_abs_level_minus1 in AVC)
- * \param absGoRice current Rice parameter for coeff_abs_level_minus3
- * \param qbits quantization step size
- * \param scaleFactor correction factor
- * \param bLast indicates if the coefficient is the last significant
- * \returns best quantized transform level for given scan position
- * This method calculates the best quantized transform level for a given scan position.
- */
-inline uint32_t TComTrQuant::xGetCodedLevel(double&  codedCost,
-                                            const double curCostSig,
-                                            double&  codedCostSig,
-                                            int      levelDouble,
-                                            uint32_t maxAbsLevel,
-                                            uint32_t baseLevel,
-                                            const int *greaterOneBits,
-                                            const int *levelAbsBits,
-                                            uint32_t absGoRice,
-                                            uint32_t c1c2Idx,
-                                            int      qbits,
-                                            double   scaleFactor,
-                                            bool     last) const
-{
-    uint32_t   bestAbsLevel = 0;
-
-    if (!last && maxAbsLevel == 0)
-    {
-        assert(0);
-    }
-
-    int32_t minAbsLevel = maxAbsLevel - 1;
-    if (minAbsLevel < 1)
-        minAbsLevel = 1;
-
-    // NOTE: (A + B) ^ 2 = (A ^ 2) + 2 * A * B + (B ^ 2)
-    assert(abs((double)levelDouble - (maxAbsLevel << qbits)) < INT_MAX);
-    const int32_t err1 = levelDouble - (maxAbsLevel << qbits);            // A
-           double err2 = (double)((int64_t)err1 * err1);                  // A^ 2
-    const int64_t err3 = (int64_t)2 * err1 * ((int64_t)1 << qbits);       // 2 * A * B
-    const int64_t err4 = ((int64_t)1 << qbits) * ((int64_t)1 << qbits);   // B ^ 2
-    const double errInc = (err3 + err4) * scaleFactor;
-
-    err2 *= scaleFactor;
-
-    double bestCodedCost = codedCost;
-    double bestCodedCostSig = codedCostSig;
-    int diffLevel = maxAbsLevel - baseLevel;
-    for (int absLevel = maxAbsLevel; absLevel >= minAbsLevel; absLevel--)
-    {
-        assert(fabs((double)err2 - double(levelDouble  - (absLevel << qbits)) * double(levelDouble  - (absLevel << qbits)) * scaleFactor) < 1e-5);
-        double curCost = err2 + xGetICRateCost(absLevel, diffLevel, greaterOneBits, levelAbsBits, absGoRice, c1c2Idx);
-        curCost       += curCostSig;
-
-        if (curCost < bestCodedCost)
-        {
-            bestAbsLevel = absLevel;
-            bestCodedCost = curCost;
-            bestCodedCostSig = curCostSig;
-        }
-        err2 += errInc;
-        diffLevel--;
-    }
-    codedCost = bestCodedCost;
-    codedCostSig = bestCodedCostSig;
-    return bestAbsLevel;
-}
-
-/** Calculates the cost for specific absolute transform level
- * \param absLevel scaled quantized level
- * \param ctxNumOne current ctxInc for coeff_abs_level_greater1 (1st bin of coeff_abs_level_minus1 in AVC)
- * \param ctxNumAbs current ctxInc for coeff_abs_level_greater2 (remaining bins of coeff_abs_level_minus1 in AVC)
- * \param absGoRice Rice parameter for coeff_abs_level_minus3
- * \returns cost of given absolute transform level
- */
-inline double TComTrQuant::xGetICRateCost(uint32_t absLevel,
-                                          int32_t  diffLevel,
-                                          const int *greaterOneBits,
-                                          const int *levelAbsBits,
-                                          uint32_t absGoRice,
-                                          uint32_t c1c2Idx) const
-{
-    assert(absLevel > 0);
-    uint32_t rate = xGetIEPRate();
-
-    if (diffLevel < 0)
-    {
-        assert((absLevel == 1) || (absLevel == 2));
-        rate += greaterOneBits[(absLevel == 2)];
-
-        if (absLevel == 2)
-        {
-            rate += levelAbsBits[0];
-        }
-    }
-    else
-    {
-        uint32_t symbol = diffLevel;
-        uint32_t length;
-        if ((symbol >> absGoRice) < COEF_REMAIN_BIN_REDUCTION)
-        {
-            length = symbol >> absGoRice;
-            rate += (length + 1 + absGoRice) << 15;
-        }
-        else
-        {
-            length = 0;
-            symbol = (symbol >> absGoRice) - COEF_REMAIN_BIN_REDUCTION;
-            if (symbol != 0)
-            {
-                unsigned long idx;
-                CLZ32(idx, symbol + 1);
-                length = idx;
-            }
-
-            rate += (COEF_REMAIN_BIN_REDUCTION + length + absGoRice + 1 + length) << 15;
-        }
-        if (c1c2Idx & 1)
-        {
-            rate += greaterOneBits[1];
-        }
-
-        if (c1c2Idx == 3)
-        {
-            rate += levelAbsBits[1];
-        }
-    }
-
-    return xGetICost(rate);
-}
-
-inline int TComTrQuant::xGetICRate(uint32_t absLevel,
-                                   int32_t  diffLevel,
-                                   const int *greaterOneBits,
-                                   const int *levelAbsBits,
-                                   uint32_t absGoRice,
-                                   uint32_t c1c2Idx) const
-{
-    assert(c1c2Idx <= 3);
-    assert(absGoRice <= 4);
-    if (absLevel == 0)
-    {
-        assert(diffLevel < 0);
-        return 0;
-    }
-    int rate = 0;
-
-    if (diffLevel < 0)
-    {
-        assert(absLevel >= 0 && absLevel <= 2);
-        rate += greaterOneBits[(absLevel == 2)];
-
-        if (absLevel == 2)
-        {
-            rate += levelAbsBits[0];
-        }
-    }
-    else
-    {
-        uint32_t symbol   = diffLevel;
-        const uint32_t maxVlc   = g_goRiceRange[absGoRice];
-        bool expGolomb = (symbol > maxVlc);
-
-        if (expGolomb)
-        {
-            absLevel = symbol - maxVlc;
-
-            // NOTE: mapping to x86 hardware instruction BSR
-            unsigned long size;
-            CLZ32(size, absLevel);
-            int egs = size * 2 + 1;
-
-            rate   += egs << 15;
-
-            // NOTE: in here, expGolomb=true means (symbol >= maxVlc + 1)
-            assert(x265_min_fast(symbol, (maxVlc + 1)) == maxVlc + 1);
-            symbol = maxVlc + 1;
-        }
-
-        uint32_t prefLen = (symbol >> absGoRice) + 1;
-        uint32_t numBins = x265_min_fast(prefLen + absGoRice, 8 /* g_goRicePrefixLen[absGoRice] + absGoRice */);
-
-        rate += numBins << 15;
-
-        if (c1c2Idx & 1)
-        {
-            rate += greaterOneBits[1];
-        }
-
-        if (c1c2Idx == 3)
-        {
-            rate += levelAbsBits[1];
-        }
-    }
-    return rate;
-}
-
-/** Calculates the cost of signaling the last significant coefficient in the block
- * \param posx X coordinate of the last significant coefficient
- * \param posy Y coordinate of the last significant coefficient
- * \returns cost of last significant coefficient
- */
-inline double TComTrQuant::xGetRateLast(uint32_t posx, uint32_t posy) const
-{
-    uint32_t ctxX = g_groupIdx[posx];
-    uint32_t ctxY = g_groupIdx[posy];
-    uint32_t cost = m_estBitsSbac->lastXBits[ctxX] + m_estBitsSbac->lastYBits[ctxY];
-
-    int32_t maskX = (int32_t)(2 - posx) >> 31;
-    int32_t maskY = (int32_t)(2 - posy) >> 31;
-    cost += maskX & (xGetIEPRate() * ((ctxX - 2) >> 1));
-    cost += maskY & (xGetIEPRate() * ((ctxY - 2) >> 1));
-    return xGetICost(cost);
-}
-
-/** Context derivation process of coeff_abs_significant_flag
- * \param sigCoeffGroupFlag significance map of L1
- * \param uiBlkX column of current scan position
- * \param uiBlkY row of current scan position
- * \param uiLog2BlkSize log2 value of block size
- * \returns ctxInc for current scan position
- */
-uint32_t TComTrQuant::getSigCoeffGroupCtxInc(const uint64_t  sigCoeffGroupFlag64,
-                                             const uint32_t  cgPosX,
-                                             const uint32_t  cgPosY,
-                                             const uint32_t  log2TrSizeCG)
-{
-    const uint32_t trSizeCG = 1 << log2TrSizeCG;
-    assert(trSizeCG <= 32);
-    const uint32_t sigPos = sigCoeffGroupFlag64 >> (1 + (cgPosY << log2TrSizeCG) + cgPosX);
-    const uint32_t sigRight = ((int32_t)(cgPosX - (trSizeCG - 1)) >> 31) & sigPos;
-    const uint32_t sigLower = ((int32_t)(cgPosY - (trSizeCG - 1)) >> 31) & (sigPos >> (trSizeCG - 1));
-
-    return (sigRight | sigLower) & 1;
-}
-
-/** set quantized matrix coefficient for encode
- * \param scalingList quantized matrix address
- */
-void TComTrQuant::setScalingList(TComScalingList *scalingList)
-{
-    uint32_t size, list;
-    uint32_t qp;
-
-    for (size = 0; size < SCALING_LIST_SIZE_NUM; size++)
-    {
-        for (list = 0; list < g_scalingListNum[size]; list++)
-        {
-            for (qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
-            {
-                xSetScalingListEnc(scalingList, list, size, qp);
-                xSetScalingListDec(scalingList, list, size, qp);
-                setErrScaleCoeff(list, size, qp);
-            }
-        }
-    }
-}
-
-/** set error scale coefficients
- * \param list List ID
- * \param uiSize Size
- * \param uiQP Quantization parameter
- */
-void TComTrQuant::setErrScaleCoeff(uint32_t list, uint32_t size, uint32_t qp)
-{
-    uint32_t log2TrSize = g_convertToBit[g_scalingListSizeX[size]] + 2;
-    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize; // Represents scaling through forward transform
-
-    uint32_t i, maxNumCoeff = g_scalingListSize[size];
-    int32_t *quantCoeff;
-    double *errScale;
-
-    quantCoeff   = getQuantCoeff(list, qp, size);
-    errScale     = getErrScaleCoeff(list, size, qp);
-
-    double scalingBits = (double)(1 << SCALE_BITS);               // Compensate for scaling of bitcount in Lagrange cost function
-    scalingBits = scalingBits * pow(2.0, -2.0 * transformShift);  // Compensate for scaling through forward transform
-    for (i = 0; i < maxNumCoeff; i++)
-    {
-        errScale[i] = scalingBits / quantCoeff[i] / quantCoeff[i] / (1 << DISTORTION_PRECISION_ADJUSTMENT(2 * (X265_DEPTH - 8)));
-    }
-}
-
-/** set quantized matrix coefficient for encode
- * \param scalingList quantized matrix address
- * \param listId List index
- * \param sizeId size index
- * \param uiQP Quantization parameter
- */
-void TComTrQuant::xSetScalingListEnc(TComScalingList *scalingList, uint32_t listId, uint32_t sizeId, uint32_t qp)
-{
-    uint32_t width = g_scalingListSizeX[sizeId];
-    uint32_t height = g_scalingListSizeX[sizeId];
-    uint32_t ratio = g_scalingListSizeX[sizeId] / X265_MIN(MAX_MATRIX_SIZE_NUM, (int)g_scalingListSizeX[sizeId]);
-    int32_t *quantcoeff;
-    int32_t *coeff = scalingList->getScalingListAddress(sizeId, listId);
-
-    quantcoeff   = getQuantCoeff(listId, qp, sizeId);
-    processScalingListEnc(coeff, quantcoeff, g_quantScales[qp] << 4, height, width, ratio, X265_MIN(MAX_MATRIX_SIZE_NUM, (int)g_scalingListSizeX[sizeId]), scalingList->getScalingListDC(sizeId, listId));
-}
-
-/** set quantized matrix coefficient for decode
- * \param scalingList quantized matrix address
- * \param list List index
- * \param size size index
- * \param uiQP Quantization parameter
- */
-void TComTrQuant::xSetScalingListDec(TComScalingList *scalingList, uint32_t listId, uint32_t sizeId, uint32_t qp)
-{
-    uint32_t width = g_scalingListSizeX[sizeId];
-    uint32_t height = g_scalingListSizeX[sizeId];
-    uint32_t ratio = g_scalingListSizeX[sizeId] / X265_MIN(MAX_MATRIX_SIZE_NUM, (int)g_scalingListSizeX[sizeId]);
-    int32_t *dequantcoeff;
-    int32_t *coeff = scalingList->getScalingListAddress(sizeId, listId);
-
-    dequantcoeff = getDequantCoeff(listId, qp, sizeId);
-    processScalingListDec(coeff, dequantcoeff, g_invQuantScales[qp], height, width, ratio, X265_MIN(MAX_MATRIX_SIZE_NUM, (int)g_scalingListSizeX[sizeId]), scalingList->getScalingListDC(sizeId, listId));
-}
-
-/** set flat matrix value to quantized coefficient
- */
-void TComTrQuant::setFlatScalingList()
-{
-    uint32_t size, list;
-    uint32_t qp;
-
-    for (size = 0; size < SCALING_LIST_SIZE_NUM; size++)
-    {
-        for (list = 0; list < g_scalingListNum[size]; list++)
-        {
-            for (qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
-            {
-                xsetFlatScalingList(list, size, qp);
-                setErrScaleCoeff(list, size, qp);
-            }
-        }
-    }
-}
-
-/** set flat matrix value to quantized coefficient
- * \param list List ID
- * \param uiQP Quantization parameter
- * \param uiSize Size
- */
-void TComTrQuant::xsetFlatScalingList(uint32_t list, uint32_t size, uint32_t qp)
-{
-    uint32_t i, num = g_scalingListSize[size];
-    int32_t *quantcoeff;
-    int32_t *dequantcoeff;
-    int quantScales = g_quantScales[qp];
-    int invQuantScales = g_invQuantScales[qp] << 4;
-
-    quantcoeff   = getQuantCoeff(list, qp, size);
-    dequantcoeff = getDequantCoeff(list, qp, size);
-
-    for (i = 0; i < num; i++)
-    {
-        *quantcoeff++ = quantScales;
-        *dequantcoeff++ = invQuantScales;
-    }
-}
-
-/** set quantized matrix coefficient for encode
- * \param coeff quantized matrix address
- * \param quantcoeff quantized matrix address
- * \param quantScales Q(QP%6)
- * \param height height
- * \param width width
- * \param ratio ratio for upscale
- * \param sizuNum matrix size
- * \param dc dc parameter
- */
-void TComTrQuant::processScalingListEnc(int32_t *coeff, int32_t *quantcoeff, int quantScales, uint32_t height, uint32_t width, uint32_t ratio, int sizuNum, uint32_t dc)
-{
-    int nsqth = (height < width) ? 4 : 1; // height ratio for NSQT
-    int nsqtw = (width < height) ? 4 : 1; // width ratio for NSQT
-
-    for (uint32_t j = 0; j < height; j++)
-    {
-        for (uint32_t i = 0; i < width; i++)
-        {
-            quantcoeff[j * width + i] = quantScales / coeff[sizuNum * (j * nsqth / ratio) + i * nsqtw / ratio];
-        }
-    }
-
-    if (ratio > 1)
-    {
-        quantcoeff[0] = quantScales / dc;
-    }
-}
-
-/** set quantized matrix coefficient for decode
- * \param coeff quantized matrix address
- * \param dequantcoeff quantized matrix address
- * \param invQuantScales IQ(QP%6))
- * \param height height
- * \param width width
- * \param ratio ratio for upscale
- * \param sizuNum matrix size
- * \param dc dc parameter
- */
-void TComTrQuant::processScalingListDec(int32_t *coeff, int32_t *dequantcoeff, int invQuantScales, uint32_t height, uint32_t width, uint32_t ratio, int sizuNum, uint32_t dc)
-{
-    for (uint32_t j = 0; j < height; j++)
-    {
-        for (uint32_t i = 0; i < width; i++)
-        {
-            dequantcoeff[j * width + i] = invQuantScales * coeff[sizuNum * (j / ratio) + i / ratio];
-        }
-    }
-
-    if (ratio > 1)
-    {
-        dequantcoeff[0] = invQuantScales * dc;
-    }
-}
-
-/** initialization process of scaling list array
- */
-void TComTrQuant::initScalingList()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            for (uint32_t qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
-            {
-                m_quantCoef[sizeId][listId][qp] = new int[g_scalingListSize[sizeId]];
-                m_dequantCoef[sizeId][listId][qp] = new int[g_scalingListSize[sizeId]];
-                m_errScale[sizeId][listId][qp] = new double[g_scalingListSize[sizeId]];
-            }
-        }
-    }
-}
-
-/** destroy quantization matrix array
- */
-void TComTrQuant::destroyScalingList()
-{
-    for (uint32_t sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (uint32_t listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            for (uint32_t qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
-            {
-                delete [] m_quantCoef[sizeId][listId][qp];
-                delete [] m_dequantCoef[sizeId][listId][qp];
-                delete [] m_errScale[sizeId][listId][qp];
-            }
-        }
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComTrQuant.h Deleted

@@ -1,219 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComTrQuant.h
-    \brief    transform and quantization class (header)
-*/
-
-#ifndef X265_TCOMTRQUANT_H
-#define X265_TCOMTRQUANT_H
-
-#include "CommonDef.h"
-#include "TComYuv.h"
-#include "TComDataCU.h"
-#include "ContextTables.h"
-
-namespace x265 {
-// private namespace
-
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-
-#define QP_BITS 15
-
-// ====================================================================================================================
-// Type definition
-// ====================================================================================================================
-
-typedef struct
-{
-    int significantCoeffGroupBits[NUM_SIG_CG_FLAG_CTX][2];
-    uint32_t significantBits[NUM_SIG_FLAG_CTX][2];
-    int lastXBits[10];
-    int lastYBits[10];
-    int greaterOneBits[NUM_ONE_FLAG_CTX][2];
-    int levelAbsBits[NUM_ABS_FLAG_CTX][2];
-
-    int blockCbpBits[NUM_QT_CBF_CTX][2];
-    int blockRootCbpBits[NUM_QT_ROOT_CBF_CTX][2];
-} estBitsSbacStruct;
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-class QpParam
-{
-public:
-
-    QpParam() {}
-
-    int m_qp;
-    int m_per;
-    int m_rem;
-
-    int m_bits;
-
-public:
-
-    void setQpParam(int qpScaled)
-    {
-        m_qp   = qpScaled;
-        m_per  = qpScaled / 6;
-        m_rem  = qpScaled % 6;
-        m_bits = QP_BITS + m_per;
-    }
-
-    void clear()
-    {
-        m_qp   = 0;
-        m_per  = 0;
-        m_rem  = 0;
-        m_bits = 0;
-    }
-
-    int per()   const { return m_per; }
-
-    int rem()   const { return m_rem; }
-
-    int bits()  const { return m_bits; }
-
-    int qp() { return m_qp; }
-};
-
-/// transform and quantization class
-class TComTrQuant
-{
-public:
-
-    TComTrQuant();
-    ~TComTrQuant();
-
-    // initialize class
-    void init(uint32_t maxTrSize, int useRDOQ, int useRDOQTS, int useTransformSkipFast);
-
-    // transform & inverse transform functions
-    uint32_t transformNxN(TComDataCU* cu, int16_t* residual, uint32_t stride, TCoeff* coeff, uint32_t trSize,
-                          TextType ttype, uint32_t absPartIdx, int32_t* lastPos, bool useTransformSkip = false, bool curUseRDOQ = true);
-
-    void invtransformNxN(bool transQuantBypass, uint32_t mode, int16_t* residual, uint32_t stride, TCoeff* coeff, uint32_t trSize, int scalingListType, bool useTransformSkip = false, int lastPos = MAX_INT);
-
-    // Misc functions
-    void setQPforQuant(int qpy, TextType ttype, int qpBdOffset, int chromaQPOffset, int chFmt);
-    void setLambda(double lambdaLuma, double lambdaChroma) { m_lumaLambda = lambdaLuma; m_chromaLambda = lambdaChroma; }
-
-    void selectLambda(TextType ttype) { m_lambda = (ttype == TEXT_LUMA) ? m_lumaLambda : m_chromaLambda; }
-
-    void initScalingList();
-    void destroyScalingList();
-    void setErrScaleCoeff(uint32_t list, uint32_t size, uint32_t qp);
-    double* getErrScaleCoeff(uint32_t list, uint32_t size, uint32_t qp) { return m_errScale[size][list][qp]; }   //!< get Error Scale Coefficent
-
-    int32_t* getQuantCoeff(uint32_t list, uint32_t qp, uint32_t size) { return m_quantCoef[size][list][qp]; }        //!< get Quant Coefficent
-
-    int32_t* getDequantCoeff(uint32_t list, uint32_t qp, uint32_t size) { return m_dequantCoef[size][list][qp]; }    //!< get DeQuant Coefficent
-
-    void setUseScalingList(bool bUseScalingList) { m_scalingListEnabledFlag = bUseScalingList; }
-
-    bool getUseScalingList() { return m_scalingListEnabledFlag; }
-
-    void setFlatScalingList();
-    void xsetFlatScalingList(uint32_t list, uint32_t size, uint32_t qp);
-    void xSetScalingListEnc(TComScalingList *scalingList, uint32_t list, uint32_t size, uint32_t qp);
-    void xSetScalingListDec(TComScalingList *scalingList, uint32_t list, uint32_t size, uint32_t qp);
-    void setScalingList(TComScalingList *scalingList);
-    void processScalingListEnc(int32_t *coeff, int32_t *quantcoeff, int quantScales, uint32_t height, uint32_t width, uint32_t ratio, int sizuNum, uint32_t dc);
-    void processScalingListDec(int32_t *coeff, int32_t *dequantcoeff, int invQuantScales, uint32_t height, uint32_t width, uint32_t ratio, int sizuNum, uint32_t dc);
-    static uint32_t calcPatternSigCtx(const uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, uint32_t log2TrSizeCG);
-    static uint32_t getSigCtxInc(uint32_t patternSigCtx, const uint32_t log2TrSize, const uint32_t trSize, const uint32_t blkPos, const TUEntropyCodingParameters &codingParameters);
-    static uint32_t getSigCoeffGroupCtxInc(const uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, const uint32_t log2TrSizeCG);
-    static void getTUEntropyCodingParameters(TComDataCU* cu, TUEntropyCodingParameters &result, uint32_t absPartIdx, uint32_t log2TrSize, TextType ttype);
-    estBitsSbacStruct* m_estBitsSbac;
-
-protected:
-
-    QpParam  m_qpParam;
-
-    double   m_lambda;
-    double   m_lumaLambda;
-    double   m_chromaLambda;
-
-    uint32_t     m_maxTrSize;
-    bool     m_useRDOQ;
-    bool     m_useRDOQTS;
-    bool     m_useTransformSkipFast;
-    bool     m_scalingListEnabledFlag;
-
-    int32_t*     m_tmpCoeff;
-    int32_t*     m_quantCoef[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM][SCALING_LIST_REM_NUM];     ///< array of quantization matrix coefficient 4x4
-    int32_t*     m_dequantCoef[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM][SCALING_LIST_REM_NUM];   ///< array of dequantization matrix coefficient 4x4
-
-    double  *m_errScale[SCALING_LIST_SIZE_NUM][SCALING_LIST_NUM][SCALING_LIST_REM_NUM];
-
-private:
-
-    void xTransformSkip(int16_t* resiBlock, uint32_t stride, int32_t* coeff, int trSize);
-    void signBitHidingHDQ(TCoeff* qcoeff, TCoeff* coeff, int32_t* deltaU, const TUEntropyCodingParameters &codingParameters);
-    uint32_t xQuant(TComDataCU* cu, int32_t* src, TCoeff* dst, int trSize, TextType ttype, uint32_t absPartIdx, int32_t *lastPos, bool curUseRDOQ = true);
-
-    // RDOQ functions
-    uint32_t xRateDistOptQuant(TComDataCU* cu, int32_t* srcCoeff, TCoeff* dstCoeff, uint32_t trSize, TextType ttype, uint32_t absPartIdx, int32_t *lastPos);
-
-    inline uint32_t xGetCodedLevel(double& codedCost, const double curCostSig, double& codedCostSig, int levelDouble,
-                                   uint32_t maxAbsLevel, uint32_t baseLevel, const int *greaterOneBits, const int *levelAbsBits, uint32_t absGoRice,
-                                   uint32_t c1c2Idx, int qbits, double scale, bool bLast) const;
-
-    inline double xGetICRateCost(uint32_t absLevel, int32_t  diffLevel, const int *greaterOneBits, const int *levelAbsBits, uint32_t absGoRice, uint32_t c1c2Idx) const;
-
-    inline int    xGetICRate(uint32_t absLevel, int32_t diffLevel, const int *greaterOneBits, const int *levelAbsBits, uint32_t absGoRice, uint32_t c1c2Idx) const;
-
-    inline double xGetRateLast(uint32_t posx, uint32_t posy) const;
-
-    inline double xGetRateSigCoeffGroup(uint16_t sigCoeffGroup, uint16_t ctxNumSig) const { return m_lambda * m_estBitsSbac->significantCoeffGroupBits[ctxNumSig][sigCoeffGroup]; }
-
-    inline double xGetRateSigCoef(uint32_t sig, uint32_t ctxNumSig) const { return m_lambda * m_estBitsSbac->significantBits[ctxNumSig][sig]; }
-
-    inline double xGetICost(double rate) const { return m_lambda * rate; } ///< Get the cost for a specific rate
-
-    inline uint32_t xGetIEPRate() const          { return 32768; }            ///< Get the cost of an equal probable bit
-
-    void xITransformSkip(int32_t* coeff, int16_t* residual, uint32_t stride, int trSize);
-};
-}
-//! \}
-
-#endif // ifndef X265_TCOMTRQUANT_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComWeightPrediction.cpp Deleted

@@ -1,699 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComWeightPrediction.h
-    \brief    weighting prediction class (header)
-*/
-
-// Include files
-#include "TComSlice.h"
-#include "TComWeightPrediction.h"
-#include "TComPrediction.h"
-#include "primitives.h"
-
-using namespace x265;
-
-static inline Pel weightBidirY(int w0, int16_t P0, int w1, int16_t P1, int round, int shift, int offset)
-{
-    return ClipY(((w0 * (P0 + IF_INTERNAL_OFFS) + w1 * (P1 + IF_INTERNAL_OFFS) + round + (offset << (shift - 1))) >> shift));
-}
-
-static inline Pel weightBidirC(int w0, int16_t P0, int w1, int16_t P1, int round, int shift, int offset)
-{
-    return ClipC(((w0 * (P0 + IF_INTERNAL_OFFS) + w1 * (P1 + IF_INTERNAL_OFFS) + round + (offset << (shift - 1))) >> shift));
-}
-
-static inline Pel weightUnidirY(int w0, int16_t P0, int round, int shift, int offset)
-{
-    return ClipY(((w0 * (P0 + IF_INTERNAL_OFFS) + round) >> shift) + offset);
-}
-
-static inline Pel weightUnidirC(int w0, int16_t P0, int round, int shift, int offset)
-{
-    return ClipC(((w0 * (P0 + IF_INTERNAL_OFFS) + round) >> shift) + offset);
-}
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-TComWeightPrediction::TComWeightPrediction()
-{}
-
-/** weighted averaging for bi-pred
- * \param TComYuv* srcYuv0
- * \param TComYuv* srcYuv1
- * \param partUnitIdx
- * \param width
- * \param height
- * \param wpScalingParam *wp0
- * \param wpScalingParam *wp1
- * \param TComYuv* outDstYuv
- * \returns void
- */
-void TComWeightPrediction::addWeightBi(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, wpScalingParam *wp1, TComYuv* outDstYuv, bool bRound, bool bLuma, bool bChroma)
-{
-    int x, y;
-
-    Pel* srcY0  = srcYuv0->getLumaAddr(partUnitIdx);
-    Pel* srcU0  = srcYuv0->getCbAddr(partUnitIdx);
-    Pel* srcV0  = srcYuv0->getCrAddr(partUnitIdx);
-
-    Pel* srcY1  = srcYuv1->getLumaAddr(partUnitIdx);
-    Pel* srcU1  = srcYuv1->getCbAddr(partUnitIdx);
-    Pel* srcV1  = srcYuv1->getCrAddr(partUnitIdx);
-
-    Pel* pDstY  = outDstYuv->getLumaAddr(partUnitIdx);
-    Pel* dstU   = outDstYuv->getCbAddr(partUnitIdx);
-    Pel* dstV   = outDstYuv->getCrAddr(partUnitIdx);
-
-    if (bLuma)
-    {
-        // Luma : --------------------------------------------
-        int w0      = wp0[0].w;
-        int offset  = wp0[0].offset;
-        int shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        int shift   = wp0[0].shift + shiftNum;
-        int round   = shift ? (1 << (shift - 1)) * bRound : 0;
-        int w1      = wp1[0].w;
-
-        uint32_t  src0Stride = srcYuv0->getStride();
-        uint32_t  src1Stride = srcYuv1->getStride();
-        uint32_t  dststride  = outDstYuv->getStride();
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: luma min width is 4
-                pDstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                pDstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                pDstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                pDstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-            }
-
-            srcY0 += src0Stride;
-            srcY1 += src1Stride;
-            pDstY  += dststride;
-        }
-    }
-
-    if (bChroma)
-    {
-        // Chroma U : --------------------------------------------
-        int w0      = wp0[1].w;
-        int offset  = wp0[1].offset;
-        int shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        int shift   = wp0[1].shift + shiftNum;
-        int round   = shift ? (1 << (shift - 1)) : 0;
-        int w1      = wp1[1].w;
-
-        uint32_t src0Stride = srcYuv0->getCStride();
-        uint32_t src1Stride = srcYuv1->getCStride();
-        uint32_t dststride  = outDstYuv->getCStride();
-
-        width  >>= 1;
-        height >>= 1;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstU[x] = weightBidirC(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
-                x--;
-                dstU[x] = weightBidirC(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
-                x--;
-            }
-
-            srcU0 += src0Stride;
-            srcU1 += src1Stride;
-            dstU  += dststride;
-        }
-
-        // Chroma V : --------------------------------------------
-        w0      = wp0[2].w;
-        offset  = wp0[2].offset;
-        shift   = wp0[2].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-        w1      = wp1[2].w;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstV[x] = weightBidirC(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
-                x--;
-                dstV[x] = weightBidirC(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
-                x--;
-            }
-
-            srcV0 += src0Stride;
-            srcV1 += src1Stride;
-            dstV  += dststride;
-        }
-    }
-}
-
-/** weighted averaging for bi-pred
- * \param TShortYuv* srcYuv0
- * \param TShortYuv* srcYuv1
- * \param partUnitIdx
- * \param width
- * \param height
- * \param wpScalingParam *wp0
- * \param wpScalingParam *wp1
- * \param TComYuv* outDstYuv
- * \returns void
- */
-void TComWeightPrediction::addWeightBi(ShortYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, wpScalingParam *wp1, TComYuv* outDstYuv, bool bRound, bool bLuma, bool bChroma)
-{
-    int x, y;
-
-    int w0, w1, offset, shiftNum, shift, round;
-    uint32_t src0Stride, src1Stride, dststride;
-
-    int16_t* srcY0  = srcYuv0->getLumaAddr(partUnitIdx);
-    int16_t* srcU0  = srcYuv0->getCbAddr(partUnitIdx);
-    int16_t* srcV0  = srcYuv0->getCrAddr(partUnitIdx);
-
-    int16_t* srcY1  = srcYuv1->getLumaAddr(partUnitIdx);
-    int16_t* srcU1  = srcYuv1->getCbAddr(partUnitIdx);
-    int16_t* srcV1  = srcYuv1->getCrAddr(partUnitIdx);
-
-    Pel* dstY   = outDstYuv->getLumaAddr(partUnitIdx);
-    Pel* dstU   = outDstYuv->getCbAddr(partUnitIdx);
-    Pel* dstV   = outDstYuv->getCrAddr(partUnitIdx);
-
-    if (bLuma)
-    {
-        // Luma : --------------------------------------------
-        w0      = wp0[0].w;
-        offset  = wp0[0].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[0].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) * bRound : 0;
-        w1      = wp1[0].w;
-
-        src0Stride = srcYuv0->m_width;
-        src1Stride = srcYuv1->m_width;
-        dststride  = outDstYuv->getStride();
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: luma min width is 4
-                dstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                dstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                dstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-                dstY[x] = weightBidirY(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
-                x--;
-            }
-
-            srcY0 += src0Stride;
-            srcY1 += src1Stride;
-            dstY  += dststride;
-        }
-    }
-
-    if (bChroma)
-    {
-        // Chroma U : --------------------------------------------
-        w0      = wp0[1].w;
-        offset  = wp0[1].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[1].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-        w1      = wp1[1].w;
-
-        src0Stride = srcYuv0->m_cwidth;
-        src1Stride = srcYuv1->m_cwidth;
-        dststride  = outDstYuv->getCStride();
-
-        width  >>= 1;
-        height >>= 1;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstU[x] = weightBidirC(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
-                x--;
-                dstU[x] = weightBidirC(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
-                x--;
-            }
-
-            srcU0 += src0Stride;
-            srcU1 += src1Stride;
-            dstU  += dststride;
-        }
-
-        // Chroma V : --------------------------------------------
-        w0      = wp0[2].w;
-        offset  = wp0[2].offset;
-        shift   = wp0[2].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-        w1      = wp1[2].w;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstV[x] = weightBidirC(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
-                x--;
-                dstV[x] = weightBidirC(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
-                x--;
-            }
-
-            srcV0 += src0Stride;
-            srcV1 += src1Stride;
-            dstV  += dststride;
-        }
-    }
-}
-
-/** weighted averaging for uni-pred
- * \param TComYuv* srcYuv0
- * \param partUnitIdx
- * \param width
- * \param height
- * \param wpScalingParam *wp0
- * \param TComYuv* outDstYuv
- * \returns void
- */
-void TComWeightPrediction::addWeightUni(TComYuv* srcYuv0, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, TComYuv* outDstYuv, bool bLuma, bool bChroma)
-{
-    int x, y;
-
-    int w0, offset, shiftNum, shift, round;
-    uint32_t src0Stride, dststride;
-
-    Pel* srcY0  = srcYuv0->getLumaAddr(partUnitIdx);
-    Pel* srcU0  = srcYuv0->getCbAddr(partUnitIdx);
-    Pel* srcV0  = srcYuv0->getCrAddr(partUnitIdx);
-
-    Pel* dstY   = outDstYuv->getLumaAddr(partUnitIdx);
-    Pel* dstU   = outDstYuv->getCbAddr(partUnitIdx);
-    Pel* dstV   = outDstYuv->getCrAddr(partUnitIdx);
-
-    if (bLuma)
-    {
-        // Luma : --------------------------------------------
-        w0      = wp0[0].w;
-        offset  = wp0[0].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[0].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-        src0Stride = srcYuv0->getStride();
-        dststride  = outDstYuv->getStride();
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: luma min width is 4
-                dstY[x] = weightUnidirY(w0, srcY0[x], round, shift, offset);
-                x--;
-                dstY[x] = weightUnidirY(w0, srcY0[x], round, shift, offset);
-                x--;
-                dstY[x] = weightUnidirY(w0, srcY0[x], round, shift, offset);
-                x--;
-                dstY[x] = weightUnidirY(w0, srcY0[x], round, shift, offset);
-                x--;
-            }
-
-            srcY0 += src0Stride;
-            dstY  += dststride;
-        }
-    }
-
-    if (bChroma)
-    {
-        // Chroma U : --------------------------------------------
-        w0      = wp0[1].w;
-        offset  = wp0[1].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[1].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-
-        src0Stride = srcYuv0->getCStride();
-        dststride  = outDstYuv->getCStride();
-
-        width  >>= 1;
-        height >>= 1;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstU[x] = weightUnidirC(w0, srcU0[x], round, shift, offset);
-                x--;
-                dstU[x] = weightUnidirC(w0, srcU0[x], round, shift, offset);
-                x--;
-            }
-
-            srcU0 += src0Stride;
-            dstU  += dststride;
-        }
-
-        // Chroma V : --------------------------------------------
-        w0      = wp0[2].w;
-        offset  = wp0[2].offset;
-        shift   = wp0[2].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstV[x] = weightUnidirC(w0, srcV0[x], round, shift, offset);
-                x--;
-                dstV[x] = weightUnidirC(w0, srcV0[x], round, shift, offset);
-                x--;
-            }
-
-            srcV0 += src0Stride;
-            dstV  += dststride;
-        }
-    }
-}
-
-/** weighted averaging for uni-pred
- * \param TShortYUV* srcYuv0
- * \param partUnitIdx
- * \param width
- * \param height
- * \param wpScalingParam *wp0
- * \param TComYuv* outDstYuv
- * \returns void
- */
-
-void TComWeightPrediction::addWeightUni(ShortYuv* srcYuv0, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, TComYuv* outDstYuv, bool bLuma, bool bChroma)
-{
-    int16_t* srcY0  = srcYuv0->getLumaAddr(partUnitIdx);
-    int16_t* srcU0  = srcYuv0->getCbAddr(partUnitIdx);
-    int16_t* srcV0  = srcYuv0->getCrAddr(partUnitIdx);
-
-    Pel* dstY   = outDstYuv->getLumaAddr(partUnitIdx);
-    Pel* dstU   = outDstYuv->getCbAddr(partUnitIdx);
-    Pel* dstV   = outDstYuv->getCrAddr(partUnitIdx);
-
-    int w0, offset, shiftNum, shift, round;
-    uint32_t srcStride, dstStride;
-
-    if (bLuma)
-    {
-        // Luma : --------------------------------------------
-        w0      = wp0[0].w;
-        offset  = wp0[0].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[0].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-        srcStride = srcYuv0->m_width;
-        dstStride  = outDstYuv->getStride();
-
-        primitives.weight_sp(srcY0, dstY, srcStride, dstStride, width, height, w0, round, shift, offset);
-    }
-
-    if (bChroma)
-    {
-        // Chroma U : --------------------------------------------
-        w0      = wp0[1].w;
-        offset  = wp0[1].offset;
-        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
-        shift   = wp0[1].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-
-        srcStride = srcYuv0->m_cwidth;
-        dstStride  = outDstYuv->getCStride();
-
-        width  >>= 1;
-        height >>= 1;
-
-        primitives.weight_sp(srcU0, dstU, srcStride, dstStride, width, height, w0, round, shift, offset);
-
-        // Chroma V : --------------------------------------------
-        w0      = wp0[2].w;
-        offset  = wp0[2].offset;
-        shift   = wp0[2].shift + shiftNum;
-        round   = shift ? (1 << (shift - 1)) : 0;
-
-        primitives.weight_sp(srcV0, dstV, srcStride, dstStride, width, height, w0, round, shift, offset);
-    }
-}
-
-//=======================================================
-//  getWpScaling()
-//=======================================================
-
-/** derivation of wp tables
- * \param TComDataCU* cu
- * \param refIdx0
- * \param refIdx1
- * \param wpScalingParam *&wp0
- * \param wpScalingParam *&wp1
- * \param ibdi
- * \returns void
- */
-void TComWeightPrediction::getWpScaling(TComDataCU* cu, int refIdx0, int refIdx1, wpScalingParam *&wp0, wpScalingParam *&wp1)
-{
-    TComSlice*      slice = cu->getSlice();
-    const TComPPS*  pps     = cu->getSlice()->getPPS();
-    bool            wpBiPred = pps->getWPBiPred();
-    wpScalingParam* pwp;
-    bool            bBiDir   = (refIdx0 >= 0 && refIdx1 >= 0);
-    bool            bUniDir  = !bBiDir;
-
-    if (bUniDir || wpBiPred)
-    { // explicit --------------------
-        if (refIdx0 >= 0)
-        {
-            slice->getWpScaling(REF_PIC_LIST_0, refIdx0, wp0);
-        }
-        if (refIdx1 >= 0)
-        {
-            slice->getWpScaling(REF_PIC_LIST_1, refIdx1, wp1);
-        }
-    }
-    else
-    {
-        assert(0);
-    }
-
-    if (refIdx0 < 0)
-    {
-        wp0 = NULL;
-    }
-    if (refIdx1 < 0)
-    {
-        wp1 = NULL;
-    }
-
-    if (bBiDir)
-    { // Bi-Dir case
-        for (int yuv = 0; yuv < 3; yuv++)
-        {
-            wp0[yuv].w      = wp0[yuv].inputWeight;
-            wp0[yuv].o      = wp0[yuv].inputOffset * (1 << (X265_DEPTH - 8));
-            wp1[yuv].w      = wp1[yuv].inputWeight;
-            wp1[yuv].o      = wp1[yuv].inputOffset * (1 << (X265_DEPTH - 8));
-            wp0[yuv].offset = wp0[yuv].o + wp1[yuv].o;
-            wp0[yuv].shift  = wp0[yuv].log2WeightDenom + 1;
-            wp0[yuv].round  = (1 << wp0[yuv].log2WeightDenom);
-            wp1[yuv].offset = wp0[yuv].offset;
-            wp1[yuv].shift  = wp0[yuv].shift;
-            wp1[yuv].round  = wp0[yuv].round;
-        }
-    }
-    else
-    { // Unidir
-        pwp = (refIdx0 >= 0) ? wp0 : wp1;
-        for (int yuv = 0; yuv < 3; yuv++)
-        {
-            pwp[yuv].w      = pwp[yuv].inputWeight;
-            pwp[yuv].offset = pwp[yuv].inputOffset * (1 << (X265_DEPTH - 8));
-            pwp[yuv].shift  = pwp[yuv].log2WeightDenom;
-            pwp[yuv].round  = (pwp[yuv].log2WeightDenom >= 1) ? (1 << (pwp[yuv].log2WeightDenom - 1)) : (0);
-        }
-    }
-}
-
-/** weighted prediction for bi-pred
- * \param TComDataCU* cu
- * \param TComYuv* srcYuv0
- * \param TComYuv* srcYuv1
- * \param refIdx0
- * \param refIdx1
- * \param partIdx
- * \param width
- * \param height
- * \param TComYuv* outDstYuv
- * \returns void
- */
-void TComWeightPrediction::xWeightedPredictionBi(TComDataCU* cu, TComYuv* srcYuv0, TComYuv* srcYuv1, int refIdx0, int refIdx1, uint32_t partIdx, int width, int height, TComYuv* outDstYuv, bool bLuma, bool bChroma)
-{
-    wpScalingParam  *pwp0, *pwp1;
-
-    getWpScaling(cu, refIdx0, refIdx1, pwp0, pwp1);
-
-    if (refIdx0 >= 0 && refIdx1 >= 0)
-    {
-        addWeightBi(srcYuv0, srcYuv1, partIdx, width, height, pwp0, pwp1, outDstYuv, true, bLuma, bChroma);
-    }
-    else if (refIdx0 >= 0 && refIdx1 <  0)
-    {
-        addWeightUni(srcYuv0, partIdx, width, height, pwp0, outDstYuv, bLuma, bChroma);
-    }
-    else if (refIdx0 <  0 && refIdx1 >= 0)
-    {
-        addWeightUni(srcYuv1, partIdx, width, height, pwp1, outDstYuv, bLuma, bChroma);
-    }
-    else
-    {
-        assert(0);
-    }
-}
-
-/** weighted prediction for bi-pred
- * \param TComDataCU* cu
- * \param TShortYuv* srcYuv0
- * \param TShortYuv* srcYuv1
- * \param refIdx0
- * \param refIdx1
- * \param partIdx
- * \param width
- * \param height
- * \param TComYuv* outDstYuv
- * \returns void
- */
-void TComWeightPrediction::xWeightedPredictionBi(TComDataCU* cu, ShortYuv* srcYuv0, ShortYuv* srcYuv1, int refIdx0, int refIdx1, uint32_t partIdx, int width, int height, TComYuv* outDstYuv, bool bLuma, bool bChroma)
-{
-    wpScalingParam  *pwp0, *pwp1;
-
-    getWpScaling(cu, refIdx0, refIdx1, pwp0, pwp1);
-
-    if (refIdx0 >= 0 && refIdx1 >= 0)
-    {
-        addWeightBi(srcYuv0, srcYuv1, partIdx, width, height, pwp0, pwp1, outDstYuv, true, bLuma, bChroma);
-    }
-    else if (refIdx0 >= 0 && refIdx1 <  0)
-    {
-        addWeightUni(srcYuv0, partIdx, width, height, pwp0, outDstYuv, bLuma, bChroma);
-    }
-    else if (refIdx0 <  0 && refIdx1 >= 0)
-    {
-        addWeightUni(srcYuv1, partIdx, width, height, pwp1, outDstYuv, bLuma, bChroma);
-    }
-    else
-    {
-        assert(0);
-    }
-}
-
-/** weighted prediction for uni-pred
- * \param TComDataCU* cu
- * \param TComYuv* srcYuv
- * \param partAddr
- * \param width
- * \param height
- * \param picList
- * \param TComYuv*& outPredYuv
- * \param partIdx
- * \param refIdx
- * \returns void
- */
-void TComWeightPrediction::xWeightedPredictionUni(TComDataCU* cu, TComYuv* srcYuv, uint32_t partAddr, int width, int height, int picList, TComYuv*& outPredYuv, int refIdx, bool bLuma, bool bChroma)
-{
-    wpScalingParam  *pwp, *pwpTmp;
-
-    if (refIdx < 0)
-    {
-        refIdx   = cu->getCUMvField(picList)->getRefIdx(partAddr);
-    }
-    assert(refIdx >= 0);
-
-    if (picList == REF_PIC_LIST_0)
-    {
-        getWpScaling(cu, refIdx, -1, pwp, pwpTmp);
-    }
-    else
-    {
-        getWpScaling(cu, -1, refIdx, pwpTmp, pwp);
-    }
-    addWeightUni(srcYuv, partAddr, width, height, pwp, outPredYuv, bLuma, bChroma);
-}
-
-/** weighted prediction for uni-pred
- * \param TComDataCU* cu
- * \param TShortYuv* srcYuv
- * \param partAddr
- * \param width
- * \param height
- * \param picList
- * \param TComYuv*& outPredYuv
- * \param partIdx
- * \param refIdx
- * \returns void
- */
-void TComWeightPrediction::xWeightedPredictionUni(TComDataCU* cu, ShortYuv* srcYuv, uint32_t partAddr, int width, int height, int picList, TComYuv*& outPredYuv, int refIdx, bool bLuma, bool bChroma)
-{
-    wpScalingParam  *pwp, *pwpTmp;
-
-    if (refIdx < 0)
-    {
-        refIdx   = cu->getCUMvField(picList)->getRefIdx(partAddr);
-    }
-    assert(refIdx >= 0);
-
-    if (picList == REF_PIC_LIST_0)
-    {
-        getWpScaling(cu, refIdx, -1, pwp, pwpTmp);
-    }
-    else
-    {
-        getWpScaling(cu, -1, refIdx, pwpTmp, pwp);
-    }
-    addWeightUni(srcYuv, partAddr, width, height, pwp, outPredYuv, bLuma, bChroma);
-}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComWeightPrediction.h Deleted

@@ -1,75 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TComWeightPrediction.h
-    \brief    weighting prediction class (header)
-*/
-
-#ifndef X265_TCOMWEIGHTPREDICTION_H
-#define X265_TCOMWEIGHTPREDICTION_H
-
-// Include files
-#include "TComPic.h"
-#include "TComMotionInfo.h"
-#include "TComPattern.h"
-
-namespace x265 {
-// private namespace
-
-class TComYuv;
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-/// weighting prediction class
-class TComWeightPrediction
-{
-public:
-
-    TComWeightPrediction();
-
-    void  getWpScaling(TComDataCU* cu, int refIdx0, int refIdx1, wpScalingParam *&wp0, wpScalingParam *&wp1);
-
-    void  addWeightBi(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, wpScalingParam *wp1, TComYuv* outDstYuv, bool bRound = true, bool bLuma = true, bool bChroma = true);
-    void  addWeightBi(ShortYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, wpScalingParam *wp1, TComYuv* outDstYuv, bool bRound = true, bool bLuma = true, bool bChroma = true);
-    void  addWeightUni(TComYuv* srcYuv0, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, TComYuv* outDstYuv, bool bLuma = true, bool bChroma = true);
-    void  addWeightUni(ShortYuv* srcYuv0, uint32_t partUnitIdx, uint32_t width, uint32_t height, wpScalingParam *wp0, TComYuv* outDstYuv, bool bLuma = true, bool bChroma = true);
-
-    void  xWeightedPredictionUni(TComDataCU* cu, TComYuv* srcYuv, uint32_t partAddr, int width, int height, int picList, TComYuv*& outPredYuv, int refIdx = -1, bool bLuma = true, bool bChroma = true);
-    void  xWeightedPredictionUni(TComDataCU* cu, ShortYuv* srcYuv, uint32_t partAddr, int width, int height, int picList, TComYuv*& outPredYuv, int refIdx = -1, bool bLuma = true, bool bChroma = true);
-    void  xWeightedPredictionBi(TComDataCU* cu, TComYuv* srcYuv0, TComYuv* srcYuv1, int refIdx0, int refIdx1, uint32_t partIdx, int width, int height, TComYuv* outDstYuv, bool bLuma = true, bool bChroma = true);
-    void  xWeightedPredictionBi(TComDataCU* cu, ShortYuv* srcYuv0, ShortYuv* srcYuv1, int refIdx0, int refIdx1, uint32_t partIdx, int width, int height, TComYuv* outDstYuv, bool bLuma = true, bool bChroma = true);
-};
-}
-
-#endif // ifndef X265_TCOMWEIGHTPREDICTION_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder Deleted

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/NALwrite.cpp Deleted

@@ -1,140 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#include "common.h"
-#include "TLibCommon/NAL.h"
-#include "TLibCommon/TComBitStream.h"
-#include "NALwrite.h"
-
-namespace x265 {
-//! \ingroup TLibEncoder
-//! \{
-
-static const char emulation_prevention_three_byte[] = { 3 };
-
-/**
- * write nalu to bytestream out, performing RBSP anti startcode
- * emulation as required.  nalu.m_RBSPayload must be byte aligned.
- * caller is responsible for freeing returned allocated pointer
- */
-uint8_t *write(const OutputNALUnit& nalu, uint32_t &packetSize)
-{
-    packetSize = 0;
-    TComOutputBitstream bsNALUHeader;
-    bsNALUHeader.write(0, 1);                        // forbidden_zero_bit
-    bsNALUHeader.write(nalu.m_nalUnitType, 6);       // nal_unit_type
-    bsNALUHeader.write(nalu.m_reservedZero6Bits, 6); // nuh_reserved_zero_6bits
-    bsNALUHeader.write(nalu.m_temporalId + 1, 3);    // nuh_temporal_id_plus1
-
-    packetSize += bsNALUHeader.getByteStreamLength();
-    uint8_t *out = (uint8_t*)malloc(packetSize);
-    ::memcpy(out, bsNALUHeader.getByteStream(), packetSize);
-
-    /* write out rsbp_byte's, inserting any required
-     * emulation_prevention_three_byte's */
-
-    /* 7.4.1 ...
-     * emulation_prevention_three_byte is a byte equal to 0x03. When an
-     * emulation_prevention_three_byte is present in the NAL unit, it shall be
-     * discarded by the decoding process.
-     * The last byte of the NAL unit shall not be equal to 0x00.
-     * Within the NAL unit, the following three-byte sequences shall not occur at
-     * any byte-aligned position:
-     *  - 0x000000
-     *  - 0x000001
-     *  - 0x000002
-     * Within the NAL unit, any four-byte sequence that starts with 0x000003
-     * other than the following sequences shall not occur at any byte-aligned
-     * position:
-     *  - 0x00000300
-     *  - 0x00000301
-     *  - 0x00000302
-     *  - 0x00000303
-     */
-    uint32_t fsize = nalu.m_bitstream.getByteStreamLength();
-    uint8_t* fifo = nalu.m_bitstream.getFIFO();
-    uint32_t emulationSize = fsize / 2;
-    uint8_t* emulation = X265_MALLOC(uint8_t, fsize + emulationSize);
-    uint32_t nalsize = 0;
-
-    if (emulation)
-    {
-        for (int count = 0; count < fsize; count++)
-        {
-            uint8_t val = fifo[count];
-            if (count > 3 && !emulation[nalsize - 2] && !emulation[nalsize - 3] && emulation[nalsize - 1] <= 0x03)
-            {
-                uint8_t tmp = emulation[nalsize - 1];
-                emulation[nalsize - 1] = emulation_prevention_three_byte[0];
-                emulation[nalsize] = tmp;
-                emulation[nalsize + 1] = val;
-                nalsize++;
-            }
-            else
-            {
-                emulation[nalsize] = val;
-            }
-            nalsize++;
-        }
-
-        uint32_t i = packetSize;
-        out = (uint8_t*)realloc(out, nalsize + 4);
-        memcpy(out + packetSize, emulation, nalsize);
-        packetSize += nalsize;
-
-        /* 7.4.1.1
-         * ... when the last byte of the RBSP data is equal to 0x00 (which can
-         * only occur when the RBSP ends in a cabac_zero_word), a final byte equal
-         * to 0x03 is appended to the end of the data.
-         */
-        if (out[packetSize - 1] == 0x00)
-        {
-            out[i] = 3;
-            packetSize += 1;
-        }
-        X265_FREE(emulation);
-    }
-
-    return out;
-}
-
-/**
- * Write rbsp_trailing_bits to bs causing it to become byte-aligned
- */
-void writeRBSPTrailingBits(TComOutputBitstream& bs)
-{
-    bs.write(1, 1);
-    bs.writeAlignZero();
-}
-}
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/NALwrite.h Deleted

@@ -1,81 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#ifndef X265_NALWRITE_H
-#define X265_NALWRITE_H
-
-#include "TLibCommon/TypeDef.h"
-#include "TLibCommon/TComBitStream.h"
-#include "TLibCommon/NAL.h"
-
-namespace x265 {
-// private namespace
-
-/**
- * A convenience wrapper to NALUnit that also provides a
- * bitstream object.
- */
-struct OutputNALUnit : public NALUnit
-{
-    /**
-     * construct an OutputNALunit structure with given header values and
-     * storage for a bitstream.  Upon construction the NALunit header is
-     * written to the bitstream.
-     */
-    OutputNALUnit(NalUnitType nalUnitType)
-        : NALUnit(nalUnitType)
-        , m_bitstream()
-    {}
-
-    void resetToType(NalUnitType nalUnitType)
-    {
-        m_nalUnitType = nalUnitType;
-        m_bitstream.clear();
-    }
-
-    TComOutputBitstream m_bitstream;
-};
-
-uint8_t *write(const OutputNALUnit& nalu, uint32_t& packetSize);
-void writeRBSPTrailingBits(TComOutputBitstream& bs);
-
-void inline NALUnitEBSP::init(const OutputNALUnit& nalu)
-{
-    m_nalUnitType = nalu.m_nalUnitType;
-    m_temporalId = nalu.m_temporalId;
-    m_reservedZero6Bits = nalu.m_reservedZero6Bits;
-    m_nalUnitData = write(nalu, m_packetSize);
-}
-}
-
-#endif // ifndef X265_NALWRITE_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/SEIwrite.cpp Deleted

@@ -1,411 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#include "TLibCommon/TComBitCounter.h"
-#include "TLibCommon/TComBitStream.h"
-#include "TLibCommon/SEI.h"
-#include "TLibCommon/TComSlice.h"
-#include "SEIwrite.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-#if ENC_DEC_TRACE
-void  xTraceSEIHeader()
-{
-    fprintf(g_hTrace, "=========== SEI message ===========\n");
-}
-
-void  xTraceSEIMessageType(SEI::PayloadType payloadType)
-{
-    switch (payloadType)
-    {
-    case SEI::DECODED_PICTURE_HASH:
-        fprintf(g_hTrace, "=========== Decoded picture hash SEI message ===========\n");
-        break;
-    case SEI::USER_DATA_UNREGISTERED:
-        fprintf(g_hTrace, "=========== User Data Unregistered SEI message ===========\n");
-        break;
-    case SEI::ACTIVE_PARAMETER_SETS:
-        fprintf(g_hTrace, "=========== Active Parameter sets SEI message ===========\n");
-        break;
-    case SEI::BUFFERING_PERIOD:
-        fprintf(g_hTrace, "=========== Buffering period SEI message ===========\n");
-        break;
-    case SEI::PICTURE_TIMING:
-        fprintf(g_hTrace, "=========== Picture timing SEI message ===========\n");
-        break;
-    case SEI::RECOVERY_POINT:
-        fprintf(g_hTrace, "=========== Recovery point SEI message ===========\n");
-        break;
-    case SEI::FRAME_PACKING:
-        fprintf(g_hTrace, "=========== Frame Packing Arrangement SEI message ===========\n");
-        break;
-    case SEI::DISPLAY_ORIENTATION:
-        fprintf(g_hTrace, "=========== Display Orientation SEI message ===========\n");
-        break;
-    case SEI::TEMPORAL_LEVEL0_INDEX:
-        fprintf(g_hTrace, "=========== Temporal Level Zero Index SEI message ===========\n");
-        break;
-    case SEI::REGION_REFRESH_INFO:
-        fprintf(g_hTrace, "=========== Gradual Decoding Refresh Information SEI message ===========\n");
-        break;
-    case SEI::DECODING_UNIT_INFO:
-        fprintf(g_hTrace, "=========== Decoding Unit Information SEI message ===========\n");
-        break;
-    case SEI::TONE_MAPPING_INFO:
-        fprintf(g_hTrace, "=========== Tone Mapping Info SEI message ===========\n");
-        break;
-    case SEI::SOP_DESCRIPTION:
-        fprintf(g_hTrace, "=========== SOP Description SEI message ===========\n");
-        break;
-    case SEI::SCALABLE_NESTING:
-        fprintf(g_hTrace, "=========== Scalable Nesting SEI message ===========\n");
-        break;
-    default:
-        fprintf(g_hTrace, "=========== Unknown SEI message ===========\n");
-        break;
-    }
-}
-
-#endif // if ENC_DEC_TRACE
-
-void SEIWriter::xWriteSEIpayloadData(const SEI& sei, TComSPS *sps)
-{
-    switch (sei.payloadType())
-    {
-    case SEI::USER_DATA_UNREGISTERED:
-        xWriteSEIuserDataUnregistered(*static_cast<const SEIuserDataUnregistered*>(&sei));
-        break;
-    case SEI::ACTIVE_PARAMETER_SETS:
-        xWriteSEIActiveParameterSets(*static_cast<const SEIActiveParameterSets*>(&sei));
-        break;
-    case SEI::DECODING_UNIT_INFO:
-        xWriteSEIDecodingUnitInfo(*static_cast<const SEIDecodingUnitInfo*>(&sei), sps);
-        break;
-    case SEI::DECODED_PICTURE_HASH:
-        xWriteSEIDecodedPictureHash(*static_cast<const SEIDecodedPictureHash*>(&sei));
-        break;
-    case SEI::BUFFERING_PERIOD:
-        xWriteSEIBufferingPeriod(*static_cast<const SEIBufferingPeriod*>(&sei), sps);
-        break;
-    case SEI::PICTURE_TIMING:
-        xWriteSEIPictureTiming(*static_cast<const SEIPictureTiming*>(&sei), sps);
-        break;
-    case SEI::RECOVERY_POINT:
-        xWriteSEIRecoveryPoint(*static_cast<const SEIRecoveryPoint*>(&sei));
-        break;
-    case SEI::DISPLAY_ORIENTATION:
-        xWriteSEIDisplayOrientation(*static_cast<const SEIDisplayOrientation*>(&sei));
-        break;
-    case SEI::REGION_REFRESH_INFO:
-        xWriteSEIGradualDecodingRefreshInfo(*static_cast<const SEIGradualDecodingRefreshInfo*>(&sei));
-        break;
-    case SEI::SOP_DESCRIPTION:
-    case SEI::TONE_MAPPING_INFO:
-    case SEI::SCALABLE_NESTING:
-    case SEI::FRAME_PACKING:
-    case SEI::TEMPORAL_LEVEL0_INDEX:
-    default:
-        assert(!"Unhandled SEI message");
-    }
-}
-
-/**
- * marshal a single SEI message sei, storing the marshalled representation
- * in bitstream bs.
- */
-void SEIWriter::writeSEImessage(TComBitIf& bs, const SEI& sei, TComSPS *sps)
-{
-    /*
-     * calculate how large the payload data is
-     * TODO: this would be far nicer if it used vectored buffers */
-    TComBitCounter bs_count;
-
-    bs_count.resetBits();
-    setBitstream(&bs_count);
-
-#if ENC_DEC_TRACE
-    bool traceEnable = g_HLSTraceEnable;
-    g_HLSTraceEnable = false;
-#endif
-    xWriteSEIpayloadData(sei, sps);
-#if ENC_DEC_TRACE
-    g_HLSTraceEnable = traceEnable;
-#endif
-
-    uint32_t payload_data_num_bits = bs_count.getNumberOfWrittenBits();
-    assert(0 == payload_data_num_bits % 8);
-
-    setBitstream(&bs);
-
-#if ENC_DEC_TRACE
-    if (g_HLSTraceEnable)
-        xTraceSEIHeader();
-#endif
-
-    uint32_t payloadType = sei.payloadType();
-    for (; payloadType >= 0xff; payloadType -= 0xff)
-    {
-        WRITE_CODE(0xff, 8, "payload_type");
-    }
-
-    WRITE_CODE(payloadType, 8, "payload_type");
-
-    uint32_t payloadSize = payload_data_num_bits / 8;
-    for (; payloadSize >= 0xff; payloadSize -= 0xff)
-    {
-        WRITE_CODE(0xff, 8, "payload_size");
-    }
-
-    WRITE_CODE(payloadSize, 8, "payload_size");
-
-    /* payloadData */
-#if ENC_DEC_TRACE
-    if (g_HLSTraceEnable)
-        xTraceSEIMessageType(sei.payloadType());
-#endif
-
-    xWriteSEIpayloadData(sei, sps);
-}
-
-/**
- * marshal a user_data_unregistered SEI message sei, storing the marshalled
- * representation in bitstream bs.
- */
-void SEIWriter::xWriteSEIuserDataUnregistered(const SEIuserDataUnregistered &sei)
-{
-    for (uint32_t i = 0; i < 16; i++)
-    {
-        WRITE_CODE(sei.uuid_iso_iec_11578[i], 8, "sei.uuid_iso_iec_11578[i]");
-    }
-
-    for (uint32_t i = 0; i < sei.userDataLength; i++)
-    {
-        WRITE_CODE(sei.userData[i], 8, "user_data");
-    }
-}
-
-/**
- * marshal a decoded picture hash SEI message, storing the marshalled
- * representation in bitstream bs.
- */
-void SEIWriter::xWriteSEIDecodedPictureHash(const SEIDecodedPictureHash& sei)
-{
-    uint32_t val;
-
-    WRITE_CODE(sei.method, 8, "hash_type");
-
-    for (int yuvIdx = 0; yuvIdx < 3; yuvIdx++)
-    {
-        if (sei.method == SEIDecodedPictureHash::MD5)
-        {
-            for (uint32_t i = 0; i < 16; i++)
-            {
-                WRITE_CODE(sei.digest[yuvIdx][i], 8, "picture_md5");
-            }
-        }
-        else if (sei.method == SEIDecodedPictureHash::CRC)
-        {
-            val = (sei.digest[yuvIdx][0] << 8)  + sei.digest[yuvIdx][1];
-            WRITE_CODE(val, 16, "picture_crc");
-        }
-        else if (sei.method == SEIDecodedPictureHash::CHECKSUM)
-        {
-            val = (sei.digest[yuvIdx][0] << 24)  + (sei.digest[yuvIdx][1] << 16) + (sei.digest[yuvIdx][2] << 8) + sei.digest[yuvIdx][3];
-            WRITE_CODE(val, 32, "picture_checksum");
-        }
-    }
-}
-
-void SEIWriter::xWriteSEIActiveParameterSets(const SEIActiveParameterSets& sei)
-{
-    WRITE_CODE(sei.activeVPSId,     4, "active_vps_id");
-    WRITE_FLAG(sei.m_fullRandomAccessFlag, "full_random_access_flag");
-    WRITE_FLAG(sei.m_noParamSetUpdateFlag, "no_param_set_update_flag");
-    WRITE_UVLC(sei.numSpsIdsMinus1,    "num_sps_ids_minus1");
-    WRITE_UVLC(sei.activeSeqParamSetId, "active_seq_param_set_id");
-
-    uint32_t bits = m_bitIf->getNumberOfWrittenBits();
-    uint32_t alignedBits = (8 - (bits & 7)) % 8;
-    if (alignedBits)
-    {
-        WRITE_FLAG(1, "alignment_bit");
-        alignedBits--;
-        while (alignedBits--)
-        {
-            WRITE_FLAG(0, "alignment_bit");
-        }
-    }
-}
-
-void SEIWriter::xWriteSEIDecodingUnitInfo(const SEIDecodingUnitInfo& sei, TComSPS *sps)
-{
-    TComVUI *vui = sps->getVuiParameters();
-
-    WRITE_UVLC(sei.m_decodingUnitIdx, "decoding_unit_idx");
-    if (vui->getHrdParameters()->getSubPicCpbParamsInPicTimingSEIFlag())
-    {
-        WRITE_CODE(sei.m_duSptCpbRemovalDelay, (vui->getHrdParameters()->getDuCpbRemovalDelayLengthMinus1() + 1), "du_spt_cpb_removal_delay");
-    }
-    WRITE_FLAG(sei.m_dpbOutputDuDelayPresentFlag, "dpb_output_du_delay_present_flag");
-    if (sei.m_dpbOutputDuDelayPresentFlag)
-    {
-        WRITE_CODE(sei.m_picSptDpbOutputDuDelay, vui->getHrdParameters()->getDpbOutputDelayDuLengthMinus1() + 1, "pic_spt_dpb_output_du_delay");
-    }
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteSEIBufferingPeriod(const SEIBufferingPeriod& sei, TComSPS *sps)
-{
-    int i, nalOrVcl;
-    TComVUI *vui = sps->getVuiParameters();
-    TComHRD *hrd = vui->getHrdParameters();
-
-    WRITE_UVLC(sei.m_bpSeqParameterSetId, "bp_seq_parameter_set_id");
-    if (!hrd->getSubPicHrdParamsPresentFlag())
-    {
-        WRITE_FLAG(sei.m_rapCpbParamsPresentFlag, "rap_cpb_params_present_flag");
-    }
-    WRITE_FLAG(sei.m_concatenationFlag, "concatenation_flag");
-    WRITE_CODE(sei.m_auCpbRemovalDelayDelta - 1, (hrd->getCpbRemovalDelayLengthMinus1() + 1), "au_cpb_removal_delay_delta_minus1");
-    if (sei.m_rapCpbParamsPresentFlag)
-    {
-        WRITE_CODE(sei.m_cpbDelayOffset, hrd->getCpbRemovalDelayLengthMinus1() + 1, "cpb_delay_offset");
-        WRITE_CODE(sei.m_dpbDelayOffset, hrd->getDpbOutputDelayLengthMinus1()  + 1, "dpb_delay_offset");
-    }
-    for (nalOrVcl = 0; nalOrVcl < 2; nalOrVcl++)
-    {
-        if (((nalOrVcl == 0) && (hrd->getNalHrdParametersPresentFlag())) ||
-            ((nalOrVcl == 1) && (hrd->getVclHrdParametersPresentFlag())))
-        {
-            for (i = 0; i < (hrd->getCpbCntMinus1(0) + 1); i++)
-            {
-                WRITE_CODE(sei.m_initialCpbRemovalDelay[i][nalOrVcl], (hrd->getInitialCpbRemovalDelayLengthMinus1() + 1),           "initial_cpb_removal_delay");
-                WRITE_CODE(sei.m_initialCpbRemovalDelayOffset[i][nalOrVcl], (hrd->getInitialCpbRemovalDelayLengthMinus1() + 1),      "initial_cpb_removal_delay_offset");
-                if (hrd->getSubPicHrdParamsPresentFlag() || sei.m_rapCpbParamsPresentFlag)
-                {
-                    WRITE_CODE(sei.m_initialAltCpbRemovalDelay[i][nalOrVcl], (hrd->getInitialCpbRemovalDelayLengthMinus1() + 1),     "initial_alt_cpb_removal_delay");
-                    WRITE_CODE(sei.m_initialAltCpbRemovalDelayOffset[i][nalOrVcl], (hrd->getInitialCpbRemovalDelayLengthMinus1() + 1), "initial_alt_cpb_removal_delay_offset");
-                }
-            }
-        }
-    }
-
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteSEIPictureTiming(const SEIPictureTiming& sei, TComSPS *sps)
-{
-    int i;
-    TComVUI *vui = sps->getVuiParameters();
-    TComHRD *hrd = vui->getHrdParameters();
-
-    if (vui->getFrameFieldInfoPresentFlag())
-    {
-        WRITE_CODE(sei.m_picStruct, 4,              "pic_struct");
-        WRITE_CODE(sei.m_sourceScanType, 2,         "source_scan_type");
-        WRITE_FLAG(sei.m_duplicateFlag ? 1 : 0,     "duplicate_flag");
-    }
-
-    if (hrd->getCpbDpbDelaysPresentFlag())
-    {
-        WRITE_CODE(sei.m_auCpbRemovalDelay - 1, (hrd->getCpbRemovalDelayLengthMinus1() + 1), "au_cpb_removal_delay_minus1");
-        WRITE_CODE(sei.m_picDpbOutputDelay, (hrd->getDpbOutputDelayLengthMinus1() + 1), "pic_dpb_output_delay");
-        if (hrd->getSubPicHrdParamsPresentFlag())
-        {
-            WRITE_CODE(sei.m_picDpbOutputDuDelay, hrd->getDpbOutputDelayDuLengthMinus1() + 1, "pic_dpb_output_du_delay");
-        }
-        if (hrd->getSubPicHrdParamsPresentFlag() && hrd->getSubPicCpbParamsInPicTimingSEIFlag())
-        {
-            WRITE_UVLC(sei.m_numDecodingUnitsMinus1,     "num_decoding_units_minus1");
-            WRITE_FLAG(sei.m_duCommonCpbRemovalDelayFlag, "du_common_cpb_removal_delay_flag");
-            if (sei.m_duCommonCpbRemovalDelayFlag)
-            {
-                WRITE_CODE(sei.m_duCommonCpbRemovalDelayMinus1, (hrd->getDuCpbRemovalDelayLengthMinus1() + 1), "du_common_cpb_removal_delay_minus1");
-            }
-            for (i = 0; i <= sei.m_numDecodingUnitsMinus1; i++)
-            {
-                WRITE_UVLC(sei.m_numNalusInDuMinus1[i],  "num_nalus_in_du_minus1");
-                if ((!sei.m_duCommonCpbRemovalDelayFlag) && (i < sei.m_numDecodingUnitsMinus1))
-                {
-                    WRITE_CODE(sei.m_duCpbRemovalDelayMinus1[i], (hrd->getDuCpbRemovalDelayLengthMinus1() + 1), "du_cpb_removal_delay_minus1");
-                }
-            }
-        }
-    }
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteSEIRecoveryPoint(const SEIRecoveryPoint& sei)
-{
-    WRITE_SVLC(sei.m_recoveryPocCnt,    "recovery_poc_cnt");
-    WRITE_FLAG(sei.m_exactMatchingFlag, "exact_matching_flag");
-    WRITE_FLAG(sei.m_brokenLinkFlag,    "broken_link_flag");
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteSEIDisplayOrientation(const SEIDisplayOrientation &sei)
-{
-    WRITE_FLAG(sei.cancelFlag,           "display_orientation_cancel_flag");
-    if (!sei.cancelFlag)
-    {
-        WRITE_FLAG(sei.horFlip,                   "hor_flip");
-        WRITE_FLAG(sei.verFlip,                   "ver_flip");
-        WRITE_CODE(sei.anticlockwiseRotation, 16, "anticlockwise_rotation");
-        WRITE_FLAG(sei.persistenceFlag,           "display_orientation_persistence_flag");
-    }
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteSEIGradualDecodingRefreshInfo(const SEIGradualDecodingRefreshInfo &sei)
-{
-    WRITE_FLAG(sei.m_gdrForegroundFlag, "gdr_foreground_flag");
-    xWriteByteAlign();
-}
-
-void SEIWriter::xWriteByteAlign()
-{
-    if (m_bitIf->getNumberOfWrittenBits() % 8 != 0)
-    {
-        WRITE_FLAG(1, "bit_equal_to_one");
-        while (m_bitIf->getNumberOfWrittenBits() % 8 != 0)
-        {
-            WRITE_FLAG(0, "bit_equal_to_zero");
-        }
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/SEIwrite.h Deleted

@@ -1,74 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#ifndef X265_SEIWRITE_H
-#define X265_SEIWRITE_H
-
-#include "SyntaxElementWriter.h"
-#include "TLibCommon/SEI.h"
-
-namespace x265 {
-// private namespace
-
-class TComBitIf;
-
-//! \ingroup TLibEncoder
-//! \{
-class SEIWriter : public SyntaxElementWriter
-{
-public:
-
-    SEIWriter() {}
-
-    virtual ~SEIWriter() {}
-
-    void writeSEImessage(TComBitIf& bs, const SEI& sei, TComSPS *sps);
-
-protected:
-
-    void xWriteSEIpayloadData(const SEI& sei, TComSPS *sps);
-    void xWriteSEIuserDataUnregistered(const SEIuserDataUnregistered &sei);
-    void xWriteSEIActiveParameterSets(const SEIActiveParameterSets& sei);
-    void xWriteSEIDecodingUnitInfo(const SEIDecodingUnitInfo& sei, TComSPS *sps);
-    void xWriteSEIDecodedPictureHash(const SEIDecodedPictureHash& sei);
-    void xWriteSEIBufferingPeriod(const SEIBufferingPeriod& sei, TComSPS *sps);
-    void xWriteSEIPictureTiming(const SEIPictureTiming& sei, TComSPS *sps);
-    void xWriteSEIRecoveryPoint(const SEIRecoveryPoint& sei);
-    void xWriteSEIDisplayOrientation(const SEIDisplayOrientation &sei);
-    void xWriteSEIGradualDecodingRefreshInfo(const SEIGradualDecodingRefreshInfo &sei);
-    void xWriteByteAlign();
-};
-}
-//! \}
-
-#endif // ifndef X265_SEIWRITE_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/SyntaxElementWriter.cpp Deleted

@@ -1,133 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     SyntaxElementWriter.cpp
-    \brief    CAVLC encoder class
-*/
-
-#include "TLibCommon/CommonDef.h"
-#include "SyntaxElementWriter.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-#if ENC_DEC_TRACE
-
-void  SyntaxElementWriter::xWriteCodeTr(uint32_t value, uint32_t  length, const char *symbolName)
-{
-    xWriteCode(value, length);
-    if (g_HLSTraceEnable)
-    {
-        fprintf(g_hTrace, "%8ld  ", g_nSymbolCounter++);
-        if (length < 10)
-        {
-            fprintf(g_hTrace, "%-50s u(%d)  : %d\n", symbolName, length, value);
-        }
-        else
-        {
-            fprintf(g_hTrace, "%-50s u(%d) : %d\n", symbolName, length, value);
-        }
-    }
-}
-
-void  SyntaxElementWriter::xWriteUvlcTr(uint32_t value, const char *symbolName)
-{
-    xWriteUvlc(value);
-    if (g_HLSTraceEnable)
-    {
-        fprintf(g_hTrace, "%8ld  ", g_nSymbolCounter++);
-        fprintf(g_hTrace, "%-50s ue(v) : %d\n", symbolName, value);
-    }
-}
-
-void  SyntaxElementWriter::xWriteSvlcTr(int value, const char *symbolName)
-{
-    xWriteSvlc(value);
-    if (g_HLSTraceEnable)
-    {
-        fprintf(g_hTrace, "%8ld  ", g_nSymbolCounter++);
-        fprintf(g_hTrace, "%-50s se(v) : %d\n", symbolName, value);
-    }
-}
-
-void  SyntaxElementWriter::xWriteFlagTr(uint32_t value, const char *symbolName)
-{
-    xWriteFlag(value);
-    if (g_HLSTraceEnable)
-    {
-        fprintf(g_hTrace, "%8ld  ", g_nSymbolCounter++);
-        fprintf(g_hTrace, "%-50s u(1)  : %d\n", symbolName, value);
-    }
-}
-
-#endif // if ENC_DEC_TRACE
-
-void SyntaxElementWriter::xWriteCode(uint32_t code, uint32_t len)
-{
-    assert(len > 0);
-    m_bitIf->write(code, len);
-}
-
-void SyntaxElementWriter::xWriteUvlc(uint32_t code)
-{
-    uint32_t len = 1;
-    uint32_t temp = ++code;
-
-    assert(temp);
-
-    while (1 != temp)
-    {
-        temp >>= 1;
-        len += 2;
-    }
-
-    // Take care of cases where len > 32
-    m_bitIf->write(0, len >> 1);
-    m_bitIf->write(code, (len + 1) >> 1);
-}
-
-void SyntaxElementWriter::xWriteSvlc(int code)
-{
-    uint32_t ucode = xConvertToUInt(code);
-
-    xWriteUvlc(ucode);
-}
-
-void SyntaxElementWriter::xWriteFlag(uint32_t code)
-{
-    m_bitIf->write(code, 1);
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/SyntaxElementWriter.h Deleted

@@ -1,97 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     SyntaxElementWriter.h
-    \brief    CAVLC encoder class (header)
-*/
-
-#ifndef X265_SYNTAXELEMENTWRITER_H
-#define X265_SYNTAXELEMENTWRITER_H
-
-#include "TLibCommon/CommonDef.h"
-#include "TLibCommon/TComBitStream.h"
-#include "TLibCommon/TComRom.h"
-
-//! \ingroup TLibEncoder
-//! \{
-
-#if ENC_DEC_TRACE
-
-#define WRITE_CODE(value, length, name)    xWriteCodeTr(value, length, name)
-#define WRITE_UVLC(value,         name)    xWriteUvlcTr(value,         name)
-#define WRITE_SVLC(value,         name)    xWriteSvlcTr(value,         name)
-#define WRITE_FLAG(value,         name)    xWriteFlagTr(value,         name)
-
-#else
-
-#define WRITE_CODE(value, length, name)     xWriteCode(value, length)
-#define WRITE_UVLC(value,         name)     xWriteUvlc(value)
-#define WRITE_SVLC(value,         name)     xWriteSvlc(value)
-#define WRITE_FLAG(value,         name)     xWriteFlag(value)
-
-#endif // if ENC_DEC_TRACE
-
-namespace x265 {
-// private namespace
-
-class SyntaxElementWriter
-{
-protected:
-
-    TComBitIf* m_bitIf;
-
-    SyntaxElementWriter()
-        : m_bitIf(NULL)
-    {}
-
-    virtual ~SyntaxElementWriter() {}
-
-    void  setBitstream(TComBitIf* p)  { m_bitIf = p;  }
-
-    void  xWriteCode(uint32_t code, uint32_t len);
-    void  xWriteUvlc(uint32_t code);
-    void  xWriteSvlc(int code);
-    void  xWriteFlag(uint32_t code);
-#if ENC_DEC_TRACE
-    void  xWriteCodeTr(uint32_t value, uint32_t  length, const char *symbolName);
-    void  xWriteUvlcTr(uint32_t value,               const char *symbolName);
-    void  xWriteSvlcTr(int value,                const char *symbolName);
-    void  xWriteFlagTr(uint32_t value,               const char *symbolName);
-#endif
-
-    uint32_t  xConvertToUInt(int val) { return (val <= 0) ? -val << 1 : (val << 1) - 1; }
-};
-}
-//! \}
-
-#endif // ifndef X265_SYNTAXELEMENTWRITER_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncBinCoderCABAC.cpp Deleted

@@ -1,380 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncBinCoderCABAC.cpp
-    \brief    binary entropy encoder of CABAC
-*/
-
-#include "TEncBinCoderCABAC.h"
-#include "TLibCommon/TComRom.h"
-#include "threading.h"  // CLZ32
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-TEncBinCABAC::TEncBinCABAC(bool isCounter)
-    : m_bitIf(0)
-    , m_fracBits(0)
-    , m_bIsCounter(isCounter)
-{}
-
-TEncBinCABAC::~TEncBinCABAC()
-{}
-
-void TEncBinCABAC::init(TComBitIf* bitIf)
-{
-    m_bitIf = bitIf;
-}
-
-void TEncBinCABAC::start()
-{
-    m_low              = 0;
-    m_range            = 510;
-    m_bitsLeft         = -12;
-    m_numBufferedBytes = 0;
-    m_bufferedByte     = 0xff;
-}
-
-void TEncBinCABAC::finish()
-{
-    if (m_bIsCounter)
-    {
-        // TODO: why write 0 bits?
-        m_bitIf->write(0, uint32_t(m_fracBits >> 15));
-        m_fracBits &= 32767;
-        assert(0);
-    }
-
-    if (m_low >> (21 + m_bitsLeft))
-    {
-        //assert( m_numBufferedBytes > 0 );
-        //assert( m_bufferedByte != 0xff );
-        m_bitIf->writeByte(m_bufferedByte + 1);
-        while (m_numBufferedBytes > 1)
-        {
-            m_bitIf->writeByte(0x00);
-            m_numBufferedBytes--;
-        }
-
-        m_low -= 1 << (21 + m_bitsLeft);
-    }
-    else
-    {
-        if (m_numBufferedBytes > 0)
-        {
-            m_bitIf->writeByte(m_bufferedByte);
-        }
-        while (m_numBufferedBytes > 1)
-        {
-            m_bitIf->writeByte(0xff);
-            m_numBufferedBytes--;
-        }
-    }
-    m_bitIf->write(m_low >> 8, 13 + m_bitsLeft);
-}
-
-void TEncBinCABAC::flush()
-{
-    encodeBinTrm(1);
-    finish();
-    m_bitIf->write(1, 1);
-    m_bitIf->writeAlignZero();
-
-    start();
-}
-
-/** Reset BAC register and counter values.
- * \returns void
- */
-void TEncBinCABAC::resetBac()
-{
-    start();
-}
-
-/** Encode PCM alignment zero bits.
- * \returns void
- */
-void TEncBinCABAC::encodePCMAlignBits()
-{
-    finish();
-    m_bitIf->write(1, 1);
-    m_bitIf->writeAlignZero(); // pcm align zero
-}
-
-/** Write a PCM code.
- * \param code code value
- * \param length code bit-depth
- * \returns void
- */
-void TEncBinCABAC::xWritePCMCode(uint32_t code, uint32_t length)
-{
-    m_bitIf->write(code, length);
-}
-
-void TEncBinCABAC::copyState(TEncBinCABAC* binCABAC)
-{
-    m_low              = binCABAC->m_low;
-    m_range            = binCABAC->m_range;
-    m_bitsLeft         = binCABAC->m_bitsLeft;
-    m_bufferedByte     = binCABAC->m_bufferedByte;
-    m_numBufferedBytes = binCABAC->m_numBufferedBytes;
-    m_fracBits         = binCABAC->m_fracBits;
-}
-
-void TEncBinCABAC::resetBits()
-{
-    m_low              = 0;
-    m_bitsLeft         = -12;
-    m_numBufferedBytes = 0;
-    m_bufferedByte     = 0xff;
-    m_fracBits        &= 32767;
-}
-
-/**
- * \brief Encode bin
- *
- * \param binValue   bin value
- * \param rcCtxModel context model
- */
-void TEncBinCABAC::encodeBin(uint32_t binValue, ContextModel &ctxModel)
-{
-    {
-        DTRACE_CABAC_VL(g_nSymbolCounter++)
-        DTRACE_CABAC_T("\tstate=")
-        DTRACE_CABAC_V((ctxModel.m_state << 1) + sbacGetMps(ctxModel.m_state))
-        DTRACE_CABAC_T("\tsymbol=")
-        DTRACE_CABAC_V(binValue)
-        DTRACE_CABAC_T("\n")
-    }
-
-    uint32_t mstate = ctxModel.m_state;
-
-    ctxModel.m_state = sbacNext(mstate, binValue);
-
-    if (m_bIsCounter)
-    {
-        m_fracBits += sbacGetEntropyBits(mstate, binValue);
-        return;
-    }
-    ctxModel.bBinsCoded = 1;
-
-    uint32_t mps = sbacGetMps(mstate);
-    uint32_t state = sbacGetState(mstate);
-    uint32_t lps = g_lpsTable[state][(m_range >> 6) & 3];
-    m_range -= lps;
-
-    assert(lps != 0);
-
-    int numBits = (uint32_t)(m_range - 256) >> 31;
-    uint32_t low = m_low;
-    uint32_t range = m_range;
-    if (binValue != mps)
-    {
-        // NOTE: lps is non-zero and the maximum of idx is 8 because lps less than 256
-        //numBits   = g_renormTable[lps >> 3];
-        unsigned long idx;
-        CLZ32(idx, lps);
-        numBits = 8 - idx;
-        if (numBits >= 6)
-            numBits = 6;
-
-        low    += range;
-        range   = lps;
-    }
-    m_low = (low << numBits);
-    m_range = (range << numBits);
-    m_bitsLeft += numBits;
-
-    if (m_bitsLeft >= 0)
-    {
-        writeOut();
-    }
-}
-
-/**
- * \brief Encode equiprobable bin
- *
- * \param binValue bin value
- */
-void TEncBinCABAC::encodeBinEP(uint32_t binValue)
-{
-    {
-        DTRACE_CABAC_VL(g_nSymbolCounter++)
-        DTRACE_CABAC_T("\tEPsymbol=")
-        DTRACE_CABAC_V(binValue)
-        DTRACE_CABAC_T("\n")
-    }
-    if (m_bIsCounter)
-    {
-        m_fracBits += 32768;
-        return;
-    }
-    m_low <<= 1;
-    if (binValue)
-    {
-        m_low += m_range;
-    }
-    m_bitsLeft++;
-
-    if (m_bitsLeft >= 0)
-    {
-        writeOut();
-    }
-}
-
-/**
- * \brief Encode equiprobable bins
- *
- * \param binValues bin values
- * \param numBins number of bins
- */
-void TEncBinCABAC::encodeBinsEP(uint32_t binValues, int numBins)
-{
-    if (m_bIsCounter)
-    {
-        m_fracBits += 32768 * numBins;
-        return;
-    }
-
-    for (int i = 0; i < numBins; i++)
-    {
-        DTRACE_CABAC_VL(g_nSymbolCounter++)
-        DTRACE_CABAC_T("\tEPsymbol=")
-        DTRACE_CABAC_V((binValues >> (numBins - 1 - i)) & 1)
-        DTRACE_CABAC_T("\n")
-    }
-
-    while (numBins > 8)
-    {
-        numBins -= 8;
-        uint32_t pattern = binValues >> numBins;
-        m_low <<= 8;
-        m_low += m_range * pattern;
-        binValues -= pattern << numBins;
-        m_bitsLeft += 8;
-
-        if (m_bitsLeft >= 0)
-        {
-            writeOut();
-        }
-    }
-
-    m_low <<= numBins;
-    m_low += m_range * binValues;
-    m_bitsLeft += numBins;
-
-    if (m_bitsLeft >= 0)
-    {
-        writeOut();
-    }
-}
-
-/**
- * \brief Encode terminating bin
- *
- * \param binValue bin value
- */
-void TEncBinCABAC::encodeBinTrm(uint32_t binValue)
-{
-    if (m_bIsCounter)
-    {
-        m_fracBits += sbacGetEntropyBitsTrm(binValue);
-        return;
-    }
-
-    m_range -= 2;
-    if (binValue)
-    {
-        m_low  += m_range;
-        m_low <<= 7;
-        m_range = 2 << 7;
-        m_bitsLeft += 7;
-    }
-    else if (m_range >= 256)
-    {
-        return;
-    }
-    else
-    {
-        m_low   <<= 1;
-        m_range <<= 1;
-        m_bitsLeft++;
-    }
-
-    if (m_bitsLeft >= 0)
-    {
-        writeOut();
-    }
-}
-
-/**
- * \brief Move bits from register into bitstream
- */
-void TEncBinCABAC::writeOut()
-{
-    uint32_t leadByte = m_low >> (13 + m_bitsLeft);
-
-    m_bitsLeft -= 8;
-    m_low &= 0xffffffffu >> (11 - m_bitsLeft);
-
-    if (leadByte == 0xff)
-    {
-        m_numBufferedBytes++;
-    }
-    else
-    {
-        if (m_numBufferedBytes > 0)
-        {
-            uint32_t carry = leadByte >> 8;
-            uint32_t byteTowrite = m_bufferedByte + carry;
-            m_bufferedByte = leadByte & 0xff;
-            m_bitIf->writeByte(byteTowrite);
-
-            byteTowrite = (0xff + carry) & 0xff;
-            while (m_numBufferedBytes > 1)
-            {
-                m_bitIf->writeByte(byteTowrite);
-                m_numBufferedBytes--;
-            }
-        }
-        else
-        {
-            m_numBufferedBytes = 1;
-            m_bufferedByte = leadByte;
-        }
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncBinCoderCABAC.h Deleted

@@ -1,104 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncBinCoderCABAC.h
-    \brief    binary entropy encoder of CABAC
-*/
-
-#ifndef X265_TENCBINCODERCABAC_H
-#define X265_TENCBINCODERCABAC_H
-
-#include "TLibCommon/ContextTables.h"
-#include "TLibCommon/TComBitStream.h"
-
-//! \ingroup TLibEncoder
-//! \{
-
-namespace x265 {
-// private namespace
-
-class TEncBinCABAC
-{
-public:
-
-    TEncBinCABAC(bool isCounter = false);
-    virtual ~TEncBinCABAC();
-
-    void  init(TComBitIf* bitIf);
-
-    void  start();
-    void  finish();
-    void  copyState(TEncBinCABAC* binIf);
-    void  flush();
-
-    void  resetBac();
-    void  encodePCMAlignBits();
-    void  xWritePCMCode(uint32_t code, uint32_t length);
-
-    void  resetBits();
-
-    uint32_t getNumWrittenBits()
-    {
-        // NOTE: the HM go here only in Counter mode
-        assert(!m_bIsCounter || (m_bitIf->getNumberOfWrittenBits() == 0));
-        assert(m_bIsCounter);
-        return uint32_t(m_fracBits >> 15);
-
-        // NOTE: I keep the old code, so someone may active if they want
-        //return m_bitIf->getNumberOfWrittenBits() + 8 * m_numBufferedBytes + 23 - m_bitsLeft;
-    }
-
-    void  encodeBin(uint32_t binValue, ContextModel& ctxModel);
-    void  encodeBinEP(uint32_t binValue);
-    void  encodeBinsEP(uint32_t binValues, int numBins);
-    void  encodeBinTrm(uint32_t binValue);
-
-protected:
-
-    void writeOut();
-
-public:
-
-    TComBitIf* m_bitIf;
-    uint32_t   m_low;
-    uint32_t   m_range;
-    uint32_t   m_bufferedByte;
-    int        m_numBufferedBytes;
-    int        m_bitsLeft;
-    uint64_t   m_fracBits;
-    bool       m_bIsCounter;
-};
-}
-//! \}
-
-#endif // ifndef X265_TENCBINCODERCABAC_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncCu.cpp Deleted

@@ -1,1637 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncCu.cpp
-    \brief    Coding Unit (CU) encoder class
-*/
-
-#include "TEncCu.h"
-
-#include "primitives.h"
-#include "encoder.h"
-#include "common.h"
-
-#include "PPA/ppa.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
-TEncCu::TEncCu()
-{
-    m_interCU_2Nx2N   = NULL;
-    m_interCU_2NxN    = NULL;
-    m_interCU_Nx2N    = NULL;
-    m_intraInInterCU  = NULL;
-    m_mergeCU         = NULL;
-    m_bestMergeCU     = NULL;
-    m_bestCU          = NULL;
-    m_tempCU          = NULL;
-
-    m_bestPredYuv     = NULL;
-    m_bestResiYuv     = NULL;
-    m_bestRecoYuv     = NULL;
-
-    m_tmpPredYuv      = NULL;
-    m_tmpResiYuv      = NULL;
-    m_tmpRecoYuv      = NULL;
-    m_bestMergeRecoYuv = NULL;
-    m_origYuv         = NULL;
-    for (int i = 0; i < MAX_PRED_TYPES; i++)
-    {
-        m_modePredYuv[i] = NULL;
-    }
-
-    m_search          = NULL;
-    m_trQuant         = NULL;
-    m_rdCost          = NULL;
-    m_bitCounter      = NULL;
-    m_entropyCoder    = NULL;
-    m_rdSbacCoders    = NULL;
-    m_rdGoOnSbacCoder = NULL;
-}
-
-/**
- \param    totalDepth  total number of allowable depth
- \param    maxWidth    largest CU width
- \param    maxHeight   largest CU height
- */
-bool TEncCu::create(UChar totalDepth, uint32_t maxWidth)
-{
-    m_totalDepth   = totalDepth + 1;
-    m_interCU_2Nx2N  = new TComDataCU*[m_totalDepth - 1];
-    m_interCU_2NxN   = new TComDataCU*[m_totalDepth - 1];
-    m_interCU_Nx2N   = new TComDataCU*[m_totalDepth - 1];
-    m_intraInInterCU = new TComDataCU*[m_totalDepth - 1];
-    m_mergeCU        = new TComDataCU*[m_totalDepth - 1];
-    m_bestMergeCU    = new TComDataCU*[m_totalDepth - 1];
-    m_bestCU      = new TComDataCU*[m_totalDepth - 1];
-    m_tempCU      = new TComDataCU*[m_totalDepth - 1];
-
-    m_bestPredYuv = new TComYuv*[m_totalDepth - 1];
-    m_bestResiYuv = new ShortYuv*[m_totalDepth - 1];
-    m_bestRecoYuv = new TComYuv*[m_totalDepth - 1];
-
-    m_tmpPredYuv = new TComYuv*[m_totalDepth - 1];
-
-    m_modePredYuv[0] = new TComYuv*[m_totalDepth - 1];
-    m_modePredYuv[1] = new TComYuv*[m_totalDepth - 1];
-    m_modePredYuv[2] = new TComYuv*[m_totalDepth - 1];
-    m_modePredYuv[3] = new TComYuv*[m_totalDepth - 1];
-    m_modePredYuv[4] = new TComYuv*[m_totalDepth - 1];
-    m_modePredYuv[5] = new TComYuv*[m_totalDepth - 1];
-
-    m_tmpResiYuv = new ShortYuv*[m_totalDepth - 1];
-    m_tmpRecoYuv = new TComYuv*[m_totalDepth - 1];
-
-    m_bestMergeRecoYuv = new TComYuv*[m_totalDepth - 1];
-
-    m_origYuv = new TComYuv*[m_totalDepth - 1];
-
-    int csp = m_param->internalCsp;
-
-    bool ok = true;
-    for (int i = 0; i < m_totalDepth - 1; i++)
-    {
-        uint32_t numPartitions = 1 << ((m_totalDepth - i - 1) << 1);
-        uint32_t width  = maxWidth  >> i;
-        uint32_t height = maxWidth >> i;
-
-        m_bestCU[i] = new TComDataCU;
-        ok &= m_bestCU[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_tempCU[i] = new TComDataCU;
-        ok &= m_tempCU[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_interCU_2Nx2N[i] = new TComDataCU;
-        ok &= m_interCU_2Nx2N[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_interCU_2NxN[i] = new TComDataCU;
-        ok &= m_interCU_2NxN[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_interCU_Nx2N[i] = new TComDataCU;
-        ok &= m_interCU_Nx2N[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_intraInInterCU[i] = new TComDataCU;
-        ok &= m_intraInInterCU[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_mergeCU[i] = new TComDataCU;
-        ok &= m_mergeCU[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_bestMergeCU[i] = new TComDataCU;
-        ok &= m_bestMergeCU[i]->create(numPartitions, width, height, maxWidth >> (m_totalDepth - 1), csp);
-
-        m_bestPredYuv[i] = new TComYuv;
-        ok &= m_bestPredYuv[i]->create(width, height, csp);
-
-        m_bestResiYuv[i] = new ShortYuv;
-        ok &= m_bestResiYuv[i]->create(width, height, csp);
-
-        m_bestRecoYuv[i] = new TComYuv;
-        ok &= m_bestRecoYuv[i]->create(width, height, csp);
-
-        m_tmpPredYuv[i] = new TComYuv;
-        ok &= m_tmpPredYuv[i]->create(width, height, csp);
-
-        for (int j = 0; j < MAX_PRED_TYPES; j++)
-        {
-            m_modePredYuv[j][i] = new TComYuv;
-            ok &= m_modePredYuv[j][i]->create(width, height, csp);
-        }
-
-        m_tmpResiYuv[i] = new ShortYuv;
-        ok &= m_tmpResiYuv[i]->create(width, height, csp);
-
-        m_tmpRecoYuv[i] = new TComYuv;
-        ok &= m_tmpRecoYuv[i]->create(width, height, csp);
-
-        m_bestMergeRecoYuv[i] = new TComYuv;
-        ok &= m_bestMergeRecoYuv[i]->create(width, height, csp);
-
-        m_origYuv[i] = new TComYuv;
-        ok &= m_origYuv[i]->create(width, height, csp);
-    }
-
-    m_bEncodeDQP = false;
-    return ok;
-}
-
-void TEncCu::destroy()
-{
-    for (int i = 0; i < m_totalDepth - 1; i++)
-    {
-        if (m_interCU_2Nx2N && m_interCU_2Nx2N[i])
-        {
-            m_interCU_2Nx2N[i]->destroy();
-            delete m_interCU_2Nx2N[i];
-            m_interCU_2Nx2N[i] = NULL;
-        }
-        if (m_interCU_2NxN && m_interCU_2NxN[i])
-        {
-            m_interCU_2NxN[i]->destroy();
-            delete m_interCU_2NxN[i];
-            m_interCU_2NxN[i] = NULL;
-        }
-        if (m_interCU_Nx2N && m_interCU_Nx2N[i])
-        {
-            m_interCU_Nx2N[i]->destroy();
-            delete m_interCU_Nx2N[i];
-            m_interCU_Nx2N[i] = NULL;
-        }
-        if (m_intraInInterCU && m_intraInInterCU[i])
-        {
-            m_intraInInterCU[i]->destroy();
-            delete m_intraInInterCU[i];
-            m_intraInInterCU[i] = NULL;
-        }
-        if (m_mergeCU && m_mergeCU[i])
-        {
-            m_mergeCU[i]->destroy();
-            delete m_mergeCU[i];
-            m_mergeCU[i] = NULL;
-        }
-        if (m_bestMergeCU && m_bestMergeCU[i])
-        {
-            m_bestMergeCU[i]->destroy();
-            delete m_bestMergeCU[i];
-            m_bestMergeCU[i] = NULL;
-        }
-        if (m_bestCU && m_bestCU[i])
-        {
-            m_bestCU[i]->destroy();
-            delete m_bestCU[i];
-            m_bestCU[i] = NULL;
-        }
-        if (m_tempCU && m_tempCU[i])
-        {
-            m_tempCU[i]->destroy();
-            delete m_tempCU[i];
-            m_tempCU[i] = NULL;
-        }
-
-        if (m_bestPredYuv && m_bestPredYuv[i])
-        {
-            m_bestPredYuv[i]->destroy();
-            delete m_bestPredYuv[i];
-            m_bestPredYuv[i] = NULL;
-        }
-        if (m_bestResiYuv && m_bestResiYuv[i])
-        {
-            m_bestResiYuv[i]->destroy();
-            delete m_bestResiYuv[i];
-            m_bestResiYuv[i] = NULL;
-        }
-        if (m_bestRecoYuv && m_bestRecoYuv[i])
-        {
-            m_bestRecoYuv[i]->destroy();
-            delete m_bestRecoYuv[i];
-            m_bestRecoYuv[i] = NULL;
-        }
-
-        if (m_tmpPredYuv && m_tmpPredYuv[i])
-        {
-            m_tmpPredYuv[i]->destroy();
-            delete m_tmpPredYuv[i];
-            m_tmpPredYuv[i] = NULL;
-        }
-        for (int j = 0; j < MAX_PRED_TYPES; j++)
-        {
-            if (m_modePredYuv[j] && m_modePredYuv[j][i])
-            {
-                m_modePredYuv[j][i]->destroy();
-                delete m_modePredYuv[j][i];
-                m_modePredYuv[j][i] = NULL;
-            }
-        }
-
-        if (m_tmpResiYuv && m_tmpResiYuv[i])
-        {
-            m_tmpResiYuv[i]->destroy();
-            delete m_tmpResiYuv[i];
-            m_tmpResiYuv[i] = NULL;
-        }
-        if (m_tmpRecoYuv && m_tmpRecoYuv[i])
-        {
-            m_tmpRecoYuv[i]->destroy();
-            delete m_tmpRecoYuv[i];
-            m_tmpRecoYuv[i] = NULL;
-        }
-        if (m_bestMergeRecoYuv && m_bestMergeRecoYuv[i])
-        {
-            m_bestMergeRecoYuv[i]->destroy();
-            delete m_bestMergeRecoYuv[i];
-            m_bestMergeRecoYuv[i] = NULL;
-        }
-
-        if (m_origYuv && m_origYuv[i])
-        {
-            m_origYuv[i]->destroy();
-            delete m_origYuv[i];
-            m_origYuv[i] = NULL;
-        }
-    }
-
-    delete [] m_interCU_2Nx2N;
-    m_interCU_2Nx2N = NULL;
-    delete [] m_interCU_2NxN;
-    m_interCU_2NxN = NULL;
-    delete [] m_interCU_Nx2N;
-    m_interCU_Nx2N = NULL;
-    delete [] m_intraInInterCU;
-    m_intraInInterCU = NULL;
-    delete [] m_mergeCU;
-    m_mergeCU = NULL;
-    delete [] m_bestMergeCU;
-    m_bestMergeCU = NULL;
-    delete [] m_bestCU;
-    m_bestCU = NULL;
-    delete [] m_tempCU;
-    m_tempCU = NULL;
-
-    delete [] m_bestPredYuv;
-    m_bestPredYuv = NULL;
-    delete [] m_bestResiYuv;
-    m_bestResiYuv = NULL;
-    delete [] m_bestRecoYuv;
-    m_bestRecoYuv = NULL;
-
-    delete [] m_bestMergeRecoYuv;
-    m_bestMergeRecoYuv = NULL;
-    delete [] m_tmpPredYuv;
-    m_tmpPredYuv = NULL;
-
-    for (int i = 0; i < MAX_PRED_TYPES; i++)
-    {
-        delete [] m_modePredYuv[i];
-        m_modePredYuv[i] = NULL;
-    }
-
-    delete [] m_tmpResiYuv;
-    m_tmpResiYuv = NULL;
-    delete [] m_tmpRecoYuv;
-    m_tmpRecoYuv = NULL;
-    delete [] m_origYuv;
-    m_origYuv = NULL;
-}
-
-/** \param    pcEncTop      pointer of encoder class
- */
-void TEncCu::init(Encoder* top)
-{
-    m_param = top->param;
-    m_CUTransquantBypassFlagValue = top->m_CUTransquantBypassFlagValue;
-}
-
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
-
-/** \param  rpcCU pointer of CU data class
- */
-
-void TEncCu::compressCU(TComDataCU* cu)
-{
-    if (cu->getSlice()->getPPS()->getUseDQP())
-    {
-        setdQPFlag(true);
-    }
-    // initialize CU data
-    m_bestCU[0]->initCU(cu->getPic(), cu->getAddr());
-    m_tempCU[0]->initCU(cu->getPic(), cu->getAddr());
-
-    // analysis of CU
-#if LOG_CU_STATISTICS
-    int numPartition = cu->getTotalNumPart();
-#endif
-    
-    if (m_bestCU[0]->getSlice()->getSliceType() == I_SLICE)
-    {
-        xCompressIntraCU(m_bestCU[0], m_tempCU[0], 0);
-#if LOG_CU_STATISTICS
-        int i = 0, part;
-        do
-        {
-            m_log->totalCu++;
-            part = cu->getDepth(i);
-            int next = numPartition >> (part * 2);
-            if (part == g_maxCUDepth - 1 && cu->getPartitionSize(i) != SIZE_2Nx2N)
-            {
-                m_log->cntIntraNxN++;
-            }
-            else
-            {
-                m_log->cntIntra[part]++;
-                if (cu->getLumaIntraDir()[i] > 1)
-                    m_log->cuIntraDistribution[part][ANGULAR_MODE_ID]++;
-                else
-                    m_log->cuIntraDistribution[part][cu->getLumaIntraDir()[i]]++;
-            }
-            i += next;
-        }
-        while (i < numPartition);
-#endif // if LOG_CU_STATISTICS
-    }
-    else
-    {
-        if (m_param->rdLevel < 5)
-        {
-            TComDataCU* outBestCU = NULL;
-
-            /* At the start of analysis, the best CU is a null pointer
-            On return, it points to the CU encode with best chosen mode*/
-            xCompressInterCU(outBestCU, m_tempCU[0], cu, 0, 0, 4);
-        }
-        else
-            xCompressCU(m_bestCU[0], m_tempCU[0], 0);
-#if LOG_CU_STATISTICS
-        int i = 0, part;
-        do
-        {
-            part = cu->getDepth(i);
-            m_log->cntTotalCu[part]++;
-            int next = numPartition >> (part * 2);
-            if (cu->isSkipped(i))
-            {
-                m_log->cntSkipCu[part]++;
-            }
-            else
-            {
-                m_log->totalCu++;
-                if (cu->getPredictionMode(0) == MODE_INTER)
-                {
-                    m_log->cntInter[part]++;
-                    if (cu->getPartitionSize(0) < AMP_ID)
-                        m_log->cuInterDistribution[part][cu->getPartitionSize(0)]++;
-                    else
-                        m_log->cuInterDistribution[part][AMP_ID]++;
-                }
-                else if (cu->getPredictionMode(0) == MODE_INTRA)
-                {
-                    if (part == g_maxCUDepth - 1 && cu->getPartitionSize(0) == SIZE_NxN)
-                    {
-                        m_log->cntIntraNxN++;
-                    }
-                    else
-                    {
-                        m_log->cntIntra[part]++;
-                        if (cu->getLumaIntraDir()[0] > 1)
-                            m_log->cuIntraDistribution[part][ANGULAR_MODE_ID]++;
-                        else
-                            m_log->cuIntraDistribution[part][cu->getLumaIntraDir()[0]]++;
-                    }
-                }
-            }
-            i = i + next;
-        }
-        while (i < numPartition);
-#endif // if LOG_CU_STATISTICS
-    }
-}
-
-/** \param  cu  pointer of CU data class
- */
-void TEncCu::encodeCU(TComDataCU* cu)
-{
-    if (cu->getSlice()->getPPS()->getUseDQP())
-    {
-        setdQPFlag(true);
-    }
-
-    // Encode CU data
-    xEncodeCU(cu, 0, 0);
-}
-
-// ====================================================================================================================
-// Protected member functions
-// ====================================================================================================================
-
-/** Derive small set of test modes for AMP encoder speed-up
- *\param   outBestCU
- *\param   parentSize
- *\param   bTestAMP_Hor
- *\param   bTestAMP_Ver
- *\param   bTestMergeAMP_Hor
- *\param   bTestMergeAMP_Ver
- *\returns void
-*/
-void TEncCu::deriveTestModeAMP(TComDataCU* outBestCU, PartSize parentSize, bool &bTestAMP_Hor, bool &bTestAMP_Ver,
-                               bool &bTestMergeAMP_Hor, bool &bTestMergeAMP_Ver)
-{
-    if (outBestCU->getPartitionSize(0) == SIZE_2NxN)
-    {
-        bTestAMP_Hor = true;
-    }
-    else if (outBestCU->getPartitionSize(0) == SIZE_Nx2N)
-    {
-        bTestAMP_Ver = true;
-    }
-    else if (outBestCU->getPartitionSize(0) == SIZE_2Nx2N && outBestCU->getMergeFlag(0) == false &&
-             outBestCU->isSkipped(0) == false)
-    {
-        bTestAMP_Hor = true;
-        bTestAMP_Ver = true;
-    }
-
-    //! Utilizing the partition size of parent PU
-    if (parentSize >= SIZE_2NxnU && parentSize <= SIZE_nRx2N)
-    {
-        bTestMergeAMP_Hor = true;
-        bTestMergeAMP_Ver = true;
-    }
-
-    if (parentSize == SIZE_NONE) //! if parent is intra
-    {
-        if (outBestCU->getPartitionSize(0) == SIZE_2NxN)
-        {
-            bTestMergeAMP_Hor = true;
-        }
-        else if (outBestCU->getPartitionSize(0) == SIZE_Nx2N)
-        {
-            bTestMergeAMP_Ver = true;
-        }
-    }
-
-    if (outBestCU->getPartitionSize(0) == SIZE_2Nx2N && outBestCU->isSkipped(0) == false)
-    {
-        bTestMergeAMP_Hor = true;
-        bTestMergeAMP_Ver = true;
-    }
-
-    if (outBestCU->getCUSize(0) == 64)
-    {
-        bTestAMP_Hor = false;
-        bTestAMP_Ver = false;
-    }
-}
-
-// ====================================================================================================================
-// Protected member functions
-// ====================================================================================================================
-
-/** Compress a CU block recursively with enabling sub-LCU-level delta QP
- *\param   outBestCU
- *\param   outTempCU
- *\param   depth
- *\returns void
- *
- *- for loop of QP value to compress the current CU with all possible QP
-*/
-
-void TEncCu::xCompressIntraCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth)
-{
-    //PPAScopeEvent(TEncCu_xCompressIntraCU + depth);
-
-    TComPic* pic = outBestCU->getPic();
-
-    if (depth == 0)
-    {
-        // get original YUV data from picture
-        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU());
-    }
-    else
-    {
-        // copy partition YUV from depth 0 CTU cache
-        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], outBestCU->getZorderIdxInCU());
-    }
-
-    // variable for Early CU determination
-    bool bSubBranch = true;
-
-    // variable for Cbf fast mode PU decision
-    bool bBoundary = false;
-
-    uint32_t lpelx = outBestCU->getCUPelX();
-    uint32_t rpelx = lpelx + outBestCU->getCUSize(0) - 1;
-    uint32_t tpelx = outBestCU->getCUPelY();
-    uint32_t bpely = tpelx + outBestCU->getCUSize(0) - 1;
-    
-    // If slice start or slice end is within this cu...
-    TComSlice * slice = outTempCU->getPic()->getSlice();
-    bool bSliceEnd = (slice->getSliceCurEndCUAddr() > outTempCU->getSCUAddr() && slice->getSliceCurEndCUAddr() < outTempCU->getSCUAddr() + outTempCU->getTotalNumPart());
-    bool bInsidePicture = (rpelx < outBestCU->getSlice()->getSPS()->getPicWidthInLumaSamples()) && (bpely < outBestCU->getSlice()->getSPS()->getPicHeightInLumaSamples());
-
-    //Data for splitting
-    UChar nextDepth = depth + 1;
-    uint32_t partUnitIdx = 0;
-    TComDataCU* subBestPartCU[4], *subTempPartCU[4];
-
-    //We need to split; so dont try these modes
-    if (!bSliceEnd && bInsidePicture)
-    {
-        outTempCU->initEstData(depth);
-
-        xCheckRDCostIntra(outBestCU, outTempCU, SIZE_2Nx2N);
-        outTempCU->initEstData(depth);
-
-        if (depth == g_maxCUDepth - g_addCUDepth)
-        {
-            if (outTempCU->getCUSize(0) > (1 << outTempCU->getSlice()->getSPS()->getQuadtreeTULog2MinSize()))
-            {
-                xCheckRDCostIntra(outBestCU, outTempCU, SIZE_NxN);
-            }
-        }
-
-        m_entropyCoder->resetBits();
-        m_entropyCoder->encodeSplitFlag(outBestCU, 0, depth, true);
-        outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-        outBestCU->m_totalCost  = m_rdCost->calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
-
-        // Early CU determination
-        if (outBestCU->isSkipped(0))
-            bSubBranch = false;
-        else
-            bSubBranch = true;
-    }
-    else if (!(bSliceEnd && bInsidePicture))
-    {
-        bBoundary = true;
-    }
-
-    outTempCU->initEstData(depth);
-
-    // further split
-    if (bSubBranch && depth < g_maxCUDepth - g_addCUDepth)
-    {
-        for (; partUnitIdx < 4; partUnitIdx++)
-        {
-            subBestPartCU[partUnitIdx]     = m_bestCU[nextDepth];
-            subTempPartCU[partUnitIdx]     = m_tempCU[nextDepth];
-
-            subBestPartCU[partUnitIdx]->initSubCU(outTempCU, partUnitIdx, nextDepth);     // clear sub partition datas or init.
-            subTempPartCU[partUnitIdx]->initSubCU(outTempCU, partUnitIdx, nextDepth);     // clear sub partition datas or init.
-
-            bool bInSlice = subBestPartCU[partUnitIdx]->getSCUAddr() < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (subBestPartCU[partUnitIdx]->getCUPelX() < slice->getSPS()->getPicWidthInLumaSamples()) && (subBestPartCU[partUnitIdx]->getCUPelY() < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                if (0 == partUnitIdx) //initialize RD with previous depth buffer
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-                }
-                else
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[nextDepth][CI_NEXT_BEST]);
-                }
-
-                xCompressIntraCU(subBestPartCU[partUnitIdx], subTempPartCU[partUnitIdx], nextDepth);
-                outTempCU->copyPartFrom(subBestPartCU[partUnitIdx], partUnitIdx, nextDepth); // Keep best part data to current temporary data.
-                xCopyYuv2Tmp(subBestPartCU[partUnitIdx]->getTotalNumPart() * partUnitIdx, nextDepth);
-            }
-            else if (bInSlice)
-            {
-                subBestPartCU[partUnitIdx]->copyToPic(nextDepth);
-                outTempCU->copyPartFrom(subBestPartCU[partUnitIdx], partUnitIdx, nextDepth);
-            }
-        }
-
-        if (!bBoundary)
-        {
-            m_entropyCoder->resetBits();
-            m_entropyCoder->encodeSplitFlag(outTempCU, 0, depth, true);
-
-            outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-        }
-        outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-        if ((g_maxCUSize >> depth) == outTempCU->getSlice()->getPPS()->getMinCuDQPSize() && outTempCU->getSlice()->getPPS()->getUseDQP())
-        {
-            bool hasResidual = false;
-            for (uint32_t uiBlkIdx = 0; uiBlkIdx < outTempCU->getTotalNumPart(); uiBlkIdx++)
-            {
-                if (outTempCU->getCbf(uiBlkIdx, TEXT_LUMA) || outTempCU->getCbf(uiBlkIdx, TEXT_CHROMA_U) |
-                    outTempCU->getCbf(uiBlkIdx, TEXT_CHROMA_V))
-                {
-                    hasResidual = true;
-                    break;
-                }
-            }
-
-            uint32_t targetPartIdx = 0;
-            if (hasResidual)
-            {
-                bool foundNonZeroCbf = false;
-                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
-                assert(foundNonZeroCbf);
-            }
-            else
-            {
-                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
-            }
-        }
-
-        m_rdSbacCoders[nextDepth][CI_NEXT_BEST]->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-        xCheckBestMode(outBestCU, outTempCU, depth); // RD compare current prediction with split prediction.
-    }
-
-    outBestCU->copyToPic(depth); // Copy Best data to Picture for next partition prediction.
-
-    // Copy Yuv data to picture Yuv
-    xCopyYuv2Pic(outBestCU->getPic(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU(), depth, depth, outBestCU, lpelx, tpelx);
-
-    if (bBoundary || (bSliceEnd && bInsidePicture)) return;
-
-    // Assert if Best prediction mode is NONE
-    // Selected mode's RD-cost must be not MAX_INT64.
-    assert(outBestCU->getPartitionSize(0) != SIZE_NONE);
-    assert(outBestCU->getPredictionMode(0) != MODE_NONE);
-    assert(outBestCU->m_totalCost != MAX_INT64);
-}
-
-void TEncCu::xCompressCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, PartSize parentSize)
-{
-    //PPAScopeEvent(TEncCu_xCompressCU + depth);
-
-    TComPic* pic = outBestCU->getPic();
-
-    // get Original YUV data from picture
-    if (depth == 0)
-    {
-        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU());
-    }
-    else
-    {
-        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], outBestCU->getZorderIdxInCU());
-    }
-
-    // variable for Early CU determination
-    bool bSubBranch = true;
-
-    // variable for Cbf fast mode PU decision
-    bool doNotBlockPu = true;
-    bool earlyDetectionSkipMode = false;
-
-    bool bBoundary = false;
-    uint32_t lpelx = outBestCU->getCUPelX();
-    uint32_t rpelx = lpelx + outBestCU->getCUSize(0) - 1;
-    uint32_t tpely = outBestCU->getCUPelY();
-    uint32_t bpely = tpely + outBestCU->getCUSize(0) - 1;
-
-    // If slice start or slice end is within this cu...
-    TComSlice* slice = outTempCU->getPic()->getSlice();
-    bool bSliceEnd = (slice->getSliceCurEndCUAddr() > outTempCU->getSCUAddr() &&
-                      slice->getSliceCurEndCUAddr() < outTempCU->getSCUAddr() + outTempCU->getTotalNumPart());
-    bool bInsidePicture = (rpelx < outBestCU->getSlice()->getSPS()->getPicWidthInLumaSamples()) &&
-        (bpely < outBestCU->getSlice()->getSPS()->getPicHeightInLumaSamples());
-
-    // We need to split, so don't try these modes.
-    if (!bSliceEnd && bInsidePicture)
-    {
-        outTempCU->initEstData(depth);
-
-        // do inter modes, SKIP and 2Nx2N
-        if (outBestCU->getSlice()->getSliceType() != I_SLICE)
-        {
-            // 2Nx2N
-            if (m_param->bEnableEarlySkip)
-            {
-                xCheckRDCostInter(outBestCU, outTempCU, SIZE_2Nx2N);
-                outTempCU->initEstData(depth); // by competition for inter_2Nx2N
-            }
-            // by Merge for inter_2Nx2N
-            xCheckRDCostMerge2Nx2N(outBestCU, outTempCU, &earlyDetectionSkipMode, m_bestPredYuv[depth], m_bestRecoYuv[depth]);
-
-            outTempCU->initEstData(depth);
-
-            if (!m_param->bEnableEarlySkip)
-            {
-                // 2Nx2N, NxN
-                xCheckRDCostInter(outBestCU, outTempCU, SIZE_2Nx2N);
-                outTempCU->initEstData(depth);
-                if (m_param->bEnableCbfFastMode)
-                {
-                    doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                }
-            }
-        }
-
-        if (!earlyDetectionSkipMode)
-        {
-            outTempCU->initEstData(depth);
-
-            // do inter modes, NxN, 2NxN, and Nx2N
-            if (outBestCU->getSlice()->getSliceType() != I_SLICE)
-            {
-                // 2Nx2N, NxN
-                if (!(outBestCU->getCUSize(0) == 8))
-                {
-                    if (depth == g_maxCUDepth - g_addCUDepth && doNotBlockPu)
-                    {
-                        xCheckRDCostInter(outBestCU, outTempCU, SIZE_NxN);
-                        outTempCU->initEstData(depth);
-                    }
-                }
-
-                if (m_param->bEnableRectInter)
-                {
-                    // 2NxN, Nx2N
-                    if (doNotBlockPu)
-                    {
-                        xCheckRDCostInter(outBestCU, outTempCU, SIZE_Nx2N);
-                        outTempCU->initEstData(depth);
-                        if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_Nx2N)
-                        {
-                            doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                        }
-                    }
-                    if (doNotBlockPu)
-                    {
-                        xCheckRDCostInter(outBestCU, outTempCU, SIZE_2NxN);
-                        outTempCU->initEstData(depth);
-                        if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxN)
-                        {
-                            doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                        }
-                    }
-                }
-
-                // Try AMP (SIZE_2NxnU, SIZE_2NxnD, SIZE_nLx2N, SIZE_nRx2N)
-                if (pic->getSlice()->getSPS()->getAMPAcc(depth))
-                {
-                    bool bTestAMP_Hor = false, bTestAMP_Ver = false;
-                    bool bTestMergeAMP_Hor = false, bTestMergeAMP_Ver = false;
-
-                    deriveTestModeAMP(outBestCU, parentSize, bTestAMP_Hor, bTestAMP_Ver, bTestMergeAMP_Hor, bTestMergeAMP_Ver);
-
-                    // Do horizontal AMP
-                    if (bTestAMP_Hor)
-                    {
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_2NxnU);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnU)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_2NxnD);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnD)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                    }
-                    else if (bTestMergeAMP_Hor)
-                    {
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_2NxnU, true);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnU)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_2NxnD, true);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnD)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                    }
-
-                    // Do horizontal AMP
-                    if (bTestAMP_Ver)
-                    {
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_nLx2N);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_nLx2N)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_nRx2N);
-                            outTempCU->initEstData(depth);
-                        }
-                    }
-                    else if (bTestMergeAMP_Ver)
-                    {
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_nLx2N, true);
-                            outTempCU->initEstData(depth);
-                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_nLx2N)
-                            {
-                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
-                            }
-                        }
-                        if (doNotBlockPu)
-                        {
-                            xCheckRDCostInter(outBestCU, outTempCU, SIZE_nRx2N, true);
-                            outTempCU->initEstData(depth);
-                        }
-                    }
-                }
-            }
-
-            // do normal intra modes
-            // speedup for inter frames
-            if (outBestCU->getSlice()->getSliceType() == I_SLICE ||
-                outBestCU->getCbf(0, TEXT_LUMA) != 0   ||
-                outBestCU->getCbf(0, TEXT_CHROMA_U) != 0   ||
-                outBestCU->getCbf(0, TEXT_CHROMA_V) != 0) // avoid very complex intra if it is unlikely
-            {
-                xCheckRDCostIntraInInter(outBestCU, outTempCU, SIZE_2Nx2N);
-                outTempCU->initEstData(depth);
-
-                if (depth == g_maxCUDepth - g_addCUDepth)
-                {
-                    if (outTempCU->getCUSize(0) > (1 << outTempCU->getSlice()->getSPS()->getQuadtreeTULog2MinSize()))
-                    {
-                        xCheckRDCostIntraInInter(outBestCU, outTempCU, SIZE_NxN);
-                        outTempCU->initEstData(depth);
-                    }
-                }
-            }
-            // test PCM
-            if (pic->getSlice()->getSPS()->getUsePCM()
-                && outTempCU->getCUSize(0) <= (1 << pic->getSlice()->getSPS()->getPCMLog2MaxSize())
-                && outTempCU->getCUSize(0) >= (1 << pic->getSlice()->getSPS()->getPCMLog2MinSize()))
-            {
-                uint32_t rawbits = (2 * X265_DEPTH + X265_DEPTH) * outBestCU->getCUSize(0) * outBestCU->getCUSize(0) / 2;
-                uint32_t bestbits = outBestCU->m_totalBits;
-                if ((bestbits > rawbits) || (outBestCU->m_totalCost > m_rdCost->calcRdCost(0, rawbits)))
-                {
-                    xCheckIntraPCM(outBestCU, outTempCU);
-                }
-            }
-        }
-
-        m_entropyCoder->resetBits();
-        m_entropyCoder->encodeSplitFlag(outBestCU, 0, depth, true);
-        outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-        outBestCU->m_totalCost  = m_rdCost->calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
-
-        // Early CU determination
-        if (outBestCU->isSkipped(0))
-            bSubBranch = false;
-        else
-            bSubBranch = true;
-    }
-    else if (!(bSliceEnd && bInsidePicture))
-    {
-        bBoundary = true;
-    }
-
-    // copy original YUV samples to PCM buffer
-    if (outBestCU->isLosslessCoded(0) && (outBestCU->getIPCMFlag(0) == false))
-    {
-        xFillPCMBuffer(outBestCU, m_origYuv[depth]);
-    }
-
-    outTempCU->initEstData(depth);
-
-    // further split
-    if (bSubBranch && depth < g_maxCUDepth - g_addCUDepth)
-    {
-        UChar       nextDepth     = depth + 1;
-        TComDataCU* subBestPartCU = m_bestCU[nextDepth];
-        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
-        uint32_t partUnitIdx = 0;
-        for (; partUnitIdx < 4; partUnitIdx++)
-        {
-            subBestPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth); // clear sub partition datas or init.
-            subTempPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth); // clear sub partition datas or init.
-
-            bool bInSlice = subBestPartCU->getSCUAddr() < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (subBestPartCU->getCUPelX() < slice->getSPS()->getPicWidthInLumaSamples()) &&
-                (subBestPartCU->getCUPelY() < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                if (0 == partUnitIdx) //initialize RD with previous depth buffer
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-                }
-                else
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[nextDepth][CI_NEXT_BEST]);
-                }
-
-                xCompressCU(subBestPartCU, subTempPartCU, nextDepth);
-                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth); // Keep best part data to current temporary data.
-                xCopyYuv2Tmp(subBestPartCU->getTotalNumPart() * partUnitIdx, nextDepth);
-            }
-            else if (bInSlice)
-            {
-                subBestPartCU->copyToPic(nextDepth);
-                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth);
-            }
-        }
-
-        if (!bBoundary)
-        {
-            m_entropyCoder->resetBits();
-            m_entropyCoder->encodeSplitFlag(outTempCU, 0, depth, true);
-
-            outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-        }
-        outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-        if ((g_maxCUSize >> depth) == outTempCU->getSlice()->getPPS()->getMinCuDQPSize() && outTempCU->getSlice()->getPPS()->getUseDQP())
-        {
-            bool hasResidual = false;
-            for (uint32_t blkIdx = 0; blkIdx < outTempCU->getTotalNumPart(); blkIdx++)
-            {
-                if (outTempCU->getCbf(blkIdx, TEXT_LUMA) || outTempCU->getCbf(blkIdx, TEXT_CHROMA_U) ||
-                    outTempCU->getCbf(blkIdx, TEXT_CHROMA_V))
-                {
-                    hasResidual = true;
-                    break;
-                }
-            }
-
-            uint32_t targetPartIdx;
-            targetPartIdx = 0;
-            if (hasResidual)
-            {
-                bool foundNonZeroCbf = false;
-                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
-                assert(foundNonZeroCbf);
-            }
-            else
-            {
-                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
-            }
-        }
-
-        m_rdSbacCoders[nextDepth][CI_NEXT_BEST]->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-        xCheckBestMode(outBestCU, outTempCU, depth); // RD compare current CU against split
-    }
-    outBestCU->copyToPic(depth); // Copy Best data to Picture for next partition prediction.
-    // Copy Yuv data to picture Yuv
-    xCopyYuv2Pic(outBestCU->getPic(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU(), depth, depth, outBestCU, lpelx, tpely);
-
-    if (bBoundary || (bSliceEnd && bInsidePicture)) return;
-
-    // Assert if Best prediction mode is NONE
-    // Selected mode's RD-cost must be not MAX_INT64.
-    assert(outBestCU->getPartitionSize(0) != SIZE_NONE);
-    assert(outBestCU->getPredictionMode(0) != MODE_NONE);
-    assert(outBestCU->m_totalCost != MAX_INT64);
-}
-
-/** finish encoding a cu and handle end-of-slice conditions
- * \param cu
- * \param absPartIdx
- * \param depth
- * \returns void
- */
-void TEncCu::finishCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
-{
-    TComPic* pic = cu->getPic();
-    TComSlice* slice = cu->getPic()->getSlice();
-
-    //Calculate end address
-    uint32_t cuAddr = cu->getSCUAddr() + absPartIdx;
-
-    uint32_t internalAddress = (slice->getSliceCurEndCUAddr() - 1) % pic->getNumPartInCU();
-    uint32_t externalAddress = (slice->getSliceCurEndCUAddr() - 1) / pic->getNumPartInCU();
-    uint32_t posx = (externalAddress % pic->getFrameWidthInCU()) * g_maxCUSize + g_rasterToPelX[g_zscanToRaster[internalAddress]];
-    uint32_t posy = (externalAddress / pic->getFrameWidthInCU()) * g_maxCUSize + g_rasterToPelY[g_zscanToRaster[internalAddress]];
-    uint32_t width = slice->getSPS()->getPicWidthInLumaSamples();
-    uint32_t height = slice->getSPS()->getPicHeightInLumaSamples();
-
-    while (posx >= width || posy >= height)
-    {
-        internalAddress--;
-        posx = (externalAddress % pic->getFrameWidthInCU()) * g_maxCUSize + g_rasterToPelX[g_zscanToRaster[internalAddress]];
-        posy = (externalAddress / pic->getFrameWidthInCU()) * g_maxCUSize + g_rasterToPelY[g_zscanToRaster[internalAddress]];
-    }
-
-    internalAddress++;
-    if (internalAddress == cu->getPic()->getNumPartInCU())
-    {
-        internalAddress = 0;
-        externalAddress = (externalAddress + 1);
-    }
-    uint32_t realEndAddress = (externalAddress * pic->getNumPartInCU() + internalAddress);
-
-    // Encode slice finish
-    bool bTerminateSlice = false;
-    if (cuAddr + (cu->getPic()->getNumPartInCU() >> (depth << 1)) == realEndAddress)
-    {
-        bTerminateSlice = true;
-    }
-    uint32_t uiGranularityWidth = g_maxCUSize;
-    posx = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absPartIdx]];
-    posy = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absPartIdx]];
-    bool granularityBoundary = ((posx + cu->getCUSize(absPartIdx)) % uiGranularityWidth == 0 || (posx + cu->getCUSize(absPartIdx) == width))
-        && ((posy + cu->getCUSize(absPartIdx)) % uiGranularityWidth == 0 || (posy + cu->getCUSize(absPartIdx) == height));
-
-    if (granularityBoundary)
-    {
-        // The 1-terminating bit is added to all streams, so don't add it here when it's 1.
-        if (!bTerminateSlice)
-            m_entropyCoder->encodeTerminatingBit(bTerminateSlice ? 1 : 0);
-    }
-
-    int numberOfWrittenBits = 0;
-    if (m_bitCounter)
-    {
-        numberOfWrittenBits = m_entropyCoder->getNumberOfWrittenBits();
-    }
-
-    if (granularityBoundary)
-    {
-        slice->setSliceBits((uint32_t)(slice->getSliceBits() + numberOfWrittenBits));
-        slice->setSliceSegmentBits(slice->getSliceSegmentBits() + numberOfWrittenBits);
-        if (m_bitCounter)
-        {
-            m_entropyCoder->resetBits();
-        }
-    }
-}
-
-/** encode a CU block recursively
- * \param cu
- * \param absPartIdx
- * \param depth
- * \returns void
- */
-void TEncCu::xEncodeCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
-{
-    TComPic* pic = cu->getPic();
-
-    bool bBoundary = false;
-    uint32_t lpelx = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absPartIdx]];
-    uint32_t rpelx = lpelx + (g_maxCUSize >> depth) - 1;
-    uint32_t tpely = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absPartIdx]];
-    uint32_t bpely = tpely + (g_maxCUSize >> depth) - 1;
-
-    TComSlice* slice = cu->getPic()->getSlice();
-
-    // If slice start is within this cu...
-
-    // We need to split, so don't try these modes.
-    if ((rpelx < slice->getSPS()->getPicWidthInLumaSamples()) && (bpely < slice->getSPS()->getPicHeightInLumaSamples()))
-    {
-        m_entropyCoder->encodeSplitFlag(cu, absPartIdx, depth);
-    }
-    else
-    {
-        bBoundary = true;
-    }
-
-    if ((g_maxCUSize >> depth) >= cu->getSlice()->getPPS()->getMinCuDQPSize() && cu->getSlice()->getPPS()->getUseDQP())
-    {
-        setdQPFlag(true);
-    }
-
-    if (((depth < cu->getDepth(absPartIdx)) && (depth < (g_maxCUDepth - g_addCUDepth))) || bBoundary)
-    {
-        uint32_t qNumParts = (pic->getNumPartInCU() >> (depth << 1)) >> 2;
-        
-        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
-        {
-            lpelx = cu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absPartIdx]];
-            tpely = cu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absPartIdx]];
-            bool bInSlice = cu->getSCUAddr() + absPartIdx < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (lpelx < slice->getSPS()->getPicWidthInLumaSamples()) && (tpely < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                xEncodeCU(cu, absPartIdx, depth + 1);
-            }
-        }
-
-        return;
-    }
-
-    
-    if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        m_entropyCoder->encodeCUTransquantBypassFlag(cu, absPartIdx);
-    }
-    if (!cu->getSlice()->isIntra())
-    {
-        m_entropyCoder->encodeSkipFlag(cu, absPartIdx);
-    }
-
-    if (cu->isSkipped(absPartIdx))
-    {
-        m_entropyCoder->encodeMergeIndex(cu, absPartIdx);
-        finishCU(cu, absPartIdx, depth);
-        return;
-    }
-    m_entropyCoder->encodePredMode(cu, absPartIdx);
-
-    m_entropyCoder->encodePartSize(cu, absPartIdx, depth);
-
-    if (cu->isIntra(absPartIdx) && cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N)
-    {
-        m_entropyCoder->encodeIPCMInfo(cu, absPartIdx);
-
-        if (cu->getIPCMFlag(absPartIdx))
-        {
-            // Encode slice finish
-            finishCU(cu, absPartIdx, depth);
-            return;
-        }
-    }
-
-    // prediction Info ( Intra : direction mode, Inter : Mv, reference idx )
-    m_entropyCoder->encodePredInfo(cu, absPartIdx);
-
-    // Encode Coefficients
-    bool bCodeDQP = getdQPFlag();
-    m_entropyCoder->encodeCoeff(cu, absPartIdx, depth, cu->getCUSize(absPartIdx), cu->getCUSize(absPartIdx), bCodeDQP);
-    setdQPFlag(bCodeDQP);
-
-    // --- write terminating bit ---
-    finishCU(cu, absPartIdx, depth);
-}
-
-/** check RD costs for a CU block encoded with merge
- * \param outBestCU
- * \param outTempCU
- * \returns void
- */
-void TEncCu::xCheckRDCostMerge2Nx2N(TComDataCU*& outBestCU, TComDataCU*& outTempCU, bool *earlyDetectionSkipMode, TComYuv*& outBestPredYuv, TComYuv*& rpcYuvReconBest)
-{
-    assert(outTempCU->getSlice()->getSliceType() != I_SLICE);
-    TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS << 1]; // double length for mv of both lists
-    UChar interDirNeighbours[MRG_MAX_NUM_CANDS];
-    int numValidMergeCand = 0;
-
-    for (uint32_t i = 0; i < outTempCU->getSlice()->getMaxNumMergeCand(); ++i)
-    {
-        interDirNeighbours[i] = 0;
-    }
-
-    UChar depth = outTempCU->getDepth(0);
-    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to LCU level
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-    outTempCU->getInterMergeCandidates(0, 0, mvFieldNeighbours, interDirNeighbours, numValidMergeCand);
-
-    int mergeCandBuffer[MRG_MAX_NUM_CANDS];
-    for (uint32_t i = 0; i < numValidMergeCand; ++i)
-    {
-        mergeCandBuffer[i] = 0;
-    }
-
-    bool bestIsSkip = false;
-
-    uint32_t iteration;
-    if (outTempCU->isLosslessCoded(0))
-    {
-        iteration = 1;
-    }
-    else
-    {
-        iteration = 2;
-    }
-
-    for (uint32_t noResidual = 0; noResidual < iteration; ++noResidual)
-    {
-        for (uint32_t mergeCand = 0; mergeCand < numValidMergeCand; ++mergeCand)
-        {
-            if (m_param->frameNumThreads > 1 &&
-                (mvFieldNeighbours[0 + 2 * mergeCand].mv.y >= (m_param->searchRange + 1) * 4 ||
-                 mvFieldNeighbours[1 + 2 * mergeCand].mv.y >= (m_param->searchRange + 1) * 4))
-            {
-                continue;
-            }
-            if (!(noResidual && mergeCandBuffer[mergeCand] == 1))
-            {
-                if (!(bestIsSkip && !noResidual))
-                {
-                    // set MC parameters
-                    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth); // interprets depth relative to LCU level
-                    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-                    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to LCU level
-                    outTempCU->setMergeFlag(0, true);
-                    outTempCU->setMergeIndex(0, mergeCand);
-                    outTempCU->setInterDirSubParts(interDirNeighbours[mergeCand], 0, 0, depth); // interprets depth relative to LCU level
-                    outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[0 + 2 * mergeCand], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
-                    outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[1 + 2 * mergeCand], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
-
-                    // do MC
-                    m_search->motionCompensation(outTempCU, m_tmpPredYuv[depth], REF_PIC_LIST_X, 0);
-                    // estimate residual and encode everything
-                    m_search->encodeResAndCalcRdInterCU(outTempCU,
-                                                        m_origYuv[depth],
-                                                        m_tmpPredYuv[depth],
-                                                        m_tmpResiYuv[depth],
-                                                        m_bestResiYuv[depth],
-                                                        m_tmpRecoYuv[depth],
-                                                        !!noResidual,
-                                                        true);
-
-                    /* Todo: Fix the satd cost estimates. Why is merge being chosen in high motion areas: estimated distortion is too low? */
-                    if (!noResidual && !outTempCU->getQtRootCbf(0))
-                        mergeCandBuffer[mergeCand] = 1;
-
-                    outTempCU->setSkipFlagSubParts(!outTempCU->getQtRootCbf(0), 0, depth);
-                    int origQP = outTempCU->getQP(0);
-                    xCheckDQP(outTempCU);
-                    if (outTempCU->m_totalCost < outBestCU->m_totalCost)
-                    {
-                        TComDataCU* tmp = outTempCU;
-                        outTempCU = outBestCU;
-                        outBestCU = tmp;
-
-                        // Change Prediction data
-                        TComYuv* yuv = NULL;
-                        yuv = outBestPredYuv;
-                        outBestPredYuv = m_tmpPredYuv[depth];
-                        m_tmpPredYuv[depth] = yuv;
-
-                        yuv = rpcYuvReconBest;
-                        rpcYuvReconBest = m_tmpRecoYuv[depth];
-                        m_tmpRecoYuv[depth] = yuv;
-                    }
-                    outTempCU->initEstData(depth, origQP);
-                    if (!bestIsSkip)
-                    {
-                        bestIsSkip = !outBestCU->getQtRootCbf(0);
-                    }
-                }
-            }
-        }
-
-        if (noResidual == 0 && m_param->bEnableEarlySkip)
-        {
-            if (outBestCU->getQtRootCbf(0) == 0)
-            {
-                if (outBestCU->getMergeFlag(0))
-                {
-                    *earlyDetectionSkipMode = true;
-                }
-                else
-                {
-                    int mvsum = 0;
-                    for (uint32_t refListIdx = 0; refListIdx < 2; refListIdx++)
-                    {
-                        if (outBestCU->getSlice()->getNumRefIdx(refListIdx) > 0)
-                        {
-                            TComCUMvField* pcCUMvField = outBestCU->getCUMvField(refListIdx);
-                            int hor = abs(pcCUMvField->getMvd(0).x);
-                            int ver = abs(pcCUMvField->getMvd(0).y);
-                            mvsum += hor + ver;
-                        }
-                    }
-
-                    if (mvsum == 0)
-                    {
-                        *earlyDetectionSkipMode = true;
-                    }
-                }
-            }
-        }
-    }
-}
-
-void TEncCu::xCheckRDCostInter(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize, bool bUseMRG)
-{
-    UChar depth = outTempCU->getDepth(0);
-
-    outTempCU->setDepthSubParts(depth);
-    outTempCU->setSkipFlagSubParts(false, 0, depth);
-    outTempCU->setPartSizeSubParts(partSize, 0, depth);
-    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    m_tmpRecoYuv[depth]->clear(); // TODO: Are either of these clears necessary?
-    m_tmpResiYuv[depth]->clear();
-    if (m_search->predInterSearch(outTempCU, m_tmpPredYuv[depth], bUseMRG, true))
-    {
-        m_search->encodeResAndCalcRdInterCU(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_bestResiYuv[depth], m_tmpRecoYuv[depth], false, true);
-        xCheckDQP(outTempCU);
-        xCheckBestMode(outBestCU, outTempCU, depth);
-    }
-}
-
-void TEncCu::xCheckRDCostIntra(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize)
-{
-    //PPAScopeEvent(TEncCU_xCheckRDCostIntra + depth);
-    uint32_t depth = outTempCU->getDepth(0);
-
-    outTempCU->setSkipFlagSubParts(false, 0, depth);
-    outTempCU->setPartSizeSubParts(partSize, 0, depth);
-    outTempCU->setPredModeSubParts(MODE_INTRA, 0, depth);
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    m_search->estIntraPredQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
-
-    m_search->estIntraPredChromaQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
-
-    m_entropyCoder->resetBits();
-    if (outTempCU->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        m_entropyCoder->encodeCUTransquantBypassFlag(outTempCU, 0, true);
-    }
-    m_entropyCoder->encodeSkipFlag(outTempCU, 0, true);
-    m_entropyCoder->encodePredMode(outTempCU, 0, true);
-    m_entropyCoder->encodePartSize(outTempCU, 0, depth, true);
-    m_entropyCoder->encodePredInfo(outTempCU, 0, true);
-    m_entropyCoder->encodeIPCMInfo(outTempCU, 0, true);
-
-    // Encode Coefficients
-    bool bCodeDQP = getdQPFlag();
-    m_entropyCoder->encodeCoeff(outTempCU, 0, depth, outTempCU->getCUSize(0), outTempCU->getCUSize(0), bCodeDQP);
-    
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-
-    outTempCU->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
-    outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-    xCheckDQP(outTempCU);
-    xCheckBestMode(outBestCU, outTempCU, depth);
-}
-
-void TEncCu::xCheckRDCostIntraInInter(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize)
-{
-    uint32_t depth = outTempCU->getDepth(0);
-
-    PPAScopeEvent(TEncCU_xCheckRDCostIntra + depth);
-
-    outTempCU->setSkipFlagSubParts(false, 0, depth);
-    outTempCU->setPartSizeSubParts(partSize, 0, depth);
-    outTempCU->setPredModeSubParts(MODE_INTRA, 0, depth);
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    m_search->estIntraPredQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
-
-    m_search->estIntraPredChromaQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
-
-    m_entropyCoder->resetBits();
-    if (outTempCU->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        m_entropyCoder->encodeCUTransquantBypassFlag(outTempCU, 0, true);
-    }
-    m_entropyCoder->encodeSkipFlag(outTempCU, 0, true);
-    m_entropyCoder->encodePredMode(outTempCU, 0, true);
-    m_entropyCoder->encodePartSize(outTempCU, 0, depth, true);
-    m_entropyCoder->encodePredInfo(outTempCU, 0, true);
-    m_entropyCoder->encodeIPCMInfo(outTempCU, 0, true);
-
-    // Encode Coefficients
-    bool bCodeDQP = getdQPFlag();
-    m_entropyCoder->encodeCoeff(outTempCU, 0, depth, outTempCU->getCUSize(0), outTempCU->getCUSize(0), bCodeDQP);
-    
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-
-    outTempCU->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
-    outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-    xCheckDQP(outTempCU);
-    xCheckBestMode(outBestCU, outTempCU, depth);
-}
-
-/** Check R-D costs for a CU with PCM mode.
- * \param outBestCU pointer to best mode CU data structure
- * \param outTempCU pointer to testing mode CU data structure
- * \returns void
- *
- * \note Current PCM implementation encodes sample values in a lossless way. The distortion of PCM mode CUs are zero. PCM mode is selected if the best mode yields bits greater than that of PCM mode.
- */
-void TEncCu::xCheckIntraPCM(TComDataCU*& outBestCU, TComDataCU*& outTempCU)
-{
-    //PPAScopeEvent(TEncCU_xCheckIntraPCM);
-
-    uint32_t depth = outTempCU->getDepth(0);
-
-    outTempCU->setSkipFlagSubParts(false, 0, depth);
-    outTempCU->setIPCMFlag(0, true);
-    outTempCU->setIPCMFlagSubParts(true, 0, outTempCU->getDepth(0));
-    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth);
-    outTempCU->setPredModeSubParts(MODE_INTRA, 0, depth);
-    outTempCU->setTrIdxSubParts(0, 0, depth);
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    m_search->IPCMSearch(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
-
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-    m_entropyCoder->resetBits();
-    if (outTempCU->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        m_entropyCoder->encodeCUTransquantBypassFlag(outTempCU, 0, true);
-    }
-    m_entropyCoder->encodeSkipFlag(outTempCU, 0, true);
-    m_entropyCoder->encodePredMode(outTempCU, 0, true);
-    m_entropyCoder->encodePartSize(outTempCU, 0, depth, true);
-    m_entropyCoder->encodeIPCMInfo(outTempCU, 0, true);
-
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-
-    outTempCU->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
-    outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-    xCheckDQP(outTempCU);
-    xCheckBestMode(outBestCU, outTempCU, depth);
-}
-
-/** check whether current try is the best with identifying the depth of current try
- * \param outBestCU
- * \param outTempCU
- * \returns void
- */
-void TEncCu::xCheckBestMode(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth)
-{
-    if (outTempCU->m_totalCost < outBestCU->m_totalCost)
-    {
-        TComYuv* yuv;
-        // Change Information data
-        TComDataCU* cu = outBestCU;
-        outBestCU = outTempCU;
-        outTempCU = cu;
-
-        // Change Prediction data
-        yuv = m_bestPredYuv[depth];
-        m_bestPredYuv[depth] = m_tmpPredYuv[depth];
-        m_tmpPredYuv[depth] = yuv;
-
-        // Change Reconstruction data
-        yuv = m_bestRecoYuv[depth];
-        m_bestRecoYuv[depth] = m_tmpRecoYuv[depth];
-        m_tmpRecoYuv[depth] = yuv;
-
-        m_rdSbacCoders[depth][CI_TEMP_BEST]->store(m_rdSbacCoders[depth][CI_NEXT_BEST]);
-    }
-}
-
-void TEncCu::xCheckDQP(TComDataCU* cu)
-{
-    uint32_t depth = cu->getDepth(0);
-
-    if (cu->getSlice()->getPPS()->getUseDQP() && (g_maxCUSize >> depth) >= cu->getSlice()->getPPS()->getMinCuDQPSize())
-    {
-        if (!cu->getCbf(0, TEXT_LUMA, 0) && !cu->getCbf(0, TEXT_CHROMA_U, 0) && !cu->getCbf(0, TEXT_CHROMA_V, 0))
-            cu->setQPSubParts(cu->getRefQP(0), 0, depth); // set QP to default QP
-    }
-}
-
-void TEncCu::xCopyYuv2Pic(TComPic* outPic, uint32_t cuAddr, uint32_t absPartIdx, uint32_t depth, uint32_t srcDepth, TComDataCU* cu, uint32_t lpelx, uint32_t tpely)
-{
-    uint32_t rpelx = lpelx + (g_maxCUSize >> depth) - 1;
-    uint32_t bpely = tpely + (g_maxCUSize >> depth) - 1;
-    TComSlice* slice = cu->getPic()->getSlice();
-    bool bSliceEnd = slice->getSliceCurEndCUAddr() > (cu->getAddr()) * cu->getPic()->getNumPartInCU() + absPartIdx &&
-        slice->getSliceCurEndCUAddr() < (cu->getAddr()) * cu->getPic()->getNumPartInCU() + absPartIdx + (cu->getPic()->getNumPartInCU() >> (depth << 1));
-
-    if (!bSliceEnd && (rpelx < slice->getSPS()->getPicWidthInLumaSamples()) && (bpely < slice->getSPS()->getPicHeightInLumaSamples()))
-    {
-        uint32_t absPartIdxInRaster = g_zscanToRaster[absPartIdx];
-        uint32_t srcBlkWidth = outPic->getNumPartInWidth() >> (srcDepth);
-        uint32_t blkWidth    = outPic->getNumPartInWidth() >> (depth);
-        uint32_t partIdxX = ((absPartIdxInRaster % outPic->getNumPartInWidth()) % srcBlkWidth) / blkWidth;
-        uint32_t partIdxY = ((absPartIdxInRaster / outPic->getNumPartInWidth()) % srcBlkWidth) / blkWidth;
-        uint32_t partIdx = partIdxY * (srcBlkWidth / blkWidth) + partIdxX;
-        m_bestRecoYuv[srcDepth]->copyToPicYuv(outPic->getPicYuvRec(), cuAddr, absPartIdx, depth - srcDepth, partIdx);
-    }
-    else
-    {
-        uint32_t qNumParts = (cu->getPic()->getNumPartInCU() >> (depth << 1)) >> 2;
-
-        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
-        {
-            uint32_t subCULPelX = lpelx + (g_maxCUSize >> (depth + 1)) * (partUnitIdx &  1);
-            uint32_t subCUTPelY = tpely + (g_maxCUSize >> (depth + 1)) * (partUnitIdx >> 1);
-
-            bool bInSlice = cu->getAddr() * cu->getPic()->getNumPartInCU() + absPartIdx < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (subCULPelX < slice->getSPS()->getPicWidthInLumaSamples()) && (subCUTPelY < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                xCopyYuv2Pic(outPic, cuAddr, absPartIdx, depth + 1, srcDepth, cu, subCULPelX, subCUTPelY); // Copy Yuv data to picture Yuv
-            }
-        }
-    }
-}
-
-void TEncCu::xCopyYuv2Tmp(uint32_t partUnitIdx, uint32_t nextDepth)
-{
-    m_bestRecoYuv[nextDepth]->copyToPartYuv(m_tmpRecoYuv[nextDepth - 1], partUnitIdx);
-}
-
-/** Function for filling the PCM buffer of a CU using its original sample array
- * \param cu pointer to current CU
- * \param fencYuv pointer to original sample array
- * \returns void
- */
-void TEncCu::xFillPCMBuffer(TComDataCU* cu, TComYuv* fencYuv)
-{
-    uint32_t width = cu->getCUSize(0);
-    uint32_t height = cu->getCUSize(0);
-
-    Pel* srcY = fencYuv->getLumaAddr();
-    Pel* dstY = cu->getPCMSampleY();
-    uint32_t srcStride = fencYuv->getStride();
-
-    for (int y = 0; y < height; y++)
-    {
-        for (int x = 0; x < width; x++)
-        {
-            dstY[x] = srcY[x];
-        }
-
-        dstY += width;
-        srcY += srcStride;
-    }
-
-    Pel* srcCb = fencYuv->getCbAddr();
-    Pel* srcCr = fencYuv->getCrAddr();
-
-    Pel* dstCb = cu->getPCMSampleCb();
-    Pel* dstCr = cu->getPCMSampleCr();
-
-    uint32_t srcStrideC = fencYuv->getCStride();
-    uint32_t heightC = height >> 1;
-    uint32_t widthC = width >> 1;
-
-    for (int y = 0; y < heightC; y++)
-    {
-        for (int x = 0; x < widthC; x++)
-        {
-            dstCb[x] = srcCb[x];
-            dstCr[x] = srcCr[x];
-        }
-
-        dstCb += widthC;
-        dstCr += widthC;
-        srcCb += srcStrideC;
-        srcCr += srcStrideC;
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncCu.h Deleted

@@ -1,196 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncCu.h
-    \brief    Coding Unit (CU) encoder class (header)
-*/
-
-#ifndef X265_TENCCU_H
-#define X265_TENCCU_H
-
-#define ANGULAR_MODE_ID 2
-#define AMP_ID 3
-#define INTER_MODES 4
-#define INTRA_MODES 3
-
-#include "TLibCommon/CommonDef.h"
-#include "TLibCommon/TComYuv.h"
-#include "TLibCommon/TComPrediction.h"
-#include "TLibCommon/TComTrQuant.h"
-#include "TLibCommon/TComBitCounter.h"
-#include "TLibCommon/TComDataCU.h"
-#include "shortyuv.h"
-
-#include "TEncEntropy.h"
-#include "TEncSearch.h"
-
-//! \ingroup TLibEncoder
-//! \{
-struct StatisticLog
-{
-    uint64_t cntInter[4];
-    uint64_t cntIntra[4];
-    uint64_t cuInterDistribution[4][INTER_MODES];
-    uint64_t cuIntraDistribution[4][INTRA_MODES];
-    uint64_t cntIntraNxN;
-    uint64_t cntSkipCu[4];
-    uint64_t cntTotalCu[4];
-    uint64_t totalCu;
-
-    StatisticLog()
-    {
-        memset(this, 0, sizeof(StatisticLog));
-    }
-};
-
-namespace x265 {
-// private namespace
-
-class Encoder;
-class TEncSbac;
-class TEncCavlc;
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// CU encoder class
-class TEncCu
-{
-private:
-
-    static const int MAX_PRED_TYPES = 6;
-
-    TComDataCU** m_interCU_2Nx2N;
-    TComDataCU** m_interCU_2NxN;
-    TComDataCU** m_interCU_Nx2N;
-    TComDataCU** m_intraInInterCU;
-    TComDataCU** m_mergeCU;
-    TComDataCU** m_bestMergeCU;
-    TComDataCU** m_bestCU;      ///< Best CUs at each depth
-    TComDataCU** m_tempCU;      ///< Temporary CUs at each depth
-
-    TComYuv**    m_bestPredYuv; ///< Best Prediction Yuv for each depth
-    ShortYuv**  m_bestResiYuv; ///< Best Residual Yuv for each depth
-    TComYuv**    m_bestRecoYuv; ///< Best Reconstruction Yuv for each depth
-
-    TComYuv**    m_tmpPredYuv;  ///< Temporary Prediction Yuv for each depth
-    ShortYuv**  m_tmpResiYuv;  ///< Temporary Residual Yuv for each depth
-    TComYuv**    m_tmpRecoYuv;  ///< Temporary Reconstruction Yuv for each depth
-    TComYuv**    m_modePredYuv[MAX_PRED_TYPES]; ///< Prediction buffers for inter, intra, rect(2) and merge
-    TComYuv**    m_bestMergeRecoYuv;
-    TComYuv**    m_origYuv;     ///< Original Yuv at each depth
-
-    x265_param*  m_param;
-    TEncSearch*  m_search;
-    TComTrQuant* m_trQuant;
-    TComRdCost*  m_rdCost;
-    TEncEntropy* m_entropyCoder;
-    TComBitCounter* m_bitCounter;
-
-    // SBAC RD
-    TEncSbac***  m_rdSbacCoders;
-    TEncSbac*    m_rdGoOnSbacCoder;
-
-    UChar        m_totalDepth;
-
-    bool         m_bEncodeDQP;
-    bool         m_CUTransquantBypassFlagValue;
-
-public:
-
-#if LOG_CU_STATISTICS
-    StatisticLog  m_sliceTypeLog[3];
-    StatisticLog* m_log;
-#endif
-    TEncCu();
-
-    void init(Encoder* top);
-    bool create(UChar totalDepth, uint32_t maxWidth);
-    void destroy();
-    void compressCU(TComDataCU* cu);
-    void encodeCU(TComDataCU* cu);
-
-    void setRDSbacCoder(TEncSbac*** rdSbacCoder) { m_rdSbacCoders = rdSbacCoder; }
-
-    void setEntropyCoder(TEncEntropy* entropyCoder) { m_entropyCoder = entropyCoder; }
-
-    void setPredSearch(TEncSearch* predSearch) { m_search = predSearch; }
-
-    void setRDGoOnSbacCoder(TEncSbac* rdGoOnSbacCoder) { m_rdGoOnSbacCoder = rdGoOnSbacCoder; }
-
-    void setTrQuant(TComTrQuant* trQuant) { m_trQuant = trQuant; }
-
-    void setRdCost(TComRdCost* rdCost) { m_rdCost = rdCost; }
-
-    void setBitCounter(TComBitCounter* pcBitCounter) { m_bitCounter = pcBitCounter; }
-
-protected:
-
-    void finishCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
-    void xCompressCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, PartSize parentSize = SIZE_NONE);
-    void xCompressIntraCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth);
-    void xCompressInterCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComDataCU*& cu, uint32_t depth, uint32_t partitionIndex, UChar minDepth);
-    void xEncodeCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
-    void xCheckBestMode(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth);
-
-    void xCheckRDCostMerge2Nx2N(TComDataCU*& outBestCU, TComDataCU*& outTempCU, bool *earlyDetectionSkipMode,
-                                TComYuv*& outBestPredYuv, TComYuv*& rpcYuvReconBest);
-    void xComputeCostIntraInInter(TComDataCU* cu, PartSize partSize);
-    void xCheckRDCostInter(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize, bool bUseMRG = false);
-    void xComputeCostInter(TComDataCU* outTempCU, TComYuv* outPredYUV, PartSize partSize, bool bUseMRG = false);
-    void xComputeCostMerge2Nx2N(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComYuv*& bestPredYuv, TComYuv*& tmpPredYuv);
-    void xEncodeIntraInInter(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* outResiYuv, TComYuv* outReconYuv);
-    void encodeResidue(TComDataCU* lcu, TComDataCU* cu, uint32_t absPartIdx, UChar depth);
-    void xCheckRDCostIntra(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize);
-    void xCheckRDCostIntraInInter(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize);
-    void xCheckDQP(TComDataCU* cu);
-
-    void xCheckIntraPCM(TComDataCU*& outBestCU, TComDataCU*& outTempCU);
-    void xCopyYuv2Pic(TComPic* outPic, uint32_t cuAddr, uint32_t absPartIdx, uint32_t depth, uint32_t uiSrcDepth, TComDataCU* cu,
-                      uint32_t lpelx, uint32_t tpely);
-    void xCopyYuv2Tmp(uint32_t uhPartUnitIdx, uint32_t depth);
-
-    bool getdQPFlag()        { return m_bEncodeDQP; }
-
-    void setdQPFlag(bool b)  { m_bEncodeDQP = b; }
-
-    void deriveTestModeAMP(TComDataCU* bestCU, PartSize parentSize, bool &bTestAMP_Hor, bool &bTestAMP_Ver,
-                           bool &bTestMergeAMP_Hor, bool &bTestMergeAMP_Ver);
-
-    void xFillPCMBuffer(TComDataCU* outCU, TComYuv* origYuv);
-};
-}
-//! \}
-
-#endif // ifndef X265_TENCCU_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncEntropy.cpp Deleted

@@ -1,738 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncEntropy.cpp
-    \brief    entropy encoder class
-*/
-
-#include "TEncEntropy.h"
-#include "TLibCommon/TypeDef.h"
-#include "TLibCommon/TComSampleAdaptiveOffset.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-void TEncEntropy::setEntropyCoder(TEncEntropyIf* e, TComSlice* slice)
-{
-    m_entropyCoderIf = e;
-    m_entropyCoderIf->setSlice(slice);
-}
-
-void TEncEntropy::encodeSliceHeader(TComSlice* slice)
-{
-    if (slice->getSPS()->getUseSAO())
-    {
-        SAOParam *saoParam = slice->getPic()->getPicSym()->getSaoParam();
-        slice->setSaoEnabledFlag(saoParam->bSaoFlag[0]);
-        {
-            slice->setSaoEnabledFlagChroma(saoParam->bSaoFlag[1]);
-        }
-    }
-
-    m_entropyCoderIf->codeSliceHeader(slice);
-}
-
-void  TEncEntropy::encodeTilesWPPEntryPoint(TComSlice* pSlice)
-{
-    m_entropyCoderIf->codeTilesWPPEntryPoint(pSlice);
-}
-
-void TEncEntropy::encodeTerminatingBit(uint32_t isLast)
-{
-    m_entropyCoderIf->codeTerminatingBit(isLast);
-}
-
-void TEncEntropy::encodeSliceFinish()
-{
-    m_entropyCoderIf->codeSliceFinish();
-}
-
-void TEncEntropy::encodePPS(TComPPS* pps)
-{
-    m_entropyCoderIf->codePPS(pps);
-}
-
-void TEncEntropy::encodeSPS(TComSPS* sps)
-{
-    m_entropyCoderIf->codeSPS(sps);
-}
-
-void TEncEntropy::encodeCUTransquantBypassFlag(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    m_entropyCoderIf->codeCUTransquantBypassFlag(cu, absPartIdx);
-}
-
-void TEncEntropy::encodeVPS(TComVPS* vps)
-{
-    m_entropyCoderIf->codeVPS(vps);
-}
-
-void TEncEntropy::encodeSkipFlag(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (cu->getSlice()->isIntra())
-    {
-        return;
-    }
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    m_entropyCoderIf->codeSkipFlag(cu, absPartIdx);
-}
-
-/** encode merge flag
- * \param cu
- * \param absPartIdx
- * \returns void
- */
-void TEncEntropy::encodeMergeFlag(TComDataCU* cu, uint32_t absPartIdx)
-{
-    // at least one merge candidate exists
-    m_entropyCoderIf->codeMergeFlag(cu, absPartIdx);
-}
-
-/** encode merge index
- * \param cu
- * \param absPartIdx
- * \param uiPUIdx
- * \param bRD
- * \returns void
- */
-void TEncEntropy::encodeMergeIndex(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-        assert(cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N);
-    }
-    m_entropyCoderIf->codeMergeIndex(cu, absPartIdx);
-}
-
-/** encode prediction mode
- * \param cu
- * \param absPartIdx
- * \param bRD
- * \returns void
- */
-void TEncEntropy::encodePredMode(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    if (cu->getSlice()->isIntra())
-    {
-        return;
-    }
-
-    m_entropyCoderIf->codePredMode(cu, absPartIdx);
-}
-
-// Split mode
-void TEncEntropy::encodeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    m_entropyCoderIf->codeSplitFlag(cu, absPartIdx, depth);
-}
-
-/** encode partition size
- * \param cu
- * \param absPartIdx
- * \param depth
- * \param bRD
- * \returns void
- */
-void TEncEntropy::encodePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    m_entropyCoderIf->codePartSize(cu, absPartIdx, depth);
-}
-
-/** Encode I_PCM information.
- * \param cu pointer to CU
- * \param absPartIdx CU index
- * \param bRD flag indicating estimation or encoding
- * \returns void
- */
-void TEncEntropy::encodeIPCMInfo(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (!cu->getSlice()->getSPS()->getUsePCM()
-        || cu->getCUSize(absPartIdx) > (1 << cu->getSlice()->getSPS()->getPCMLog2MaxSize())
-        || cu->getCUSize(absPartIdx) < (1 << cu->getSlice()->getSPS()->getPCMLog2MinSize()))
-    {
-        return;
-    }
-
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-
-    m_entropyCoderIf->codeIPCMInfo(cu, absPartIdx);
-}
-
-void TEncEntropy::xEncodeTransform(TComDataCU* cu, uint32_t offsetLuma, uint32_t offsetChroma, uint32_t absPartIdx, uint32_t depth, uint32_t width, uint32_t height, uint32_t trIdx, bool& bCodeDQP)
-{
-    const uint32_t subdiv = cu->getTransformIdx(absPartIdx) + cu->getDepth(absPartIdx) > depth;
-    const uint32_t log2TrafoSize = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize()] + 2 - depth;
-    uint32_t cbfY = cu->getCbf(absPartIdx, TEXT_LUMA, trIdx);
-    uint32_t cbfU = cu->getCbf(absPartIdx, TEXT_CHROMA_U, trIdx);
-    uint32_t cbfV = cu->getCbf(absPartIdx, TEXT_CHROMA_V, trIdx);
-
-    if (trIdx == 0)
-    {
-        m_bakAbsPartIdxCU = absPartIdx;
-    }
-
-    if ((log2TrafoSize == 2) && !(cu->getChromaFormat() == CHROMA_444))
-    {
-        uint32_t partNum = cu->getPic()->getNumPartInCU() >> ((depth - 1) << 1);
-        if ((absPartIdx % partNum) == 0)
-        {
-            m_bakAbsPartIdx   = absPartIdx;
-            m_bakChromaOffset = offsetChroma;
-        }
-        else if ((absPartIdx % partNum) == (partNum - 1))
-        {
-            cbfU = cu->getCbf(m_bakAbsPartIdx, TEXT_CHROMA_U, trIdx);
-            cbfV = cu->getCbf(m_bakAbsPartIdx, TEXT_CHROMA_V, trIdx);
-        }
-    }
-
-    if (cu->getPredictionMode(absPartIdx) == MODE_INTRA && cu->getPartitionSize(absPartIdx) == SIZE_NxN && depth == cu->getDepth(absPartIdx))
-    {
-        assert(subdiv);
-    }
-    else if (cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N) && depth == cu->getDepth(absPartIdx) &&  (cu->getSlice()->getSPS()->getQuadtreeTUMaxDepthInter() == 1))
-    {
-        if (log2TrafoSize > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx))
-        {
-            assert(subdiv);
-        }
-        else
-        {
-            assert(!subdiv);
-        }
-    }
-    else if (log2TrafoSize > cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize())
-    {
-        assert(subdiv);
-    }
-    else if (log2TrafoSize == cu->getSlice()->getSPS()->getQuadtreeTULog2MinSize())
-    {
-        assert(!subdiv);
-    }
-    else if (log2TrafoSize == cu->getQuadtreeTULog2MinSizeInCU(absPartIdx))
-    {
-        assert(!subdiv);
-    }
-    else
-    {
-        assert(log2TrafoSize > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-        m_entropyCoderIf->codeTransformSubdivFlag(subdiv, 5 - log2TrafoSize);
-    }
-
-    const uint32_t trDepthCurr = depth - cu->getDepth(absPartIdx);
-    const bool bFirstCbfOfCU = trDepthCurr == 0;
-
-    bool mCodeAll = true;
-    const uint32_t numPels = (width >> cu->getHorzChromaShift()) * (height >> cu->getHorzChromaShift());
-    if (numPels < (MIN_TU_SIZE * MIN_TU_SIZE))
-    {
-        mCodeAll = false;
-    }
-
-    if (bFirstCbfOfCU || mCodeAll)
-    {
-        if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr - 1))
-        {
-            m_entropyCoderIf->codeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, trDepthCurr);
-        }
-        if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr - 1))
-        {
-            m_entropyCoderIf->codeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, trDepthCurr);
-        }
-    }
-    else
-    {
-        assert(cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr) == cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr - 1));
-        assert(cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr) == cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr - 1));
-    }
-
-    if (subdiv)
-    {
-        uint32_t size;
-        width  >>= 1;
-        height >>= 1;
-        size = width * height;
-        trIdx++;
-        ++depth;
-        const uint32_t partNum = cu->getPic()->getNumPartInCU() >> (depth << 1);
-
-        xEncodeTransform(cu, offsetLuma, offsetChroma, absPartIdx, depth, width, height, trIdx, bCodeDQP);
-
-        absPartIdx += partNum;
-        offsetLuma += size;
-        offsetChroma += (size >> (cu->getHorzChromaShift() + cu->getVertChromaShift()));
-        xEncodeTransform(cu, offsetLuma, offsetChroma, absPartIdx, depth, width, height, trIdx, bCodeDQP);
-
-        absPartIdx += partNum;
-        offsetLuma += size;
-        offsetChroma += (size >> (cu->getHorzChromaShift() + cu->getVertChromaShift()));
-        xEncodeTransform(cu, offsetLuma, offsetChroma, absPartIdx, depth, width, height, trIdx, bCodeDQP);
-
-        absPartIdx += partNum;
-        offsetLuma += size;
-        offsetChroma += (size >> (cu->getHorzChromaShift() + cu->getVertChromaShift()));
-        xEncodeTransform(cu, offsetLuma, offsetChroma, absPartIdx, depth, width, height, trIdx, bCodeDQP);
-    }
-    else
-    {
-        {
-            DTRACE_CABAC_VL(g_nSymbolCounter++);
-            DTRACE_CABAC_T("\tTrIdx: abspart=");
-            DTRACE_CABAC_V(absPartIdx);
-            DTRACE_CABAC_T("\tdepth=");
-            DTRACE_CABAC_V(depth);
-            DTRACE_CABAC_T("\ttrdepth=");
-            DTRACE_CABAC_V(cu->getTransformIdx(absPartIdx));
-            DTRACE_CABAC_T("\n");
-        }
-
-        if (cu->getPredictionMode(absPartIdx) != MODE_INTRA && depth == cu->getDepth(absPartIdx) && !cu->getCbf(absPartIdx, TEXT_CHROMA_U, 0) && !cu->getCbf(absPartIdx, TEXT_CHROMA_V, 0))
-        {
-            assert(cu->getCbf(absPartIdx, TEXT_LUMA, 0));
-            //      printf( "saved one bin! " );
-        }
-        else
-        {
-            m_entropyCoderIf->codeQtCbf(cu, absPartIdx, TEXT_LUMA, cu->getTransformIdx(absPartIdx));
-        }
-
-        if (cbfY || cbfU || cbfV)
-        {
-            // dQP: only for LCU once
-            if (cu->getSlice()->getPPS()->getUseDQP())
-            {
-                if (bCodeDQP)
-                {
-                    encodeQP(cu, m_bakAbsPartIdxCU);
-                    bCodeDQP = false;
-                }
-            }
-        }
-        if (cbfY)
-        {
-            m_entropyCoderIf->codeCoeffNxN(cu, (cu->getCoeffY() + offsetLuma), absPartIdx, width, depth, TEXT_LUMA);
-        }
-
-        if ((log2TrafoSize == 2) && !(cu->getChromaFormat() == CHROMA_444))
-        {
-            uint32_t partNum = cu->getPic()->getNumPartInCU() >> ((depth - 1) << 1);
-            if ((absPartIdx % partNum) == (partNum - 1))
-            {
-                // TODO: 4:2:2
-                assert(width == height);
-                if (cbfU)
-                {
-                    m_entropyCoderIf->codeCoeffNxN(cu, (cu->getCoeffCb() + m_bakChromaOffset), m_bakAbsPartIdx, width, depth, TEXT_CHROMA_U);
-                }
-                if (cbfV)
-                {
-                    m_entropyCoderIf->codeCoeffNxN(cu, (cu->getCoeffCr() + m_bakChromaOffset), m_bakAbsPartIdx, width, depth, TEXT_CHROMA_V);
-                }
-            }
-        }
-        else
-        {
-            int trWidth  = width >> cu->getHorzChromaShift();
-#if _DEBUG
-            // TODO: 4:2:2
-            int trHeight = height >> cu->getVertChromaShift();
-            assert(trWidth == trHeight);
-#endif
-            if (cbfU)
-            {
-                m_entropyCoderIf->codeCoeffNxN(cu, (cu->getCoeffCb() + offsetChroma), absPartIdx, trWidth, depth, TEXT_CHROMA_U);
-            }
-            if (cbfV)
-            {
-                m_entropyCoderIf->codeCoeffNxN(cu, (cu->getCoeffCr() + offsetChroma), absPartIdx, trWidth, depth, TEXT_CHROMA_V);
-            }
-        }
-    }
-}
-
-// Intra direction for Luma
-void TEncEntropy::encodeIntraDirModeLuma(TComDataCU* cu, uint32_t absPartIdx, bool isMultiplePU)
-{
-    m_entropyCoderIf->codeIntraDirLumaAng(cu, absPartIdx, isMultiplePU);
-}
-
-// Intra direction for Chroma
-void TEncEntropy::encodeIntraDirModeChroma(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-
-    m_entropyCoderIf->codeIntraDirChroma(cu, absPartIdx);
-}
-
-void TEncEntropy::encodePredInfo(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-    if (cu->isIntra(absPartIdx)) // If it is Intra mode, encode intra prediction mode.
-    {
-        encodeIntraDirModeLuma(cu, absPartIdx, true);
-        if (cu->getChromaFormat() != CHROMA_400)
-        {
-            encodeIntraDirModeChroma(cu, absPartIdx, bRD);
-
-            if ((cu->getChromaFormat() == CHROMA_444) && (cu->getPartitionSize(absPartIdx) == SIZE_NxN))
-            {
-                uint32_t partOffset = (cu->getPic()->getNumPartInCU() >> (cu->getDepth(absPartIdx) << 1)) >> 2;
-                encodeIntraDirModeChroma(cu, absPartIdx + partOffset, bRD);
-                encodeIntraDirModeChroma(cu, absPartIdx + partOffset * 2, bRD);
-                encodeIntraDirModeChroma(cu, absPartIdx + partOffset * 3, bRD);
-            }
-        }
-    }
-    else                        // if it is Inter mode, encode motion vector and reference index
-    {
-        encodePUWise(cu, absPartIdx, bRD);
-    }
-}
-
-/** encode motion information for every PU block
- * \param cu
- * \param absPartIdx
- * \param bRD
- * \returns void
- */
-void TEncEntropy::encodePUWise(TComDataCU* cu, uint32_t absPartIdx, bool bRD)
-{
-    if (bRD)
-    {
-        absPartIdx = 0;
-    }
-
-    PartSize partSize = cu->getPartitionSize(absPartIdx);
-    uint32_t numPU = (partSize == SIZE_2Nx2N ? 1 : (partSize == SIZE_NxN ? 4 : 2));
-    uint32_t depth = cu->getDepth(absPartIdx);
-    uint32_t puOffset = (g_puOffset[uint32_t(partSize)] << ((cu->getSlice()->getSPS()->getMaxCUDepth() - depth) << 1)) >> 4;
-
-    for (uint32_t partIdx = 0, subPartIdx = absPartIdx; partIdx < numPU; partIdx++, subPartIdx += puOffset)
-    {
-        encodeMergeFlag(cu, subPartIdx);
-        if (cu->getMergeFlag(subPartIdx))
-        {
-            encodeMergeIndex(cu, subPartIdx);
-        }
-        else
-        {
-            uint32_t interDir = cu->getInterDir(subPartIdx);
-            encodeInterDirPU(cu, subPartIdx);
-            for (uint32_t refListIdx = 0; refListIdx < 2; refListIdx++)
-            {
-                if (interDir & (1 << refListIdx))
-                {
-                    assert(cu->getSlice()->getNumRefIdx(refListIdx) > 0);
-
-                    encodeRefFrmIdxPU(cu, subPartIdx, refListIdx);
-                    encodeMvdPU(cu, subPartIdx, refListIdx);
-                    encodeMVPIdxPU(cu, subPartIdx, refListIdx);
-                }
-            }
-        }
-    }
-}
-
-void TEncEntropy::encodeInterDirPU(TComDataCU* cu, uint32_t absPartIdx)
-{
-    if (!cu->getSlice()->isInterB())
-    {
-        return;
-    }
-
-    m_entropyCoderIf->codeInterDir(cu, absPartIdx);
-}
-
-/** encode reference frame index for a PU block
- * \param cu
- * \param absPartIdx
- * \param eRefList
- * \returns void
- */
-void TEncEntropy::encodeRefFrmIdxPU(TComDataCU* cu, uint32_t absPartIdx, int list)
-{
-    assert(!cu->isIntra(absPartIdx));
-    {
-        if ((cu->getSlice()->getNumRefIdx(list) == 1))
-        {
-            return;
-        }
-
-        assert(cu->getInterDir(absPartIdx) & (1 << list));
-        {
-            m_entropyCoderIf->codeRefFrmIdx(cu, absPartIdx, list);
-        }
-    }
-}
-
-/** encode motion vector difference for a PU block
- * \param cu
- * \param absPartIdx
- * \param eRefList
- * \returns void
- */
-void TEncEntropy::encodeMvdPU(TComDataCU* cu, uint32_t absPartIdx, int list)
-{
-    assert(!cu->isIntra(absPartIdx));
-
-    assert(cu->getInterDir(absPartIdx) & (1 << list));
-    {
-        m_entropyCoderIf->codeMvd(cu, absPartIdx, list);
-    }
-}
-
-void TEncEntropy::encodeMVPIdxPU(TComDataCU* cu, uint32_t absPartIdx, int list)
-{
-    assert(cu->getInterDir(absPartIdx) & (1 << list));
-    {
-        m_entropyCoderIf->codeMVPIdx(cu->getMVPIdx(list, absPartIdx));
-    }
-}
-
-void TEncEntropy::encodeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth)
-{
-    m_entropyCoderIf->codeQtCbf(cu, absPartIdx, ttype, trDepth);
-}
-
-void TEncEntropy::encodeTransformSubdivFlag(uint32_t symbol, uint32_t ctx)
-{
-    m_entropyCoderIf->codeTransformSubdivFlag(symbol, ctx);
-}
-
-void TEncEntropy::encodeQtRootCbf(TComDataCU* cu, uint32_t absPartIdx)
-{
-    m_entropyCoderIf->codeQtRootCbf(cu, absPartIdx);
-}
-
-void TEncEntropy::encodeQtCbfZero(TComDataCU* cu, TextType ttype, uint32_t trDepth)
-{
-    m_entropyCoderIf->codeQtCbfZero(cu, ttype, trDepth);
-}
-
-void TEncEntropy::encodeQtRootCbfZero(TComDataCU* cu)
-{
-    m_entropyCoderIf->codeQtRootCbfZero(cu);
-}
-
-// dQP
-void TEncEntropy::encodeQP(TComDataCU* cu, uint32_t absPartIdx)
-{
-    m_entropyCoderIf->codeDeltaQP(cu, absPartIdx);
-}
-
-// texture
-
-/** encode coefficients
- * \param cu
- * \param absPartIdx
- * \param depth
- * \param width
- * \param height
- */
-void TEncEntropy::encodeCoeff(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, uint32_t width, uint32_t height, bool& bCodeDQP)
-{
-    uint32_t minCoeffSize = cu->getPic()->getMinCUSize() * cu->getPic()->getMinCUSize();
-    uint32_t lumaOffset   = minCoeffSize * absPartIdx;
-    uint32_t chromaOffset = lumaOffset >> (cu->getHorzChromaShift() + cu->getVertChromaShift());
-
-    if (cu->isIntra(absPartIdx))
-    {
-        DTRACE_CABAC_VL(g_nSymbolCounter++)
-        DTRACE_CABAC_T("\tdecodeTransformIdx()\tCUDepth=")
-        DTRACE_CABAC_V(depth)
-        DTRACE_CABAC_T("\n")
-    }
-    else
-    {
-        if (!(cu->getMergeFlag(absPartIdx) && cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N))
-        {
-            m_entropyCoderIf->codeQtRootCbf(cu, absPartIdx);
-        }
-        if (!cu->getQtRootCbf(absPartIdx))
-        {
-            return;
-        }
-    }
-
-    xEncodeTransform(cu, lumaOffset, chromaOffset, absPartIdx, depth, width, height, 0, bCodeDQP);
-}
-
-void TEncEntropy::encodeCoeffNxN(TComDataCU* cu, TCoeff* coeff, uint32_t absPartIdx, uint32_t trWidth, uint32_t, uint32_t depth, TextType ttype)
-{
-    // This is for Transform unit processing. This may be used at mode selection stage for Inter.
-    // TODO: 4:2:2
-    //assert(trWidth == trHeight);
-    m_entropyCoderIf->codeCoeffNxN(cu, coeff, absPartIdx, trWidth, depth, ttype);
-}
-
-void TEncEntropy::estimateBit(estBitsSbacStruct* estBitsSBac, int trSize, TextType ttype)
-{
-    ttype = ttype == TEXT_LUMA ? TEXT_LUMA : TEXT_CHROMA;
-
-    m_entropyCoderIf->estBit(estBitsSBac, trSize, ttype);
-}
-
-/** Encode SAO Offset
- * \param  saoLcuParam SAO LCU paramters
- */
-void TEncEntropy::encodeSaoOffset(SaoLcuParam* saoLcuParam, uint32_t compIdx)
-{
-    uint32_t symbol;
-    int i;
-
-    symbol = saoLcuParam->typeIdx + 1;
-    if (compIdx != 2)
-    {
-        m_entropyCoderIf->codeSaoTypeIdx(symbol);
-    }
-    if (symbol)
-    {
-        if (saoLcuParam->typeIdx < 4 && compIdx != 2)
-        {
-            saoLcuParam->subTypeIdx = saoLcuParam->typeIdx;
-        }
-        int offsetTh = 1 << X265_MIN(X265_DEPTH - 5, 5);
-        if (saoLcuParam->typeIdx == SAO_BO)
-        {
-            for (i = 0; i < saoLcuParam->length; i++)
-            {
-                uint32_t absOffset = ((saoLcuParam->offset[i] < 0) ? -saoLcuParam->offset[i] : saoLcuParam->offset[i]);
-                m_entropyCoderIf->codeSaoMaxUvlc(absOffset, offsetTh - 1);
-            }
-
-            for (i = 0; i < saoLcuParam->length; i++)
-            {
-                if (saoLcuParam->offset[i] != 0)
-                {
-                    uint32_t sign = (saoLcuParam->offset[i] < 0) ? 1 : 0;
-                    m_entropyCoderIf->codeSAOSign(sign);
-                }
-            }
-
-            symbol = (uint32_t)(saoLcuParam->subTypeIdx);
-            m_entropyCoderIf->codeSaoUflc(5, symbol);
-        }
-        else if (saoLcuParam->typeIdx < 4)
-        {
-            m_entropyCoderIf->codeSaoMaxUvlc(saoLcuParam->offset[0], offsetTh - 1);
-            m_entropyCoderIf->codeSaoMaxUvlc(saoLcuParam->offset[1], offsetTh - 1);
-            m_entropyCoderIf->codeSaoMaxUvlc(-saoLcuParam->offset[2], offsetTh - 1);
-            m_entropyCoderIf->codeSaoMaxUvlc(-saoLcuParam->offset[3], offsetTh - 1);
-            if (compIdx != 2)
-            {
-                symbol = (uint32_t)(saoLcuParam->subTypeIdx);
-                m_entropyCoderIf->codeSaoUflc(2, symbol);
-            }
-        }
-    }
-}
-
-/** Encode SAO unit interleaving
-* \param  rx
-* \param  ry
-* \param  pSaoParam
-* \param  cu
-* \param  iCUAddrInSlice
-* \param  iCUAddrUpInSlice
-* \param  bLFCrossSliceBoundaryFlag
- */
-void TEncEntropy::encodeSaoUnitInterleaving(int compIdx, bool saoFlag, int rx, int ry, SaoLcuParam* saoLcuParam, int cuAddrInSlice, int cuAddrUpInSlice, int allowMergeLeft, int allowMergeUp)
-{
-    if (saoFlag)
-    {
-        if (rx > 0 && cuAddrInSlice != 0 && allowMergeLeft)
-        {
-            m_entropyCoderIf->codeSaoMerge(saoLcuParam->mergeLeftFlag);
-        }
-        else
-        {
-            saoLcuParam->mergeLeftFlag = 0;
-        }
-        if (saoLcuParam->mergeLeftFlag == 0)
-        {
-            if ((ry > 0) && (cuAddrUpInSlice >= 0) && allowMergeUp)
-            {
-                m_entropyCoderIf->codeSaoMerge(saoLcuParam->mergeUpFlag);
-            }
-            else
-            {
-                saoLcuParam->mergeUpFlag = 0;
-            }
-            if (!saoLcuParam->mergeUpFlag)
-            {
-                encodeSaoOffset(saoLcuParam, compIdx);
-            }
-        }
-    }
-}
-
-/** encode quantization matrix
- * \param scalingList quantization matrix information
- */
-void TEncEntropy::encodeScalingList(TComScalingList* scalingList)
-{
-    m_entropyCoderIf->codeScalingList(scalingList);
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncEntropy.h Deleted

@@ -1,196 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncEntropy.h
-    \brief    entropy encoder class (header)
-*/
-
-#ifndef X265_TENCENTROPY_H
-#define X265_TENCENTROPY_H
-
-#include "TLibCommon/TComSlice.h"
-#include "TLibCommon/TComDataCU.h"
-#include "TLibCommon/TComBitStream.h"
-#include "TLibCommon/TComPic.h"
-#include "TLibCommon/TComTrQuant.h"
-#include "TLibCommon/TComSampleAdaptiveOffset.h"
-
-namespace x265 {
-// private namespace
-
-class TEncSbac;
-class TEncCavlc;
-class SEI;
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// entropy encoder pure class
-class TEncEntropyIf
-{
-public:
-
-    virtual void resetEntropy() = 0;
-    virtual void determineCabacInitIdx() = 0;
-    virtual void setBitstream(TComBitIf* p) = 0;
-    virtual void setSlice(TComSlice* p) = 0;
-    virtual void resetBits() = 0;
-    virtual uint32_t getNumberOfWrittenBits() = 0;
-
-    virtual void codeVPS(TComVPS* vps) = 0;
-    virtual void codeSPS(TComSPS* sps) = 0;
-    virtual void codePPS(TComPPS* pps) = 0;
-    virtual void codeSliceHeader(TComSlice* slice) = 0;
-
-    virtual void codeTilesWPPEntryPoint(TComSlice* slice) = 0;
-    virtual void codeTerminatingBit(uint32_t isLast) = 0;
-    virtual void codeSliceFinish() = 0;
-    virtual void codeMVPIdx(uint32_t symbol) = 0;
-    virtual void codeScalingList(TComScalingList* scalingList) = 0;
-
-public:
-
-    virtual void codeCUTransquantBypassFlag(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeSkipFlag(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeMergeFlag(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeMergeIndex(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth) = 0;
-
-    virtual void codePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth) = 0;
-    virtual void codePredMode(TComDataCU* cu, uint32_t absPartIdx) = 0;
-
-    virtual void codeIPCMInfo(TComDataCU* cu, uint32_t absPartIdx) = 0;
-
-    virtual void codeTransformSubdivFlag(uint32_t symbol, uint32_t ctx) = 0;
-    virtual void codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth) = 0;
-    virtual void codeQtRootCbf(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeQtCbfZero(TComDataCU* cu, TextType ttype, uint32_t trDepth) = 0;
-    virtual void codeQtRootCbfZero(TComDataCU* cu) = 0;
-    virtual void codeIntraDirLumaAng(TComDataCU* cu, uint32_t absPartIdx, bool isMultiplePU) = 0;
-
-    virtual void codeIntraDirChroma(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeInterDir(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeRefFrmIdx(TComDataCU* cu, uint32_t absPartIdx, int eRefList) = 0;
-    virtual void codeMvd(TComDataCU* cu, uint32_t absPartIdx, int eRefList) = 0;
-    virtual void codeDeltaQP(TComDataCU* cu, uint32_t absPartIdx) = 0;
-    virtual void codeCoeffNxN(TComDataCU* cu, TCoeff* pcCoef, uint32_t absPartIdx, uint32_t trSize, uint32_t depth, TextType ttype) = 0;
-    virtual void codeSAOSign(uint32_t code) = 0;
-    virtual void codeSaoMaxUvlc(uint32_t code, uint32_t maxSymbol) = 0;
-    virtual void codeSaoMerge(uint32_t code) = 0;
-    virtual void codeSaoTypeIdx(uint32_t code) = 0;
-    virtual void codeSaoUflc(uint32_t length, uint32_t code) = 0;
-    virtual void estBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype) = 0;
-
-    virtual ~TEncEntropyIf() {}
-};
-
-/// entropy encoder class
-class TEncEntropy
-{
-private:
-
-    uint32_t    m_bakAbsPartIdx;
-    uint32_t    m_bakChromaOffset;
-    uint32_t    m_bakAbsPartIdxCU;
-
-public:
-
-    void    setEntropyCoder(TEncEntropyIf* e, TComSlice* slice);
-    void    setBitstream(TComBitIf* p) { m_entropyCoderIf->setBitstream(p);  }
-
-    void    resetBits() { m_entropyCoderIf->resetBits();      }
-
-    uint32_t    getNumberOfWrittenBits() { return m_entropyCoderIf->getNumberOfWrittenBits(); }
-
-    void    resetEntropy() { m_entropyCoderIf->resetEntropy();  }
-
-    void    determineCabacInitIdx() { m_entropyCoderIf->determineCabacInitIdx(); }
-
-    void    encodeSliceHeader(TComSlice* slice);
-    void    encodeTilesWPPEntryPoint(TComSlice* pSlice);
-    void    encodeTerminatingBit(uint32_t uiIsLast);
-    void    encodeSliceFinish();
-    TEncEntropyIf*      m_entropyCoderIf;
-
-public:
-
-    void encodeVPS(TComVPS* vps);
-    // SPS
-    void encodeSPS(TComSPS* sps);
-    void encodePPS(TComPPS* pps);
-    void encodeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bRD = false);
-    void encodeCUTransquantBypassFlag(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodeSkipFlag(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodePUWise(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodeInterDirPU(TComDataCU* pcSubCU, uint32_t absPartIdx);
-    void encodeRefFrmIdxPU(TComDataCU* pcSubCU, uint32_t absPartIdx, int eRefList);
-    void encodeMvdPU(TComDataCU* pcSubCU, uint32_t absPartIdx, int eRefList);
-    void encodeMVPIdxPU(TComDataCU* pcSubCU, uint32_t absPartIdx, int eRefList);
-    void encodeMergeFlag(TComDataCU* cu, uint32_t absPartIdx);
-    void encodeMergeIndex(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodePredMode(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bRD = false);
-    void encodeIPCMInfo(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodePredInfo(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-    void encodeIntraDirModeLuma(TComDataCU* cu, uint32_t absPartIdx, bool isMultiplePU = false);
-
-    void encodeIntraDirModeChroma(TComDataCU* cu, uint32_t absPartIdx, bool bRD = false);
-
-    void encodeTransformSubdivFlag(uint32_t symbol, uint32_t ctx);
-    void encodeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth);
-    void encodeQtCbfZero(TComDataCU* cu, TextType ttype, uint32_t trDepth);
-    void encodeQtRootCbfZero(TComDataCU* cu);
-    void encodeQtRootCbf(TComDataCU* cu, uint32_t absPartIdx);
-    void encodeQP(TComDataCU* cu, uint32_t absPartIdx);
-
-    void encodeScalingList(TComScalingList* scalingList);
-
-private:
-
-    void xEncodeTransform(TComDataCU* cu, uint32_t offsetLumaOffset, uint32_t offsetChroma, uint32_t absPartIdx, uint32_t depth, uint32_t width, uint32_t height, uint32_t uiTrIdx, bool& bCodeDQP);
-
-public:
-
-    void encodeCoeff(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, uint32_t width, uint32_t height, bool& bCodeDQP);
-
-    void encodeCoeffNxN(TComDataCU* cu, TCoeff* pcCoeff, uint32_t absPartIdx, uint32_t trWidth, uint32_t trHeight, uint32_t depth, TextType ttype);
-
-    void estimateBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype);
-    void encodeSaoOffset(SaoLcuParam* saoLcuParam, uint32_t compIdx);
-    void encodeSaoUnitInterleaving(int compIdx, bool saoFlag, int rx, int ry, SaoLcuParam* saoLcuParam, int cuAddrInSlice, int cuAddrUpInSlice, int allowMergeLeft, int allowMergeUp);
-}; // END CLASS DEFINITION TEncEntropy
-}
-//! \}
-
-#endif // ifndef X265_TENCENTROPY_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSampleAdaptiveOffset.cpp Deleted

@@ -1,1962 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/**
- \file     TEncSampleAdaptiveOffset.cpp
- \brief       estimation part of sample adaptive offset class
- */
-
-#include "TEncSampleAdaptiveOffset.h"
-#include "common.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-TEncSampleAdaptiveOffset::TEncSampleAdaptiveOffset()
-    : m_entropyCoder(NULL)
-    , m_rdSbacCoders(NULL)
-    , m_rdGoOnSbacCoder(NULL)
-    , m_binCoderCABAC(NULL)
-    , m_count(NULL)
-    , m_offset(NULL)
-    , m_offsetOrg(NULL)
-    , m_countPreDblk(NULL)
-    , m_offsetOrgPreDblk(NULL)
-    , m_rate(NULL)
-    , m_dist(NULL)
-    , m_cost(NULL)
-    , m_costPartBest(NULL)
-    , m_distOrg(NULL)
-    , m_typePartBest(NULL)
-    , lumaLambda(0.)
-    , chromaLambda(0.)
-    , depth(0)
-{
-    m_depthSaoRate[0][0] = 0;
-    m_depthSaoRate[0][1] = 0;
-    m_depthSaoRate[0][2] = 0;
-    m_depthSaoRate[0][3] = 0;
-    m_depthSaoRate[1][0] = 0;
-    m_depthSaoRate[1][1] = 0;
-    m_depthSaoRate[1][2] = 0;
-    m_depthSaoRate[1][3] = 0;
-
-    m_saoBitIncreaseY = X265_MAX(X265_DEPTH - 10, 0);
-    m_saoBitIncreaseC = X265_MAX(X265_DEPTH - 10, 0);
-    m_offsetThY = 1 << X265_MIN(X265_DEPTH - 5, 5);
-    m_offsetThC = 1 << X265_MIN(X265_DEPTH - 5, 5);
-}
-
-TEncSampleAdaptiveOffset::~TEncSampleAdaptiveOffset()
-{}
-
-// ====================================================================================================================
-// Static
-// ====================================================================================================================
-
-// ====================================================================================================================
-// Constants
-// ====================================================================================================================
-
-// ====================================================================================================================
-// Tables
-// ====================================================================================================================
-
-#if HIGH_BIT_DEPTH
-inline double xRoundIbdi2(double x)
-{
-    return ((x) > 0) ? (int)(((int)(x) + (1 << (X265_DEPTH - 8 - 1))) / (1 << (X265_DEPTH - 8))) : ((int)(((int)(x) - (1 << (X265_DEPTH - 8 - 1))) / (1 << (X265_DEPTH - 8))));
-}
-
-#endif
-
-/** rounding with IBDI
- * \param  x
- */
-inline double xRoundIbdi(double x)
-{
-#if HIGH_BIT_DEPTH
-    return X265_DEPTH > 8 ? xRoundIbdi2(x) : ((x) >= 0 ? ((int)((x) + 0.5)) : ((int)((x) - 0.5)));
-#else
-    return (x) >= 0 ? ((int)((x) + 0.5)) : ((int)((x) - 0.5));
-#endif
-}
-
-/** process SAO for one partition
- * \param  *psQTPart, partIdx, lambda
- */
-void TEncSampleAdaptiveOffset::rdoSaoOnePart(SAOQTPart *psQTPart, int partIdx, double lambda, int yCbCr)
-{
-    int typeIdx;
-    int numTotalType = MAX_NUM_SAO_TYPE;
-    SAOQTPart* onePart = &(psQTPart[partIdx]);
-
-    int64_t estDist;
-    int classIdx;
-    int shift = 2 * DISTORTION_PRECISION_ADJUSTMENT(X265_DEPTH - 8);
-
-    m_distOrg[partIdx] =  0;
-
-    double bestRDCostTableBo = MAX_DOUBLE;
-    int    bestClassTableBo  = 0;
-    int    currentDistortionTableBo[MAX_NUM_SAO_CLASS];
-    double currentRdCostTableBo[MAX_NUM_SAO_CLASS];
-
-    int allowMergeLeft;
-    int allowMergeUp;
-    SaoLcuParam saoLcuParamRdo;
-
-    for (typeIdx = -1; typeIdx < numTotalType; typeIdx++)
-    {
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[onePart->partLevel][CI_CURR_BEST]);
-        m_rdGoOnSbacCoder->resetBits();
-
-        if (typeIdx == -1)
-        {
-            for (int ry = onePart->startCUY; ry <= onePart->endCUY; ry++)
-            {
-                for (int rx = onePart->startCUX; rx <= onePart->endCUX; rx++)
-                {
-                    // get bits for iTypeIdx = -1
-                    allowMergeLeft = 1;
-                    allowMergeUp   = 1;
-
-                    // reset
-                    resetSaoUnit(&saoLcuParamRdo);
-
-                    // set merge flag
-                    saoLcuParamRdo.mergeUpFlag   = 1;
-                    saoLcuParamRdo.mergeLeftFlag = 1;
-
-                    if (ry == onePart->startCUY)
-                    {
-                        saoLcuParamRdo.mergeUpFlag = 0;
-                    }
-
-                    if (rx == onePart->startCUX)
-                    {
-                        saoLcuParamRdo.mergeLeftFlag = 0;
-                    }
-
-                    m_entropyCoder->encodeSaoUnitInterleaving(yCbCr, 1, rx, ry,  &saoLcuParamRdo, 1,  1,  allowMergeLeft, allowMergeUp);
-                }
-            }
-        }
-
-        if (typeIdx >= 0)
-        {
-            estDist = estSaoTypeDist(partIdx, typeIdx, shift, lambda, currentDistortionTableBo, currentRdCostTableBo);
-            if (typeIdx == SAO_BO)
-            {
-                // Estimate Best Position
-                double currentRDCost = 0.0;
-
-                for (int i = 0; i < SAO_MAX_BO_CLASSES - SAO_BO_LEN + 1; i++)
-                {
-                    currentRDCost = 0.0;
-                    for (uint32_t uj = i; uj < i + SAO_BO_LEN; uj++)
-                    {
-                        currentRDCost += currentRdCostTableBo[uj];
-                    }
-
-                    if (currentRDCost < bestRDCostTableBo)
-                    {
-                        bestRDCostTableBo = currentRDCost;
-                        bestClassTableBo  = i;
-                    }
-                }
-
-                // Re code all Offsets
-                // Code Center
-                for (classIdx = bestClassTableBo; classIdx < bestClassTableBo + SAO_BO_LEN; classIdx++)
-                {
-                    estDist += currentDistortionTableBo[classIdx];
-                }
-            }
-
-            for (int ry = onePart->startCUY; ry <= onePart->endCUY; ry++)
-            {
-                for (int rx = onePart->startCUX; rx <= onePart->endCUX; rx++)
-                {
-                    // get bits for iTypeIdx = -1
-                    allowMergeLeft = 1;
-                    allowMergeUp   = 1;
-
-                    // reset
-                    resetSaoUnit(&saoLcuParamRdo);
-
-                    // set merge flag
-                    saoLcuParamRdo.mergeUpFlag   = 1;
-                    saoLcuParamRdo.mergeLeftFlag = 1;
-
-                    if (ry == onePart->startCUY)
-                    {
-                        saoLcuParamRdo.mergeUpFlag = 0;
-                    }
-
-                    if (rx == onePart->startCUX)
-                    {
-                        saoLcuParamRdo.mergeLeftFlag = 0;
-                    }
-
-                    // set type and offsets
-                    saoLcuParamRdo.typeIdx = typeIdx;
-                    saoLcuParamRdo.subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo : 0;
-                    saoLcuParamRdo.length = m_numClass[typeIdx];
-                    for (classIdx = 0; classIdx < saoLcuParamRdo.length; classIdx++)
-                    {
-                        saoLcuParamRdo.offset[classIdx] = (int)m_offset[partIdx][typeIdx][classIdx + saoLcuParamRdo.subTypeIdx + 1];
-                    }
-
-                    m_entropyCoder->encodeSaoUnitInterleaving(yCbCr, 1, rx, ry,  &saoLcuParamRdo, 1,  1,  allowMergeLeft, allowMergeUp);
-                }
-            }
-
-            m_dist[partIdx][typeIdx] = estDist;
-            m_rate[partIdx][typeIdx] = m_entropyCoder->getNumberOfWrittenBits();
-
-            m_cost[partIdx][typeIdx] = (double)((double)m_dist[partIdx][typeIdx] + lambda * (double)m_rate[partIdx][typeIdx]);
-
-            if (m_cost[partIdx][typeIdx] < m_costPartBest[partIdx])
-            {
-                m_distOrg[partIdx] = 0;
-                m_costPartBest[partIdx] = m_cost[partIdx][typeIdx];
-                m_typePartBest[partIdx] = typeIdx;
-                m_rdGoOnSbacCoder->store(m_rdSbacCoders[onePart->partLevel][CI_TEMP_BEST]);
-            }
-        }
-        else
-        {
-            if (m_distOrg[partIdx] < m_costPartBest[partIdx])
-            {
-                m_costPartBest[partIdx] = (double)m_distOrg[partIdx] + m_entropyCoder->getNumberOfWrittenBits() * lambda;
-                m_typePartBest[partIdx] = -1;
-                m_rdGoOnSbacCoder->store(m_rdSbacCoders[onePart->partLevel][CI_TEMP_BEST]);
-            }
-        }
-    }
-
-    onePart->bProcessed = true;
-    onePart->bSplit    = false;
-    onePart->minDist   =       m_typePartBest[partIdx] >= 0 ? m_dist[partIdx][m_typePartBest[partIdx]] : m_distOrg[partIdx];
-    onePart->minRate   = (int)(m_typePartBest[partIdx] >= 0 ? m_rate[partIdx][m_typePartBest[partIdx]] : 0);
-    onePart->minCost   = onePart->minDist + lambda * onePart->minRate;
-    onePart->bestType  = m_typePartBest[partIdx];
-    if (onePart->bestType != -1)
-    {
-        // pOnePart->bEnableFlag =  1;
-        onePart->length = m_numClass[onePart->bestType];
-        int minIndex = 0;
-        if (onePart->bestType == SAO_BO)
-        {
-            onePart->subTypeIdx = bestClassTableBo;
-            minIndex = onePart->subTypeIdx;
-        }
-        for (int i = 0; i < onePart->length; i++)
-        {
-            onePart->offset[i] = (int)m_offset[partIdx][onePart->bestType][minIndex + i + 1];
-        }
-    }
-    else
-    {
-        // pOnePart->bEnableFlag = 0;
-        onePart->length     = 0;
-    }
-}
-
-/** Run partition tree disable
- */
-void TEncSampleAdaptiveOffset::disablePartTree(SAOQTPart *psQTPart, int partIdx)
-{
-    SAOQTPart* pOnePart = &(psQTPart[partIdx]);
-
-    pOnePart->bSplit     = false;
-    pOnePart->length     =  0;
-    pOnePart->bestType   = -1;
-
-    if (pOnePart->partLevel < m_maxSplitLevel)
-    {
-        for (int i = 0; i < NUM_DOWN_PART; i++)
-        {
-            disablePartTree(psQTPart, pOnePart->downPartsIdx[i]);
-        }
-    }
-}
-
-/** Run quadtree decision function
- * \param  partIdx, pcPicOrg, pcPicDec, pcPicRest, &costFinal
- */
-void TEncSampleAdaptiveOffset::runQuadTreeDecision(SAOQTPart *qtPart, int partIdx, double &costFinal, int maxLevel, double lambda, int yCbCr)
-{
-    SAOQTPart* onePart = &(qtPart[partIdx]);
-
-    uint32_t nextDepth = onePart->partLevel + 1;
-
-    if (partIdx == 0)
-    {
-        costFinal = 0;
-    }
-
-    // SAO for this part
-    if (!onePart->bProcessed)
-    {
-        rdoSaoOnePart(qtPart, partIdx, lambda, yCbCr);
-    }
-
-    // SAO for sub 4 parts
-    if (onePart->partLevel < maxLevel)
-    {
-        double costNotSplit = lambda + onePart->minCost;
-        double costSplit    = lambda;
-
-        for (int i = 0; i < NUM_DOWN_PART; i++)
-        {
-            if (0 == i) //initialize RD with previous depth buffer
-            {
-                m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[onePart->partLevel][CI_CURR_BEST]);
-            }
-            else
-            {
-                m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[nextDepth][CI_NEXT_BEST]);
-            }
-            runQuadTreeDecision(qtPart, onePart->downPartsIdx[i], costFinal, maxLevel, lambda, yCbCr);
-            costSplit += costFinal;
-            m_rdSbacCoders[nextDepth][CI_NEXT_BEST]->load(m_rdSbacCoders[nextDepth][CI_TEMP_BEST]);
-        }
-
-        if (costSplit < costNotSplit)
-        {
-            costFinal = costSplit;
-            onePart->bSplit   = true;
-            onePart->length   =  0;
-            onePart->bestType = -1;
-            m_rdSbacCoders[onePart->partLevel][CI_NEXT_BEST]->load(m_rdSbacCoders[nextDepth][CI_NEXT_BEST]);
-        }
-        else
-        {
-            costFinal = costNotSplit;
-            onePart->bSplit = false;
-            for (int i = 0; i < NUM_DOWN_PART; i++)
-            {
-                disablePartTree(qtPart, onePart->downPartsIdx[i]);
-            }
-
-            m_rdSbacCoders[onePart->partLevel][CI_NEXT_BEST]->load(m_rdSbacCoders[onePart->partLevel][CI_TEMP_BEST]);
-        }
-    }
-    else
-    {
-        costFinal = onePart->minCost;
-    }
-}
-
-/** delete allocated memory of TEncSampleAdaptiveOffset class.
- */
-void TEncSampleAdaptiveOffset::destroyEncBuffer()
-{
-    for (int i = 0; i < m_numTotalParts; i++)
-    {
-        for (int j = 0; j < MAX_NUM_SAO_TYPE; j++)
-        {
-            X265_FREE(m_count[i][j]);
-            X265_FREE(m_offset[i][j]);
-            X265_FREE(m_offsetOrg[i][j]);
-        }
-
-        X265_FREE(m_rate[i]);
-        X265_FREE(m_dist[i]);
-        X265_FREE(m_cost[i]);
-        X265_FREE(m_count[i]);
-        X265_FREE(m_offset[i]);
-        X265_FREE(m_offsetOrg[i]);
-    }
-
-    X265_FREE(m_distOrg);
-    m_distOrg = NULL;
-    X265_FREE(m_costPartBest);
-    m_costPartBest = NULL;
-    X265_FREE(m_typePartBest);
-    m_typePartBest = NULL;
-    X265_FREE(m_rate);
-    m_rate = NULL;
-    X265_FREE(m_dist);
-    m_dist = NULL;
-    X265_FREE(m_cost);
-    m_cost = NULL;
-    X265_FREE(m_count);
-    m_count = NULL;
-    X265_FREE(m_offset);
-    m_offset = NULL;
-    X265_FREE(m_offsetOrg);
-    m_offsetOrg = NULL;
-
-    delete[] m_countPreDblk;
-    m_countPreDblk = NULL;
-
-    delete[] m_offsetOrgPreDblk;
-    m_offsetOrgPreDblk = NULL;
-
-    int maxDepth = 4;
-    for (int d = 0; d < maxDepth + 1; d++)
-    {
-        for (int iCIIdx = 0; iCIIdx < CI_NUM_SAO; iCIIdx++)
-        {
-            X265_FREE(m_rdSbacCoders[d][iCIIdx]);
-            X265_FREE(m_binCoderCABAC[d][iCIIdx]);
-        }
-    }
-
-    for (int d = 0; d < maxDepth + 1; d++)
-    {
-        X265_FREE(m_rdSbacCoders[d]);
-        X265_FREE(m_binCoderCABAC[d]);
-    }
-
-    X265_FREE(m_rdSbacCoders);
-    X265_FREE(m_binCoderCABAC);
-}
-
-/** create Encoder Buffer for SAO
- * \param
- */
-void TEncSampleAdaptiveOffset::createEncBuffer()
-{
-    m_distOrg = X265_MALLOC(int64_t, m_numTotalParts);
-    m_costPartBest = X265_MALLOC(double, m_numTotalParts);
-    m_typePartBest = X265_MALLOC(int, m_numTotalParts);
-
-    m_rate = X265_MALLOC(int64_t*, m_numTotalParts);
-    m_dist = X265_MALLOC(int64_t*, m_numTotalParts);
-    m_cost = X265_MALLOC(double*, m_numTotalParts);
-
-    m_count  = X265_MALLOC(int64_t**, m_numTotalParts);
-    m_offset = X265_MALLOC(int64_t**, m_numTotalParts);
-    m_offsetOrg = X265_MALLOC(int64_t**, m_numTotalParts);
-
-    for (int i = 0; i < m_numTotalParts; i++)
-    {
-        m_rate[i] = X265_MALLOC(int64_t, MAX_NUM_SAO_TYPE);
-        m_dist[i] = X265_MALLOC(int64_t, MAX_NUM_SAO_TYPE);
-        m_cost[i] = X265_MALLOC(double, MAX_NUM_SAO_TYPE);
-
-        m_count[i] = X265_MALLOC(int64_t*, MAX_NUM_SAO_TYPE);
-        m_offset[i] = X265_MALLOC(int64_t*, MAX_NUM_SAO_TYPE);
-        m_offsetOrg[i] = X265_MALLOC(int64_t*, MAX_NUM_SAO_TYPE);
-
-        for (int j = 0; j < MAX_NUM_SAO_TYPE; j++)
-        {
-            m_count[i][j] = X265_MALLOC(int64_t, MAX_NUM_SAO_CLASS);
-            m_offset[i][j] = X265_MALLOC(int64_t, MAX_NUM_SAO_CLASS);
-            m_offsetOrg[i][j] = X265_MALLOC(int64_t, MAX_NUM_SAO_CLASS);
-        }
-    }
-
-    int numLcu = m_numCuInWidth * m_numCuInHeight;
-    if (m_countPreDblk == NULL)
-    {
-        assert(m_offsetOrgPreDblk == NULL);
-
-        m_countPreDblk  = new int64_t[numLcu][3][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-        m_offsetOrgPreDblk = new int64_t[numLcu][3][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-    }
-
-    int maxDepth = 4;
-    m_rdSbacCoders = X265_MALLOC(TEncSbac**, maxDepth + 1);
-    m_binCoderCABAC = X265_MALLOC(TEncBinCABAC**, maxDepth + 1);
-
-    for (int d = 0; d < maxDepth + 1; d++)
-    {
-        m_rdSbacCoders[d] = X265_MALLOC(TEncSbac*, CI_NUM_SAO);
-        m_binCoderCABAC[d] = X265_MALLOC(TEncBinCABAC*, CI_NUM_SAO);
-        for (int ciIdx = 0; ciIdx < CI_NUM_SAO; ciIdx++)
-        {
-            m_rdSbacCoders[d][ciIdx] = X265_MALLOC(TEncSbac, 1);
-            m_binCoderCABAC[d][ciIdx] = X265_MALLOC(TEncBinCABAC, true);
-            m_rdSbacCoders[d][ciIdx]->init(m_binCoderCABAC[d][ciIdx]);
-        }
-    }
-}
-
-/** Start SAO encoder
- * \param pic, entropyCoder, rdSbacCoder, rdGoOnSbacCoder
- */
-void TEncSampleAdaptiveOffset::startSaoEnc(TComPic* pic, TEncEntropy* entropyCoder, TEncSbac* rdGoOnSbacCoder)
-{
-    m_pic = pic;
-    m_entropyCoder = entropyCoder;
-
-    m_rdGoOnSbacCoder = rdGoOnSbacCoder;
-    m_entropyCoder->setEntropyCoder(m_rdGoOnSbacCoder, pic->getSlice());
-    m_entropyCoder->resetEntropy();
-    m_entropyCoder->resetBits();
-
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_NEXT_BEST]);
-    m_rdSbacCoders[0][CI_CURR_BEST]->load(m_rdSbacCoders[0][CI_NEXT_BEST]);
-}
-
-/** End SAO encoder
- */
-void TEncSampleAdaptiveOffset::endSaoEnc()
-{
-    m_pic = NULL;
-    m_entropyCoder = NULL;
-}
-
-inline int xSign(int x)
-{
-    return (x >> 31) | ((int)((((uint32_t)-x)) >> 31));
-}
-
-/** Calculate SAO statistics for current LCU without non-crossing slice
- * \param  addr,  partIdx,  yCbCr
- */
-void TEncSampleAdaptiveOffset::calcSaoStatsCu(int addr, int partIdx, int yCbCr)
-{
-    int x, y;
-    TComDataCU *pTmpCu = m_pic->getCU(addr);
-    TComSPS *pTmpSPS =  m_pic->getSlice()->getSPS();
-
-    pixel* fenc;
-    pixel* pRec;
-    int stride;
-    int iLcuHeight = pTmpSPS->getMaxCUSize();
-    int iLcuWidth  = pTmpSPS->getMaxCUSize();
-    uint32_t lpelx   = pTmpCu->getCUPelX();
-    uint32_t tpely   = pTmpCu->getCUPelY();
-    uint32_t rpelx;
-    uint32_t bpely;
-    int64_t* iStats;
-    int64_t* iCount;
-    int iClassIdx;
-    int iPicWidthTmp;
-    int iPicHeightTmp;
-    int iStartX;
-    int iStartY;
-    int iEndX;
-    int iEndY;
-    pixel* pTableBo = (yCbCr == 0) ? m_lumaTableBo : m_chromaTableBo;
-    int32_t *tmp_swap;
-
-    int iIsChroma = (yCbCr != 0) ? 1 : 0;
-    int numSkipLine = iIsChroma ? 4 - (2 * m_vChromaShift) : 4;
-
-    if (m_saoLcuBasedOptimization == 0)
-    {
-        numSkipLine = 0;
-    }
-    int numSkipLineRight = iIsChroma ? 5 - (2 * m_hChromaShift) : 5;
-
-    if (m_saoLcuBasedOptimization == 0)
-    {
-        numSkipLineRight = 0;
-    }
-    iPicWidthTmp  = (iIsChroma == 0) ? m_picWidth  : m_picWidth  >> m_hChromaShift;
-    iPicHeightTmp = (iIsChroma == 0) ? m_picHeight : m_picHeight >> m_vChromaShift;
-    iLcuWidth     = (iIsChroma == 0) ? iLcuWidth   : iLcuWidth   >> m_hChromaShift;
-    iLcuHeight    = (iIsChroma == 0) ? iLcuHeight  : iLcuHeight  >> m_vChromaShift;
-    lpelx         = (iIsChroma == 0) ? lpelx       : lpelx       >> m_hChromaShift;
-    tpely         = (iIsChroma == 0) ? tpely       : tpely       >> m_vChromaShift;
-
-    rpelx       = lpelx + iLcuWidth;
-    bpely       = tpely + iLcuHeight;
-    rpelx       = rpelx > iPicWidthTmp  ? iPicWidthTmp  : rpelx;
-    bpely       = bpely > iPicHeightTmp ? iPicHeightTmp : bpely;
-    iLcuWidth     = rpelx - lpelx;
-    iLcuHeight    = bpely - tpely;
-
-    stride    =  (yCbCr == 0) ? m_pic->getStride() : m_pic->getCStride();
-
-//if(iSaoType == BO_0 || iSaoType == BO_1)
-    {
-        if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-        {
-            numSkipLine      = iIsChroma ? 3 - (2 * m_vChromaShift) : 3;
-            numSkipLineRight = iIsChroma ? 4 - (2 * m_hChromaShift) : 4;
-        }
-        iStats = m_offsetOrg[partIdx][SAO_BO];
-        iCount = m_count[partIdx][SAO_BO];
-
-        fenc = getPicYuvAddr(m_pic->getPicYuvOrg(), yCbCr, addr);
-        pRec = getPicYuvAddr(m_pic->getPicYuvRec(), yCbCr, addr);
-
-        iEndX   = (rpelx == iPicWidthTmp) ? iLcuWidth : iLcuWidth - numSkipLineRight;
-        iEndY   = (bpely == iPicHeightTmp) ? iLcuHeight : iLcuHeight - numSkipLine;
-        for (y = 0; y < iEndY; y++)
-        {
-            for (x = 0; x < iEndX; x++)
-            {
-                iClassIdx = pTableBo[pRec[x]];
-                if (iClassIdx)
-                {
-                    iStats[iClassIdx] += (fenc[x] - pRec[x]);
-                    iCount[iClassIdx]++;
-                }
-            }
-
-            fenc += stride;
-            pRec += stride;
-        }
-    }
-    int iSignLeft;
-    int iSignRight;
-    int iSignDown;
-    int iSignDown1;
-    int iSignDown2;
-
-    uint32_t uiEdgeType;
-
-//if (iSaoType == EO_0  || iSaoType == EO_1 || iSaoType == EO_2 || iSaoType == EO_3)
-    {
-        //if (iSaoType == EO_0)
-        {
-            if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-            {
-                numSkipLine      = iIsChroma ? 3 - (2 * m_vChromaShift) : 3;
-                numSkipLineRight = iIsChroma ? 5 - (2 * m_hChromaShift) : 5;
-            }
-            iStats = m_offsetOrg[partIdx][SAO_EO_0];
-            iCount = m_count[partIdx][SAO_EO_0];
-
-            fenc = getPicYuvAddr(m_pic->getPicYuvOrg(), yCbCr, addr);
-            pRec = getPicYuvAddr(m_pic->getPicYuvRec(), yCbCr, addr);
-
-            iStartX = (lpelx == 0) ? 1 : 0;
-            iEndX   = (rpelx == iPicWidthTmp) ? iLcuWidth - 1 : iLcuWidth - numSkipLineRight;
-            for (y = 0; y < iLcuHeight - numSkipLine; y++)
-            {
-                iSignLeft = xSign(pRec[iStartX] - pRec[iStartX - 1]);
-                for (x = iStartX; x < iEndX; x++)
-                {
-                    iSignRight =  xSign(pRec[x] - pRec[x + 1]);
-                    uiEdgeType =  iSignRight + iSignLeft + 2;
-                    iSignLeft  = -iSignRight;
-
-                    iStats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                    iCount[m_eoTable[uiEdgeType]]++;
-                }
-
-                fenc += stride;
-                pRec += stride;
-            }
-        }
-
-        //if (iSaoType == EO_1)
-        {
-            if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-            {
-                numSkipLine      = iIsChroma ? 4 - (2 * m_vChromaShift) : 4;
-                numSkipLineRight = iIsChroma ? 4 - (2 * m_hChromaShift) : 4;
-            }
-            iStats = m_offsetOrg[partIdx][SAO_EO_1];
-            iCount = m_count[partIdx][SAO_EO_1];
-
-            fenc = getPicYuvAddr(m_pic->getPicYuvOrg(), yCbCr, addr);
-            pRec = getPicYuvAddr(m_pic->getPicYuvRec(), yCbCr, addr);
-
-            iStartY = (tpely == 0) ? 1 : 0;
-            iEndX   = (rpelx == iPicWidthTmp) ? iLcuWidth : iLcuWidth - numSkipLineRight;
-            iEndY   = (bpely == iPicHeightTmp) ? iLcuHeight - 1 : iLcuHeight - numSkipLine;
-            if (tpely == 0)
-            {
-                fenc += stride;
-                pRec += stride;
-            }
-
-            for (x = 0; x < iLcuWidth; x++)
-            {
-                m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride]);
-            }
-
-            for (y = iStartY; y < iEndY; y++)
-            {
-                for (x = 0; x < iEndX; x++)
-                {
-                    iSignDown     =  xSign(pRec[x] - pRec[x + stride]);
-                    uiEdgeType    =  iSignDown + m_upBuff1[x] + 2;
-                    m_upBuff1[x] = -iSignDown;
-
-                    iStats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                    iCount[m_eoTable[uiEdgeType]]++;
-                }
-
-                fenc += stride;
-                pRec += stride;
-            }
-        }
-        //if (iSaoType == EO_2)
-        {
-            if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-            {
-                numSkipLine      = iIsChroma ? 4 - (2 * m_vChromaShift) : 4;
-                numSkipLineRight = iIsChroma ? 5 - (2 * m_hChromaShift) : 5;
-            }
-            iStats = m_offsetOrg[partIdx][SAO_EO_2];
-            iCount = m_count[partIdx][SAO_EO_2];
-
-            fenc = getPicYuvAddr(m_pic->getPicYuvOrg(), yCbCr, addr);
-            pRec = getPicYuvAddr(m_pic->getPicYuvRec(), yCbCr, addr);
-
-            iStartX = (lpelx == 0) ? 1 : 0;
-            iEndX   = (rpelx == iPicWidthTmp) ? iLcuWidth - 1 : iLcuWidth - numSkipLineRight;
-
-            iStartY = (tpely == 0) ? 1 : 0;
-            iEndY   = (bpely == iPicHeightTmp) ? iLcuHeight - 1 : iLcuHeight - numSkipLine;
-            if (tpely == 0)
-            {
-                fenc += stride;
-                pRec += stride;
-            }
-
-            for (x = iStartX; x < iEndX; x++)
-            {
-                m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride - 1]);
-            }
-
-            for (y = iStartY; y < iEndY; y++)
-            {
-                iSignDown2 = xSign(pRec[stride + iStartX] - pRec[iStartX - 1]);
-                for (x = iStartX; x < iEndX; x++)
-                {
-                    iSignDown1      =  xSign(pRec[x] - pRec[x + stride + 1]);
-                    uiEdgeType      =  iSignDown1 + m_upBuff1[x] + 2;
-                    m_upBufft[x + 1] = -iSignDown1;
-                    iStats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                    iCount[m_eoTable[uiEdgeType]]++;
-                }
-
-                m_upBufft[iStartX] = iSignDown2;
-                tmp_swap  = m_upBuff1;
-                m_upBuff1 = m_upBufft;
-                m_upBufft = tmp_swap;
-
-                pRec += stride;
-                fenc += stride;
-            }
-        }
-        //if (iSaoType == EO_3  )
-        {
-            if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-            {
-                numSkipLine      = iIsChroma ? 4 - (2 * m_vChromaShift) : 4;
-                numSkipLineRight = iIsChroma ? 5 - (2 * m_hChromaShift) : 5;
-            }
-            iStats = m_offsetOrg[partIdx][SAO_EO_3];
-            iCount = m_count[partIdx][SAO_EO_3];
-
-            fenc = getPicYuvAddr(m_pic->getPicYuvOrg(), yCbCr, addr);
-            pRec = getPicYuvAddr(m_pic->getPicYuvRec(), yCbCr, addr);
-
-            iStartX = (lpelx == 0) ? 1 : 0;
-            iEndX   = (rpelx == iPicWidthTmp) ? iLcuWidth - 1 : iLcuWidth - numSkipLineRight;
-
-            iStartY = (tpely == 0) ? 1 : 0;
-            iEndY   = (bpely == iPicHeightTmp) ? iLcuHeight - 1 : iLcuHeight - numSkipLine;
-            if (iStartY == 1)
-            {
-                fenc += stride;
-                pRec += stride;
-            }
-
-            for (x = iStartX - 1; x < iEndX; x++)
-            {
-                m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride + 1]);
-            }
-
-            for (y = iStartY; y < iEndY; y++)
-            {
-                for (x = iStartX; x < iEndX; x++)
-                {
-                    iSignDown1      =  xSign(pRec[x] - pRec[x + stride - 1]);
-                    uiEdgeType      =  iSignDown1 + m_upBuff1[x] + 2;
-                    m_upBuff1[x - 1] = -iSignDown1;
-                    iStats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                    iCount[m_eoTable[uiEdgeType]]++;
-                }
-
-                m_upBuff1[iEndX - 1] = xSign(pRec[iEndX - 1 + stride] - pRec[iEndX]);
-
-                pRec += stride;
-                fenc += stride;
-            }
-        }
-    }
-}
-
-void TEncSampleAdaptiveOffset::calcSaoStatsRowCus_BeforeDblk(TComPic* pic, int idxY)
-{
-    int addr, yCbCr;
-    int x, y;
-    TComSPS *pTmpSPS =  pic->getSlice()->getSPS();
-
-    pixel* fenc;
-    pixel* pRec;
-    int stride;
-    int lcuHeight;
-    int lcuWidth;
-    uint32_t rPelX;
-    uint32_t bPelY;
-    int64_t* stats;
-    int64_t* count;
-    int classIdx;
-    int picWidthTmp = 0;
-    int picHeightTmp = 0;
-    int startX;
-    int startY;
-    int endX;
-    int endY;
-    int firstX, firstY;
-
-    int idxX;
-    int frameWidthInCU  = m_numCuInWidth;
-
-    int isChroma;
-    int numSkipLine, numSkipLineRight;
-
-    uint32_t lPelX, tPelY;
-    TComDataCU *pTmpCu;
-    pixel* pTableBo;
-    int32_t *tmp_swap;
-
-    {
-        for (idxX = 0; idxX < frameWidthInCU; idxX++)
-        {
-            lcuHeight = pTmpSPS->getMaxCUSize();
-            lcuWidth  = pTmpSPS->getMaxCUSize();
-            addr     = idxX  + frameWidthInCU * idxY;
-            pTmpCu = pic->getCU(addr);
-            lPelX   = pTmpCu->getCUPelX();
-            tPelY   = pTmpCu->getCUPelY();
-
-            memset(m_countPreDblk[addr], 0, 3 * MAX_NUM_SAO_TYPE * MAX_NUM_SAO_CLASS * sizeof(int64_t));
-            memset(m_offsetOrgPreDblk[addr], 0, 3 * MAX_NUM_SAO_TYPE * MAX_NUM_SAO_CLASS * sizeof(int64_t));
-            for (yCbCr = 0; yCbCr < 3; yCbCr++)
-            {
-                isChroma = (yCbCr != 0) ? 1 : 0;
-
-                if (yCbCr == 0)
-                {
-                    picWidthTmp  = m_picWidth;
-                    picHeightTmp = m_picHeight;
-                }
-                else if (yCbCr == 1)
-                {
-                    picWidthTmp  = m_picWidth  >> isChroma;
-                    picHeightTmp = m_picHeight >> isChroma;
-                    lcuWidth     = lcuWidth    >> isChroma;
-                    lcuHeight    = lcuHeight   >> isChroma;
-                    lPelX       = lPelX      >> isChroma;
-                    tPelY       = tPelY      >> isChroma;
-                }
-                rPelX       = lPelX + lcuWidth;
-                bPelY       = tPelY + lcuHeight;
-                rPelX       = rPelX > picWidthTmp  ? picWidthTmp  : rPelX;
-                bPelY       = bPelY > picHeightTmp ? picHeightTmp : bPelY;
-                lcuWidth     = rPelX - lPelX;
-                lcuHeight    = bPelY - tPelY;
-
-                stride    =  (yCbCr == 0) ? pic->getStride() : pic->getCStride();
-                pTableBo = (yCbCr == 0) ? m_lumaTableBo : m_chromaTableBo;
-
-                //if(iSaoType == BO)
-
-                numSkipLine = isChroma ? 1 : 3;
-                numSkipLineRight = isChroma ? 2 : 4;
-
-                stats = m_offsetOrgPreDblk[addr][yCbCr][SAO_BO];
-                count = m_countPreDblk[addr][yCbCr][SAO_BO];
-
-                fenc = getPicYuvAddr(pic->getPicYuvOrg(), yCbCr, addr);
-                pRec = getPicYuvAddr(pic->getPicYuvRec(), yCbCr, addr);
-
-                startX   = (rPelX == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
-                startY   = (bPelY == picHeightTmp) ? lcuHeight : lcuHeight - numSkipLine;
-
-                for (y = 0; y < lcuHeight; y++)
-                {
-                    for (x = 0; x < lcuWidth; x++)
-                    {
-                        if (x < startX && y < startY)
-                            continue;
-
-                        classIdx = pTableBo[pRec[x]];
-                        if (classIdx)
-                        {
-                            stats[classIdx] += (fenc[x] - pRec[x]);
-                            count[classIdx]++;
-                        }
-                    }
-
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                int signLeft;
-                int signRight;
-                int signDown;
-                int signDown1;
-                int signDown2;
-
-                uint32_t uiEdgeType;
-
-                //if (iSaoType == EO_0)
-
-                numSkipLine = isChroma ? 1 : 3;
-                numSkipLineRight = isChroma ? 3 : 5;
-
-                stats = m_offsetOrgPreDblk[addr][yCbCr][SAO_EO_0];
-                count = m_countPreDblk[addr][yCbCr][SAO_EO_0];
-
-                fenc = getPicYuvAddr(pic->getPicYuvOrg(), yCbCr, addr);
-                pRec = getPicYuvAddr(pic->getPicYuvRec(), yCbCr, addr);
-
-                startX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
-                startY   = (bPelY == picHeightTmp) ? lcuHeight : lcuHeight - numSkipLine;
-                firstX   = (lPelX == 0) ? 1 : 0;
-                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
-
-                for (y = 0; y < lcuHeight; y++)
-                {
-                    signLeft = xSign(pRec[firstX] - pRec[firstX - 1]);
-                    for (x = firstX; x < endX; x++)
-                    {
-                        signRight =  xSign(pRec[x] - pRec[x + 1]);
-                        uiEdgeType =  signRight + signLeft + 2;
-                        signLeft  = -signRight;
-
-                        if (x < startX && y < startY)
-                            continue;
-
-                        stats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                        count[m_eoTable[uiEdgeType]]++;
-                    }
-
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                //if (iSaoType == EO_1)
-
-                numSkipLine = isChroma ? 2 : 4;
-                numSkipLineRight = isChroma ? 2 : 4;
-
-                stats = m_offsetOrgPreDblk[addr][yCbCr][SAO_EO_1];
-                count = m_countPreDblk[addr][yCbCr][SAO_EO_1];
-
-                fenc = getPicYuvAddr(pic->getPicYuvOrg(), yCbCr, addr);
-                pRec = getPicYuvAddr(pic->getPicYuvRec(), yCbCr, addr);
-
-                startX   = (rPelX == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
-                startY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
-                firstY = (tPelY == 0) ? 1 : 0;
-                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-                if (firstY == 1)
-                {
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                for (x = 0; x < lcuWidth; x++)
-                {
-                    m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride]);
-                }
-
-                for (y = firstY; y < endY; y++)
-                {
-                    for (x = 0; x < lcuWidth; x++)
-                    {
-                        signDown     =  xSign(pRec[x] - pRec[x + stride]);
-                        uiEdgeType    =  signDown + m_upBuff1[x] + 2;
-                        m_upBuff1[x] = -signDown;
-
-                        if (x < startX && y < startY)
-                            continue;
-
-                        stats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                        count[m_eoTable[uiEdgeType]]++;
-                    }
-
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                //if (iSaoType == EO_2)
-
-                numSkipLine = isChroma ? 2 : 4;
-                numSkipLineRight = isChroma ? 3 : 5;
-
-                stats = m_offsetOrgPreDblk[addr][yCbCr][SAO_EO_2];
-                count = m_countPreDblk[addr][yCbCr][SAO_EO_2];
-
-                fenc = getPicYuvAddr(pic->getPicYuvOrg(), yCbCr, addr);
-                pRec = getPicYuvAddr(pic->getPicYuvRec(), yCbCr, addr);
-
-                startX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
-                startY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
-                firstX   = (lPelX == 0) ? 1 : 0;
-                firstY = (tPelY == 0) ? 1 : 0;
-                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
-                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-                if (firstY == 1)
-                {
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                for (x = firstX; x < endX; x++)
-                {
-                    m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride - 1]);
-                }
-
-                for (y = firstY; y < endY; y++)
-                {
-                    signDown2 = xSign(pRec[stride + startX] - pRec[startX - 1]);
-                    for (x = firstX; x < endX; x++)
-                    {
-                        signDown1      =  xSign(pRec[x] - pRec[x + stride + 1]);
-                        uiEdgeType      =  signDown1 + m_upBuff1[x] + 2;
-                        m_upBufft[x + 1] = -signDown1;
-
-                        if (x < startX && y < startY)
-                            continue;
-
-                        stats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                        count[m_eoTable[uiEdgeType]]++;
-                    }
-
-                    m_upBufft[firstX] = signDown2;
-                    tmp_swap  = m_upBuff1;
-                    m_upBuff1 = m_upBufft;
-                    m_upBufft = tmp_swap;
-
-                    pRec += stride;
-                    fenc += stride;
-                }
-
-                //if (iSaoType == EO_3)
-
-                numSkipLine = isChroma ? 2 : 4;
-                numSkipLineRight = isChroma ? 3 : 5;
-
-                stats = m_offsetOrgPreDblk[addr][yCbCr][SAO_EO_3];
-                count = m_countPreDblk[addr][yCbCr][SAO_EO_3];
-
-                fenc = getPicYuvAddr(pic->getPicYuvOrg(), yCbCr, addr);
-                pRec = getPicYuvAddr(pic->getPicYuvRec(), yCbCr, addr);
-
-                startX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
-                startY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
-                firstX   = (lPelX == 0) ? 1 : 0;
-                firstY = (tPelY == 0) ? 1 : 0;
-                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
-                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
-                if (firstY == 1)
-                {
-                    fenc += stride;
-                    pRec += stride;
-                }
-
-                for (x = firstX - 1; x < endX; x++)
-                {
-                    m_upBuff1[x] = xSign(pRec[x] - pRec[x - stride + 1]);
-                }
-
-                for (y = firstY; y < endY; y++)
-                {
-                    for (x = firstX; x < endX; x++)
-                    {
-                        signDown1      =  xSign(pRec[x] - pRec[x + stride - 1]);
-                        uiEdgeType      =  signDown1 + m_upBuff1[x] + 2;
-                        m_upBuff1[x - 1] = -signDown1;
-
-                        if (x < startX && y < startY)
-                            continue;
-
-                        stats[m_eoTable[uiEdgeType]] += (fenc[x] - pRec[x]);
-                        count[m_eoTable[uiEdgeType]]++;
-                    }
-
-                    m_upBuff1[endX - 1] = xSign(pRec[endX - 1 + stride] - pRec[endX]);
-
-                    pRec += stride;
-                    fenc += stride;
-                }
-            }
-        }
-    }
-}
-
-/** get SAO statistics
- * \param  *psQTPart,  yCbCr
- */
-void TEncSampleAdaptiveOffset::getSaoStats(SAOQTPart *psQTPart, int yCbCr)
-{
-    int levelIdx, partIdx, typeIdx, classIdx;
-    int i;
-    int numTotalType = MAX_NUM_SAO_TYPE;
-    int LcuIdxX;
-    int LcuIdxY;
-    int addr;
-    int frameWidthInCU = m_pic->getFrameWidthInCU();
-    int downPartIdx;
-    int partStart;
-    int partEnd;
-    SAOQTPart* onePart;
-
-    if (m_maxSplitLevel == 0)
-    {
-        partIdx = 0;
-        onePart = &(psQTPart[partIdx]);
-        for (LcuIdxY = onePart->startCUY; LcuIdxY <= onePart->endCUY; LcuIdxY++)
-        {
-            for (LcuIdxX = onePart->startCUX; LcuIdxX <= onePart->endCUX; LcuIdxX++)
-            {
-                addr = LcuIdxY * frameWidthInCU + LcuIdxX;
-                calcSaoStatsCu(addr, partIdx, yCbCr);
-            }
-        }
-    }
-    else
-    {
-        for (partIdx = m_numCulPartsLevel[m_maxSplitLevel - 1]; partIdx < m_numCulPartsLevel[m_maxSplitLevel]; partIdx++)
-        {
-            onePart = &(psQTPart[partIdx]);
-            for (LcuIdxY = onePart->startCUY; LcuIdxY <= onePart->endCUY; LcuIdxY++)
-            {
-                for (LcuIdxX = onePart->startCUX; LcuIdxX <= onePart->endCUX; LcuIdxX++)
-                {
-                    addr = LcuIdxY * frameWidthInCU + LcuIdxX;
-                    calcSaoStatsCu(addr, partIdx, yCbCr);
-                }
-            }
-        }
-
-        for (levelIdx = m_maxSplitLevel - 1; levelIdx >= 0; levelIdx--)
-        {
-            partStart = (levelIdx > 0) ? m_numCulPartsLevel[levelIdx - 1] : 0;
-            partEnd   = m_numCulPartsLevel[levelIdx];
-
-            for (partIdx = partStart; partIdx < partEnd; partIdx++)
-            {
-                onePart = &(psQTPart[partIdx]);
-                for (i = 0; i < NUM_DOWN_PART; i++)
-                {
-                    downPartIdx = onePart->downPartsIdx[i];
-                    for (typeIdx = 0; typeIdx < numTotalType; typeIdx++)
-                    {
-                        for (classIdx = 0; classIdx < (typeIdx < SAO_BO ? m_numClass[typeIdx] : SAO_MAX_BO_CLASSES) + 1; classIdx++)
-                        {
-                            m_offsetOrg[partIdx][typeIdx][classIdx] += m_offsetOrg[downPartIdx][typeIdx][classIdx];
-                            m_count[partIdx][typeIdx][classIdx]    += m_count[downPartIdx][typeIdx][classIdx];
-                        }
-                    }
-                }
-            }
-        }
-    }
-}
-
-/** reset offset statistics
- * \param
- */
-void TEncSampleAdaptiveOffset::resetStats()
-{
-    for (int i = 0; i < m_numTotalParts; i++)
-    {
-        m_costPartBest[i] = MAX_DOUBLE;
-        m_typePartBest[i] = -1;
-        m_distOrg[i] = 0;
-        for (int j = 0; j < MAX_NUM_SAO_TYPE; j++)
-        {
-            m_dist[i][j] = 0;
-            m_rate[i][j] = 0;
-            m_cost[i][j] = 0;
-            for (int k = 0; k < MAX_NUM_SAO_CLASS; k++)
-            {
-                m_count[i][j][k] = 0;
-                m_offset[i][j][k] = 0;
-                m_offsetOrg[i][j][k] = 0;
-            }
-        }
-    }
-}
-
-/** Sample adaptive offset process
- * \param saoParam
- * \param dLambdaLuma
- * \param lambdaChroma
- */
-void TEncSampleAdaptiveOffset::SAOProcess(SAOParam *saoParam)
-{
-    assert(m_saoLcuBasedOptimization == false);
-    double costFinal = 0;
-    saoParam->bSaoFlag[0] = 1;
-    saoParam->bSaoFlag[1] = 0;
-    costFinal = 0;
-    getSaoStats(saoParam->saoPart[0], 0);
-    runQuadTreeDecision(saoParam->saoPart[0], 0, costFinal, m_maxSplitLevel, lumaLambda, 0);
-    saoParam->bSaoFlag[0] = costFinal < 0 ? 1 : 0;
-    if (saoParam->bSaoFlag[0])
-    {
-        convertQT2SaoUnit(saoParam, 0, 0);
-        assignSaoUnitSyntax(saoParam->saoLcuParam[0], saoParam->saoPart[0], saoParam->oneUnitFlag[0]);
-        processSaoUnitAll(saoParam->saoLcuParam[0], saoParam->oneUnitFlag[0], 0);
-    }
-    if (saoParam->bSaoFlag[1])
-    {
-        processSaoUnitAll(saoParam->saoLcuParam[1], saoParam->oneUnitFlag[1], 1);
-        processSaoUnitAll(saoParam->saoLcuParam[2], saoParam->oneUnitFlag[2], 2);
-    }
-}
-
-/** Check merge SAO unit
- * \param saoUnitCurr current SAO unit
- * \param saoUnitCheck SAO unit tobe check
- * \param dir direction
- */
-void TEncSampleAdaptiveOffset::checkMerge(SaoLcuParam * saoUnitCurr, SaoLcuParam * saoUnitCheck, int dir)
-{
-    int i;
-    int countDiff = 0;
-
-    if (saoUnitCurr->partIdx != saoUnitCheck->partIdx)
-    {
-        if (saoUnitCurr->typeIdx != -1)
-        {
-            if (saoUnitCurr->typeIdx == saoUnitCheck->typeIdx)
-            {
-                for (i = 0; i < saoUnitCurr->length; i++)
-                {
-                    countDiff += (saoUnitCurr->offset[i] != saoUnitCheck->offset[i]);
-                }
-
-                countDiff += (saoUnitCurr->subTypeIdx != saoUnitCheck->subTypeIdx);
-                if (countDiff == 0)
-                {
-                    saoUnitCurr->partIdx = saoUnitCheck->partIdx;
-                    if (dir == 1)
-                    {
-                        saoUnitCurr->mergeUpFlag = 1;
-                        saoUnitCurr->mergeLeftFlag = 0;
-                    }
-                    else
-                    {
-                        saoUnitCurr->mergeUpFlag = 0;
-                        saoUnitCurr->mergeLeftFlag = 1;
-                    }
-                }
-            }
-        }
-        else
-        {
-            if (saoUnitCurr->typeIdx == saoUnitCheck->typeIdx)
-            {
-                saoUnitCurr->partIdx = saoUnitCheck->partIdx;
-                if (dir == 1)
-                {
-                    saoUnitCurr->mergeUpFlag = 1;
-                    saoUnitCurr->mergeLeftFlag = 0;
-                }
-                else
-                {
-                    saoUnitCurr->mergeUpFlag = 0;
-                    saoUnitCurr->mergeLeftFlag = 1;
-                }
-            }
-        }
-    }
-}
-
-/** Assign SAO unit syntax from picture-based algorithm
- * \param saoLcuParam SAO LCU parameters
- * \param saoPart SAO part
- * \param oneUnitFlag SAO one unit flag
- * \param yCbCr color component Index
- */
-void TEncSampleAdaptiveOffset::assignSaoUnitSyntax(SaoLcuParam* saoLcuParam,  SAOQTPart* saoPart, bool &oneUnitFlag)
-{
-    if (saoPart->bSplit == 0)
-    {
-        oneUnitFlag = 1;
-    }
-    else
-    {
-        int i, j, addr, addrUp, addrLeft,  idx, idxUp, idxLeft,  idxCount;
-
-        oneUnitFlag = 0;
-
-        idxCount = -1;
-        saoLcuParam[0].mergeUpFlag = 0;
-        saoLcuParam[0].mergeLeftFlag = 0;
-
-        for (j = 0; j < m_numCuInHeight; j++)
-        {
-            for (i = 0; i < m_numCuInWidth; i++)
-            {
-                addr     = i + j * m_numCuInWidth;
-                addrLeft = (addr % m_numCuInWidth == 0) ? -1 : addr - 1;
-                addrUp   = (addr < m_numCuInWidth)      ? -1 : addr - m_numCuInWidth;
-                idx      = saoLcuParam[addr].partIdxTmp;
-                idxLeft  = (addrLeft == -1) ? -1 : saoLcuParam[addrLeft].partIdxTmp;
-                idxUp    = (addrUp == -1)   ? -1 : saoLcuParam[addrUp].partIdxTmp;
-
-                if (idx != idxLeft && idx != idxUp)
-                {
-                    saoLcuParam[addr].mergeUpFlag   = 0;
-                    idxCount++;
-                    saoLcuParam[addr].mergeLeftFlag = 0;
-                    saoLcuParam[addr].partIdx = idxCount;
-                }
-                else if (idx == idxLeft)
-                {
-                    saoLcuParam[addr].mergeUpFlag   = 1;
-                    saoLcuParam[addr].mergeLeftFlag = 1;
-                    saoLcuParam[addr].partIdx = saoLcuParam[addrLeft].partIdx;
-                }
-                else if (idx == idxUp)
-                {
-                    saoLcuParam[addr].mergeUpFlag   = 1;
-                    saoLcuParam[addr].mergeLeftFlag = 0;
-                    saoLcuParam[addr].partIdx = saoLcuParam[addrUp].partIdx;
-                }
-                if (addrUp != -1)
-                {
-                    checkMerge(&saoLcuParam[addr], &saoLcuParam[addrUp], 1);
-                }
-                if (addrLeft != -1)
-                {
-                    checkMerge(&saoLcuParam[addr], &saoLcuParam[addrLeft], 0);
-                }
-            }
-        }
-    }
-}
-
-void TEncSampleAdaptiveOffset::rdoSaoUnitRowInit(SAOParam *saoParam)
-{
-    saoParam->bSaoFlag[0] = true;
-    saoParam->bSaoFlag[1] = true;
-    saoParam->oneUnitFlag[0] = false;
-    saoParam->oneUnitFlag[1] = false;
-    saoParam->oneUnitFlag[2] = false;
-
-    numNoSao[0] = 0; // Luma
-    numNoSao[1] = 0; // Chroma
-    if (depth > 0 && m_depthSaoRate[0][depth - 1] > SAO_ENCODING_RATE)
-    {
-        saoParam->bSaoFlag[0] = false;
-    }
-    if (depth > 0 && m_depthSaoRate[1][depth - 1] > SAO_ENCODING_RATE_CHROMA)
-    {
-        saoParam->bSaoFlag[1] = false;
-    }
-}
-
-void TEncSampleAdaptiveOffset::rdoSaoUnitRowEnd(SAOParam *saoParam, int numlcus)
-{
-    if (!saoParam->bSaoFlag[0])
-    {
-        m_depthSaoRate[0][depth] = 1.0;
-    }
-    else
-    {
-        m_depthSaoRate[0][depth] = numNoSao[0] / ((double)numlcus);
-    }
-    if (!saoParam->bSaoFlag[1])
-    {
-        m_depthSaoRate[1][depth] = 1.0;
-    }
-    else
-    {
-        m_depthSaoRate[1][depth] = numNoSao[1] / ((double)numlcus * 2);
-    }
-}
-
-void TEncSampleAdaptiveOffset::rdoSaoUnitRow(SAOParam *saoParam, int idxY)
-{
-    int idxX;
-    int frameWidthInCU  = saoParam->numCuInWidth;
-    int j, k;
-    int addr = 0;
-    int addrUp = -1;
-    int addrLeft = -1;
-    int compIdx = 0;
-    SaoLcuParam mergeSaoParam[3][2];
-    double compDistortion[3];
-
-    {
-        for (idxX = 0; idxX < frameWidthInCU; idxX++)
-        {
-            addr     = idxX  + frameWidthInCU * idxY;
-            addrUp   = addr < frameWidthInCU ? -1 : idxX   + frameWidthInCU * (idxY - 1);
-            addrLeft = idxX == 0               ? -1 : idxX - 1 + frameWidthInCU * idxY;
-            int allowMergeLeft = 1;
-            int allowMergeUp   = 1;
-            uint32_t rate;
-            double bestCost, mergeCost;
-            if (idxX == 0)
-            {
-                allowMergeLeft = 0;
-            }
-            if (idxY == 0)
-            {
-                allowMergeUp = 0;
-            }
-
-            compDistortion[0] = 0;
-            compDistortion[1] = 0;
-            compDistortion[2] = 0;
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_CURR_BEST]);
-            if (allowMergeLeft)
-            {
-                m_entropyCoder->m_entropyCoderIf->codeSaoMerge(0);
-            }
-            if (allowMergeUp)
-            {
-                m_entropyCoder->m_entropyCoderIf->codeSaoMerge(0);
-            }
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_TEMP_BEST]);
-            // reset stats Y, Cb, Cr
-            for (compIdx = 0; compIdx < 3; compIdx++)
-            {
-                for (j = 0; j < MAX_NUM_SAO_TYPE; j++)
-                {
-                    for (k = 0; k < MAX_NUM_SAO_CLASS; k++)
-                    {
-                        m_offset[compIdx][j][k] = 0;
-                        if (m_saoLcuBasedOptimization && m_saoLcuBoundary)
-                        {
-                            m_count[compIdx][j][k] = m_countPreDblk[addr][compIdx][j][k];
-                            m_offsetOrg[compIdx][j][k] = m_offsetOrgPreDblk[addr][compIdx][j][k];
-                        }
-                        else
-                        {
-                            m_count[compIdx][j][k] = 0;
-                            m_offsetOrg[compIdx][j][k] = 0;
-                        }
-                    }
-                }
-
-                saoParam->saoLcuParam[compIdx][addr].typeIdx       =  -1;
-                saoParam->saoLcuParam[compIdx][addr].mergeUpFlag   = 0;
-                saoParam->saoLcuParam[compIdx][addr].mergeLeftFlag = 0;
-                saoParam->saoLcuParam[compIdx][addr].subTypeIdx    = 0;
-                if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
-                {
-                    calcSaoStatsCu(addr, compIdx,  compIdx);
-                }
-            }
-
-            saoComponentParamDist(allowMergeLeft, allowMergeUp, saoParam, addr, addrUp, addrLeft, 0,  lumaLambda, &mergeSaoParam[0][0], &compDistortion[0]);
-            sao2ChromaParamDist(allowMergeLeft, allowMergeUp, saoParam, addr, addrUp, addrLeft, chromaLambda, &mergeSaoParam[1][0], &mergeSaoParam[2][0], &compDistortion[0]);
-            if (saoParam->bSaoFlag[0] || saoParam->bSaoFlag[1])
-            {
-                // Cost of new SAO_params
-                m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_CURR_BEST]);
-                m_rdGoOnSbacCoder->resetBits();
-                if (allowMergeLeft)
-                {
-                    m_entropyCoder->m_entropyCoderIf->codeSaoMerge(0);
-                }
-                if (allowMergeUp)
-                {
-                    m_entropyCoder->m_entropyCoderIf->codeSaoMerge(0);
-                }
-                for (compIdx = 0; compIdx < 3; compIdx++)
-                {
-                    if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
-                    {
-                        m_entropyCoder->encodeSaoOffset(&saoParam->saoLcuParam[compIdx][addr], compIdx);
-                    }
-                }
-
-                rate = m_entropyCoder->getNumberOfWrittenBits();
-                bestCost = compDistortion[0] + (double)rate;
-                m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_TEMP_BEST]);
-
-                // Cost of Merge
-                for (int mergeUp = 0; mergeUp < 2; ++mergeUp)
-                {
-                    if ((allowMergeLeft && (mergeUp == 0)) || (allowMergeUp && (mergeUp == 1)))
-                    {
-                        m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_CURR_BEST]);
-                        m_rdGoOnSbacCoder->resetBits();
-                        if (allowMergeLeft)
-                        {
-                            m_entropyCoder->m_entropyCoderIf->codeSaoMerge(1 - mergeUp);
-                        }
-                        if (allowMergeUp && (mergeUp == 1))
-                        {
-                            m_entropyCoder->m_entropyCoderIf->codeSaoMerge(1);
-                        }
-
-                        rate = m_entropyCoder->getNumberOfWrittenBits();
-                        mergeCost = compDistortion[mergeUp + 1] + (double)rate;
-                        if (mergeCost < bestCost)
-                        {
-                            bestCost = mergeCost;
-                            m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_TEMP_BEST]);
-                            for (compIdx = 0; compIdx < 3; compIdx++)
-                            {
-                                mergeSaoParam[compIdx][mergeUp].mergeLeftFlag = 1 - mergeUp;
-                                mergeSaoParam[compIdx][mergeUp].mergeUpFlag = mergeUp;
-                                if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
-                                {
-                                    copySaoUnit(&saoParam->saoLcuParam[compIdx][addr], &mergeSaoParam[compIdx][mergeUp]);
-                                }
-                            }
-                        }
-                    }
-                }
-
-                if (saoParam->saoLcuParam[0][addr].typeIdx == -1)
-                {
-                    numNoSao[0]++;
-                }
-                if (saoParam->saoLcuParam[1][addr].typeIdx == -1)
-                {
-                    numNoSao[1] += 2;
-                }
-                m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-                m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_CURR_BEST]);
-            }
-        }
-    }
-}
-
-/** rate distortion optimization of SAO unit
- * \param saoParam SAO parameters
- * \param addr address
- * \param addrUp above address
- * \param addrLeft left address
- * \param yCbCr color component index
- * \param lambda
- */
-inline int64_t TEncSampleAdaptiveOffset::estSaoTypeDist(int compIdx, int typeIdx, int shift, double lambda, int32_t *currentDistortionTableBo, double *currentRdCostTableBo)
-{
-    int64_t estDist = 0;
-    int classIdx;
-    int saoBitIncrease = (compIdx == 0) ? m_saoBitIncreaseY : m_saoBitIncreaseC;
-    int saoOffsetTh = (compIdx == 0) ? m_offsetThY : m_offsetThC;
-
-    for (classIdx = 1; classIdx < ((typeIdx < SAO_BO) ?  m_numClass[typeIdx] + 1 : SAO_MAX_BO_CLASSES + 1); classIdx++)
-    {
-        if (typeIdx == SAO_BO)
-        {
-            currentDistortionTableBo[classIdx - 1] = 0;
-            currentRdCostTableBo[classIdx - 1] = lambda;
-        }
-        if (m_count[compIdx][typeIdx][classIdx])
-        {
-            m_offset[compIdx][typeIdx][classIdx] = (int64_t)xRoundIbdi((double)(m_offsetOrg[compIdx][typeIdx][classIdx] << (X265_DEPTH - 8)) / (double)(m_count[compIdx][typeIdx][classIdx] << saoBitIncrease));
-            m_offset[compIdx][typeIdx][classIdx] = Clip3(-saoOffsetTh + 1, saoOffsetTh - 1, (int)m_offset[compIdx][typeIdx][classIdx]);
-            if (typeIdx < 4)
-            {
-                if (m_offset[compIdx][typeIdx][classIdx] < 0 && classIdx < 3)
-                {
-                    m_offset[compIdx][typeIdx][classIdx] = 0;
-                }
-                if (m_offset[compIdx][typeIdx][classIdx] > 0 && classIdx >= 3)
-                {
-                    m_offset[compIdx][typeIdx][classIdx] = 0;
-                }
-            }
-            m_offset[compIdx][typeIdx][classIdx] = estIterOffset(typeIdx, classIdx, lambda, m_offset[compIdx][typeIdx][classIdx], m_count[compIdx][typeIdx][classIdx], m_offsetOrg[compIdx][typeIdx][classIdx], shift, saoBitIncrease, currentDistortionTableBo, currentRdCostTableBo, saoOffsetTh);
-        }
-        else
-        {
-            m_offsetOrg[compIdx][typeIdx][classIdx] = 0;
-            m_offset[compIdx][typeIdx][classIdx] = 0;
-        }
-        if (typeIdx != SAO_BO)
-        {
-            estDist += estSaoDist(m_count[compIdx][typeIdx][classIdx], m_offset[compIdx][typeIdx][classIdx] << saoBitIncrease, m_offsetOrg[compIdx][typeIdx][classIdx], shift);
-        }
-    }
-
-    return estDist;
-}
-
-inline int64_t TEncSampleAdaptiveOffset::estSaoDist(int64_t count, int64_t offset, int64_t offsetOrg, int shift)
-{
-    return (count * offset * offset - offsetOrg * offset * 2) >> shift;
-}
-
-inline int64_t TEncSampleAdaptiveOffset::estIterOffset(int typeIdx, int classIdx, double lambda, int64_t offsetInput, int64_t count, int64_t offsetOrg, int shift, int bitIncrease, int32_t *currentDistortionTableBo, double *currentRdCostTableBo, int offsetTh)
-{
-    //Clean up, best_q_offset.
-    int64_t iterOffset, tempOffset;
-    int64_t tempDist, tempRate;
-    double tempCost, tempMinCost;
-    int64_t offsetOutput = 0;
-
-    iterOffset = offsetInput;
-    // Assuming sending quantized value 0 results in zero offset and sending the value zero needs 1 bit. entropy coder can be used to measure the exact rate here.
-    tempMinCost = lambda;
-    while (iterOffset != 0)
-    {
-        // Calculate the bits required for signalling the offset
-        tempRate = (typeIdx == SAO_BO) ? (abs((int)iterOffset) + 2) : (abs((int)iterOffset) + 1);
-        if (abs((int)iterOffset) == offsetTh - 1)
-        {
-            tempRate--;
-        }
-        // Do the dequntization before distorion calculation
-        tempOffset  = iterOffset << bitIncrease;
-        tempDist    = estSaoDist(count, tempOffset, offsetOrg, shift);
-        tempCost    = ((double)tempDist + lambda * (double)tempRate);
-        if (tempCost < tempMinCost)
-        {
-            tempMinCost = tempCost;
-            offsetOutput = iterOffset;
-            if (typeIdx == SAO_BO)
-            {
-                currentDistortionTableBo[classIdx - 1] = (int)tempDist;
-                currentRdCostTableBo[classIdx - 1] = tempCost;
-            }
-        }
-        iterOffset = (iterOffset > 0) ? (iterOffset - 1) : (iterOffset + 1);
-    }
-
-    return offsetOutput;
-}
-
-void TEncSampleAdaptiveOffset::saoComponentParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, int yCbCr, double lambda, SaoLcuParam *compSaoParam, double *compDistortion)
-{
-    int typeIdx;
-
-    int64_t estDist;
-    int classIdx;
-    int shift = 2 * DISTORTION_PRECISION_ADJUSTMENT(X265_DEPTH - 8);
-    int64_t bestDist;
-
-    SaoLcuParam*  saoLcuParam = &(saoParam->saoLcuParam[yCbCr][addr]);
-    SaoLcuParam*  saoLcuParamNeighbor = NULL;
-
-    resetSaoUnit(saoLcuParam);
-    resetSaoUnit(&compSaoParam[0]);
-    resetSaoUnit(&compSaoParam[1]);
-
-    double dCostPartBest = MAX_DOUBLE;
-
-    double  bestRDCostTableBo = MAX_DOUBLE;
-    int     bestClassTableBo    = 0;
-    int     currentDistortionTableBo[MAX_NUM_SAO_CLASS];
-    double  currentRdCostTableBo[MAX_NUM_SAO_CLASS];
-
-    SaoLcuParam   saoLcuParamRdo;
-    double   estRate = 0;
-
-    resetSaoUnit(&saoLcuParamRdo);
-
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-    m_rdGoOnSbacCoder->resetBits();
-    m_entropyCoder->encodeSaoOffset(&saoLcuParamRdo, yCbCr);
-
-    dCostPartBest = m_entropyCoder->getNumberOfWrittenBits() * lambda;
-    copySaoUnit(saoLcuParam, &saoLcuParamRdo);
-    bestDist = 0;
-
-    for (typeIdx = 0; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
-    {
-        estDist = estSaoTypeDist(yCbCr, typeIdx, shift, lambda, currentDistortionTableBo, currentRdCostTableBo);
-
-        if (typeIdx == SAO_BO)
-        {
-            // Estimate Best Position
-            double currentRDCost = 0.0;
-
-            for (int i = 0; i < SAO_MAX_BO_CLASSES - SAO_BO_LEN + 1; i++)
-            {
-                currentRDCost = 0.0;
-                for (uint32_t uj = i; uj < i + SAO_BO_LEN; uj++)
-                {
-                    currentRDCost += currentRdCostTableBo[uj];
-                }
-
-                if (currentRDCost < bestRDCostTableBo)
-                {
-                    bestRDCostTableBo = currentRDCost;
-                    bestClassTableBo  = i;
-                }
-            }
-
-            // Re code all Offsets
-            // Code Center
-            estDist = 0;
-            for (classIdx = bestClassTableBo; classIdx < bestClassTableBo + SAO_BO_LEN; classIdx++)
-            {
-                estDist += currentDistortionTableBo[classIdx];
-            }
-        }
-        resetSaoUnit(&saoLcuParamRdo);
-        saoLcuParamRdo.length = m_numClass[typeIdx];
-        saoLcuParamRdo.typeIdx = typeIdx;
-        saoLcuParamRdo.mergeLeftFlag = 0;
-        saoLcuParamRdo.mergeUpFlag   = 0;
-        saoLcuParamRdo.subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo : 0;
-        for (classIdx = 0; classIdx < saoLcuParamRdo.length; classIdx++)
-        {
-            saoLcuParamRdo.offset[classIdx] = (int)m_offset[yCbCr][typeIdx][classIdx + saoLcuParamRdo.subTypeIdx + 1];
-        }
-
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-        m_rdGoOnSbacCoder->resetBits();
-        m_entropyCoder->encodeSaoOffset(&saoLcuParamRdo, yCbCr);
-
-        estRate = m_entropyCoder->getNumberOfWrittenBits();
-        m_cost[yCbCr][typeIdx] = (double)((double)estDist + lambda * (double)estRate);
-
-        if (m_cost[yCbCr][typeIdx] < dCostPartBest)
-        {
-            dCostPartBest = m_cost[yCbCr][typeIdx];
-            copySaoUnit(saoLcuParam, &saoLcuParamRdo);
-            bestDist = estDist;
-        }
-    }
-
-    compDistortion[0] += ((double)bestDist / lambda);
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-    m_entropyCoder->encodeSaoOffset(saoLcuParam, yCbCr);
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_TEMP_BEST]);
-
-    // merge left or merge up
-
-    for (int idxNeighbor = 0; idxNeighbor < 2; idxNeighbor++)
-    {
-        saoLcuParamNeighbor = NULL;
-        if (allowMergeLeft && addrLeft >= 0 && idxNeighbor == 0)
-        {
-            saoLcuParamNeighbor = &(saoParam->saoLcuParam[yCbCr][addrLeft]);
-        }
-        else if (allowMergeUp && addrUp >= 0 && idxNeighbor == 1)
-        {
-            saoLcuParamNeighbor = &(saoParam->saoLcuParam[yCbCr][addrUp]);
-        }
-        if (saoLcuParamNeighbor != NULL)
-        {
-            estDist = 0;
-            typeIdx = saoLcuParamNeighbor->typeIdx;
-            if (typeIdx >= 0)
-            {
-                int mergeBandPosition = (typeIdx == SAO_BO) ? saoLcuParamNeighbor->subTypeIdx : 0;
-                int   merge_iOffset;
-                for (classIdx = 0; classIdx < m_numClass[typeIdx]; classIdx++)
-                {
-                    merge_iOffset = saoLcuParamNeighbor->offset[classIdx];
-                    estDist   += estSaoDist(m_count[yCbCr][typeIdx][classIdx + mergeBandPosition + 1], merge_iOffset, m_offsetOrg[yCbCr][typeIdx][classIdx + mergeBandPosition + 1],  shift);
-                }
-            }
-            else
-            {
-                estDist = 0;
-            }
-
-            copySaoUnit(&compSaoParam[idxNeighbor], saoLcuParamNeighbor);
-            compSaoParam[idxNeighbor].mergeUpFlag   = idxNeighbor;
-            compSaoParam[idxNeighbor].mergeLeftFlag = !idxNeighbor;
-
-            compDistortion[idxNeighbor + 1] += ((double)estDist / lambda);
-        }
-    }
-}
-
-void TEncSampleAdaptiveOffset::sao2ChromaParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, double lambda, SaoLcuParam *crSaoParam, SaoLcuParam *cbSaoParam, double *distortion)
-{
-    int typeIdx;
-
-    int64_t estDist[2];
-    int classIdx;
-    int shift = 2 * DISTORTION_PRECISION_ADJUSTMENT(X265_DEPTH - 8);
-    int64_t bestDist = 0;
-
-    SaoLcuParam*  saoLcuParam[2] = { &(saoParam->saoLcuParam[1][addr]), &(saoParam->saoLcuParam[2][addr]) };
-    SaoLcuParam*  saoLcuParamNeighbor[2] = { NULL, NULL };
-    SaoLcuParam*  saoMergeParam[2][2];
-
-    saoMergeParam[0][0] = &crSaoParam[0];
-    saoMergeParam[0][1] = &crSaoParam[1];
-    saoMergeParam[1][0] = &cbSaoParam[0];
-    saoMergeParam[1][1] = &cbSaoParam[1];
-
-    resetSaoUnit(saoLcuParam[0]);
-    resetSaoUnit(saoLcuParam[1]);
-    resetSaoUnit(saoMergeParam[0][0]);
-    resetSaoUnit(saoMergeParam[0][1]);
-    resetSaoUnit(saoMergeParam[1][0]);
-    resetSaoUnit(saoMergeParam[1][1]);
-
-    double costPartBest = MAX_DOUBLE;
-
-    double  bestRDCostTableBo;
-    int     bestClassTableBo[2]    = { 0, 0 };
-    int     currentDistortionTableBo[MAX_NUM_SAO_CLASS];
-    double  currentRdCostTableBo[MAX_NUM_SAO_CLASS];
-
-    SaoLcuParam   saoLcuParamRdo[2];
-    double   estRate = 0;
-
-    resetSaoUnit(&saoLcuParamRdo[0]);
-    resetSaoUnit(&saoLcuParamRdo[1]);
-
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-    m_rdGoOnSbacCoder->resetBits();
-    m_entropyCoder->encodeSaoOffset(&saoLcuParamRdo[0], 1);
-    m_entropyCoder->encodeSaoOffset(&saoLcuParamRdo[1], 2);
-
-    costPartBest = m_entropyCoder->getNumberOfWrittenBits() * lambda;
-    copySaoUnit(saoLcuParam[0], &saoLcuParamRdo[0]);
-    copySaoUnit(saoLcuParam[1], &saoLcuParamRdo[1]);
-
-    for (typeIdx = 0; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
-    {
-        if (typeIdx == SAO_BO)
-        {
-            // Estimate Best Position
-            for (int compIdx = 0; compIdx < 2; compIdx++)
-            {
-                double currentRDCost = 0.0;
-                bestRDCostTableBo = MAX_DOUBLE;
-                estDist[compIdx] = estSaoTypeDist(compIdx + 1, typeIdx, shift, lambda, currentDistortionTableBo, currentRdCostTableBo);
-                for (int i = 0; i < SAO_MAX_BO_CLASSES - SAO_BO_LEN + 1; i++)
-                {
-                    currentRDCost = 0.0;
-                    for (uint32_t uj = i; uj < i + SAO_BO_LEN; uj++)
-                    {
-                        currentRDCost += currentRdCostTableBo[uj];
-                    }
-
-                    if (currentRDCost < bestRDCostTableBo)
-                    {
-                        bestRDCostTableBo = currentRDCost;
-                        bestClassTableBo[compIdx]  = i;
-                    }
-                }
-
-                // Re code all Offsets
-                // Code Center
-                estDist[compIdx] = 0;
-                for (classIdx = bestClassTableBo[compIdx]; classIdx < bestClassTableBo[compIdx] + SAO_BO_LEN; classIdx++)
-                {
-                    estDist[compIdx] += currentDistortionTableBo[classIdx];
-                }
-            }
-        }
-        else
-        {
-            estDist[0] = estSaoTypeDist(1, typeIdx, shift, lambda, currentDistortionTableBo, currentRdCostTableBo);
-            estDist[1] = estSaoTypeDist(2, typeIdx, shift, lambda, currentDistortionTableBo, currentRdCostTableBo);
-        }
-
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-        m_rdGoOnSbacCoder->resetBits();
-
-        for (int compIdx = 0; compIdx < 2; compIdx++)
-        {
-            resetSaoUnit(&saoLcuParamRdo[compIdx]);
-            saoLcuParamRdo[compIdx].length = m_numClass[typeIdx];
-            saoLcuParamRdo[compIdx].typeIdx = typeIdx;
-            saoLcuParamRdo[compIdx].mergeLeftFlag = 0;
-            saoLcuParamRdo[compIdx].mergeUpFlag   = 0;
-            saoLcuParamRdo[compIdx].subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo[compIdx] : 0;
-            for (classIdx = 0; classIdx < saoLcuParamRdo[compIdx].length; classIdx++)
-            {
-                saoLcuParamRdo[compIdx].offset[classIdx] = (int)m_offset[compIdx + 1][typeIdx][classIdx + saoLcuParamRdo[compIdx].subTypeIdx + 1];
-            }
-
-            m_entropyCoder->encodeSaoOffset(&saoLcuParamRdo[compIdx], compIdx + 1);
-        }
-
-        estRate = m_entropyCoder->getNumberOfWrittenBits();
-        m_cost[1][typeIdx] = (double)((double)(estDist[0] + estDist[1])  + lambda * (double)estRate);
-
-        if (m_cost[1][typeIdx] < costPartBest)
-        {
-            costPartBest = m_cost[1][typeIdx];
-            copySaoUnit(saoLcuParam[0], &saoLcuParamRdo[0]);
-            copySaoUnit(saoLcuParam[1], &saoLcuParamRdo[1]);
-            bestDist = (estDist[0] + estDist[1]);
-        }
-    }
-
-    distortion[0] += ((double)bestDist / lambda);
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[0][CI_TEMP_BEST]);
-    m_entropyCoder->encodeSaoOffset(saoLcuParam[0], 1);
-    m_entropyCoder->encodeSaoOffset(saoLcuParam[1], 2);
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[0][CI_TEMP_BEST]);
-
-    // merge left or merge up
-
-    for (int idxNeighbor = 0; idxNeighbor < 2; idxNeighbor++)
-    {
-        for (int compIdx = 0; compIdx < 2; compIdx++)
-        {
-            saoLcuParamNeighbor[compIdx] = NULL;
-            if (allowMergeLeft && addrLeft >= 0 && idxNeighbor == 0)
-            {
-                saoLcuParamNeighbor[compIdx] = &(saoParam->saoLcuParam[compIdx + 1][addrLeft]);
-            }
-            else if (allowMergeUp && addrUp >= 0 && idxNeighbor == 1)
-            {
-                saoLcuParamNeighbor[compIdx] = &(saoParam->saoLcuParam[compIdx + 1][addrUp]);
-            }
-            if (saoLcuParamNeighbor[compIdx] != NULL)
-            {
-                estDist[compIdx] = 0;
-                typeIdx = saoLcuParamNeighbor[compIdx]->typeIdx;
-                if (typeIdx >= 0)
-                {
-                    int mergeBandPosition = (typeIdx == SAO_BO) ? saoLcuParamNeighbor[compIdx]->subTypeIdx : 0;
-                    int   merge_iOffset;
-                    for (classIdx = 0; classIdx < m_numClass[typeIdx]; classIdx++)
-                    {
-                        merge_iOffset = saoLcuParamNeighbor[compIdx]->offset[classIdx];
-                        estDist[compIdx]   += estSaoDist(m_count[compIdx + 1][typeIdx][classIdx + mergeBandPosition + 1], merge_iOffset, m_offsetOrg[compIdx + 1][typeIdx][classIdx + mergeBandPosition + 1],  shift);
-                    }
-                }
-                else
-                {
-                    estDist[compIdx] = 0;
-                }
-
-                copySaoUnit(saoMergeParam[compIdx][idxNeighbor], saoLcuParamNeighbor[compIdx]);
-                saoMergeParam[compIdx][idxNeighbor]->mergeUpFlag   = idxNeighbor;
-                saoMergeParam[compIdx][idxNeighbor]->mergeLeftFlag = !idxNeighbor;
-                distortion[idxNeighbor + 1] += ((double)estDist[compIdx] / lambda);
-            }
-        }
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSampleAdaptiveOffset.h Deleted

@@ -1,125 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/**
- \file     TEncSampleAdaptiveOffset.h
- \brief    estimation part of sample adaptive offset class (header)
- */
-
-#ifndef X265_TENCSAMPLEADAPTIVEOFFSET_H
-#define X265_TENCSAMPLEADAPTIVEOFFSET_H
-
-#include "TLibCommon/TComSampleAdaptiveOffset.h"
-#include "TLibCommon/TComPic.h"
-
-#include "TEncEntropy.h"
-#include "TEncSbac.h"
-#include "TLibCommon/TComBitCounter.h"
-
-//! \ingroup TLibEncoder
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-namespace x265 {
-// private namespace
-
-class TEncSampleAdaptiveOffset : public TComSampleAdaptiveOffset
-{
-private:
-
-    TEncEntropy*      m_entropyCoder;
-    TEncSbac***       m_rdSbacCoders;            ///< for CABAC
-    TEncSbac*         m_rdGoOnSbacCoder;
-    TEncBinCABAC***   m_binCoderCABAC;          ///< temporal CABAC state storage for RD computation
-
-    int64_t  ***m_count;    //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-    int64_t  ***m_offset;   //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-    int64_t  ***m_offsetOrg; //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE];
-    int64_t(*m_countPreDblk)[3][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];    //[LCU][YCbCr][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-    int64_t(*m_offsetOrgPreDblk)[3][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS]; //[LCU][YCbCr][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS];
-    int64_t  **m_rate;      //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE];
-    int64_t  **m_dist;      //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE];
-    double **m_cost;      //[MAX_NUM_SAO_PART][MAX_NUM_SAO_TYPE];
-    double *m_costPartBest; //[MAX_NUM_SAO_PART];
-    int64_t  *m_distOrg;    //[MAX_NUM_SAO_PART];
-    int    *m_typePartBest; //[MAX_NUM_SAO_PART];
-    int     m_offsetThY;
-    int     m_offsetThC;
-    double  m_depthSaoRate[2][4];
-
-public:
-
-    double  lumaLambda;
-    double  chromaLambda;
-    int     depth;
-    int     numNoSao[2];
-
-    TEncSampleAdaptiveOffset();
-    virtual ~TEncSampleAdaptiveOffset();
-
-    void startSaoEnc(TComPic* pic, TEncEntropy* entropyCoder, TEncSbac* rdGoOnSbacCoder);
-    void endSaoEnc();
-    void resetStats();
-    void SAOProcess(SAOParam *saoParam);
-
-    void runQuadTreeDecision(SAOQTPart *psQTPart, int partIdx, double &costFinal, int maxLevel, double lambda, int yCbCr);
-    void rdoSaoOnePart(SAOQTPart *psQTPart, int partIdx, double lambda, int yCbCr);
-
-    void disablePartTree(SAOQTPart *psQTPart, int partIdx);
-    void getSaoStats(SAOQTPart *psQTPart, int yCbCr);
-    void calcSaoStatsCu(int addr, int partIdx, int yCbCr);
-    void calcSaoStatsRowCus_BeforeDblk(TComPic* pic, int idxY);
-    void destroyEncBuffer();
-    void createEncBuffer();
-    void assignSaoUnitSyntax(SaoLcuParam* saoLcuParam,  SAOQTPart* saoPart, bool &oneUnitFlag);
-    void checkMerge(SaoLcuParam* lcuParamCurr, SaoLcuParam * lcuParamCheck, int dir);
-    void saoComponentParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, int yCbCr, double lambda, SaoLcuParam *compSaoParam, double *distortion);
-    void sao2ChromaParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, double lambda, SaoLcuParam *crSaoParam, SaoLcuParam *cbSaoParam, double *distortion);
-    inline int64_t estSaoDist(int64_t count, int64_t offset, int64_t offsetOrg, int shift);
-    inline int64_t estIterOffset(int typeIdx, int classIdx, double lambda, int64_t offsetInput, int64_t count, int64_t offsetOrg, int shift, int bitIncrease, int32_t *currentDistortionTableBo, double *currentRdCostTableBo, int offsetTh);
-    inline int64_t estSaoTypeDist(int compIdx, int typeIdx, int shift, double lambda, int32_t *currentDistortionTableBo, double *currentRdCostTableBo);
-    void setMaxNumOffsetsPerPic(int val) { m_maxNumOffsetsPerPic = val; }
-
-    int  getMaxNumOffsetsPerPic() { return m_maxNumOffsetsPerPic; }
-
-    void rdoSaoUnitRowInit(SAOParam *saoParam);
-    void rdoSaoUnitRowEnd(SAOParam *saoParam, int numlcus);
-    void rdoSaoUnitRow(SAOParam *saoParam, int idxY);
-};
-}
-//! \}
-
-#endif // ifndef X265_TENCSAMPLEADAPTIVEOFFSET_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSbac.cpp Deleted

@@ -1,2525 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncSbac.cpp
-    \brief    SBAC encoder class
-*/
-
-#include "TEncSbac.h"
-#include "primitives.h"
-
-namespace x265 {
-//! \ingroup TLibEncoder
-//! \{
-
-#if ENC_DEC_TRACE
-
-void  xTraceSPSHeader(TComSPS *sps)
-{
-    fprintf(g_hTrace, "=========== Sequence Parameter Set ID: %d ===========\n", sps->getSPSId());
-}
-
-void  xTracePPSHeader(TComPPS *pps)
-{
-    fprintf(g_hTrace, "=========== Picture Parameter Set ID: %d ===========\n", pps->getPPSId());
-}
-
-void  xTraceSliceHeader(TComSlice *)
-{
-    fprintf(g_hTrace, "=========== Slice ===========\n");
-}
-
-#endif // if ENC_DEC_TRACE
-
-const uint32_t g_entropyBits[128] =
-{
-    // Corrected table, most notably for last state
-    0x07b23, 0x085f9, 0x074a0, 0x08cbc, 0x06ee4, 0x09354, 0x067f4, 0x09c1b, 0x060b0, 0x0a62a, 0x05a9c, 0x0af5b, 0x0548d, 0x0b955, 0x04f56, 0x0c2a9,
-    0x04a87, 0x0cbf7, 0x045d6, 0x0d5c3, 0x04144, 0x0e01b, 0x03d88, 0x0e937, 0x039e0, 0x0f2cd, 0x03663, 0x0fc9e, 0x03347, 0x10600, 0x03050, 0x10f95,
-    0x02d4d, 0x11a02, 0x02ad3, 0x12333, 0x0286e, 0x12cad, 0x02604, 0x136df, 0x02425, 0x13f48, 0x021f4, 0x149c4, 0x0203e, 0x1527b, 0x01e4d, 0x15d00,
-    0x01c99, 0x166de, 0x01b18, 0x17017, 0x019a5, 0x17988, 0x01841, 0x18327, 0x016df, 0x18d50, 0x015d9, 0x19547, 0x0147c, 0x1a083, 0x0138e, 0x1a8a3,
-    0x01251, 0x1b418, 0x01166, 0x1bd27, 0x01068, 0x1c77b, 0x00f7f, 0x1d18e, 0x00eda, 0x1d91a, 0x00e19, 0x1e254, 0x00d4f, 0x1ec9a, 0x00c90, 0x1f6e0,
-    0x00c01, 0x1fef8, 0x00b5f, 0x208b1, 0x00ab6, 0x21362, 0x00a15, 0x21e46, 0x00988, 0x2285d, 0x00934, 0x22ea8, 0x008a8, 0x239b2, 0x0081d, 0x24577,
-    0x007c9, 0x24ce6, 0x00763, 0x25663, 0x00710, 0x25e8f, 0x006a0, 0x26a26, 0x00672, 0x26f23, 0x005e8, 0x27ef8, 0x005ba, 0x284b5, 0x0055e, 0x29057,
-    0x0050c, 0x29bab, 0x004c1, 0x2a674, 0x004a7, 0x2aa5e, 0x0046f, 0x2b32f, 0x0041f, 0x2c0ad, 0x003e7, 0x2ca8d, 0x003ba, 0x2d323, 0x0010c, 0x3bfbb
-};
-
-const uint8_t g_nextState[128][2] =
-{
-    { 2, 1 }, { 0, 3 }, { 4, 0 }, { 1, 5 }, { 6, 2 }, { 3, 7 }, { 8, 4 }, { 5, 9 },
-    { 10, 4 }, { 5, 11 }, { 12, 8 }, { 9, 13 }, { 14, 8 }, { 9, 15 }, { 16, 10 }, { 11, 17 },
-    { 18, 12 }, { 13, 19 }, { 20, 14 }, { 15, 21 }, { 22, 16 }, { 17, 23 }, { 24, 18 }, { 19, 25 },
-    { 26, 18 }, { 19, 27 }, { 28, 22 }, { 23, 29 }, { 30, 22 }, { 23, 31 }, { 32, 24 }, { 25, 33 },
-    { 34, 26 }, { 27, 35 }, { 36, 26 }, { 27, 37 }, { 38, 30 }, { 31, 39 }, { 40, 30 }, { 31, 41 },
-    { 42, 32 }, { 33, 43 }, { 44, 32 }, { 33, 45 }, { 46, 36 }, { 37, 47 }, { 48, 36 }, { 37, 49 },
-    { 50, 38 }, { 39, 51 }, { 52, 38 }, { 39, 53 }, { 54, 42 }, { 43, 55 }, { 56, 42 }, { 43, 57 },
-    { 58, 44 }, { 45, 59 }, { 60, 44 }, { 45, 61 }, { 62, 46 }, { 47, 63 }, { 64, 48 }, { 49, 65 },
-    { 66, 48 }, { 49, 67 }, { 68, 50 }, { 51, 69 }, { 70, 52 }, { 53, 71 }, { 72, 52 }, { 53, 73 },
-    { 74, 54 }, { 55, 75 }, { 76, 54 }, { 55, 77 }, { 78, 56 }, { 57, 79 }, { 80, 58 }, { 59, 81 },
-    { 82, 58 }, { 59, 83 }, { 84, 60 }, { 61, 85 }, { 86, 60 }, { 61, 87 }, { 88, 60 }, { 61, 89 },
-    { 90, 62 }, { 63, 91 }, { 92, 64 }, { 65, 93 }, { 94, 64 }, { 65, 95 }, { 96, 66 }, { 67, 97 },
-    { 98, 66 }, { 67, 99 }, { 100, 66 }, { 67, 101 }, { 102, 68 }, { 69, 103 }, { 104, 68 }, { 69, 105 },
-    { 106, 70 }, { 71, 107 }, { 108, 70 }, { 71, 109 }, { 110, 70 }, { 71, 111 }, { 112, 72 }, { 73, 113 },
-    { 114, 72 }, { 73, 115 }, { 116, 72 }, { 73, 117 }, { 118, 74 }, { 75, 119 }, { 120, 74 }, { 75, 121 },
-    { 122, 74 }, { 75, 123 }, { 124, 76 }, { 77, 125 }, { 124, 76 }, { 77, 125 }, { 126, 126 }, { 127, 127 }
-};
-
-/**
- - initialize context model with respect to QP and initialization value
- .
- \param  qp         input QP value
- \param  initValue  8 bit initialization value
- */
-uint8_t sbacInit(int qp, int initValue)
-{
-    qp = Clip3(0, 51, qp);
-
-    int  slope      = (initValue >> 4) * 5 - 45;
-    int  offset     = ((initValue & 15) << 3) - 16;
-    int  initState  =  X265_MIN(X265_MAX(1, (((slope * qp) >> 4) + offset)), 126);
-    uint32_t mpState = (initState >= 64);
-    uint8_t m_state = ((mpState ? (initState - 64) : (63 - initState)) << 1) + mpState;
-
-    return m_state;
-}
-
-static void initBuffer(ContextModel* contextModel, SliceType sliceType, int qp, UChar* ctxModel, int size)
-{
-    ctxModel += sliceType * size;
-
-    for (int n = 0; n < size; n++)
-    {
-        contextModel[n].m_state = sbacInit(qp, ctxModel[n]);
-        contextModel[n].bBinsCoded = 0;
-    }
-}
-
-static uint32_t calcCost(ContextModel *contextModel, SliceType sliceType, int qp, UChar* ctxModel, int size)
-{
-    uint32_t cost = 0;
-
-    ctxModel += sliceType * size;
-
-    for (int n = 0; n < size; n++)
-    {
-        ContextModel tmpContextModel;
-        tmpContextModel.m_state = sbacInit(qp, ctxModel[n]);
-
-        // Map the 64 CABAC states to their corresponding probability values
-        static const double aStateToProbLPS[] = { 0.50000000, 0.47460857, 0.45050660, 0.42762859, 0.40591239, 0.38529900, 0.36573242, 0.34715948, 0.32952974, 0.31279528, 0.29691064, 0.28183267, 0.26752040, 0.25393496, 0.24103941, 0.22879875, 0.21717969, 0.20615069, 0.19568177, 0.18574449, 0.17631186, 0.16735824, 0.15885931, 0.15079198, 0.14313433, 0.13586556, 0.12896592, 0.12241667, 0.11620000, 0.11029903, 0.10469773, 0.09938088, 0.09433404, 0.08954349, 0.08499621, 0.08067986, 0.07658271, 0.07269362, 0.06900203, 0.06549791, 0.06217174, 0.05901448, 0.05601756, 0.05317283, 0.05047256, 0.04790942, 0.04547644, 0.04316702, 0.04097487, 0.03889405, 0.03691890, 0.03504406, 0.03326442, 0.03157516, 0.02997168, 0.02844963, 0.02700488, 0.02563349, 0.02433175, 0.02309612, 0.02192323, 0.02080991, 0.01975312, 0.01875000 };
-
-        double probLPS = aStateToProbLPS[sbacGetState(contextModel[n].m_state)];
-        double prob0, prob1;
-        if (sbacGetMps(contextModel[n].m_state) == 1)
-        {
-            prob0 = probLPS;
-            prob1 = 1.0 - prob0;
-        }
-        else
-        {
-            prob1 = probLPS;
-            prob0 = 1.0 - prob1;
-        }
-
-        if (contextModel[n].bBinsCoded > 0)
-        {
-            cost += (uint32_t)(prob0 * sbacGetEntropyBits(tmpContextModel.m_state, 0) + prob1 * sbacGetEntropyBits(tmpContextModel.m_state, 1));
-        }
-    }
-
-    return cost;
-}
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
-TEncSbac::TEncSbac()
-// new structure here
-    : m_slice(NULL)
-    , m_binIf(NULL)
-{
-    memset(m_contextModels, 0, sizeof(m_contextModels));
-}
-
-TEncSbac::~TEncSbac()
-{}
-
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
-
-void TEncSbac::resetEntropy()
-{
-    int  qp              = m_slice->getSliceQp();
-    SliceType sliceType  = m_slice->getSliceType();
-
-    int encCABACTableIdx = m_slice->getPPS()->getEncCABACTableIdx();
-
-    if (!m_slice->isIntra() && (encCABACTableIdx == B_SLICE || encCABACTableIdx == P_SLICE) && m_slice->getPPS()->getCabacInitPresentFlag())
-    {
-        sliceType = (SliceType)encCABACTableIdx;
-    }
-
-    initBuffer(&m_contextModels[OFF_SPLIT_FLAG_CTX], sliceType, qp, (UChar*)INIT_SPLIT_FLAG, NUM_SPLIT_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_SKIP_FLAG_CTX], sliceType, qp, (UChar*)INIT_SKIP_FLAG, NUM_SKIP_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_MERGE_FLAG_EXT_CTX], sliceType, qp, (UChar*)INIT_MERGE_FLAG_EXT, NUM_MERGE_FLAG_EXT_CTX);
-    initBuffer(&m_contextModels[OFF_MERGE_IDX_EXT_CTX], sliceType, qp, (UChar*)INIT_MERGE_IDX_EXT, NUM_MERGE_IDX_EXT_CTX);
-    initBuffer(&m_contextModels[OFF_PART_SIZE_CTX], sliceType, qp, (UChar*)INIT_PART_SIZE, NUM_PART_SIZE_CTX);
-    initBuffer(&m_contextModels[OFF_PRED_MODE_CTX], sliceType, qp, (UChar*)INIT_PRED_MODE, NUM_PRED_MODE_CTX);
-    initBuffer(&m_contextModels[OFF_ADI_CTX], sliceType, qp, (UChar*)INIT_INTRA_PRED_MODE, NUM_ADI_CTX);
-    initBuffer(&m_contextModels[OFF_CHROMA_PRED_CTX], sliceType, qp, (UChar*)INIT_CHROMA_PRED_MODE, NUM_CHROMA_PRED_CTX);
-    initBuffer(&m_contextModels[OFF_DELTA_QP_CTX], sliceType, qp, (UChar*)INIT_DQP, NUM_DELTA_QP_CTX);
-    initBuffer(&m_contextModels[OFF_INTER_DIR_CTX], sliceType, qp, (UChar*)INIT_INTER_DIR, NUM_INTER_DIR_CTX);
-    initBuffer(&m_contextModels[OFF_REF_NO_CTX], sliceType, qp, (UChar*)INIT_REF_PIC, NUM_REF_NO_CTX);
-    initBuffer(&m_contextModels[OFF_MV_RES_CTX], sliceType, qp, (UChar*)INIT_MVD, NUM_MV_RES_CTX);
-    initBuffer(&m_contextModels[OFF_QT_CBF_CTX], sliceType, qp, (UChar*)INIT_QT_CBF, NUM_QT_CBF_CTX);
-    initBuffer(&m_contextModels[OFF_TRANS_SUBDIV_FLAG_CTX], sliceType, qp, (UChar*)INIT_TRANS_SUBDIV_FLAG, NUM_TRANS_SUBDIV_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_QT_ROOT_CBF_CTX], sliceType, qp, (UChar*)INIT_QT_ROOT_CBF, NUM_QT_ROOT_CBF_CTX);
-    initBuffer(&m_contextModels[OFF_SIG_CG_FLAG_CTX], sliceType, qp, (UChar*)INIT_SIG_CG_FLAG, 2 * NUM_SIG_CG_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_SIG_FLAG_CTX], sliceType, qp, (UChar*)INIT_SIG_FLAG, NUM_SIG_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_CTX_LAST_FLAG_X], sliceType, qp, (UChar*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
-    initBuffer(&m_contextModels[OFF_CTX_LAST_FLAG_Y], sliceType, qp, (UChar*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
-    initBuffer(&m_contextModels[OFF_ONE_FLAG_CTX], sliceType, qp, (UChar*)INIT_ONE_FLAG, NUM_ONE_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_ABS_FLAG_CTX], sliceType, qp, (UChar*)INIT_ABS_FLAG, NUM_ABS_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_MVP_IDX_CTX], sliceType, qp, (UChar*)INIT_MVP_IDX, NUM_MVP_IDX_CTX);
-    initBuffer(&m_contextModels[OFF_SAO_MERGE_FLAG_CTX], sliceType, qp, (UChar*)INIT_SAO_MERGE_FLAG, NUM_SAO_MERGE_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_SAO_TYPE_IDX_CTX], sliceType, qp, (UChar*)INIT_SAO_TYPE_IDX, NUM_SAO_TYPE_IDX_CTX);
-    initBuffer(&m_contextModels[OFF_TRANSFORMSKIP_FLAG_CTX], sliceType, qp, (UChar*)INIT_TRANSFORMSKIP_FLAG, 2 * NUM_TRANSFORMSKIP_FLAG_CTX);
-    initBuffer(&m_contextModels[OFF_CU_TRANSQUANT_BYPASS_FLAG_CTX], sliceType, qp, (UChar*)INIT_CU_TRANSQUANT_BYPASS_FLAG, NUM_CU_TRANSQUANT_BYPASS_FLAG_CTX);
-    // new structure
-
-    m_binIf->start();
-}
-
-/** The function does the following:
- * If current slice type is P/B then it determines the distance of initialisation type 1 and 2 from the current CABAC states and
- * stores the index of the closest table.  This index is used for the next P/B slice when cabac_init_present_flag is true.
- */
-void TEncSbac::determineCabacInitIdx()
-{
-    int qp = m_slice->getSliceQp();
-
-    if (!m_slice->isIntra())
-    {
-        SliceType aSliceTypeChoices[] = { B_SLICE, P_SLICE };
-
-        uint32_t bestCost       = MAX_UINT;
-        SliceType bestSliceType = aSliceTypeChoices[0];
-        for (uint32_t idx = 0; idx < 2; idx++)
-        {
-            uint32_t curCost = 0;
-            SliceType curSliceType = aSliceTypeChoices[idx];
-
-            curCost  = calcCost(&m_contextModels[OFF_SPLIT_FLAG_CTX], curSliceType, qp, (UChar*)INIT_SPLIT_FLAG, NUM_SPLIT_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_SKIP_FLAG_CTX], curSliceType, qp, (UChar*)INIT_SKIP_FLAG, NUM_SKIP_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_MERGE_FLAG_EXT_CTX], curSliceType, qp, (UChar*)INIT_MERGE_FLAG_EXT, NUM_MERGE_FLAG_EXT_CTX);
-            curCost += calcCost(&m_contextModels[OFF_MERGE_IDX_EXT_CTX], curSliceType, qp, (UChar*)INIT_MERGE_IDX_EXT, NUM_MERGE_IDX_EXT_CTX);
-            curCost += calcCost(&m_contextModels[OFF_PART_SIZE_CTX], curSliceType, qp, (UChar*)INIT_PART_SIZE, NUM_PART_SIZE_CTX);
-            curCost += calcCost(&m_contextModels[OFF_PRED_MODE_CTX], curSliceType, qp, (UChar*)INIT_PRED_MODE, NUM_PRED_MODE_CTX);
-            curCost += calcCost(&m_contextModels[OFF_ADI_CTX], curSliceType, qp, (UChar*)INIT_INTRA_PRED_MODE, NUM_ADI_CTX);
-            curCost += calcCost(&m_contextModels[OFF_CHROMA_PRED_CTX], curSliceType, qp, (UChar*)INIT_CHROMA_PRED_MODE, NUM_CHROMA_PRED_CTX);
-            curCost += calcCost(&m_contextModels[OFF_DELTA_QP_CTX], curSliceType, qp, (UChar*)INIT_DQP, NUM_DELTA_QP_CTX);
-            curCost += calcCost(&m_contextModels[OFF_INTER_DIR_CTX], curSliceType, qp, (UChar*)INIT_INTER_DIR, NUM_INTER_DIR_CTX);
-            curCost += calcCost(&m_contextModels[OFF_REF_NO_CTX], curSliceType, qp, (UChar*)INIT_REF_PIC, NUM_REF_NO_CTX);
-            curCost += calcCost(&m_contextModels[OFF_MV_RES_CTX], curSliceType, qp, (UChar*)INIT_MVD, NUM_MV_RES_CTX);
-            curCost += calcCost(&m_contextModels[OFF_QT_CBF_CTX], curSliceType, qp, (UChar*)INIT_QT_CBF, NUM_QT_CBF_CTX);
-            curCost += calcCost(&m_contextModels[OFF_TRANS_SUBDIV_FLAG_CTX], curSliceType, qp, (UChar*)INIT_TRANS_SUBDIV_FLAG, NUM_TRANS_SUBDIV_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_QT_ROOT_CBF_CTX], curSliceType, qp, (UChar*)INIT_QT_ROOT_CBF, NUM_QT_ROOT_CBF_CTX);
-            curCost += calcCost(&m_contextModels[OFF_SIG_CG_FLAG_CTX], curSliceType, qp, (UChar*)INIT_SIG_CG_FLAG, 2 * NUM_SIG_CG_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_SIG_FLAG_CTX], curSliceType, qp, (UChar*)INIT_SIG_FLAG, NUM_SIG_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_CTX_LAST_FLAG_X], curSliceType, qp, (UChar*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
-            curCost += calcCost(&m_contextModels[OFF_CTX_LAST_FLAG_Y], curSliceType, qp, (UChar*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
-            curCost += calcCost(&m_contextModels[OFF_ONE_FLAG_CTX], curSliceType, qp, (UChar*)INIT_ONE_FLAG, NUM_ONE_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_ABS_FLAG_CTX], curSliceType, qp, (UChar*)INIT_ABS_FLAG, NUM_ABS_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_MVP_IDX_CTX], curSliceType, qp, (UChar*)INIT_MVP_IDX, NUM_MVP_IDX_CTX);
-            curCost += calcCost(&m_contextModels[OFF_SAO_MERGE_FLAG_CTX], curSliceType, qp, (UChar*)INIT_SAO_MERGE_FLAG, NUM_SAO_MERGE_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_SAO_TYPE_IDX_CTX], curSliceType, qp, (UChar*)INIT_SAO_TYPE_IDX, NUM_SAO_TYPE_IDX_CTX);
-            curCost += calcCost(&m_contextModels[OFF_TRANSFORMSKIP_FLAG_CTX], curSliceType, qp, (UChar*)INIT_TRANSFORMSKIP_FLAG, 2 * NUM_TRANSFORMSKIP_FLAG_CTX);
-            curCost += calcCost(&m_contextModels[OFF_CU_TRANSQUANT_BYPASS_FLAG_CTX], curSliceType, qp, (UChar*)INIT_CU_TRANSQUANT_BYPASS_FLAG, NUM_CU_TRANSQUANT_BYPASS_FLAG_CTX);
-            if (curCost < bestCost)
-            {
-                bestSliceType = curSliceType;
-                bestCost      = curCost;
-            }
-        }
-
-        m_slice->getPPS()->setEncCABACTableIdx(bestSliceType);
-    }
-    else
-    {
-        m_slice->getPPS()->setEncCABACTableIdx(I_SLICE);
-    }
-}
-
-void TEncSbac::codeVPS(TComVPS* vps)
-{
-    WRITE_CODE(vps->getVPSId(),                    4,        "vps_video_parameter_set_id");
-    WRITE_CODE(3,                                  2,        "vps_reserved_three_2bits");
-    WRITE_CODE(0,                                  6,        "vps_reserved_zero_6bits");
-    WRITE_CODE(vps->getMaxTLayers() - 1,           3,        "vps_max_sub_layers_minus1");
-    WRITE_FLAG(vps->getTemporalNestingFlag(),                "vps_temporal_id_nesting_flag");
-    assert(vps->getMaxTLayers() > 1 || vps->getTemporalNestingFlag());
-    WRITE_CODE(0xffff,                            16,        "vps_reserved_ffff_16bits");
-    codePTL(vps->getPTL(), true, vps->getMaxTLayers() - 1);
-    WRITE_FLAG(true,             "vps_sub_layer_ordering_info_present_flag");
-    for (uint32_t i = 0; i <= vps->getMaxTLayers() - 1; i++)
-    {
-        WRITE_UVLC(vps->getMaxDecPicBuffering(i) - 1,       "vps_max_dec_pic_buffering_minus1[i]");
-        WRITE_UVLC(vps->getNumReorderPics(i),               "vps_num_reorder_pics[i]");
-        WRITE_UVLC(vps->getMaxLatencyIncrease(i),           "vps_max_latency_increase_plus1[i]");
-    }
-
-    assert(vps->getNumHrdParameters() <= MAX_VPS_NUM_HRD_PARAMETERS);
-    assert(vps->getMaxNuhReservedZeroLayerId() < MAX_VPS_NUH_RESERVED_ZERO_LAYER_ID_PLUS1);
-    WRITE_CODE(vps->getMaxNuhReservedZeroLayerId(), 6,     "vps_max_nuh_reserved_zero_layer_id");
-    vps->setMaxOpSets(1);
-    WRITE_UVLC(vps->getMaxOpSets() - 1,                    "vps_max_op_sets_minus1");
-    for (uint32_t opsIdx = 1; opsIdx <= (vps->getMaxOpSets() - 1); opsIdx++)
-    {
-        // Operation point set
-        for (uint32_t i = 0; i <= vps->getMaxNuhReservedZeroLayerId(); i++)
-        {
-            // Only applicable for version 1
-            vps->setLayerIdIncludedFlag(true, opsIdx, i);
-            WRITE_FLAG(vps->getLayerIdIncludedFlag(opsIdx, i) ? 1 : 0, "layer_id_included_flag[opsIdx][i]");
-        }
-    }
-
-    TimingInfo *timingInfo = vps->getTimingInfo();
-    WRITE_FLAG(timingInfo->getTimingInfoPresentFlag(),          "vps_timing_info_present_flag");
-    if (timingInfo->getTimingInfoPresentFlag())
-    {
-        WRITE_CODE(timingInfo->getNumUnitsInTick(), 32,           "vps_num_units_in_tick");
-        WRITE_CODE(timingInfo->getTimeScale(),      32,           "vps_time_scale");
-        WRITE_FLAG(timingInfo->getPocProportionalToTimingFlag(),  "vps_poc_proportional_to_timing_flag");
-        if (timingInfo->getPocProportionalToTimingFlag())
-        {
-            WRITE_UVLC(timingInfo->getNumTicksPocDiffOneMinus1(),   "vps_num_ticks_poc_diff_one_minus1");
-        }
-        vps->setNumHrdParameters(0);
-        WRITE_UVLC(vps->getNumHrdParameters(),                 "vps_num_hrd_parameters");
-
-        if (vps->getNumHrdParameters() > 0)
-        {
-            vps->createHrdParamBuffer();
-        }
-        for (uint32_t i = 0; i < vps->getNumHrdParameters(); i++)
-        {
-            // Only applicable for version 1
-            vps->setHrdOpSetIdx(0, i);
-            WRITE_UVLC(vps->getHrdOpSetIdx(i),                "hrd_op_set_idx");
-            if (i > 0)
-            {
-                WRITE_FLAG(vps->getCprmsPresentFlag(i) ? 1 : 0, "cprms_present_flag[i]");
-            }
-            codeHrdParameters(vps->getHrdParameters(i), vps->getCprmsPresentFlag(i), vps->getMaxTLayers() - 1);
-        }
-    }
-    WRITE_FLAG(0,                     "vps_extension_flag");
-
-    //future extensions here..
-}
-
-void TEncSbac::codeShortTermRefPicSet(TComReferencePictureSet* rps, bool calledFromSliceHeader, int idx)
-{
-    if (idx > 0)
-    {
-        WRITE_FLAG(rps->getInterRPSPrediction(), "inter_ref_pic_set_prediction_flag"); // inter_RPS_prediction_flag
-    }
-    if (rps->getInterRPSPrediction())
-    {
-        int deltaRPS = rps->getDeltaRPS();
-        if (calledFromSliceHeader)
-        {
-            WRITE_UVLC(rps->getDeltaRIdxMinus1(), "delta_idx_minus1"); // delta index of the Reference Picture Set used for prediction minus 1
-        }
-
-        WRITE_CODE((deltaRPS >= 0 ? 0 : 1), 1, "delta_rps_sign"); //delta_rps_sign
-        WRITE_UVLC(abs(deltaRPS) - 1, "abs_delta_rps_minus1"); // absolute delta RPS minus 1
-
-        for (int j = 0; j < rps->getNumRefIdc(); j++)
-        {
-            int refIdc = rps->getRefIdc(j);
-            WRITE_CODE((refIdc == 1 ? 1 : 0), 1, "used_by_curr_pic_flag"); //first bit is "1" if Idc is 1
-            if (refIdc != 1)
-            {
-                WRITE_CODE(refIdc >> 1, 1, "use_delta_flag"); //second bit is "1" if Idc is 2, "0" otherwise.
-            }
-        }
-    }
-    else
-    {
-        WRITE_UVLC(rps->getNumberOfNegativePictures(), "num_negative_pics");
-        WRITE_UVLC(rps->getNumberOfPositivePictures(), "num_positive_pics");
-        int prev = 0;
-        for (int j = 0; j < rps->getNumberOfNegativePictures(); j++)
-        {
-            WRITE_UVLC(prev - rps->getDeltaPOC(j) - 1, "delta_poc_s0_minus1");
-            prev = rps->getDeltaPOC(j);
-            WRITE_FLAG(rps->getUsed(j), "used_by_curr_pic_s0_flag");
-        }
-
-        prev = 0;
-        for (int j = rps->getNumberOfNegativePictures(); j < rps->getNumberOfNegativePictures() + rps->getNumberOfPositivePictures(); j++)
-        {
-            WRITE_UVLC(rps->getDeltaPOC(j) - prev - 1, "delta_poc_s1_minus1");
-            prev = rps->getDeltaPOC(j);
-            WRITE_FLAG(rps->getUsed(j), "used_by_curr_pic_s1_flag");
-        }
-    }
-}
-
-void TEncSbac::codeSPS(TComSPS* sps)
-{
-#if ENC_DEC_TRACE
-    xTraceSPSHeader(sps);
-#endif
-    WRITE_CODE(sps->getVPSId(),          4,       "sps_video_parameter_set_id");
-    WRITE_CODE(sps->getMaxTLayers() - 1,  3,       "sps_max_sub_layers_minus1");
-    WRITE_FLAG(sps->getTemporalIdNestingFlag() ? 1 : 0, "sps_temporal_id_nesting_flag");
-    codePTL(sps->getPTL(), 1, sps->getMaxTLayers() - 1);
-    WRITE_UVLC(sps->getSPSId(),                   "sps_seq_parameter_set_id");
-    WRITE_UVLC(sps->getChromaFormatIdc(),         "chroma_format_idc");
-
-    if (sps->getChromaFormatIdc() == CHROMA_444)
-    {
-        WRITE_FLAG(0,                             "separate_colour_plane_flag");
-    }
-
-    WRITE_UVLC(sps->getPicWidthInLumaSamples(),   "pic_width_in_luma_samples");
-    WRITE_UVLC(sps->getPicHeightInLumaSamples(),  "pic_height_in_luma_samples");
-    Window conf = sps->getConformanceWindow();
-
-    WRITE_FLAG(conf.m_enabledFlag,          "conformance_window_flag");
-    if (conf.m_enabledFlag)
-    {
-        WRITE_UVLC(conf.m_winLeftOffset   / TComSPS::getWinUnitX(sps->getChromaFormatIdc()), "conf_win_left_offset");
-        WRITE_UVLC(conf.m_winRightOffset  / TComSPS::getWinUnitX(sps->getChromaFormatIdc()), "conf_win_right_offset");
-        WRITE_UVLC(conf.m_winTopOffset    / TComSPS::getWinUnitY(sps->getChromaFormatIdc()), "conf_win_top_offset");
-        WRITE_UVLC(conf.m_winBottomOffset / TComSPS::getWinUnitY(sps->getChromaFormatIdc()), "conf_win_bottom_offset");
-    }
-
-    WRITE_UVLC(sps->getBitDepthY() - 8,             "bit_depth_luma_minus8");
-    WRITE_UVLC(sps->getBitDepthC() - 8,             "bit_depth_chroma_minus8");
-
-    WRITE_UVLC(sps->getBitsForPOC() - 4,            "log2_max_pic_order_cnt_lsb_minus4");
-
-    WRITE_FLAG(true,     "sps_sub_layer_ordering_info_present_flag");
-    for (uint32_t i = 0; i <= sps->getMaxTLayers() - 1; i++)
-    {
-        WRITE_UVLC(sps->getMaxDecPicBuffering(i) - 1, "sps_max_dec_pic_buffering_minus1[i]");
-        WRITE_UVLC(sps->getNumReorderPics(i),       "sps_num_reorder_pics[i]");
-        WRITE_UVLC(sps->getMaxLatencyIncrease(i),   "sps_max_latency_increase_plus1[i]");
-    }
-
-    WRITE_UVLC(sps->getLog2MinCodingBlockSize() - 3,    "log2_min_coding_block_size_minus3");
-    WRITE_UVLC(sps->getLog2DiffMaxMinCodingBlockSize(), "log2_diff_max_min_coding_block_size");
-    WRITE_UVLC(sps->getQuadtreeTULog2MinSize() - 2,     "log2_min_transform_block_size_minus2");
-    WRITE_UVLC(sps->getQuadtreeTULog2MaxSize() - sps->getQuadtreeTULog2MinSize(), "log2_diff_max_min_transform_block_size");
-    WRITE_UVLC(sps->getQuadtreeTUMaxDepthInter() - 1,   "max_transform_hierarchy_depth_inter");
-    WRITE_UVLC(sps->getQuadtreeTUMaxDepthIntra() - 1,   "max_transform_hierarchy_depth_intra");
-    WRITE_FLAG(sps->getScalingListFlag() ? 1 : 0,       "scaling_list_enabled_flag");
-    if (sps->getScalingListFlag())
-    {
-        WRITE_FLAG(sps->getScalingListPresentFlag() ? 1 : 0, "sps_scaling_list_data_present_flag");
-        if (sps->getScalingListPresentFlag())
-        {
-            codeScalingList(m_slice->getScalingList());
-        }
-    }
-    WRITE_FLAG(sps->getUseAMP() ? 1 : 0, "amp_enabled_flag");
-    WRITE_FLAG(sps->getUseSAO() ? 1 : 0, "sample_adaptive_offset_enabled_flag");
-
-    WRITE_FLAG(sps->getUsePCM() ? 1 : 0, "pcm_enabled_flag");
-    if (sps->getUsePCM())
-    {
-        WRITE_CODE(sps->getPCMBitDepthLuma() - 1, 4,                     "pcm_sample_bit_depth_luma_minus1");
-        WRITE_CODE(sps->getPCMBitDepthChroma() - 1, 4,                   "pcm_sample_bit_depth_chroma_minus1");
-        WRITE_UVLC(sps->getPCMLog2MinSize() - 3,                         "log2_min_pcm_luma_coding_block_size_minus3");
-        WRITE_UVLC(sps->getPCMLog2MaxSize() - sps->getPCMLog2MinSize(),  "log2_diff_max_min_pcm_luma_coding_block_size");
-        WRITE_FLAG(sps->getPCMFilterDisableFlag() ? 1 : 0,               "pcm_loop_filter_disable_flag");
-    }
-
-    assert(sps->getMaxTLayers() > 0);
-
-    TComRPSList* rpsList = sps->getRPSList();
-    TComReferencePictureSet* rps;
-
-    WRITE_UVLC(rpsList->getNumberOfReferencePictureSets(), "num_short_term_ref_pic_sets");
-    for (int i = 0; i < rpsList->getNumberOfReferencePictureSets(); i++)
-    {
-        rps = rpsList->getReferencePictureSet(i);
-        codeShortTermRefPicSet(rps, false, i);
-    }
-
-    WRITE_FLAG(sps->getLongTermRefsPresent() ? 1 : 0,      "long_term_ref_pics_present_flag");
-    if (sps->getLongTermRefsPresent())
-    {
-        WRITE_UVLC(sps->getNumLongTermRefPicSPS(), "num_long_term_ref_pic_sps");
-        for (uint32_t k = 0; k < sps->getNumLongTermRefPicSPS(); k++)
-        {
-            WRITE_CODE(sps->getLtRefPicPocLsbSps(k), sps->getBitsForPOC(), "lt_ref_pic_poc_lsb_sps");
-            WRITE_FLAG(sps->getUsedByCurrPicLtSPSFlag(k), "used_by_curr_pic_lt_sps_flag");
-        }
-    }
-    WRITE_FLAG(sps->getTMVPFlagsPresent()  ? 1 : 0,        "sps_temporal_mvp_enable_flag");
-
-    WRITE_FLAG(sps->getUseStrongIntraSmoothing(),          "sps_strong_intra_smoothing_enable_flag");
-
-    WRITE_FLAG(sps->getVuiParametersPresentFlag(),         "vui_parameters_present_flag");
-    if (sps->getVuiParametersPresentFlag())
-    {
-        codeVUI(sps->getVuiParameters(), sps);
-    }
-
-    WRITE_FLAG(0, "sps_extension_flag");
-}
-
-void TEncSbac::codePPS(TComPPS* pps)
-{
-#if ENC_DEC_TRACE
-    xTracePPSHeader(pps);
-#endif
-
-    WRITE_UVLC(pps->getPPSId(),                            "pps_pic_parameter_set_id");
-    WRITE_UVLC(pps->getSPSId(),                            "pps_seq_parameter_set_id");
-    WRITE_FLAG(0,                                          "dependent_slice_segments_enabled_flag");
-    WRITE_FLAG(pps->getOutputFlagPresentFlag() ? 1 : 0,    "output_flag_present_flag");
-    WRITE_CODE(pps->getNumExtraSliceHeaderBits(), 3,       "num_extra_slice_header_bits");
-    WRITE_FLAG(pps->getSignHideFlag(),                     "sign_data_hiding_flag");
-    WRITE_FLAG(pps->getCabacInitPresentFlag() ? 1 : 0,     "cabac_init_present_flag");
-    WRITE_UVLC(pps->getNumRefIdxL0DefaultActive() - 1,     "num_ref_idx_l0_default_active_minus1");
-    WRITE_UVLC(pps->getNumRefIdxL1DefaultActive() - 1,     "num_ref_idx_l1_default_active_minus1");
-
-    WRITE_SVLC(pps->getPicInitQPMinus26(),                 "init_qp_minus26");
-    WRITE_FLAG(pps->getConstrainedIntraPred() ? 1 : 0,     "constrained_intra_pred_flag");
-    WRITE_FLAG(pps->getUseTransformSkip() ? 1 : 0,         "transform_skip_enabled_flag");
-    WRITE_FLAG(pps->getUseDQP() ? 1 : 0,                   "cu_qp_delta_enabled_flag");
-    if (pps->getUseDQP())
-    {
-        WRITE_UVLC(pps->getMaxCuDQPDepth(),                "diff_cu_qp_delta_depth");
-    }
-    WRITE_SVLC(pps->getChromaCbQpOffset(),                 "pps_cb_qp_offset");
-    WRITE_SVLC(pps->getChromaCrQpOffset(),                 "pps_cr_qp_offset");
-    WRITE_FLAG(pps->getSliceChromaQpFlag() ? 1 : 0,        "pps_slice_chroma_qp_offsets_present_flag");
-
-    WRITE_FLAG(pps->getUseWP() ? 1 : 0,                    "weighted_pred_flag");   // Use of Weighting Prediction (P_SLICE)
-    WRITE_FLAG(pps->getWPBiPred() ? 1 : 0,                 "weighted_bipred_flag");  // Use of Weighting Bi-Prediction (B_SLICE)
-    WRITE_FLAG(pps->getTransquantBypassEnableFlag() ? 1 : 0, "transquant_bypass_enable_flag");
-    WRITE_FLAG(0,                                          "tiles_enabled_flag");
-    WRITE_FLAG(pps->getEntropyCodingSyncEnabledFlag() ? 1 : 0, "entropy_coding_sync_enabled_flag");
-    WRITE_FLAG(1,                                          "loop_filter_across_slices_enabled_flag");
-
-    // TODO: Here have some time sequence problem, we set below field in initEncSlice(), but use them in getStreamHeaders() early
-    WRITE_FLAG(pps->getDeblockingFilterControlPresentFlag() ? 1 : 0, "deblocking_filter_control_present_flag");
-    if (pps->getDeblockingFilterControlPresentFlag())
-    {
-        WRITE_FLAG(pps->getDeblockingFilterOverrideEnabledFlag() ? 1 : 0,  "deblocking_filter_override_enabled_flag");
-        WRITE_FLAG(pps->getPicDisableDeblockingFilterFlag() ? 1 : 0,       "pps_disable_deblocking_filter_flag");
-        if (!pps->getPicDisableDeblockingFilterFlag())
-        {
-            WRITE_SVLC(pps->getDeblockingFilterBetaOffsetDiv2(), "pps_beta_offset_div2");
-            WRITE_SVLC(pps->getDeblockingFilterTcOffsetDiv2(),   "pps_tc_offset_div2");
-        }
-    }
-    WRITE_FLAG(pps->getScalingListPresentFlag() ? 1 : 0,         "pps_scaling_list_data_present_flag");
-    if (pps->getScalingListPresentFlag())
-    {
-        codeScalingList(m_slice->getScalingList());
-    }
-    WRITE_FLAG(pps->getListsModificationPresentFlag(), "lists_modification_present_flag");
-    WRITE_UVLC(pps->getLog2ParallelMergeLevelMinus2(), "log2_parallel_merge_level_minus2");
-    WRITE_FLAG(pps->getSliceHeaderExtensionPresentFlag() ? 1 : 0, "slice_segment_header_extension_present_flag");
-    WRITE_FLAG(0, "pps_extension_flag");
-}
-
-void TEncSbac::codeVUI(TComVUI *vui, TComSPS* sps)
-{
-#if ENC_DEC_TRACE
-    fprintf(g_hTrace, "----------- vui_parameters -----------\n");
-#endif
-    WRITE_FLAG(vui->getAspectRatioInfoPresentFlag(),  "aspect_ratio_info_present_flag");
-    if (vui->getAspectRatioInfoPresentFlag())
-    {
-        WRITE_CODE(vui->getAspectRatioIdc(), 8,       "aspect_ratio_idc");
-        if (vui->getAspectRatioIdc() == 255)
-        {
-            WRITE_CODE(vui->getSarWidth(), 16,        "sar_width");
-            WRITE_CODE(vui->getSarHeight(), 16,       "sar_height");
-        }
-    }
-    WRITE_FLAG(vui->getOverscanInfoPresentFlag(),     "overscan_info_present_flag");
-    if (vui->getOverscanInfoPresentFlag())
-    {
-        WRITE_FLAG(vui->getOverscanAppropriateFlag(), "overscan_appropriate_flag");
-    }
-    WRITE_FLAG(vui->getVideoSignalTypePresentFlag(),  "video_signal_type_present_flag");
-    if (vui->getVideoSignalTypePresentFlag())
-    {
-        WRITE_CODE(vui->getVideoFormat(), 3,          "video_format");
-        WRITE_FLAG(vui->getVideoFullRangeFlag(),      "video_full_range_flag");
-        WRITE_FLAG(vui->getColourDescriptionPresentFlag(), "colour_description_present_flag");
-        if (vui->getColourDescriptionPresentFlag())
-        {
-            WRITE_CODE(vui->getColourPrimaries(), 8,         "colour_primaries");
-            WRITE_CODE(vui->getTransferCharacteristics(), 8, "transfer_characteristics");
-            WRITE_CODE(vui->getMatrixCoefficients(), 8,      "matrix_coefficients");
-        }
-    }
-
-    WRITE_FLAG(vui->getChromaLocInfoPresentFlag(),           "chroma_loc_info_present_flag");
-    if (vui->getChromaLocInfoPresentFlag())
-    {
-        WRITE_UVLC(vui->getChromaSampleLocTypeTopField(),    "chroma_sample_loc_type_top_field");
-        WRITE_UVLC(vui->getChromaSampleLocTypeBottomField(), "chroma_sample_loc_type_bottom_field");
-    }
-
-    WRITE_FLAG(vui->getNeutralChromaIndicationFlag(),        "neutral_chroma_indication_flag");
-    WRITE_FLAG(vui->getFieldSeqFlag(),                       "field_seq_flag");
-    WRITE_FLAG(vui->getFrameFieldInfoPresentFlag(),          "frame_field_info_present_flag");
-
-    Window defaultDisplayWindow = vui->getDefaultDisplayWindow();
-    WRITE_FLAG(defaultDisplayWindow.m_enabledFlag,           "default_display_window_flag");
-    if (defaultDisplayWindow.m_enabledFlag)
-    {
-        WRITE_UVLC(defaultDisplayWindow.m_winLeftOffset,     "def_disp_win_left_offset");
-        WRITE_UVLC(defaultDisplayWindow.m_winRightOffset,    "def_disp_win_right_offset");
-        WRITE_UVLC(defaultDisplayWindow.m_winTopOffset,      "def_disp_win_top_offset");
-        WRITE_UVLC(defaultDisplayWindow.m_winBottomOffset,   "def_disp_win_bottom_offset");
-    }
-    TimingInfo *timingInfo = vui->getTimingInfo();
-    WRITE_FLAG(timingInfo->getTimingInfoPresentFlag(),       "vui_timing_info_present_flag");
-    if (timingInfo->getTimingInfoPresentFlag())
-    {
-        WRITE_CODE(timingInfo->getNumUnitsInTick(), 32,      "vui_num_units_in_tick");
-        WRITE_CODE(timingInfo->getTimeScale(),      32,      "vui_time_scale");
-        WRITE_FLAG(timingInfo->getPocProportionalToTimingFlag(),  "vui_poc_proportional_to_timing_flag");
-        if (timingInfo->getPocProportionalToTimingFlag())
-        {
-            WRITE_UVLC(timingInfo->getNumTicksPocDiffOneMinus1(),   "vui_num_ticks_poc_diff_one_minus1");
-        }
-        WRITE_FLAG(vui->getHrdParametersPresentFlag(),              "hrd_parameters_present_flag");
-        if (vui->getHrdParametersPresentFlag())
-        {
-            codeHrdParameters(vui->getHrdParameters(), 1, sps->getMaxTLayers() - 1);
-        }
-    }
-    WRITE_FLAG(vui->getBitstreamRestrictionFlag(),                "bitstream_restriction_flag");
-    if (vui->getBitstreamRestrictionFlag())
-    {
-        WRITE_FLAG(vui->getTilesFixedStructureFlag(),             "tiles_fixed_structure_flag");
-        WRITE_FLAG(vui->getMotionVectorsOverPicBoundariesFlag(),  "motion_vectors_over_pic_boundaries_flag");
-        WRITE_FLAG(vui->getRestrictedRefPicListsFlag(),           "restricted_ref_pic_lists_flag");
-        WRITE_UVLC(vui->getMinSpatialSegmentationIdc(),           "min_spatial_segmentation_idc");
-        WRITE_UVLC(vui->getMaxBytesPerPicDenom(),                 "max_bytes_per_pic_denom");
-        WRITE_UVLC(vui->getMaxBitsPerMinCuDenom(),                "max_bits_per_mincu_denom");
-        WRITE_UVLC(vui->getLog2MaxMvLengthHorizontal(),           "log2_max_mv_length_horizontal");
-        WRITE_UVLC(vui->getLog2MaxMvLengthVertical(),             "log2_max_mv_length_vertical");
-    }
-}
-
-void TEncSbac::codeHrdParameters(TComHRD *hrd, bool commonInfPresentFlag, uint32_t maxNumSubLayersMinus1)
-{
-    if (commonInfPresentFlag)
-    {
-        WRITE_FLAG(hrd->getNalHrdParametersPresentFlag() ? 1 : 0,  "nal_hrd_parameters_present_flag");
-        WRITE_FLAG(hrd->getVclHrdParametersPresentFlag() ? 1 : 0,  "vcl_hrd_parameters_present_flag");
-        if (hrd->getNalHrdParametersPresentFlag() || hrd->getVclHrdParametersPresentFlag())
-        {
-            WRITE_FLAG(hrd->getSubPicHrdParamsPresentFlag() ? 1 : 0,  "sub_pic_hrd_params_present_flag");
-            if (hrd->getSubPicHrdParamsPresentFlag())
-            {
-                WRITE_CODE(hrd->getTickDivisorMinus2(), 8,              "tick_divisor_minus2");
-                WRITE_CODE(hrd->getDuCpbRemovalDelayLengthMinus1(), 5,  "du_cpb_removal_delay_length_minus1");
-                WRITE_FLAG(hrd->getSubPicCpbParamsInPicTimingSEIFlag() ? 1 : 0, "sub_pic_cpb_params_in_pic_timing_sei_flag");
-                WRITE_CODE(hrd->getDpbOutputDelayDuLengthMinus1(), 5,   "dpb_output_delay_du_length_minus1");
-            }
-            WRITE_CODE(hrd->getBitRateScale(), 4,                     "bit_rate_scale");
-            WRITE_CODE(hrd->getCpbSizeScale(), 4,                     "cpb_size_scale");
-            if (hrd->getSubPicHrdParamsPresentFlag())
-            {
-                WRITE_CODE(hrd->getDuCpbSizeScale(), 4,                "du_cpb_size_scale");
-            }
-            WRITE_CODE(hrd->getInitialCpbRemovalDelayLengthMinus1(), 5, "initial_cpb_removal_delay_length_minus1");
-            WRITE_CODE(hrd->getCpbRemovalDelayLengthMinus1(),        5, "au_cpb_removal_delay_length_minus1");
-            WRITE_CODE(hrd->getDpbOutputDelayLengthMinus1(),         5, "dpb_output_delay_length_minus1");
-        }
-    }
-    int i, j, nalOrVcl;
-    for (i = 0; i <= maxNumSubLayersMinus1; i++)
-    {
-        WRITE_FLAG(hrd->getFixedPicRateFlag(i) ? 1 : 0,          "fixed_pic_rate_general_flag");
-        if (!hrd->getFixedPicRateFlag(i))
-        {
-            WRITE_FLAG(hrd->getFixedPicRateWithinCvsFlag(i) ? 1 : 0, "fixed_pic_rate_within_cvs_flag");
-        }
-        else
-        {
-            hrd->setFixedPicRateWithinCvsFlag(i, true);
-        }
-        if (hrd->getFixedPicRateWithinCvsFlag(i))
-        {
-            WRITE_UVLC(hrd->getPicDurationInTcMinus1(i), "elemental_duration_in_tc_minus1");
-        }
-        else
-        {
-            WRITE_FLAG(hrd->getLowDelayHrdFlag(i) ? 1 : 0, "low_delay_hrd_flag");
-        }
-        if (!hrd->getLowDelayHrdFlag(i))
-        {
-            WRITE_UVLC(hrd->getCpbCntMinus1(i), "cpb_cnt_minus1");
-        }
-
-        for (nalOrVcl = 0; nalOrVcl < 2; nalOrVcl++)
-        {
-            if (((nalOrVcl == 0) && (hrd->getNalHrdParametersPresentFlag())) ||
-                ((nalOrVcl == 1) && (hrd->getVclHrdParametersPresentFlag())))
-            {
-                for (j = 0; j <= (hrd->getCpbCntMinus1(i)); j++)
-                {
-                    WRITE_UVLC(hrd->getBitRateValueMinus1(i, j, nalOrVcl), "bit_rate_value_minus1");
-                    WRITE_UVLC(hrd->getCpbSizeValueMinus1(i, j, nalOrVcl), "cpb_size_value_minus1");
-                    if (hrd->getSubPicHrdParamsPresentFlag())
-                    {
-                        WRITE_UVLC(hrd->getDuCpbSizeValueMinus1(i, j, nalOrVcl), "cpb_size_du_value_minus1");
-                        WRITE_UVLC(hrd->getDuBitRateValueMinus1(i, j, nalOrVcl), "bit_rate_du_value_minus1");
-                    }
-                    WRITE_FLAG(hrd->getCbrFlag(i, j, nalOrVcl) ? 1 : 0, "cbr_flag");
-                }
-            }
-        }
-    }
-}
-
-void TEncSbac::codePTL(TComPTL* ptl, bool profilePresentFlag, int maxNumSubLayersMinus1)
-{
-    if (profilePresentFlag)
-    {
-        codeProfileTier(ptl->getGeneralPTL()); // general_...
-    }
-    WRITE_CODE(ptl->getGeneralPTL()->getLevelIdc(), 8, "general_level_idc");
-
-    for (int i = 0; i < maxNumSubLayersMinus1; i++)
-    {
-        if (profilePresentFlag)
-        {
-            WRITE_FLAG(ptl->getSubLayerProfilePresentFlag(i), "sub_layer_profile_present_flag[i]");
-        }
-
-        WRITE_FLAG(ptl->getSubLayerLevelPresentFlag(i), "sub_layer_level_present_flag[i]");
-    }
-
-    if (maxNumSubLayersMinus1 > 0)
-    {
-        for (int i = maxNumSubLayersMinus1; i < 8; i++)
-        {
-            WRITE_CODE(0, 2, "reserved_zero_2bits");
-        }
-    }
-
-    for (int i = 0; i < maxNumSubLayersMinus1; i++)
-    {
-        if (profilePresentFlag && ptl->getSubLayerProfilePresentFlag(i))
-        {
-            codeProfileTier(ptl->getSubLayerPTL(i)); // sub_layer_...
-        }
-        if (ptl->getSubLayerLevelPresentFlag(i))
-        {
-            WRITE_CODE(ptl->getSubLayerPTL(i)->getLevelIdc(), 8, "sub_layer_level_idc[i]");
-        }
-    }
-}
-
-void TEncSbac::codeProfileTier(ProfileTierLevel* ptl)
-{
-    WRITE_CODE(ptl->getProfileSpace(), 2, "XXX_profile_space[]");
-    WRITE_FLAG(ptl->getTierFlag(),        "XXX_tier_flag[]");
-    WRITE_CODE(ptl->getProfileIdc(), 5,   "XXX_profile_idc[]");
-    for (int j = 0; j < 32; j++)
-    {
-        WRITE_FLAG(ptl->getProfileCompatibilityFlag(j), "XXX_profile_compatibility_flag[][j]");
-    }
-
-    WRITE_FLAG(ptl->getProgressiveSourceFlag(),   "general_progressive_source_flag");
-    WRITE_FLAG(ptl->getInterlacedSourceFlag(),    "general_interlaced_source_flag");
-    WRITE_FLAG(ptl->getNonPackedConstraintFlag(), "general_non_packed_constraint_flag");
-    WRITE_FLAG(ptl->getFrameOnlyConstraintFlag(), "general_frame_only_constraint_flag");
-
-    WRITE_CODE(0, 16, "XXX_reserved_zero_44bits[0..15]");
-    WRITE_CODE(0, 16, "XXX_reserved_zero_44bits[16..31]");
-    WRITE_CODE(0, 12, "XXX_reserved_zero_44bits[32..43]");
-}
-
-/* code explicit wp tables */
-void TEncSbac::xCodePredWeightTable(TComSlice* slice)
-{
-    wpScalingParam  *wp;
-    bool            bChroma      = true; // color always present in HEVC ?
-    int             numRefDirs   = (slice->getSliceType() == B_SLICE) ? (2) : (1);
-    bool            bDenomCoded  = false;
-    uint32_t        mode = 0;
-    uint32_t        totalSignalledWeightFlags = 0;
-
-    if ((slice->getSliceType() == P_SLICE && slice->getPPS()->getUseWP()) || (slice->getSliceType() == B_SLICE && slice->getPPS()->getWPBiPred()))
-    {
-        mode = 1; // explicit
-    }
-    if (mode == 1)
-    {
-        for (int picList = 0; picList < numRefDirs; picList++)
-        {
-            for (int refIdx = 0; refIdx < slice->getNumRefIdx(picList); refIdx++)
-            {
-                slice->getWpScaling(picList, refIdx, wp);
-                if (!bDenomCoded)
-                {
-                    int iDeltaDenom;
-                    WRITE_UVLC(wp[0].log2WeightDenom, "luma_log2_weight_denom"); // ue(v): luma_log2_weight_denom
-
-                    if (bChroma)
-                    {
-                        iDeltaDenom = (wp[1].log2WeightDenom - wp[0].log2WeightDenom);
-                        WRITE_SVLC(iDeltaDenom, "delta_chroma_log2_weight_denom"); // se(v): delta_chroma_log2_weight_denom
-                    }
-                    bDenomCoded = true;
-                }
-                WRITE_FLAG(wp[0].bPresentFlag, "luma_weight_lX_flag");         // u(1): luma_weight_lX_flag
-                totalSignalledWeightFlags += wp[0].bPresentFlag;
-            }
-
-            if (bChroma)
-            {
-                for (int refIdx = 0; refIdx < slice->getNumRefIdx(picList); refIdx++)
-                {
-                    slice->getWpScaling(picList, refIdx, wp);
-                    WRITE_FLAG(wp[1].bPresentFlag, "chroma_weight_lX_flag");   // u(1): chroma_weight_lX_flag
-                    totalSignalledWeightFlags += 2 * wp[1].bPresentFlag;
-                }
-            }
-
-            for (int refIdx = 0; refIdx < slice->getNumRefIdx(picList); refIdx++)
-            {
-                slice->getWpScaling(picList, refIdx, wp);
-                if (wp[0].bPresentFlag)
-                {
-                    int iDeltaWeight = (wp[0].inputWeight - (1 << wp[0].log2WeightDenom));
-                    WRITE_SVLC(iDeltaWeight, "delta_luma_weight_lX");          // se(v): delta_luma_weight_lX
-                    WRITE_SVLC(wp[0].inputOffset, "luma_offset_lX");               // se(v): luma_offset_lX
-                }
-
-                if (bChroma)
-                {
-                    if (wp[1].bPresentFlag)
-                    {
-                        for (int j = 1; j < 3; j++)
-                        {
-                            int iDeltaWeight = (wp[j].inputWeight - (1 << wp[1].log2WeightDenom));
-                            WRITE_SVLC(iDeltaWeight, "delta_chroma_weight_lX"); // se(v): delta_chroma_weight_lX
-
-                            int pred = (128 - ((128 * wp[j].inputWeight) >> (wp[j].log2WeightDenom)));
-                            int iDeltaChroma = (wp[j].inputOffset - pred);
-                            WRITE_SVLC(iDeltaChroma, "delta_chroma_offset_lX"); // se(v): delta_chroma_offset_lX
-                        }
-                    }
-                }
-            }
-        }
-
-        assert(totalSignalledWeightFlags <= 24);
-    }
-}
-
-/** code quantization matrix
- *  \param scalingList quantization matrix information
- */
-void TEncSbac::codeScalingList(TComScalingList* scalingList)
-{
-    uint32_t listId, sizeId;
-    bool scalingListPredModeFlag;
-
-    //for each size
-    for (sizeId = 0; sizeId < SCALING_LIST_SIZE_NUM; sizeId++)
-    {
-        for (listId = 0; listId < g_scalingListNum[sizeId]; listId++)
-        {
-            scalingListPredModeFlag = scalingList->checkPredMode(sizeId, listId);
-            WRITE_FLAG(scalingListPredModeFlag, "scaling_list_pred_mode_flag");
-            if (!scalingListPredModeFlag) // Copy Mode
-            {
-                WRITE_UVLC((int)listId - (int)scalingList->getRefMatrixId(sizeId, listId), "scaling_list_pred_matrix_id_delta");
-            }
-            else // DPCM Mode
-            {
-                xCodeScalingList(scalingList, sizeId, listId);
-            }
-        }
-    }
-}
-
-/** code DPCM
- * \param scalingList quantization matrix information
- * \param sizeIdc size index
- * \param listIdc list index
- */
-void TEncSbac::xCodeScalingList(TComScalingList* scalingList, uint32_t sizeId, uint32_t listId)
-{
-    int coefNum = X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId]);
-    const uint16_t* scan = g_scanOrder[SCAN_UNGROUPED][SCAN_DIAG][sizeId == 0 ? 2 : 3];
-    int nextCoef = SCALING_LIST_START_VALUE;
-    int data;
-    int32_t *src = scalingList->getScalingListAddress(sizeId, listId);
-
-    if (sizeId > SCALING_LIST_8x8)
-    {
-        WRITE_SVLC(scalingList->getScalingListDC(sizeId, listId) - 8, "scaling_list_dc_coef_minus8");
-        nextCoef = scalingList->getScalingListDC(sizeId, listId);
-    }
-    for (int i = 0; i < coefNum; i++)
-    {
-        data = src[scan[i]] - nextCoef;
-        nextCoef = src[scan[i]];
-        if (data > 127)
-        {
-            data = data - 256;
-        }
-        if (data < -128)
-        {
-            data = data + 256;
-        }
-
-        WRITE_SVLC(data,  "scaling_list_delta_coef");
-    }
-}
-
-bool TEncSbac::findMatchingLTRP(TComSlice* slice, uint32_t *ltrpsIndex, int ltrpPOC, bool usedFlag)
-{
-    // bool state = true, state2 = false;
-    int lsb = ltrpPOC % (1 << slice->getSPS()->getBitsForPOC());
-
-    for (int k = 0; k < slice->getSPS()->getNumLongTermRefPicSPS(); k++)
-    {
-        if ((lsb == slice->getSPS()->getLtRefPicPocLsbSps(k)) && (usedFlag == slice->getSPS()->getUsedByCurrPicLtSPSFlag(k)))
-        {
-            *ltrpsIndex = k;
-            return true;
-        }
-    }
-
-    return false;
-}
-
-bool TComScalingList::checkPredMode(uint32_t sizeId, uint32_t listId)
-{
-    for (int predListIdx = (int)listId; predListIdx >= 0; predListIdx--)
-    {
-        if (!memcmp(getScalingListAddress(sizeId, listId),
-                    ((listId == predListIdx) ? getScalingListDefaultAddress(sizeId, predListIdx) : getScalingListAddress(sizeId, predListIdx)),
-                    sizeof(int) * X265_MIN(MAX_MATRIX_COEF_NUM, (int)g_scalingListSize[sizeId])) // check value of matrix
-            && ((sizeId < SCALING_LIST_16x16) || (getScalingListDC(sizeId, listId) == getScalingListDC(sizeId, predListIdx)))) // check DC value
-        {
-            setRefMatrixId(sizeId, listId, predListIdx);
-            return false;
-        }
-    }
-
-    return true;
-}
-
-void TEncSbac::codeSliceHeader(TComSlice* slice)
-{
-#if ENC_DEC_TRACE
-    xTraceSliceHeader(slice);
-#endif
-
-    //calculate number of bits required for slice address
-    int maxSliceSegmentAddress = slice->getPic()->getNumCUsInFrame();
-    int bitsSliceSegmentAddress = 0;
-    while (maxSliceSegmentAddress > (1 << bitsSliceSegmentAddress))
-    {
-        bitsSliceSegmentAddress++;
-    }
-
-    //write slice address
-    int sliceSegmentAddress = 0;
-
-    WRITE_FLAG(sliceSegmentAddress == 0, "first_slice_segment_in_pic_flag");
-    if (slice->getRapPicFlag())
-    {
-        WRITE_FLAG(0, "no_output_of_prior_pics_flag");
-    }
-    WRITE_UVLC(slice->getPPS()->getPPSId(), "slice_pic_parameter_set_id");
-    slice->setDependentSliceSegmentFlag(!slice->isNextSlice());
-    if (sliceSegmentAddress > 0)
-    {
-        WRITE_CODE(sliceSegmentAddress, bitsSliceSegmentAddress, "slice_segment_address");
-    }
-    if (!slice->getDependentSliceSegmentFlag())
-    {
-        for (int i = 0; i < slice->getPPS()->getNumExtraSliceHeaderBits(); i++)
-        {
-            assert(!!"slice_reserved_undetermined_flag[]");
-            WRITE_FLAG(0, "slice_reserved_undetermined_flag[]");
-        }
-
-        WRITE_UVLC(slice->getSliceType(),       "slice_type");
-
-        if (slice->getPPS()->getOutputFlagPresentFlag())
-        {
-            WRITE_FLAG(slice->getPicOutputFlag() ? 1 : 0, "pic_output_flag");
-        }
-        if (slice->getSPS()->getChromaFormatIdc() == CHROMA_444)
-        {
-            //In this version separate_color_plane_flag is 0
-            if (slice->getSPS()->getSeparateColorPlaneFlag())
-            {
-                //plane_id values 0, 1, and 2 correspond to the Y, Cb, and Cr planes, respectively.
-                //WRITE_FLAG(0, "colour_plane_id");
-            }
-        }
-        if (!slice->getIdrPicFlag())
-        {
-            int picOrderCntLSB = (slice->getPOC() - slice->getLastIDR() + (1 << slice->getSPS()->getBitsForPOC())) % (1 << slice->getSPS()->getBitsForPOC());
-            WRITE_CODE(picOrderCntLSB, slice->getSPS()->getBitsForPOC(), "pic_order_cnt_lsb");
-            TComReferencePictureSet* rps = slice->getRPS();
-
-            // check for bitstream restriction stating that:
-            // If the current picture is a BLA or CRA picture, the value of NumPocTotalCurr shall be equal to 0.
-            // Ideally this process should not be repeated for each slice in a picture
-            if (slice->isIRAP())
-            {
-                for (int picIdx = 0; picIdx < rps->getNumberOfPictures(); picIdx++)
-                {
-                    assert(!rps->getUsed(picIdx));
-                }
-            }
-
-            if (slice->getRPSidx() < 0)
-            {
-                WRITE_FLAG(0, "short_term_ref_pic_set_sps_flag");
-                codeShortTermRefPicSet(rps, true, slice->getSPS()->getRPSList()->getNumberOfReferencePictureSets());
-            }
-            else
-            {
-                WRITE_FLAG(1, "short_term_ref_pic_set_sps_flag");
-                int numBits = 0;
-                while ((1 << numBits) < slice->getSPS()->getRPSList()->getNumberOfReferencePictureSets())
-                {
-                    numBits++;
-                }
-
-                if (numBits > 0)
-                {
-                    WRITE_CODE(slice->getRPSidx(), numBits, "short_term_ref_pic_set_idx");
-                }
-            }
-            if (slice->getSPS()->getLongTermRefsPresent())
-            {
-                int numLtrpInSH = rps->getNumberOfLongtermPictures();
-                int ltrpInSPS[MAX_NUM_REF_PICS];
-                int numLtrpInSPS = 0;
-                uint32_t ltrpIndex;
-                int counter = 0;
-                for (int k = rps->getNumberOfPictures() - 1; k > rps->getNumberOfPictures() - rps->getNumberOfLongtermPictures() - 1; k--)
-                {
-                    if (findMatchingLTRP(slice, &ltrpIndex, rps->getPOC(k), rps->getUsed(k)))
-                    {
-                        ltrpInSPS[numLtrpInSPS] = ltrpIndex;
-                        numLtrpInSPS++;
-                    }
-                    else
-                    {
-                        counter++;
-                    }
-                }
-
-                numLtrpInSH -= numLtrpInSPS;
-
-                int bitsForLtrpInSPS = 0;
-                while (slice->getSPS()->getNumLongTermRefPicSPS() > (1 << bitsForLtrpInSPS))
-                {
-                    bitsForLtrpInSPS++;
-                }
-
-                if (slice->getSPS()->getNumLongTermRefPicSPS() > 0)
-                {
-                    WRITE_UVLC(numLtrpInSPS, "num_long_term_sps");
-                }
-                WRITE_UVLC(numLtrpInSH, "num_long_term_pics");
-                // Note that the LSBs of the LT ref. pic. POCs must be sorted before.
-                // Not sorted here because LT ref indices will be used in setRefPicList()
-                int prevDeltaMSB = 0, prevLSB = 0;
-                int offset = rps->getNumberOfNegativePictures() + rps->getNumberOfPositivePictures();
-                for (int i = rps->getNumberOfPictures() - 1; i > offset - 1; i--)
-                {
-                    if (counter < numLtrpInSPS)
-                    {
-                        if (bitsForLtrpInSPS > 0)
-                        {
-                            WRITE_CODE(ltrpInSPS[counter], bitsForLtrpInSPS, "lt_idx_sps[i]");
-                        }
-                    }
-                    else
-                    {
-                        WRITE_CODE(rps->getPocLSBLT(i), slice->getSPS()->getBitsForPOC(), "poc_lsb_lt");
-                        WRITE_FLAG(rps->getUsed(i), "used_by_curr_pic_lt_flag");
-                    }
-                    WRITE_FLAG(rps->getDeltaPocMSBPresentFlag(i), "delta_poc_msb_present_flag");
-
-                    if (rps->getDeltaPocMSBPresentFlag(i))
-                    {
-                        bool deltaFlag = false;
-                        //  First LTRP from SPS                 ||  First LTRP from SH                              || curr LSB            != prev LSB
-                        if ((i == rps->getNumberOfPictures() - 1) || (i == rps->getNumberOfPictures() - 1 - numLtrpInSPS) || (rps->getPocLSBLT(i) != prevLSB))
-                        {
-                            deltaFlag = true;
-                        }
-                        if (deltaFlag)
-                        {
-                            WRITE_UVLC(rps->getDeltaPocMSBCycleLT(i), "delta_poc_msb_cycle_lt[i]");
-                        }
-                        else
-                        {
-                            int differenceInDeltaMSB = rps->getDeltaPocMSBCycleLT(i) - prevDeltaMSB;
-                            assert(differenceInDeltaMSB >= 0);
-                            WRITE_UVLC(differenceInDeltaMSB, "delta_poc_msb_cycle_lt[i]");
-                        }
-                        prevLSB = rps->getPocLSBLT(i);
-                        prevDeltaMSB = rps->getDeltaPocMSBCycleLT(i);
-                    }
-                }
-            }
-            if (slice->getSPS()->getTMVPFlagsPresent())
-            {
-                WRITE_FLAG(slice->getEnableTMVPFlag() ? 1 : 0, "slice_temporal_mvp_enable_flag");
-            }
-        }
-        if (slice->getSPS()->getUseSAO())
-        {
-            if (slice->getSPS()->getUseSAO())
-            {
-                WRITE_FLAG(slice->getSaoEnabledFlag(), "slice_sao_luma_flag");
-                {
-                    SAOParam *saoParam = slice->getPic()->getPicSym()->getSaoParam();
-                    WRITE_FLAG(saoParam->bSaoFlag[1], "slice_sao_chroma_flag");
-                }
-            }
-        }
-
-        //check if numrefidxes match the defaults. If not, override
-
-        if (!slice->isIntra())
-        {
-            bool overrideFlag = (slice->getNumRefIdx(REF_PIC_LIST_0) != slice->getPPS()->getNumRefIdxL0DefaultActive() || (slice->isInterB() && slice->getNumRefIdx(REF_PIC_LIST_1) != slice->getPPS()->getNumRefIdxL1DefaultActive()));
-            WRITE_FLAG(overrideFlag ? 1 : 0,                               "num_ref_idx_active_override_flag");
-            if (overrideFlag)
-            {
-                WRITE_UVLC(slice->getNumRefIdx(REF_PIC_LIST_0) - 1,      "num_ref_idx_l0_active_minus1");
-                if (slice->isInterB())
-                {
-                    WRITE_UVLC(slice->getNumRefIdx(REF_PIC_LIST_1) - 1,    "num_ref_idx_l1_active_minus1");
-                }
-                else
-                {
-                    slice->setNumRefIdx(REF_PIC_LIST_1, 0);
-                }
-            }
-        }
-        else
-        {
-            slice->setNumRefIdx(REF_PIC_LIST_0, 0);
-            slice->setNumRefIdx(REF_PIC_LIST_1, 0);
-        }
-
-        if (slice->isInterB())
-        {
-            WRITE_FLAG(slice->getMvdL1ZeroFlag() ? 1 : 0,   "mvd_l1_zero_flag");
-        }
-
-        if (!slice->isIntra())
-        {
-            if (!slice->isIntra() && slice->getPPS()->getCabacInitPresentFlag())
-            {
-                SliceType sliceType   = slice->getSliceType();
-                int  encCABACTableIdx = slice->getPPS()->getEncCABACTableIdx();
-                bool encCabacInitFlag = (sliceType != encCABACTableIdx && encCABACTableIdx != I_SLICE) ? true : false;
-                slice->setCabacInitFlag(encCabacInitFlag);
-                WRITE_FLAG(encCabacInitFlag ? 1 : 0, "cabac_init_flag");
-            }
-        }
-
-        if (slice->getEnableTMVPFlag())
-        {
-            if (slice->getSliceType() == B_SLICE)
-            {
-                WRITE_FLAG(slice->getColFromL0Flag(), "collocated_from_l0_flag");
-            }
-
-            if (slice->getSliceType() != I_SLICE &&
-                ((slice->getColFromL0Flag() == 1 && slice->getNumRefIdx(REF_PIC_LIST_0) > 1) ||
-                 (slice->getColFromL0Flag() == 0  && slice->getNumRefIdx(REF_PIC_LIST_1) > 1)))
-            {
-                WRITE_UVLC(slice->getColRefIdx(), "collocated_ref_idx");
-            }
-        }
-        if ((slice->getPPS()->getUseWP() && slice->getSliceType() == P_SLICE) || (slice->getPPS()->getWPBiPred() && slice->getSliceType() == B_SLICE))
-        {
-            xCodePredWeightTable(slice);
-        }
-        assert(slice->getMaxNumMergeCand() <= MRG_MAX_NUM_CANDS);
-        if (!slice->isIntra())
-        {
-            WRITE_UVLC(MRG_MAX_NUM_CANDS - slice->getMaxNumMergeCand(), "five_minus_max_num_merge_cand");
-        }
-        int code = slice->getSliceQp() - (slice->getPPS()->getPicInitQPMinus26() + 26);
-        WRITE_SVLC(code, "slice_qp_delta");
-        if (slice->getPPS()->getSliceChromaQpFlag())
-        {
-            code = slice->getSliceQpDeltaCb();
-            WRITE_SVLC(code, "slice_qp_delta_cb");
-            code = slice->getSliceQpDeltaCr();
-            WRITE_SVLC(code, "slice_qp_delta_cr");
-        }
-        if (slice->getPPS()->getDeblockingFilterControlPresentFlag())
-        {
-            if (slice->getPPS()->getDeblockingFilterOverrideEnabledFlag())
-            {
-                WRITE_FLAG(slice->getDeblockingFilterOverrideFlag(), "deblocking_filter_override_flag");
-            }
-            if (slice->getDeblockingFilterOverrideFlag())
-            {
-                WRITE_FLAG(slice->getDeblockingFilterDisable(), "slice_disable_deblocking_filter_flag");
-                if (!slice->getDeblockingFilterDisable())
-                {
-                    WRITE_SVLC(slice->getDeblockingFilterBetaOffsetDiv2(), "slice_beta_offset_div2");
-                    WRITE_SVLC(slice->getDeblockingFilterTcOffsetDiv2(),   "slice_tc_offset_div2");
-                }
-            }
-        }
-
-        bool isSAOEnabled = (!slice->getSPS()->getUseSAO()) ? (false) : (slice->getSaoEnabledFlag() || slice->getSaoEnabledFlagChroma());
-        bool isDBFEnabled = (!slice->getDeblockingFilterDisable());
-
-        if (isSAOEnabled || isDBFEnabled)
-        {
-            WRITE_FLAG(1, "slice_loop_filter_across_slices_enabled_flag");
-        }
-    }
-    if (slice->getPPS()->getSliceHeaderExtensionPresentFlag())
-    {
-        WRITE_UVLC(0, "slice_header_extension_length");
-    }
-}
-
-/**
- - write wavefront substreams sizes for the slice header.
- .
- \param slice Where we find the substream size information.
- */
-void  TEncSbac::codeTilesWPPEntryPoint(TComSlice* slice)
-{
-    if (!slice->getPPS()->getEntropyCodingSyncEnabledFlag())
-    {
-        return;
-    }
-    uint32_t numEntryPointOffsets = 0, offsetLenMinus1 = 0, maxOffset = 0;
-    uint32_t *entryPointOffset = NULL;
-    if (slice->getPPS()->getEntropyCodingSyncEnabledFlag())
-    {
-        uint32_t* substreamSizes          = slice->getSubstreamSizes();
-        int maxNumParts                   = slice->getPic()->getNumPartInCU();
-        int numZeroSubstreamsAtEndOfSlice = slice->getPic()->getFrameHeightInCU() - 1 - ((slice->getSliceCurEndCUAddr() - 1) / maxNumParts / slice->getPic()->getFrameWidthInCU());
-        numEntryPointOffsets              = slice->getPic()->getFrameHeightInCU() - numZeroSubstreamsAtEndOfSlice - 1;
-        slice->setNumEntryPointOffsets(numEntryPointOffsets);
-        entryPointOffset = new uint32_t[numEntryPointOffsets];
-        for (int idx = 0; idx < numEntryPointOffsets; idx++)
-        {
-            entryPointOffset[idx] = (substreamSizes[idx] >> 3);
-            if (entryPointOffset[idx] > maxOffset)
-            {
-                maxOffset = entryPointOffset[idx];
-            }
-        }
-    }
-    // Determine number of bits "offsetLenMinus1+1" required for entry point information
-    offsetLenMinus1 = 0;
-    while (maxOffset >= (1u << (offsetLenMinus1 + 1)))
-    {
-        offsetLenMinus1++;
-        assert(offsetLenMinus1 + 1 < 32);
-    }
-
-    WRITE_UVLC(numEntryPointOffsets, "num_entry_point_offsets");
-    if (numEntryPointOffsets > 0)
-    {
-        WRITE_UVLC(offsetLenMinus1, "offset_len_minus1");
-    }
-
-    for (uint32_t idx = 0; idx < numEntryPointOffsets; idx++)
-    {
-        WRITE_CODE(entryPointOffset[idx] - 1, offsetLenMinus1 + 1, "entry_point_offset_minus1");
-    }
-
-    delete [] entryPointOffset;
-}
-
-void TEncSbac::codeTerminatingBit(uint32_t lsLast)
-{
-    m_binIf->encodeBinTrm(lsLast);
-}
-
-void TEncSbac::codeSliceFinish()
-{
-    m_binIf->finish();
-}
-
-void TEncSbac::xWriteUnaryMaxSymbol(uint32_t symbol, ContextModel* scmModel, int offset, uint32_t maxSymbol)
-{
-    assert(maxSymbol > 0);
-
-    m_binIf->encodeBin(symbol ? 1 : 0, scmModel[0]);
-
-    if (symbol == 0)
-    {
-        return;
-    }
-
-    bool bCodeLast = (maxSymbol > symbol);
-
-    while (--symbol)
-    {
-        m_binIf->encodeBin(1, scmModel[offset]);
-    }
-
-    if (bCodeLast)
-    {
-        m_binIf->encodeBin(0, scmModel[offset]);
-    }
-}
-
-void TEncSbac::xWriteEpExGolomb(uint32_t symbol, uint32_t count)
-{
-    uint32_t bins = 0;
-    int numBins = 0;
-
-    while (symbol >= (uint32_t)(1 << count))
-    {
-        bins = 2 * bins + 1;
-        numBins++;
-        symbol -= 1 << count;
-        count++;
-    }
-
-    bins = 2 * bins + 0;
-    numBins++;
-
-    bins = (bins << count) | symbol;
-    numBins += count;
-
-    assert(numBins <= 32);
-    m_binIf->encodeBinsEP(bins, numBins);
-}
-
-/** Coding of coeff_abs_level_minus3
- * \param symbol value of coeff_abs_level_minus3
- * \param ruiGoRiceParam reference to Rice parameter
- * \returns void
- */
-void TEncSbac::xWriteCoefRemainExGolomb(uint32_t codeNumber, const uint32_t absGoRice)
-{
-    uint32_t length;
-    const uint32_t codeRemain = codeNumber & ((1 << absGoRice) - 1);
-
-    if ((codeNumber >> absGoRice) < COEF_REMAIN_BIN_REDUCTION)
-    {
-        length = codeNumber >> absGoRice;
-
-        assert(codeNumber - (length << absGoRice) == (codeNumber & ((1 << absGoRice) - 1)));
-
-        assert(length + 1 + absGoRice < 32);
-        m_binIf->encodeBinsEP((((1 << (length + 1)) - 2) << absGoRice) + codeRemain, length + 1 + absGoRice);
-    }
-    else
-    {
-        length = 0;
-        codeNumber = (codeNumber >> absGoRice) - COEF_REMAIN_BIN_REDUCTION;
-        if (codeNumber != 0)
-        {
-            unsigned long idx;
-            CLZ32(idx, codeNumber + 1);
-            length = idx;
-            codeNumber -= (1 << idx) - 1;
-        }
-        codeNumber = (codeNumber << absGoRice) + codeRemain;
-
-        m_binIf->encodeBinsEP((1 << (COEF_REMAIN_BIN_REDUCTION + length + 1)) - 2, COEF_REMAIN_BIN_REDUCTION + length + 1);
-        m_binIf->encodeBinsEP(codeNumber, length + absGoRice);
-    }
-}
-
-void  TEncSbac::setBitstream(TComBitIf* p)
-{
-    m_bitIf = p;
-    // NOTE: When write header, it isn't initial
-    if (m_binIf)
-        m_binIf->init(p);
-}
-
-// SBAC RD
-void  TEncSbac::load(TEncSbac* src)
-{
-    this->xCopyFrom(src);
-}
-
-void  TEncSbac::loadIntraDirModeLuma(TEncSbac* src)
-{
-    m_binIf->copyState(src->m_binIf);
-
-    ::memcpy(&this->m_contextModels[OFF_ADI_CTX], &src->m_contextModels[OFF_ADI_CTX], sizeof(ContextModel) * NUM_ADI_CTX);
-}
-
-void  TEncSbac::store(TEncSbac* pDest)
-{
-    pDest->xCopyFrom(this);
-}
-
-void TEncSbac::xCopyFrom(TEncSbac* src)
-{
-    m_binIf->copyState(src->m_binIf);
-
-    memcpy(m_contextModels, src->m_contextModels, MAX_OFF_CTX_MOD * sizeof(ContextModel));
-}
-
-void TEncSbac::codeMVPIdx(uint32_t symbol)
-{
-    m_binIf->encodeBin(symbol, m_contextModels[OFF_MVP_IDX_CTX]);
-}
-
-void TEncSbac::codePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
-{
-    PartSize partSize = cu->getPartitionSize(absPartIdx);
-
-    if (cu->isIntra(absPartIdx))
-    {
-        if (depth == g_maxCUDepth - g_addCUDepth)
-        {
-            m_binIf->encodeBin(partSize == SIZE_2Nx2N ? 1 : 0, m_contextModels[OFF_PART_SIZE_CTX]);
-        }
-        return;
-    }
-
-    switch (partSize)
-    {
-    case SIZE_2Nx2N:
-        m_binIf->encodeBin(1, m_contextModels[OFF_PART_SIZE_CTX]);
-        break;
-
-    case SIZE_2NxN:
-    case SIZE_2NxnU:
-    case SIZE_2NxnD:
-        m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 0]);
-        m_binIf->encodeBin(1, m_contextModels[OFF_PART_SIZE_CTX + 1]);
-        if (cu->getSlice()->getSPS()->getAMPAcc(depth))
-        {
-            m_binIf->encodeBin((partSize == SIZE_2NxN) ? 1 : 0, m_contextModels[OFF_PART_SIZE_CTX + 3]);
-            if (partSize != SIZE_2NxN)
-            {
-                m_binIf->encodeBinEP((partSize == SIZE_2NxnU ? 0 : 1));
-            }
-        }
-        break;
-
-    case SIZE_Nx2N:
-    case SIZE_nLx2N:
-    case SIZE_nRx2N:
-        m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 0]);
-        m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 1]);
-        if (depth == g_maxCUDepth - g_addCUDepth && !(cu->getCUSize(absPartIdx) == 8))
-        {
-            m_binIf->encodeBin(1, m_contextModels[OFF_PART_SIZE_CTX + 2]);
-        }
-        if (cu->getSlice()->getSPS()->getAMPAcc(depth))
-        {
-            m_binIf->encodeBin((partSize == SIZE_Nx2N) ? 1 : 0, m_contextModels[OFF_PART_SIZE_CTX + 3]);
-            if (partSize != SIZE_Nx2N)
-            {
-                m_binIf->encodeBinEP((partSize == SIZE_nLx2N ? 0 : 1));
-            }
-        }
-        break;
-
-    case SIZE_NxN:
-        if (depth == g_maxCUDepth - g_addCUDepth && !(cu->getCUSize(absPartIdx) == 8))
-        {
-            m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 0]);
-            m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 1]);
-            m_binIf->encodeBin(0, m_contextModels[OFF_PART_SIZE_CTX + 2]);
-        }
-        break;
-
-    default:
-        assert(0);
-    }
-}
-
-/** code prediction mode
- * \param cu
- * \param absPartIdx
- * \returns void
- */
-void TEncSbac::codePredMode(TComDataCU* cu, uint32_t absPartIdx)
-{
-    // get context function is here
-    int predMode = cu->getPredictionMode(absPartIdx);
-
-    m_binIf->encodeBin(predMode == MODE_INTER ? 0 : 1, m_contextModels[OFF_PRED_MODE_CTX]);
-}
-
-void TEncSbac::codeCUTransquantBypassFlag(TComDataCU* cu, uint32_t absPartIdx)
-{
-    uint32_t symbol = cu->getCUTransquantBypass(absPartIdx);
-
-    m_binIf->encodeBin(symbol, m_contextModels[OFF_CU_TRANSQUANT_BYPASS_FLAG_CTX]);
-}
-
-/** code skip flag
- * \param cu
- * \param absPartIdx
- * \returns void
- */
-void TEncSbac::codeSkipFlag(TComDataCU* cu, uint32_t absPartIdx)
-{
-    // get context function is here
-    uint32_t symbol = cu->isSkipped(absPartIdx) ? 1 : 0;
-    uint32_t ctxSkip = cu->getCtxSkipFlag(absPartIdx);
-
-    m_binIf->encodeBin(symbol, m_contextModels[OFF_SKIP_FLAG_CTX + ctxSkip]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++);
-    DTRACE_CABAC_T("\tSkipFlag");
-    DTRACE_CABAC_T("\tuiCtxSkip: ");
-    DTRACE_CABAC_V(ctxSkip);
-    DTRACE_CABAC_T("\tuiSymbol: ");
-    DTRACE_CABAC_V(symbol);
-    DTRACE_CABAC_T("\n");
-}
-
-/** code merge flag
- * \param cu
- * \param absPartIdx
- * \returns void
- */
-void TEncSbac::codeMergeFlag(TComDataCU* cu, uint32_t absPartIdx)
-{
-    const uint32_t symbol = cu->getMergeFlag(absPartIdx) ? 1 : 0;
-
-    m_binIf->encodeBin(symbol, m_contextModels[OFF_MERGE_FLAG_EXT_CTX]);
-
-    DTRACE_CABAC_VL(g_nSymbolCounter++);
-    DTRACE_CABAC_T("\tMergeFlag: ");
-    DTRACE_CABAC_V(symbol);
-    DTRACE_CABAC_T("\tAddress: ");
-    DTRACE_CABAC_V(cu->getAddr());
-    DTRACE_CABAC_T("\tuiAbsPartIdx: ");
-    DTRACE_CABAC_V(absPartIdx);
-    DTRACE_CABAC_T("\n");
-}
-
-/** code merge index
- * \param cu
- * \param absPartIdx
- * \returns void
- */
-void TEncSbac::codeMergeIndex(TComDataCU* cu, uint32_t absPartIdx)
-{
-    uint32_t numCand = cu->getSlice()->getMaxNumMergeCand();
-
-    if (numCand > 1)
-    {
-        uint32_t unaryIdx = cu->getMergeIndex(absPartIdx);
-        m_binIf->encodeBin((unaryIdx != 0), m_contextModels[OFF_MERGE_IDX_EXT_CTX]);
-
-        assert(unaryIdx < numCand);
-
-        if (unaryIdx != 0)
-        {
-            uint32_t mask = (1 << unaryIdx) - 2;
-            mask >>= (unaryIdx == numCand - 1) ? 1 : 0;
-            m_binIf->encodeBinsEP(mask, unaryIdx - (unaryIdx == numCand - 1));
-        }
-    }
-    DTRACE_CABAC_VL(g_nSymbolCounter++);
-    DTRACE_CABAC_T("\tparseMergeIndex()");
-    DTRACE_CABAC_T("\tuiMRGIdx= ");
-    DTRACE_CABAC_V(cu->getMergeIndex(absPartIdx));
-    DTRACE_CABAC_T("\n");
-}
-
-void TEncSbac::codeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
-{
-    if (depth == g_maxCUDepth - g_addCUDepth)
-        return;
-
-    uint32_t ctx           = cu->getCtxSplitFlag(absPartIdx, depth);
-    uint32_t currSplitFlag = (cu->getDepth(absPartIdx) > depth) ? 1 : 0;
-
-    assert(ctx < 3);
-    m_binIf->encodeBin(currSplitFlag, m_contextModels[OFF_SPLIT_FLAG_CTX + ctx]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tSplitFlag\n")
-}
-
-void TEncSbac::codeTransformSubdivFlag(uint32_t symbol, uint32_t ctx)
-{
-    m_binIf->encodeBin(symbol, m_contextModels[OFF_TRANS_SUBDIV_FLAG_CTX + ctx]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tparseTransformSubdivFlag()")
-    DTRACE_CABAC_T("\tsymbol=")
-    DTRACE_CABAC_V(symbol)
-    DTRACE_CABAC_T("\tctx=")
-    DTRACE_CABAC_V(ctx)
-    DTRACE_CABAC_T("\n")
-}
-
-void TEncSbac::codeIntraDirLumaAng(TComDataCU* cu, uint32_t absPartIdx, bool isMultiple)
-{
-    uint32_t dir[4], j;
-    int preds[4][3];
-    int predIdx[4];
-    PartSize mode = cu->getPartitionSize(absPartIdx);
-    uint32_t partNum = isMultiple ? (mode == SIZE_NxN ? 4 : 1) : 1;
-    uint32_t partOffset = (cu->getPic()->getNumPartInCU() >> (cu->getDepth(absPartIdx) << 1)) >> 2;
-
-    for (j = 0; j < partNum; j++)
-    {
-        dir[j] = cu->getLumaIntraDir(absPartIdx + partOffset * j);
-        cu->getIntraDirLumaPredictor(absPartIdx + partOffset * j, preds[j]);
-        predIdx[j] = -1;
-        for (uint32_t i = 0; i < 3; i++)
-        {
-            if (dir[j] == preds[j][i])
-            {
-                predIdx[j] = i;
-            }
-        }
-
-        m_binIf->encodeBin((predIdx[j] != -1) ? 1 : 0, m_contextModels[OFF_ADI_CTX]);
-    }
-
-    for (j = 0; j < partNum; j++)
-    {
-        if (predIdx[j] != -1)
-        {
-            assert((predIdx[j] >= 0) && (predIdx[j] <= 2));
-            // NOTE: Mapping
-            //       0 = 0
-            //       1 = 10
-            //       2 = 11
-            int nonzero = (!!predIdx[j]);
-            m_binIf->encodeBinsEP(predIdx[j] + nonzero, 1 + nonzero);
-        }
-        else
-        {
-            if (preds[j][0] > preds[j][1])
-            {
-                std::swap(preds[j][0], preds[j][1]);
-            }
-            if (preds[j][0] > preds[j][2])
-            {
-                std::swap(preds[j][0], preds[j][2]);
-            }
-            if (preds[j][1] > preds[j][2])
-            {
-                std::swap(preds[j][1], preds[j][2]);
-            }
-            dir[j] += (dir[j] > preds[j][2]) ? -1 : 0;
-            dir[j] += (dir[j] > preds[j][1]) ? -1 : 0;
-            dir[j] += (dir[j] > preds[j][0]) ? -1 : 0;
-
-            m_binIf->encodeBinsEP(dir[j], 5);
-        }
-    }
-}
-
-void TEncSbac::codeIntraDirChroma(TComDataCU* cu, uint32_t absPartIdx)
-{
-    uint32_t intraDirChroma = cu->getChromaIntraDir(absPartIdx);
-
-    if (intraDirChroma == DM_CHROMA_IDX)
-    {
-        m_binIf->encodeBin(0, m_contextModels[OFF_CHROMA_PRED_CTX]);
-    }
-    else
-    {
-        uint32_t allowedChromaDir[NUM_CHROMA_MODE];
-        cu->getAllowedChromaDir(absPartIdx, allowedChromaDir);
-
-        for (int i = 0; i < NUM_CHROMA_MODE - 1; i++)
-        {
-            if (intraDirChroma == allowedChromaDir[i])
-            {
-                intraDirChroma = i;
-                break;
-            }
-        }
-
-        m_binIf->encodeBin(1, m_contextModels[OFF_CHROMA_PRED_CTX]);
-
-        m_binIf->encodeBinsEP(intraDirChroma, 2);
-    }
-}
-
-void TEncSbac::codeInterDir(TComDataCU* cu, uint32_t absPartIdx)
-{
-    const uint32_t interDir = cu->getInterDir(absPartIdx) - 1;
-    const uint32_t ctx      = cu->getCtxInterDir(absPartIdx);
-
-    if (cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N || cu->getCUSize(absPartIdx) != 8)
-    {
-        m_binIf->encodeBin(interDir == 2 ? 1 : 0, m_contextModels[OFF_INTER_DIR_CTX + ctx]);
-    }
-    if (interDir < 2)
-    {
-        m_binIf->encodeBin(interDir, m_contextModels[OFF_INTER_DIR_CTX + 4]);
-    }
-}
-
-void TEncSbac::codeRefFrmIdx(TComDataCU* cu, uint32_t absPartIdx, int list)
-{
-    int refFrame = cu->getCUMvField(list)->getRefIdx(absPartIdx);
-
-    m_binIf->encodeBin(refFrame > 0, m_contextModels[OFF_REF_NO_CTX]);
-
-    if (refFrame > 0)
-    {
-        uint32_t refNum = cu->getSlice()->getNumRefIdx(list) - 2;
-        if (refNum == 0)
-        {
-            return;
-        }
-        refFrame--;
-        m_binIf->encodeBin(refFrame > 0, m_contextModels[OFF_REF_NO_CTX + 1]);
-        if (refFrame > 0)
-        {
-            uint32_t mask = (1 << refFrame) - 2;
-            mask >>= (refFrame == refNum) ? 1 : 0;
-            m_binIf->encodeBinsEP(mask, refFrame - (refFrame == refNum));
-        }
-    }
-}
-
-void TEncSbac::codeMvd(TComDataCU* cu, uint32_t absPartIdx, int list)
-{
-    if (list == REF_PIC_LIST_1 && cu->getSlice()->getMvdL1ZeroFlag() && cu->getInterDir(absPartIdx) == 3)
-    {
-        return;
-    }
-
-    const TComCUMvField* cuMvField = cu->getCUMvField(list);
-    const int hor = cuMvField->getMvd(absPartIdx).x;
-    const int ver = cuMvField->getMvd(absPartIdx).y;
-
-    m_binIf->encodeBin(hor != 0 ? 1 : 0, m_contextModels[OFF_MV_RES_CTX]);
-    m_binIf->encodeBin(ver != 0 ? 1 : 0, m_contextModels[OFF_MV_RES_CTX]);
-
-    const bool bHorAbsGr0 = hor != 0;
-    const bool bVerAbsGr0 = ver != 0;
-    const uint32_t horAbs   = 0 > hor ? -hor : hor;
-    const uint32_t verAbs   = 0 > ver ? -ver : ver;
-
-    if (bHorAbsGr0)
-    {
-        m_binIf->encodeBin(horAbs > 1 ? 1 : 0, m_contextModels[OFF_MV_RES_CTX + 1]);
-    }
-
-    if (bVerAbsGr0)
-    {
-        m_binIf->encodeBin(verAbs > 1 ? 1 : 0, m_contextModels[OFF_MV_RES_CTX + 1]);
-    }
-
-    if (bHorAbsGr0)
-    {
-        if (horAbs > 1)
-        {
-            xWriteEpExGolomb(horAbs - 2, 1);
-        }
-
-        m_binIf->encodeBinEP(0 > hor ? 1 : 0);
-    }
-
-    if (bVerAbsGr0)
-    {
-        if (verAbs > 1)
-        {
-            xWriteEpExGolomb(verAbs - 2, 1);
-        }
-
-        m_binIf->encodeBinEP(0 > ver ? 1 : 0);
-    }
-}
-
-void TEncSbac::codeDeltaQP(TComDataCU* cu, uint32_t absPartIdx)
-{
-    int dqp = cu->getQP(absPartIdx) - cu->getRefQP(absPartIdx);
-
-    int qpBdOffsetY = QP_BD_OFFSET;
-
-    dqp = (dqp + 78 + qpBdOffsetY + (qpBdOffsetY / 2)) % (52 + qpBdOffsetY) - 26 - (qpBdOffsetY / 2);
-
-    uint32_t absDQp = (uint32_t)((dqp > 0) ? dqp  : (-dqp));
-    uint32_t TUValue = X265_MIN((int)absDQp, CU_DQP_TU_CMAX);
-    xWriteUnaryMaxSymbol(TUValue, &m_contextModels[OFF_DELTA_QP_CTX], 1, CU_DQP_TU_CMAX);
-    if (absDQp >= CU_DQP_TU_CMAX)
-    {
-        xWriteEpExGolomb(absDQp - CU_DQP_TU_CMAX, CU_DQP_EG_k);
-    }
-
-    if (absDQp > 0)
-    {
-        uint32_t sign = (dqp > 0 ? 0 : 1);
-        m_binIf->encodeBinEP(sign);
-    }
-}
-
-void TEncSbac::codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth)
-{
-    uint32_t cbf = cu->getCbf(absPartIdx, ttype, trDepth);
-    uint32_t ctx = cu->getCtxQtCbf(ttype, trDepth);
-
-    m_binIf->encodeBin(cbf, m_contextModels[OFF_QT_CBF_CTX + ctx]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tparseQtCbf()")
-    DTRACE_CABAC_T("\tsymbol=")
-    DTRACE_CABAC_V(cbf)
-    DTRACE_CABAC_T("\tctx=")
-    DTRACE_CABAC_V(ctx)
-    DTRACE_CABAC_T("\tetype=")
-    DTRACE_CABAC_V(ttype)
-    DTRACE_CABAC_T("\tuiAbsPartIdx=")
-    DTRACE_CABAC_V(absPartIdx)
-    DTRACE_CABAC_T("\n")
-}
-
-void TEncSbac::codeTransformSkipFlags(TComDataCU* cu, uint32_t absPartIdx, uint32_t trSize, TextType ttype)
-{
-    if (cu->getCUTransquantBypass(absPartIdx))
-    {
-        return;
-    }
-    if (trSize != 4)
-    {
-        return;
-    }
-
-    uint32_t useTransformSkip = cu->getTransformSkip(absPartIdx, ttype);
-    m_binIf->encodeBin(useTransformSkip, m_contextModels[OFF_TRANSFORMSKIP_FLAG_CTX + (ttype ? NUM_TRANSFORMSKIP_FLAG_CTX : 0)]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tparseTransformSkip()");
-    DTRACE_CABAC_T("\tsymbol=")
-    DTRACE_CABAC_V(useTransformSkip)
-    DTRACE_CABAC_T("\tAddr=")
-    DTRACE_CABAC_V(cu->getAddr())
-    DTRACE_CABAC_T("\tetype=")
-    DTRACE_CABAC_V(ttype)
-    DTRACE_CABAC_T("\tuiAbsPartIdx=")
-    DTRACE_CABAC_V(absPartIdx)
-    DTRACE_CABAC_T("\n")
-}
-
-/** Code I_PCM information.
- * \param cu pointer to CU
- * \param absPartIdx CU index
- * \returns void
- */
-void TEncSbac::codeIPCMInfo(TComDataCU* cu, uint32_t absPartIdx)
-{
-    uint32_t ipcm = (cu->getIPCMFlag(absPartIdx) == true) ? 1 : 0;
-
-    bool writePCMSampleFlag = cu->getIPCMFlag(absPartIdx);
-
-    m_binIf->encodeBinTrm(ipcm);
-
-    if (writePCMSampleFlag)
-    {
-        m_binIf->encodePCMAlignBits();
-
-        uint32_t minCoeffSize = cu->getPic()->getMinCUSize() * cu->getPic()->getMinCUSize();
-        uint32_t lumaOffset   = minCoeffSize * absPartIdx;
-        uint32_t chromaOffset = lumaOffset >> 2;
-        uint32_t width;
-        uint32_t height;
-        uint32_t sampleBits;
-        uint32_t x, y;
-
-        pixel *pcmSample = cu->getPCMSampleY() + lumaOffset;
-        width = cu->getCUSize(absPartIdx);
-        height = cu->getCUSize(absPartIdx);
-        sampleBits = cu->getSlice()->getSPS()->getPCMBitDepthLuma();
-
-        for (y = 0; y < height; y++)
-        {
-            for (x = 0; x < width; x++)
-            {
-                uint32_t sample = pcmSample[x];
-
-                m_binIf->xWritePCMCode(sample, sampleBits);
-            }
-
-            pcmSample += width;
-        }
-
-        pcmSample = cu->getPCMSampleCb() + chromaOffset;
-        width = cu->getCUSize(absPartIdx) / 2;
-        height = cu->getCUSize(absPartIdx) / 2;
-        sampleBits = cu->getSlice()->getSPS()->getPCMBitDepthChroma();
-
-        for (y = 0; y < height; y++)
-        {
-            for (x = 0; x < width; x++)
-            {
-                uint32_t sample = pcmSample[x];
-
-                m_binIf->xWritePCMCode(sample, sampleBits);
-            }
-
-            pcmSample += width;
-        }
-
-        pcmSample = cu->getPCMSampleCr() + chromaOffset;
-
-        for (y = 0; y < height; y++)
-        {
-            for (x = 0; x < width; x++)
-            {
-                uint32_t sample = pcmSample[x];
-
-                m_binIf->xWritePCMCode(sample, sampleBits);
-            }
-
-            pcmSample += width;
-        }
-
-        m_binIf->resetBac();
-    }
-}
-
-void TEncSbac::codeQtRootCbf(TComDataCU* cu, uint32_t absPartIdx)
-{
-    uint32_t cbf = cu->getQtRootCbf(absPartIdx);
-    uint32_t ctx = 0;
-
-    m_binIf->encodeBin(cbf, m_contextModels[OFF_QT_ROOT_CBF_CTX + ctx]);
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tparseQtRootCbf()")
-    DTRACE_CABAC_T("\tsymbol=")
-    DTRACE_CABAC_V(cbf)
-    DTRACE_CABAC_T("\tctx=")
-    DTRACE_CABAC_V(ctx)
-    DTRACE_CABAC_T("\tuiAbsPartIdx=")
-    DTRACE_CABAC_V(absPartIdx)
-    DTRACE_CABAC_T("\n")
-}
-
-void TEncSbac::codeQtCbfZero(TComDataCU* cu, TextType ttype, uint32_t trDepth)
-{
-    // this function is only used to estimate the bits when cbf is 0
-    // and will never be called when writing the bistream. do not need to write log
-    uint32_t cbf = 0;
-    uint32_t ctx = cu->getCtxQtCbf(ttype, trDepth);
-
-    m_binIf->encodeBin(cbf, m_contextModels[OFF_QT_CBF_CTX + ctx]);
-}
-
-void TEncSbac::codeQtRootCbfZero(TComDataCU*)
-{
-    // this function is only used to estimate the bits when cbf is 0
-    // and will never be called when writing the bistream. do not need to write log
-    uint32_t cbf = 0;
-    uint32_t ctx = 0;
-
-    m_binIf->encodeBin(cbf, m_contextModels[OFF_QT_ROOT_CBF_CTX + ctx]);
-}
-
-/** Encode (X,Y) position of the last significant coefficient
- * \param posx X component of last coefficient
- * \param posy Y component of last coefficient
- * \param width  Block width
- * \param height Block height
- * \param ttype plane type / luminance or chrominance
- * \param uiScanIdx scan type (zig-zag, hor, ver)
- * This method encodes the X and Y component within a block of the last significant coefficient.
- */
-void TEncSbac::codeLastSignificantXY(uint32_t posx, uint32_t posy, uint32_t log2TrSize, TextType ttype, uint32_t scanIdx)
-{
-    assert((ttype == TEXT_LUMA) || (ttype == TEXT_CHROMA));
-    // swap
-    if (scanIdx == SCAN_VER)
-    {
-        std::swap(posx, posy);
-    }
-
-    uint32_t ctxLast;
-    uint32_t groupIdxX = g_groupIdx[posx];
-    uint32_t groupIdxY = g_groupIdx[posy];
-
-    int blkSizeOffset = ttype ? NUM_CTX_LAST_FLAG_XY_LUMA : ((log2TrSize - 2) * 3 + ((log2TrSize - 1) >> 2));
-    int ctxShift = ttype ? log2TrSize - 2 : ((log2TrSize + 1) >> 2);
-    uint32_t maxGroupIdx = log2TrSize * 2 - 1;
-    // posX
-    ContextModel *ctxX = &m_contextModels[OFF_CTX_LAST_FLAG_X];
-    for (ctxLast = 0; ctxLast < groupIdxX; ctxLast++)
-    {
-        m_binIf->encodeBin(1, *(ctxX + blkSizeOffset + (ctxLast >> ctxShift)));
-    }
-    if (groupIdxX < maxGroupIdx)
-    {
-        m_binIf->encodeBin(0, *(ctxX + blkSizeOffset + (ctxLast >> ctxShift)));
-    }
-
-    // posY
-    ContextModel *ctxY = &m_contextModels[OFF_CTX_LAST_FLAG_Y];
-    for (ctxLast = 0; ctxLast < groupIdxY; ctxLast++)
-    {
-        m_binIf->encodeBin(1, *(ctxY + blkSizeOffset + (ctxLast >> ctxShift)));
-    }
-    if (groupIdxY < maxGroupIdx)
-    {
-        m_binIf->encodeBin(0, *(ctxY + blkSizeOffset + (ctxLast >> ctxShift)));
-    }
-
-    if (groupIdxX > 3)
-    {
-        uint32_t count = (groupIdxX - 2) >> 1;
-        posx = posx - g_minInGroup[groupIdxX];
-        m_binIf->encodeBinsEP(posx, count);
-    }
-    if (groupIdxY > 3)
-    {
-        uint32_t count = (groupIdxY - 2) >> 1;
-        posy = posy - g_minInGroup[groupIdxY];
-        m_binIf->encodeBinsEP(posy, count);
-    }
-}
-
-void TEncSbac::codeCoeffNxN(TComDataCU* cu, TCoeff* coeff, uint32_t absPartIdx, uint32_t trSize, uint32_t depth, TextType ttype)
-{
-#if ENC_DEC_TRACE
-    DTRACE_CABAC_VL(g_nSymbolCounter++)
-    DTRACE_CABAC_T("\tparseCoeffNxN()\teType=")
-    DTRACE_CABAC_V(ttype)
-    DTRACE_CABAC_T("\twidth=")
-    DTRACE_CABAC_V(trSize)
-    DTRACE_CABAC_T("\theight=")
-    DTRACE_CABAC_V(trSize)
-    DTRACE_CABAC_T("\tdepth=")
-    DTRACE_CABAC_V(depth)
-    DTRACE_CABAC_T("\tabspartidx=")
-    DTRACE_CABAC_V(absPartIdx)
-    DTRACE_CABAC_T("\ttoCU-X=")
-    DTRACE_CABAC_V(cu->getCUPelX())
-    DTRACE_CABAC_T("\ttoCU-Y=")
-    DTRACE_CABAC_V(cu->getCUPelY())
-    DTRACE_CABAC_T("\tCU-addr=")
-    DTRACE_CABAC_V(cu->getAddr())
-    DTRACE_CABAC_T("\tinCU-X=")
-    DTRACE_CABAC_V(g_rasterToPelX[g_zscanToRaster[absPartIdx]])
-    DTRACE_CABAC_T("\tinCU-Y=")
-    DTRACE_CABAC_V(g_rasterToPelY[g_zscanToRaster[absPartIdx]])
-    DTRACE_CABAC_T("\tpredmode=")
-    DTRACE_CABAC_V(cu->getPredictionMode(absPartIdx))
-    DTRACE_CABAC_T("\n")
-#else // if ENC_DEC_TRACE
-    (void)depth;
-#endif // if ENC_DEC_TRACE
-
-    assert(trSize <= m_slice->getSPS()->getMaxTrSize());
-
-    // compute number of significant coefficients
-    uint32_t numSig = primitives.count_nonzero(coeff, trSize * trSize);
-
-    if (numSig == 0)
-        return;
-
-    bool beValid;
-    if (cu->getCUTransquantBypass(absPartIdx))
-    {
-        beValid = false;
-    }
-    else
-    {
-        beValid = cu->getSlice()->getPPS()->getSignHideFlag() > 0;
-    }
-    if (cu->getSlice()->getPPS()->getUseTransformSkip())
-    {
-        codeTransformSkipFlags(cu, absPartIdx, trSize, ttype);
-    }
-
-    ttype = ttype == TEXT_LUMA ? TEXT_LUMA : TEXT_CHROMA;
-    const uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-
-    //select scans
-    TUEntropyCodingParameters codingParameters;
-    TComTrQuant::getTUEntropyCodingParameters(cu, codingParameters, absPartIdx, log2TrSize, ttype);
-
-    //----- encode significance map -----
-
-    // Find position of last coefficient
-    int scanPosLast = -1;
-    int posLast;
-    uint64_t sigCoeffGroupFlag64 = 0;
-    const uint32_t maskPosXY = (1 << (log2TrSize - MLS_CG_LOG2_SIZE)) - 1;
-    do
-    {
-        posLast = codingParameters.scan[++scanPosLast];
-        if (coeff[posLast] != 0)
-        {
-            // get L1 sig map
-            // NOTE: the new algorithm is complicated, so I keep reference code here
-            //uint32_t posy   = posLast >> log2TrSize;
-            //uint32_t posx   = posLast - (posy << log2TrSize);
-            //uint32_t blkIdx0 = ((posy >> MLS_CG_LOG2_SIZE) << codingParameters.log2TrSizeCG) + (posx >> MLS_CG_LOG2_SIZE);
-            uint32_t blkIdx = ((posLast >> (2 * MLS_CG_LOG2_SIZE)) & ~maskPosXY) + ((posLast >> MLS_CG_LOG2_SIZE) & maskPosXY);
-            sigCoeffGroupFlag64 |= ((uint64_t)1 << blkIdx);
-
-            numSig--;
-        }
-    }
-    while (numSig > 0);
-
-    // Code position of last coefficient
-    int posLastY = posLast >> log2TrSize;
-    int posLastX = posLast - (posLastY << log2TrSize);
-    codeLastSignificantXY(posLastX, posLastY, log2TrSize, ttype, codingParameters.scanType);
-    //===== code significance flag =====
-    ContextModel * const baseCoeffGroupCtx = &m_contextModels[OFF_SIG_CG_FLAG_CTX + (ttype ? NUM_SIG_CG_FLAG_CTX : 0)];
-    ContextModel * const baseCtx = (ttype == TEXT_LUMA) ? &m_contextModels[OFF_SIG_FLAG_CTX] : &m_contextModels[OFF_SIG_FLAG_CTX + NUM_SIG_FLAG_CTX_LUMA];
-    const int lastScanSet = scanPosLast >> MLS_CG_SIZE;
-    uint32_t c1 = 1;
-    uint32_t goRiceParam  = 0;
-    int  scanPosSig       = scanPosLast;
-
-    for (int subSet = lastScanSet; subSet >= 0; subSet--)
-    {
-        int numNonZero = 0;
-        int subPos     = subSet << MLS_CG_SIZE;
-        goRiceParam    = 0;
-        int absCoeff[1 << MLS_CG_SIZE];
-        uint32_t coeffSigns = 0;
-        int lastNZPosInCG = -1;
-        int firstNZPosInCG = 1 << MLS_CG_SIZE;
-        if (scanPosSig == scanPosLast)
-        {
-            absCoeff[0] = int(abs(coeff[posLast]));
-            coeffSigns    = (coeff[posLast] < 0);
-            numNonZero    = 1;
-            lastNZPosInCG  = scanPosSig;
-            firstNZPosInCG = scanPosSig;
-            scanPosSig--;
-        }
-        // encode significant_coeffgroup_flag
-        const int cgBlkPos = codingParameters.scanCG[subSet];
-        const int cgPosY   = cgBlkPos >> codingParameters.log2TrSizeCG;
-        const int cgPosX   = cgBlkPos - (cgPosY << codingParameters.log2TrSizeCG);
-        const uint64_t cgBlkPosMask = ((uint64_t)1 << cgBlkPos);
-
-        if (subSet == lastScanSet || subSet == 0)
-        {
-            sigCoeffGroupFlag64 |= cgBlkPosMask;
-        }
-        else
-        {
-            uint32_t sigCoeffGroup = ((sigCoeffGroupFlag64 & cgBlkPosMask) != 0);
-            uint32_t ctxSig = TComTrQuant::getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
-            m_binIf->encodeBin(sigCoeffGroup, baseCoeffGroupCtx[ctxSig]);
-        }
-        // encode significant_coeff_flag
-        if (sigCoeffGroupFlag64 & cgBlkPosMask)
-        {
-            const int patternSigCtx = TComTrQuant::calcPatternSigCtx(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
-            uint32_t blkPos, sig, ctxSig;
-            for (; scanPosSig >= subPos; scanPosSig--)
-            {
-                blkPos  = codingParameters.scan[scanPosSig];
-                sig     = (coeff[blkPos] != 0);
-                if (scanPosSig > subPos || subSet == 0 || numNonZero)
-                {
-                    ctxSig  = TComTrQuant::getSigCtxInc(patternSigCtx, log2TrSize, trSize, blkPos, codingParameters);
-                    m_binIf->encodeBin(sig, baseCtx[ctxSig]);
-                }
-                if (sig)
-                {
-                    absCoeff[numNonZero] = int(abs(coeff[blkPos]));
-                    coeffSigns = 2 * coeffSigns + (coeff[blkPos] < 0);
-                    numNonZero++;
-                    if (lastNZPosInCG == -1)
-                    {
-                        lastNZPosInCG = scanPosSig;
-                    }
-                    firstNZPosInCG = scanPosSig;
-                }
-            }
-        }
-        else
-        {
-            scanPosSig = subPos - 1;
-        }
-
-        if (numNonZero > 0)
-        {
-            bool signHidden = (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD);
-            uint32_t ctxSet = (subSet > 0 && ttype == TEXT_LUMA) ? 2 : 0;
-
-            if (c1 == 0)
-            {
-                ctxSet++;
-            }
-            c1 = 1;
-            ContextModel *baseCtxMod = (ttype == TEXT_LUMA) ? &m_contextModels[OFF_ONE_FLAG_CTX + 4 * ctxSet] : &m_contextModels[OFF_ONE_FLAG_CTX + NUM_ONE_FLAG_CTX_LUMA + 4 * ctxSet];
-
-            int numC1Flag = X265_MIN(numNonZero, C1FLAG_NUMBER);
-            int firstC2FlagIdx = -1;
-            for (int idx = 0; idx < numC1Flag; idx++)
-            {
-                uint32_t symbol = absCoeff[idx] > 1;
-                m_binIf->encodeBin(symbol, baseCtxMod[c1]);
-                if (symbol)
-                {
-                    c1 = 0;
-
-                    if (firstC2FlagIdx == -1)
-                    {
-                        firstC2FlagIdx = idx;
-                    }
-                }
-                else if ((c1 < 3) && (c1 > 0))
-                {
-                    c1++;
-                }
-            }
-
-            if (c1 == 0)
-            {
-                baseCtxMod = (ttype == TEXT_LUMA) ? &m_contextModels[OFF_ABS_FLAG_CTX + ctxSet] : &m_contextModels[OFF_ABS_FLAG_CTX + NUM_ABS_FLAG_CTX_LUMA + ctxSet];
-                if (firstC2FlagIdx != -1)
-                {
-                    uint32_t symbol = absCoeff[firstC2FlagIdx] > 2;
-                    m_binIf->encodeBin(symbol, baseCtxMod[0]);
-                }
-            }
-
-            if (beValid && signHidden)
-            {
-                m_binIf->encodeBinsEP((coeffSigns >> 1), numNonZero - 1);
-            }
-            else
-            {
-                m_binIf->encodeBinsEP(coeffSigns, numNonZero);
-            }
-
-            int firstCoeff2 = 1;
-            if (c1 == 0 || numNonZero > C1FLAG_NUMBER)
-            {
-                for (int idx = 0; idx < numNonZero; idx++)
-                {
-                    uint32_t baseLevel  = (idx < C1FLAG_NUMBER) ? (2 + firstCoeff2) : 1;
-
-                    if (absCoeff[idx] >= baseLevel)
-                    {
-                        xWriteCoefRemainExGolomb(absCoeff[idx] - baseLevel, goRiceParam);
-                        if (absCoeff[idx] > 3 * (1 << goRiceParam))
-                        {
-                            goRiceParam = std::min<uint32_t>(goRiceParam + 1, 4);
-                        }
-                    }
-                    if (absCoeff[idx] >= 2)
-                    {
-                        firstCoeff2 = 0;
-                    }
-                }
-            }
-        }
-    }
-}
-
-void TEncSbac::codeSaoMaxUvlc(uint32_t code, uint32_t maxSymbol)
-{
-    assert(maxSymbol > 0);
-
-    uint32_t isCodeLast = (maxSymbol > code) ? 1 : 0;
-    uint32_t isCodeNonZero = (code != 0) ? 1 : 0;
-
-    m_binIf->encodeBinEP(isCodeNonZero);
-    if (isCodeNonZero)
-    {
-        uint32_t mask = (1 << (code - 1)) - 1;
-        uint32_t len = code - 1 + isCodeLast;
-        mask <<= isCodeLast;
-
-        m_binIf->encodeBinsEP(mask, len);
-    }
-}
-
-/** Code SAO EO class or BO band position
- * \param length
- * \param code
- */
-void TEncSbac::codeSaoUflc(uint32_t length, uint32_t code)
-{
-    m_binIf->encodeBinsEP(code, length);
-}
-
-/** Code SAO merge flags
- * \param code
- * \param uiCompIdx
- */
-void TEncSbac::codeSaoMerge(uint32_t code)
-{
-    assert((code == 0) || (code == 1));
-    m_binIf->encodeBin(code, m_contextModels[OFF_SAO_MERGE_FLAG_CTX]);
-}
-
-/** Code SAO type index
- * \param code
- */
-void TEncSbac::codeSaoTypeIdx(uint32_t code)
-{
-    m_binIf->encodeBin((code == 0) ? 0 : 1, m_contextModels[OFF_SAO_TYPE_IDX_CTX]);
-    if (code != 0)
-    {
-        m_binIf->encodeBinEP(code <= 4 ? 1 : 0);
-    }
-}
-
-/*!
- ****************************************************************************
- * \brief
- *   estimate bit cost for CBP, significant map and significant coefficients
- ****************************************************************************
- */
-void TEncSbac::estBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype)
-{
-    estCBFBit(estBitsSbac);
-
-    estSignificantCoeffGroupMapBit(estBitsSbac, ttype);
-
-    // encode significance map
-    estSignificantMapBit(estBitsSbac, trSize, ttype);
-
-    // encode significant coefficients
-    estSignificantCoefficientsBit(estBitsSbac, ttype);
-}
-
-/*!
- ****************************************************************************
- * \brief
- *    estimate bit cost for each CBP bit
- ****************************************************************************
- */
-void TEncSbac::estCBFBit(estBitsSbacStruct* estBitsSbac)
-{
-    ContextModel *ctx = &m_contextModels[OFF_QT_CBF_CTX];
-
-    for (uint32_t ctxInc = 0; ctxInc < NUM_QT_CBF_CTX; ctxInc++)
-    {
-        estBitsSbac->blockCbpBits[ctxInc][0] = sbacGetEntropyBits(ctx[ctxInc].m_state, 0);
-        estBitsSbac->blockCbpBits[ctxInc][1] = sbacGetEntropyBits(ctx[ctxInc].m_state, 1);
-    }
-
-    ctx = &m_contextModels[OFF_QT_ROOT_CBF_CTX];
-
-    for (uint32_t ctxInc = 0; ctxInc < NUM_QT_ROOT_CBF_CTX; ctxInc++)
-    {
-        estBitsSbac->blockRootCbpBits[ctxInc][0] = sbacGetEntropyBits(ctx[ctxInc].m_state, 0);
-        estBitsSbac->blockRootCbpBits[ctxInc][1] = sbacGetEntropyBits(ctx[ctxInc].m_state, 1);
-    }
-}
-
-/*!
- ****************************************************************************
- * \brief
- *    estimate SAMBAC bit cost for significant coefficient group map
- ****************************************************************************
- */
-void TEncSbac::estSignificantCoeffGroupMapBit(estBitsSbacStruct* estBitsSbac, TextType ttype)
-{
-    assert((ttype == TEXT_LUMA) || (ttype == TEXT_CHROMA));
-    int firstCtx = 0, numCtx = NUM_SIG_CG_FLAG_CTX;
-
-    for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
-    {
-        for (uint32_t bin = 0; bin < 2; bin++)
-        {
-            estBitsSbac->significantCoeffGroupBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextModels[OFF_SIG_CG_FLAG_CTX + ((ttype ? NUM_SIG_CG_FLAG_CTX : 0) + ctxIdx)].m_state, bin);
-        }
-    }
-}
-
-/*!
- ****************************************************************************
- * \brief
- *    estimate SAMBAC bit cost for significant coefficient map
- ****************************************************************************
- */
-void TEncSbac::estSignificantMapBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype)
-{
-    int firstCtx = 1, numCtx = 8;
-
-    if (trSize >= 16)
-    {
-        firstCtx = (ttype == TEXT_LUMA) ? 21 : 12;
-        numCtx = (ttype == TEXT_LUMA) ? 6 : 3;
-    }
-    else if (trSize == 8)
-    {
-        firstCtx = 9;
-        numCtx = (ttype == TEXT_LUMA) ? 12 : 3;
-    }
-
-    if (ttype == TEXT_LUMA)
-    {
-        for (uint32_t bin = 0; bin < 2; bin++)
-        {
-            estBitsSbac->significantBits[0][bin] = sbacGetEntropyBits(m_contextModels[OFF_SIG_FLAG_CTX].m_state, bin);
-        }
-
-        for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
-        {
-            for (uint32_t bin = 0; bin < 2; bin++)
-            {
-                estBitsSbac->significantBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextModels[OFF_SIG_FLAG_CTX + ctxIdx].m_state, bin);
-            }
-        }
-    }
-    else
-    {
-        for (uint32_t bin = 0; bin < 2; bin++)
-        {
-            estBitsSbac->significantBits[0][bin] = sbacGetEntropyBits(m_contextModels[OFF_SIG_FLAG_CTX + (NUM_SIG_FLAG_CTX_LUMA + 0)].m_state, bin);
-        }
-
-        for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
-        {
-            for (uint32_t bin = 0; bin < 2; bin++)
-            {
-                estBitsSbac->significantBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextModels[OFF_SIG_FLAG_CTX + (NUM_SIG_FLAG_CTX_LUMA + ctxIdx)].m_state, bin);
-            }
-        }
-    }
-    int bitsX = 0, bitsY = 0;
-
-    uint32_t log2TrSize = g_convertToBit[trSize] + 2;
-    int blkSizeOffset = ttype ? NUM_CTX_LAST_FLAG_XY_LUMA : ((log2TrSize - 2) * 3 + ((log2TrSize - 1) >> 2));
-    int ctxShift = ttype ? log2TrSize - 2 : ((log2TrSize + 1) >> 2);
-    uint32_t maxGroupIdx = log2TrSize * 2 - 1;
-
-    assert((ttype == TEXT_LUMA) || (ttype == TEXT_CHROMA));
-    int ctx;
-    const ContextModel *ctxX = &m_contextModels[OFF_CTX_LAST_FLAG_X];
-    for (ctx = 0; ctx < maxGroupIdx; ctx++)
-    {
-        int ctxOffset = blkSizeOffset + (ctx >> ctxShift);
-        estBitsSbac->lastXBits[ctx] = bitsX + sbacGetEntropyBits(ctxX[ctxOffset].m_state, 0);
-        bitsX += sbacGetEntropyBits(ctxX[ctxOffset].m_state, 1);
-    }
-    estBitsSbac->lastXBits[ctx] = bitsX;
-
-    const ContextModel *ctxY = &m_contextModels[OFF_CTX_LAST_FLAG_Y];
-    for (ctx = 0; ctx < maxGroupIdx; ctx++)
-    {
-        int ctxOffset = blkSizeOffset + (ctx >> ctxShift);
-        estBitsSbac->lastYBits[ctx] = bitsY + sbacGetEntropyBits(ctxY[ctxOffset].m_state, 0);
-        bitsY += sbacGetEntropyBits(ctxY[ctxOffset].m_state, 1);
-    }
-    estBitsSbac->lastYBits[ctx] = bitsY;
-}
-
-/*!
- ****************************************************************************
- * \brief
- *    estimate bit cost of significant coefficient
- ****************************************************************************
- */
-void TEncSbac::estSignificantCoefficientsBit(estBitsSbacStruct* estBitsSbac, TextType ttype)
-{
-    if (ttype == TEXT_LUMA)
-    {
-        ContextModel *ctxOne = &m_contextModels[OFF_ONE_FLAG_CTX];
-        ContextModel *ctxAbs = &m_contextModels[OFF_ABS_FLAG_CTX];
-
-        for (int ctxIdx = 0; ctxIdx < NUM_ONE_FLAG_CTX_LUMA; ctxIdx++)
-        {
-            estBitsSbac->greaterOneBits[ctxIdx][0] = sbacGetEntropyBits(ctxOne[ctxIdx].m_state, 0);
-            estBitsSbac->greaterOneBits[ctxIdx][1] = sbacGetEntropyBits(ctxOne[ctxIdx].m_state, 1);
-        }
-
-        for (int ctxIdx = 0; ctxIdx < NUM_ABS_FLAG_CTX_LUMA; ctxIdx++)
-        {
-            estBitsSbac->levelAbsBits[ctxIdx][0] = sbacGetEntropyBits(ctxAbs[ctxIdx].m_state, 0);
-            estBitsSbac->levelAbsBits[ctxIdx][1] = sbacGetEntropyBits(ctxAbs[ctxIdx].m_state, 1);
-        }
-    }
-    else
-    {
-        ContextModel *ctxOne = &m_contextModels[OFF_ONE_FLAG_CTX + NUM_ONE_FLAG_CTX_LUMA];
-        ContextModel *ctxAbs = &m_contextModels[OFF_ABS_FLAG_CTX + NUM_ABS_FLAG_CTX_LUMA];
-
-        for (int ctxIdx = 0; ctxIdx < NUM_ONE_FLAG_CTX_CHROMA; ctxIdx++)
-        {
-            estBitsSbac->greaterOneBits[ctxIdx][0] = sbacGetEntropyBits(ctxOne[ctxIdx].m_state, 0);
-            estBitsSbac->greaterOneBits[ctxIdx][1] = sbacGetEntropyBits(ctxOne[ctxIdx].m_state, 1);
-        }
-
-        for (int ctxIdx = 0; ctxIdx < NUM_ABS_FLAG_CTX_CHROMA; ctxIdx++)
-        {
-            estBitsSbac->levelAbsBits[ctxIdx][0] = sbacGetEntropyBits(ctxAbs[ctxIdx].m_state, 0);
-            estBitsSbac->levelAbsBits[ctxIdx][1] = sbacGetEntropyBits(ctxAbs[ctxIdx].m_state, 1);
-        }
-    }
-}
-
-/**
- - Initialize our context information from the nominated source.
- .
- \param src From where to copy context information.
- */
-void TEncSbac::xCopyContextsFrom(TEncSbac* src)
-{
-    memcpy(m_contextModels, src->m_contextModels, MAX_OFF_CTX_MOD * sizeof(m_contextModels[0]));
-}
-
-void  TEncSbac::loadContexts(TEncSbac* src)
-{
-    this->xCopyContextsFrom(src);
-}
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSbac.h Deleted

@@ -1,162 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncSbac.h
-    \brief    Context-adaptive entropy encoder class (header)
-*/
-
-#ifndef X265_TENCSBAC_H
-#define X265_TENCSBAC_H
-
-#include "TLibCommon/TComBitStream.h"
-#include "TLibCommon/ContextTables.h"
-#include "TEncEntropy.h"
-#include "TEncBinCoderCABAC.h"
-#include "SyntaxElementWriter.h"
-
-namespace x265 {
-// private namespace
-
-//! \ingroup TLibEncoder
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// SBAC encoder class
-class TEncSbac : public SyntaxElementWriter, public TEncEntropyIf
-{
-public:
-    uint64_t pad;
-    ContextModel m_contextModels[MAX_OFF_CTX_MOD];
-
-    TComSlice*    m_slice;
-    TEncBinCABAC* m_binIf;
-
-    TEncSbac();
-    virtual ~TEncSbac();
-
-    void  init(TEncBinCABAC* p)       { m_binIf = p; }
-    void  setSlice(TComSlice* p)      { m_slice = p; }
-    void  resetBits()                 { m_binIf->resetBits(); m_bitIf->resetBits(); }
-    uint32_t getNumberOfWrittenBits() { return m_binIf->getNumWrittenBits(); }
-
-
-    //  Virtual list
-    void resetEntropy();
-    void determineCabacInitIdx();
-    void setBitstream(TComBitIf* p);
-
-    // SBAC RD
-    void load(TEncSbac* src);
-    void loadIntraDirModeLuma(TEncSbac* src);
-    void store(TEncSbac* dest);
-    void loadContexts(TEncSbac* src);
-
-    void codeVPS(TComVPS* vps);
-    void codeSPS(TComSPS* sps);
-    void codePPS(TComPPS* pps);
-    void codeVUI(TComVUI* vui, TComSPS* sps);
-    void codeSliceHeader(TComSlice* slice);
-    void codePTL(TComPTL* ptl, bool profilePresentFlag, int maxNumSubLayersMinus1);
-    void codeProfileTier(ProfileTierLevel* ptl);
-    void codeHrdParameters(TComHRD* hrd, bool commonInfPresentFlag, uint32_t maxNumSubLayersMinus1);
-    void codeTilesWPPEntryPoint(TComSlice* slice);
-    void codeTerminatingBit(uint32_t lsLast);
-    void codeSliceFinish();
-    void codeSaoMaxUvlc(uint32_t code, uint32_t maxSymbol);
-    void codeSaoMerge(uint32_t code);
-    void codeSaoTypeIdx(uint32_t code);
-    void codeSaoUflc(uint32_t length, uint32_t code);
-    void codeShortTermRefPicSet(TComReferencePictureSet* pcRPS, bool calledFromSliceHeader, int idx);
-    bool findMatchingLTRP(TComSlice* slice, uint32_t *ltrpsIndex, int ltrpPOC, bool usedFlag);
-
-    void codeSAOSign(uint32_t code) { m_binIf->encodeBinEP(code); }
-    void codeScalingList(TComScalingList*);
-
-    void codeCUTransquantBypassFlag(TComDataCU* cu, uint32_t absPartIdx);
-    void codeSkipFlag(TComDataCU* cu, uint32_t absPartIdx);
-    void codeMergeFlag(TComDataCU* cu, uint32_t absPartIdx);
-    void codeMergeIndex(TComDataCU* cu, uint32_t absPartIdx);
-    void codeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
-    void codeMVPIdx(uint32_t symbol);
-
-    void codePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
-    void codePredMode(TComDataCU* cu, uint32_t absPartIdx);
-    void codeIPCMInfo(TComDataCU* cu, uint32_t absPartIdx);
-    void codeTransformSubdivFlag(uint32_t symbol, uint32_t ctx);
-    void codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth);
-    void codeQtRootCbf(TComDataCU* cu, uint32_t absPartIdx);
-    void codeQtCbfZero(TComDataCU* cu, TextType ttype, uint32_t trDepth);
-    void codeQtRootCbfZero(TComDataCU* cu);
-    void codeIntraDirLumaAng(TComDataCU* cu, uint32_t absPartIdx, bool isMultiple);
-
-    void codeIntraDirChroma(TComDataCU* cu, uint32_t absPartIdx);
-    void codeInterDir(TComDataCU* cu, uint32_t absPartIdx);
-    void codeRefFrmIdx(TComDataCU* cu, uint32_t absPartIdx, int eRefList);
-    void codeMvd(TComDataCU* cu, uint32_t absPartIdx, int eRefList);
-
-    void codeDeltaQP(TComDataCU* cu, uint32_t absPartIdx);
-
-    void codeLastSignificantXY(uint32_t posx, uint32_t posy, uint32_t log2TrSize, TextType ttype, uint32_t scanIdx);
-    void codeCoeffNxN(TComDataCU* cu, TCoeff* coef, uint32_t absPartIdx, uint32_t trSize, uint32_t depth, TextType ttype);
-    void codeTransformSkipFlags(TComDataCU* cu, uint32_t absPartIdx, uint32_t trSize, TextType ttype);
-
-    // -------------------------------------------------------------------------------------------------------------------
-    // for RD-optimizatioon
-    // -------------------------------------------------------------------------------------------------------------------
-
-    void estBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype);
-    void estCBFBit(estBitsSbacStruct* estBitsSbac);
-    void estSignificantCoeffGroupMapBit(estBitsSbacStruct* estBitsSbac, TextType ttype);
-    void estSignificantMapBit(estBitsSbacStruct* estBitsSbac, int trSize, TextType ttype);
-    void estSignificantCoefficientsBit(estBitsSbacStruct* estBitsSbac, TextType ttype);
-
-    TEncBinCABAC* getEncBinIf()  { return m_binIf; }
-
-private:
-
-    void xWriteUnaryMaxSymbol(uint32_t symbol, ContextModel* scmModel, int offset, uint32_t maxSymbol);
-    void xWriteEpExGolomb(uint32_t symbol, uint32_t count);
-    void xWriteCoefRemainExGolomb(uint32_t symbol, const uint32_t absGoRice);
-
-    void xCopyFrom(TEncSbac* src);
-    void xCopyContextsFrom(TEncSbac* src);
-    void xCodePredWeightTable(TComSlice* slice);
-    void xCodeScalingList(TComScalingList* scalingList, uint32_t sizeId, uint32_t listId);
-};
-}
-//! \}
-
-#endif // ifndef X265_TENCSBAC_H

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSearch.cpp Deleted

@@ -1,3822 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncSearch.cpp
- \brief    encoder search class
- */
-
-#include "TLibCommon/TypeDef.h"
-#include "TLibCommon/TComRom.h"
-#include "TLibCommon/TComMotionInfo.h"
-#include "TEncSearch.h"
-#include "encoder.h"
-
-#include "common.h"
-#include "primitives.h"
-#include "PPA/ppa.h"
-
-using namespace x265;
-
-//! \ingroup TLibEncoder
-//! \{
-
-TEncSearch::TEncSearch()
-{
-    m_qtTempCoeffY  = NULL;
-    m_qtTempCoeffCb = NULL;
-    m_qtTempCoeffCr = NULL;
-    m_qtTempTrIdx = NULL;
-    m_qtTempShortYuv = NULL;
-    m_qtTempTUCoeffY  = NULL;
-    m_qtTempTUCoeffCb = NULL;
-    m_qtTempTUCoeffCr = NULL;
-    for (int i = 0; i < 3; i++)
-    {
-        m_qtTempTransformSkipFlag[i] = NULL;
-        m_qtTempCbf[i] = NULL;
-    }
-
-    m_cfg = NULL;
-    m_rdCost  = NULL;
-    m_trQuant = NULL;
-    m_entropyCoder = NULL;
-    m_rdSbacCoders    = NULL;
-    m_rdGoOnSbacCoder = NULL;
-}
-
-TEncSearch::~TEncSearch()
-{
-    if (m_cfg)
-    {
-        const uint32_t numLayersToAllocate = m_cfg->m_quadtreeTULog2MaxSize - m_cfg->m_quadtreeTULog2MinSize + 1;
-        for (uint32_t i = 0; i < numLayersToAllocate; ++i)
-        {
-            X265_FREE(m_qtTempCoeffY[i]);
-            X265_FREE(m_qtTempCoeffCb[i]);
-            X265_FREE(m_qtTempCoeffCr[i]);
-            m_qtTempShortYuv[i].destroy();
-        }
-    }
-    X265_FREE(m_qtTempTUCoeffY);
-    X265_FREE(m_qtTempTUCoeffCb);
-    X265_FREE(m_qtTempTUCoeffCr);
-    X265_FREE(m_qtTempTrIdx);
-    for (uint32_t i = 0; i < 3; ++i)
-    {
-        X265_FREE(m_qtTempCbf[i]);
-        X265_FREE(m_qtTempTransformSkipFlag[i]);
-    }
-
-    delete[] m_qtTempCoeffY;
-    delete[] m_qtTempCoeffCb;
-    delete[] m_qtTempCoeffCr;
-    delete[] m_qtTempShortYuv;
-    m_qtTempTransformSkipYuv.destroy();
-}
-
-bool TEncSearch::init(Encoder* cfg, TComRdCost* rdCost, TComTrQuant* trQuant)
-{
-    m_cfg     = cfg;
-    m_trQuant = trQuant;
-    m_rdCost  = rdCost;
-
-    initTempBuff(cfg->param->internalCsp);
-    m_me.setSearchMethod(cfg->param->searchMethod);
-    m_me.setSubpelRefine(cfg->param->subpelRefine);
-
-    /* When frame parallelism is active, only 'refLagPixels' of reference frames will be guaranteed
-     * available for motion reference.  See refLagRows in FrameEncoder::compressCTURows() */
-    m_refLagPixels = cfg->param->frameNumThreads > 1 ? cfg->param->searchRange : cfg->param->sourceHeight;
-
-    const uint32_t numLayersToAllocate = cfg->m_quadtreeTULog2MaxSize - cfg->m_quadtreeTULog2MinSize + 1;
-    m_qtTempCoeffY  = new TCoeff*[numLayersToAllocate];
-    m_qtTempCoeffCb = new TCoeff*[numLayersToAllocate];
-    m_qtTempCoeffCr = new TCoeff*[numLayersToAllocate];
-    m_qtTempShortYuv = new ShortYuv[numLayersToAllocate];
-    for (uint32_t i = 0; i < numLayersToAllocate; ++i)
-    {
-        m_qtTempCoeffY[i]  = X265_MALLOC(TCoeff, g_maxCUSize * g_maxCUSize);
-        m_qtTempCoeffCb[i] = X265_MALLOC(TCoeff, (g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift));
-        m_qtTempCoeffCr[i] = X265_MALLOC(TCoeff, (g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift));
-        m_qtTempShortYuv[i].create(MAX_CU_SIZE, MAX_CU_SIZE, cfg->param->internalCsp);
-    }
-
-    const uint32_t numPartitions = 1 << (g_maxCUDepth << 1);
-    CHECKED_MALLOC(m_qtTempTrIdx, uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempCbf[0], uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempCbf[1], uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempCbf[2], uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempTransformSkipFlag[0], uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempTransformSkipFlag[1], uint8_t, numPartitions);
-    CHECKED_MALLOC(m_qtTempTransformSkipFlag[2], uint8_t, numPartitions);
-
-    CHECKED_MALLOC(m_qtTempTUCoeffY, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
-    CHECKED_MALLOC(m_qtTempTUCoeffCb, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
-    CHECKED_MALLOC(m_qtTempTUCoeffCr, TCoeff, MAX_TS_WIDTH * MAX_TS_HEIGHT);
-
-    return m_qtTempTransformSkipYuv.create(g_maxCUSize, g_maxCUSize, cfg->param->internalCsp);
-
-fail:
-    return false;
-}
-
-void TEncSearch::setQPLambda(int QP, double lambdaLuma, double lambdaChroma)
-{
-    m_trQuant->setLambda(lambdaLuma, lambdaChroma);
-    m_me.setQP(QP);
-}
-
-void TEncSearch::xEncSubdivCbfQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma)
-{
-    uint32_t fullDepth  = cu->getDepth(0) + trDepth;
-    uint32_t trMode     = cu->getTransformIdx(absPartIdx);
-    uint32_t subdiv     = (trMode > trDepth ? 1 : 0);
-    uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize()] + 2 - fullDepth;
-    int      chFmt      = cu->getChromaFormat();
-
-    if (cu->getPredictionMode(0) == MODE_INTRA && cu->getPartitionSize(0) == SIZE_NxN && trDepth == 0)
-    {
-        assert(subdiv);
-    }
-    else if (trSizeLog2 > cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize())
-    {
-        assert(subdiv);
-    }
-    else if (trSizeLog2 == cu->getSlice()->getSPS()->getQuadtreeTULog2MinSize())
-    {
-        assert(!subdiv);
-    }
-    else if (trSizeLog2 == cu->getQuadtreeTULog2MinSizeInCU(absPartIdx))
-    {
-        assert(!subdiv);
-    }
-    else
-    {
-        assert(trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-        if (bLuma)
-        {
-            m_entropyCoder->encodeTransformSubdivFlag(subdiv, 5 - trSizeLog2);
-        }
-    }
-
-    if (bChroma)
-    {
-        if ((trSizeLog2 > 2) && !(chFmt == CHROMA_444))
-        {
-            if (trDepth == 0 || cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepth - 1))
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, trDepth);
-            if (trDepth == 0 || cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepth - 1))
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, trDepth);
-        }
-    }
-
-    if (subdiv)
-    {
-        uint32_t qtPartNum = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        for (uint32_t part = 0; part < 4; part++)
-        {
-            xEncSubdivCbfQT(cu, trDepth + 1, absPartIdx + part * qtPartNum, bLuma, bChroma);
-        }
-
-        return;
-    }
-
-    //===== Cbfs =====
-    if (bLuma)
-    {
-        m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
-    }
-}
-
-void TEncSearch::xEncCoeffQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TextType ttype)
-{
-    uint32_t fullDepth  = cu->getDepth(0) + trDepth;
-    uint32_t trMode     = cu->getTransformIdx(absPartIdx);
-    uint32_t subdiv     = (trMode > trDepth ? 1 : 0);
-    uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize()] + 2 - fullDepth;
-    uint32_t chroma     = (ttype != TEXT_LUMA ? 1 : 0);
-    int chFmt           = cu->getChromaFormat();
-
-    if (subdiv)
-    {
-        uint32_t qtPartNum = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        for (uint32_t part = 0; part < 4; part++)
-        {
-            xEncCoeffQT(cu, trDepth + 1, absPartIdx + part * qtPartNum, ttype);
-        }
-
-        return;
-    }
-
-    if ((ttype != TEXT_LUMA) && (trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-    {
-        assert(trDepth > 0);
-        trDepth--;
-        uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trDepth) << 1);
-        bool bFirstQ = ((absPartIdx % qpdiv) == 0);
-        if (!bFirstQ)
-        {
-            return;
-        }
-    }
-
-    //===== coefficients =====
-    int cspx = chroma ? m_hChromaShift : 0;
-    int cspy = chroma ? m_vChromaShift : 0;
-    uint32_t width = cu->getCUSize(0) >> (trDepth + cspx);
-    uint32_t height = cu->getCUSize(0) >> (trDepth + cspy);
-    uint32_t coeffOffset = (cu->getPic()->getMinCUSize() >> cspx) * (cu->getPic()->getMinCUSize() >> cspy) * absPartIdx;
-    uint32_t qtLayer = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-    TCoeff* coeff = 0;
-    switch (ttype)
-    {
-    case TEXT_LUMA:     coeff = m_qtTempCoeffY[qtLayer];
-        break;
-    case TEXT_CHROMA_U: coeff = m_qtTempCoeffCb[qtLayer];
-        break;
-    case TEXT_CHROMA_V: coeff = m_qtTempCoeffCr[qtLayer];
-        break;
-    default: assert(0);
-    }
-
-    coeff += coeffOffset;
-
-    m_entropyCoder->encodeCoeffNxN(cu, coeff, absPartIdx, width, height, fullDepth, ttype);
-}
-
-void TEncSearch::xEncIntraHeader(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma)
-{
-    if (bLuma)
-    {
-        // CU header
-        if (absPartIdx == 0)
-        {
-            if (!cu->getSlice()->isIntra())
-            {
-                if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-                {
-                    m_entropyCoder->encodeCUTransquantBypassFlag(cu, 0, true);
-                }
-                m_entropyCoder->encodeSkipFlag(cu, 0, true);
-                m_entropyCoder->encodePredMode(cu, 0, true);
-            }
-
-            m_entropyCoder->encodePartSize(cu, 0, cu->getDepth(0), true);
-
-            if (cu->isIntra(0) && cu->getPartitionSize(0) == SIZE_2Nx2N)
-            {
-                m_entropyCoder->encodeIPCMInfo(cu, 0, true);
-
-                if (cu->getIPCMFlag(0))
-                {
-                    return;
-                }
-            }
-        }
-        // luma prediction mode
-        if (cu->getPartitionSize(0) == SIZE_2Nx2N)
-        {
-            if (absPartIdx == 0)
-            {
-                m_entropyCoder->encodeIntraDirModeLuma(cu, 0);
-            }
-        }
-        else
-        {
-            uint32_t qtNumParts = cu->getTotalNumPart() >> 2;
-            if (trDepth == 0)
-            {
-                assert(absPartIdx == 0);
-                for (uint32_t part = 0; part < 4; part++)
-                {
-                    m_entropyCoder->encodeIntraDirModeLuma(cu, part * qtNumParts);
-                }
-            }
-            else if ((absPartIdx % qtNumParts) == 0)
-            {
-                m_entropyCoder->encodeIntraDirModeLuma(cu, absPartIdx);
-            }
-        }
-    }
-    if (bChroma)
-    {
-        // chroma prediction mode
-        if ((cu->getPartitionSize(0) == SIZE_2Nx2N) || !(cu->getChromaFormat() == CHROMA_444))
-        {
-            if (absPartIdx == 0)
-            {
-                m_entropyCoder->encodeIntraDirModeChroma(cu, absPartIdx, true);
-            }
-        }
-        else
-        {
-            uint32_t qtNumParts = cu->getTotalNumPart() >> 2;
-            assert(trDepth > 0);
-            if ((absPartIdx % qtNumParts) == 0)
-                m_entropyCoder->encodeIntraDirModeChroma(cu, absPartIdx, true);
-        }
-    }
-}
-
-uint32_t TEncSearch::xGetIntraBitsQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma)
-{
-    m_entropyCoder->resetBits();
-    xEncIntraHeader(cu, trDepth, absPartIdx, bLuma, bChroma);
-    xEncSubdivCbfQT(cu, trDepth, absPartIdx, bLuma, bChroma);
-
-    if (bLuma)
-    {
-        xEncCoeffQT(cu, trDepth, absPartIdx, TEXT_LUMA);
-    }
-    if (bChroma)
-    {
-        xEncCoeffQT(cu, trDepth, absPartIdx, TEXT_CHROMA_U);
-        xEncCoeffQT(cu, trDepth, absPartIdx, TEXT_CHROMA_V);
-    }
-    return m_entropyCoder->getNumberOfWrittenBits();
-}
-
-uint32_t TEncSearch::xGetIntraBitsQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t chromaId)
-{
-    m_entropyCoder->resetBits();
-    if (chromaId == TEXT_CHROMA_U)
-    {
-        xEncCoeffQT(cu, trDepth, absPartIdx, TEXT_CHROMA_U);
-    }
-    else if (chromaId == TEXT_CHROMA_V)
-    {
-        xEncCoeffQT(cu, trDepth, absPartIdx, TEXT_CHROMA_V);
-    }
-    return m_entropyCoder->getNumberOfWrittenBits();
-}
-
-void TEncSearch::xIntraCodingLumaBlk(TComDataCU* cu,
-                                     uint32_t    trDepth,
-                                     uint32_t    absPartIdx,
-                                     TComYuv*    fencYuv,
-                                     TComYuv*    predYuv,
-                                     ShortYuv*   resiYuv,
-                                     uint32_t&   outDist,
-                                     bool        bReusePred)
-{
-    uint32_t fullDepth    = cu->getDepth(0)  + trDepth;
-    uint32_t width        = cu->getCUSize(0) >> trDepth;
-    uint32_t height       = cu->getCUSize(0) >> trDepth;
-    uint32_t stride       = fencYuv->getStride();
-    Pel*     fenc         = fencYuv->getLumaAddr(absPartIdx);
-    Pel*     pred         = predYuv->getLumaAddr(absPartIdx);
-    int16_t* residual     = resiYuv->getLumaAddr(absPartIdx);
-    int      chFmt        = cu->getChromaFormat();
-    int      part         = partitionFromSizes(width, height);
-
-    uint32_t trSizeLog2     = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-    uint32_t qtLayer        = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-    uint32_t numCoeffPerInc = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-    TCoeff*  coeff          = m_qtTempCoeffY[qtLayer] + numCoeffPerInc * absPartIdx;
-
-    int16_t* reconQt        = m_qtTempShortYuv[qtLayer].getLumaAddr(absPartIdx);
-
-    assert(m_qtTempShortYuv[qtLayer].m_width == MAX_CU_SIZE);
-
-    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
-    Pel*     reconIPred       = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
-    uint32_t reconIPredStride = cu->getPic()->getPicYuvRec()->getStride();
-    bool     useTransformSkip = cu->getTransformSkip(absPartIdx, TEXT_LUMA);
-
-    if (!bReusePred)
-    {
-        //===== init availability pattern =====
-        TComPattern::initAdiPattern(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt);
-        uint32_t lumaPredMode = cu->getLumaIntraDir(absPartIdx);
-        //===== get prediction signal =====
-        predIntraLumaAng(lumaPredMode, pred, stride, width);
-    }
-
-    //===== get residual signal =====
-    assert(!((uint32_t)(size_t)fenc & (width - 1)));
-    assert(!((uint32_t)(size_t)pred & (width - 1)));
-    assert(!((uint32_t)(size_t)residual & (width - 1)));
-    primitives.calcresidual[(int)g_convertToBit[width]](fenc, pred, residual, stride);
-
-    //===== transform and quantization =====
-    //--- init rate estimation arrays for RDOQ ---
-    if (useTransformSkip ? m_cfg->bEnableRDOQTS : m_cfg->bEnableRDOQ)
-    {
-        m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, width, TEXT_LUMA);
-    }
-
-    //--- transform and quantization ---
-    uint32_t absSum = 0;
-    int lastPos = -1;
-    cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
-
-    m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-    m_trQuant->selectLambda(TEXT_LUMA);
-
-    absSum = m_trQuant->transformNxN(cu, residual, stride, coeff, width, TEXT_LUMA, absPartIdx, &lastPos, useTransformSkip);
-
-    //--- set coded block flag ---
-    cu->setCbfSubParts((absSum ? 1 : 0) << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
-
-    //--- inverse transform ---
-    int size = g_convertToBit[width];
-    if (absSum)
-    {
-        int scalingListType = 0 + TEXT_LUMA;
-        assert(scalingListType < 6);
-        m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), cu->getLumaIntraDir(absPartIdx), residual, stride, coeff, width, scalingListType, useTransformSkip, lastPos);
-    }
-    else
-    {
-        int16_t* resiTmp = residual;
-        memset(coeff, 0, sizeof(TCoeff) * width * height);
-        primitives.blockfill_s[size](resiTmp, stride, 0);
-    }
-
-    assert(width <= 32);
-#if NEW_CALCRECON
-    //===== reconstruction =====
-    primitives.calcrecon[size](pred, residual, 0, reconQt, reconIPred, stride, MAX_CU_SIZE, reconIPredStride);
-    //===== update distortion =====
-    outDist += primitives.sse_sp[part](reconQt, MAX_CU_SIZE, fenc, stride);
-#else
-    ALIGN_VAR_32(pixel, recon[MAX_CU_SIZE * MAX_CU_SIZE]);
-    //===== reconstruction =====
-    primitives.calcrecon[size](pred, residual, recon, reconQt, reconIPred, stride, MAX_CU_SIZE, reconIPredStride);
-    //===== update distortion =====
-    outDist += primitives.sse_pp[part](fenc, stride, recon, stride);
-#endif
-}
-
-void TEncSearch::xIntraCodingChromaBlk(TComDataCU* cu,
-                                       uint32_t    trDepth,
-                                       uint32_t    absPartIdx,
-                                       TComYuv*    fencYuv,
-                                       TComYuv*    predYuv,
-                                       ShortYuv*   resiYuv,
-                                       uint32_t&   outDist,
-                                       uint32_t    chromaId,
-                                       bool        bReusePred)
-{
-    uint32_t origTrDepth = trDepth;
-    uint32_t fullDepth   = cu->getDepth(0) + trDepth;
-    uint32_t trSizeLog2  = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-    int      chFmt       = cu->getChromaFormat();
-
-    if ((trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-    {
-        assert(trDepth > 0);
-        trDepth--;
-        uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trDepth) << 1);
-        bool bFirstQ = ((absPartIdx % qpdiv) == 0);
-        if (!bFirstQ)
-        {
-            return;
-        }
-    }
-
-    TextType ttype          = (chromaId > 0 ? TEXT_CHROMA_V : TEXT_CHROMA_U);
-    uint32_t width          = cu->getCUSize(0) >> (trDepth + m_hChromaShift);
-    uint32_t height         = cu->getCUSize(0) >> (trDepth + m_vChromaShift);
-    uint32_t stride         = fencYuv->getCStride();
-    Pel*     fenc           = (chromaId > 0 ? fencYuv->getCrAddr(absPartIdx) : fencYuv->getCbAddr(absPartIdx));
-    Pel*     pred           = (chromaId > 0 ? predYuv->getCrAddr(absPartIdx) : predYuv->getCbAddr(absPartIdx));
-    int16_t* residual       = (chromaId > 0 ? resiYuv->getCrAddr(absPartIdx) : resiYuv->getCbAddr(absPartIdx));
-
-    uint32_t qtlayer        = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-    uint32_t numCoeffPerInc = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1)) >> (m_hChromaShift + m_vChromaShift);
-    TCoeff*  coeff          = (chromaId > 0 ? m_qtTempCoeffCr[qtlayer] : m_qtTempCoeffCb[qtlayer]) + numCoeffPerInc * absPartIdx;
-    int16_t* reconQt        = (chromaId > 0 ? m_qtTempShortYuv[qtlayer].getCrAddr(absPartIdx) : m_qtTempShortYuv[qtlayer].getCbAddr(absPartIdx));
-    uint32_t reconQtStride  = m_qtTempShortYuv[qtlayer].m_cwidth;
-    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
-    Pel*     reconIPred       = (chromaId > 0 ? cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder) : cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder));
-    uint32_t reconIPredStride = cu->getPic()->getPicYuvRec()->getCStride();
-    bool     useTransformSkipChroma = cu->getTransformSkip(absPartIdx, ttype);
-    int      part = partitionFromSizes(width, height);
-
-    if (!bReusePred)
-    {
-        //===== init availability pattern =====
-        TComPattern::initAdiPatternChroma(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, chromaId);
-        Pel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, height, m_predBuf);
-
-        uint32_t chromaPredMode = cu->getChromaIntraDir(absPartIdx);
-        //===== update chroma mode =====
-        if (chromaPredMode == DM_CHROMA_IDX)
-        {
-            chromaPredMode = cu->getLumaIntraDir(absPartIdx);
-        }
-
-        //===== get prediction signal =====
-        predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, width, height, chFmt);
-    }
-
-    //===== get residual signal =====
-    assert(!((uint32_t)(size_t)fenc & (width - 1)));
-    assert(!((uint32_t)(size_t)pred & (width - 1)));
-    assert(!((uint32_t)(size_t)residual & (width - 1)));
-    int size = g_convertToBit[width];
-    primitives.calcresidual[size](fenc, pred, residual, stride);
-
-    //===== transform and quantization =====
-    {
-        //--- init rate estimation arrays for RDOQ ---
-        if (useTransformSkipChroma ? m_cfg->bEnableRDOQTS : m_cfg->bEnableRDOQ)
-        {
-            m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, width, ttype);
-        }
-        //--- transform and quantization ---
-        uint32_t absSum = 0;
-        int lastPos = -1;
-
-        int curChromaQpOffset;
-        if (ttype == TEXT_CHROMA_U)
-        {
-            curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-        }
-        else
-        {
-            curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-        }
-        m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-        m_trQuant->selectLambda(TEXT_CHROMA);
-
-        absSum = m_trQuant->transformNxN(cu, residual, stride, coeff, width, ttype, absPartIdx, &lastPos, useTransformSkipChroma);
-
-        //--- set coded block flag ---
-        cu->setCbfSubParts((absSum ? 1 : 0) << origTrDepth, ttype, absPartIdx, cu->getDepth(0) + trDepth);
-
-        //--- inverse transform ---
-        if (absSum)
-        {
-            int scalingListType = 0 + ttype;
-            assert(scalingListType < 6);
-            m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, residual, stride, coeff, width, scalingListType, useTransformSkipChroma, lastPos);
-        }
-        else
-        {
-            int16_t* resiTmp = residual;
-            memset(coeff, 0, sizeof(TCoeff) * width * height);
-            primitives.blockfill_s[size](resiTmp, stride, 0);
-        }
-    }
-
-    assert(((intptr_t)residual & (width - 1)) == 0);
-    assert(width <= 32);
-#if NEW_CALCRECON
-    //===== reconstruction =====
-    primitives.calcrecon[size](pred, residual, 0, reconQt, reconIPred, stride, reconQtStride, reconIPredStride);
-    //===== update distortion =====
-    uint32_t dist = primitives.sse_sp[part](reconQt, reconQtStride, fenc, stride);
-#else
-    ALIGN_VAR_32(pixel, recon[MAX_CU_SIZE * MAX_CU_SIZE]);
-    //===== reconstruction =====
-    primitives.calcrecon[size](pred, residual, recon, reconQt, reconIPred, stride, reconQtStride, reconIPredStride);
-    //===== update distortion =====
-    uint32_t dist = primitives.sse_pp[part](fenc, stride, recon, stride);
-#endif
-    if (ttype == TEXT_CHROMA_U)
-    {
-        outDist += m_rdCost->scaleChromaDistCb(dist);
-    }
-    else if (ttype == TEXT_CHROMA_V)
-    {
-        outDist += m_rdCost->scaleChromaDistCr(dist);
-    }
-    else
-    {
-        outDist += dist;
-    }
-}
-
-void TEncSearch::xRecurIntraCodingQT(TComDataCU* cu,
-                                     uint32_t    trDepth,
-                                     uint32_t    absPartIdx,
-                                     TComYuv*    fencYuv,
-                                     TComYuv*    predYuv,
-                                     ShortYuv*   resiYuv,
-                                     uint32_t&   outDistY,
-                                     bool        bCheckFirst,
-                                     uint64_t&   rdCost)
-{
-    uint32_t fullDepth   = cu->getDepth(0) +  trDepth;
-    uint32_t trSizeLog2  = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-    bool     bCheckFull  = (trSizeLog2 <= cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize());
-    bool     bCheckSplit = (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-
-    int maxTuSize = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize();
-    int isIntraSlice = (cu->getSlice()->getSliceType() == I_SLICE);
-
-    // don't check split if TU size is less or equal to max TU size
-    bool noSplitIntraMaxTuSize = bCheckFull;
-
-    if (m_cfg->param->rdPenalty && !isIntraSlice)
-    {
-        // in addition don't check split if TU size is less or equal to 16x16 TU size for non-intra slice
-        noSplitIntraMaxTuSize = (trSizeLog2 <= X265_MIN(maxTuSize, 4));
-
-        // if maximum RD-penalty don't check TU size 32x32
-        if (m_cfg->param->rdPenalty == 2)
-        {
-            bCheckFull = (trSizeLog2 <= X265_MIN(maxTuSize, 4));
-        }
-    }
-    if (bCheckFirst && noSplitIntraMaxTuSize)
-    {
-        bCheckSplit = false;
-    }
-
-    uint64_t singleCost  = MAX_INT64;
-    uint32_t singleDistY = 0;
-    uint32_t singleCbfY  = 0;
-    bool   checkTransformSkip  = cu->getSlice()->getPPS()->getUseTransformSkip();
-    uint32_t widthTransformSkip  = cu->getCUSize(0) >> trDepth;
-    uint32_t heightTransformSkip = cu->getCUSize(0) >> trDepth;
-    int    bestModeId    = 0;
-
-    checkTransformSkip &= (widthTransformSkip == 4 && heightTransformSkip == 4);
-    checkTransformSkip &= (!cu->getCUTransquantBypass(0));
-    checkTransformSkip &= (!((cu->getQP(0) == 0) && (cu->getSlice()->getSPS()->getUseLossless())));
-    if (m_cfg->param->bEnableTSkipFast)
-    {
-        checkTransformSkip &= (cu->getPartitionSize(absPartIdx) == SIZE_NxN);
-    }
-
-    if (bCheckFull)
-    {
-        if (checkTransformSkip == true)
-        {
-            //----- store original entropy coding status -----
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-
-            uint32_t singleDistYTmp     = 0;
-            uint32_t singleCbfYTmp      = 0;
-            uint64_t singleCostTmp      = 0;
-            const int firstCheckId      = 0;
-
-            for (int modeId = firstCheckId; modeId < 2; modeId++)
-            {
-                singleDistYTmp = 0;
-                cu->setTransformSkipSubParts(modeId, TEXT_LUMA, absPartIdx, fullDepth);
-                //----- code luma block with given intra prediction mode and store Cbf-----
-                bool bReusePred = modeId != firstCheckId;
-                xIntraCodingLumaBlk(cu, trDepth, absPartIdx, fencYuv, predYuv, resiYuv, singleDistYTmp, bReusePred);
-                singleCbfYTmp = cu->getCbf(absPartIdx, TEXT_LUMA, trDepth);
-                //----- determine rate and r-d cost -----
-                if (modeId == 1 && singleCbfYTmp == 0)
-                {
-                    //In order not to code TS flag when cbf is zero, the case for TS with cbf being zero is forbidden.
-                    singleCostTmp = MAX_INT64;
-                }
-                else
-                {
-                    uint32_t singleBits = xGetIntraBitsQT(cu, trDepth, absPartIdx, true, false);
-                    singleCostTmp = m_rdCost->calcRdCost(singleDistYTmp, singleBits);
-                }
-
-                if (singleCostTmp < singleCost)
-                {
-                    singleCost   = singleCostTmp;
-                    singleDistY = singleDistYTmp;
-                    singleCbfY  = singleCbfYTmp;
-                    bestModeId    = modeId;
-                    if (bestModeId == firstCheckId)
-                    {
-                        xStoreIntraResultQT(cu, trDepth, absPartIdx);
-                        m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_TEMP_BEST]);
-                    }
-                }
-                if (modeId == firstCheckId)
-                {
-                    m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-                }
-            }
-
-            cu->setTransformSkipSubParts(bestModeId, TEXT_LUMA, absPartIdx, fullDepth);
-
-            if (bestModeId == firstCheckId)
-            {
-                xLoadIntraResultQT(cu, trDepth, absPartIdx);
-                cu->setCbfSubParts(singleCbfY << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
-                m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_TEMP_BEST]);
-            }
-        }
-        else
-        {
-            cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, fullDepth);
-
-            //----- store original entropy coding status -----
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-
-            //----- code luma block with given intra prediction mode and store Cbf-----
-            xIntraCodingLumaBlk(cu, trDepth, absPartIdx, fencYuv, predYuv, resiYuv, singleDistY);
-            if (bCheckSplit)
-            {
-                singleCbfY = cu->getCbf(absPartIdx, TEXT_LUMA, trDepth);
-            }
-            //----- determine rate and r-d cost -----
-            uint32_t singleBits = xGetIntraBitsQT(cu, trDepth, absPartIdx, true, false);
-            if (m_cfg->param->rdPenalty && (trSizeLog2 == 5) && !isIntraSlice)
-            {
-                singleBits = singleBits * 4;
-            }
-            singleCost = m_rdCost->calcRdCost(singleDistY, singleBits);
-        }
-    }
-
-    if (bCheckSplit)
-    {
-        //----- store full entropy coding status, load original entropy coding status -----
-        if (bCheckFull)
-        {
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_TEST]);
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-        }
-        else
-        {
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-        }
-
-        //----- code splitted block -----
-        uint64_t splitCost     = 0;
-        uint32_t splitDistY    = 0;
-        uint32_t qPartsDiv     = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        uint32_t absPartIdxSub = absPartIdx;
-
-        uint32_t splitCbfY = 0;
-
-        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
-        {
-            xRecurIntraCodingQT(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, splitDistY, bCheckFirst, splitCost);
-
-            splitCbfY |= cu->getCbf(absPartIdxSub, TEXT_LUMA, trDepth + 1);
-        }
-
-        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
-        {
-            cu->getCbf(TEXT_LUMA)[absPartIdx + offs] |= (splitCbfY << trDepth);
-        }
-
-        //----- restore context states -----
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-
-        //----- determine rate and r-d cost -----
-        uint32_t splitBits = xGetIntraBitsQT(cu, trDepth, absPartIdx, true, false);
-        splitCost = m_rdCost->calcRdCost(splitDistY, splitBits);
-
-        //===== compare and set best =====
-        if (splitCost < singleCost)
-        {
-            //--- update cost ---
-            outDistY += splitDistY;
-            rdCost   += splitCost;
-            return;
-        }
-
-        //----- set entropy coding status -----
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_TEST]);
-
-        //--- set transform index and Cbf values ---
-        cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
-        cu->setCbfSubParts(singleCbfY << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
-        cu->setTransformSkipSubParts(bestModeId, TEXT_LUMA, absPartIdx, fullDepth);
-
-        //--- set reconstruction for next intra prediction blocks ---
-        uint32_t width     = cu->getCUSize(0) >> trDepth;
-        uint32_t height    = cu->getCUSize(0) >> trDepth;
-        uint32_t qtLayer   = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-        uint32_t zorder    = cu->getZorderIdxInCU() + absPartIdx;
-        int16_t* src       = m_qtTempShortYuv[qtLayer].getLumaAddr(absPartIdx);
-        assert(m_qtTempShortYuv[qtLayer].m_width == MAX_CU_SIZE);
-        Pel*     dst       = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
-        uint32_t dststride = cu->getPic()->getPicYuvRec()->getStride();
-        primitives.blockcpy_ps(width, height, dst, dststride, src, MAX_CU_SIZE);
-    }
-
-    outDistY += singleDistY;
-    rdCost   += singleCost;
-}
-
-void TEncSearch::residualTransformQuantIntra(TComDataCU* cu,
-                                             uint32_t    trDepth,
-                                             uint32_t    absPartIdx,
-                                             TComYuv*    fencYuv,
-                                             TComYuv*    predYuv,
-                                             ShortYuv*   resiYuv,
-                                             TComYuv*    reconYuv)
-{
-    uint32_t fullDepth   = cu->getDepth(0) +  trDepth;
-    uint32_t trSizeLog2  = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-    bool     bCheckFull  = (trSizeLog2 <= cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize());
-    bool     bCheckSplit = (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-
-    int maxTuSize = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize();
-    int isIntraSlice = (cu->getSlice()->getSliceType() == I_SLICE);
-
-    if (m_cfg->param->rdPenalty && !isIntraSlice)
-    {
-        // if maximum RD-penalty don't check TU size 32x32
-        if (m_cfg->param->rdPenalty == 2)
-        {
-            bCheckFull = (trSizeLog2 <= X265_MIN(maxTuSize, 4));
-        }
-    }
-    if (bCheckFull)
-    {
-        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, fullDepth);
-
-        //----- code luma block with given intra prediction mode and store Cbf-----
-        uint32_t lumaPredMode = cu->getLumaIntraDir(absPartIdx);
-        uint32_t width        = cu->getCUSize(0) >> trDepth;
-        uint32_t height       = cu->getCUSize(0) >> trDepth;
-        int      chFmt        = cu->getChromaFormat();
-        uint32_t stride       = fencYuv->getStride();
-        Pel*     fenc         = fencYuv->getLumaAddr(absPartIdx);
-        Pel*     pred         = predYuv->getLumaAddr(absPartIdx);
-        int16_t* residual     = resiYuv->getLumaAddr(absPartIdx);
-        Pel*     recon        = reconYuv->getLumaAddr(absPartIdx);
-
-        uint32_t numCoeffPerInc = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-        TCoeff*  coeff          = cu->getCoeffY() + numCoeffPerInc * absPartIdx;
-
-        uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
-        Pel*     reconIPred       = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
-        uint32_t reconIPredStride = cu->getPic()->getPicYuvRec()->getStride();
-
-        bool     useTransformSkip = cu->getTransformSkip(absPartIdx, TEXT_LUMA);
-
-        //===== init availability pattern =====
-
-        TComPattern::initAdiPattern(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt);
-        //===== get prediction signal =====
-        predIntraLumaAng(lumaPredMode, pred, stride, width);
-
-        //===== get residual signal =====
-        assert(!((uint32_t)(size_t)fenc & (width - 1)));
-        assert(!((uint32_t)(size_t)pred & (width - 1)));
-        assert(!((uint32_t)(size_t)residual & (width - 1)));
-        primitives.calcresidual[(int)g_convertToBit[width]](fenc, pred, residual, stride);
-
-        //===== transform and quantization =====
-        uint32_t absSum = 0;
-        int lastPos = -1;
-        cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
-
-        m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-        m_trQuant->selectLambda(TEXT_LUMA);
-        absSum = m_trQuant->transformNxN(cu, residual, stride, coeff, width, TEXT_LUMA, absPartIdx, &lastPos, useTransformSkip);
-
-        //--- set coded block flag ---
-        cu->setCbfSubParts((absSum ? 1 : 0) << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
-
-        //--- inverse transform ---
-        int size = g_convertToBit[width];
-        if (absSum)
-        {
-            int scalingListType = 0 + TEXT_LUMA;
-            assert(scalingListType < 6);
-            m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), cu->getLumaIntraDir(absPartIdx), residual, stride, coeff, width, scalingListType, useTransformSkip, lastPos);
-        }
-        else
-        {
-            int16_t* resiTmp = residual;
-            memset(coeff, 0, sizeof(TCoeff) * width * height);
-            primitives.blockfill_s[size](resiTmp, stride, 0);
-        }
-
-        //Generate Recon
-        assert(width <= 32);
-        int part = partitionFromSizes(width, height);
-        primitives.luma_add_ps[part](recon, stride, pred, residual, stride, stride);
-        primitives.blockcpy_pp(width, height, reconIPred, reconIPredStride, recon, stride);
-    }
-
-    if (bCheckSplit && !bCheckFull)
-    {
-        //----- code splitted block -----
-
-        uint32_t qPartsDiv     = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        uint32_t absPartIdxSub = absPartIdx;
-        uint32_t splitCbfY = 0;
-
-        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
-        {
-            residualTransformQuantIntra(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, reconYuv);
-            splitCbfY |= cu->getCbf(absPartIdxSub, TEXT_LUMA, trDepth + 1);
-        }
-
-        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
-        {
-            cu->getCbf(TEXT_LUMA)[absPartIdx + offs] |= (splitCbfY << trDepth);
-        }
-
-        return;
-    }
-}
-
-void TEncSearch::xSetIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-
-    if (trMode == trDepth)
-    {
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-        uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-        //===== copy transform coefficients =====
-        uint32_t numCoeffY    = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullDepth << 1);
-        uint32_t numCoeffIncY = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-        TCoeff* coeffSrcY = m_qtTempCoeffY[qtlayer] + (numCoeffIncY * absPartIdx);
-        TCoeff* coeffDestY = cu->getCoeffY()        + (numCoeffIncY * absPartIdx);
-        ::memcpy(coeffDestY, coeffSrcY, sizeof(TCoeff) * numCoeffY);
-
-        //===== copy reconstruction =====
-        m_qtTempShortYuv[qtlayer].copyPartToPartLuma(reconYuv, absPartIdx, 1 << trSizeLog2, 1 << trSizeLog2);
-    }
-    else
-    {
-        uint32_t numQPart = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        for (uint32_t part = 0; part < 4; part++)
-        {
-            xSetIntraResultQT(cu, trDepth + 1, absPartIdx + part * numQPart, reconYuv);
-        }
-    }
-}
-
-void TEncSearch::xStoreIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx)
-{
-    uint32_t fullMode = cu->getDepth(0) + trDepth;
-
-    uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullMode] + 2;
-    uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-    //===== copy transform coefficients =====
-    uint32_t numCoeffY    = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullMode << 1);
-    uint32_t numCoeffIncY = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-    TCoeff* coeffSrcY = m_qtTempCoeffY[qtlayer] + (numCoeffIncY * absPartIdx);
-    TCoeff* coeffDstY = m_qtTempTUCoeffY;
-    ::memcpy(coeffDstY, coeffSrcY, sizeof(TCoeff) * numCoeffY);
-
-    //===== copy reconstruction =====
-    m_qtTempShortYuv[qtlayer].copyPartToPartLuma(&m_qtTempTransformSkipYuv, absPartIdx, 1 << trSizeLog2, 1 << trSizeLog2);
-}
-
-void TEncSearch::xLoadIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-
-    uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-    uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-    //===== copy transform coefficients =====
-    uint32_t numCoeffY    = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullDepth << 1);
-    uint32_t numCoeffIncY = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-    TCoeff* coeffDstY = m_qtTempCoeffY[qtlayer] + (numCoeffIncY * absPartIdx);
-    TCoeff* coeffSrcY = m_qtTempTUCoeffY;
-    ::memcpy(coeffDstY, coeffSrcY, sizeof(TCoeff) * numCoeffY);
-
-    //===== copy reconstruction =====
-    uint32_t trSize = 1 << trSizeLog2;
-    m_qtTempTransformSkipYuv.copyPartToPartLuma(&m_qtTempShortYuv[qtlayer], absPartIdx, trSize);
-
-    uint32_t   zOrder           = cu->getZorderIdxInCU() + absPartIdx;
-    pixel*     reconIPred       = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), zOrder);
-    uint32_t   reconIPredStride = cu->getPic()->getPicYuvRec()->getStride();
-    int16_t*   reconQt          = m_qtTempShortYuv[qtlayer].getLumaAddr(absPartIdx);
-    primitives.blockcpy_ps(trSize, trSize, reconIPred, reconIPredStride, reconQt, MAX_CU_SIZE);
-    assert(m_qtTempShortYuv[qtlayer].m_width == MAX_CU_SIZE);
-}
-
-void TEncSearch::xStoreIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t stateU0V1Both2)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-
-    if (trMode == trDepth)
-    {
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-        uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-        bool bChromaSame = false;
-        if (trSizeLog2 == 2)
-        {
-            assert(trDepth > 0);
-            trDepth--;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trDepth) << 1);
-            if ((absPartIdx % qpdiv) != 0)
-            {
-                return;
-            }
-            bChromaSame = true;
-        }
-
-        //===== copy transform coefficients =====
-        uint32_t numCoeffC = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullDepth << 1);
-        if (!bChromaSame)
-        {
-            numCoeffC >>= 2;
-        }
-        uint32_t numCoeffIncC = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> ((cu->getSlice()->getSPS()->getMaxCUDepth() << 1) + 2);
-        if (stateU0V1Both2 == 0 || stateU0V1Both2 == 2)
-        {
-            TCoeff* coeffSrcU = m_qtTempCoeffCb[qtlayer] + (numCoeffIncC * absPartIdx);
-            TCoeff* coeffDstU = m_qtTempTUCoeffCb;
-            ::memcpy(coeffDstU, coeffSrcU, sizeof(TCoeff) * numCoeffC);
-        }
-        if (stateU0V1Both2 == 1 || stateU0V1Both2 == 2)
-        {
-            TCoeff* coeffSrcV = m_qtTempCoeffCr[qtlayer] + (numCoeffIncC * absPartIdx);
-            TCoeff* coeffDstV = m_qtTempTUCoeffCr;
-            ::memcpy(coeffDstV, coeffSrcV, sizeof(TCoeff) * numCoeffC);
-        }
-
-        //===== copy reconstruction =====
-        uint32_t lumaSize = 1 << (bChromaSame ? trSizeLog2 + 1 : trSizeLog2);
-        m_qtTempShortYuv[qtlayer].copyPartToPartYuvChroma(&m_qtTempTransformSkipYuv, absPartIdx, lumaSize, stateU0V1Both2);
-    }
-}
-
-void TEncSearch::xLoadIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t stateU0V1Both2)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-
-    if (trMode == trDepth)
-    {
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-        uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-        bool bChromaSame = false;
-        if (trSizeLog2 == 2)
-        {
-            assert(trDepth > 0);
-            trDepth--;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trDepth) << 1);
-            if ((absPartIdx % qpdiv) != 0)
-            {
-                return;
-            }
-            bChromaSame = true;
-        }
-
-        //===== copy transform coefficients =====
-        uint32_t numCoeffC = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullDepth << 1);
-        if (!bChromaSame)
-        {
-            numCoeffC >>= 2;
-        }
-        uint32_t numCoeffIncC = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> ((cu->getSlice()->getSPS()->getMaxCUDepth() << 1) + 2);
-
-        if (stateU0V1Both2 == 0 || stateU0V1Both2 == 2)
-        {
-            TCoeff* coeffDstU = m_qtTempCoeffCb[qtlayer] + (numCoeffIncC * absPartIdx);
-            TCoeff* coeffSrcU = m_qtTempTUCoeffCb;
-            ::memcpy(coeffDstU, coeffSrcU, sizeof(TCoeff) * numCoeffC);
-        }
-        if (stateU0V1Both2 == 1 || stateU0V1Both2 == 2)
-        {
-            TCoeff* coeffDstV = m_qtTempCoeffCr[qtlayer] + (numCoeffIncC * absPartIdx);
-            TCoeff* coeffSrcV = m_qtTempTUCoeffCr;
-            ::memcpy(coeffDstV, coeffSrcV, sizeof(TCoeff) * numCoeffC);
-        }
-
-        //===== copy reconstruction =====
-        uint32_t lumaSize = 1 << (bChromaSame ? trSizeLog2 + 1 : trSizeLog2);
-        m_qtTempTransformSkipYuv.copyPartToPartChroma(&m_qtTempShortYuv[qtlayer], absPartIdx, lumaSize, stateU0V1Both2);
-
-        uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
-        uint32_t width            = cu->getCUSize(0) >> (trDepth + 1);
-        uint32_t height           = cu->getCUSize(0) >> (trDepth + 1);
-        uint32_t reconQtStride    = m_qtTempShortYuv[qtlayer].m_cwidth;
-        uint32_t reconIPredStride = cu->getPic()->getPicYuvRec()->getCStride();
-
-        if (stateU0V1Both2 == 0 || stateU0V1Both2 == 2)
-        {
-            Pel* reconIPred = cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder);
-            int16_t* reconQt  = m_qtTempShortYuv[qtlayer].getCbAddr(absPartIdx);
-            primitives.blockcpy_ps(width, height, reconIPred, reconIPredStride, reconQt, reconQtStride);
-        }
-        if (stateU0V1Both2 == 1 || stateU0V1Both2 == 2)
-        {
-            Pel* reconIPred = cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder);
-            int16_t* reconQt  = m_qtTempShortYuv[qtlayer].getCrAddr(absPartIdx);
-            primitives.blockcpy_ps(width, height, reconIPred, reconIPredStride, reconQt, reconQtStride);
-        }
-    }
-}
-
-void TEncSearch::xRecurIntraChromaCodingQT(TComDataCU* cu,
-                                           uint32_t    trDepth,
-                                           uint32_t    absPartIdx,
-                                           TComYuv*    fencYuv,
-                                           TComYuv*    predYuv,
-                                           ShortYuv*  resiYuv,
-                                           uint32_t&   outDist)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-
-    if (trMode == trDepth)
-    {
-        bool checkTransformSkip = cu->getSlice()->getPPS()->getUseTransformSkip();
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-
-        uint32_t actualTrDepth = trDepth;
-        if ((trSizeLog2 == 2) && !(cu->getChromaFormat() == CHROMA_444))
-        {
-            assert(trDepth > 0);
-            actualTrDepth--;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + actualTrDepth) << 1);
-            bool bFirstQ = ((absPartIdx % qpdiv) == 0);
-            if (!bFirstQ)
-            {
-                return;
-            }
-        }
-
-        checkTransformSkip &= (trSizeLog2 <= 3);
-        if (m_cfg->param->bEnableTSkipFast)
-        {
-            checkTransformSkip &= (trSizeLog2 < 3);
-            if (checkTransformSkip)
-            {
-                int nbLumaSkip = 0;
-                for (uint32_t absPartIdxSub = absPartIdx; absPartIdxSub < absPartIdx + 4; absPartIdxSub++)
-                {
-                    nbLumaSkip += cu->getTransformSkip(absPartIdxSub, TEXT_LUMA);
-                }
-
-                checkTransformSkip &= (nbLumaSkip > 0);
-            }
-        }
-
-        if (checkTransformSkip)
-        {
-            // use RDO to decide whether Cr/Cb takes TS
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-
-            for (int chromaId = 0; chromaId < 2; chromaId++)
-            {
-                uint64_t singleCost     = MAX_INT64;
-                int      bestModeId     = 0;
-                uint32_t singleDistC    = 0;
-                uint32_t singleCbfC     = 0;
-                uint32_t singleDistCTmp = 0;
-                uint64_t singleCostTmp  = 0;
-                uint32_t singleCbfCTmp  = 0;
-
-                const int firstCheckId  = 0;
-
-                for (int chromaModeId = firstCheckId; chromaModeId < 2; chromaModeId++)
-                {
-                    cu->setTransformSkipSubParts(chromaModeId, (TextType)(chromaId + TEXT_CHROMA_U), absPartIdx, cu->getDepth(0) + actualTrDepth);
-                    singleDistCTmp = 0;
-                    bool bReusePred = chromaModeId != firstCheckId;
-                    xIntraCodingChromaBlk(cu, trDepth, absPartIdx, fencYuv, predYuv, resiYuv, singleDistCTmp, chromaId, bReusePred);
-                    singleCbfCTmp = cu->getCbf(absPartIdx, (TextType)(chromaId + TEXT_CHROMA_U), trDepth);
-
-                    if (chromaModeId == 1 && singleCbfCTmp == 0)
-                    {
-                        //In order not to code TS flag when cbf is zero, the case for TS with cbf being zero is forbidden.
-                        singleCostTmp = MAX_INT64;
-                    }
-                    else
-                    {
-                        uint32_t bitsTmp = xGetIntraBitsQTChroma(cu, trDepth, absPartIdx, chromaId + TEXT_CHROMA_U);
-                        singleCostTmp = m_rdCost->calcRdCost(singleDistCTmp, bitsTmp);
-                    }
-
-                    if (singleCostTmp < singleCost)
-                    {
-                        singleCost  = singleCostTmp;
-                        singleDistC = singleDistCTmp;
-                        bestModeId  = chromaModeId;
-                        singleCbfC  = singleCbfCTmp;
-
-                        if (bestModeId == firstCheckId)
-                        {
-                            xStoreIntraResultChromaQT(cu, trDepth, absPartIdx, chromaId);
-                            m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_TEMP_BEST]);
-                        }
-                    }
-                    if (chromaModeId == firstCheckId)
-                    {
-                        m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-                    }
-                }
-
-                if (bestModeId == firstCheckId)
-                {
-                    xLoadIntraResultChromaQT(cu, trDepth, absPartIdx, chromaId);
-                    cu->setCbfSubParts(singleCbfC << trDepth, (TextType)(chromaId + TEXT_CHROMA_U), absPartIdx, cu->getDepth(0) + actualTrDepth);
-                    m_rdGoOnSbacCoder->load(m_rdSbacCoders[fullDepth][CI_TEMP_BEST]);
-                }
-                cu->setTransformSkipSubParts(bestModeId, (TextType)(chromaId + TEXT_CHROMA_U), absPartIdx, cu->getDepth(0) + actualTrDepth);
-                outDist += singleDistC;
-
-                if (chromaId == 0)
-                {
-                    m_rdGoOnSbacCoder->store(m_rdSbacCoders[fullDepth][CI_QT_TRAFO_ROOT]);
-                }
-            }
-        }
-        else
-        {
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + actualTrDepth);
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + actualTrDepth);
-            xIntraCodingChromaBlk(cu, trDepth, absPartIdx, fencYuv, predYuv, resiYuv, outDist, 0);
-            xIntraCodingChromaBlk(cu, trDepth, absPartIdx, fencYuv, predYuv, resiYuv, outDist, 1);
-        }
-    }
-    else
-    {
-        uint32_t splitCbfU     = 0;
-        uint32_t splitCbfV     = 0;
-        uint32_t qPartsDiv     = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        uint32_t absPartIdxSub = absPartIdx;
-        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
-        {
-            xRecurIntraChromaCodingQT(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, outDist);
-            splitCbfU |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_U, trDepth + 1);
-            splitCbfV |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_V, trDepth + 1);
-        }
-
-        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
-        {
-            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + offs] |= (splitCbfU << trDepth);
-            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + offs] |= (splitCbfV << trDepth);
-        }
-    }
-}
-
-void TEncSearch::xSetIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-    int      chFmt     = cu->getChromaFormat();
-
-    if (trMode == trDepth)
-    {
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-        uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-        bool bChromaSame  = false;
-        if ((trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-        {
-            assert(trDepth > 0);
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trDepth - 1) << 1);
-            if ((absPartIdx % qpdiv) != 0)
-            {
-                return;
-            }
-            bChromaSame = true;
-        }
-
-        //===== copy transform coefficients =====
-        uint32_t numCoeffC = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize()) >> (fullDepth << 1);
-        if (!bChromaSame)
-        {
-            numCoeffC = ((cu->getSlice()->getSPS()->getMaxCUSize() >> m_hChromaShift) * (cu->getSlice()->getSPS()->getMaxCUSize() >> m_vChromaShift)) >> (fullDepth << 1);
-        }
-
-        uint32_t numCoeffIncC = ((cu->getSlice()->getSPS()->getMaxCUSize() >> m_hChromaShift) * (cu->getSlice()->getSPS()->getMaxCUSize() >> m_vChromaShift)) >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-
-        TCoeff* coeffSrcU = m_qtTempCoeffCb[qtlayer] + (numCoeffIncC * absPartIdx);
-        TCoeff* coeffSrcV = m_qtTempCoeffCr[qtlayer] + (numCoeffIncC * absPartIdx);
-        TCoeff* coeffDstU = cu->getCoeffCb()         + (numCoeffIncC * absPartIdx);
-        TCoeff* coeffDstV = cu->getCoeffCr()         + (numCoeffIncC * absPartIdx);
-        ::memcpy(coeffDstU, coeffSrcU, sizeof(TCoeff) * numCoeffC);
-        ::memcpy(coeffDstV, coeffSrcV, sizeof(TCoeff) * numCoeffC);
-
-        //===== copy reconstruction =====
-        m_qtTempShortYuv[qtlayer].copyPartToPartChroma(reconYuv, absPartIdx, 1 << trSizeLog2, bChromaSame);
-    }
-    else
-    {
-        uint32_t numQPart = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        for (uint32_t part = 0; part < 4; part++)
-        {
-            xSetIntraResultChromaQT(cu, trDepth + 1, absPartIdx + part * numQPart, reconYuv);
-        }
-    }
-}
-
-void TEncSearch::residualQTIntrachroma(TComDataCU* cu,
-                                       uint32_t    trDepth,
-                                       uint32_t    absPartIdx,
-                                       TComYuv*    fencYuv,
-                                       TComYuv*    predYuv,
-                                       ShortYuv*  resiYuv,
-                                       TComYuv*    reconYuv)
-{
-    uint32_t fullDepth = cu->getDepth(0) + trDepth;
-    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
-    int      chFmt     = cu->getChromaFormat();
-
-    if (trMode == trDepth)
-    {
-        uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> fullDepth] + 2;
-        uint32_t actualTrDepth = trDepth;
-        if (trSizeLog2 == 2)
-        {
-            assert(trDepth > 0);
-            actualTrDepth--;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + actualTrDepth) << 1);
-            bool bFirstQ = ((absPartIdx % qpdiv) == 0);
-            if (!bFirstQ)
-            {
-                return;
-            }
-        }
-
-        cu->setTransformSkipSubParts(0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) +  actualTrDepth);
-        cu->setTransformSkipSubParts(0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) +  actualTrDepth);
-        uint32_t width          = cu->getCUSize(0) >> (trDepth + m_hChromaShift);
-        uint32_t height         = cu->getCUSize(0) >> (trDepth + m_vChromaShift);
-        uint32_t stride         = fencYuv->getCStride();
-
-        for (uint32_t chromaId = 0; chromaId < 2; chromaId++)
-        {
-            TextType ttype          = (chromaId > 0 ? TEXT_CHROMA_V : TEXT_CHROMA_U);
-            uint32_t chromaPredMode = cu->getChromaIntraDir(absPartIdx);
-            Pel*     fenc           = (chromaId > 0 ? fencYuv->getCrAddr(absPartIdx) : fencYuv->getCbAddr(absPartIdx));
-            Pel*     pred           = (chromaId > 0 ? predYuv->getCrAddr(absPartIdx) : predYuv->getCbAddr(absPartIdx));
-            int16_t* residual       = (chromaId > 0 ? resiYuv->getCrAddr(absPartIdx) : resiYuv->getCbAddr(absPartIdx));
-            Pel*     recon          = (chromaId > 0 ? reconYuv->getCrAddr(absPartIdx) : reconYuv->getCbAddr(absPartIdx));
-            uint32_t numCoeffPerInc = (cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1)) >> 2;
-            TCoeff*  coeff          = (chromaId > 0 ? cu->getCoeffCr() : cu->getCoeffCb()) + numCoeffPerInc * absPartIdx;
-
-            uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
-            Pel*     reconIPred       = (chromaId > 0 ? cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder) : cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder));
-            uint32_t reconIPredStride = cu->getPic()->getPicYuvRec()->getCStride();
-            bool     useTransformSkipChroma = cu->getTransformSkip(absPartIdx, ttype);
-            //===== update chroma mode =====
-            if (chromaPredMode == DM_CHROMA_IDX)
-            {
-                chromaPredMode = cu->getLumaIntraDir(0);
-            }
-            //===== init availability pattern =====
-            TComPattern::initAdiPatternChroma(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, chromaId);
-            Pel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, height, m_predBuf);
-
-            //===== get prediction signal =====
-            predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, width, height, chFmt);
-
-            //===== get residual signal =====
-            assert(!((uint32_t)(size_t)fenc & (width - 1)));
-            assert(!((uint32_t)(size_t)pred & (width - 1)));
-            assert(!((uint32_t)(size_t)residual & (width - 1)));
-            int size = g_convertToBit[width];
-            primitives.calcresidual[size](fenc, pred, residual, stride);
-
-            //--- transform and quantization ---
-            uint32_t absSum = 0;
-            int lastPos = -1;
-
-            int curChromaQpOffset;
-            if (ttype == TEXT_CHROMA_U)
-            {
-                curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-            }
-            else
-            {
-                curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-            }
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-            m_trQuant->selectLambda(TEXT_CHROMA);
-
-            absSum = m_trQuant->transformNxN(cu, residual, stride, coeff, width, ttype, absPartIdx, &lastPos, useTransformSkipChroma);
-
-            //--- set coded block flag ---
-            cu->setCbfSubParts((absSum ? 1 : 0) << trDepth, ttype, absPartIdx, cu->getDepth(0) + trDepth);
-
-            //--- inverse transform ---
-            if (absSum)
-            {
-                int scalingListType = 0 + ttype;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, residual, stride, coeff, width, scalingListType, useTransformSkipChroma, lastPos);
-            }
-            else
-            {
-                int16_t* resiTmp = residual;
-                memset(coeff, 0, sizeof(TCoeff) * width * height);
-                primitives.blockfill_s[size](resiTmp, stride, 0);
-            }
-
-            //===== reconstruction =====
-            assert(((uint32_t)(size_t)residual & (width - 1)) == 0);
-            assert(width <= 32);
-            int part = partitionFromSizes(cu->getCUSize(0) >> (trDepth), cu->getCUSize(0) >> (trDepth));
-            primitives.chroma[m_cfg->param->internalCsp].add_ps[part](recon, stride, pred, residual, stride, stride);
-            primitives.chroma[m_cfg->param->internalCsp].copy_pp[part](reconIPred, reconIPredStride, recon, stride);
-        }
-    }
-    else
-    {
-        uint32_t splitCbfU     = 0;
-        uint32_t splitCbfV     = 0;
-        uint32_t qPartsDiv     = cu->getPic()->getNumPartInCU() >> ((fullDepth + 1) << 1);
-        uint32_t absPartIdxSub = absPartIdx;
-        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
-        {
-            residualQTIntrachroma(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, reconYuv);
-            splitCbfU |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_U, trDepth + 1);
-            splitCbfV |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_V, trDepth + 1);
-        }
-
-        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
-        {
-            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + offs] |= (splitCbfU << trDepth);
-            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + offs] |= (splitCbfV << trDepth);
-        }
-    }
-}
-
-void TEncSearch::estIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv)
-{
-    uint32_t depth        = cu->getDepth(0);
-    uint32_t initTrDepth  = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
-    uint32_t numPU        = 1 << (2 * initTrDepth);
-    uint32_t puSize       = cu->getCUSize(0) >> initTrDepth;
-    uint32_t qNumParts    = cu->getTotalNumPart() >> 2;
-    uint32_t qPartNum     = cu->getPic()->getNumPartInCU() >> ((depth + initTrDepth) << 1);
-    uint32_t overallDistY = 0;
-    uint32_t candNum;
-    uint64_t candCostList[FAST_UDI_MAX_RDMODE_NUM];
-    uint32_t puSizeIdx    = g_convertToBit[puSize]; // log2(puSize) - 2
-    static const uint8_t intraModeNumFast[] = { 8, 8, 3, 3, 3 }; // 4x4, 8x8, 16x16, 32x32, 64x64
-
-    //===== loop over partitions =====
-    uint32_t partOffset = 0;
-    for (uint32_t pu = 0; pu < numPU; pu++, partOffset += qNumParts)
-    {
-        // Reference sample smoothing
-        TComPattern::initAdiPattern(cu, partOffset, initTrDepth, m_predBuf, m_predBufStride, m_predBufHeight, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt);
-
-        //===== determine set of modes to be tested (using prediction signal only) =====
-        const int numModesAvailable = 35; //total number of Intra modes
-        Pel* fenc   = fencYuv->getLumaAddr(pu, puSize);
-        uint32_t stride = predYuv->getStride();
-        uint32_t rdModeList[FAST_UDI_MAX_RDMODE_NUM];
-        int numModesForFullRD = intraModeNumFast[puSizeIdx];
-
-        bool doFastSearch = (numModesForFullRD != numModesAvailable);
-        if (doFastSearch)
-        {
-            assert(numModesForFullRD < numModesAvailable);
-
-            for (int i = 0; i < numModesForFullRD; i++)
-            {
-                candCostList[i] = MAX_INT64;
-            }
-
-            candNum = 0;
-            uint32_t modeCosts[35];
-
-            Pel *above         = m_refAbove    + puSize - 1;
-            Pel *aboveFiltered = m_refAboveFlt + puSize - 1;
-            Pel *left          = m_refLeft     + puSize - 1;
-            Pel *leftFiltered  = m_refLeftFlt  + puSize - 1;
-
-            // 33 Angle modes once
-            ALIGN_VAR_32(Pel, buf_trans[32 * 32]);
-            ALIGN_VAR_32(Pel, tmp[33 * 32 * 32]);
-            int scaleSize = puSize;
-            int scaleStride = stride;
-            int costShift = 0;
-
-            if (puSize > 32)
-            {
-                // origin is 64x64, we scale to 32x32 and setup required parameters
-                ALIGN_VAR_32(Pel, bufScale[32 * 32]);
-                primitives.scale2D_64to32(bufScale, fenc, stride);
-                fenc = bufScale;
-
-                // reserve space in case primitives need to store data in above
-                // or left buffers
-                Pel _above[4 * 32 + 1];
-                Pel _left[4 * 32 + 1];
-                Pel *aboveScale  = _above + 2 * 32;
-                Pel *leftScale   = _left + 2 * 32;
-                aboveScale[0] = leftScale[0] = above[0];
-                primitives.scale1D_128to64(aboveScale + 1, above + 1, 0);
-                primitives.scale1D_128to64(leftScale + 1, left + 1, 0);
-
-                scaleSize = 32;
-                scaleStride = 32;
-                costShift = 2;
-
-                // Filtered and Unfiltered refAbove and refLeft pointing to above and left.
-                above         = aboveScale;
-                left          = leftScale;
-                aboveFiltered = aboveScale;
-                leftFiltered  = leftScale;
-            }
-
-            int log2SizeMinus2 = g_convertToBit[scaleSize];
-            pixelcmp_t sa8d = primitives.sa8d[log2SizeMinus2];
-
-            // DC
-            primitives.intra_pred[log2SizeMinus2][DC_IDX](tmp, scaleStride, left, above, 0, (scaleSize <= 16));
-            modeCosts[DC_IDX] = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
-
-            Pel *abovePlanar   = above;
-            Pel *leftPlanar    = left;
-
-            if (puSize >= 8 && puSize <= 32)
-            {
-                abovePlanar = aboveFiltered;
-                leftPlanar  = leftFiltered;
-            }
-
-            // PLANAR
-            primitives.intra_pred[log2SizeMinus2][PLANAR_IDX](tmp, scaleStride, leftPlanar, abovePlanar, 0, 0);
-            modeCosts[PLANAR_IDX] = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
-
-            // Transpose NxN
-            primitives.transpose[log2SizeMinus2](buf_trans, fenc, scaleStride);
-
-            primitives.intra_pred_allangs[log2SizeMinus2](tmp, above, left, aboveFiltered, leftFiltered, (scaleSize <= 16));
-
-            for (uint32_t mode = 2; mode < numModesAvailable; mode++)
-            {
-                bool modeHor = (mode < 18);
-                Pel *cmp = (modeHor ? buf_trans : fenc);
-                intptr_t srcStride = (modeHor ? scaleSize : scaleStride);
-                modeCosts[mode] = sa8d(cmp, srcStride, &tmp[(mode - 2) * (scaleSize * scaleSize)], scaleSize) << costShift;
-            }
-
-            // Find N least cost modes. N = numModesForFullRD
-            for (uint32_t mode = 0; mode < numModesAvailable; mode++)
-            {
-                uint32_t sad = modeCosts[mode];
-                uint32_t bits = xModeBitsIntra(cu, mode, partOffset, depth, initTrDepth);
-                uint64_t cost = m_rdCost->calcRdSADCost(sad, bits);
-                candNum += xUpdateCandList(mode, cost, numModesForFullRD, rdModeList, candCostList);
-            }
-
-            int preds[3];
-            int numCand = cu->getIntraDirLumaPredictor(partOffset, preds);
-
-            for (int j = 0; j < numCand; j++)
-            {
-                bool mostProbableModeIncluded = false;
-                int mostProbableMode = preds[j];
-
-                for (int i = 0; i < numModesForFullRD; i++)
-                {
-                    if (mostProbableMode == rdModeList[i])
-                    {
-                        mostProbableModeIncluded = true;
-                        break;
-                    }
-                }
-
-                if (!mostProbableModeIncluded)
-                {
-                    rdModeList[numModesForFullRD++] = mostProbableMode;
-                }
-            }
-        }
-        else
-        {
-            for (int i = 0; i < numModesForFullRD; i++)
-            {
-                rdModeList[i] = i;
-            }
-        }
-        x265_emms();
-
-        //===== check modes (using r-d costs) =====
-        uint32_t bestPUMode  = 0;
-        uint32_t bestPUDistY = 0;
-        uint64_t bestPUCost  = MAX_INT64;
-        for (uint32_t mode = 0; mode < numModesForFullRD; mode++)
-        {
-            // set luma prediction mode
-            uint32_t origMode = rdModeList[mode];
-
-            cu->setLumaIntraDirSubParts(origMode, partOffset, depth + initTrDepth);
-
-            // set context models
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-            // determine residual for partition
-            uint32_t puDistY = 0;
-            uint64_t puCost  = 0;
-            xRecurIntraCodingQT(cu, initTrDepth, partOffset, fencYuv, predYuv, resiYuv, puDistY, true, puCost);
-
-            // check r-d cost
-            if (puCost < bestPUCost)
-            {
-                bestPUMode  = origMode;
-                bestPUDistY = puDistY;
-                bestPUCost  = puCost;
-
-                xSetIntraResultQT(cu, initTrDepth, partOffset, reconYuv);
-
-                ::memcpy(m_qtTempTrIdx,  cu->getTransformIdx()     + partOffset, qPartNum * sizeof(uint8_t));
-                ::memcpy(m_qtTempCbf[0], cu->getCbf(TEXT_LUMA)     + partOffset, qPartNum * sizeof(uint8_t));
-                ::memcpy(m_qtTempTransformSkipFlag[0], cu->getTransformSkip(TEXT_LUMA)     + partOffset, qPartNum * sizeof(uint8_t));
-            }
-        } // Mode loop
-
-        {
-            uint32_t origMode = bestPUMode;
-
-            cu->setLumaIntraDirSubParts(origMode, partOffset, depth + initTrDepth);
-
-            // set context models
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-            // determine residual for partition
-            uint32_t puDistY = 0;
-            uint64_t puCost  = 0;
-            xRecurIntraCodingQT(cu, initTrDepth, partOffset, fencYuv, predYuv, resiYuv, puDistY, false, puCost);
-
-            // check r-d cost
-            if (puCost < bestPUCost)
-            {
-                bestPUMode  = origMode;
-                bestPUDistY = puDistY;
-
-                xSetIntraResultQT(cu, initTrDepth, partOffset, reconYuv);
-
-                ::memcpy(m_qtTempTrIdx,  cu->getTransformIdx()     + partOffset, qPartNum * sizeof(uint8_t));
-                ::memcpy(m_qtTempCbf[0], cu->getCbf(TEXT_LUMA)     + partOffset, qPartNum * sizeof(uint8_t));
-                ::memcpy(m_qtTempTransformSkipFlag[0], cu->getTransformSkip(TEXT_LUMA)     + partOffset, qPartNum * sizeof(uint8_t));
-            }
-        } // Mode loop
-
-        //--- update overall distortion ---
-        overallDistY += bestPUDistY;
-
-        //--- update transform index and cbf ---
-        ::memcpy(cu->getTransformIdx()     + partOffset, m_qtTempTrIdx,  qPartNum * sizeof(uint8_t));
-        ::memcpy(cu->getCbf(TEXT_LUMA)     + partOffset, m_qtTempCbf[0], qPartNum * sizeof(uint8_t));
-        ::memcpy(cu->getTransformSkip(TEXT_LUMA)     + partOffset, m_qtTempTransformSkipFlag[0], qPartNum * sizeof(uint8_t));
-        //--- set reconstruction for next intra prediction blocks ---
-        if (pu != numPU - 1)
-        {
-            uint32_t zorder      = cu->getZorderIdxInCU() + partOffset;
-            int      part        = partitionFromSizes(puSize, puSize);
-            Pel*     dst         = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
-            uint32_t dststride   = cu->getPic()->getPicYuvRec()->getStride();
-            Pel*     src         = reconYuv->getLumaAddr(partOffset);
-            uint32_t srcstride   = reconYuv->getStride();
-            primitives.luma_copy_pp[part](dst, dststride, src, srcstride);
-        }
-
-        //=== update PU data ====
-        cu->setLumaIntraDirSubParts(bestPUMode, partOffset, depth + initTrDepth);
-        cu->copyToPic(depth, pu, initTrDepth);
-    } // PU loop
-
-    if (numPU > 1)
-    { // set Cbf for all blocks
-        uint32_t combCbfY = 0;
-        uint32_t partIdx  = 0;
-        for (uint32_t part = 0; part < 4; part++, partIdx += qNumParts)
-        {
-            combCbfY |= cu->getCbf(partIdx, TEXT_LUMA,     1);
-        }
-
-        for (uint32_t offs = 0; offs < 4 * qNumParts; offs++)
-        {
-            cu->getCbf(TEXT_LUMA)[offs] |= combCbfY;
-        }
-    }
-
-    //===== reset context models =====
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-    //===== set distortion (rate and r-d costs are determined later) =====
-    cu->m_totalDistortion = overallDistY;
-}
-
-void TEncSearch::getBestIntraModeChroma(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv)
-{
-    uint32_t depth     = cu->getDepth(0);
-    uint32_t trDepth = 0;
-    uint32_t absPartIdx = 0;
-    uint32_t bestMode  = 0;
-    uint64_t bestCost  = MAX_INT64;
-    //----- init mode list -----
-    uint32_t minMode = 0;
-    uint32_t maxMode = NUM_CHROMA_MODE;
-    uint32_t modeList[NUM_CHROMA_MODE];
-
-    uint32_t width          = cu->getCUSize(0) >> (trDepth + m_hChromaShift);
-    uint32_t height         = cu->getCUSize(0) >> (trDepth + m_vChromaShift);
-    int      chFmt          = cu->getChromaFormat();
-    uint32_t stride         = fencYuv->getCStride();
-    int scaleWidth = width;
-    int scaleStride = stride;
-    int costShift = 0;
-
-    if (width > 32)
-    {
-        scaleWidth = 32;
-        scaleStride = 32;
-        costShift = 2;
-    }
-
-    TComPattern::initAdiPatternChroma(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, 0);
-    TComPattern::initAdiPatternChroma(cu, absPartIdx, trDepth, m_predBuf, m_predBufStride, m_predBufHeight, 1);
-    cu->getAllowedChromaDir(0, modeList);
-    //----- check chroma modes -----
-    for (uint32_t mode = minMode; mode < maxMode; mode++)
-    {
-        uint64_t cost = 0;
-        for (int chromaId = 0; chromaId < 2; chromaId++)
-        {
-            int sad = 0;
-            uint32_t chromaPredMode = modeList[mode];
-            if (chromaPredMode == DM_CHROMA_IDX)
-                chromaPredMode = cu->getLumaIntraDir(0);
-            Pel*     fenc           = (chromaId > 0 ? fencYuv->getCrAddr(absPartIdx) : fencYuv->getCbAddr(absPartIdx));
-            Pel*     pred           = (chromaId > 0 ? predYuv->getCrAddr(absPartIdx) : predYuv->getCbAddr(absPartIdx));
-
-            Pel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, height, m_predBuf);
-
-            //===== get prediction signal =====
-            predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, width, height, chFmt);
-            int log2SizeMinus2 = g_convertToBit[scaleWidth];
-            pixelcmp_t sa8d = primitives.sa8d[log2SizeMinus2];
-            sad = sa8d(fenc, scaleStride, pred, scaleStride) << costShift;
-            cost += sad;
-        }
-
-        //----- compare -----
-        if (cost < bestCost)
-        {
-            bestCost = cost;
-            bestMode = modeList[mode];
-        }
-    }
-
-    cu->setChromIntraDirSubParts(bestMode, 0, depth);
-}
-
-bool TEncSearch::isNextSection()
-{
-    if (m_splitMode == DONT_SPLIT)
-    {
-        m_section++;
-        return false;
-    }
-    else
-    {
-        m_absPartIdxTURelCU += m_absPartIdxStep;
-
-        m_section++;
-        return m_section < (1 << m_splitMode);
-    }
-}
-
-bool TEncSearch::isLastSection()
-{
-    return (m_section + 1) >= (1 << m_splitMode);
-}
-
-void TEncSearch::estIntraPredChromaQT(TComDataCU* cu,
-                                      TComYuv*    fencYuv,
-                                      TComYuv*    predYuv,
-                                      ShortYuv*   resiYuv,
-                                      TComYuv*    reconYuv)
-{
-    uint32_t depth              = cu->getDepth(0);
-    uint32_t initTrDepth        = (cu->getPartitionSize(0) != SIZE_2Nx2N) && (cu->getChromaFormat() == CHROMA_444 ? 1 : 0);
-
-    m_splitMode                 = (initTrDepth == 0) ? DONT_SPLIT : QUAD_SPLIT;
-    m_absPartIdxStep            = (cu->getPic()->getNumPartInCU() >> (depth << 1)) >> partIdxStepShift[m_splitMode];
-    m_partOffset                = 0;
-    m_section                   = 0;
-    m_absPartIdxTURelCU         = 0;
-
-    do
-    {
-        uint32_t bestMode           = 0;
-        uint32_t bestDist           = 0;
-        uint64_t bestCost           = MAX_INT64;
-
-        //----- init mode list -----
-        uint32_t minMode = 0;
-        uint32_t maxMode = NUM_CHROMA_MODE;
-        uint32_t modeList[NUM_CHROMA_MODE];
-
-        m_partOffset = m_absPartIdxTURelCU;
-
-        cu->getAllowedChromaDir(m_partOffset, modeList);
-
-        //----- check chroma modes -----
-        for (uint32_t mode = minMode; mode < maxMode; mode++)
-        {
-            //----- restore context models -----
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-            //----- chroma coding -----
-            uint32_t dist = 0;
-
-            cu->setChromIntraDirSubParts(modeList[mode], m_partOffset, depth + initTrDepth);
-
-            xRecurIntraChromaCodingQT(cu, initTrDepth, m_absPartIdxTURelCU, fencYuv, predYuv, resiYuv, dist);
-
-            if (cu->getSlice()->getPPS()->getUseTransformSkip())
-            {
-                m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-            }
-
-            uint32_t bits = xGetIntraBitsQT(cu, initTrDepth, m_absPartIdxTURelCU, false, true);
-            uint64_t cost = m_rdCost->calcRdCost(dist, bits);
-
-            //----- compare -----
-            if (cost < bestCost)
-            {
-                bestCost = cost;
-                bestDist = dist;
-                bestMode = modeList[mode];
-                xSetIntraResultChromaQT(cu, initTrDepth, m_absPartIdxTURelCU, reconYuv);
-                ::memcpy(m_qtTempCbf[1], cu->getCbf(TEXT_CHROMA_U) + m_partOffset, m_absPartIdxStep * sizeof(uint8_t));
-                ::memcpy(m_qtTempCbf[2], cu->getCbf(TEXT_CHROMA_V) + m_partOffset, m_absPartIdxStep * sizeof(uint8_t));
-                ::memcpy(m_qtTempTransformSkipFlag[1], cu->getTransformSkip(TEXT_CHROMA_U) + m_partOffset, m_absPartIdxStep * sizeof(uint8_t));
-                ::memcpy(m_qtTempTransformSkipFlag[2], cu->getTransformSkip(TEXT_CHROMA_V) + m_partOffset, m_absPartIdxStep * sizeof(uint8_t));
-            }
-        }
-
-        if (!isLastSection())
-        {
-            uint32_t compWidth   = (cu->getCUSize(0) >> m_hChromaShift)  >> initTrDepth;
-            uint32_t compHeight  = (cu->getCUSize(0) >> m_vChromaShift) >> initTrDepth;
-            uint32_t zorder      = cu->getZorderIdxInCU() + m_partOffset;
-            Pel*     dst         = cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder);
-            uint32_t dststride   = cu->getPic()->getPicYuvRec()->getCStride();
-            Pel*     src         = reconYuv->getCbAddr(m_partOffset);
-            uint32_t srcstride   = reconYuv->getCStride();
-
-            primitives.blockcpy_pp(compWidth, compHeight, dst, dststride, src, srcstride);
-
-            dst                 = cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder);
-            src                 = reconYuv->getCrAddr(m_partOffset);
-            primitives.blockcpy_pp(compWidth, compHeight, dst, dststride, src, srcstride);
-        }
-
-        //----- set data -----
-        ::memcpy(cu->getCbf(TEXT_CHROMA_U) + m_partOffset, m_qtTempCbf[1], m_absPartIdxStep * sizeof(uint8_t));
-        ::memcpy(cu->getCbf(TEXT_CHROMA_V) + m_partOffset, m_qtTempCbf[2], m_absPartIdxStep * sizeof(uint8_t));
-        ::memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + m_partOffset, m_qtTempTransformSkipFlag[1], m_absPartIdxStep * sizeof(uint8_t));
-        ::memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + m_partOffset, m_qtTempTransformSkipFlag[2], m_absPartIdxStep * sizeof(uint8_t));
-        cu->setChromIntraDirSubParts(bestMode, m_partOffset, depth + initTrDepth);
-        cu->m_totalDistortion += bestDist;
-    }
-    while (isNextSection());
-
-    //----- restore context models -----
-    if (initTrDepth != 0)
-    {   // set Cbf for all blocks
-        uint32_t uiCombCbfU = 0;
-        uint32_t uiCombCbfV = 0;
-        uint32_t uiPartIdx  = 0;
-        for (uint32_t uiPart = 0; uiPart < 4; uiPart++, uiPartIdx += m_absPartIdxStep)
-        {
-            uiCombCbfU |= cu->getCbf(uiPartIdx, TEXT_CHROMA_U, 1);
-            uiCombCbfV |= cu->getCbf(uiPartIdx, TEXT_CHROMA_V, 1);
-        }
-
-        for (uint32_t uiOffs = 0; uiOffs < 4 * m_absPartIdxStep; uiOffs++)
-        {
-            cu->getCbf(TEXT_CHROMA_U)[uiOffs] |= uiCombCbfU;
-            cu->getCbf(TEXT_CHROMA_V)[uiOffs] |= uiCombCbfV;
-        }
-    }
-    //----- restore context models -----
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-}
-
-/** Function for encoding and reconstructing luma/chroma samples of a PCM mode CU.
- * \param cu pointer to current CU
- * \param absPartIdx part index
- * \param fenc pointer to original sample arrays
- * \param pcm pointer to PCM code arrays
- * \param pred pointer to prediction signal arrays
- * \param resi pointer to residual signal arrays
- * \param reco pointer to reconstructed sample arrays
- * \param stride stride of the original/prediction/residual sample arrays
- * \param width block width
- * \param height block height
- * \param ttText texture component type
- * \returns void
- */
-void TEncSearch::xEncPCM(TComDataCU* cu, uint32_t absPartIdx, Pel* fenc, Pel* pcm, Pel* pred, int16_t* resi, Pel* recon, uint32_t stride, uint32_t width, uint32_t height, TextType eText)
-{
-    uint32_t x, y;
-    uint32_t reconStride;
-    Pel* pcmTmp = pcm;
-    Pel* reconPic;
-    int shiftPcm;
-
-    if (eText == TEXT_LUMA)
-    {
-        reconStride = cu->getPic()->getPicYuvRec()->getStride();
-        reconPic    = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + absPartIdx);
-        shiftPcm = X265_DEPTH - cu->getSlice()->getSPS()->getPCMBitDepthLuma();
-    }
-    else
-    {
-        reconStride = cu->getPic()->getPicYuvRec()->getCStride();
-        if (eText == TEXT_CHROMA_U)
-        {
-            reconPic = cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), cu->getZorderIdxInCU() + absPartIdx);
-        }
-        else
-        {
-            reconPic = cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), cu->getZorderIdxInCU() + absPartIdx);
-        }
-        shiftPcm = X265_DEPTH - cu->getSlice()->getSPS()->getPCMBitDepthChroma();
-    }
-
-    // zero prediction and residual
-    for (y = 0; y < height; y++)
-    {
-        for (x = 0; x < width; x++)
-        {
-            pred[x] = resi[x] = 0;
-        }
-
-        pred += stride;
-        resi += stride;
-    }
-
-    // Encode
-    for (y = 0; y < height; y++)
-    {
-        for (x = 0; x < width; x++)
-        {
-            pcmTmp[x] = fenc[x] >> shiftPcm;
-        }
-
-        pcmTmp += width;
-        fenc += stride;
-    }
-
-    pcmTmp = pcm;
-
-    // Reconstruction
-    for (y = 0; y < height; y++)
-    {
-        for (x = 0; x < width; x++)
-        {
-            recon[x] = pcmTmp[x] << shiftPcm;
-            reconPic[x] = recon[x];
-        }
-
-        pcmTmp   += width;
-        recon    += stride;
-        reconPic += reconStride;
-    }
-}
-
-/**  Function for PCM mode estimation.
- * \param cu
- * \param fencYuv
- * \param rpcPredYuv
- * \param rpcResiYuv
- * \param rpcRecoYuv
- * \returns void
- */
-void TEncSearch::IPCMSearch(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv)
-{
-    uint32_t depth      = cu->getDepth(0);
-    uint32_t width      = cu->getCUSize(0);
-    uint32_t height     = cu->getCUSize(0);
-    uint32_t stride     = predYuv->getStride();
-    uint32_t strideC    = predYuv->getCStride();
-    uint32_t widthC     = width  >> 1;
-    uint32_t heightC    = height >> 1;
-    uint32_t distortion = 0;
-    uint32_t bits;
-    uint64_t cost;
-
-    uint32_t absPartIdx = 0;
-    uint32_t minCoeffSize = cu->getPic()->getMinCUSize() * cu->getPic()->getMinCUSize();
-    uint32_t lumaOffset   = minCoeffSize * absPartIdx;
-    uint32_t chromaOffset = lumaOffset >> 2;
-
-    // Luminance
-    Pel*   fenc = fencYuv->getLumaAddr();
-    int16_t* resi = resiYuv->getLumaAddr();
-    Pel*   pred = predYuv->getLumaAddr();
-    Pel*   recon = reconYuv->getLumaAddr();
-    Pel*   pcm  = cu->getPCMSampleY() + lumaOffset;
-
-    xEncPCM(cu, 0, fenc, pcm, pred, resi, recon, stride, width, height, TEXT_LUMA);
-
-    // Chroma U
-    fenc = fencYuv->getCbAddr();
-    resi = resiYuv->getCbAddr();
-    pred = predYuv->getCbAddr();
-    recon = reconYuv->getCbAddr();
-    pcm  = cu->getPCMSampleCb() + chromaOffset;
-
-    xEncPCM(cu, 0, fenc, pcm, pred, resi, recon, strideC, widthC, heightC, TEXT_CHROMA_U);
-
-    // Chroma V
-    fenc = fencYuv->getCrAddr();
-    resi = resiYuv->getCrAddr();
-    pred = predYuv->getCrAddr();
-    recon = reconYuv->getCrAddr();
-    pcm  = cu->getPCMSampleCr() + chromaOffset;
-
-    xEncPCM(cu, 0, fenc, pcm, pred, resi, recon, strideC, widthC, heightC, TEXT_CHROMA_V);
-
-    m_entropyCoder->resetBits();
-    xEncIntraHeader(cu, depth, absPartIdx, true, false);
-    bits = m_entropyCoder->getNumberOfWrittenBits();
-    cost = m_rdCost->calcRdCost(distortion, bits);
-
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-    cu->m_totalBits       = bits;
-    cu->m_totalCost       = cost;
-    cu->m_totalDistortion = distortion;
-
-    cu->copyToPic(depth, 0, 0);
-}
-
-/** estimation of best merge coding
- * \param cu
- * \param puIdx
- * \param m
- * \returns void
- */
-uint32_t TEncSearch::xMergeEstimation(TComDataCU* cu, int puIdx, MergeData& m)
-{
-    uint32_t depth    = cu->getDepth(m.absPartIdx);
-    PartSize partSize = cu->getPartitionSize(0);
-
-    if (cu->getSlice()->getPPS()->getLog2ParallelMergeLevelMinus2() && partSize != SIZE_2Nx2N && cu->getCUSize(0) <= 8)
-    {
-        cu->setPartSizeSubParts(SIZE_2Nx2N, 0, depth);
-        if (puIdx == 0)
-        {
-            cu->getInterMergeCandidates(0, 0, m.mvFieldNeighbours, m.interDirNeighbours, m.numValidMergeCand);
-        }
-        cu->setPartSizeSubParts(partSize, 0, depth);
-    }
-    else
-    {
-        cu->getInterMergeCandidates(m.absPartIdx, puIdx, m.mvFieldNeighbours, m.interDirNeighbours, m.numValidMergeCand);
-    }
-    /* convert bidir merge candidates into unidir
-     * TODO: why did the HM do this?, why use MV pairs below? */
-    if (cu->isBipredRestriction())
-    {
-        for (uint32_t mergeCand = 0; mergeCand < m.numValidMergeCand; ++mergeCand)
-        {
-            if (m.interDirNeighbours[mergeCand] == 3)
-            {
-                m.interDirNeighbours[mergeCand] = 1;
-                m.mvFieldNeighbours[(mergeCand << 1) + 1].setMvField(MV(0, 0), -1);
-            }
-        }
-    }
-
-    uint32_t outCost = MAX_UINT;
-    for (uint32_t mergeCand = 0; mergeCand < m.numValidMergeCand; ++mergeCand)
-    {
-        /* Prevent TMVP candidates from using unavailable reference pixels */
-        if (m_cfg->param->frameNumThreads > 1 &&
-            (m.mvFieldNeighbours[0 + 2 * mergeCand].mv.y >= (m_cfg->param->searchRange + 1) * 4 ||
-             m.mvFieldNeighbours[1 + 2 * mergeCand].mv.y >= (m_cfg->param->searchRange + 1) * 4))
-        {
-            continue;
-        }
-
-        cu->getCUMvField(REF_PIC_LIST_0)->m_mv[m.absPartIdx] = m.mvFieldNeighbours[0 + 2 * mergeCand].mv;
-        cu->getCUMvField(REF_PIC_LIST_0)->m_refIdx[m.absPartIdx] = m.mvFieldNeighbours[0 + 2 * mergeCand].refIdx;
-        cu->getCUMvField(REF_PIC_LIST_1)->m_mv[m.absPartIdx] = m.mvFieldNeighbours[1 + 2 * mergeCand].mv;
-        cu->getCUMvField(REF_PIC_LIST_1)->m_refIdx[m.absPartIdx] = m.mvFieldNeighbours[1 + 2 * mergeCand].refIdx;
-
-        motionCompensation(cu, &m_predTempYuv, REF_PIC_LIST_X, puIdx, true, false);
-        uint32_t costCand = m_me.bufSATD(m_predTempYuv.getLumaAddr(m.absPartIdx), m_predTempYuv.getStride());
-        uint32_t bitsCand = mergeCand + 1;
-        if (mergeCand == m_cfg->param->maxNumMergeCand - 1)
-        {
-            bitsCand--;
-        }
-        costCand = costCand + m_rdCost->getCost(bitsCand);
-        if (costCand < outCost)
-        {
-            outCost = costCand;
-            m.bits = bitsCand;
-            m.mvField[0] = m.mvFieldNeighbours[0 + 2 * mergeCand];
-            m.mvField[1] = m.mvFieldNeighbours[1 + 2 * mergeCand];
-            m.interDir = m.interDirNeighbours[mergeCand];
-            m.index = mergeCand;
-        }
-    }
-
-    return outCost;
-}
-
-/** search of the best candidate for inter prediction
- * \param cu
- * \param predYuv    - output buffer for motion compensation prediction
- * \param bMergeOnly - try merge predictions only, do not perform motion estimation
- * \param bChroma    - generate a chroma prediction
- * \returns true if predYuv was filled with a motion compensated prediction
- */
-bool TEncSearch::predInterSearch(TComDataCU* cu, TComYuv* predYuv, bool bMergeOnly, bool bChroma)
-{
-    AMVPInfo amvpInfo[2][MAX_NUM_REF];
-
-    TComPicYuv *fenc    = cu->getSlice()->getPic()->getPicYuvOrg();
-    PartSize partSize   = cu->getPartitionSize(0);
-    int      numPart    = cu->getNumPartInter();
-    int      numPredDir = cu->getSlice()->isInterP() ? 1 : 2;
-    uint32_t lastMode = 0;
-    int      totalmebits = 0;
-
-    MergeData merge;
-    memset(&merge, 0, sizeof(merge));
-
-    for (int partIdx = 0; partIdx < numPart; partIdx++)
-    {
-        uint32_t partAddr;
-        int      roiWidth, roiHeight;
-        cu->getPartIndexAndSize(partIdx, partAddr, roiWidth, roiHeight);
-
-        Pel* pu = fenc->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr);
-        m_me.setSourcePU(pu - fenc->getLumaAddr(), roiWidth, roiHeight);
-
-        uint32_t mrgCost = MAX_UINT;
-
-        /* find best cost merge candidate */
-        if (cu->getPartitionSize(partAddr) != SIZE_2Nx2N)
-        {
-            merge.absPartIdx = partAddr;
-            merge.width = roiWidth;
-            merge.height = roiHeight;
-            mrgCost = xMergeEstimation(cu, partIdx, merge);
-
-            if (bMergeOnly && cu->getCUSize(0) > 8)
-            {
-                if (mrgCost == MAX_UINT)
-                {
-                    /* No valid merge modes were found, there is no possible way to
-                     * perform a valid motion compensation prediction, so early-exit */
-                    return false;
-                }
-                // set merge result
-                cu->setMergeFlag(partAddr, true);
-                cu->setMergeIndex(partAddr, merge.index);
-                cu->setInterDirSubParts(merge.interDir, partAddr, partIdx, cu->getDepth(partAddr));
-                cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(merge.mvField[0], partSize, partAddr, 0, partIdx);
-                cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(merge.mvField[1], partSize, partAddr, 0, partIdx);
-                totalmebits += merge.bits;
-
-                motionCompensation(cu, predYuv, REF_PIC_LIST_X, partIdx, true, bChroma);
-                continue;
-            }
-        }
-
-        MotionData list[2];
-        MotionData bidir[2];
-        uint32_t listSelBits[3]; // cost in bits of selecting a particular ref list
-        uint32_t bidirCost = MAX_UINT;
-        int bidirBits = 0;
-
-        list[0].cost = MAX_UINT;
-        list[1].cost = MAX_UINT;
-
-        xGetBlkBits(partSize, cu->getSlice()->isInterP(), partIdx, lastMode, listSelBits);
-
-        // Uni-directional prediction
-        for (int l = 0; l < numPredDir; l++)
-        {
-            for (int ref = 0; ref < cu->getSlice()->getNumRefIdx(l); ref++)
-            {
-                uint32_t bits = listSelBits[l] + MVP_IDX_BITS;
-                if (cu->getSlice()->getNumRefIdx(l) > 1)
-                {
-                    bits += ref + 1;
-                    if (ref == cu->getSlice()->getNumRefIdx(l) - 1)
-                        bits--;
-                }
-
-                cu->fillMvpCand(partIdx, partAddr, l, ref, &amvpInfo[l][ref]);
-
-                // Pick the best possible MVP from AMVP candidates based on least residual
-                MV mvc[AMVP_MAX_NUM_CANDS];
-                uint32_t bestCost = MAX_INT;
-                int mvpIdx = 0;
-                int numMvc = 0;
-                for (int i = 0; i < amvpInfo[l][ref].m_num; i++)
-                {
-                    MV mvCand = amvpInfo[l][ref].m_mvCand[i];
-                    if (mvCand.notZero())
-                        mvc[numMvc++] = mvCand;
-
-                    // TODO: skip mvCand if Y is > merange and -FN>1
-                    cu->clipMv(mvCand);
-
-                    xPredInterLumaBlk(cu, cu->getSlice()->getRefPic(l, ref)->getPicYuvRec(), partAddr, &mvCand, roiWidth, roiHeight, &m_predTempYuv);
-                    uint32_t cost = m_me.bufSAD(m_predTempYuv.getLumaAddr(partAddr), m_predTempYuv.getStride());
-                    cost = m_rdCost->calcRdSADCost(cost, MVP_IDX_BITS);
-
-                    if (bestCost > cost)
-                    {
-                        bestCost = cost;
-                        mvpIdx  = i;
-                    }
-                }
-
-                MV mvmin, mvmax, outmv, mvp = amvpInfo[l][ref].m_mvCand[mvpIdx];
-
-                int merange = m_cfg->param->searchRange;
-                xSetSearchRange(cu, mvp, merange, mvmin, mvmax);
-                int satdCost = m_me.motionEstimate(m_mref[l][ref], mvmin, mvmax, mvp, numMvc, mvc, merange, outmv);
-
-                /* Get total cost of partition, but only include MV bit cost once */
-                bits += m_me.bitcost(outmv);
-                uint32_t cost = (satdCost - m_me.mvcost(outmv)) + m_rdCost->getCost(bits);
-
-                /* Refine MVP selection, updates: mvp, mvpIdx, bits, cost */
-                xCheckBestMVP(&amvpInfo[l][ref], outmv, mvp, mvpIdx, bits, cost);
-
-                if (cost < list[l].cost)
-                {
-                    list[l].mv = outmv;
-                    list[l].mvp = mvp;
-                    list[l].mvpIdx = mvpIdx;
-                    list[l].ref = ref;
-                    list[l].cost = cost;
-                    list[l].bits = bits;
-                }
-            }
-        }
-
-        // Bi-directional prediction
-        if (cu->getSlice()->isInterB() && !cu->isBipredRestriction() && list[0].cost != MAX_UINT && list[1].cost != MAX_UINT)
-        {
-            ALIGN_VAR_32(pixel, avg[MAX_CU_SIZE * MAX_CU_SIZE]);
-
-            bidir[0] = list[0];
-            bidir[1] = list[1];
-
-            // Generate reference subpels
-            TComPicYuv *refPic0 = cu->getSlice()->getRefPic(REF_PIC_LIST_0, list[0].ref)->getPicYuvRec();
-            TComPicYuv *refPic1 = cu->getSlice()->getRefPic(REF_PIC_LIST_1, list[1].ref)->getPicYuvRec();
-            xPredInterLumaBlk(cu, refPic0, partAddr, &list[0].mv, roiWidth, roiHeight, &m_predYuv[0]);
-            xPredInterLumaBlk(cu, refPic1, partAddr, &list[1].mv, roiWidth, roiHeight, &m_predYuv[1]);
-
-            pixel *pred0 = m_predYuv[0].getLumaAddr(partAddr);
-            pixel *pred1 = m_predYuv[1].getLumaAddr(partAddr);
-
-            int partEnum = partitionFromSizes(roiWidth, roiHeight);
-            primitives.pixelavg_pp[partEnum](avg, roiWidth, pred0, m_predYuv[0].getStride(), pred1, m_predYuv[1].getStride(), 32);
-            int satdCost = m_me.bufSATD(avg, roiWidth);
-
-            bidirBits = list[0].bits + list[1].bits + listSelBits[2] - (listSelBits[0] + listSelBits[1]);
-            bidirCost = satdCost + m_rdCost->getCost(bidirBits);
-
-            MV mvzero(0, 0);
-            bool bTryZero = list[0].mv.notZero() || list[1].mv.notZero();
-            if (bTryZero)
-            {
-                /* Do not try zero MV if unidir motion predictors are beyond
-                 * valid search area */
-                MV mvmin, mvmax;
-                int merange = X265_MAX(m_cfg->param->sourceWidth, m_cfg->param->sourceHeight);
-                xSetSearchRange(cu, mvzero, merange, mvmin, mvmax);
-                mvmax.y += 2; // there is some pad for subpel refine
-                mvmin <<= 2;
-                mvmax <<= 2;
-
-                bTryZero &= list[0].mvp.checkRange(mvmin, mvmax);
-                bTryZero &= list[1].mvp.checkRange(mvmin, mvmax);
-            }
-            if (bTryZero)
-            {
-                // coincident blocks of the two reference pictures
-                pixel *ref0 = m_mref[0][list[0].ref]->fpelPlane + (pu - fenc->getLumaAddr());
-                pixel *ref1 = m_mref[1][list[1].ref]->fpelPlane + (pu - fenc->getLumaAddr());
-                intptr_t refStride = m_mref[0][0]->lumaStride;
-
-                primitives.pixelavg_pp[partEnum](avg, roiWidth, ref0, refStride, ref1, refStride, 32);
-                satdCost = m_me.bufSATD(avg, roiWidth);
-
-                MV mvp0 = list[0].mvp;
-                int mvpIdx0 = list[0].mvpIdx;
-                m_me.setMVP(mvp0);
-                uint32_t bits0 = list[0].bits - m_me.bitcost(list[0].mv) + m_me.bitcost(mvzero);
-
-                MV mvp1 = list[1].mvp;
-                int mvpIdx1 = list[1].mvpIdx;
-                m_me.setMVP(mvp1);
-                uint32_t bits1 = list[1].bits - m_me.bitcost(list[1].mv) + m_me.bitcost(mvzero);
-
-                uint32_t cost = satdCost + m_rdCost->getCost(bits0) + m_rdCost->getCost(bits1);
-
-                /* refine MVP selection for zero mv, updates: mvp, mvpidx, bits, cost */
-                xCheckBestMVP(&amvpInfo[0][list[0].ref], mvzero, mvp0, mvpIdx0, bits0, cost);
-                xCheckBestMVP(&amvpInfo[1][list[1].ref], mvzero, mvp1, mvpIdx1, bits1, cost);
-
-                if (cost < bidirCost)
-                {
-                    bidir[0].mv = mvzero;
-                    bidir[1].mv = mvzero;
-                    bidir[0].mvp = mvp0;
-                    bidir[1].mvp = mvp1;
-                    bidir[0].mvpIdx = mvpIdx0;
-                    bidir[1].mvpIdx = mvpIdx1;
-                    bidirCost = cost;
-                    bidirBits = bits0 + bits1 + listSelBits[2] - (listSelBits[0] + listSelBits[1]);
-                }
-            }
-        }
-
-        /* select best option and store into CU */
-        cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(TComMvField(), partSize, partAddr, 0, partIdx);
-        cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(TComMvField(), partSize, partAddr, 0, partIdx);
-
-        if (mrgCost < bidirCost && mrgCost < list[0].cost && mrgCost < list[1].cost)
-        {
-            cu->setMergeFlag(partAddr, true);
-            cu->setMergeIndex(partAddr, merge.index);
-            cu->setInterDirSubParts(merge.interDir, partAddr, partIdx, cu->getDepth(partAddr));
-            cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(merge.mvField[0], partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(merge.mvField[1], partSize, partAddr, 0, partIdx);
-
-            totalmebits += merge.bits;
-        }
-        else if (bidirCost < list[0].cost && bidirCost < list[1].cost)
-        {
-            lastMode = 2;
-
-            cu->setMergeFlag(partAddr, false);
-            cu->setInterDirSubParts(3, partAddr, partIdx, cu->getDepth(0));
-            cu->getCUMvField(REF_PIC_LIST_0)->setAllMv(bidir[0].mv, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_0)->setAllRefIdx(list[0].ref, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_0)->setMvd(partAddr, bidir[0].mv - bidir[0].mvp);
-            cu->setMVPIdx(REF_PIC_LIST_0, partAddr, bidir[0].mvpIdx);
-
-            cu->getCUMvField(REF_PIC_LIST_1)->setAllMv(bidir[1].mv, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_1)->setAllRefIdx(list[1].ref, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_1)->setMvd(partAddr, bidir[1].mv - bidir[1].mvp);
-            cu->setMVPIdx(REF_PIC_LIST_1, partAddr, bidir[1].mvpIdx);
-
-            totalmebits += bidirBits;
-        }
-        else if (list[0].cost <= list[1].cost)
-        {
-            lastMode = 0;
-
-            cu->setMergeFlag(partAddr, false);
-            cu->setInterDirSubParts(1, partAddr, partIdx, cu->getDepth(0));
-            cu->getCUMvField(REF_PIC_LIST_0)->setAllMv(list[0].mv, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_0)->setAllRefIdx(list[0].ref, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_0)->setMvd(partAddr, list[0].mv - list[0].mvp);
-            cu->setMVPIdx(REF_PIC_LIST_0, partAddr, list[0].mvpIdx);
-
-            totalmebits += list[0].bits;
-        }
-        else
-        {
-            lastMode = 1;
-
-            cu->setMergeFlag(partAddr, false);
-            cu->setInterDirSubParts(2, partAddr, partIdx, cu->getDepth(0));
-            cu->getCUMvField(REF_PIC_LIST_1)->setAllMv(list[1].mv, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_1)->setAllRefIdx(list[1].ref, partSize, partAddr, 0, partIdx);
-            cu->getCUMvField(REF_PIC_LIST_1)->setMvd(partAddr, list[1].mv - list[1].mvp);
-            cu->setMVPIdx(REF_PIC_LIST_1, partAddr, list[1].mvpIdx);
-
-            totalmebits += list[1].bits;
-        }
-
-        motionCompensation(cu, predYuv, REF_PIC_LIST_X, partIdx, true, bChroma);
-    }
-
-    x265_emms();
-    cu->m_totalBits = totalmebits;
-    return true;
-}
-
-void TEncSearch::xGetBlkBits(PartSize cuMode, bool bPSlice, int partIdx, uint32_t lastMode, uint32_t blockBit[3])
-{
-    if (cuMode == SIZE_2Nx2N)
-    {
-        blockBit[0] = (!bPSlice) ? 3 : 1;
-        blockBit[1] = 3;
-        blockBit[2] = 5;
-    }
-    else if ((cuMode == SIZE_2NxN || cuMode == SIZE_2NxnU) || cuMode == SIZE_2NxnD)
-    {
-        static const uint32_t listBits[2][3][3] = {
-            { { 0, 0, 3 }, { 0, 0, 0 }, { 0, 0, 0 } },
-            { { 5, 7, 7 }, { 7, 5, 7 }, { 9 - 3, 9 - 3, 9 - 3 } }
-        };
-        if (bPSlice)
-        {
-            blockBit[0] = 3;
-            blockBit[1] = 0;
-            blockBit[2] = 0;
-        }
-        else
-        {
-            ::memcpy(blockBit, listBits[partIdx][lastMode], 3 * sizeof(uint32_t));
-        }
-    }
-    else if ((cuMode == SIZE_Nx2N || cuMode == SIZE_nLx2N) || cuMode == SIZE_nRx2N)
-    {
-        static const uint32_t listBits[2][3][3] = {
-            { { 0, 2, 3 }, { 0, 0, 0 }, { 0, 0, 0 } },
-            { { 5, 7, 7 }, { 7 - 2, 7 - 2, 9 - 2 }, { 9 - 3, 9 - 3, 9 - 3 } }
-        };
-        if (bPSlice)
-        {
-            blockBit[0] = 3;
-            blockBit[1] = 0;
-            blockBit[2] = 0;
-        }
-        else
-        {
-            ::memcpy(blockBit, listBits[partIdx][lastMode], 3 * sizeof(uint32_t));
-        }
-    }
-    else if (cuMode == SIZE_NxN)
-    {
-        blockBit[0] = (!bPSlice) ? 3 : 1;
-        blockBit[1] = 3;
-        blockBit[2] = 5;
-    }
-    else
-    {
-        printf("Wrong!\n");
-        assert(0);
-    }
-}
-
-/* Check if using an alternative MVP would result in a smaller MVD + signal bits */
-void TEncSearch::xCheckBestMVP(AMVPInfo* amvpInfo, MV mv, MV& mvPred, int& outMvpIdx, uint32_t& outBits, uint32_t& outCost)
-{
-    assert(amvpInfo->m_mvCand[outMvpIdx] == mvPred);
-
-    m_me.setMVP(mvPred);
-    int bestMvpIdx = outMvpIdx;
-    int mvBitsOrig = m_me.bitcost(mv) + MVP_IDX_BITS;
-    int bestMvBits = mvBitsOrig;
-
-    for (int mvpIdx = 0; mvpIdx < AMVP_MAX_NUM_CANDS; mvpIdx++)
-    {
-        if (mvpIdx == outMvpIdx)
-            continue;
-
-        m_me.setMVP(amvpInfo->m_mvCand[mvpIdx]);
-        int mvbits = m_me.bitcost(mv) + MVP_IDX_BITS;
-
-        if (mvbits < bestMvBits)
-        {
-            bestMvBits = mvbits;
-            bestMvpIdx = mvpIdx;
-        }
-    }
-
-    if (bestMvpIdx != outMvpIdx) // if changed
-    {
-        mvPred = amvpInfo->m_mvCand[bestMvpIdx];
-
-        outMvpIdx = bestMvpIdx;
-        uint32_t origOutBits = outBits;
-        outBits = origOutBits - mvBitsOrig + bestMvBits;
-        outCost = (outCost - m_rdCost->getCost(origOutBits))  + m_rdCost->getCost(outBits);
-    }
-}
-
-void TEncSearch::xSetSearchRange(TComDataCU* cu, MV mvp, int merange, MV& mvmin, MV& mvmax)
-{
-    cu->clipMv(mvp);
-
-    MV dist(merange << 2, merange << 2);
-    mvmin = mvp - dist;
-    mvmax = mvp + dist;
-
-    cu->clipMv(mvmin);
-    cu->clipMv(mvmax);
-
-    mvmin >>= 2;
-    mvmax >>= 2;
-
-    /* conditional clipping for frame parallelism */
-    mvmin.y = X265_MIN(mvmin.y, m_refLagPixels);
-    mvmax.y = X265_MIN(mvmax.y, m_refLagPixels);
-}
-
-/** encode residual and calculate rate-distortion for a CU block
- * \param cu
- * \param fencYuv
- * \param predYuv
- * \param outResiYuv
- * \param rpcYuvResiBest
- * \param outReconYuv
- * \param bSkipRes
- * \returns void
- */
-void TEncSearch::encodeResAndCalcRdInterCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* outResiYuv,
-                                           ShortYuv* outBestResiYuv, TComYuv* outReconYuv, bool bSkipRes, bool curUseRDOQ)
-{
-    if (cu->isIntra(0))
-    {
-        return;
-    }
-
-    uint32_t bits = 0, bestBits = 0;
-    uint32_t distortion = 0, bdist = 0;
-
-    uint32_t width  = cu->getCUSize(0);
-    uint32_t height = cu->getCUSize(0);
-
-    //  No residual coding : SKIP mode
-    if (bSkipRes)
-    {
-        cu->setSkipFlagSubParts(true, 0, cu->getDepth(0));
-
-        outResiYuv->clear();
-
-        predYuv->copyToPartYuv(outReconYuv, 0);
-        //Luma
-        int part = partitionFromSizes(width, height);
-        distortion = primitives.sse_pp[part](fencYuv->getLumaAddr(), fencYuv->getStride(), outReconYuv->getLumaAddr(), outReconYuv->getStride());
-        //Chroma
-        part = partitionFromSizes(width >> m_hChromaShift, height >> m_vChromaShift);
-        distortion += m_rdCost->scaleChromaDistCb(primitives.sse_pp[part](fencYuv->getCbAddr(), fencYuv->getCStride(), outReconYuv->getCbAddr(), outReconYuv->getCStride()));
-        distortion += m_rdCost->scaleChromaDistCr(primitives.sse_pp[part](fencYuv->getCrAddr(), fencYuv->getCStride(), outReconYuv->getCrAddr(), outReconYuv->getCStride()));
-
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[cu->getDepth(0)][CI_CURR_BEST]);
-        m_entropyCoder->resetBits();
-        if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-        {
-            m_entropyCoder->encodeCUTransquantBypassFlag(cu, 0, true);
-        }
-        m_entropyCoder->encodeSkipFlag(cu, 0, true);
-        m_entropyCoder->encodeMergeIndex(cu, 0, true);
-
-        bits = m_entropyCoder->getNumberOfWrittenBits();
-        m_rdGoOnSbacCoder->store(m_rdSbacCoders[cu->getDepth(0)][CI_TEMP_BEST]);
-
-        cu->m_totalBits       = bits;
-        cu->m_totalDistortion = distortion;
-        cu->m_totalCost       = m_rdCost->calcRdCost(distortion, bits);
-
-        m_rdGoOnSbacCoder->store(m_rdSbacCoders[cu->getDepth(0)][CI_TEMP_BEST]);
-
-        cu->setCbfSubParts(0, 0, 0, 0, cu->getDepth(0));
-        cu->setTrIdxSubParts(0, 0, cu->getDepth(0));
-        return;
-    }
-
-    //  Residual coding.
-    uint64_t cost, bcost = MAX_INT64;
-    
-    outResiYuv->subtract(fencYuv, predYuv, width);
-
-    cost = 0;
-    bits = 0;
-    distortion = 0;
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[cu->getDepth(0)][CI_CURR_BEST]);
-
-    uint32_t zeroDistortion = 0;
-    xEstimateResidualQT(cu, 0, 0, outResiYuv, cu->getDepth(0), cost, bits, distortion, &zeroDistortion, curUseRDOQ);
-
-    m_entropyCoder->resetBits();
-    m_entropyCoder->encodeQtRootCbfZero(cu);
-    uint32_t zeroResiBits = m_entropyCoder->getNumberOfWrittenBits();
-
-    uint64_t zeroCost = m_rdCost->calcRdCost(zeroDistortion, zeroResiBits);
-    if (cu->isLosslessCoded(0))
-    {
-        zeroCost = cost + 1;
-    }
-    if (zeroCost < cost)
-    {
-        distortion = zeroDistortion;
-
-        const uint32_t qpartnum = cu->getPic()->getNumPartInCU() >> (cu->getDepth(0) << 1);
-        ::memset(cu->getTransformIdx(), 0, qpartnum * sizeof(uint8_t));
-        ::memset(cu->getCbf(TEXT_LUMA), 0, qpartnum * sizeof(uint8_t));
-        ::memset(cu->getCbf(TEXT_CHROMA_U), 0, qpartnum * sizeof(uint8_t));
-        ::memset(cu->getCbf(TEXT_CHROMA_V), 0, qpartnum * sizeof(uint8_t));
-        ::memset(cu->getCoeffY(), 0, width * height * sizeof(TCoeff));
-        ::memset(cu->getCoeffCb(), 0, width * height * sizeof(TCoeff) >> 2);
-        ::memset(cu->getCoeffCr(), 0, width * height * sizeof(TCoeff) >> 2);
-        cu->setTransformSkipSubParts(0, 0, 0, 0, cu->getDepth(0));
-    }
-    else
-    {
-        xSetResidualQTData(cu, 0, 0, NULL, cu->getDepth(0), false);
-    }
-
-    m_rdGoOnSbacCoder->load(m_rdSbacCoders[cu->getDepth(0)][CI_CURR_BEST]);
-
-    bits = xSymbolBitsInter(cu);
-
-    cost = m_rdCost->calcRdCost(distortion, bits);
-
-    if (cost < bcost)
-    {
-        if (!cu->getQtRootCbf(0))
-        {
-            outBestResiYuv->clear();
-        }
-        else
-        {
-            xSetResidualQTData(cu, 0, 0, outBestResiYuv, cu->getDepth(0), true);
-        }
-
-        bestBits = bits;
-        bcost    = cost;
-        m_rdGoOnSbacCoder->store(m_rdSbacCoders[cu->getDepth(0)][CI_TEMP_BEST]);
-    }
-
-    assert(bcost != MAX_INT64);
-
-    if (cu->getQtRootCbf(0))
-    {
-        outReconYuv->addClip(predYuv, outBestResiYuv, width);
-    }
-    else
-    {
-        predYuv->copyToPartYuv(outReconYuv, 0);
-    }
-
-    // update with clipped distortion and cost (qp estimation loop uses unclipped values)
-    int part = partitionFromSizes(width, height);
-    bdist = primitives.sse_pp[part](fencYuv->getLumaAddr(), fencYuv->getStride(), outReconYuv->getLumaAddr(), outReconYuv->getStride());
-    part = partitionFromSizes(width >> cu->getHorzChromaShift(), height >> cu->getVertChromaShift());
-    bdist += m_rdCost->scaleChromaDistCb(primitives.sse_pp[part](fencYuv->getCbAddr(), fencYuv->getCStride(), outReconYuv->getCbAddr(), outReconYuv->getCStride()));
-    bdist += m_rdCost->scaleChromaDistCr(primitives.sse_pp[part](fencYuv->getCrAddr(), fencYuv->getCStride(), outReconYuv->getCrAddr(), outReconYuv->getCStride()));
-    bcost = m_rdCost->calcRdCost(bdist, bestBits);
-
-    cu->m_totalBits       = bestBits;
-    cu->m_totalDistortion = bdist;
-    cu->m_totalCost       = bcost;
-
-    if (cu->isSkipped(0))
-    {
-        cu->setCbfSubParts(0, 0, 0, 0, cu->getDepth(0));
-    }    
-}
-
-void TEncSearch::generateCoeffRecon(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, bool skipRes)
-{
-    if (skipRes && cu->getPredictionMode(0) == MODE_INTER && cu->getMergeFlag(0) && cu->getPartitionSize(0) == SIZE_2Nx2N)
-    {
-        predYuv->copyToPartYuv(reconYuv, 0);
-        cu->setCbfSubParts(0, TEXT_LUMA, 0, 0, cu->getDepth(0));
-        cu->setCbfSubParts(0, TEXT_CHROMA_U, 0, 0, cu->getDepth(0));
-        cu->setCbfSubParts(0, TEXT_CHROMA_V, 0, 0, cu->getDepth(0));
-        return;
-    }
-    if (cu->getPredictionMode(0) == MODE_INTER)
-    {
-        residualTransformQuantInter(cu, 0, 0, resiYuv, cu->getDepth(0), true);
-        uint32_t width  = cu->getCUSize(0);
-        if (cu->getQtRootCbf(0))
-        {
-            reconYuv->addClip(predYuv, resiYuv, width);
-        }
-        else
-        {
-            predYuv->copyToPartYuv(reconYuv, 0);
-            if (cu->getMergeFlag(0) && cu->getPartitionSize(0) == SIZE_2Nx2N)
-            {
-                cu->setSkipFlagSubParts(true, 0, cu->getDepth(0));
-            }
-        }
-    }
-    else if (cu->getPredictionMode(0) == MODE_INTRA)
-    {
-        uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
-        residualTransformQuantIntra(cu, initTrDepth, 0, fencYuv, predYuv, resiYuv, reconYuv);
-        getBestIntraModeChroma(cu, fencYuv, predYuv);
-        residualQTIntrachroma(cu, 0, 0, fencYuv, predYuv, resiYuv, reconYuv);
-    }
-}
-
-#if _MSC_VER
-#pragma warning(disable: 4701) // potentially uninitialized local variable
-#endif
-
-void TEncSearch::residualTransformQuantInter(TComDataCU* cu, uint32_t absPartIdx, uint32_t absTUPartIdx, ShortYuv* resiYuv, const uint32_t depth, bool curuseRDOQ)
-{
-    assert(cu->getDepth(0) == cu->getDepth(absPartIdx));
-    const uint32_t trMode = depth - cu->getDepth(0);
-    const uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> depth] + 2;
-    int chFmt                 = cu->getChromaFormat();
-
-    bool bSplitFlag = ((cu->getSlice()->getSPS()->getQuadtreeTUMaxDepthInter() == 1) && cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N));
-    bool bCheckFull;
-    if (bSplitFlag && depth == cu->getDepth(absPartIdx) && (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx)))
-        bCheckFull = false;
-    else
-        bCheckFull = (trSizeLog2 <= cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize());
-    const bool bCheckSplit = (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-    assert(bCheckFull || bCheckSplit);
-
-    bool bCodeChroma = true;
-    uint32_t trModeC = trMode;
-    uint32_t trSizeCLog2 = trSizeLog2 - 1;
-    if (trSizeLog2 == 2)
-    {
-        trSizeCLog2++;
-        trModeC--;
-        uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trModeC) << 1);
-        bCodeChroma = ((absPartIdx % qpdiv) == 0);
-    }
-
-    const uint32_t setCbf = 1 << trMode;
-    // code full block
-    uint32_t absSumY = 0, absSumU = 0, absSumV = 0;
-    int lastPosY = -1, lastPosU = -1, lastPosV = -1;
-    if (bCheckFull)
-    {
-        const uint32_t numCoeffPerAbsPartIdxIncrement = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-
-        TCoeff *coeffCurY = cu->getCoeffY() + (numCoeffPerAbsPartIdxIncrement * absPartIdx);
-        TCoeff *coeffCurU = cu->getCoeffCb() + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> 2);
-        TCoeff *coeffCurV = cu->getCoeffCr() + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> 2);
-
-        int trWidth = 0, trHeight = 0, trWidthC = 0, trHeightC = 0;
-        uint32_t absTUPartIdxC = absPartIdx;
-
-        trWidth  = trHeight  = 1 << trSizeLog2;
-        trWidthC = trHeightC = 1 << trSizeCLog2;
-        cu->setTrIdxSubParts(depth - cu->getDepth(0), absPartIdx, depth);
-
-        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
-        if (bCodeChroma)
-        {
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-        m_trQuant->selectLambda(TEXT_LUMA);
-
-        absSumY = m_trQuant->transformNxN(cu, resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width, coeffCurY,
-                                          trWidth, TEXT_LUMA, absPartIdx, &lastPosY, false, curuseRDOQ);
-
-        cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-
-        if (bCodeChroma)
-        {
-            int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-            m_trQuant->selectLambda(TEXT_CHROMA);
-
-            absSumU = m_trQuant->transformNxN(cu, resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurU,
-                                              trWidthC, TEXT_CHROMA_U, absPartIdx, &lastPosU, false, curuseRDOQ);
-
-            curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-            absSumV = m_trQuant->transformNxN(cu, resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurV,
-                                              trWidthC, TEXT_CHROMA_V, absPartIdx, &lastPosV, false, curuseRDOQ);
-
-            cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        if (absSumY)
-        {
-            int16_t *curResiY = resiYuv->getLumaAddr(absTUPartIdx);
-
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-
-            int scalingListType = 3 + TEXT_LUMA;
-            assert(scalingListType < 6);
-            m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiY, resiYuv->m_width,  coeffCurY, trWidth, scalingListType, false, lastPosY); //this is for inter mode only
-        }
-        else
-        {
-            int16_t *ptr =  resiYuv->getLumaAddr(absTUPartIdx);
-            assert(trWidth == trHeight);
-            primitives.blockfill_s[(int)g_convertToBit[trWidth]](ptr, resiYuv->m_width, 0);
-        }
-
-        if (bCodeChroma)
-        {
-            if (absSumU)
-            {
-                int16_t *pcResiCurrU = resiYuv->getCbAddr(absTUPartIdxC);
-
-                int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_U;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, pcResiCurrU, resiYuv->m_cwidth, coeffCurU, trWidthC, scalingListType, false, lastPosU);
-            }
-            else
-            {
-                int16_t *ptr = resiYuv->getCbAddr(absTUPartIdxC);
-                assert(trWidthC == trHeightC);
-                primitives.blockfill_s[(int)g_convertToBit[trWidthC]](ptr, resiYuv->m_cwidth, 0);
-            }
-            if (absSumV)
-            {
-                int16_t *curResiV = resiYuv->getCrAddr(absTUPartIdxC);
-                int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_V;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiV, resiYuv->m_cwidth, coeffCurV, trWidthC, scalingListType, false, lastPosV);
-            }
-            else
-            {
-                int16_t *ptr =  resiYuv->getCrAddr(absTUPartIdxC);
-                assert(trWidthC == trHeightC);
-                primitives.blockfill_s[(int)g_convertToBit[trWidthC]](ptr, resiYuv->m_cwidth, 0);
-            }
-        }
-        cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-        if (bCodeChroma)
-        {
-            cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-    }
-
-    // code sub-blocks
-    if (bCheckSplit && !bCheckFull)
-    {
-        const uint32_t qPartNumSubdiv = cu->getPic()->getNumPartInCU() >> ((depth + 1) << 1);
-        for (uint32_t i = 0; i < 4; ++i)
-        {
-            uint32_t nsAddr = absPartIdx + i * qPartNumSubdiv;
-            residualTransformQuantInter(cu, absPartIdx + i * qPartNumSubdiv, nsAddr, resiYuv, depth + 1, curuseRDOQ);
-        }
-
-        uint32_t ycbf = 0;
-        uint32_t ucbf = 0;
-        uint32_t vcbf = 0;
-        for (uint32_t i = 0; i < 4; ++i)
-        {
-            ycbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_LUMA,     trMode + 1);
-            ucbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_U, trMode + 1);
-            vcbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_V, trMode + 1);
-        }
-
-        for (uint32_t i = 0; i < 4 * qPartNumSubdiv; ++i)
-        {
-            cu->getCbf(TEXT_LUMA)[absPartIdx + i] |= ycbf << trMode;
-            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + i] |= ucbf << trMode;
-            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + i] |= vcbf << trMode;
-        }
-
-        return;
-    }
-
-    cu->setTrIdxSubParts(trMode, absPartIdx, depth);
-    cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-
-    if (bCodeChroma)
-    {
-        cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-        cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-    }
-}
-
-void TEncSearch::xEstimateResidualQT(TComDataCU*    cu,
-                                     uint32_t       absPartIdx,
-                                     uint32_t       absTUPartIdx,
-                                     ShortYuv*     resiYuv,
-                                     const uint32_t depth,
-                                     uint64_t &     rdCost,
-                                     uint32_t &     outBits,
-                                     uint32_t &     outDist,
-                                     uint32_t *     outZeroDist,
-                                     bool           curuseRDOQ)
-{
-    assert(cu->getDepth(0) == cu->getDepth(absPartIdx));
-    const uint32_t trMode = depth - cu->getDepth(0);
-    const uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> depth] + 2;
-    uint32_t  trSizeCLog2     = g_convertToBit[(cu->getSlice()->getSPS()->getMaxCUSize() >> m_hChromaShift) >> depth] + 2;
-    int chFmt                 = cu->getChromaFormat();
-
-    bool bSplitFlag = ((cu->getSlice()->getSPS()->getQuadtreeTUMaxDepthInter() == 1) && cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N));
-    bool bCheckFull;
-    if (bSplitFlag && depth == cu->getDepth(absPartIdx) && (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx)))
-        bCheckFull = false;
-    else
-        bCheckFull = (trSizeLog2 <= cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize());
-    const bool bCheckSplit = (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx));
-    assert(bCheckFull || bCheckSplit);
-
-    bool bCodeChroma = true;
-    uint32_t trModeC = trMode;
-    if ((trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-    {
-        trSizeCLog2++;
-        trModeC--;
-        uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((depth - 1) << 1);
-        bCodeChroma = ((absPartIdx % qpdiv) == 0);
-    }
-
-    const uint32_t setCbf = 1 << trMode;
-    // code full block
-    uint64_t singleCost = MAX_INT64;
-    uint32_t singleBits = 0;
-    uint32_t singleDist = 0;
-    uint32_t absSumY = 0, absSumU = 0, absSumV = 0;
-    int lastPosY = -1, lastPosU = -1, lastPosV = -1;
-    uint32_t bestTransformMode[3] = { 0 };
-
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-
-    if (bCheckFull)
-    {
-        const uint32_t numCoeffPerAbsPartIdxIncrement = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-        const uint32_t qtlayer = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-        TCoeff *coeffCurY = m_qtTempCoeffY[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx);
-        TCoeff *coeffCurU = m_qtTempCoeffCb[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-        TCoeff *coeffCurV = m_qtTempCoeffCr[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-        int trWidth = 0, trHeight = 0, trWidthC = 0, trHeightC = 0;
-        uint32_t absTUPartIdxC = absPartIdx;
-
-        trWidth  = trHeight  = 1 << trSizeLog2;
-        trWidthC = trHeightC = 1 << trSizeCLog2;
-        cu->setTrIdxSubParts(depth - cu->getDepth(0), absPartIdx, depth);
-        uint64_t minCostY = MAX_INT64;
-        uint64_t minCostU = MAX_INT64;
-        uint64_t minCostV = MAX_INT64;
-        bool checkTransformSkipY  = cu->getSlice()->getPPS()->getUseTransformSkip() && trWidth == 4 && trHeight == 4;
-        bool checkTransformSkipUV = cu->getSlice()->getPPS()->getUseTransformSkip() && trWidthC == 4 && trHeightC == 4;
-
-        checkTransformSkipY         &= (!cu->isLosslessCoded(0));
-        checkTransformSkipUV        &= (!cu->isLosslessCoded(0));
-
-        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
-        if (bCodeChroma)
-        {
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setTransformSkipSubParts(0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        if (m_cfg->bEnableRDOQ && curuseRDOQ)
-        {
-            m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, trWidth, TEXT_LUMA);
-        }
-
-        m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-        m_trQuant->selectLambda(TEXT_LUMA);
-
-        absSumY = m_trQuant->transformNxN(cu, resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width, coeffCurY,
-                                          trWidth, TEXT_LUMA, absPartIdx, &lastPosY, false, curuseRDOQ);
-
-        cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-
-        if (bCodeChroma)
-        {
-            if (m_cfg->bEnableRDOQ && curuseRDOQ)
-            {
-                m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, trWidthC, TEXT_CHROMA);
-            }
-            //Cb transform
-            int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-            m_trQuant->selectLambda(TEXT_CHROMA);
-
-            absSumU = m_trQuant->transformNxN(cu, resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurU,
-                                              trWidthC, TEXT_CHROMA_U, absPartIdx, &lastPosU, false, curuseRDOQ);
-            //Cr transform
-            curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-            absSumV = m_trQuant->transformNxN(cu, resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurV,
-                                              trWidthC, TEXT_CHROMA_V, absPartIdx, &lastPosV, false, curuseRDOQ);
-
-            cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        m_entropyCoder->resetBits();
-        m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
-        m_entropyCoder->encodeCoeffNxN(cu, coeffCurY, absPartIdx,  trWidth,  trHeight, depth, TEXT_LUMA);
-        const uint32_t uiSingleBitsY = m_entropyCoder->getNumberOfWrittenBits();
-
-        uint32_t singleBitsU = 0;
-        uint32_t singleBitsV = 0;
-        if (bCodeChroma)
-        {
-            m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, trMode);
-            m_entropyCoder->encodeCoeffNxN(cu, coeffCurU, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_U);
-            singleBitsU = m_entropyCoder->getNumberOfWrittenBits() - uiSingleBitsY;
-
-            m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, trMode);
-            m_entropyCoder->encodeCoeffNxN(cu, coeffCurV, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_V);
-            singleBitsV = m_entropyCoder->getNumberOfWrittenBits() - (uiSingleBitsY + singleBitsU);
-        }
-
-        const uint32_t numSamplesLuma = 1 << (trSizeLog2 << 1);
-        const uint32_t numSamplesChroma = 1 << (trSizeCLog2 << 1);
-
-        ALIGN_VAR_32(static const pixel, zeroPel[MAX_CU_SIZE * MAX_CU_SIZE]) = { 0 };
-
-        int partSize = partitionFromSizes(trWidth, trHeight);
-        uint32_t distY = primitives.sse_sp[partSize](resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width, (pixel*)zeroPel, trWidth);
-
-        if (outZeroDist)
-        {
-            *outZeroDist += distY;
-        }
-        if (absSumY)
-        {
-            int16_t *curResiY = m_qtTempShortYuv[qtlayer].getLumaAddr(absTUPartIdx);
-
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-
-            int scalingListType = 3 + TEXT_LUMA;
-            assert(scalingListType < 6);
-            assert(m_qtTempShortYuv[qtlayer].m_width == MAX_CU_SIZE);
-            m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiY, MAX_CU_SIZE,  coeffCurY, trWidth, scalingListType, false, lastPosY); //this is for inter mode only
-
-            const uint32_t nonZeroDistY = primitives.sse_ss[partSize](resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width, m_qtTempShortYuv[qtlayer].getLumaAddr(absTUPartIdx), MAX_CU_SIZE);
-            if (cu->isLosslessCoded(0))
-            {
-                distY = nonZeroDistY;
-            }
-            else
-            {
-                const uint64_t singleCostY = m_rdCost->calcRdCost(nonZeroDistY, uiSingleBitsY);
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeQtCbfZero(cu, TEXT_LUMA,     trMode);
-                const uint32_t nullBitsY = m_entropyCoder->getNumberOfWrittenBits();
-                const uint64_t nullCostY = m_rdCost->calcRdCost(distY, nullBitsY);
-                if (nullCostY < singleCostY)
-                {
-                    absSumY = 0;
-                    ::memset(coeffCurY, 0, sizeof(TCoeff) * numSamplesLuma);
-                    if (checkTransformSkipY)
-                    {
-                        minCostY = nullCostY;
-                    }
-                }
-                else
-                {
-                    distY = nonZeroDistY;
-                    if (checkTransformSkipY)
-                    {
-                        minCostY = singleCostY;
-                    }
-                }
-            }
-        }
-        else if (checkTransformSkipY)
-        {
-            m_entropyCoder->resetBits();
-            m_entropyCoder->encodeQtCbfZero(cu, TEXT_LUMA, trMode);
-            const uint32_t nullBitsY = m_entropyCoder->getNumberOfWrittenBits();
-            minCostY = m_rdCost->calcRdCost(distY, nullBitsY);
-        }
-
-        if (!absSumY)
-        {
-            int16_t *ptr =  m_qtTempShortYuv[qtlayer].getLumaAddr(absTUPartIdx);
-            assert(m_qtTempShortYuv[qtlayer].m_width == MAX_CU_SIZE);
-
-            assert(trWidth == trHeight);
-            primitives.blockfill_s[(int)g_convertToBit[trWidth]](ptr, MAX_CU_SIZE, 0);
-        }
-
-        uint32_t distU = 0;
-        uint32_t distV = 0;
-
-        int partSizeC = partitionFromSizes(trWidthC, trHeightC);
-        if (bCodeChroma)
-        {
-            distU = m_rdCost->scaleChromaDistCb(primitives.sse_sp[partSizeC](resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth, (pixel*)zeroPel, trWidthC));
-
-            if (outZeroDist)
-            {
-                *outZeroDist += distU;
-            }
-            if (absSumU)
-            {
-                int16_t *pcResiCurrU = m_qtTempShortYuv[qtlayer].getCbAddr(absTUPartIdxC);
-
-                int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_U;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, pcResiCurrU, m_qtTempShortYuv[qtlayer].m_cwidth, coeffCurU, trWidthC, scalingListType, false, lastPosU);
-                uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth,
-                                                             m_qtTempShortYuv[qtlayer].getCbAddr(absTUPartIdxC),
-                                                             m_qtTempShortYuv[qtlayer].m_cwidth);
-                const uint32_t nonZeroDistU = m_rdCost->scaleChromaDistCb(dist);
-
-                if (cu->isLosslessCoded(0))
-                {
-                    distU = nonZeroDistU;
-                }
-                else
-                {
-                    const uint64_t singleCostU = m_rdCost->calcRdCost(nonZeroDistU, singleBitsU);
-                    m_entropyCoder->resetBits();
-                    m_entropyCoder->encodeQtCbfZero(cu, TEXT_CHROMA_U, trMode);
-                    const uint32_t nullBitsU = m_entropyCoder->getNumberOfWrittenBits();
-                    const uint64_t nullCostU = m_rdCost->calcRdCost(distU, nullBitsU);
-                    if (nullCostU < singleCostU)
-                    {
-                        absSumU = 0;
-                        ::memset(coeffCurU, 0, sizeof(TCoeff) * numSamplesChroma);
-                        if (checkTransformSkipUV)
-                        {
-                            minCostU = nullCostU;
-                        }
-                    }
-                    else
-                    {
-                        distU = nonZeroDistU;
-                        if (checkTransformSkipUV)
-                        {
-                            minCostU = singleCostU;
-                        }
-                    }
-                }
-            }
-            else if (checkTransformSkipUV)
-            {
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeQtCbfZero(cu, TEXT_CHROMA_U, trModeC);
-                const uint32_t nullBitsU = m_entropyCoder->getNumberOfWrittenBits();
-                minCostU = m_rdCost->calcRdCost(distU, nullBitsU);
-            }
-            if (!absSumU)
-            {
-                int16_t *ptr = m_qtTempShortYuv[qtlayer].getCbAddr(absTUPartIdxC);
-                const uint32_t stride = m_qtTempShortYuv[qtlayer].m_cwidth;
-                assert(trWidthC == trHeightC);
-                primitives.blockfill_s[(int)g_convertToBit[trWidthC]](ptr, stride, 0);
-            }
-
-            distV = m_rdCost->scaleChromaDistCr(primitives.sse_sp[partSizeC](resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth, (pixel*)zeroPel, trWidthC));
-            if (outZeroDist)
-            {
-                *outZeroDist += distV;
-            }
-            if (absSumV)
-            {
-                int16_t *curResiV = m_qtTempShortYuv[qtlayer].getCrAddr(absTUPartIdxC);
-                int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_V;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiV, m_qtTempShortYuv[qtlayer].m_cwidth, coeffCurV, trWidthC, scalingListType, false, lastPosV);
-                uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth,
-                                                             m_qtTempShortYuv[qtlayer].getCrAddr(absTUPartIdxC),
-                                                             m_qtTempShortYuv[qtlayer].m_cwidth);
-                const uint32_t nonZeroDistV = m_rdCost->scaleChromaDistCr(dist);
-
-                if (cu->isLosslessCoded(0))
-                {
-                    distV = nonZeroDistV;
-                }
-                else
-                {
-                    const uint64_t singleCostV = m_rdCost->calcRdCost(nonZeroDistV, singleBitsV);
-                    m_entropyCoder->resetBits();
-                    m_entropyCoder->encodeQtCbfZero(cu, TEXT_CHROMA_V, trMode);
-                    const uint32_t nullBitsV = m_entropyCoder->getNumberOfWrittenBits();
-                    const uint64_t nullCostV = m_rdCost->calcRdCost(distV, nullBitsV);
-                    if (nullCostV < singleCostV)
-                    {
-                        absSumV = 0;
-                        ::memset(coeffCurV, 0, sizeof(TCoeff) * numSamplesChroma);
-                        if (checkTransformSkipUV)
-                        {
-                            minCostV = nullCostV;
-                        }
-                    }
-                    else
-                    {
-                        distV = nonZeroDistV;
-                        if (checkTransformSkipUV)
-                        {
-                            minCostV = singleCostV;
-                        }
-                    }
-                }
-            }
-            else if (checkTransformSkipUV)
-            {
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeQtCbfZero(cu, TEXT_CHROMA_V, trModeC);
-                const uint32_t nullBitsV = m_entropyCoder->getNumberOfWrittenBits();
-                minCostV = m_rdCost->calcRdCost(distV, nullBitsV);
-            }
-            if (!absSumV)
-            {
-                int16_t *ptr =  m_qtTempShortYuv[qtlayer].getCrAddr(absTUPartIdxC);
-                const uint32_t stride = m_qtTempShortYuv[qtlayer].m_cwidth;
-                assert(trWidthC == trHeightC);
-                primitives.blockfill_s[(int)g_convertToBit[trWidthC]](ptr, stride, 0);
-            }
-        }
-        cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-        if (bCodeChroma)
-        {
-            cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        if (checkTransformSkipY)
-        {
-            uint32_t nonZeroDistY = 0, absSumTransformSkipY;
-            int lastPosTransformSkipY = -1;
-            uint64_t singleCostY = MAX_INT64;
-
-            int16_t *curResiY = m_qtTempShortYuv[qtlayer].getLumaAddr(absTUPartIdx);
-            assert(m_qtTempShortYuv[qtlayer].m_width == MAX_CU_SIZE);
-
-            TCoeff bestCoeffY[32 * 32];
-            memcpy(bestCoeffY, coeffCurY, sizeof(TCoeff) * numSamplesLuma);
-
-            int16_t bestResiY[32 * 32];
-            for (int i = 0; i < trHeight; ++i)
-            {
-                memcpy(bestResiY + i * trWidth, curResiY + i * MAX_CU_SIZE, sizeof(int16_t) * trWidth);
-            }
-
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-
-            cu->setTransformSkipSubParts(1, TEXT_LUMA, absPartIdx, depth);
-
-            if (m_cfg->bEnableRDOQTS)
-            {
-                m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, trWidth, TEXT_LUMA);
-            }
-
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-
-            m_trQuant->selectLambda(TEXT_LUMA);
-            absSumTransformSkipY = m_trQuant->transformNxN(cu, resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width, coeffCurY,
-                                                           trWidth, TEXT_LUMA, absPartIdx, &lastPosTransformSkipY, true, curuseRDOQ);
-            cu->setCbfSubParts(absSumTransformSkipY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-
-            if (absSumTransformSkipY != 0)
-            {
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
-                m_entropyCoder->encodeCoeffNxN(cu, coeffCurY, absPartIdx, trWidth, trHeight, depth, TEXT_LUMA);
-                const uint32_t skipSingleBitsY = m_entropyCoder->getNumberOfWrittenBits();
-
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_LUMA, QP_BD_OFFSET, 0, chFmt);
-
-                int scalingListType = 3 + TEXT_LUMA;
-                assert(scalingListType < 6);
-                assert(m_qtTempShortYuv[qtlayer].m_width == MAX_CU_SIZE);
-
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiY, MAX_CU_SIZE,  coeffCurY, trWidth, scalingListType, true, lastPosTransformSkipY);
-
-                nonZeroDistY = primitives.sse_ss[partSize](resiYuv->getLumaAddr(absTUPartIdx), resiYuv->m_width,
-                                                           m_qtTempShortYuv[qtlayer].getLumaAddr(absTUPartIdx),
-                                                           MAX_CU_SIZE);
-
-                singleCostY = m_rdCost->calcRdCost(nonZeroDistY, skipSingleBitsY);
-            }
-
-            if (!absSumTransformSkipY || minCostY < singleCostY)
-            {
-                cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
-                memcpy(coeffCurY, bestCoeffY, sizeof(TCoeff) * numSamplesLuma);
-                for (int i = 0; i < trHeight; ++i)
-                {
-                    memcpy(curResiY + i * MAX_CU_SIZE, &bestResiY[i * trWidth], sizeof(int16_t) * trWidth);
-                }
-            }
-            else
-            {
-                distY = nonZeroDistY;
-                absSumY = absSumTransformSkipY;
-                bestTransformMode[0] = 1;
-            }
-
-            cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-        }
-
-        if (bCodeChroma && checkTransformSkipUV)
-        {
-            uint32_t nonZeroDistU = 0, nonZeroDistV = 0, absSumTransformSkipU, absSumTransformSkipV;
-            int lastPosTransformSkipU = -1, lastPosTransformSkipV = -1;
-            uint64_t singleCostU = MAX_INT64;
-            uint64_t singleCostV = MAX_INT64;
-
-            int16_t *curResiU = m_qtTempShortYuv[qtlayer].getCbAddr(absTUPartIdxC);
-            int16_t *curResiV = m_qtTempShortYuv[qtlayer].getCrAddr(absTUPartIdxC);
-            uint32_t stride = m_qtTempShortYuv[qtlayer].m_cwidth;
-            TCoeff bestCoeffU[32 * 32], bestCoeffV[32 * 32];
-            memcpy(bestCoeffU, coeffCurU, sizeof(TCoeff) * numSamplesChroma);
-            memcpy(bestCoeffV, coeffCurV, sizeof(TCoeff) * numSamplesChroma);
-
-            int16_t bestResiU[32 * 32], bestResiV[32 * 32];
-            for (int i = 0; i < trHeightC; ++i)
-            {
-                memcpy(&bestResiU[i * trWidthC], curResiU + i * stride, sizeof(int16_t) * trWidthC);
-                memcpy(&bestResiV[i * trWidthC], curResiV + i * stride, sizeof(int16_t) * trWidthC);
-            }
-
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-
-            cu->setTransformSkipSubParts(1, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setTransformSkipSubParts(1, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-
-            if (m_cfg->bEnableRDOQTS)
-            {
-                m_entropyCoder->estimateBit(m_trQuant->m_estBitsSbac, trWidthC, TEXT_CHROMA);
-            }
-
-            int curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-            m_trQuant->selectLambda(TEXT_CHROMA);
-
-            absSumTransformSkipU = m_trQuant->transformNxN(cu, resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurU,
-                                                           trWidthC, TEXT_CHROMA_U, absPartIdx, &lastPosTransformSkipU, true, curuseRDOQ);
-            curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-            m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-            absSumTransformSkipV = m_trQuant->transformNxN(cu, resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth, coeffCurV,
-                                                           trWidthC, TEXT_CHROMA_V, absPartIdx, &lastPosTransformSkipV, true, curuseRDOQ);
-
-            cu->setCbfSubParts(absSumTransformSkipU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumTransformSkipV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-
-            m_entropyCoder->resetBits();
-            singleBitsU = 0;
-
-            if (absSumTransformSkipU)
-            {
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, trMode);
-                m_entropyCoder->encodeCoeffNxN(cu, coeffCurU, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_U);
-                singleBitsU = m_entropyCoder->getNumberOfWrittenBits();
-
-                curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCbQpOffset() + cu->getSlice()->getSliceQpDeltaCb();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_U;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiU, m_qtTempShortYuv[qtlayer].m_cwidth, coeffCurU, trWidthC, scalingListType, true, lastPosTransformSkipU);
-                uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCbAddr(absTUPartIdxC), resiYuv->m_cwidth,
-                                                             m_qtTempShortYuv[qtlayer].getCbAddr(absTUPartIdxC),
-                                                             m_qtTempShortYuv[qtlayer].m_cwidth);
-                nonZeroDistU = m_rdCost->scaleChromaDistCb(dist);
-                singleCostU = m_rdCost->calcRdCost(nonZeroDistU, singleBitsU);
-            }
-
-            if (!absSumTransformSkipU || minCostU < singleCostU)
-            {
-                cu->setTransformSkipSubParts(0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-
-                memcpy(coeffCurU, bestCoeffU, sizeof(TCoeff) * numSamplesChroma);
-                for (int i = 0; i < trHeightC; ++i)
-                {
-                    memcpy(curResiU + i * stride, &bestResiU[i * trWidthC], sizeof(int16_t) * trWidthC);
-                }
-            }
-            else
-            {
-                distU = nonZeroDistU;
-                absSumU = absSumTransformSkipU;
-                bestTransformMode[1] = 1;
-            }
-
-            if (absSumTransformSkipV)
-            {
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, trMode);
-                m_entropyCoder->encodeCoeffNxN(cu, coeffCurV, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_V);
-                singleBitsV = m_entropyCoder->getNumberOfWrittenBits() - singleBitsU;
-
-                curChromaQpOffset = cu->getSlice()->getPPS()->getChromaCrQpOffset() + cu->getSlice()->getSliceQpDeltaCr();
-                m_trQuant->setQPforQuant(cu->getQP(0), TEXT_CHROMA, cu->getSlice()->getSPS()->getQpBDOffsetC(), curChromaQpOffset, chFmt);
-
-                int scalingListType = 3 + TEXT_CHROMA_V;
-                assert(scalingListType < 6);
-                m_trQuant->invtransformNxN(cu->getCUTransquantBypass(absPartIdx), REG_DCT, curResiV, m_qtTempShortYuv[qtlayer].m_cwidth, coeffCurV, trWidthC, scalingListType, true, lastPosTransformSkipV);
-                uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCrAddr(absTUPartIdxC), resiYuv->m_cwidth,
-                                                             m_qtTempShortYuv[qtlayer].getCrAddr(absTUPartIdxC),
-                                                             m_qtTempShortYuv[qtlayer].m_cwidth);
-                nonZeroDistV = m_rdCost->scaleChromaDistCr(dist);
-                singleCostV = m_rdCost->calcRdCost(nonZeroDistV, singleBitsV);
-            }
-
-            if (!absSumTransformSkipV || minCostV < singleCostV)
-            {
-                cu->setTransformSkipSubParts(0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-
-                memcpy(coeffCurV, bestCoeffV, sizeof(TCoeff) * numSamplesChroma);
-                for (int i = 0; i < trHeightC; ++i)
-                {
-                    memcpy(curResiV + i * stride, &bestResiV[i * trWidthC], sizeof(int16_t) * trWidthC);
-                }
-            }
-            else
-            {
-                distV = nonZeroDistV;
-                absSumV = absSumTransformSkipV;
-                bestTransformMode[2] = 1;
-            }
-
-            cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-
-        m_entropyCoder->resetBits();
-
-        if (trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx))
-        {
-            m_entropyCoder->encodeTransformSubdivFlag(0, 5 - trSizeLog2);
-        }
-
-        if (bCodeChroma)
-        {
-            m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, trMode);
-            m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, trMode);
-        }
-
-        m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_LUMA,     trMode);
-        m_entropyCoder->encodeCoeffNxN(cu, coeffCurY, absPartIdx, trWidth, trHeight, depth, TEXT_LUMA);
-
-        if (bCodeChroma)
-        {
-            m_entropyCoder->encodeCoeffNxN(cu, coeffCurU, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_U);
-            m_entropyCoder->encodeCoeffNxN(cu, coeffCurV, absPartIdx, trWidthC, trHeightC, depth, TEXT_CHROMA_V);
-        }
-
-        singleBits = m_entropyCoder->getNumberOfWrittenBits();
-        singleDist = distY + distU + distV;
-        singleCost = m_rdCost->calcRdCost(singleDist, singleBits);
-    }
-
-    // code sub-blocks
-    if (bCheckSplit)
-    {
-        if (bCheckFull)
-        {
-            m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_QT_TRAFO_TEST]);
-            m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-        }
-        uint32_t subdivDist = 0;
-        uint32_t subdivBits = 0;
-        uint64_t subDivCost = 0;
-
-        const uint32_t qPartNumSubdiv = cu->getPic()->getNumPartInCU() >> ((depth + 1) << 1);
-        for (uint32_t i = 0; i < 4; ++i)
-        {
-            uint32_t nsAddr = absPartIdx + i * qPartNumSubdiv;
-            xEstimateResidualQT(cu, absPartIdx + i * qPartNumSubdiv, nsAddr, resiYuv, depth + 1, subDivCost, subdivBits, subdivDist, bCheckFull ? NULL : outZeroDist);
-        }
-
-        uint32_t ycbf = 0;
-        uint32_t ucbf = 0;
-        uint32_t vcbf = 0;
-        for (uint32_t i = 0; i < 4; ++i)
-        {
-            ycbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_LUMA,     trMode + 1);
-            ucbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_U, trMode + 1);
-            vcbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_V, trMode + 1);
-        }
-
-        for (uint32_t i = 0; i < 4 * qPartNumSubdiv; ++i)
-        {
-            cu->getCbf(TEXT_LUMA)[absPartIdx + i] |= ycbf << trMode;
-            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + i] |= ucbf << trMode;
-            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + i] |= vcbf << trMode;
-        }
-
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_ROOT]);
-        m_entropyCoder->resetBits();
-
-        xEncodeResidualQT(cu, absPartIdx, depth, true,  TEXT_LUMA);
-        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_LUMA);
-        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_CHROMA_U);
-        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_CHROMA_V);
-
-        subdivBits = m_entropyCoder->getNumberOfWrittenBits();
-        subDivCost  = m_rdCost->calcRdCost(subdivDist, subdivBits);
-
-        if (ycbf || ucbf || vcbf || !bCheckFull)
-        {
-            if (subDivCost < singleCost)
-            {
-                rdCost += subDivCost;
-                outBits += subdivBits;
-                outDist += subdivDist;
-                return;
-            }
-        }
-
-        cu->setTransformSkipSubParts(bestTransformMode[0], TEXT_LUMA, absPartIdx, depth);
-        if (bCodeChroma)
-        {
-            cu->setTransformSkipSubParts(bestTransformMode[1], TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-            cu->setTransformSkipSubParts(bestTransformMode[2], TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-        }
-        assert(bCheckFull);
-        m_rdGoOnSbacCoder->load(m_rdSbacCoders[depth][CI_QT_TRAFO_TEST]);
-    }
-
-    rdCost += singleCost;
-    outBits += singleBits;
-    outDist += singleDist;
-
-    cu->setTrIdxSubParts(trMode, absPartIdx, depth);
-    cu->setCbfSubParts(absSumY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
-
-    if (bCodeChroma)
-    {
-        cu->setCbfSubParts(absSumU ? setCbf : 0, TEXT_CHROMA_U, absPartIdx, cu->getDepth(0) + trModeC);
-        cu->setCbfSubParts(absSumV ? setCbf : 0, TEXT_CHROMA_V, absPartIdx, cu->getDepth(0) + trModeC);
-    }
-}
-
-void TEncSearch::xEncodeResidualQT(TComDataCU* cu, uint32_t absPartIdx, const uint32_t depth, bool bSubdivAndCbf, TextType ttype)
-{
-    assert(cu->getDepth(0) == cu->getDepth(absPartIdx));
-    const uint32_t curTrMode   = depth - cu->getDepth(0);
-    const uint32_t trMode      = cu->getTransformIdx(absPartIdx);
-    const bool     bSubdiv     = curTrMode != trMode;
-    const uint32_t trSizeLog2  = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> depth] + 2;
-    uint32_t       trSizeCLog2 = g_convertToBit[(cu->getSlice()->getSPS()->getMaxCUSize() >> m_hChromaShift) >> depth] + 2;
-    int            chFmt       = cu->getChromaFormat();
-
-    if (bSubdivAndCbf && trSizeLog2 <= cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() && trSizeLog2 > cu->getQuadtreeTULog2MinSizeInCU(absPartIdx))
-    {
-        m_entropyCoder->encodeTransformSubdivFlag(bSubdiv, 5 - trSizeLog2);
-    }
-
-    assert(cu->getPredictionMode(absPartIdx) != MODE_INTRA);
-
-    bool mCodeAll = true;
-    int width  = 1 << trSizeLog2;
-    int height = 1 << trSizeLog2;
-    const uint32_t numPels = (width >> cu->getHorzChromaShift()) * (height >> cu->getHorzChromaShift());
-    if (numPels < (MIN_TU_SIZE * MIN_TU_SIZE))
-    {
-        mCodeAll = false;
-    }
-
-    if (bSubdivAndCbf)
-    {
-        const bool bFirstCbfOfCU = curTrMode == 0;
-        if (bFirstCbfOfCU || mCodeAll)
-        {
-            if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode - 1))
-            {
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_U, curTrMode);
-            }
-            if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode - 1))
-            {
-                m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_CHROMA_V, curTrMode);
-            }
-        }
-        else
-        {
-            assert(cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode) == cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode - 1));
-            assert(cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode) == cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode - 1));
-        }
-    }
-
-    if (!bSubdiv)
-    {
-        //Luma
-        const uint32_t numCoeffPerAbsPartIdxIncrement = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-        const uint32_t qtlayer = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-        TCoeff *coeffCurY = m_qtTempCoeffY[qtlayer] +  numCoeffPerAbsPartIdxIncrement * absPartIdx;
-
-        //Chroma
-        TCoeff *coeffCurU = m_qtTempCoeffCb[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-        TCoeff *coeffCurV = m_qtTempCoeffCr[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-        bool bCodeChroma = true;
-        if ((trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-        {
-            trSizeCLog2++;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((depth - 1) << 1);
-            bCodeChroma = ((absPartIdx % qpdiv) == 0);
-        }
-
-        if (bSubdivAndCbf)
-        {
-            m_entropyCoder->encodeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
-        }
-        else
-        {
-            if (ttype == TEXT_LUMA && cu->getCbf(absPartIdx, TEXT_LUMA, trMode))
-            {
-                m_entropyCoder->encodeCoeffNxN(cu, coeffCurY, absPartIdx, 1 << trSizeLog2, 1 << trSizeLog2, depth, TEXT_LUMA);
-            }
-            if (bCodeChroma)
-            {
-                if (ttype == TEXT_CHROMA_U && cu->getCbf(absPartIdx, TEXT_CHROMA_U, trMode))
-                {
-                    m_entropyCoder->encodeCoeffNxN(cu, coeffCurU, absPartIdx, 1 << trSizeCLog2, 1 << trSizeCLog2, depth, TEXT_CHROMA_U);
-                }
-                if (ttype == TEXT_CHROMA_V && cu->getCbf(absPartIdx, TEXT_CHROMA_V, trMode))
-                {
-                    m_entropyCoder->encodeCoeffNxN(cu, coeffCurV, absPartIdx, 1 << trSizeCLog2, 1 << trSizeCLog2, depth, TEXT_CHROMA_V);
-                }
-            }
-        }
-    }
-    else
-    {
-        if (bSubdivAndCbf || cu->getCbf(absPartIdx, ttype, curTrMode))
-        {
-            const uint32_t qpartNumSubdiv = cu->getPic()->getNumPartInCU() >> ((depth + 1) << 1);
-            for (uint32_t i = 0; i < 4; ++i)
-            {
-                xEncodeResidualQT(cu, absPartIdx + i * qpartNumSubdiv, depth + 1, bSubdivAndCbf, ttype);
-            }
-        }
-    }
-}
-
-void TEncSearch::xSetResidualQTData(TComDataCU* cu, uint32_t absPartIdx, uint32_t absTUPartIdx, ShortYuv* resiYuv, uint32_t depth, bool bSpatial)
-{
-    assert(cu->getDepth(0) == cu->getDepth(absPartIdx));
-    int            chFmt     = cu->getChromaFormat();
-    const uint32_t curTrMode = depth - cu->getDepth(0);
-    const uint32_t trMode = cu->getTransformIdx(absPartIdx);
-
-    if (curTrMode == trMode)
-    {
-        const uint32_t trSizeLog2 = g_convertToBit[cu->getSlice()->getSPS()->getMaxCUSize() >> depth] + 2;
-        uint32_t  trSizeCLog2     = g_convertToBit[(cu->getSlice()->getSPS()->getMaxCUSize() >> cu->getHorzChromaShift()) >> depth] + 2;
-        const uint32_t qtlayer    = cu->getSlice()->getSPS()->getQuadtreeTULog2MaxSize() - trSizeLog2;
-
-        bool bCodeChroma = true;
-        bool bChromaSame = false;
-        uint32_t trModeC = trMode;
-        if ((trSizeLog2 == 2) && !(chFmt == CHROMA_444))
-        {
-            trSizeCLog2++;
-            trModeC--;
-            uint32_t qpdiv = cu->getPic()->getNumPartInCU() >> ((cu->getDepth(0) + trModeC) << 1);
-            bCodeChroma = ((absPartIdx % qpdiv) == 0);
-            bChromaSame = true;
-        }
-
-        if (bSpatial)
-        {
-            int trWidth  = 1 << trSizeLog2;
-            int trHeight = 1 << trSizeLog2;
-            m_qtTempShortYuv[qtlayer].copyPartToPartLuma(resiYuv, absTUPartIdx, trWidth, trHeight);
-
-            if (bCodeChroma)
-            {
-                m_qtTempShortYuv[qtlayer].copyPartToPartChroma(resiYuv, absPartIdx, 1 << trSizeLog2, bChromaSame);
-            }
-        }
-        else
-        {
-            uint32_t numCoeffPerAbsPartIdxIncrement = cu->getSlice()->getSPS()->getMaxCUSize() * cu->getSlice()->getSPS()->getMaxCUSize() >> (cu->getSlice()->getSPS()->getMaxCUDepth() << 1);
-            uint32_t numCoeffY = (1 << (trSizeLog2 << 1));
-            TCoeff* coeffSrcY = m_qtTempCoeffY[qtlayer] +  numCoeffPerAbsPartIdxIncrement * absPartIdx;
-            TCoeff* coeffDstY = cu->getCoeffY() + numCoeffPerAbsPartIdxIncrement * absPartIdx;
-            ::memcpy(coeffDstY, coeffSrcY, sizeof(TCoeff) * numCoeffY);
-            if (bCodeChroma)
-            {
-                uint32_t numCoeffC = (1 << (trSizeCLog2 << 1));
-                TCoeff* coeffSrcU = m_qtTempCoeffCb[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-                TCoeff* coeffSrcV = m_qtTempCoeffCr[qtlayer] + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-                TCoeff* coeffDstU = cu->getCoeffCb() + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-                TCoeff* coeffDstV = cu->getCoeffCr() + (numCoeffPerAbsPartIdxIncrement * absPartIdx >> (m_hChromaShift + m_vChromaShift));
-                ::memcpy(coeffDstU, coeffSrcU, sizeof(TCoeff) * numCoeffC);
-                ::memcpy(coeffDstV, coeffSrcV, sizeof(TCoeff) * numCoeffC);
-            }
-        }
-    }
-    else
-    {
-        const uint32_t qPartNumSubdiv = cu->getPic()->getNumPartInCU() >> ((depth + 1) << 1);
-        for (uint32_t i = 0; i < 4; ++i)
-        {
-            uint32_t nsAddr = absPartIdx + i * qPartNumSubdiv;
-            xSetResidualQTData(cu, absPartIdx + i * qPartNumSubdiv, nsAddr, resiYuv, depth + 1, bSpatial);
-        }
-    }
-}
-
-uint32_t TEncSearch::xModeBitsIntra(TComDataCU* cu, uint32_t mode, uint32_t partOffset, uint32_t depth, uint32_t initTrDepth)
-{
-    // Reload only contexts required for coding intra mode information
-    m_rdGoOnSbacCoder->loadIntraDirModeLuma(m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-    cu->setLumaIntraDirSubParts(mode, partOffset, depth + initTrDepth);
-
-    m_entropyCoder->resetBits();
-    m_entropyCoder->encodeIntraDirModeLuma(cu, partOffset);
-
-    return m_entropyCoder->getNumberOfWrittenBits();
-}
-
-uint32_t TEncSearch::xUpdateCandList(uint32_t mode, uint64_t cost, uint32_t fastCandNum, uint32_t* CandModeList, uint64_t* CandCostList)
-{
-    uint32_t i;
-    uint32_t shift = 0;
-
-    while (shift < fastCandNum && cost < CandCostList[fastCandNum - 1 - shift])
-    {
-        shift++;
-    }
-
-    if (shift != 0)
-    {
-        for (i = 1; i < shift; i++)
-        {
-            CandModeList[fastCandNum - i] = CandModeList[fastCandNum - 1 - i];
-            CandCostList[fastCandNum - i] = CandCostList[fastCandNum - 1 - i];
-        }
-
-        CandModeList[fastCandNum - shift] = mode;
-        CandCostList[fastCandNum - shift] = cost;
-        return 1;
-    }
-
-    return 0;
-}
-
-/** add inter-prediction syntax elements for a CU block
- * \param cu
- */
-uint32_t TEncSearch::xSymbolBitsInter(TComDataCU* cu)
-{
-    if (cu->getMergeFlag(0) && cu->getPartitionSize(0) == SIZE_2Nx2N && !cu->getQtRootCbf(0))
-    {
-        cu->setSkipFlagSubParts(true, 0, cu->getDepth(0));
-
-        m_entropyCoder->resetBits();
-        if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-        {
-            m_entropyCoder->encodeCUTransquantBypassFlag(cu, 0, true);
-        }
-        m_entropyCoder->encodeSkipFlag(cu, 0, true);
-        m_entropyCoder->encodeMergeIndex(cu, 0, true);
-        return m_entropyCoder->getNumberOfWrittenBits();
-    }
-    else
-    {
-        m_entropyCoder->resetBits();
-        if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-        {
-            m_entropyCoder->encodeCUTransquantBypassFlag(cu, 0, true);
-        }
-        m_entropyCoder->encodeSkipFlag(cu, 0, true);
-        m_entropyCoder->encodePredMode(cu, 0, true);
-        m_entropyCoder->encodePartSize(cu, 0, cu->getDepth(0), true);
-        m_entropyCoder->encodePredInfo(cu, 0, true);
-        bool bDummy = false;
-        m_entropyCoder->encodeCoeff(cu, 0, cu->getDepth(0), cu->getCUSize(0), cu->getCUSize(0), bDummy);
-        return m_entropyCoder->getNumberOfWrittenBits();
-    }
-}
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibEncoder/TEncSearch.h Deleted

@@ -1,268 +0,0 @@
-/* The copyright in this software is being made available under the BSD
- * License, included below. This software may be subject to other third party
- * and contributor rights, including patent rights, and no such rights are
- * granted under this license.
- *
- * Copyright (c) 2010-2013, ITU/ISO/IEC
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- *  * Redistributions of source code must retain the above copyright notice,
- *    this list of conditions and the following disclaimer.
- *  * Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
- *    be used to endorse or promote products derived from this software without
- *    specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
- * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- * THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-/** \file     TEncSearch.h
-    \brief    encoder search class (header)
-*/
-
-#ifndef X265_TENCSEARCH_H
-#define X265_TENCSEARCH_H
-
-// Include files
-#include "TLibCommon/TComYuv.h"
-#include "TLibCommon/TComMotionInfo.h"
-#include "TLibCommon/TComPattern.h"
-#include "TLibCommon/TComPrediction.h"
-#include "TLibCommon/TComTrQuant.h"
-#include "TLibCommon/TComPic.h"
-#include "TEncEntropy.h"
-#include "TEncSbac.h"
-#include "primitives.h"
-#include "bitcost.h"
-#include "motion.h"
-
-#define MVP_IDX_BITS 1
-
-//! \ingroup TLibEncoder
-//! \{
-
-namespace x265 {
-// private namespace
-#define DONT_SPLIT            0
-#define VERTICAL_SPLIT        1
-#define QUAD_SPLIT            2
-#define NUMBER_OF_SPLIT_MODES 3
-
-static const uint32_t partIdxStepShift[NUMBER_OF_SPLIT_MODES] = { 0, 1, 2 };
-
-class TEncCu;
-class Encoder;
-
-struct MotionData
-{
-    MV  mv;
-    MV  mvp;
-    int mvpIdx;
-    int ref;
-    uint32_t cost;
-    int bits;
-};
-
-struct MergeData
-{
-    /* merge candidate data, cached between calls to xMergeEstimation */
-    TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS << 1];
-    uint8_t     interDirNeighbours[MRG_MAX_NUM_CANDS];
-    int         numValidMergeCand;
-
-    /* data updated for each partition */
-    uint32_t    absPartIdx;
-    int         width;
-    int         height;
-
-    /* outputs */
-    TComMvField mvField[2];
-    uint32_t    interDir;
-    uint32_t    index;
-    uint32_t    bits;
-};
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// encoder search class
-class TEncSearch : public TComPrediction
-{
-public:
-
-    MotionEstimate   m_me;
-    MotionReference* m_mref[2][MAX_NUM_REF + 1];
-
-protected:
-
-    ShortYuv*       m_qtTempShortYuv;
-
-    TCoeff**        m_qtTempCoeffY;
-    TCoeff**        m_qtTempCoeffCb;
-    TCoeff**        m_qtTempCoeffCr;
-    uint8_t*        m_qtTempTrIdx;
-    uint8_t*        m_qtTempCbf[3];
-
-    TCoeff*         m_qtTempTUCoeffY;
-    TCoeff*         m_qtTempTUCoeffCb;
-    TCoeff*         m_qtTempTUCoeffCr;
-    uint8_t*        m_qtTempTransformSkipFlag[3];
-    TComYuv         m_qtTempTransformSkipYuv;
-
-    // interface to option
-    Encoder*        m_cfg;
-
-    // interface to classes
-    TComTrQuant*    m_trQuant;
-    TComRdCost*     m_rdCost;
-    TEncEntropy*    m_entropyCoder;
-
-    // ME parameters
-    int             m_refLagPixels;
-
-    // Color space parameters
-    uint32_t        m_section;
-    uint32_t        m_splitMode;
-    uint32_t        m_absPartIdxTURelCU;
-    uint32_t        m_absPartIdxStep;
-    uint32_t        m_partOffset;
-
-public:
-
-    TEncSbac***     m_rdSbacCoders;
-    TEncSbac*       m_rdGoOnSbacCoder;
-
-    void setRDSbacCoder(TEncSbac*** rdSbacCoders) { m_rdSbacCoders = rdSbacCoders; }
-
-    void setEntropyCoder(TEncEntropy* entropyCoder) { m_entropyCoder = entropyCoder; }
-
-    void setRDGoOnSbacCoder(TEncSbac* rdGoOnSbacCoder) { m_rdGoOnSbacCoder = rdGoOnSbacCoder; }
-
-    void setQPLambda(int QP, double lambdaLuma, double lambdaChroma);
-
-    TEncSearch();
-    virtual ~TEncSearch();
-
-    bool init(Encoder* cfg, TComRdCost* rdCost, TComTrQuant *trQuant);
-
-public:
-
-    uint32_t xModeBitsIntra(TComDataCU* cu, uint32_t mode, uint32_t partOffset, uint32_t depth, uint32_t initTrDepth);
-    uint32_t xUpdateCandList(uint32_t mode, uint64_t cost, uint32_t fastCandNum, uint32_t* CandModeList, uint64_t* CandCostList);
-
-    void estIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
-
-    void getBestIntraModeChroma(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv);
-
-    void estIntraPredChromaQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
-                              TComYuv* reconYuv);
-
-    /// encoder estimation - inter prediction (non-skip)
-    bool predInterSearch(TComDataCU* cu, TComYuv* predYuv, bool bMergeOnly, bool bChroma);
-
-    /// encode residual and compute rd-cost for inter mode
-    void encodeResAndCalcRdInterCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, ShortYuv* bestResiYuv,
-                                   TComYuv* reconYuv, bool bSkipRes, bool curUseRDOQ);
-
-    void xEncPCM(TComDataCU* cu, uint32_t absPartIdx, Pel* fenc, Pel* pcm, Pel* pred, int16_t* residual, Pel* recon, uint32_t stride,
-                 uint32_t width, uint32_t height, TextType ttype);
-
-    void IPCMSearch(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
-
-    void xRecurIntraCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
-                             TComYuv* predYuv, ShortYuv* resiYuv, uint32_t& distY, bool bCheckFirst,
-                             uint64_t& dRDCost);
-    void xSetIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv);
-
-    void generateCoeffRecon(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, bool skipRes);
-
-    void xEstimateResidualQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t absTUPartIdx, ShortYuv* resiYuv, uint32_t depth,
-                             uint64_t &rdCost, uint32_t &outBits, uint32_t &outDist, uint32_t *puiZeroDist, bool curUseRDOQ = true);
-    void xSetResidualQTData(TComDataCU* cu, uint32_t absPartIdx, uint32_t absTUPartIdx, ShortYuv* resiYuv, uint32_t depth, bool bSpatial);
-
-    void residualTransformQuantInter(TComDataCU* cu, uint32_t absPartIdx, uint32_t absTUPartIdx, ShortYuv* resiYuv, uint32_t depth, bool curUseRDOQ = true);
-
-    // -------------------------------------------------------------------------------------------------------------------
-    // compute symbol bits
-    // -------------------------------------------------------------------------------------------------------------------
-
-    uint32_t xSymbolBitsInter(TComDataCU* cu);
-    bool isNextSection();
-    bool isLastSection();
-
-protected:
-
-    // --------------------------------------------------------------------------------------------
-    // Intra search
-    // --------------------------------------------------------------------------------------------
-
-    void xEncSubdivCbfQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma);
-
-    void xEncCoeffQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TextType ttype);
-    void xEncIntraHeader(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma);
-    uint32_t xGetIntraBitsQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, bool bLuma, bool bChroma);
-    uint32_t xGetIntraBitsQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t uiChromaId);
-    void xIntraCodingLumaBlk(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv,
-                             ShortYuv* resiYuv, uint32_t& outDist, bool bReusePred = false);
-
-    void xIntraCodingChromaBlk(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv,
-                               ShortYuv* resiYuv, uint32_t& outDist, uint32_t uiChromaId, bool bReusePred = false);
-
-    void xRecurIntraChromaCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
-                                   TComYuv* predYuv, ShortYuv* resiYuv, uint32_t& outDist);
-
-    void residualTransformQuantIntra(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
-                                     TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
-    void residualQTIntrachroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
-                               TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
-
-    void xSetIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv);
-
-    void xStoreIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx);
-    void xLoadIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx);
-    void xStoreIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t stateU0V1Both2);
-    void xLoadIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t stateU0V1Both2);
-
-    // --------------------------------------------------------------------------------------------
-    // Inter search (AMP)
-    // --------------------------------------------------------------------------------------------
-
-    void xCheckBestMVP(AMVPInfo* amvpInfo, MV cMv, MV& mvPred, int& mvpIdx,
-                       uint32_t& outBits, uint32_t& outCost);
-
-    void xGetBlkBits(PartSize cuMode, bool bPSlice, int partIdx, uint32_t lastMode, uint32_t blockBit[3]);
-
-    uint32_t xMergeEstimation(TComDataCU* cu, int partIdx, MergeData& m);
-
-    // -------------------------------------------------------------------------------------------------------------------
-    // motion estimation
-    // -------------------------------------------------------------------------------------------------------------------
-
-    void xSetSearchRange(TComDataCU* cu, MV mvp, int merange, MV& mvmin, MV& mvmax);
-
-    // -------------------------------------------------------------------------------------------------------------------
-    // T & Q & Q-1 & T-1
-    // -------------------------------------------------------------------------------------------------------------------
-
-    void xEncodeResidualQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bSubdivAndCbf, TextType ttype);
-};
-}
-//! \}
-
-#endif // ifndef X265_TENCSEARCH_H

x265-hg20140324.tar.bz2/source/common/vec/blockcopy-sse3.cpp Deleted

@@ -1,229 +0,0 @@
-/*****************************************************************************
- * Copyright (C) 2013 x265 project
- *
- * Authors: Steve Borho <steve@borho.org>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
- *
- * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com
- *****************************************************************************/
-
-#include "TLibCommon/TComRom.h"
-#include "primitives.h"
-#include <xmmintrin.h> // SSE
-#include <pmmintrin.h> // SSE3
-
-namespace {
-#if HIGH_BIT_DEPTH
-void blockcopy_pp(int bx, int by, pixel *dst, intptr_t dstride, pixel *src, intptr_t sstride)
-{
-    if ((bx & 7) || (((size_t)dst | (size_t)src | sstride | dstride) & 15))
-    {
-        // slow path, irregular memory alignments or sizes
-        for (int y = 0; y < by; y++)
-        {
-            memcpy(dst, src, bx * sizeof(pixel));
-            src += sstride;
-            dst += dstride;
-        }
-    }
-    else
-    {
-        // fast path, multiples of 8 pixel wide blocks
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x += 8)
-            {
-                __m128i word = _mm_load_si128((__m128i const*)(src + x));
-                _mm_store_si128((__m128i*)&dst[x], word);
-            }
-
-            src += sstride;
-            dst += dstride;
-        }
-    }
-}
-
-#else // if HIGH_BIT_DEPTH
-void blockcopy_pp(int bx, int by, pixel *dst, intptr_t dstride, pixel *src, intptr_t sstride)
-{
-    size_t aligncheck = (size_t)dst | (size_t)src | bx | sstride | dstride;
-
-    if (!(aligncheck & 15))
-    {
-        // fast path, multiples of 16 pixel wide blocks
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x += 16)
-            {
-                __m128i word0 = _mm_load_si128((__m128i const*)(src + x)); // load block of 16 byte from src
-                _mm_store_si128((__m128i*)&dst[x], word0); // store block into dst
-            }
-
-            src += sstride;
-            dst += dstride;
-        }
-    }
-    else
-    {
-        // slow path, irregular memory alignments or sizes
-        for (int y = 0; y < by; y++)
-        {
-            memcpy(dst, src, bx * sizeof(pixel));
-            src += sstride;
-            dst += dstride;
-        }
-    }
-}
-
-void blockcopy_ps(int bx, int by, pixel *dst, intptr_t dstride, int16_t *src, intptr_t sstride)
-{
-    size_t aligncheck = (size_t)dst | (size_t)src | bx | sstride | dstride;
-
-    if (!(aligncheck & 15))
-    {
-        // fast path, multiples of 16 pixel wide blocks
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x += 16)
-            {
-                __m128i word0 = _mm_load_si128((__m128i const*)(src + x));       // load block of 16 byte from src
-                __m128i word1 = _mm_load_si128((__m128i const*)(src + x + 8));
-
-                __m128i mask = _mm_set1_epi32(0x00FF00FF);                  // mask for low bytes
-                __m128i low_mask = _mm_and_si128(word0, mask);              // bytes of low
-                __m128i high_mask = _mm_and_si128(word1, mask);             // bytes of high
-                __m128i word01 = _mm_packus_epi16(low_mask, high_mask);     // unsigned pack
-                _mm_store_si128((__m128i*)&dst[x], word01);                 // store block into dst
-            }
-
-            src += sstride;
-            dst += dstride;
-        }
-    }
-    else
-    {
-        // slow path, irregular memory alignments or sizes
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x++)
-            {
-                dst[x] = (pixel)src[x];
-            }
-
-            src += sstride;
-            dst += dstride;
-        }
-    }
-}
-
-#endif /* if HIGH_BIT_DEPTH */
-
-void pixeladd_ss(int bx, int by, int16_t *dst, intptr_t dstride, int16_t *src0, int16_t *src1, intptr_t sstride0, intptr_t sstride1)
-{
-    size_t aligncheck = (size_t)dst | (size_t)src0 | sstride0 | sstride1 | dstride;
-
-    if (!(aligncheck & 15) && !(bx & 7))
-    {
-        __m128i maxval = _mm_set1_epi16((1 << X265_DEPTH) - 1);
-        __m128i zero = _mm_setzero_si128();
-
-        // fast path, multiples of 8 pixel wide blocks
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x += 8)
-            {
-                __m128i word0, word1, sum;
-
-                word0 = _mm_load_si128((__m128i*)(src0 + x));    // load 16 bytes from src1
-                word1 = _mm_load_si128((__m128i*)(src1 + x));    // load 16 bytes from src2
-
-                sum = _mm_adds_epi16(word0, word1);
-                sum = _mm_max_epi16(sum, zero);
-                sum = _mm_min_epi16(sum, maxval);
-
-                _mm_store_si128((__m128i*)&dst[x], sum);    // store block into dst
-            }
-
-            src0 += sstride0;
-            src1 += sstride1;
-            dst += dstride;
-        }
-    }
-    else if (!(bx & 7))
-    {
-        __m128i maxval = _mm_set1_epi16((1 << X265_DEPTH) - 1);
-        __m128i zero = _mm_setzero_si128();
-
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x += 8)
-            {
-                __m128i word0, word1, sum;
-
-                word0 = _mm_load_si128((__m128i*)(src0 + x));    // load 16 bytes from src1
-                word1 = _mm_load_si128((__m128i*)(src1 + x));    // load 16 bytes from src2
-
-                sum = _mm_adds_epi16(word0, word1);
-                sum = _mm_max_epi16(sum, zero);
-                sum = _mm_min_epi16(sum, maxval);
-
-                _mm_store_si128((__m128i*)&dst[x], sum);    // store block into dst
-            }
-
-            src0 += sstride0;
-            src1 += sstride1;
-            dst += dstride;
-        }
-    }
-    else
-    {
-        int tmp;
-        int max = (1 << X265_DEPTH) - 1;
-        // slow path, irregular memory alignments or sizes
-        for (int y = 0; y < by; y++)
-        {
-            for (int x = 0; x < bx; x++)
-            {
-                tmp = src0[x] + src1[x];
-                tmp = tmp < 0 ? 0 : tmp;
-                tmp = tmp > max ? max : tmp;
-                dst[x] = (int16_t)tmp;
-            }
-
-            src0 += sstride0;
-            src1 += sstride1;
-            dst += dstride;
-        }
-    }
-}
-}
-
-namespace x265 {
-void Setup_Vec_BlockCopyPrimitives_sse3(EncoderPrimitives &p)
-{
-#if HIGH_BIT_DEPTH
-    p.blockcpy_pp = blockcopy_pp;
-    p.blockcpy_ps = (blockcpy_ps_t)blockcopy_pp;
-    // At high bit depth, a pixel is a short
-    p.pixeladd_ss = pixeladd_ss;
-#else
-    p.blockcpy_pp = blockcopy_pp;
-    p.blockcpy_ps = blockcopy_ps;
-    p.pixeladd_ss = pixeladd_ss;
-#endif // if HIGH_BIT_DEPTH
-}
-}

x265-hg20140324.tar.bz2/source/encoder/compress.cpp Deleted

@@ -1,958 +0,0 @@
-/*****************************************************************************
- * Copyright (C) 2013 x265 project
- *
- * Authors: Deepthi Nandakumar <deepthi@multicorewareinc.com>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
- *
- * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
- *****************************************************************************/
-
-#include "TLibEncoder/TEncCu.h"
-#include "encoder.h"
-#include "common.h"
-
-/* Lambda Partition Select adjusts the threshold value for Early Exit in No-RDO flow */
-#define LAMBDA_PARTITION_SELECT     0.9
-#define EARLY_EXIT                  1
-#define TOPSKIP                     1
-
-using namespace x265;
-
-void TEncCu::xEncodeIntraInInter(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv,  ShortYuv* outResiYuv, TComYuv* outReconYuv)
-{
-    uint64_t puCost = 0;
-    uint32_t puDistY = 0;
-    uint32_t depth = cu->getDepth(0);
-    uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
-
-    // set context models
-    m_search->m_rdGoOnSbacCoder->load(m_search->m_rdSbacCoders[depth][CI_CURR_BEST]);
-
-    m_search->xRecurIntraCodingQT(cu, initTrDepth, 0, fencYuv, predYuv, outResiYuv, puDistY, false, puCost);
-    m_search->xSetIntraResultQT(cu, initTrDepth, 0, outReconYuv);
-
-    //=== update PU data ====
-    cu->copyToPic(cu->getDepth(0), 0, initTrDepth);
-
-    //===== set distortion (rate and r-d costs are determined later) =====
-    cu->m_totalDistortion = puDistY;
-
-    m_search->estIntraPredChromaQT(cu, fencYuv, predYuv, outResiYuv, outReconYuv);
-    m_entropyCoder->resetBits();
-    if (cu->getSlice()->getPPS()->getTransquantBypassEnableFlag())
-    {
-        m_entropyCoder->encodeCUTransquantBypassFlag(cu, 0, true);
-    }
-    m_entropyCoder->encodeSkipFlag(cu, 0, true);
-    m_entropyCoder->encodePredMode(cu, 0, true);
-    m_entropyCoder->encodePartSize(cu, 0, depth, true);
-    m_entropyCoder->encodePredInfo(cu, 0, true);
-    m_entropyCoder->encodeIPCMInfo(cu, 0, true);
-
-    // Encode Coefficients
-    bool bCodeDQP = getdQPFlag();
-    m_entropyCoder->encodeCoeff(cu, 0, depth, cu->getCUSize(0), cu->getCUSize(0), bCodeDQP);
-    
-    m_rdGoOnSbacCoder->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-
-    cu->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
-    cu->m_totalCost = m_rdCost->calcRdCost(cu->m_totalDistortion, cu->m_totalBits);
-}
-
-void TEncCu::xComputeCostIntraInInter(TComDataCU* cu, PartSize partSize)
-{
-    uint32_t depth = cu->getDepth(0);
-
-    cu->setPartSizeSubParts(partSize, 0, depth);
-    cu->setPredModeSubParts(MODE_INTRA, 0, depth);
-    cu->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
-    uint32_t width       = cu->getCUSize(0) >> initTrDepth;
-    uint32_t partOffset  = 0;
-
-    // Reference sample smoothing
-    TComPattern::initAdiPattern(cu, partOffset, initTrDepth, m_search->m_predBuf, m_search->m_predBufStride,
-                                m_search->m_predBufHeight, m_search->m_refAbove, m_search->m_refLeft,
-                                m_search->m_refAboveFlt, m_search->m_refLeftFlt);
-
-    Pel* fenc   = m_origYuv[depth]->getLumaAddr();
-    uint32_t stride = m_modePredYuv[5][depth]->getStride();
-
-    Pel *above         = m_search->m_refAbove    + width - 1;
-    Pel *aboveFiltered = m_search->m_refAboveFlt + width - 1;
-    Pel *left          = m_search->m_refLeft     + width - 1;
-    Pel *leftFiltered  = m_search->m_refLeftFlt  + width - 1;
-    int sad, bsad;
-    uint32_t bits, bbits, mode, bmode;
-    uint64_t cost, bcost;
-
-    // 33 Angle modes once
-    ALIGN_VAR_32(Pel, buf_trans[32 * 32]);
-    ALIGN_VAR_32(Pel, tmp[33 * 32 * 32]);
-    int scaleWidth = width;
-    int scaleStride = stride;
-    int costMultiplier = 1;
-
-    if (width > 32)
-    {
-        // origin is 64x64, we scale to 32x32 and setup required parameters
-        ALIGN_VAR_32(Pel, bufScale[32 * 32]);
-        primitives.scale2D_64to32(bufScale, fenc, stride);
-        fenc = bufScale;
-
-        // reserve space in case primitives need to store data in above
-        // or left buffers
-        Pel _above[4 * 32 + 1];
-        Pel _left[4 * 32 + 1];
-        Pel *aboveScale  = _above + 2 * 32;
-        Pel *leftScale   = _left + 2 * 32;
-        aboveScale[0] = leftScale[0] = above[0];
-        primitives.scale1D_128to64(aboveScale + 1, above + 1, 0);
-        primitives.scale1D_128to64(leftScale + 1, left + 1, 0);
-
-        scaleWidth = 32;
-        scaleStride = 32;
-        costMultiplier = 4;
-
-        // Filtered and Unfiltered refAbove and refLeft pointing to above and left.
-        above         = aboveScale;
-        left          = leftScale;
-        aboveFiltered = aboveScale;
-        leftFiltered  = leftScale;
-    }
-
-    int log2SizeMinus2 = g_convertToBit[scaleWidth];
-    pixelcmp_t sa8d = primitives.sa8d[log2SizeMinus2];
-
-    // DC
-    primitives.intra_pred[log2SizeMinus2][DC_IDX](tmp, scaleStride, left, above, 0, (scaleWidth <= 16));
-    bsad = costMultiplier * sa8d(fenc, scaleStride, tmp, scaleStride);
-    bmode = mode = DC_IDX;
-    bbits  = m_search->xModeBitsIntra(cu, mode, partOffset, depth, initTrDepth);
-    bcost = m_rdCost->calcRdSADCost(bsad, bbits);
-
-    Pel *abovePlanar   = above;
-    Pel *leftPlanar    = left;
-
-    if (width >= 8 && width <= 32)
-    {
-        abovePlanar = aboveFiltered;
-        leftPlanar  = leftFiltered;
-    }
-
-    // PLANAR
-    primitives.intra_pred[log2SizeMinus2][PLANAR_IDX](tmp, scaleStride, leftPlanar, abovePlanar, 0, 0);
-    sad = costMultiplier * sa8d(fenc, scaleStride, tmp, scaleStride);
-    mode = PLANAR_IDX;
-    bits = m_search->xModeBitsIntra(cu, mode, partOffset, depth, initTrDepth);
-    cost = m_rdCost->calcRdSADCost(sad, bits);
-    COPY4_IF_LT(bcost, cost, bmode, mode, bsad, sad, bbits, bits);
-
-    // Transpose NxN
-    primitives.transpose[log2SizeMinus2](buf_trans, fenc, scaleStride);
-
-    primitives.intra_pred_allangs[log2SizeMinus2](tmp, above, left, aboveFiltered, leftFiltered, (scaleWidth <= 16));
-
-    for (mode = 2; mode < 35; mode++)
-    {
-        bool modeHor = (mode < 18);
-        Pel *cmp = (modeHor ? buf_trans : fenc);
-        intptr_t srcStride = (modeHor ? scaleWidth : scaleStride);
-        sad  = costMultiplier * sa8d(cmp, srcStride, &tmp[(mode - 2) * (scaleWidth * scaleWidth)], scaleWidth);
-        bits = m_search->xModeBitsIntra(cu, mode, partOffset, depth, initTrDepth);
-        cost = m_rdCost->calcRdSADCost(sad, bits);
-        COPY4_IF_LT(bcost, cost, bmode, mode, bsad, sad, bbits, bits);
-    }
-
-    cu->m_totalBits = bbits;
-    cu->m_totalDistortion = bsad;
-    cu->m_totalCost = bcost;
-
-    cu->m_totalBits = 0;
-    cu->m_totalCost = m_rdCost->calcRdSADCost(cu->m_totalDistortion, cu->m_totalBits);
-
-    // generate predYuv for the best mode
-    cu->setLumaIntraDirSubParts(bmode, partOffset, depth + initTrDepth);
-}
-
-/** check RD costs for a CU block encoded with merge */
-void TEncCu::xComputeCostInter(TComDataCU* outTempCU, TComYuv* outPredYuv, PartSize partSize, bool bUseMRG)
-{
-    UChar depth = outTempCU->getDepth(0);
-
-    outTempCU->setPartSizeSubParts(partSize, 0, depth);
-    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-
-    // do motion compensation only for Luma since luma cost alone is calculated
-    outTempCU->m_totalBits = 0;
-    if (m_search->predInterSearch(outTempCU, outPredYuv, bUseMRG, false))
-    {
-        int part = g_convertToBit[outTempCU->getCUSize(0)];
-        uint32_t distortion = primitives.sa8d[part](m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
-                                                    outPredYuv->getLumaAddr(), outPredYuv->getStride());
-        outTempCU->m_totalDistortion = distortion;
-        outTempCU->m_totalCost = m_rdCost->calcRdSADCost(distortion, outTempCU->m_totalBits);
-    }
-    else
-    {
-        outTempCU->m_totalDistortion = MAX_UINT;
-        outTempCU->m_totalCost = MAX_INT64;
-    }
-}
-
-void TEncCu::xComputeCostMerge2Nx2N(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComYuv*& bestPredYuv, TComYuv*& yuvReconBest)
-{
-    assert(outTempCU->getSlice()->getSliceType() != I_SLICE);
-    TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS << 1]; // double length for mv of both lists
-    UChar interDirNeighbours[MRG_MAX_NUM_CANDS];
-    int numValidMergeCand = 0;
-
-    for (uint32_t i = 0; i < outTempCU->getSlice()->getMaxNumMergeCand(); ++i)
-    {
-        interDirNeighbours[i] = 0;
-    }
-
-    UChar depth = outTempCU->getDepth(0);
-    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to LCU level
-    outTempCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-    outTempCU->getInterMergeCandidates(0, 0, mvFieldNeighbours, interDirNeighbours, numValidMergeCand);
-    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
-    outTempCU->setMergeFlag(0, true);
-
-    outBestCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to LCU level
-    outBestCU->setCUTransquantBypassSubParts(m_CUTransquantBypassFlagValue, 0, depth);
-    outBestCU->setPredModeSubParts(MODE_INTER, 0, depth);
-    outBestCU->setMergeFlag(0, true);
-
-    int part = g_convertToBit[outTempCU->getCUSize(0)];
-    int bestMergeCand = -1;
-    uint32_t bitsCand = 0;
-
-    for (int mergeCand = 0; mergeCand < numValidMergeCand; ++mergeCand)
-    {
-        if (m_param->frameNumThreads <= 1 ||
-            (mvFieldNeighbours[0 + 2 * mergeCand].mv.y < (m_param->searchRange + 1) * 4 &&
-             mvFieldNeighbours[1 + 2 * mergeCand].mv.y < (m_param->searchRange + 1) * 4))
-        {
-            // set MC parameters, interprets depth relative to LCU level
-            outTempCU->setMergeIndex(0, mergeCand);
-            outTempCU->setInterDirSubParts(interDirNeighbours[mergeCand], 0, 0, depth);
-            outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[0 + 2 * mergeCand], SIZE_2Nx2N, 0, 0); // interprets depth relative to rpcTempCU level
-            outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[1 + 2 * mergeCand], SIZE_2Nx2N, 0, 0); // interprets depth relative to rpcTempCU level
-
-            // do MC only for Luma part
-            m_search->motionCompensation(outTempCU, m_tmpPredYuv[depth], REF_PIC_LIST_X, 0, true, false);
-            bitsCand = mergeCand + 1;
-            if (mergeCand == (int)m_param->maxNumMergeCand - 1)
-            {
-                bitsCand--;
-            }
-            outTempCU->m_totalBits = bitsCand;
-            outTempCU->m_totalDistortion = primitives.sa8d[part](m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
-                                                                 m_tmpPredYuv[depth]->getLumaAddr(), m_tmpPredYuv[depth]->getStride());
-            outTempCU->m_totalCost = m_rdCost->calcRdSADCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-            if (outTempCU->m_totalCost < outBestCU->m_totalCost)
-            {
-                bestMergeCand = mergeCand;
-                TComDataCU* tmp = outTempCU;
-                outTempCU = outBestCU;
-                outBestCU = tmp;
-                // Change Prediction data
-                TComYuv* yuv = bestPredYuv;
-                bestPredYuv = m_tmpPredYuv[depth];
-                m_tmpPredYuv[depth] = yuv;
-            }
-        }
-    }
-
-    if (bestMergeCand < 0)
-    {
-        outBestCU->setMergeFlag(0, false);
-        outBestCU->initEstData(depth, outBestCU->getQP(0));
-    }
-    else
-    {
-        outTempCU->setMergeIndex(0, bestMergeCand);
-        outTempCU->setInterDirSubParts(interDirNeighbours[bestMergeCand], 0, 0, depth);
-        outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[0 + 2 * bestMergeCand], SIZE_2Nx2N, 0, 0);
-        outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[1 + 2 * bestMergeCand], SIZE_2Nx2N, 0, 0);
-        outTempCU->m_totalBits = outBestCU->m_totalBits;
-        outTempCU->m_totalDistortion = outBestCU->m_totalDistortion;
-        outTempCU->m_totalBits = 0;
-        outTempCU->m_totalCost = m_rdCost->calcRdSADCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-        outTempCU->m_sa8dCost = outTempCU->m_totalCost;
-        outBestCU->m_sa8dCost = outTempCU->m_sa8dCost;
-        if (m_param->rdLevel >= 1)
-        {
-            //calculate the motion compensation for chroma for the best mode selected
-            int numPart = outBestCU->getNumPartInter();
-            for (int partIdx = 0; partIdx < numPart; partIdx++)
-            {
-                m_search->motionCompensation(outBestCU, bestPredYuv, REF_PIC_LIST_X, partIdx, false, true);
-            }
-
-            //No-residue mode
-            m_search->encodeResAndCalcRdInterCU(outBestCU, m_origYuv[depth], bestPredYuv, m_tmpResiYuv[depth], m_bestResiYuv[depth], m_tmpRecoYuv[depth], true, true);
-
-            TComYuv* yuv = yuvReconBest;
-            yuvReconBest = m_tmpRecoYuv[depth];
-            m_tmpRecoYuv[depth] = yuv;
-
-            //Encode with residue
-            m_search->encodeResAndCalcRdInterCU(outTempCU, m_origYuv[depth], bestPredYuv, m_tmpResiYuv[depth], m_bestResiYuv[depth], m_tmpRecoYuv[depth], false, true);
-
-            if (outTempCU->m_totalCost < outBestCU->m_totalCost)    //Choose best from no-residue mode and residue mode
-            {
-                TComDataCU* tmp = outTempCU;
-                outTempCU = outBestCU;
-                outBestCU = tmp;
-
-                yuv = yuvReconBest;
-                yuvReconBest = m_tmpRecoYuv[depth];
-                m_tmpRecoYuv[depth] = yuv;
-            }
-        }
-    }
-    x265_emms();
-}
-
-void TEncCu::xCompressInterCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComDataCU*& cu, uint32_t depth, uint32_t PartitionIndex, UChar minDepth)
-{
-    TComPic* pic = outTempCU->getPic();
-
-    if (depth == 0)
-    {
-        // get original YUV data from picture
-        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), outTempCU->getAddr(), outTempCU->getZorderIdxInCU());
-    }
-    else
-    {
-        // copy partition YUV from depth 0 CTU cache
-        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], outTempCU->getZorderIdxInCU());
-    }
-
-    // variables for fast encoder decision
-    bool bSubBranch = true;
-    bool bBoundary = false;
-    uint32_t lpelx = outTempCU->getCUPelX();
-    uint32_t rpelx = lpelx + outTempCU->getCUSize(0) - 1;
-    uint32_t tpely = outTempCU->getCUPelY();
-    uint32_t bpely = tpely + outTempCU->getCUSize(0) - 1;
-    TComDataCU* subTempPartCU, * subBestPartCU;
-    int qp = outTempCU->getQP(0);
-
-    // If slice start or slice end is within this cu...
-    TComSlice * slice = outTempCU->getPic()->getSlice();
-    bool bSliceEnd = slice->getSliceCurEndCUAddr() > outTempCU->getSCUAddr() &&
-        slice->getSliceCurEndCUAddr() < outTempCU->getSCUAddr() + outTempCU->getTotalNumPart();
-    bool bInsidePicture = (rpelx < outTempCU->getSlice()->getSPS()->getPicWidthInLumaSamples()) &&
-        (bpely < outTempCU->getSlice()->getSPS()->getPicHeightInLumaSamples());
-
-    if (depth == 0 && m_param->rdLevel == 0)
-    {
-        m_origYuv[depth]->copyToPicYuv(cu->getPic()->getPicYuvRec(), cu->getAddr(), 0, 0, 0);
-    }
-    // We need to split, so don't try these modes.
-    TComYuv* tempYuv = NULL;
-#if TOPSKIP
-    TComDataCU* colocated0 = outTempCU->getCUColocated(REF_PIC_LIST_0);
-    TComDataCU* colocated1 = outTempCU->getCUColocated(REF_PIC_LIST_1);
-    char currentQP = outTempCU->getQP(0);
-    char previousQP = colocated0->getQP(0);
-    UChar delta = 0, minDepth0 = 4, minDepth1 = 4;
-    if (depth == 0)
-    {
-        double sum0 = 0, sum1 = 0, avgDepth0 = 0, avgDepth1 = 0, avgDepth = 0;
-        for (uint32_t i = 0; i < outTempCU->getTotalNumPart(); i = i + 4)
-        {
-            if (colocated0 && colocated0->getDepth(i) < minDepth0)
-                minDepth0 = colocated0->getDepth(i);
-            if (colocated1 && colocated1->getDepth(i) < minDepth1)
-                minDepth1 = colocated1->getDepth(i);
-            if (colocated0)
-                sum0 += (colocated0->getDepth(i) * 4);
-            if (colocated1)
-                sum1 += (colocated1->getDepth(i) * 4);
-        }
-
-        avgDepth0 = sum0 / outTempCU->getTotalNumPart();
-        avgDepth1 = sum1 / outTempCU->getTotalNumPart();
-        avgDepth = (avgDepth0 + avgDepth1) / 2;
-
-        minDepth = X265_MIN(minDepth0, minDepth1);
-        if (((currentQP - previousQP) < 0) || (((currentQP - previousQP) >= 0) && ((avgDepth - minDepth) > 0.5)))
-            delta = 0;
-        else
-            delta = 1;
-        if (minDepth > 0)
-            minDepth = minDepth - delta;
-    }
-    if (!(depth < minDepth)) //topskip
-#endif // if TOPSKIP
-    {
-        if (!bSliceEnd && bInsidePicture)
-        {
-            /* Initialise all Mode-CUs based on parentCU */
-            if (depth == 0)
-            {
-                m_interCU_2Nx2N[depth]->initCU(cu->getPic(), cu->getAddr());
-                m_interCU_Nx2N[depth]->initCU(cu->getPic(), cu->getAddr());
-                m_interCU_2NxN[depth]->initCU(cu->getPic(), cu->getAddr());
-                m_intraInInterCU[depth]->initCU(cu->getPic(), cu->getAddr());
-                m_mergeCU[depth]->initCU(cu->getPic(), cu->getAddr());
-                m_bestMergeCU[depth]->initCU(cu->getPic(), cu->getAddr());
-            }
-            else
-            {
-                m_interCU_2Nx2N[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-                m_interCU_2NxN[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-                m_interCU_Nx2N[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-                m_intraInInterCU[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-                m_mergeCU[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-                m_bestMergeCU[depth]->initSubCU(cu, PartitionIndex, depth, qp);
-            }
-
-            /* Compute  Merge Cost */
-            xComputeCostMerge2Nx2N(m_bestMergeCU[depth], m_mergeCU[depth], m_modePredYuv[3][depth], m_bestMergeRecoYuv[depth]);
-            bool earlyskip = false;
-            if (m_param->rdLevel >= 1)
-                earlyskip = (m_param->bEnableEarlySkip && m_bestMergeCU[depth]->isSkipped(0));
-
-            if (!earlyskip)
-            {
-                /*Compute 2Nx2N mode costs*/
-                {
-                    xComputeCostInter(m_interCU_2Nx2N[depth], m_modePredYuv[0][depth], SIZE_2Nx2N);
-                    /*Choose best mode; initialise outBestCU to 2Nx2N*/
-                    outBestCU = m_interCU_2Nx2N[depth];
-                    tempYuv = m_modePredYuv[0][depth];
-                    m_modePredYuv[0][depth] = m_bestPredYuv[depth];
-                    m_bestPredYuv[depth] = tempYuv;
-                }
-
-                /*Compute Rect costs*/
-                if (m_param->bEnableRectInter)
-                {
-                    xComputeCostInter(m_interCU_Nx2N[depth], m_modePredYuv[1][depth], SIZE_Nx2N);
-                    xComputeCostInter(m_interCU_2NxN[depth], m_modePredYuv[2][depth], SIZE_2NxN);
-                }
-
-                if (m_interCU_Nx2N[depth]->m_totalCost < outBestCU->m_totalCost)
-                {
-                    outBestCU = m_interCU_Nx2N[depth];
-
-                    tempYuv = m_modePredYuv[1][depth];
-                    m_modePredYuv[1][depth] = m_bestPredYuv[depth];
-                    m_bestPredYuv[depth] = tempYuv;
-                }
-                if (m_interCU_2NxN[depth]->m_totalCost < outBestCU->m_totalCost)
-                {
-                    outBestCU = m_interCU_2NxN[depth];
-
-                    tempYuv = m_modePredYuv[2][depth];
-                    m_modePredYuv[2][depth] = m_bestPredYuv[depth];
-                    m_bestPredYuv[depth] = tempYuv;
-                }
-                if (m_param->rdLevel > 2)
-                {
-                    //calculate the motion compensation for chroma for the best mode selected
-                    int numPart = outBestCU->getNumPartInter();
-                    for (int partIdx = 0; partIdx < numPart; partIdx++)
-                    {
-                        m_search->motionCompensation(outBestCU, m_bestPredYuv[depth], REF_PIC_LIST_X, partIdx, false, true);
-                    }
-
-                    m_search->encodeResAndCalcRdInterCU(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],
-                                                        m_bestResiYuv[depth], m_bestRecoYuv[depth], false, true);
-                    if (m_bestMergeCU[depth]->m_totalCost < outBestCU->m_totalCost)
-                    {
-                        outBestCU = m_bestMergeCU[depth];
-                        tempYuv = m_modePredYuv[3][depth];
-                        m_modePredYuv[3][depth] = m_bestPredYuv[depth];
-                        m_bestPredYuv[depth] = tempYuv;
-
-                        tempYuv = m_bestRecoYuv[depth];
-                        m_bestRecoYuv[depth] = m_bestMergeRecoYuv[depth];
-                        m_bestMergeRecoYuv[depth] = tempYuv;
-                    }
-                }
-
-                /* Check for Intra in inter frames only if its a P-slice*/
-                if (outBestCU->getSlice()->getSliceType() == P_SLICE)
-                {
-                    /*compute intra cost */
-                    bool bdoIntra = true;
-                    if (m_param->rdLevel > 2)
-                    {
-                        bdoIntra = (outBestCU->getCbf(0, TEXT_LUMA) ||  outBestCU->getCbf(0, TEXT_CHROMA_U) ||
-                                    outBestCU->getCbf(0, TEXT_CHROMA_V));
-                    }
-                    if (bdoIntra)
-                    {
-                        xComputeCostIntraInInter(m_intraInInterCU[depth], SIZE_2Nx2N);
-                        if (m_param->rdLevel > 2)
-                        {
-                            xEncodeIntraInInter(m_intraInInterCU[depth], m_origYuv[depth], m_modePredYuv[5][depth],
-                                                m_tmpResiYuv[depth],  m_tmpRecoYuv[depth]);
-                        }
-                        if (m_intraInInterCU[depth]->m_totalCost < outBestCU->m_totalCost)
-                        {
-                            outBestCU = m_intraInInterCU[depth];
-                            tempYuv = m_modePredYuv[5][depth];
-                            m_modePredYuv[5][depth] = m_bestPredYuv[depth];
-                            m_bestPredYuv[depth] = tempYuv;
-
-                            TComYuv* tmpPic = m_bestRecoYuv[depth];
-                            m_bestRecoYuv[depth] = m_tmpRecoYuv[depth];
-                            m_tmpRecoYuv[depth] = tmpPic;
-                        }
-                    }
-                }
-                if (m_param->rdLevel == 2)
-                {
-                    if (m_bestMergeCU[depth]->m_sa8dCost < outBestCU->m_totalCost)
-                    {
-                        outBestCU = m_bestMergeCU[depth];
-                        tempYuv = m_modePredYuv[3][depth];
-                        m_modePredYuv[3][depth] = m_bestPredYuv[depth];
-                        m_bestPredYuv[depth] = tempYuv;
-
-                        tempYuv = m_bestRecoYuv[depth];
-                        m_bestRecoYuv[depth] = m_bestMergeRecoYuv[depth];
-                        m_bestMergeRecoYuv[depth] = tempYuv;
-                    }
-                    else if (outBestCU->getPredictionMode(0) == MODE_INTER)
-                    {
-                        int numPart = outBestCU->getNumPartInter();
-                        for (int partIdx = 0; partIdx < numPart; partIdx++)
-                        {
-                            m_search->motionCompensation(outBestCU, m_bestPredYuv[depth], REF_PIC_LIST_X, partIdx, false, true);
-                        }
-
-                        m_search->encodeResAndCalcRdInterCU(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],
-                                                            m_bestResiYuv[depth], m_bestRecoYuv[depth], false, true);
-                    }
-                    else if (outBestCU->getPredictionMode(0) == MODE_INTRA)
-                    {
-                        xEncodeIntraInInter(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],  m_bestRecoYuv[depth]);
-                    }
-                }
-                if (m_param->rdLevel == 1)
-                {
-                    if (m_bestMergeCU[depth]->m_sa8dCost < outBestCU->m_totalCost)
-                    {
-                        outBestCU = m_bestMergeCU[depth];
-                        tempYuv = m_modePredYuv[3][depth];
-                        m_modePredYuv[3][depth] = m_bestPredYuv[depth];
-                        m_bestPredYuv[depth] = tempYuv;
-
-                        tempYuv = m_bestRecoYuv[depth];
-                        m_bestRecoYuv[depth] = m_bestMergeRecoYuv[depth];
-                        m_bestMergeRecoYuv[depth] = tempYuv;
-                    }
-                    else if (outBestCU->getPredictionMode(0) == MODE_INTER)
-                    {
-                        int numPart = outBestCU->getNumPartInter();
-                        for (int partIdx = 0; partIdx < numPart; partIdx++)
-                        {
-                            m_search->motionCompensation(outBestCU, m_bestPredYuv[depth], REF_PIC_LIST_X, partIdx, false, true);
-                        }
-
-                        m_tmpResiYuv[depth]->subtract(m_origYuv[depth], m_bestPredYuv[depth], outBestCU->getCUSize(0));
-                        m_search->generateCoeffRecon(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth], m_bestRecoYuv[depth], false);
-                    }
-                    else
-                    {
-                        m_search->generateCoeffRecon(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth], m_bestRecoYuv[depth], false);
-                    }
-                }
-                if (m_param->rdLevel == 0)
-                {
-                    if (outBestCU->getPredictionMode(0) == MODE_INTER)
-                    {
-                        int numPart = outBestCU->getNumPartInter();
-                        for (int partIdx = 0; partIdx < numPart; partIdx++)
-                        {
-                            m_search->motionCompensation(outBestCU, m_bestPredYuv[depth], REF_PIC_LIST_X, partIdx, false, true);
-                        }
-                    }
-                }
-            }
-            else
-            {
-                outBestCU = m_bestMergeCU[depth];
-                tempYuv = m_modePredYuv[3][depth];
-                m_modePredYuv[3][depth] = m_bestPredYuv[depth];
-                m_bestPredYuv[depth] = tempYuv;
-
-                tempYuv = m_bestRecoYuv[depth];
-                m_bestRecoYuv[depth] = m_bestMergeRecoYuv[depth];
-                m_bestMergeRecoYuv[depth] = tempYuv;
-            }
-
-            if (m_param->rdLevel > 0) //checkDQP can be done only after residual encoding is done
-                xCheckDQP(outBestCU);
-            /* Disable recursive analysis for whole CUs temporarily */
-            if ((outBestCU != 0) && (outBestCU->isSkipped(0)))
-                bSubBranch = false;
-            else
-                bSubBranch = true;
-
-            if (m_param->rdLevel > 1)
-            {
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeSplitFlag(outBestCU, 0, depth, true);
-                outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-                outBestCU->m_totalCost  = m_rdCost->calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
-            }
-        }
-        else if (!(bSliceEnd && bInsidePicture))
-        {
-            bBoundary = true;
-        }
-    }
-
-    // further split
-    if (bSubBranch && depth < g_maxCUDepth - g_addCUDepth)
-    {
-#if EARLY_EXIT // turn ON this to enable early exit
-        // early exit when the RD cost of best mode at depth n is less than the sum of avgerage of RD cost of the neighbour
-        // CU's(above, aboveleft, aboveright, left, colocated) and avg cost of that CU at depth "n"  with weightage for each quantity
-#if TOPSKIP
-        if (outBestCU != 0 && !(depth < minDepth)) //topskip
-#else
-        if (outBestCU != 0)
-#endif
-        {
-            uint64_t totalCostNeigh = 0, totalCostCU = 0, totalCountNeigh = 0, totalCountCU = 0;
-            double avgCost = 0;
-            TComDataCU* above = outTempCU->getCUAbove();
-            TComDataCU* aboveLeft = outTempCU->getCUAboveLeft();
-            TComDataCU* aboveRight = outTempCU->getCUAboveRight();
-            TComDataCU* left = outTempCU->getCULeft();
-            TComDataCU* rootCU = outTempCU->getPic()->getPicSym()->getCU(outTempCU->getAddr());
-
-            totalCostCU += rootCU->m_avgCost[depth] * rootCU->m_count[depth];
-            totalCountCU += rootCU->m_count[depth];
-            if (above)
-            {
-                totalCostNeigh += above->m_avgCost[depth] * above->m_count[depth];
-                totalCountNeigh += above->m_count[depth];
-            }
-            if (aboveLeft)
-            {
-                totalCostNeigh += aboveLeft->m_avgCost[depth] * aboveLeft->m_count[depth];
-                totalCountNeigh += aboveLeft->m_count[depth];
-            }
-            if (aboveRight)
-            {
-                totalCostNeigh += aboveRight->m_avgCost[depth] * aboveRight->m_count[depth];
-                totalCountNeigh += aboveRight->m_count[depth];
-            }
-            if (left)
-            {
-                totalCostNeigh += left->m_avgCost[depth] * left->m_count[depth];
-                totalCountNeigh += left->m_count[depth];
-            }
-
-            //giving 60% weight to all CU's and 40% weight to neighbour CU's
-            if (totalCountNeigh + totalCountCU)
-                avgCost = ((0.6 * totalCostCU) + (0.4 * totalCostNeigh)) / ((0.6 * totalCountCU) + (0.4 * totalCountNeigh));
-
-            float lambda = 1.0f;
-
-            if (outBestCU->m_totalCost < lambda * avgCost && avgCost != 0 && depth != 0)
-            {
-                /* Copy Best data to Picture for next partition prediction. */
-                outBestCU->copyToPic((UChar)depth);
-
-                /* Copy Yuv data to picture Yuv */
-                if (m_param->rdLevel != 0)
-                    xCopyYuv2Pic(outBestCU->getPic(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU(), depth, depth, outBestCU, lpelx, tpely);
-                return;
-            }
-        }
-#endif // if EARLY_EXIT
-        outTempCU->initEstData(depth, qp);
-        UChar nextDepth = (UChar)(depth + 1);
-        subTempPartCU = m_tempCU[nextDepth];
-        for (uint32_t nextDepth_partIndex = 0; nextDepth_partIndex < 4; nextDepth_partIndex++)
-        {
-            subBestPartCU = NULL;
-            subTempPartCU->initSubCU(outTempCU, nextDepth_partIndex, nextDepth, qp); // clear sub partition datas or init.
-
-            bool bInSlice = subTempPartCU->getSCUAddr() < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (subTempPartCU->getCUPelX() < slice->getSPS()->getPicWidthInLumaSamples()) &&
-                (subTempPartCU->getCUPelY() < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                if (0 == nextDepth_partIndex) //initialize RD with previous depth buffer
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[depth][CI_CURR_BEST]);
-                }
-                else
-                {
-                    m_rdSbacCoders[nextDepth][CI_CURR_BEST]->load(m_rdSbacCoders[nextDepth][CI_NEXT_BEST]);
-                }
-                xCompressInterCU(subBestPartCU, subTempPartCU, outTempCU, nextDepth, nextDepth_partIndex, minDepth);
-#if EARLY_EXIT
-                if (subBestPartCU->getPredictionMode(0) != MODE_INTRA)
-                {
-                    uint64_t tempavgCost = subBestPartCU->m_totalCost;
-                    TComDataCU* rootCU = outTempCU->getPic()->getPicSym()->getCU(outTempCU->getAddr());
-                    uint64_t temp = rootCU->m_avgCost[depth + 1] * rootCU->m_count[depth + 1];
-                    rootCU->m_count[depth + 1] += 1;
-                    rootCU->m_avgCost[depth + 1] = (temp + tempavgCost) / rootCU->m_count[depth + 1];
-                }
-#endif // if EARLY_EXIT
-                /* Adding costs from best SUbCUs */
-                outTempCU->copyPartFrom(subBestPartCU, nextDepth_partIndex, nextDepth, true); // Keep best part data to current temporary data.
-                if (m_param->rdLevel != 0)
-                    xCopyYuv2Tmp(subBestPartCU->getTotalNumPart() * nextDepth_partIndex, nextDepth);
-                if (m_param->rdLevel == 0)
-                    m_bestPredYuv[nextDepth]->copyToPartYuv(m_tmpPredYuv[depth], subBestPartCU->getTotalNumPart() * nextDepth_partIndex);
-            }
-            else if (bInSlice)
-            {
-                subTempPartCU->copyToPic((UChar)nextDepth);
-                outTempCU->copyPartFrom(subTempPartCU, nextDepth_partIndex, nextDepth, false);
-            }
-        }
-
-        if (!bBoundary)
-        {
-            if (m_param->rdLevel > 1)
-            {
-                m_entropyCoder->resetBits();
-                m_entropyCoder->encodeSplitFlag(outTempCU, 0, depth, true);
-                outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
-            }
-        }
-        if (m_param->rdLevel > 1)
-            outTempCU->m_totalCost = m_rdCost->calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-        else
-            outTempCU->m_totalCost = m_rdCost->calcRdSADCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
-
-        if ((g_maxCUSize >> depth) == outTempCU->getSlice()->getPPS()->getMinCuDQPSize() && outTempCU->getSlice()->getPPS()->getUseDQP())
-        {
-            bool hasResidual = false;
-            for (uint32_t uiBlkIdx = 0; uiBlkIdx < outTempCU->getTotalNumPart(); uiBlkIdx++)
-            {
-                if (outTempCU->getCbf(uiBlkIdx, TEXT_LUMA) || outTempCU->getCbf(uiBlkIdx, TEXT_CHROMA_U) ||
-                    outTempCU->getCbf(uiBlkIdx, TEXT_CHROMA_V))
-                {
-                    hasResidual = true;
-                    break;
-                }
-            }
-
-            uint32_t targetPartIdx = 0;
-            if (hasResidual)
-            {
-                bool foundNonZeroCbf = false;
-                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
-                assert(foundNonZeroCbf);
-            }
-            else
-            {
-                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
-            }
-        }
-
-        m_rdSbacCoders[nextDepth][CI_NEXT_BEST]->store(m_rdSbacCoders[depth][CI_TEMP_BEST]);
-
-        /* If Best Mode is not NULL; then compare costs. Else assign best mode to Sub-CU costs
-         * Copy recon data from Temp structure to Best structure */
-        if (outBestCU)
-        {
-#if EARLY_EXIT
-            if (depth == 0)
-            {
-                uint64_t tempavgCost = outBestCU->m_totalCost;
-                TComDataCU* rootCU = outTempCU->getPic()->getPicSym()->getCU(outTempCU->getAddr());
-                uint64_t temp = rootCU->m_avgCost[depth] * rootCU->m_count[depth];
-                rootCU->m_count[depth] += 1;
-                rootCU->m_avgCost[depth] = (temp + tempavgCost) / rootCU->m_count[depth];
-            }
-#endif
-            if (outTempCU->m_totalCost < outBestCU->m_totalCost)
-            {
-                outBestCU = outTempCU;
-                tempYuv = m_tmpRecoYuv[depth];
-                m_tmpRecoYuv[depth] = m_bestRecoYuv[depth];
-                m_bestRecoYuv[depth] = tempYuv;
-                tempYuv = m_tmpPredYuv[depth];
-                m_tmpPredYuv[depth] = m_bestPredYuv[depth];
-                m_bestPredYuv[depth] = tempYuv;
-            }
-        }
-        else
-        {
-            outBestCU = outTempCU;
-            tempYuv = m_tmpRecoYuv[depth];
-            m_tmpRecoYuv[depth] = m_bestRecoYuv[depth];
-            m_bestRecoYuv[depth] = tempYuv;
-            tempYuv = m_tmpPredYuv[depth];
-            m_tmpPredYuv[depth] = m_bestPredYuv[depth];
-            m_bestPredYuv[depth] = tempYuv;
-        }
-    }
-
-    /* Copy Best data to Picture for next partition prediction. */
-    outBestCU->copyToPic((UChar)depth);
-
-    if (m_param->rdLevel == 0 && depth == 0)
-    {
-        encodeResidue(outBestCU, outBestCU, 0, 0);
-    }
-    else if (m_param->rdLevel != 0)
-    {
-        /* Copy Yuv data to picture Yuv */
-        xCopyYuv2Pic(outBestCU->getPic(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU(), depth, depth, outBestCU, lpelx, tpely);
-    }
-
-    if (bBoundary || (bSliceEnd && bInsidePicture)) return;
-
-    /* Assert if Best prediction mode is NONE
-     * Selected mode's RD-cost must be not MAX_INT64 */
-    assert(outBestCU->getPartitionSize(0) != SIZE_NONE);
-    assert(outBestCU->getPredictionMode(0) != MODE_NONE);
-    assert(outBestCU->m_totalCost != MAX_INT64);
-}
-
-void TEncCu::encodeResidue(TComDataCU* lcu, TComDataCU* cu, uint32_t absPartIdx, UChar depth)
-{
-    UChar nextDepth = (UChar)(depth + 1);
-    TComDataCU* subTempPartCU = m_tempCU[nextDepth];
-    TComPic* pic = cu->getPic();
-    TComSlice* slice = cu->getPic()->getSlice();
-
-    if (((depth < lcu->getDepth(absPartIdx)) && (depth < (g_maxCUDepth - g_addCUDepth))))
-    {
-        uint32_t qNumParts = (pic->getNumPartInCU() >> (depth << 1)) >> 2;
-        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
-        {
-            uint32_t lpelx = lcu->getCUPelX() + g_rasterToPelX[g_zscanToRaster[absPartIdx]];
-            uint32_t tpely = lcu->getCUPelY() + g_rasterToPelY[g_zscanToRaster[absPartIdx]];
-            bool bInSlice = lcu->getSCUAddr() + absPartIdx < slice->getSliceCurEndCUAddr();
-            if (bInSlice && (lpelx < slice->getSPS()->getPicWidthInLumaSamples()) && (tpely < slice->getSPS()->getPicHeightInLumaSamples()))
-            {
-                subTempPartCU->copyToSubCU(cu, partUnitIdx, depth + 1);
-                encodeResidue(lcu, subTempPartCU, absPartIdx, depth + 1);
-            }
-        }
-
-        return;
-    }
-    if (lcu->getPredictionMode(absPartIdx) == MODE_INTER)
-    {
-        if (!lcu->getSkipFlag(absPartIdx))
-        {
-            //Calculate Residue
-            Pel* src2 = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
-            Pel* src1 = m_origYuv[0]->getLumaAddr(absPartIdx);
-            int16_t* dst = m_tmpResiYuv[depth]->getLumaAddr();
-            uint32_t src2stride = m_bestPredYuv[0]->getStride();
-            uint32_t src1stride = m_origYuv[0]->getStride();
-            uint32_t dststride = m_tmpResiYuv[depth]->m_width;
-            int part = partitionFromSizes(cu->getCUSize(0), cu->getCUSize(0));
-            primitives.luma_sub_ps[part](dst, dststride, src1, src2, src1stride, src2stride);
-
-            src2 = m_bestPredYuv[0]->getCbAddr(absPartIdx);
-            src1 = m_origYuv[0]->getCbAddr(absPartIdx);
-            dst = m_tmpResiYuv[depth]->getCbAddr();
-            src2stride = m_bestPredYuv[0]->getCStride();
-            src1stride = m_origYuv[0]->getCStride();
-            dststride = m_tmpResiYuv[depth]->m_cwidth;
-            primitives.chroma[m_param->internalCsp].sub_ps[part](dst, dststride, src1, src2, src1stride, src2stride);
-
-            src2 = m_bestPredYuv[0]->getCrAddr(absPartIdx);
-            src1 = m_origYuv[0]->getCrAddr(absPartIdx);
-            dst = m_tmpResiYuv[depth]->getCrAddr();
-            dststride = m_tmpResiYuv[depth]->m_cwidth;
-            primitives.chroma[m_param->internalCsp].sub_ps[part](dst, dststride, src1, src2, src1stride, src2stride);
-
-            //Residual encoding
-            m_search->residualTransformQuantInter(cu, 0, 0, m_tmpResiYuv[depth], cu->getDepth(0), true);
-            xCheckDQP(cu);
-
-            if (lcu->getMergeFlag(absPartIdx) && cu->getPartitionSize(0) == SIZE_2Nx2N && !cu->getQtRootCbf(0))
-            {
-                cu->setSkipFlagSubParts(true, 0, depth);
-                cu->copyCodedToPic(depth);
-            }
-            else
-            {
-                cu->copyCodedToPic(depth);
-
-                //Generate Recon
-                Pel* pred = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
-                int16_t* res = m_tmpResiYuv[depth]->getLumaAddr();
-                Pel* reco = m_bestRecoYuv[depth]->getLumaAddr();
-                dststride = m_bestRecoYuv[depth]->getStride();
-                src1stride = m_bestPredYuv[0]->getStride();
-                src2stride = m_tmpResiYuv[depth]->m_width;
-                primitives.luma_add_ps[part](reco, dststride, pred, res, src1stride, src2stride);
-
-                pred = m_bestPredYuv[0]->getCbAddr(absPartIdx);
-                res = m_tmpResiYuv[depth]->getCbAddr();
-                reco = m_bestRecoYuv[depth]->getCbAddr();
-                dststride = m_bestRecoYuv[depth]->getCStride();
-                src1stride = m_bestPredYuv[0]->getCStride();
-                src2stride = m_tmpResiYuv[depth]->m_cwidth;
-                primitives.chroma[m_param->internalCsp].add_ps[part](reco, dststride, pred, res, src1stride, src2stride);
-
-                pred = m_bestPredYuv[0]->getCrAddr(absPartIdx);
-                res = m_tmpResiYuv[depth]->getCrAddr();
-                reco = m_bestRecoYuv[depth]->getCrAddr();
-                reco = m_bestRecoYuv[depth]->getCrAddr();
-                primitives.chroma[m_param->internalCsp].add_ps[part](reco, dststride, pred, res, src1stride, src2stride);
-                m_bestRecoYuv[depth]->copyToPicYuv(lcu->getPic()->getPicYuvRec(), lcu->getAddr(), absPartIdx, 0, 0);
-                return;
-            }
-        }
-
-        //Generate Recon
-        TComPicYuv* rec = lcu->getPic()->getPicYuvRec();
-        int part = partitionFromSizes(cu->getCUSize(0), cu->getCUSize(0));
-        Pel* src = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
-        Pel* dst = rec->getLumaAddr(cu->getAddr(), absPartIdx);
-        uint32_t srcstride = m_bestPredYuv[0]->getStride();
-        uint32_t dststride = rec->getStride();
-        primitives.luma_copy_pp[part](dst, dststride, src, srcstride);
-
-        src = m_bestPredYuv[0]->getCbAddr(absPartIdx);
-        dst = rec->getCbAddr(cu->getAddr(), absPartIdx);
-        srcstride = m_bestPredYuv[0]->getCStride();
-        dststride = rec->getCStride();
-        primitives.chroma[m_param->internalCsp].copy_pp[part](dst, dststride, src, srcstride);
-
-        src = m_bestPredYuv[0]->getCrAddr(absPartIdx);
-        dst = rec->getCrAddr(cu->getAddr(), absPartIdx);
-        primitives.chroma[m_param->internalCsp].copy_pp[part](dst, dststride, src, srcstride);
-    }
-    else
-    {
-        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], absPartIdx);
-        m_search->generateCoeffRecon(cu, m_origYuv[depth], m_modePredYuv[5][depth], m_tmpResiYuv[depth],  m_tmpRecoYuv[depth], false);
-        xCheckDQP(cu);
-        m_tmpRecoYuv[depth]->copyToPicYuv(cu->getPic()->getPicYuvRec(), lcu->getAddr(), absPartIdx, 0, 0);
-        cu->copyCodedToPic(depth);
-    }
-}

x265-hg20140324.tar.bz2/source/encoder/cturow.cpp Deleted

@@ -1,119 +0,0 @@
-/*****************************************************************************
- * Copyright (C) 2013 x265 project
- *
- * Authors: Chung Shin Yee <shinyee@multicorewareinc.com>
- *          Min Chen <chenm003@163.com>
- *          Steve Borho <steve@borho.org>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
- *
- * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
- *****************************************************************************/
-
-#include "encoder.h"
-#include "PPA/ppa.h"
-#include "cturow.h"
-
-using namespace x265;
-
-bool CTURow::create(Encoder* top)
-{
-    m_rdGoOnSbacCoder.init(&m_rdGoOnBinCodersCABAC);
-    m_sbacCoder.init(&m_binCoderCABAC);
-    m_trQuant.init(1 << top->m_quadtreeTULog2MaxSize, top->bEnableRDOQ, top->bEnableRDOQTS, top->param->bEnableTSkipFast);
-
-    m_rdSbacCoders = new TEncSbac * *[g_maxCUDepth + 1];
-    m_binCodersCABAC = new TEncBinCABAC * *[g_maxCUDepth + 1];
-
-    for (uint32_t depth = 0; depth < g_maxCUDepth + 1; depth++)
-    {
-        m_rdSbacCoders[depth]  = new TEncSbac*[CI_NUM];
-        m_binCodersCABAC[depth] = new TEncBinCABAC*[CI_NUM];
-
-        for (int ciIdx = 0; ciIdx < CI_NUM; ciIdx++)
-        {
-            m_rdSbacCoders[depth][ciIdx] = new TEncSbac;
-            m_binCodersCABAC[depth][ciIdx] = new TEncBinCABAC(true);
-            m_rdSbacCoders[depth][ciIdx]->init(m_binCodersCABAC[depth][ciIdx]);
-        }
-    }
-
-    m_cuCoder.init(top);
-    m_cuCoder.setRdCost(&m_rdCost);
-    m_cuCoder.setRDSbacCoder(m_rdSbacCoders);
-    m_cuCoder.setEntropyCoder(&m_entropyCoder);
-    m_cuCoder.setPredSearch(&m_search);
-    m_cuCoder.setTrQuant(&m_trQuant);
-    m_cuCoder.setRdCost(&m_rdCost);
-    m_search.setRDSbacCoder(m_rdSbacCoders);
-    m_search.setEntropyCoder(&m_entropyCoder);
-    m_search.setRDGoOnSbacCoder(&m_rdGoOnSbacCoder);
-
-    return m_search.init(top, &m_rdCost, &m_trQuant) &&
-           m_cuCoder.create((UChar)g_maxCUDepth, g_maxCUSize);
-}
-
-
-void CTURow::processCU(TComDataCU *cu, TComSlice *slice, TEncSbac *bufferSbac, bool bSaveSBac)
-{
-    if (bufferSbac)
-    {
-        // Load SBAC coder context from previous row.
-        m_rdSbacCoders[0][CI_CURR_BEST]->loadContexts(bufferSbac);
-    }
-
-    m_entropyCoder.setEntropyCoder(&m_rdGoOnSbacCoder, slice);
-    m_entropyCoder.setBitstream(&m_bitCounter);
-    m_cuCoder.setRDGoOnSbacCoder(&m_rdGoOnSbacCoder);
-
-    m_cuCoder.compressCU(cu); // Does all the CU analysis
-
-    // restore entropy coder to an initial state
-    m_entropyCoder.setEntropyCoder(m_rdSbacCoders[0][CI_CURR_BEST], slice);
-    m_entropyCoder.setBitstream(&m_bitCounter);
-    m_cuCoder.setBitCounter(&m_bitCounter);
-    m_bitCounter.resetBits();
-
-    m_cuCoder.encodeCU(cu);  // Count bits
-
-    if (bSaveSBac)
-    {
-        // Save CABAC state for next row
-        m_bufferSbacCoder.loadContexts(m_rdSbacCoders[0][CI_CURR_BEST]);
-    }
-}
-
-void CTURow::destroy()
-{
-    for (uint32_t depth = 0; depth < g_maxCUDepth + 1; depth++)
-    {
-        for (int ciIdx = 0; ciIdx < CI_NUM; ciIdx++)
-        {
-            delete m_rdSbacCoders[depth][ciIdx];
-            delete m_binCodersCABAC[depth][ciIdx];
-        }
-    }
-
-    for (uint32_t depth = 0; depth < g_maxCUDepth + 1; depth++)
-    {
-        delete [] m_rdSbacCoders[depth];
-        delete [] m_binCodersCABAC[depth];
-    }
-
-    delete[] m_rdSbacCoders;
-    delete[] m_binCodersCABAC;
-    m_cuCoder.destroy();
-}

x265-hg20140324.tar.bz2/source/encoder/cturow.h Deleted

@@ -1,108 +0,0 @@
-/*****************************************************************************
- * Copyright (C) 2013 x265 project
- *
- * Authors: Chung Shin Yee <shinyee@multicorewareinc.com>
- *          Min Chen <chenm003@163.com>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
- *
- * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
- *****************************************************************************/
-
-#ifndef X265_CTUROW_H
-#define X265_CTUROW_H
-
-#include "TLibCommon/TComBitCounter.h"
-#include "TLibCommon/TComPic.h"
-
-#include "TLibEncoder/TEncCu.h"
-#include "TLibEncoder/TEncSearch.h"
-#include "TLibEncoder/TEncSbac.h"
-#include "TLibEncoder/TEncBinCoderCABAC.h"
-
-namespace x265 {
-// private x265 namespace
-
-class Encoder;
-
-/* manages the state of encoding one row of CTU blocks.  When
- * WPP is active, several rows will be simultaneously encoded.
- * When WPP is inactive, only one CTURow instance is used. */
-class CTURow
-{
-public:
-
-    CTURow() : m_rdGoOnBinCodersCABAC(true) {}
-
-    TEncSbac               m_sbacCoder;
-    TEncSbac               m_rdGoOnSbacCoder;
-    TEncSbac               m_bufferSbacCoder;
-    TEncBinCABAC           m_binCoderCABAC;
-    TEncBinCABAC           m_rdGoOnBinCodersCABAC;
-    TComBitCounter         m_bitCounter;
-    TComRdCost             m_rdCost;
-    TEncEntropy            m_entropyCoder;
-    TEncSearch             m_search;
-    TEncCu                 m_cuCoder;
-    TComTrQuant            m_trQuant;
-    TEncSbac            ***m_rdSbacCoders;
-    TEncBinCABAC        ***m_binCodersCABAC;
-
-    bool create(Encoder* top);
-
-    void destroy();
-
-    void init(TComSlice *slice)
-    {
-        m_active = 0;
-
-        // Note: Reset status to avoid frame parallelism output mistake on different thread number
-        for (uint32_t depth = 0; depth < g_maxCUDepth + 1; depth++)
-        {
-            for (int ciIdx = 0; ciIdx < CI_NUM; ciIdx++)
-            {
-                m_rdSbacCoders[depth][ciIdx]->setSlice(slice);
-                m_rdSbacCoders[depth][ciIdx]->resetEntropy();
-                m_binCodersCABAC[depth][ciIdx]->m_fracBits = 0;
-            }
-        }
-
-        m_rdGoOnSbacCoder.setSlice(slice);
-        m_rdGoOnSbacCoder.resetEntropy();
-    }
-
-    void processCU(TComDataCU *cu, TComSlice *slice, TEncSbac *bufferSBac, bool bSaveCabac);
-
-    /* Threading variables */
-    Lock                m_lock;
-
-    /* row is ready to run, has no neighbor dependencies. The row may have
-     * external dependencies (reference frame pixels) that prevent it from being
-     * processed, so it may stay with m_active=true for some time before it is
-     * encoded by a worker thread. */
-    volatile bool       m_active;
-
-    /* row is being processed by a worker thread.  This flag is only true when a
-     * worker thread is within the context of FrameEncoder::processRow(). This
-     * flag is used to detect multiple possible wavefront problems. */
-    volatile bool       m_busy;
-
-    /* count of completed CUs in this row */
-    volatile uint32_t   m_completed;
-};
-}
-
-#endif // ifndef X265_CTUROW_H

x265-hg20140324.tar.bz2/COPYING -> x265-hg20140928.tar.bz2/COPYING Changed

x265-hg20140928.tar.bz2/doc/reST Added

x265-hg20140928.tar.bz2/doc/reST/Makefile Added

@@ -0,0 +1,97 @@
+# Makefile for (Sphinx based) restructured text documentation
+#
+
+# You can set these variables from the command line.
+SPHINXOPTS    =
+SPHINXBUILD   = sphinx-build
+PAPER         =
+
+QCOLLECTIONGENERATOR = qcollectiongenerator
+
+# Internal variables.
+PAPEROPT_a4     = -D latex_paper_size=a4
+PAPEROPT_letter = -D latex_paper_size=letter
+ALLSPHINXOPTS   = -d build/doctrees $(PAPEROPT_$(PAPER)) $(SPHINXOPTS) .
+
+.PHONY: help clean html web pickle htmlhelp qthelp qhc latex changes linkcheck
+
+help:
+	@echo "Please use \`make <target>' where <target> is one of"
+	@echo "  html      to make standalone HTML files"
+	@echo "  pickle    to make pickle files"
+	@echo "  json      to make JSON files"
+	@echo "  htmlhelp  to make HTML files and a HTML help project"
+	@echo "  qthelp    to make HTML files and a qthelp project"
+	@echo "  qhc       to make QHC file"
+	@echo "  latex     to make LaTeX files, you can set PAPER=a4 or PAPER=letter"
+	@echo "  changes   to make an overview over all changed/added/deprecated items"
+	@echo "  linkcheck to check all external links for integrity"
+
+clean:
+	-rm -rf build/*
+
+html:
+	mkdir -p build/html build/doctrees
+	$(SPHINXBUILD) -b html $(ALLSPHINXOPTS) build/html
+	@echo
+	@echo "Build finished. The HTML pages are in build/html."
+
+zip: html
+	(cd build ; zip TortoiseHg.html.zip -r html)
+
+pickle:
+	mkdir -p build/pickle build/doctrees
+	$(SPHINXBUILD) -b pickle $(ALLSPHINXOPTS) build/pickle
+	@echo
+	@echo "Build finished; now you can process the pickle files."
+
+web: pickle
+
+json:
+	mkdir -p build/json build/doctrees
+	$(SPHINXBUILD) -b json $(ALLSPHINXOPTS) build/json
+	@echo
+	@echo "Build finished; now you can process the JSON files."
+
+htmlhelp:
+	mkdir -p build/htmlhelp build/doctrees
+	$(SPHINXBUILD) -b htmlhelp $(ALLSPHINXOPTS) build/htmlhelp
+	@echo
+	@echo "Build finished; now you can run HTML Help Workshop with the" \
+	      ".hhp project file in build/htmlhelp."
+
+qthelp:
+	mkdir -p build/qthelp build/doctrees
+	$(SPHINXBUILD) -b qthelp $(ALLSPHINXOPTS) build/qthelp
+	@echo
+	@echo "Build finished; now you can run "qcollectiongenerator" with the" \
+	      ".qhcp project file in build/qthelp, like this:"
+	@echo "# qcollectiongenerator build/qthelp/foo.qhcp"
+	@echo "To view the help file:"
+	@echo "# assistant -collectionFile build/qthelp/foo.qhc"
+
+qhc: qthelp
+	$(QCOLLECTIONGENERATOR) build/qthelp/TortoiseHg.qhcp
+	@echo "Build finished. To view the help file:"
+	@echo "# assistant -collectionFile build/qthelp/TortoiseHg.qhc"
+
+latex:
+	mkdir -p build/latex build/doctrees
+	$(SPHINXBUILD) -b latex $(ALLSPHINXOPTS) build/latex
+	@echo
+	@echo "Build finished; the LaTeX files are in build/latex."
+	@echo "Run \`make all-pdf' or \`make all-ps' in that directory to" \
+	      "run these through (pdf)latex."
+
+changes:
+	mkdir -p build/changes build/doctrees
+	$(SPHINXBUILD) -b changes $(ALLSPHINXOPTS) build/changes
+	@echo
+	@echo "The overview file is in build/changes."
+
+linkcheck:
+	mkdir -p build/linkcheck build/doctrees
+	$(SPHINXBUILD) -b linkcheck $(ALLSPHINXOPTS) build/linkcheck
+	@echo
+	@echo "Link check complete; look for any errors in the above output " \
+	      "or in build/linkcheck/output.txt."

x265-hg20140928.tar.bz2/doc/reST/api.rst Added

@@ -0,0 +1,340 @@
+*********************************
+Application Programming Interface
+*********************************
+
+Introduction
+============
+
+x265 is written primarily in C++ and x86 assembly language but the
+public facing programming interface is C for the widest possible
+portability.  This C interface is wholly defined within :file:`x265.h`
+in the source/ folder of our source tree.  All of the functions and
+variables and enumerations meant to be used by the end-user are present
+in this header.
+
+Where possible, x265 has tried to keep its public API as close as
+possible to x264's public API. So those familiar with using x264 through
+its C interface will find x265 quite familiar.
+
+This file is meant to be read in-order; the narrative follows linearly
+through the various sections
+
+Build Considerations
+====================
+
+The choice of Main or Main10 profile encodes is made at compile time;
+the internal pixel depth influences a great deal of variable sizes and
+thus 8 and 10bit pixels are handled as different build options
+(primarily to maintain the performance of the 8bit builds). libx265
+exports a variable **x265_max_bit_depth** which indicates how the
+library was compiled (it will contain a value of 8 or 10). Further,
+**x265_version_str** is a pointer to a string indicating the version of
+x265 which was compiled, and **x265_build_info_str** is a pointer to a
+string identifying the compiler and build options.
+
+x265 will accept input pixels of any depth between 8 and 16 bits
+regardless of the depth of its internal pixels (8 or 10).  It will shift
+and mask input pixels as required to reach the internal depth. If
+downshifting is being performed using our CLI application, the
+:option:`--dither` option may be enabled to reduce banding. This feature
+is not available through the C interface.
+
+Encoder
+=======
+
+The primary object in x265 is the encoder object, and this is
+represented in the public API as an opaque typedef **x265_encoder**.
+Pointers of this type are passed to most encoder functions.
+
+A single encoder generates a single output bitstream from a sequence of
+raw input pictures.  Thus if you need multiple output bitstreams you
+must allocate multiple encoders.  You may pass the same input pictures
+to multiple encoders, the encode function does not modify the input
+picture structures (the pictures are copied into the encoder as the
+first step of encode).
+
+Encoder allocation is a reentrant function, so multiple encoders may be
+safely allocated in a single process. The encoder access functions are
+not reentrant for a single encoder, so the recommended use case is to
+allocate one client thread per encoder instance (one thread for all
+encoder instances is possible, but some encoder access functions are
+blocking and thus this would be less efficient).
+
+.. Note::
+
+	There is one caveat to having multiple encoders within a single
+	process. All of the encoders must use the same maximum CTU size
+	because many global variables are configured based on this size.
+	Encoder allocation will fail if a mis-matched CTU size is attempted.
+
+An encoder is allocated by calling **x265_encoder_open()**::
+
+	/* x265_encoder_open:
+	*      create a new encoder handler, all parameters from x265_param are copied */
+	x265_encoder* x265_encoder_open(x265_param *);
+
+The returned pointer is then passed to all of the functions pertaining
+to this encode. A large amount of memory is allocated during this
+function call, but the encoder will continue to allocate memory as the
+first pictures are passed to the encoder; until its pool of picture
+structures is large enough to handle all of the pictures it must keep
+internally.  The pool size is determined by the lookahead depth, the
+number of frame threads, and the maximum number of references.
+
+As indicated in the comment, **x265_param** is copied internally so the user
+may release their copy after allocating the encoder.  Changes made to
+their copy of the param structure have no affect on the encoder after it
+has been allocated.
+
+Param
+=====
+
+The **x265_param** structure describes everything the encoder needs to
+know about the input pictures and the output bitstream and most
+everything in between.
+
+The recommended way to handle these param structures is to allocate them
+from libx265 via::
+
+	/* x265_param_alloc:
+	 *  Allocates an x265_param instance. The returned param structure is not
+	 *  special in any way, but using this method together with x265_param_free()
+	 *  and x265_param_parse() to set values by name allows the application to treat
+	 *  x265_param as an opaque data struct for version safety */
+	x265_param *x265_param_alloc();
+
+In this way, your application does not need to know the exact size of
+the param structure (the build of x265 could potentially be a bit newer
+than the copy of :file:`x265.h` that your application compiled against).
+
+Next you perform the initial *rough cut* configuration of the encoder by
+chosing a performance preset and optional tune factor
+**x265_preset_names** and **x265_tune_names** respectively hold the
+string names of the presets and tune factors (see :ref:`presets
+<preset-tune-ref>` for more detail on presets and tune factors)::
+
+	/*      returns 0 on success, negative on failure (e.g. invalid preset/tune name). */
+	int x265_param_default_preset(x265_param *, const char *preset, const char *tune);
+
+Now you may optionally specify a profile. **x265_profile_names**
+contains the string names this function accepts::
+
+	/*      (can be NULL, in which case the function will do nothing)
+	 *      returns 0 on success, negative on failure (e.g. invalid profile name). */
+	int x265_param_apply_profile(x265_param *, const char *profile);
+
+Finally you configure any remaining options by name using repeated calls to::
+
+	/* x265_param_parse:
+	 *  set one parameter by name.
+	 *  returns 0 on success, or returns one of the following errors.
+	 *  note: BAD_VALUE occurs only if it can't even parse the value,
+	 *  numerical range is not checked until x265_encoder_open().
+	 *  value=NULL means "true" for boolean options, but is a BAD_VALUE for non-booleans. */
+	#define X265_PARAM_BAD_NAME  (-1)
+	#define X265_PARAM_BAD_VALUE (-2)
+	int x265_param_parse(x265_param *p, const char *name, const char *value);
+
+See :ref:`string options <string-options-ref>` for the list of options (and their
+descriptions) which can be set by **x265_param_parse()**.
+
+After the encoder has been created, you may release the param structure::
+
+	/* x265_param_free:
+	 *  Use x265_param_free() to release storage for an x265_param instance
+	 *  allocated by x265_param_alloc() */
+	void x265_param_free(x265_param *);
+
+.. Note::
+
+	Using these methods to allocate and release the param structures
+	helps future-proof your code in many ways, but the x265 API is
+	versioned in such a way that we prevent linkage against a build of
+	x265 that does not match the version of the header you are compiling
+	against. This is function of the X265_BUILD macro.
+
+**x265_encoder_parameters()** may be used to get a copy of the param
+structure from the encoder after it has been opened, in order to see the
+changes made to the parameters for auto-detection and other reasons::
+
+	/* x265_encoder_parameters:
+	 *      copies the current internal set of parameters to the pointer provided
+	 *      by the caller.  useful when the calling application needs to know
+	 *      how x265_encoder_open has changed the parameters.
+	 *      note that the data accessible through pointers in the returned param struct
+	 *      (e.g. filenames) should not be modified by the calling application. */
+	void x265_encoder_parameters(x265_encoder *, x265_param *);                                                                      
+
+Pictures
+========
+
+Raw pictures are passed to the encoder via the **x265_picture** structure.
+Just like the param structure we recommend you allocate this structure
+from the encoder to avoid potential size mismatches::
+
+	/* x265_picture_alloc:
+	 *  Allocates an x265_picture instance. The returned picture structure is not
+	 *  special in any way, but using this method together with x265_picture_free()
+	 *  and x265_picture_init() allows some version safety. New picture fields will
+	 *  always be added to the end of x265_picture */
+	x265_picture *x265_picture_alloc();
+
+Regardless of whether you allocate your picture structure this way or
+whether you simply declare it on the stack, your next step is to
+initialize the structure via::
+
+	/***
+	 * Initialize an x265_picture structure to default values. It sets the pixel
+	 * depth and color space to the encoder's internal values and sets the slice
+	 * type to auto - so the lookahead will determine slice type.
+	 */
+	void x265_picture_init(x265_param *param, x265_picture *pic);
+
+x265 does not perform any color space conversions, so the raw picture's
+color space (chroma sampling) must match the color space specified in
+the param structure used to allocate the encoder. **x265_picture_init**
+initializes this field to the internal color space and it is best to
+leave it unmodified.
+
+The picture bit depth is initialized to be the encoder's internal bit
+depth but this value should be changed to the actual depth of the pixels
+being passed into the encoder.  If the picture bit depth is more than 8,
+the encoder assumes two bytes are used to represent each sample
+(little-endian shorts).
+
+The user is responsible for setting the plane pointers and plane strides
+(in units of bytes, not pixels). The presentation time stamp (**pts**)
+is optional, depending on whether you need accurate decode time stamps
+(**dts**) on output.
+
+If you wish to override the lookahead or rate control for a given
+picture you may specify a slicetype other than X265_TYPE_AUTO, or a
+forceQP value other than 0.
+
+x265 does not modify the picture structure provided as input, so you may
+reuse a single **x265_picture** for all pictures passed to a single
+encoder, or even all pictures passed to multiple encoders.
+
+Structures allocated from the library should eventually be released::
+
+	/* x265_picture_free:
+	 *  Use x265_picture_free() to release storage for an x265_picture instance
+	 *  allocated by x265_picture_alloc() */
+	void x265_picture_free(x265_picture *);
+
+
+Analysis Buffers
+================
+
+Analysis information can be saved and reused to between encodes of the
+same video sequence (generally for multiple bitrate encodes).  The best
+results are attained by saving the analysis information of the highest
+bitrate encode and reuse it in lower bitrate encodes.
+
+When saving or loading analysis data, buffers must be allocated for
+every picture passed into the encoder using::
+
+	/* x265_alloc_analysis_data:
+	 *  Allocate memory to hold analysis meta data, returns 1 on success else 0 */
+	int x265_alloc_analysis_data(x265_picture*);
+
+Note that this is very different from the typical semantics of
+**x265_picture**, which can be reused many times. The analysis buffers must
+be re-allocated for every input picture.
+
+Analysis buffers passed to the encoder are owned by the encoder until
+they pass the buffers back via an output **x265_picture**. The user is
+responsible for releasing the buffers when they are finished with them
+via::
+
+	/* x265_free_analysis_data:
+	 *  Use x265_free_analysis_data to release storage of members allocated by
+	 *  x265_alloc_analysis_data */
+	void x265_free_analysis_data(x265_picture*);
+
+
+Encode Process
+==============
+
+The output of the encoder is a series of NAL packets, which are always
+returned concatenated in consecutive memory. HEVC streams have SPS and
+PPS and VPS headers which describe how the following packets are to be
+decoded. If you specified :option:`--repeat-headers` then those headers
+will be output with every keyframe.  Otherwise you must explicitly query
+those headers using::
+
+	/* x265_encoder_headers:
+	 *      return the SPS and PPS that will be used for the whole stream.
+	 *      *pi_nal is the number of NAL units outputted in pp_nal.
+	 *      returns negative on error, total byte size of payload data on success
+	 *      the payloads of all output NALs are guaranteed to be sequential in memory. */
+	int x265_encoder_headers(x265_encoder *, x265_nal **pp_nal, uint32_t *pi_nal);
+
+Now we get to the main encode loop. Raw input pictures are passed to the
+encoder in display order via::
+
+	/* x265_encoder_encode:
+	 *      encode one picture.
+	 *      *pi_nal is the number of NAL units outputted in pp_nal.
+	 *      returns negative on error, zero if no NAL units returned.
+	 *      the payloads of all output NALs are guaranteed to be sequential in memory. */
+	int x265_encoder_encode(x265_encoder *encoder, x265_nal **pp_nal, uint32_t *pi_nal, x265_picture *pic_in, x265_picture *pic_out);
+
+These pictures are queued up until the lookahead is full, and then the
+frame encoders in turn are filled, and then finally you begin receiving
+a output NALs (corresponding to a single output picture) with each input
+picture you pass into the encoder.
+
+Once the pipeline is completely full, **x265_encoder_encode()** will
+block until the next output picture is complete.
+
+.. note:: 
+
+	Optionally, if the pointer of a second **x265_picture** structure is
+	provided, the encoder will fill it with data pertaining to the
+	output picture corresponding to the output NALs, including the
+	recontructed image, POC and decode timestamp. These pictures will be
+	in encode (or decode) order.
+
+When the last of the raw input pictures has been sent to the encoder,
+**x265_encoder_encode()** must still be called repeatedly with a
+*pic_in* argument of 0, indicating a pipeline flush, until the function
+returns a value less than or equal to 0 (indicating the output bitstream
+is complete).
+
+At any time during this process, the application may query running
+statistics from the encoder::
+
+	/* x265_encoder_get_stats:
+	 *       returns encoder statistics */
+	void x265_encoder_get_stats(x265_encoder *encoder, x265_stats *, uint32_t statsSizeBytes);
+
+Cleanup
+=======
+
+At the end of the encode, the application will want to trigger logging
+of the final encode statistics, if :option:`--csv` had been specified::
+
+	/* x265_encoder_log:
+	 *       write a line to the configured CSV file.  If a CSV filename was not
+	 *       configured, or file open failed, or the log level indicated frame level
+	 *       logging, this function will perform no write. */
+	void x265_encoder_log(x265_encoder *encoder, int argc, char **argv);
+
+Finally, the encoder must be closed in order to free all of its
+resources. An encoder that has been flushed cannot be restarted and
+reused. Once **x265_encoder_close()** has been called, the encoder
+handle must be discarded::
+
+	/* x265_encoder_close:
+	 *      close an encoder handler */
+	void x265_encoder_close(x265_encoder *);
+
+When the application has completed all encodes, it should call
+**x265_cleanup()** to free process global resources like the thread pool;
+particularly if a memory-leak detection tool is being used::
+
+	/***
+	 * Release library static allocations
+	 */
+	void x265_cleanup(void);

x265-hg20140928.tar.bz2/doc/reST/cli.rst Added

@@ -0,0 +1,1146 @@
+*********************
+Command Line Options
+*********************
+
+.. _string-options-ref:
+
+Note that unless an option is listed as **CLI ONLY** the option is also
+supported by x265_param_parse(). The CLI uses getopt to parse the
+command line options so the short or long versions may be used and the
+long options may be truncated to the shortest unambiguous abbreviation.
+Users of the API must pass x265_param_parse() the full option name.
+
+Preset and tune have special implications. The API user must call
+x265_param_default_preset() with the preset and tune parameters they
+wish to use, prior to calling x265_param_parse() to set any additional
+fields. The CLI does this for the user implicitly, so all CLI options
+are applied after the user's preset and tune choices, regardless of the
+order of the arguments on the command line.
+
+If there is an extra command line argument (not an option or an option
+value) the CLI will treat it as the input filename.  This effectively
+makes the :option:`--input` specifier optional for the input file. If
+there are two extra arguments, the second is treated as the output
+bitstream filename, making :option:`--output` also optional if the input
+filename was implied. This makes :command:`x265 in.y4m out.hevc` a valid
+command line. If there are more than two extra arguments, the CLI will
+consider this an error and abort.
+
+Generally, when an option expects a string value from a list of strings
+the user may specify the integer ordinal of the value they desire. ie:
+:option:`--log-level` 3 is equivalent to :option:`--log-level` debug.
+
+Standalone Executable Options
+=============================
+
+.. option:: --help, -h
+
+	Display help text
+
+	**CLI ONLY**
+
+.. option:: --version, -V
+
+	Display version details
+
+	**CLI ONLY**
+
+.. option:: --asm <integer:false:string>, --no-asm
+
+	x265 will use all detected CPU SIMD architectures by default. You can
+	disable all assembly by using :option:`--no-asm` or you can specify
+	a comma separated list of SIMD architectures to use, matching these
+	strings: MMX2, SSE, SSE2, SSE3, SSSE3, SSE4, SSE4.1, SSE4.2, AVX, XOP, FMA4, AVX2, FMA3
+
+	Some higher architectures imply lower ones being present, this is
+	handled implicitly.
+
+	One may also directly supply the CPU capability bitmap as an integer.
+
+.. option:: --threads <integer>
+
+	Number of threads for thread pool. Default 0 (detected CPU core
+	count)
+
+.. option:: --preset, -p <integer|string>
+
+	Sets parameters to preselected values, trading off compression efficiency against 
+	encoding speed. These parameters are applied before all other input parameters are 
+	applied, and so you can override any parameters that these values control.
+
+	0. ultrafast
+	1. superfast
+	2. veryfast
+	3. faster
+	4. fast
+	5. medium **(default)**
+	6. slow
+	7. slower
+	8. veryslow
+	9. placebo
+
+.. option:: --tune, -t <string>
+
+	Tune the settings for a particular type of source or situation. The changes will
+	be applied after :option:`--preset` but before all other parameters. Default none
+
+	**Values:** psnr, ssim, zero-latency, fast-decode.
+
+.. option:: --frame-threads, -F <integer>
+
+	Number of concurrently encoded frames. Using a single frame thread
+	gives a slight improvement in compression, since the entire reference
+	frames are always available for motion compensation, but it has
+	severe performance implications. Default is an autodetected count
+	based on the number of CPU cores and whether WPP is enabled or not.
+
+.. option:: --log-level <integer|string>
+
+	Logging level. Debug level enables per-frame QP, metric, and bitrate
+	logging. If a CSV file is being generated, debug level makes the log
+	be per-frame rather than per-encode. Full level enables hash and
+	weight logging. -1 disables all logging, except certain fatal
+	errors, and can be specified by the string "none".
+
+	0. error
+	1. warning
+	2. info **(default)**
+	3. debug
+	4. full
+
+.. option:: --csv <filename>
+
+	Writes encoding results to a comma separated value log file. Creates
+	the file if it doesnt already exist, else adds one line per run.  if
+	:option:`--log-level` is debug or above, it writes one line per
+	frame. Default none
+
+.. option:: --cu-stats, --no-cu-stats
+
+	Records statistics on how each CU was coded (split depths and other
+	mode decisions) and reports those statistics at the end of the
+	encode. Default disabled
+
+.. option:: --output, -o <filename>
+
+	Bitstream output file name. If there are two extra CLI options, the
+	first is implicitly the input filename and the second is the output
+	filename, making the :option:`--output` option optional.
+
+	The output file will always contain a raw HEVC bitstream, the CLI
+	does not support any container file formats.
+
+	**CLI ONLY**
+
+.. option:: --no-progress
+
+	Disable CLI periodic progress reports
+
+	**CLI ONLY**
+
+Quality reporting metrics
+=========================
+
+.. option:: --ssim, --no-ssim
+
+	Calculate and report Structural Similarity values. It is
+	recommended to use :option:`--tune` ssim if you are measuring ssim,
+	else the results should not be used for comparison purposes.
+	Default disabled
+
+.. option:: --psnr, --no-psnr
+
+	Calculate and report Peak Signal to Noise Ratio.  It is recommended
+	to use :option:`--tune` psnr if you are measuring PSNR, else the
+	results should not be used for comparison purposes.  Default
+	disabled
+
+Input Options
+=============
+
+.. option:: --input <filename>
+
+	Input filename, only raw YUV or Y4M supported. Use single dash for
+	stdin. This option name will be implied for the first "extra"
+	command line argument.
+
+	**CLI ONLY**
+
+.. option:: --y4m
+
+	Parse input stream as YUV4MPEG2 regardless of file extension,
+	primarily intended for use with stdin (ie: :option:`--input` -
+	:option:`--y4m`).  This option is implied if the input filename has
+	a ".y4m" extension
+
+	**CLI ONLY**
+
+.. option:: --input-depth <integer>
+
+	YUV only: Bit-depth of input file or stream
+
+	**Values:** any value between 8 and 16. Default is internal depth.
+
+	**CLI ONLY**
+
+.. option:: --dither
+
+	Enable high quality downscaling. Dithering is based on the diffusion
+	of errors from one row of pixels to the next row of pixels in a
+	picture. Only applicable when the input bit depth is larger than
+	8bits and internal bit depth is 8bits. Default disabled
+
+	**CLI ONLY**
+
+.. option:: --nr <integer>
+
+	Noise reduction - an adaptive deadzone applied after DCT
+	(subtracting from DCT coefficients), before quantization, on inter
+	blocks. It does no pixel-level filtering, doesn't cross DCT block
+	boundaries, has no overlap, doesn't affect intra blocks. The higher
+	the strength value parameter, the more aggressively it will reduce
+	noise.
+
+	Enabling noise reduction will make outputs diverge between different
+	numbers of frame threads. Outputs will be deterministic but the
+	outputs of -F2 will no longer match the outputs of -F3, etc.
+
+	**Values:** any value in range of 100 to 1000. Default disabled.
+
+.. option:: --input-res <wxh>
+
+	YUV only: Source picture size [w x h]
+
+	**CLI ONLY**
+
+.. option:: --input-csp <integer|string>
+
+	YUV only: Source color space. Only i420, i422, and i444 are
+	supported at this time. The internal color space is always the
+	same as the source color space (libx265 does not support any color
+	space conversions).
+
+	0. i400
+	1. i420 **(default)**
+	2. i422
+	3. i444
+	4. nv12
+	5. nv16
+
+.. option:: --fps <integer|float|numerator/denominator>
+
+	YUV only: Source frame rate
+
+	**Range of values:** positive int or float, or num/denom
+
+.. option:: --interlaceMode <false|tff|bff>, --no-interlaceMode
+
+	**EXPERIMENTAL** Specify interlace type of source pictures. 
+	
+	0. progressive pictures **(default)**
+	1. top field first 
+	2. bottom field first
+
+	HEVC encodes interlaced content as fields. Fields must be provided to
+	the encoder in the correct temporal order. The source dimensions
+	must be field dimensions and the FPS must be in units of fields per
+	second. The decoder must re-combine the fields in their correct
+	orientation for display.
+
+.. option:: --seek <integer>
+
+	Number of frames to skip at start of input file. Default 0
+
+	**CLI ONLY**
+
+.. option:: --frames, -f <integer>
+
+	Number of frames to be encoded. Default 0 (all)
+
+	**CLI ONLY**
+
+.. option:: --qpfile <filename>
+
+	Specify a text file which contains frametypes and QPs for some or
+	all frames. The format of each line is:
+
+	framenumber frametype QP
+
+	Frametype can be one of [I,i,P,B,b]. **B** is a referenced B frame,
+	**b** is an unreferenced B frame.  **I** is a keyframe (random
+	access point) while **i** is a I frame that is not a keyframe
+	(references are not broken).
+
+	Specifying QP (integer) is optional, and if specified they are
+	clamped within the encoder to qpmin/qpmax.
+
+.. option:: --scaling-list <filename>
+
+	Quantization scaling lists. HEVC supports 6 quantization scaling
+	lists to be defined; one each for Y, Cb, Cr for intra prediction and
+	one each for inter prediction.
+
+	x265 does not use scaling lists by default, but this can also be
+	made explicit by :option:`--scaling-list` *off*.
+
+	HEVC specifies a default set of scaling lists which may be enabled
+	without requiring them to be signaled in the SPS. Those scaling
+	lists can be enabled via :option:`--scaling-list` *default*.
+    
+	All other strings indicate a filename containing custom scaling
+	lists in the HM format. The encode will abort if the file is not
+	parsed correctly. Custom lists must be signaled in the SPS
+
+.. option:: --lambda-file <filename>
+
+	Specify a text file containing values for x265_lambda_tab and
+	x265_lambda2_tab. Each table requires MAX_MAX_QP+1 (70) float
+	values.
+	
+	The text file syntax is simple. Comma is considered to be
+	white-space. All white-space is ignored. Lines must be less than 2k
+	bytes in length. Content following hash (#) characters are ignored.
+	The values read from the file are logged at :option:`--log-level`
+	debug.
+
+	Note that the lambda tables are process-global and so the new values
+	affect all encoders running in the same process. 
+	
+	Lambda values affect encoder mode decisions, the lower the lambda
+	the more bits it will try to spend on signaling information (motion
+	vectors and splits) and less on residual. This feature is intended
+	for experimentation.
+
+Profile, Level, Tier
+====================
+
+.. option:: --profile <string>
+
+	Enforce the requirements of the specified profile, ensuring the
+	output stream will be decodable by a decoder which supports that
+	profile.  May abort the encode if the specified profile is
+	impossible to be supported by the compile options chosen for the
+	encoder (a high bit depth encoder will be unable to output
+	bitstreams compliant with Main or Mainstillpicture).
+
+	API users must use x265_param_apply_profile() after configuring
+	their param structure. Any changes made to the param structure after
+	this call might make the encode non-compliant.
+
+	**Values:** main, main10, mainstillpicture
+
+	**CLI ONLY**
+
+.. option:: --level-idc <integer|float>
+
+	Minimum decoder requirement level. Defaults to 0, which implies
+	auto-detection by the encoder. If specified, the encoder will
+	attempt to bring the encode specifications within that specified
+	level. If the encoder is unable to reach the level it issues a
+	warning and aborts the encode. If the requested requirement level is
+	higher than the actual level, the actual requirement level is
+	signaled.
+
+	Beware, specifying a decoder level will force the encoder to enable
+	VBV for constant rate factor encodes, which may introduce
+	non-determinism.
+
+	The value is specified as a float or as an integer with the level
+	times 10, for example level **5.1** is specified as "5.1" or "51",
+	and level **5.0** is specified as "5.0" or "50".
+
+	Annex A levels: 1, 2, 2.1, 3, 3.1, 4, 4.1, 5, 5.1, 5.2, 6, 6.1, 6.2
+
+.. option:: --high-tier, --no-high-tier
+
+	If :option:`--level-idc` has been specified, the option adds the
+	intention to support the High tier of that level. If your specified
+	level does not support a High tier, a warning is issued and this
+	modifier flag is ignored.
+
+.. note::
+	:option:`--profile`, :option:`--level-idc`, and
+	:option:`--high-tier` are only intended for use when you are
+	targeting a particular decoder (or decoders) with fixed resource
+	limitations and must constrain the bitstream within those limits.
+	Specifying a profile or level may lower the encode quality
+	parameters to meet those requirements but it will never raise
+	them.
+
+Quad-Tree analysis
+==================
+
+.. option:: --wpp, --no-wpp
+
+	Enable Wavefront Parallel Processing. The encoder may begin encoding
+	a row as soon as the row above it is at least two CTUs ahead in the
+	encode process. This gives a 3-5x gain in parallelism for about 1%
+	overhead in compression efficiency. Default: Enabled
+
+.. option:: --ctu, -s <64|32|16>
+
+	Maximum CU size (width and height). The larger the maximum CU size,
+	the more efficiently x265 can encode flat areas of the picture,
+	giving large reductions in bitrate. However this comes at a loss of
+	parallelism with fewer rows of CUs that can be encoded in parallel,
+	and less frame parallelism as well. Because of this the faster
+	presets use a CU size of 32. Default: 64
+
+.. option:: --tu-intra-depth <1..4>
+
+	The transform unit (residual) quad-tree begins with the same depth
+	as the coding unit quad-tree, but the encoder may decide to further
+	split the transform unit tree if it improves compression efficiency.
+	This setting limits the number of extra recursion depth which can be
+	attempted for intra coded units. Default: 1
+
+.. option:: --tu-inter-depth <1..4>
+
+	The transform unit (residual) quad-tree begins with the same depth
+	as the coding unit quad-tree, but the encoder may decide to further
+	split the transform unit tree if it improves compression efficiency.
+	This setting limits the number of extra recursion depth which can be
+	attempted for inter coded units. Default: 1
+
+Temporal / motion search options
+================================
+
+.. option:: --me <integer|string>
+
+	Motion search method. Generally, the higher the number the harder
+	the ME method will try to find an optimal match. Diamond search is
+	the simplest. Hexagon search is a little better. Uneven
+	Multi-Hexegon is an adaption of the search method used by x264 for
+	slower presets. Star is a three step search adapted from the HM
+	encoder: a star-pattern search followed by an optional radix scan
+	followed by an optional star-search refinement. Full is an
+	exhaustive search; an order of magnitude slower than all other
+	searches but not much better than umh or star.
+
+	0. dia
+	1. hex **(default)**
+	2. umh
+	3. star
+	4. full
+
+.. option:: --subme, -m <0..7>
+
+	Amount of subpel refinement to perform. The higher the number the
+	more subpel iterations and steps are performed. Default 2
+
+	+----+------------+-----------+------------+-----------+-----------+
+	| -m | HPEL iters | HPEL dirs | QPEL iters | QPEL dirs | HPEL SATD |
+	+====+============+===========+============+===========+===========+
+	|  0 | 1          | 4         | 0          | 4         | false     |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  1 | 1          | 4         | 1          | 4         | false     |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  2 | 1          | 4         | 1          | 4         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  3 | 2          | 4         | 1          | 4         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  4 | 2          | 4         | 2          | 4         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  5 | 1          | 8         | 1          | 8         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  6 | 2          | 8         | 1          | 8         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+	|  7 | 2          | 8         | 2          | 8         | true      |
+	+----+------------+-----------+------------+-----------+-----------+
+
+.. option:: --merange <integer>
+
+	Motion search range. Default 57
+
+	The default is derived from the default CTU size (64) minus the luma
+	interpolation half-length (4) minus maximum subpel distance (2)
+	minus one extra pixel just in case the hex search method is used. If
+	the search range were any larger than this, another CTU row of
+	latency would be required for reference frames.
+
+	**Range of values:** an integer from 0 to 32768
+
+.. option:: --max-merge <1..5>
+
+	Maximum number of neighbor (spatial and temporal) candidate blocks
+	that the encoder may consider for merging motion predictions. If a
+	merge candidate results in no residual, it is immediately selected
+	as a "skip".  Otherwise the merge candidates are tested as part of
+	motion estimation when searching for the least cost inter option.
+	The max candidate number is encoded in the SPS and determines the
+	bit cost of signaling merge CUs. Default 2
+
+Spatial/intra options
+=====================
+
+.. option:: --rdpenalty <0..2>
+
+	Penalty for 32x32 intra TU in non-I slices. Default 0
+
+	**Values:** 0:disabled 1:RD-penalty 2:maximum
+
+.. option:: --b-intra, --no-b-intra
+
+	Enables the use of intra modes in very slow presets (:option:`--rd`
+	5 or 6). Presets slow to ultrafast do not try intra in B frames
+	regardless of this setting. Default enabled.
+
+.. option:: --tskip, --no-tskip
+
+	Enable intra transform skipping (encode residual as coefficients)
+	for intra coded blocks. Default disabled
+
+.. option:: --tskip-fast, --no-tskip-fast
+
+	Enable fast intra transform skip decisions. Only applicable if
+	transform skip is enabled. Default disabled
+
+.. option:: --strong-intra-smoothing, --no-strong-intra-smoothing
+
+	Enable strong intra smoothing for 32x32 intra blocks. Default enabled
+
+.. option:: --constrained-intra, --no-constrained-intra
+
+	Constrained intra prediction. When generating intra predictions for
+	blocks in inter slices, only intra-coded reference pixels are used.
+	Inter-coded reference pixels are replaced with intra-coded neighbor
+	pixels or default values. The general idea is to block the
+	propagation of reference errors that may have resulted from lossy
+	signals. Default disabled
+
+Mode decision / Analysis
+========================
+
+.. option:: --rect, --no-rect
+
+	Enable analysis of rectangular motion partitions Nx2N and 2NxN
+	(50/50 splits, two directions). Default disabled
+
+.. option:: --amp, --no-amp
+
+	Enable analysis of asymmetric motion partitions (75/25 splits, four
+	directions). This setting has no effect if rectangular partitions
+	are disabled. Even though there are four possible AMP partitions,
+	only the most likely candidate is tested, based on the results of
+	the rectangular mode tests. Default disabled
+
+.. option:: --early-skip, --no-early-skip
+
+	Measure full CU size (2Nx2N) merge candidates first; if no residual
+	is found the analysis is short circuited. Default disabled
+
+.. option:: --fast-cbf, --no-fast-cbf
+
+	Short circuit analysis if a prediction is found that does not set
+	the coded block flag (aka: no residual was encoded).  It prevents
+	the encoder from perhaps finding other predictions that also have no
+	residual but require less signaling bits. Default disabled
+
+.. option:: --fast-intra, --no-fast-intra
+
+	Perform an initial scan of every fifth intra angular mode, then
+	check modes +/- 2 distance from the best mode, then +/- 1 distance
+	from the best mode, effectively performing a gradient descent. When
+	enabled 10 modes in total are checked. When disabled all 33 angular
+	modes are checked.  Only applicable for :option:`--rd` levels 3 and
+	below (medium preset and faster).
+
+.. option:: --weightp, -w, --no-weightp
+
+	Enable weighted prediction in P slices. This enables weighting
+	analysis in the lookahead, which influences slice decisions, and
+	enables weighting analysis in the main encoder which allows P
+	reference samples to have a weight function applied to them prior to
+	using them for motion compensation.  In video which has lighting
+	changes, it can give a large improvement in compression efficiency.
+	Default is enabled
+
+.. option:: --weightb, --no-weightb
+
+	Enable weighted prediction in B slices. Default disabled
+
+.. option:: --rd <0..6>
+
+	Level of RDO in mode decision. The higher the value, the more
+	exhaustive the analysis and the more rate distortion optimization is
+	used. The lower the value the faster the encode, the higher the
+	value the smaller the bitstream (in general). Default 3
+
+	Note that this table aims for accuracy, but is not necessarily our
+	final target behavior for each mode.
+
+	+-------+---------------------------------------------------------------+
+	| Level | Description                                                   |
+	+=======+===============================================================+
+	| 0     | sa8d mode and split decisions, intra w/ source pixels         |
+	+-------+---------------------------------------------------------------+
+	| 1     | recon generated (better intra), RDO merge residual            |
+	+-------+---------------------------------------------------------------+
+	| 2     | RDO splits and merge residual choice                          |
+	+-------+---------------------------------------------------------------+
+	| 3     | RDO mode and split decisions                                  |
+	+-------+---------------------------------------------------------------+
+	| 4     | Adds RDO Quant                                                |
+	+-------+---------------------------------------------------------------+
+	| 5     | Adds RDO prediction decisions, enables intra modes in B slices|
+	+-------+---------------------------------------------------------------+
+	| 6     | Currently same as 5                                           |
+	+-------+---------------------------------------------------------------+
+
+	**Range of values:** 0: least .. 6: full RDO analysis
+
+.. option:: --cu-lossless, --no-cu-lossless
+
+	For each CU, evaluate lossless encode (transform and quant bypass)
+	as a potential rate distortion optimization. If the global option
+	:option:`--lossless` has been specified, all CUs will be encoded
+	this way unconditionally regardless of whether this option was
+	enabled. Default disabled.
+
+.. option:: --signhide, --no-signhide
+
+	Hide sign bit of one coeff per TU (rdo). The last sign is implied.
+	This requires analyzing all the coefficients to determine if a sign
+	must be toggled, and then to determine which one can be toggled with
+	the least amount of distortion. Default enabled
+ 
+Psycho-visual options
+=====================
+
+Left to its own devices, the encoder will make mode decisions based on a
+simple rate distortion formula, trading distortion for bitrate. This is
+generally effective except for the manner in which this distortion is
+measured. It tends to favor blurred reconstructed blocks over blocks
+which have wrong motion. The human eye generally prefers the wrong
+motion over the blur and thus x265 offers psycho-visual adjustments to
+the rate distortion algorithm.
+
+:option:`--psy-rd` will add an extra cost to reconstructed blocks which
+do not match the visual energy of the source block. The higher the
+strength of :option:`--psy-rd` the more strongly it will favor similar
+energy over blur and the more aggressively it will ignore rate
+distortion. If it is too high, it will introduce visal artifacts and
+increase bitrate enough for rate control to increase quantization
+globally, reducing overall quality. psy-rd will tend to reduce the use
+of blurred prediction modes, like DC and planar intra and bi-directional
+inter prediction.
+
+:option:`--psy-rdoq` will adjust the distortion cost used in
+rate-distortion optimized quantization (RDO quant), enabled in
+:option:`--rd` 4 and above, favoring the preservation of energy in the
+reconstructed image.  :option:`--psy-rdoq` prevents RDOQ from blurring
+all of the encoding options which psy-rd has to chose from.  At low
+strength levels, psy-rdoq will influence the quantization level
+decisions, favoring higher AC energy in the reconstructed image. As
+psy-rdoq strength is increased, more non-zero coefficient levels are
+added and fewer coefficients are zeroed by RDOQ's rate distortion
+analysis. High levels of psy-rdoq can double the bitrate which can have
+a drastic effect on rate control, forcing higher overall QP, and can
+cause ringing artifacts. psy-rdoq is less accurate than psy-rd, it is
+biasing towards energy in general while psy-rd biases towards the energy
+of the source image. But very large psy-rdoq values can sometimes be
+beneficial, preserving film grain for instance.
+
+As a general rule, when both psycho-visual features are disabled, the
+encoder will tend to blur blocks in areas of difficult motion. Turning
+on small amounts of psy-rd and psy-rdoq will improve the perceived
+visual quality. Increasing psycho-visual strength further will improve
+quality and begin introducing artifacts and increase bitrate, which may
+force rate control to increase global QP. Finding the optimal
+psycho-visual parameters for a given video requires experimentation. Our
+recommended defaults (1.0 for both) are generally on the low end of the
+spectrum. And generally the lower the bitrate, the lower the optimal
+psycho-visual settings.
+
+.. option:: --psy-rd <float>
+
+	Influence rate distortion optimizated mode decision to preserve the
+	energy of the source image in the encoded image at the expense of
+	compression efficiency. It only has effect on presets which use
+	RDO-based mode decisions (:option:`--rd` 3 and above).  1.0 is a
+	typical value. Default disabled.  Experimental
+
+	**Range of values:** 0 .. 2.0
+
+.. option:: --psy-rdoq <float>
+
+	Influence rate distortion optimized quantization by favoring higher
+	energy in the reconstructed image. This generally improves perceived
+	visual quality at the cost of lower quality metric scores.  It only
+	has effect on slower presets which use RDO Quantization
+	(:option:`--rd` 4, 5 and 6). 1.0 is a typical value. Default
+	disabled. Experimental
+
+	**Range of values:** 0 .. 10.0
+
+
+Slice decision options
+======================
+
+.. option:: --open-gop, --no-open-gop
+
+	Enable open GOP, allow I-slices to be non-IDR. Default enabled
+
+.. option:: --keyint, -I <integer>
+
+	Max intra period in frames. A special case of infinite-gop (single
+	keyframe at the beginning of the stream) can be triggered with
+	argument -1. Use 1 to force all-intra. Default 250
+
+.. option:: --min-keyint, -i <integer>
+
+	Minimum GOP size. Scenecuts closer together than this are coded as I
+	or P, not IDR. Minimum keyint is clamped to be at least half of
+	:option:`--keyint`. If you wish to force regular keyframe intervals
+	and disable adaptive I frame placement, you must use
+	:option:`--no-scenecut`.
+
+	**Range of values:** >=0 (0: auto)
+
+.. option:: --scenecut <integer>, --no-scenecut
+
+	How aggressively I-frames need to be inserted. The higher the
+	threshold value, the more aggressive the I-frame placement.
+	:option:`--scenecut` 0 or :option:`--no-scenecut` disables adaptive
+	I frame placement. Default 40
+
+.. option:: --rc-lookahead <integer>
+
+	Number of frames for slice-type decision lookahead (a key
+	determining factor for encoder latency). The longer the lookahead
+	buffer the more accurate scenecut decisions will be, and the more
+	effective cuTree will be at improving adaptive quant. Having a
+	lookahead larger than the max keyframe interval is not helpful.
+	Default 20
+
+	**Range of values:** Between the maximum consecutive bframe count (:option:`--bframes`) and 250
+
+.. option:: --b-adapt <integer>
+
+	Adaptive B frame scheduling. Default 2
+
+	**Values:** 0:none; 1:fast; 2:full(trellis)
+
+.. option:: --bframes, -b <0..16>
+
+	Maximum number of consecutive b-frames. Use :option:`--bframes` 0 to
+	force all P/I low-latency encodes. Default 4. This parameter has a
+	quadratic effect on the amount of memory allocated and the amount of
+	work performed by the full trellis version of :option:`--b-adapt`
+	lookahead.
+
+.. option:: --bframe-bias <integer>
+
+	Bias towards B frames in slicetype decision. The higher the bias the
+	more likely x265 is to use B frames. Can be any value between -90
+	and 100 and is clipped to that range. Default 0
+
+.. option:: --b-pyramid, --no-b-pyramid
+
+	Use B-frames as references, when possible. Default enabled
+
+.. option:: --ref <1..16>
+
+	Max number of L0 references to be allowed. This number has a linear
+	multiplier effect on the amount of work performed in motion search,
+	but will generally have a beneficial affect on compression and
+	distortion. Default 3
+
+Quality, rate control and rate distortion options
+=================================================
+
+.. option:: --bitrate <integer>
+
+	Enables single-pass ABR rate control. Specify the target bitrate in
+	kbps. Default is 0 (CRF)
+
+	**Range of values:** An integer greater than 0
+
+.. option:: --crf <0..51.0>
+
+	Quality-controlled variable bitrate. CRF is the default rate control
+	method; it does not try to reach any particular bitrate target,
+	instead it tries to achieve a given uniform quality and the size of
+	the bitstream is determined by the complexity of the source video.
+	The higher the rate factor the higher the quantization and the lower
+	the quality. Default rate factor is 28.0.
+
+.. option:: --crf-max <0..51.0>
+
+	Specify an upper limit to the rate factor which may be assigned to
+	any given frame (ensuring a max QP).  This is dangerous when CRF is
+	used in combination with VBV as it may result in buffer underruns.
+	Default disabled
+        
+.. option:: --crf-min <0..51.0>
+
+	Specify an lower limit to the rate factor which may be assigned to
+	any given frame (ensuring a min QP).  This is dangerous when CRF is
+	used in combination with VBV as it may result in buffer underruns.
+	Default disabled
+
+.. option:: --vbv-bufsize <integer>
+
+	Specify the size of the VBV buffer (kbits). Enables VBV in ABR
+	mode.  In CRF mode, :option:`--vbv-maxrate` must also be specified.
+	Default 0 (vbv disabled)
+
+.. option:: --vbv-maxrate <integer>
+
+	Maximum local bitrate (kbits/sec). Will be used only if vbv-bufsize
+	is also non-zero. Both vbv-bufsize and vbv-maxrate are required to
+	enable VBV in CRF mode. Default 0 (disabled)
+
+.. option:: --vbv-init <float>
+
+	Initial buffer occupancy. The portion of the decode buffer which
+	must be full before the decoder will begin decoding.  Determines
+	absolute maximum frame size. May be specified as a fractional value
+	between 0 and 1, or in kbits. In other words these two option pairs
+	are equivalent::
+
+	:option:`--vbv-bufsize` 1000 :option:`--vbv-init` 900
+	:option:`--vbv-bufsize` 1000 :option:`--vbv-init` 0.9
+
+	Default 0.9
+
+	**Range of values:** fractional: 0 - 1.0, or kbits: 2 .. bufsize
+
+.. option:: --qp, -q <integer>
+
+	Specify base quantization parameter for Constant QP rate control.
+	Using this option enables Constant QP rate control. The specified QP
+	is assigned to P slices. I and B slices are given QPs relative to P
+	slices using param->rc.ipFactor and param->rc.pbFactor unless QP 0
+	is specified, in which case QP 0 is used for all slice types.  Note
+	that QP 0 does not cause lossless encoding, it only disables
+	quantization. Default disabled (CRF)
+
+	**Range of values:** an integer from 0 to 51
+
+.. option:: --ipratio <float>
+
+	QP ratio factor between I and P slices. This ratio is used in all of
+	the rate control modes. Some :option:`--tune` options may change the
+	default value. It is not typically manually specified. Default 1.4
+
+.. option:: --pbratio <float>
+
+	QP ratio factor between P and B slices. This ratio is used in all of
+	the rate control modes. Some :option:`--tune` options may change the
+	default value. It is not typically manually specified. Default 1.3
+
+.. option:: --lossless, --no-lossless
+
+	Enables true lossless coding by bypassing scaling, transform,
+	quantization and in-loop filter processes. This is used for
+	ultra-high bitrates with zero loss of quality. Reconstructed output
+	pictures are bit-exact to the input pictures. Lossless encodes
+	implicitly have no rate control, all rate control options are
+	ignored. Slower presets will generally achieve better compression
+	efficiency (and generate smaller bitstreams). Default disabled.
+
+.. option:: --aq-mode <0|1|2>
+
+	Adaptive Quantization operating mode. Raise or lower per-block
+	quantization based on complexity analysis of the source image. The
+	more complex the block, the more quantization is used. This offsets
+	the tendency of the encoder to spend too many bits on complex areas
+	and not enough in flat areas.
+
+	0. disabled
+	1. AQ enabled
+	2. AQ enabled with auto-variance **(default)**
+
+.. option:: --aq-strength <float>
+
+	Adjust the strength of the adaptive quantization offsets. Setting
+	:option:`--aq-strength` to 0 disables AQ. Default 1.0.
+
+	**Range of values:** 0.0 to 3.0
+
+.. option:: --cutree, --no-cutree
+
+	Enable the use of lookahead's lowres motion vector fields to
+	determine the amount of reuse of each block to tune adaptive
+	quantization factors. CU blocks which are heavily reused as motion
+	reference for later frames are given a lower QP (more bits) while CU
+	blocks which are quickly changed and are not referenced are given
+	less bits. This tends to improve detail in the backgrounds of video
+	with less detail in areas of high motion. Default enabled
+
+.. option:: --cbqpoffs <integer>
+
+	Offset of Cb chroma QP from the luma QP selected by rate control.
+	This is a general way to spend more or less bits on the chroma
+	channel.  Default 0
+
+	**Range of values:** -12 to 12
+
+.. option:: --crqpoffs <integer>
+
+	Offset of Cr chroma QP from the luma QP selected by rate control.
+	This is a general way to spend more or less bits on the chroma
+	channel.  Default 0
+
+	**Range of values:**  -12 to 12
+
+.. option:: --pass <integer>
+
+	Enable multipass rate control mode. Input is encoded multiple times,
+	storing the encoded information of each pass in a stats file from which
+	the consecutive pass tunes the qp of each frame to improve the quality
+	of the output. Default disabled
+
+	1. First pass, creates stats file
+	2. Last pass, does not overwrite stats file
+	3. Nth pass, overwrites stats file
+
+	**Range of values:** 1 to 3
+
+.. option:: --slow-firstpass, --no-slow-firstpass
+
+	Enable a slow and more detailed first pass encode in Multipass rate
+	control mode.  Speed of the first pass encode is slightly lesser and
+	quality midly improved when compared to the default settings in a
+	multipass encode. Default disabled (turbo mode enabled)
+
+	When **turbo** first pass is not disabled, these options are
+	set on the first pass to improve performance:
+	
+	* :option:`--fast-intra`
+	* :option:`--no-rect`
+	* :option:`--no-amp`
+	* :option:`--early-skip`
+	* :option:`--ref` = 1
+	* :option:`--max-merge` = 1
+	* :option:`--me` = DIA
+	* :option:`--subme` = MIN(2, :option:`--subme`)
+	* :option:`--rd` = MIN(2, :option:`--rd`)
+
+.. option:: --analysis-mode <string|int>
+
+	Specify whether analysis information of each frame is output by encoder
+	or input for reuse. By reading the analysis data writen by an
+	earlier encode of the same sequence, substantial redundant work may
+	be avoided.
+
+	The following data may be stored and reused:
+	I frames   - split decisions and luma intra directions of all CUs.
+	P/B frames - motion vectors are dumped at each depth for all CUs.
+
+	**Values:** off(0), save(1): dump analysis data, load(2): read analysis data
+
+.. option:: --analysis-file <filename>
+
+	Specify a filename for analysis data (see :option:`--analysis-mode`)
+	If no filename is specified, x265_analysis.dat is used.
+
+Loop filters
+============
+
+.. option:: --lft, --no-lft
+
+	Toggle deblocking loop filter, default enabled
+
+.. option:: --sao, --no-sao
+
+	Toggle Sample Adaptive Offset loop filter, default enabled
+
+.. option:: --sao-lcu-bounds <0|1>
+
+	How to handle depencency with deblocking filter
+
+	0. right/bottom boundary areas skipped **(default)**
+	1. non-deblocked pixels are used
+
+.. option:: --sao-lcu-opt <0|1>
+
+	Frame level or block level optimization
+
+	0. SAO picture-based optimization (prevents frame parallelism,
+	   effectively causes :option:`--frame-threads` 1)
+	1. SAO LCU-based optimization **(default)**
+
+VUI (Video Usability Information) options
+=========================================
+
+x265 emits a VUI with only the timing info by default. If the SAR is
+specified (or read from a Y4M header) it is also included.  All other
+VUI fields must be manually specified.
+
+.. option:: --sar <integer|w:h>
+
+	Sample Aspect Ratio, the ratio of width to height of an individual
+	sample (pixel). The user may supply the width and height explicitly
+	or specify an integer from the predefined list of aspect ratios
+	defined in the HEVC specification.  Default undefined (not signaled)
+
+	1. 1:1 (square)
+	2. 12:11
+	3. 10:11
+	4. 16:11
+	5. 40:33
+	6. 24:11
+	7. 20:11
+	8. 32:11
+	9. 80:33
+	10. 18:11
+	11. 15:11
+	12. 64:33
+	13. 160:99
+	14. 4:3
+	15. 3:2
+	16. 2:1
+
+.. option:: --crop-rect <left,top,right,bottom>
+
+	Define the (overscan) region of the image that does not contain
+	information because it was added to achieve certain resolution or
+	aspect ratio. The decoder may be directed to crop away this region
+	before displaying the images via the :option:`--overscan` option.
+	Default undefined (not signaled)
+
+.. option:: --overscan <show|crop>
+
+	Specify whether it is appropriate for the decoder to display or crop
+	the overscan area. Default unspecified (not signaled)
+
+.. option:: --videoformat <integer|string>
+
+	Specify the source format of the original analog video prior to
+	digitizing and encoding. Default undefined (not signaled)
+
+	0. component
+	1. pal
+	2. ntsc
+	3. secam
+	4. mac
+	5. undefined
+
+.. option:: --range <full|limited>
+
+	Specify output range of black level and range of luma and chroma
+	signals. Default undefined (not signaled)
+
+.. option:: --colorprim <integer|string>
+
+	Specify color primitive to use when converting to RGB. Default
+	undefined (not signaled)
+
+	1. bt709
+	2. undef
+	3. **reserved**
+	4. bt470m
+	5. bt470bg
+	6. smpte170m
+	7. smpte240m
+	8. film
+	9. bt2020
+
+.. option:: --transfer <integer|string>
+
+	Specify transfer characteristics. Default undefined (not signaled)
+
+	1. bt709
+	2. undef
+	3. **reserved**
+	4. bt470m
+	5. bt470bg
+	6. smpte170m
+	7. smpte240m
+	8. linear
+	9. log100
+	10. log316
+	11. iec61966-2-4
+	12. bt1361e
+	13. iec61966-2-1
+	14. bt2020-10
+	15. bt2020-12
+
+.. option:: --colormatrix <integer|string>
+
+	Specify color matrix setting i.e set the matrix coefficients used in
+	deriving the luma and chroma. Default undefined (not signaled)
+
+	0. GBR
+	1. bt709
+	2. undef 
+	3. **reserved**
+	4. fcc
+	5. bt470bg
+	6. smpte170m
+	7. smpte240m
+	8. YCgCo
+	9. bt2020nc
+	10. bt2020c
+
+.. option:: --chromalocs <0..5>
+
+	Specify chroma sample location for 4:2:0 inputs. Consult the HEVC
+	specification for a description of these values. Default undefined
+	(not signaled)
+
+Bitstream options
+=================
+
+.. option:: --repeat-headers, --no-repeat-headers
+
+	If enabled, x265 will emit VPS, SPS, and PPS headers with every
+	keyframe. This is intended for use when you do not have a container
+	to keep the stream headers for you and you want keyframes to be
+	random access points. Default disabled
+
+.. option:: --info, --no-info
+
+	Emit an informational SEI with the stream headers which describes
+	the encoder version, build info, and encode parameters. This is very
+	helpful for debugging purposes but encoding version numbers and
+	build info could make your bitstreams diverge and interfere with
+	regression testing. Default enabled
+
+.. option:: --hrd, --no-hrd
+
+	Enable the signalling of HRD parameters to the decoder. The HRD
+	parameters are carried by the Buffering Period SEI messages and
+	Picture Timing SEI messages providing timing information to the
+	decoder. Default disabled
+
+.. option:: --aud, --no-aud
+
+	Emit an access unit delimiter NAL at the start of each slice access
+	unit. If option:`--repeat-headers` is not enabled (indicating the
+	user will be writing headers manually at the start of the stream)
+	the very first AUD will be skipped since it cannot be placed at the
+	start of the access unit, where it belongs. Default disabled
+
+.. option:: --hash <integer>
+
+	Emit decoded picture hash SEI, so the decoder may validate the
+	reconstructed pictures and detect data loss. Also useful as a
+	debug feature to validate the encoder state. Default None
+
+	1. MD5
+	2. CRC
+	3. Checksum
+
+Debugging options
+=================
+
+.. option:: --recon, -r <filename>
+
+	Output file containing reconstructed images in display order. If the
+	file extension is ".y4m" the file will contain a YUV4MPEG2 stream
+	header and frame headers. Otherwise it will be a raw YUV file in the
+	encoder's internal bit depth.
+
+	**CLI ONLY**
+
+.. option:: --recon-depth <integer>
+
+	Bit-depth of output file. This value defaults to the internal bit
+	depth and currently cannot to be modified.
+
+	**CLI ONLY**
+
+.. vim: noet

x265-hg20140928.tar.bz2/doc/reST/conf.py Added

x265-hg20140928.tar.bz2/doc/reST/index.rst Added

x265-hg20140928.tar.bz2/doc/reST/introduction.rst Added

@@ -0,0 +1,82 @@
+************
+Introduction
+************
+
+Increasing demand for high definition and ultra-high definition video,
+along with an increasing desire for video on demand has led to
+exponential growth in demand for bandwidth and storage requirements.
+These challenges can be met by the new High Efficiency Video Coding
+(HEVC) standard, also known as H.265. The x265 HEVC encoder project was
+launched by MulticoreWare in 2013, aiming to provide the most efficient,
+highest performance HEVC video encoder.
+
+About HEVC
+==========
+
+The High Efficiency Video Coding (HEVC) was developed by the ISO/IEC
+Moving Picture Experts Group (MPEG) and ITU-T Video Coding Experts Group
+(VCEG), through their Joint Collaborative Team on Video Coding (JCT-VC).
+HEVC is also known as ISO/IEC 23008-2 MPEG-H Part 2 and ITU-T H.265.
+HEVC provides superior video quality and up to twice the data
+compression as the previous standard (H.264/MPEG-4 AVC).  HEVC can
+support 8K Ultra High Definition video, with a picture size up to
+8192x4320 pixels.
+
+About x265
+==========
+
+The primary objective of x265 is to become the best H.265/HEVC encoder
+available anywhere, offering the highest compression efficiency and the
+highest performance on a wide variety of hardware platforms. The x265
+encoder is available as an open source library, published under the
+GPLv2 license. It is also available under a commercial license, enabling
+commercial companies to utilize and distribute x265 in their solutions
+without being subject to the restrictions of the GPL license.
+
+x265 is developed by `MulticoreWare <http://www.multicorewareinc.com>`_,
+leaders in high performance software solutions, with backing from
+leading video technology providers including `Telestream
+<http://www.telestream.com>`_ and `Doremi Labs
+<http://www.doremilabs.com>`_ (and other companies who want to remain
+anonymous at this time), and with contributions from open source
+developers.  x265 leverages many of the outstanding video encoding
+features and optimizations from the x264 AVC encoder project.
+
+The x265 software is available for free under the GNU GPL 2 license,
+from https://bitbucket.org/multicoreware/x265.  For commercial companies
+that wish to distribute x265 without being subject to the open source
+requirements of the GPL 2 license, commercial licenses are available
+with competitive terms.  Contact license @ x265.com to inquire about
+commercial license terms.  
+
+While x265 is primarily designed as a video encoder software library, a
+command-line executable is provided to facilitate testing and
+development.  We expect x265 to be utilized in many leading video
+hardware and software products and services in the coming months.
+
+LEGAL NOTICES
+=============
+
+The x265 software is owned and copyrighted by MulticoreWare, Inc.
+MulticoreWare is committed to offering the x265 software under the GNU
+GPL v2 license.  Companies who do not wish to integrate the x265
+Software in their products under the terms of the GPL license can
+contact MulticoreWare (license @ x265.com) to obtain a commercial
+license agreement.  Companies who use x265 under the GPL may also wish
+to work with MulticoreWare to accelerate the development of specific
+features or optimized support for specific hardware or software
+platforms, or to contract for support.
+
+The GNU GPL v2 license or the x265 commercial license agreement govern
+your rights to access the copyrighted x265 software source code, but do
+not cover any patents that may be applicable to the function of binary
+executable software created from the x265 source code.  You are
+responsible for understanding the laws in your country, and for
+licensing all applicable patent rights needed for use or distribution of
+software applications created from the x265 source code.  A good place
+to start is with the `Motion Picture Experts Group - Licensing Authority
+- HEVC Licensing Program<http://www.mpegla.com/main/PID/HEVC/default.aspx>`_.
+
+x265 is a registered trademark of MulticoreWare, Inc.  The x265 logo is
+a trademark of MulticoreWare, and may only be used with explicit written
+permission.  All rights reserved.

x265-hg20140928.tar.bz2/doc/reST/lossless.rst Added

@@ -0,0 +1,162 @@
+Lossless
+--------
+
+Lossless Encoding
+=================
+
+x265 can encode HEVC bitstreams that are entirely lossless (the
+reconstructed images are bit-exact to the source images) by using the
+:option:`--lossless` option.  Lossless operation is theoretically
+simple. Rate control, by definition, is disabled and the encoder
+disables all quality metrics since they would only waste CPU cycles.
+Instead, x265 reports only a compression factor at the end of the
+encode.
+
+In HEVC, lossless coding means bypassing both the DCT transforms and
+bypassing quantization (often referred to as transquant bypass).  Normal
+predictions are still allowed, so the encoder will find optimal inter or
+intra predictions and then losslessly code the residual (with transquant
+bypass).
+
+All :option:`--preset` options are capable of generating lossless video
+streams, but in general the slower the preset the better the compression
+ratio (and the slower the encode). Here are some examples::
+
+	./x265 ../test-720p.y4m o.bin --preset ultrafast --lossless
+	... <snip> ...
+	encoded 721 frames in 238.38s (3.02 fps), 57457.94 kb/s
+
+	./x265 ../test-720p.y4m o.bin --preset faster --lossless
+	... <snip> ...
+	x265 [info]: lossless compression ratio 3.11::1
+	encoded 721 frames in 258.46s (2.79 fps), 56787.65 kb/s
+
+	./x265 ../test-720p.y4m o.bin --preset slow --lossless
+	... <snip> ...
+	x265 [info]: lossless compression ratio 3.36::1
+	encoded 721 frames in 576.73s (1.25 fps), 52668.25 kb/s
+
+	./x265 ../test-720p.y4m o.bin --preset veryslow --lossless
+	x265 [info]: lossless compression ratio 3.76::1
+	encoded 721 frames in 6298.22s (0.11 fps), 47008.65 kb/s
+ 
+.. Note::
+	In HEVC, only QP=4 is truly lossless quantization, and thus when
+	encoding losslesly x265 uses QP=4 internally in its RDO decisions.
+
+Near-lossless Encoding
+======================
+
+Near-lossless conditions are a quite a bit more interesting.  Normal ABR
+rate control will allow one to scale the bitrate up to the point where 
+quantization is entirely bypassed (QP <= 4), but even at this point
+there is a lot of SSIM left on the table because of the DCT transforms,
+which are not lossless::
+
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 40000 --ssim
+	encoded 721 frames in 326.62s (2.21 fps), 39750.56 kb/s, SSIM Mean Y: 0.9990703 (30.317 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 50000 --ssim
+	encoded 721 frames in 349.27s (2.06 fps), 44326.84 kb/s, SSIM Mean Y: 0.9994134 (32.316 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 60000 --ssim
+	encoded 721 frames in 360.04s (2.00 fps), 45394.50 kb/s, SSIM Mean Y: 0.9994823 (32.859 dB)
+
+For the encoder to get over this quality plateau, one must enable
+lossless coding at the CU level with :option:`--cu-lossless`.  It tells
+the encoder to evaluate trans-quant bypass as a coding option for each
+CU, and to pick the option with the best rate-distortion
+characteristics.
+
+The :option:`--cu-lossless` option is very expensive, computationally,
+and it only has a positive effect when the QP is extremely low, allowing
+RDO to spend a large amount of bits to make small improvements to
+quality.  So this option should only be enabled when you are encoding
+near-lossless bitstreams::
+
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 40000 --ssim --cu-lossless
+	encoded 721 frames in 500.51s (1.44 fps), 40017.10 kb/s, SSIM Mean Y: 0.9997790 (36.557 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 50000 --ssim --cu-lossless
+	encoded 721 frames in 524.60s (1.37 fps), 46083.37 kb/s, SSIM Mean Y: 0.9999432 (42.456 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 60000 --ssim --cu-lossless
+	encoded 721 frames in 523.63s (1.38 fps), 46552.92 kb/s, SSIM Mean Y: 0.9999489 (42.917 dB)
+
+.. Note::
+	It is not unusual for bitrate to drop as you increase lossless coding.
+	Having "perfectly coded" reference blocks reduces residual in later
+	frames. It is quite possible for a near-lossless encode to spend
+	more bits than a lossless encode.
+
+Enabling psycho-visual rate distortion will improve lossless coding.
+:option:`--psy-rd` influences the RDO decisions in favor of energy
+(detail) preservation over bit cost and results in more blocks being
+losslessly coded.  Our psy-rd feature is not yet assembly optimized, so
+this makes the encodes run even slower::
+
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 40000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 581.83s (1.24 fps), 40112.15 kb/s, SSIM Mean Y: 0.9998632 (38.638 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 50000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 587.54s (1.23 fps), 46284.55 kb/s, SSIM Mean Y: 0.9999663 (44.721 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset medium --bitrate 60000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 592.93s (1.22 fps), 46839.51 kb/s, SSIM Mean Y: 0.9999707 (45.334 dB)
+
+:option:`--cu-lossless` will also be more effective at slower
+presets which perform RDO at more levels and thus may find smaller
+blocks that would benefit from lossless coding::
+
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 40000 --ssim --cu-lossless
+	encoded 721 frames in 12969.25s (0.06 fps), 37331.96 kb/s, SSIM Mean Y: 0.9998108 (37.231 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 50000 --ssim --cu-lossless
+	encoded 721 frames in 46217.84s (0.05 fps), 42976.28 kb/s, SSIM Mean Y: 0.9999482 (42.856 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 60000 --ssim --cu-lossless
+	encoded 721 frames in 13738.17s (0.05 fps), 43864.21 kb/s, SSIM Mean Y: 0.9999633 (44.348 dB)
+	
+And with psy-rd and a slow preset together, very high SSIMs are
+possible::
+
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 40000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 11675.81s (0.06 fps), 37819.45 kb/s, SSIM Mean Y: 0.9999181 (40.867 dB)
+	    
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 50000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 12414.56s (0.06 fps), 42815.75 kb/s, SSIM Mean Y: 0.9999758 (46.168 dB)
+	
+	./x265 ../test-720p.y4m o.bin --preset veryslow --bitrate 60000 --ssim --cu-lossless --psy-rd 1.0
+	encoded 721 frames in 11684.89s (0.06 fps), 43324.48 kb/s, SSIM Mean Y: 0.9999793 (46.835 dB)
+
+
+It's important to note in the end that it is easier (less work) for the
+encoder to encode the video losslessly than it is to encode it
+near-losslessly. If the encoder knows up front the encode must be
+lossless, it does not need to evaluate any lossy coding methods. The
+encoder only needs to find the most efficient prediction for each block
+and then entropy code the residual.
+
+It is not feasible for :option:`--cu-lossless` to turn itself on when
+the encoder determines it is encoding a near-lossless bitstream (ie:
+when rate control nearly disables all quantization) because the feature
+requires a flag to be enabled in the stream headers. At the time the
+stream headers are being coded we do not know whether
+:option:`--cu-lossless` would be a help or a hinder.  If very few or no
+blocks end up being coded as lossless, then having the feature enabled
+is a net loss in compression efficiency because it adds a flag that must
+be coded for every CU. So ignoring even the performance aspects of the
+feature, it can be a compression loss if enabled without being used. So
+it is up to the user to only enable this feature when they are coding at
+near-lossless quality.
+
+Transform Skip
+==============
+
+A somewhat related feature, :option:`--tskip` tells the encoder to
+evaluate transform-skip (bypass DCT but with quantization still enabled)
+when coding small intra blocks. This feature is intended to improve the
+coding efficiency of screen content (aka: text on a screen) and is not
+really intended for lossless coding.  This feature should only be
+enabled if the content has a lot of very sharp edges in it, and is
+mostly unrelated to lossless coding.

x265-hg20140928.tar.bz2/doc/reST/presets.rst Added

@@ -0,0 +1,104 @@
+Preset Options
+--------------
+
+Presets
+=======
+
+.. _preset-tune-ref:
+
+x265 has a number of predefined :option:`--preset` options that make
+trade-offs between encode speed (encoded frames per second) and
+compression efficiency (quality per bit in the bitstream).  The default
+preset is medium, it does a reasonably good job of finding the best
+possible quality without spending enormous CPU cycles looking for the
+absolute most efficient way to achieve that quality.  As you go higher
+than medium, the encoder takes shortcuts to improve performance at the
+expense of quality and compression efficiency.  As you go lower than
+medium, the encoder tries harder and harder to achieve the best quailty
+per bit compression ratio.
+
+The presets adjust encoder parameters to affect these trade-offs.
+
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+|              | ultrafast | superfast | veryfast | faster | fast | medium | slow | slower | veryslow | placebo |
++==============+===========+===========+==========+========+======+========+======+========+==========+=========+
+| ctu          |   32      |    32     |   32     |  64    |  64  |   64   |  64  |  64    |   64     |   64    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| bframes      |    4      |     4     |    4     |   4    |  4   |    4   |  4   |   8    |    8     |    8    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| b-adapt      |    0      |     0     |    0     |   0    |  2   |    2   |  2   |   2    |    2     |    2    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| rc-lookahead |   10      |    10     |   15     |  15    |  15  |   20   |  25  |   30   |   40     |   60    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| scenecut     |    0      |    40     |   40     |  40    |  40  |   40   |  40  |   40   |   40     |   40    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| refs         |    1      |     1     |    1     |   1    |  3   |    3   |  3   |   3    |    5     |    5    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| me           |   dia     |   hex     |   hex    |  hex   | hex  |   hex  | star |  star  |   star   |   star  |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| merange      |   25      |    44     |   57     |  57    |  57  |   57   | 57   |  57    |   57     |   92    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| subme        |    0      |     1     |    1     |   2    |  2   |    2   |  3   |   3    |    4     |    5    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| rect         |    0      |     0     |    0     |   0    |  0   |    0   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| amp          |    0      |     0     |    0     |   0    |  0   |    0   |  0   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| max-merge    |    2      |     2     |    2     |   2    |  2   |    2   |  3   |   3    |    4     |    5    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| early-skip   |    1      |     1     |    1     |   1    |  0   |    0   |  0   |   0    |    0     |    0    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| fast-intra   |    1      |     1     |    1     |   1    |  1   |    0   |  0   |   0    |    0     |    0    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| fast-cbf     |    1      |     1     |    1     |   1    |  0   |    0   |  0   |   0    |    0     |    0    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| sao          |    0      |     0     |    1     |   1    |  1   |    1   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| signhide     |    0      |     1     |    1     |   1    |  1   |    1   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| weightp      |    0      |     0     |    1     |   1    |  1   |    1   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| weightb      |    0      |     0     |    0     |   0    |  0   |    0   |  0   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| aq-mode      |    0      |     0     |    2     |   2    |  2   |    2   |  2   |   2    |    2     |    2    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| cuTree       |    0      |     0     |    0     |   0    |  1   |    1   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| rdLevel      |    2      |     2     |    2     |   2    |  2   |    3   |  4   |   6    |    6     |    6    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| lft          |    0      |     1     |    1     |   1    |  1   |    1   |  1   |   1    |    1     |    1    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| tu-intra     |    1      |     1     |    1     |   1    |  1   |    1   |  1   |   2    |    3     |    4    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+| tu-inter     |    1      |     1     |    1     |   1    |  1   |    1   |  1   |   2    |    3     |    4    |
++--------------+-----------+-----------+----------+--------+------+--------+------+--------+----------+---------+
+
+Placebo mode further enables transform-skip prediction analysis
+(lossless).
+
+Tuning
+======
+
+There are a few :option:`--tune` options available, which are applied
+after the preset.
+
+.. Note::
+
+	The *psnr* and *ssim* tune options disable all optimizations that
+	sacrafice metric scores for perceived visual quality (also known as
+	psycho-visual optimizations). By default x265 always tunes for
+	highest perceived visual quality but if one intends to measure an
+	encode using PSNR or SSIM for the purpose of benchmarking, we highly
+	recommend you configure x265 to tune for that particular metric.
+
++--------------+-----------------------------------------------------+
+| --tune       | effect                                              |
++==============+=====================================================+
+| psnr         | disables adaptive quant, psy-rd, and cutree         |
++--------------+-----------------------------------------------------+
+| ssim         | enables adaptive quant auto-mode, disables psy-rd   |
++--------------+-----------------------------------------------------+
+| fastdecode   | no loop filters, no weighted pred, no intra in B    |
++--------------+-----------------------------------------------------+
+| zerolatency  | no lookahead, no B frames, no cutree                |
++--------------+-----------------------------------------------------+

x265-hg20140928.tar.bz2/doc/reST/threading.rst Added

@@ -0,0 +1,227 @@
+*********
+Threading
+*********
+
+Thread Pool
+===========
+
+x265 creates a pool of worker threads and shares this thread pool
+with all encoders within the same process (it is process global, aka a
+singleton).  The number of threads within the thread pool is determined
+by the encoder which first allocates the pool, which by definition is
+the first encoder created within each process.
+
+:option:`--threads` specifies the number of threads the encoder will
+try to allocate for its thread pool.  If the thread pool was already
+allocated this parameter is ignored.  By default x265 allocates one
+thread per (hyperthreaded) CPU core in your system.
+
+Work distribution is job based.  Idle worker threads ask their parent
+pool object for jobs to perform.  When no jobs are available, idle
+worker threads block and consume no CPU cycles.
+
+Objects which desire to distribute work to worker threads are known as
+job providers (and they derive from the JobProvider class).  When job
+providers have work they enqueue themselves into the pool's provider
+list (and dequeue themselves when they no longer have work).  The thread
+pool has a method to **poke** awake a blocked idle thread, and job
+providers are recommended to call this method when they make new jobs
+available.
+
+Worker jobs are not allowed to block except when abosultely necessary
+for data locking. If a job becomes blocked, the worker thread is
+expected to drop that job and go back to the pool and find more work.
+
+.. note::
+
+	x265_cleanup() frees the process-global thread pool, allowing
+	it to be reallocated if necessary, but only if no encoders are
+	allocated at the time it is called.
+
+Wavefront Parallel Processing
+=============================
+
+New with HEVC, Wavefront Parallel Processing allows each row of CTUs to
+be encoded in parallel, so long as each row stays at least two CTUs
+behind the row above it, to ensure the intra references and other data
+of the blocks above and above-right are available. WPP has almost no
+effect on the analysis and compression of each CTU and so it has a very
+small impact on compression efficiency relative to slices or tiles. The
+compression loss from WPP has been found to be less than 1% in most of
+our tests.
+
+WPP has three effects which can impact efficiency. The first is the row
+starts must be signaled in the slice header, the second is each row must
+be padded to an even byte in length, and the third is the state of the
+entropy coder is transferred from the second CTU of each row to the
+first CTU of the row below it.  In some conditions this transfer of
+state actually improves compression since the above-right state may have
+better locality than the end of the previous row.
+
+Parabola Research have published an excellent HEVC
+`animation <http://www.parabolaresearch.com/blog/2013-12-01-hevc-wavefront-animation.html>`_
+which visualizes WPP very well.  It even correctly visualizes some of
+WPPs key drawbacks, such as:
+
+1. the low thread utilization at the start and end of each frame
+2. a difficult block may stall the wave-front and it takes a while for
+   the wave-front to recover.
+3. 64x64 CTUs are big! there are much fewer rows than with H.264 and
+   similar codecs
+
+Because of these stall issues you rarely get the full parallelisation
+benefit one would expect from row threading. 30% to 50% of the
+theoretical perfect threading is typical.
+
+In x265 WPP is enabled by default since it not only improves performance
+at encode but it also makes it possible for the decoder to be threaded.
+
+If WPP is disabled by :option:`--no-wpp` the frame will be encoded in
+scan order and the entropy overheads will be avoided.  If frame
+threading is not disabled, the encoder will change the default frame
+thread count to be higher than if WPP was enabled.  The exact formulas
+are described in the next section.
+
+
+Frame Threading
+===============
+
+Frame threading is the act of encoding multiple frames at the same time.
+It is a challenge because each frame will generally use one or more of
+the previously encoded frames as motion references and those frames may
+still be in the process of being encoded themselves.
+
+Previous encoders such as x264 worked around this problem by limiting
+the motion search region within these reference frames to just one
+macroblock row below the coincident row being encoded. Thus a frame
+could be encoded at the same time as its reference frames so long as it
+stayed one row behind the encode progress of its references (glossing
+over a few details). 
+
+x265 has the same frame threading mechanism, but we generally have much
+less frame parallelism to exploit than x264 because of the size of our
+CTU rows. For instance, with 1080p video x264 has 68 16x16 macroblock
+rows available each frame while x265 only has 17 64x64 CTU rows.
+
+The second extenuating circumstance is the loop filters. The pixels used
+for motion reference must be processed by the loop filters and the loop
+filters cannot run until a full row has been encoded, and it must run a
+full row behind the encode process so that the pixels below the row
+being filtered are available. When you add up all the row lags each
+frame ends up being 3 CTU rows behind its reference frames (the
+equivalent of 12 macroblock rows for x264)
+
+The third extenuating circumstance is that when a frame being encoded
+becomes blocked by a reference frame row being available, that frame's
+wave-front becomes completely stalled and when the row becomes available
+again it can take quite some time for the wave to be restarted, if it
+ever does. This makes WPP many times less effective when frame
+parallelism is in use.
+
+:option:`--merange` can have a negative impact on frame parallelism. If
+the range is too large, more rows of CTU lag must be added to ensure
+those pixels are available in the reference frames.  Similarly
+:option:`--sao-lcu-opt` 0 will cause SAO to be performed over the
+entire picture at once (rather than being CTU based), which prevents any
+motion reference pixels from being available until the entire frame has
+been encoded, which prevents any real frame parallelism at all.
+
+.. note::
+
+	Even though the merange is used to determine the amount of reference
+	pixels that must be available in the reference frames, the actual
+	motion search is not necessarily centered around the coincident
+	block. The motion search is actually centered around the motion
+	predictor, but the available pixel area (mvmin, mvmax) is determined
+	by merange and the interpolation filter half-heights.
+
+When frame threading is disabled, the entirety of all reference frames
+are always fully available (by definition) and thus the available pixel
+area is not restricted at all, and this can sometimes improve
+compression efficiency. Because of this, the output of encodes with
+frame parallelism disabled will not match the output of encodes with
+frame parallelism enabled; but when enabled the number of frame threads
+should have no effect on the output bitstream except when using ABR or
+VBV rate control or noise reduction.
+
+When :option:`--nr` is enabled, the outputs of each number of frame threads
+will be deterministic but none of them will match becaue each frame
+encoder maintains a cumulative noise reduction state.
+
+VBV introduces non-determinism in the encoder, at this point in time,
+regardless of the amount of frame parallelism.
+
+By default frame parallelism and WPP are enabled together. The number of
+frame threads used is auto-detected from the (hyperthreaded) CPU core
+count, but may be manually specified via :option:`--frame-threads`
+
+	+-------+--------+
+	| Cores | Frames |
+	+=======+========+
+	|  > 32 |   6    |
+	+-------+--------+
+	| >= 16 |   5    |
+	+-------+--------+
+	| >= 8  |   3    |
+	+-------+--------+
+	| >= 4  |   2    |
+	+-------+--------+
+
+If WPP is disabled, then the frame thread count defaults to **min(cpuCount, ctuRows / 2)**
+
+Over-allocating frame threads can be very counter-productive. They
+each allocate a large amount of memory and because of the limited number
+of CTU rows and the reference lag, you generally get limited benefit
+from adding frame encoders beyond the auto-detected count, and often
+the extra frame encoders reduce performance.
+
+Given these considerations, you can understand why the faster presets
+lower the max CTU size to 32x32 (making twice as many CTU rows available
+for WPP and for finer grained frame parallelism) and reduce
+:option:`--merange`
+
+Each frame encoder runs in its own thread (allocated separately from the
+worker pool). This frame thread has some pre-processing responsibilities
+and some post-processing responsibilities for each frame, but it spends
+the bulk of its time managing the wave-front processing by making CTU
+rows available to the worker threads when their dependencies are
+resolved.  The frame encoder threads spend nearly all of their time
+blocked in one of 4 possible locations:
+
+1. blocked, waiting for a frame to process
+2. blocked on a reference frame, waiting for a CTU row of reconstructed
+   and loop-filtered reference pixels to become available
+3. blocked waiting for wave-front completion
+4. blocked waiting for the main thread to consume an encoded frame
+
+Lookahead
+=========
+
+The lookahead module of x265 (the lowres pre-encode which determines
+scene cuts and slice types) uses the thread pool to distribute the
+lowres cost analysis to worker threads. It follows the same wave-front
+pattern as the main encoder except it works in reverse-scan order.
+
+The function slicetypeDecide() itself may also be performed by a worker
+thread if your system has enough CPU cores to make this a beneficial
+trade-off, else it runs within the context of the thread which calls the
+x265_encoder_encode().
+
+SAO
+===
+
+The Sample Adaptive Offset loopfilter has a large effect on encode
+performance because of the peculiar way it must be analyzed and coded.
+
+SAO flags and data are encoded at the CTU level before the CTU itself is
+coded, but SAO analysis (deciding whether to enable SAO and with what
+parameters) cannot be performed until that CTU is completely analyzed
+(reconstructed pixels are available) as well as the CTUs to the right
+and below.  So in effect the encoder must perform SAO analysis in a
+wavefront at least a full row behind the CTU compression wavefront.
+
+This extra latency forces the encoder to save the encode data of every
+CTU until the entire frame has been analyzed, at which point a function
+can code the final slice bitstream with the decided SAO flags and data
+coded between each CTU.  This second pass over the CTUs can be
+expensive, particularly at large resolutions and high bitrates.

x265-hg20140324.tar.bz2/source/CMakeLists.txt -> x265-hg20140928.tar.bz2/source/CMakeLists.txt Changed

@@ -6,9 +6,12 @@
         FORCE)
 endif()
 message(STATUS "cmake version ${CMAKE_VERSION}")
-if(NOT CMAKE_VERSION VERSION_LESS "2.8.12.20131121")
+if(POLICY CMP0025)
     cmake_policy(SET CMP0025 OLD) # report Apple's Clang as just Clang
 endif()
+if(POLICY CMP0042)
+    cmake_policy(SET CMP0042 NEW) # MACOSX_RPATH
+endif()
 
 project (x265)
 cmake_minimum_required (VERSION 2.8.8) # OBJECT libraries require 2.8.8
@@ -18,7 +21,7 @@
 include(CheckCXXCompilerFlag)
 
 # X265_BUILD must be incremented each time the public API is changed
-set(X265_BUILD 10)
+set(X265_BUILD 32)
 configure_file("${PROJECT_SOURCE_DIR}/x265.def.in"
                "${PROJECT_BINARY_DIR}/x265.def")
 configure_file("${PROJECT_SOURCE_DIR}/x265_config.h.in"
@@ -26,12 +29,17 @@
 
 SET(CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake" "${CMAKE_MODULE_PATH}")
 
+option(CHECKED_BUILD "Enable run-time sanity checks (debugging)" OFF)
+if(CHECKED_BUILD)
+    add_definitions(-DCHECKED_BUILD=1)
+endif()
+
 # System architecture detection
 string(TOLOWER "${CMAKE_SYSTEM_PROCESSOR}" SYSPROC)
 set(X86_ALIASES x86 i386 i686 x86_64 amd64)
 list(FIND X86_ALIASES "${SYSPROC}" X86MATCH)
 if("${SYSPROC}" STREQUAL "" OR X86MATCH GREATER "-1")
-    message(STATUS "Detected x86 system processor")
+    message(STATUS "Detected x86 target processor")
     set(X86 1)
     add_definitions(-DX265_ARCH_X86=1)
     if("${CMAKE_SIZEOF_VOID_P}" MATCHES 8)
@@ -39,7 +47,7 @@
         add_definitions(-DX86_64=1)
     endif()
 elseif(${SYSPROC} STREQUAL "armv6l")
-    message(STATUS "Detected ARM system processor")
+    message(STATUS "Detected ARM target processor")
     set(ARM 1)
     add_definitions(-DX265_ARCH_ARM=1 -DHAVE_ARMV6=1)
 else()
@@ -49,9 +57,10 @@
 
 if(UNIX)
     SET(PLATFORM_LIBS pthread)
-    if(NOT ${CMAKE_SYSTEM_NAME} MATCHES "Darwin")
+    find_library(LIBRT rt)
+    if(LIBRT)
         SET(PLATFORM_LIBS ${PLATFORM_LIBS} rt)
-    endif(NOT ${CMAKE_SYSTEM_NAME} MATCHES "Darwin")
+    endif()
 endif(UNIX)
 
 # Compiler detection
@@ -97,24 +106,21 @@
     endif()
 endif(MSVC)
 
+check_include_files(inttypes.h HAVE_INT_TYPES_H)
+if(HAVE_INT_TYPES_H)
+    add_definitions(-DHAVE_INT_TYPES_H=1)
+endif()
+
 if(INTEL_CXX AND UNIX)
-    # treat icpc roughly like gcc
-    set(GCC 1)
-    add_definitions(-Wall -Wextra -Wshadow -no-vec)
+    set(GCC 1) # treat icpc roughly like gcc
 elseif(CLANG)
-    # treat clang roughly like gcc
-    set(GCC 1)
-    add_definitions(-Wall -Wextra -Wshadow -ffast-math)
+    set(GCC 1) # treat clang roughly like gcc
 elseif(CMAKE_COMPILER_IS_GNUCXX)
-    add_definitions(-Wall -Wextra -Wshadow -ffast-math)
-    check_cxx_compiler_flag(-Wno-narrowing GCC_HAS_NO_NARROWING) 
-    check_cxx_compiler_flag(-mstackrealign GCC_HAS_STACK_REALIGN) 
-    if (GCC_HAS_STACK_REALIGN)
-        add_definitions(-mstackrealign)
-    endif()
-    execute_process(COMMAND ${CMAKE_CXX_COMPILER} -dumpversion OUTPUT_VARIABLE GCC_VERSION)
+    set(GCC 1)
 endif()
 if(GCC)
+    add_definitions(-Wall -Wextra -Wshadow)
+    add_definitions(-D__STDC_LIMIT_MACROS=1)
     if(X64 AND NOT WIN32)
         add_definitions(-fPIC)
     endif(X64 AND NOT WIN32)
@@ -124,6 +130,21 @@
     if(ARM)
         add_definitions(-march=armv6 -mfloat-abi=hard -mfpu=vfp)
     endif()
+    check_cxx_compiler_flag(-Wno-narrowing CC_HAS_NO_NARROWING) 
+    check_cxx_compiler_flag(-ffast-math CC_HAS_FAST_MATH) 
+    if (CC_HAS_FAST_MATH)
+        add_definitions(-ffast-math)
+    endif()
+    check_cxx_compiler_flag(-mstackrealign CC_HAS_STACK_REALIGN) 
+    if (CC_HAS_STACK_REALIGN)
+        add_definitions(-mstackrealign)
+    endif()
+    # Disable exceptions. Reduce executable size, increase compability.
+    check_cxx_compiler_flag(-fno-exceptions CC_HAS_FNO_EXCEPTIONS_FLAG)
+    if(CC_HAS_FNO_EXCEPTIONS_FLAG)
+        add_definitions(-fno-exceptions)
+    endif()
+    execute_process(COMMAND ${CMAKE_CXX_COMPILER} -dumpversion OUTPUT_VARIABLE CC_VERSION)
 endif(GCC)
 
 find_package(Yasm)
@@ -165,12 +186,7 @@
     endif()
 endif(WARNINGS_AS_ERRORS)
 
-option(LOG_CU_STATISTICS "Log Mode Decision Statistics at the CU Level" OFF)
-if(LOG_CU_STATISTICS)
-    add_definitions(-DLOG_CU_STATISTICS=1)
-else(LOG_CU_STATISTICS)
-    add_definitions(-DLOG_CU_STATISTICS=0)
-endif(LOG_CU_STATISTICS)
+
 
 option(ENABLE_PPA "Enable PPA profiling instrumentation" OFF)
 if(ENABLE_PPA)
@@ -191,10 +207,12 @@
         set(PLATFORM_LIBS ${PLATFORM_LIBS} ${VLD_LIBRARIES})
         link_directories(${VLD_LIBRARY_DIRS})
     endif()
-    if(MINGW)
-        # MinGW requires a forced Windows version in order to use CONDITION_VARIABLE
-        add_definitions(-D_WIN32_WINNT=_WIN32_WINNT_VISTA)
-    endif(MINGW)
+    option(WINXP_SUPPORT "Make binaries compatible with Windows XP" OFF)
+    if(WINXP_SUPPORT)
+        # force use of workarounds for CONDITION_VARIABLE and atomic
+        # intrinsics introduced after XP
+        add_definitions(-D_WIN32_WINNT=_WIN32_WINNT_WINXP)
+    endif()
 endif()
 
 include(version) # determine X265_VERSION and X265_LATEST_TAG
@@ -203,43 +221,25 @@
 add_subdirectory(common)
 
 if((MSVC_IDE OR XCODE) AND ENABLE_ASSEMBLY)
-    # this is horrible. ugly, and hacky, and it reproduces logic found
-    # in the yasm CMake modules, but this is required because of this cmake bug
+    # this is required because of this cmake bug
     # http://www.cmake.org/Bug/print_bug_page.php?bug_id=8170
-    if (X64)
-        if(APPLE)
-            set(FLAGS -f macho64 -m amd64 -DPREFIX -DPIC -DARCH_X86_64=1 -DHAVE_ALIGNED_STACK=1)
-        else()
-            set(FLAGS -f win64 -m amd64 -DARCH_X86_64=1 -DHAVE_ALIGNED_STACK=0)
-        endif()
-    else()
-        if(APPLE)
-            set(FLAGS -f macho -DARCH_X86_64=0 -DHAVE_ALIGNED_STACK=1 -DPREFIX)
-        else()
-            set(FLAGS -f win32 -DARCH_X86_64=0 -DHAVE_ALIGNED_STACK=0 -DPREFIX)
-        endif()
-    endif()
     if(WIN32)
         set(SUFFIX obj)
     else()
         set(SUFFIX o)
     endif()
-    if (HIGH_BIT_DEPTH)
-        set(FLAGS ${FLAGS} -DHIGH_BIT_DEPTH=1 -DBIT_DEPTH=10)
-    else()
-        set(FLAGS ${FLAGS} -DHIGH_BIT_DEPTH=0 -DBIT_DEPTH=8)
-    endif()
     foreach(ASM ${MSVC_ASMS})
         set(YASM_SRC ${CMAKE_CURRENT_SOURCE_DIR}/common/x86/${ASM})
-        set(YASM_SRCS ${YASM_SRCS} ${YASM_SRC})
-        set(YASM_OBJS ${YASM_OBJS} ${ASM}.${SUFFIX})
+        list(APPEND YASM_SRCS ${YASM_SRC})
+        list(APPEND YASM_OBJS ${ASM}.${SUFFIX})
         add_custom_command(
             OUTPUT ${ASM}.${SUFFIX}
-            COMMAND ${YASM_EXECUTABLE} ARGS ${FLAGS} ${YASM_SRC} -o ${ASM}.${SUFFIX}
+            COMMAND ${YASM_EXECUTABLE} ARGS ${YASM_FLAGS} ${YASM_SRC} -o ${ASM}.${SUFFIX}
             DEPENDS ${YASM_SRC})
     endforeach()
 endif()
 
+source_group(ASM FILES ${YASM_SRCS})
 add_library(x265-static STATIC $<TARGET_OBJECTS:encoder> $<TARGET_OBJECTS:common> ${YASM_OBJS} ${YASM_SRCS})
 if(NOT MSVC)
     set_target_properties(x265-static PROPERTIES OUTPUT_NAME x265)
@@ -265,6 +265,10 @@
     configure_file("${CMAKE_CURRENT_SOURCE_DIR}/x265.rc.in" "${X265_RC_FILE}" @ONLY)
 endif()
 
+if(NOT (MSVC_IDE OR XCODE))
+    add_custom_target(clean-generated COMMAND ${CMAKE_COMMAND} -P ${CMAKE_CURRENT_SOURCE_DIR}/cmake/clean-generated.cmake)
+endif()
+
 option(ENABLE_SHARED "Build shared library" ON)
 if(ENABLE_SHARED)
     add_library(x265-shared SHARED "${PROJECT_BINARY_DIR}/x265.def" ${YASM_OBJS}
@@ -275,22 +279,35 @@
     else()
         set_target_properties(x265-shared PROPERTIES OUTPUT_NAME x265)
     endif()
-    if(UNIX AND NOT APPLE)
-        set_target_properties(x265-shared PROPERTIES LINK_FLAGS "-Wl,-Bsymbolic,-znoexecstack")
+    if(UNIX)
+        set_target_properties(x265-shared PROPERTIES VERSION ${X265_BUILD})
+        if(APPLE)
+            set_target_properties(x265-shared PROPERTIES MACOSX_RPATH 1)
+        else()
+            set_target_properties(x265-shared PROPERTIES LINK_FLAGS "-Wl,-Bsymbolic,-znoexecstack")
+        endif()
     endif()
+    set_target_properties(x265-shared PROPERTIES SOVERSION ${X265_BUILD})
     if(X265_LATEST_TAG)
+        if(WINDOWS)
+            set_target_properties(x265-shared PROPERTIES VERSION ${X265_LATEST_TAG})
+        endif()
         # shared library is not installed if a tag is not found
-        set_target_properties(x265-shared PROPERTIES VERSION ${X265_LATEST_TAG} SOVERSION ${X265_BUILD})
         install(TARGETS x265-shared
                 LIBRARY DESTINATION ${LIB_INSTALL_DIR}
-                ARCHIVE DESTINATION ${LIB_INSTALL_DIR})
+                ARCHIVE DESTINATION ${LIB_INSTALL_DIR}
+                RUNTIME DESTINATION ${BIN_INSTALL_DIR})
     endif()
 endif()
 
 if(X265_LATEST_TAG)
     # convert lists of link libraries into -lstdc++ -lm etc..
     foreach(LIB ${CMAKE_CXX_IMPLICIT_LINK_LIBRARIES} ${PLATFORM_LIBS})
-        list(APPEND PLIBLIST "-l${LIB}")
+        if(IS_ABSOLUTE ${LIB} AND EXISTS ${LIB})
+            list(APPEND PLIBLIST "${LIB}")
+        else()
+            list(APPEND PLIBLIST "-l${LIB}")
+        endif()
     endforeach()
     if(PLIBLIST)
         # blacklist of libraries that should not be in Libs.private
@@ -319,8 +336,10 @@
 if(ENABLE_CLI)
     file(GLOB InputFiles input/*.cpp input/*.h)
     file(GLOB OutputFiles output/*.cpp output/*.h)
+    file(GLOB FilterFiles filters/*.cpp filters/*.h)
     source_group(input FILES ${InputFiles})
     source_group(output FILES ${OutputFiles})
+    source_group(filters FILES ${FilterFiles})
 
     check_include_files(getopt.h HAVE_GETOPT_H)
     if(NOT HAVE_GETOPT_H)
@@ -333,10 +352,10 @@
 
     if(XCODE)
         # Xcode seems unable to link the CLI with libs, so link as one targget
-        add_executable(cli ../COPYING ${InputFiles} ${OutputFiles} ${GETOPT} x265.cpp x265.h
+        add_executable(cli ../COPYING ${InputFiles} ${OutputFiles} ${FilterFiles} ${GETOPT} x265.cpp x265.h
                            $<TARGET_OBJECTS:encoder> $<TARGET_OBJECTS:common> ${YASM_OBJS} ${YASM_SRCS})
     else()
-        add_executable(cli ../COPYING ${InputFiles} ${OutputFiles} ${GETOPT} ${X265_RC_FILE} x265.cpp x265.h)
+        add_executable(cli ../COPYING ${InputFiles} ${OutputFiles} ${FilterFiles} ${GETOPT} ${X265_RC_FILE} x265.cpp x265.h)
         if(WIN32 OR NOT ENABLE_SHARED OR INTEL_CXX)
             # The CLI cannot link to the shared library on Windows, it
             # requires internal APIs not exported from the DLL

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/CommonDef.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/CommonDef.h Changed

@@ -31,131 +31,73 @@
  * THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-/** \file     CommonDef.h
-    \brief    Defines constants, macros and tool parameters
-*/
-
 #ifndef X265_COMMONDEF_H
 #define X265_COMMONDEF_H
 
-#include <cstdlib>
-#include "common.h"
 #include "TypeDef.h"
 
-//! \ingroup TLibCommon
-//! \{
-
 #ifndef NULL
 #define NULL 0
 #endif
 
-// ====================================================================================================================
-// Common constants
-// ====================================================================================================================
-
-#define MAX_NUM_PICS_IN_SOP         1024
-
-#define MAX_NESTING_NUM_OPS         1024
-#define MAX_NESTING_NUM_LAYER       64
-
-#define MAX_VPS_NUM_HRD_PARAMETERS  1
-#define MAX_VPS_OP_SETS_PLUS1       1024
-#define MAX_VPS_NUH_RESERVED_ZERO_LAYER_ID_PLUS1  1
-
-#define MAX_CPB_CNT                 32 ///< Upper bound of (cpb_cnt_minus1 + 1)
-#define MAX_NUM_LAYER_IDS           64
+#define MAX_UINT                    0xFFFFFFFFU // max. value of unsigned 32-bit integer
+#define MAX_INT                     2147483647  // max. value of signed 32-bit integer
+#define MAX_INT64                   0x7FFFFFFFFFFFFFFFLL  // max. value of signed 64-bit integer
+#define MAX_DOUBLE                  1.7e+308    // max. value of double-type value
 
-#define COEF_REMAIN_BIN_REDUCTION   3 ///< indicates the level at which the VLC
+#define COEF_REMAIN_BIN_REDUCTION   3 // indicates the level at which the VLC
+                                      // transitions from Golomb-Rice to TU+EG(k)
 
-///< transitions from Golomb-Rice to TU+EG(k)
+#define SBH_THRESHOLD               4 // fixed sign bit hiding controlling threshold
 
-#define CU_DQP_TU_CMAX              5 ///< max number bins for truncated unary
-#define CU_DQP_EG_k                 0 ///< exp-golomb order
-
-#define SBH_THRESHOLD               4 ///< I0156: value of the fixed SBH controlling threshold
-
-#define C1FLAG_NUMBER               8 // maximum number of largerThan1 flag coded in one chunk :  16 in HM5
+#define C1FLAG_NUMBER               8 // maximum number of largerThan1 flag coded in one chunk:  16 in HM5
 #define C2FLAG_NUMBER               1 // maximum number of largerThan2 flag coded in one chunk:  16 in HM5
 
 #define SAO_ENCODING_RATE           0.75
 #define SAO_ENCODING_RATE_CHROMA    0.5
 
-#define MAX_NUM_VPS                 16
-#define MAX_NUM_SPS                 16
-#define MAX_NUM_PPS                 64
-
-#define MIN_SCAN_POS_CROSS          4
-
-#define MLS_GRP_NUM                 64 ///< G644 : Max number of coefficient groups, max(16, 64)
-#define MLS_CG_SIZE                 4 ///< G644 : Coefficient group size of 4x4
+#define MLS_GRP_NUM                 64 // Max number of coefficient groups, max(16, 64)
+#define MLS_CG_SIZE                 4  // Coefficient group size of 4x4
 #define MLS_CG_LOG2_SIZE            2
 
-#define ARL_C_PRECISION             7 ///< G382: 7-bit arithmetic precision
-#define LEVEL_RANGE                 30 ///< G382: max coefficient level in statistics collection
+#define QUANT_IQUANT_SHIFT          20 // Q(QP%6) * IQ(QP%6) = 2^20
+#define QUANT_SHIFT                 14 // Q(4) = 2^14
+#define SCALE_BITS                  15 // Inherited from TMuC, presumably for fractional bit estimates in RDOQ
+#define MAX_TR_DYNAMIC_RANGE        15 // Maximum transform dynamic range (excluding sign bit)
+
+#define SHIFT_INV_1ST               7  // Shift after first inverse transform stage
+#define SHIFT_INV_2ND               12 // Shift after second inverse transform stage
 
 #define AMVP_DECIMATION_FACTOR      4
 
 #define SCAN_SET_SIZE               16
 #define LOG2_SCAN_SET_SIZE          4
 
-#define FAST_UDI_MAX_RDMODE_NUM     35 ///< maximum number of RD comparison in fast-UDI estimation loop
-
+#define ALL_IDX                     -1
 #define PLANAR_IDX                  0
 #define VER_IDX                     26 // index for intra VERTICAL   mode
 #define HOR_IDX                     10 // index for intra HORIZONTAL mode
-#define DC_IDX                      1 // index for intra DC mode
-#define NUM_CHROMA_MODE             5 // total number of chroma modes
+#define DC_IDX                      1  // index for intra DC mode
+#define NUM_CHROMA_MODE             5  // total number of chroma modes
 #define DM_CHROMA_IDX               36 // chroma mode index for derived from luma intra mode
 
-#define FULL_NBIT 1 ///< When enabled, compute costs using full sample bitdepth.  When disabled, compute costs as if it is 8-bit source video.
-#if FULL_NBIT || !HIGH_BIT_DEPTH
-# define DISTORTION_PRECISION_ADJUSTMENT(x) 0
-#else
-# define DISTORTION_PRECISION_ADJUSTMENT(x) (x)
-#endif
-
-#define LOG2_MAX_NUM_COLUMNS_MINUS1 7
-#define LOG2_MAX_NUM_ROWS_MINUS1    7
-#define LOG2_MAX_COLUMN_WIDTH       13
-#define LOG2_MAX_ROW_HEIGHT         13
-
-#define REG_DCT                     65535
-
-#define CABAC_INIT_PRESENT_FLAG     1
-
-#define MAX_GOP                     64          ///< max. value of hierarchical GOP size
-
-#define MAX_NUM_REF_PICS            16          ///< max. number of pictures used for reference
-#define MAX_NUM_REF                 16          ///< max. number of entries in picture reference list
-
-#define MAX_UINT                    0xFFFFFFFFU ///< max. value of unsigned 32-bit integer
-#define MAX_INT                     2147483647  ///< max. value of signed 32-bit integer
-#define MAX_INT64                   0x7FFFFFFFFFFFFFFFLL  ///< max. value of signed 64-bit integer
-#define MAX_DOUBLE                  1.7e+308    ///< max. value of double-type value
-
-#define MIN_QP                      0
-#define MAX_QP                      51
-#define MAX_MAX_QP                  69
-
-#define NOT_VALID                  -1
-
-// ====================================================================================================================
-// Coding tool configuration
-// ====================================================================================================================
+#define MDCS_ANGLE_LIMIT            4 // distance from true angle that horiz or vertical scan is allowed
+#define MDCS_LOG2_MAX_SIZE          3 // TUs with log2 of size greater than this can only use diagonal scan
 
-// AMVP: advanced motion vector prediction
-#define AMVP_MAX_NUM_CANDS          2 ///< max number of final candidates
-#define AMVP_MAX_NUM_CANDS_MEM      3 ///< max number of candidates
-#define MRG_MAX_NUM_CANDS           5
+#define MAX_NUM_REF_PICS            16 // max. number of pictures used for reference
+#define MAX_NUM_REF                 16 // max. number of entries in picture reference list
 
-// Explicit temporal layer QP offset
-#define MAX_TLAYER                  8 ///< max number of temporal layer
+#define REF_PIC_LIST_0              0
+#define REF_PIC_LIST_1              1
+#define REF_PIC_LIST_X              100
+#define NOT_VALID                   -1
 
-// Adaptive search range depending on POC difference
-#define ADAPT_SR_SCALE              1 ///< division factor for adaptive search range
+#define AMVP_NUM_CANDS              2 // number of AMVP candidates
+#define MRG_MAX_NUM_CANDS           5 // max number of final merge candidates
 
-#define MAX_CHROMA_FORMAT_IDC       3
+#define MAX_CHROMA_FORMAT_IDC       3 //  TODO: Remove me
 
-//! \}
+#define CHROMA_H_SHIFT(x) (x == X265_CSP_I420 || x == X265_CSP_I422)
+#define CHROMA_V_SHIFT(x) (x == X265_CSP_I420)
 
 #endif // ifndef X265_COMMONDEF_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/ContextTables.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/ContextTables.h Changed

@@ -65,7 +65,7 @@
 
 #define NUM_REF_NO_CTX                2       ///< number of context models for reference index
 #define NUM_TRANS_SUBDIV_FLAG_CTX     3       ///< number of context models for transform subdivision flags
-#define NUM_QT_CBF_CTX                6       ///< number of context models for QT CBF
+#define NUM_QT_CBF_CTX                7       ///< number of context models for QT CBF
 #define NUM_QT_ROOT_CBF_CTX           1       ///< number of context models for QT ROOT CBF
 #define NUM_DELTA_QP_CTX              3       ///< number of context models for dQP
 
@@ -127,27 +127,16 @@
 namespace x265 {
 // private namespace
 
-// ====================================================================================================================
-// Sbac interface
-// ====================================================================================================================
-typedef struct ContextModel
-{
-    uint8_t m_state;  ///< internal state variable
-    uint8_t bBinsCoded;
-} ContextModel;
-
 extern const uint32_t g_entropyBits[128];
 extern const uint8_t g_nextState[128][2];
-uint8_t sbacInit(int qp, int initValue);   ///< initialize state with initial probability
 
 #define sbacGetMps(S)               ((S) & 1)
 #define sbacGetState(S)             ((S) >> 1)
 #define sbacNext(S, V)              (g_nextState[(S)][(V)])
 #define sbacGetEntropyBits(S, V)    (g_entropyBits[(S) ^ (V)])
 #define sbacGetEntropyBitsTrm(V)    (g_entropyBits[126 ^ (V)])
-#define  CHANNEL_TYPE_LUMA         0
-#define  CHANNEL_TYPE_CHROMA       1
-#define  MAX_NUM_CHANNEL_TYPE      2
+
+#define MAX_NUM_CHANNEL_TYPE        2
 
 // ====================================================================================================================
 // Tables
@@ -155,6 +144,7 @@
 #define NEIGHBOURHOOD_00_CONTEXT_1_THRESHOLD_4x4  3
 #define NEIGHBOURHOOD_00_CONTEXT_2_THRESHOLD_4x4  1
 
+static const uint32_t ctxCbf[3][5] = { { 1, 0, 0, 0, 0 }, { 2, 3, 4, 5, 6 }, { 2, 3, 4, 5, 6 } };
 static const uint32_t significanceMapContextSetStart[MAX_NUM_CHANNEL_TYPE][3] = { { 0,  9, 21 }, { 0,  9, 12 } };
 static const uint32_t significanceMapContextSetSize[MAX_NUM_CHANNEL_TYPE][3]  = { { 9, 12,  6 }, { 9,  3,  3 } };
 static const uint32_t nonDiagonalScan8x8ContextOffset[MAX_NUM_CHANNEL_TYPE]   = {  6, 0  };
@@ -269,9 +259,9 @@
 static const uint8_t
     INIT_QT_CBF[3][NUM_QT_CBF_CTX] =
 {
-    { 153,  111,  149,   92,  167,  154, },
-    { 153,  111,  149,  107,  167,  154, },
-    { 111,  141,   94,  138,  182,  154, },
+    { 153,  111,  149,   92,  167,  154,  154 },
+    { 153,  111,  149,  107,  167,  154,  154 },
+    { 111,  141,   94,  138,  182,  154,  154 },
 };
 
 static const uint8_t

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComDataCU.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComDataCU.cpp Changed

@@ -36,8 +36,9 @@
     \todo     not all entities are documented
 */
 
+#include "common.h"
 #include "TComDataCU.h"
-#include "TComPic.h"
+#include "frame.h"
 #include "mv.h"
 
 using namespace x265;
@@ -60,104 +61,64 @@
 TComDataCU::TComDataCU()
 {
     /* TODO: can we use memset here? */
-    m_pic = NULL;
+    m_pic   = NULL;
     m_slice = NULL;
-    m_depth = NULL;
-    m_skipFlag = NULL;
-    m_partSizes = NULL;
-    m_predModes = NULL;
-    m_cuTransquantBypass = NULL;
-    m_cuSize = NULL;
-    m_qp = NULL;
-    m_bMergeFlags = NULL;
-    m_lumaIntraDir = NULL;
-    m_chromaIntraDir = NULL;
-    m_interDir = NULL;
-    m_trIdx = NULL;
-    m_transformSkip[0] = NULL;
-    m_transformSkip[1] = NULL;
-    m_transformSkip[2] = NULL;
-    m_cbf[0] = NULL;
-    m_cbf[1] = NULL;
-    m_cbf[2] = NULL;
-    m_trCoeffY = NULL;
-    m_trCoeffCb = NULL;
-    m_trCoeffCr = NULL;
-    m_iPCMFlags = NULL;
-    m_iPCMSampleY = NULL;
-    m_iPCMSampleCb = NULL;
-    m_iPCMSampleCr = NULL;
-    m_cuAboveLeft = NULL;
+    m_cuAboveLeft  = NULL;
     m_cuAboveRight = NULL;
     m_cuAbove = NULL;
-    m_cuLeft = NULL;
-    m_cuColocated[0] = NULL;
-    m_cuColocated[1] = NULL;
-    m_mvpIdx[0] = NULL;
-    m_mvpIdx[1] = NULL;
+    m_cuLeft  = NULL;
     m_chromaFormat = 0;
-    m_baseQp = 0;
+    m_baseQp       = 0;
+    m_DataCUMemPool.qpMemBlock             = NULL;
+    m_DataCUMemPool.depthMemBlock          = NULL;
+    m_DataCUMemPool.log2CUSizeMemBlock     = NULL;
+    m_DataCUMemPool.skipFlagMemBlock       = NULL;
+    m_DataCUMemPool.partSizeMemBlock       = NULL;
+    m_DataCUMemPool.predModeMemBlock       = NULL;
+    m_DataCUMemPool.cuTQBypassMemBlock     = NULL;
+    m_DataCUMemPool.mergeFlagMemBlock      = NULL;
+    m_DataCUMemPool.lumaIntraDirMemBlock   = NULL;
+    m_DataCUMemPool.chromaIntraDirMemBlock = NULL;
+    m_DataCUMemPool.interDirMemBlock       = NULL;
+    m_DataCUMemPool.trIdxMemBlock          = NULL;
+    m_DataCUMemPool.transformSkipMemBlock  = NULL;
+    m_DataCUMemPool.cbfMemBlock            = NULL;
+    m_DataCUMemPool.mvpIdxMemBlock         = NULL;
+    m_DataCUMemPool.trCoeffMemBlock        = NULL;
+    m_DataCUMemPool.m_tqBypassYuvMemBlock  = NULL;
 }
 
-TComDataCU::~TComDataCU()
-{}
 
-bool TComDataCU::create(uint32_t numPartition, uint32_t width, uint32_t height, int unitSize, int csp)
+bool TComDataCU::initialize(uint32_t numPartition, uint32_t sizeL, uint32_t sizeC, uint32_t numBlocks, bool isLossless)
 {
-    m_hChromaShift = CHROMA_H_SHIFT(csp);
-    m_vChromaShift = CHROMA_V_SHIFT(csp);
-    m_chromaFormat = csp;
+    bool ok = true;
 
-    m_pic           = NULL;
-    m_slice         = NULL;
-    m_numPartitions = numPartition;
-    m_unitSize = unitSize;
+    ok &= m_cuMvFieldMemPool.initialize(numPartition, numBlocks);
 
-    uint32_t tmp = 4 * AMVP_DECIMATION_FACTOR / unitSize;
-    tmp = tmp * tmp;
-    assert(tmp == (1 << (g_convertToBit[tmp] + 2)));
-    tmp = g_convertToBit[tmp] + 2;
-    m_unitMask = ~((1 << tmp) - 1);
+    CHECKED_MALLOC(m_DataCUMemPool.qpMemBlock, char,  numPartition * numBlocks);
 
-    bool ok = true;
-    ok &= m_cuMvField[0].create(numPartition);
-    ok &= m_cuMvField[1].create(numPartition);
+    CHECKED_MALLOC(m_DataCUMemPool.depthMemBlock, uint8_t, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.log2CUSizeMemBlock, uint8_t, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.skipFlagMemBlock, bool, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.partSizeMemBlock, char, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.predModeMemBlock, char, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.cuTQBypassMemBlock, bool, numPartition * numBlocks);
+
+    CHECKED_MALLOC(m_DataCUMemPool.mergeFlagMemBlock, bool,  numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.lumaIntraDirMemBlock, uint8_t, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.chromaIntraDirMemBlock, uint8_t, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.interDirMemBlock, uint8_t, numPartition * numBlocks);
+
+    CHECKED_MALLOC(m_DataCUMemPool.trIdxMemBlock, uint8_t, numPartition * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.transformSkipMemBlock, uint8_t, numPartition * 3 * numBlocks);
+
+    CHECKED_MALLOC(m_DataCUMemPool.cbfMemBlock, uint8_t, numPartition * 3 * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.mvpIdxMemBlock, uint8_t, numPartition * 2 * numBlocks);
+    CHECKED_MALLOC(m_DataCUMemPool.trCoeffMemBlock, coeff_t, (sizeL + sizeC * 2) * numBlocks);
 
-    CHECKED_MALLOC(m_qp, char,  numPartition);
-    CHECKED_MALLOC(m_depth, UChar, numPartition);
-    CHECKED_MALLOC(m_cuSize, uint8_t, numPartition);
-    CHECKED_MALLOC(m_skipFlag, bool, numPartition);
-    CHECKED_MALLOC(m_partSizes, char, numPartition);
-    CHECKED_MALLOC(m_predModes, char, numPartition);
-    CHECKED_MALLOC(m_cuTransquantBypass, bool, numPartition);
-
-    CHECKED_MALLOC(m_bMergeFlags, bool,  numPartition);
-    CHECKED_MALLOC(m_lumaIntraDir, UChar, numPartition);
-    CHECKED_MALLOC(m_chromaIntraDir, UChar, numPartition);
-    CHECKED_MALLOC(m_interDir, UChar, numPartition);
-
-    CHECKED_MALLOC(m_trIdx, UChar, numPartition);
-    CHECKED_MALLOC(m_transformSkip[0], UChar, numPartition);
-    CHECKED_MALLOC(m_transformSkip[1], UChar, numPartition);
-    CHECKED_MALLOC(m_transformSkip[2], UChar, numPartition);
-
-    CHECKED_MALLOC(m_cbf[0], UChar, numPartition);
-    CHECKED_MALLOC(m_cbf[1], UChar, numPartition);
-    CHECKED_MALLOC(m_cbf[2], UChar, numPartition);
-
-    CHECKED_MALLOC(m_mvpIdx[0], uint8_t, numPartition * 2);
-    m_mvpIdx[1] = m_mvpIdx[0] + numPartition;
-
-    CHECKED_MALLOC(m_trCoeffY, TCoeff, width * height);
-    CHECKED_MALLOC(m_trCoeffCb, TCoeff, (width >> m_hChromaShift) * (height >> m_vChromaShift));
-    CHECKED_MALLOC(m_trCoeffCr, TCoeff, (width >> m_hChromaShift) * (height >> m_vChromaShift));
-
-    CHECKED_MALLOC(m_iPCMFlags, bool, numPartition);
-    CHECKED_MALLOC(m_iPCMSampleY, Pel, width * height);
-    CHECKED_MALLOC(m_iPCMSampleCb, Pel, (width >> m_hChromaShift) * (height >> m_vChromaShift));
-    CHECKED_MALLOC(m_iPCMSampleCr, Pel, (width >> m_hChromaShift) * (height >> m_vChromaShift));
+    if (isLossless)
+        CHECKED_MALLOC(m_DataCUMemPool.m_tqBypassYuvMemBlock, pixel, (sizeL + sizeC * 2) * numBlocks);
 
-    memset(m_partSizes, SIZE_NONE, numPartition * sizeof(*m_partSizes));
     return ok;
 
 fail:
@@ -165,323 +126,290 @@
     return ok;
 }
 
-void TComDataCU::destroy()
+void TComDataCU::create(TComDataCU *cu, uint32_t numPartition, uint32_t cuSize, int csp, int index, bool isLossless)
 {
-    X265_FREE(m_qp);
-    X265_FREE(m_depth);
-    X265_FREE(m_cuSize);
-    X265_FREE(m_cbf[0]);
-    X265_FREE(m_cbf[1]);
-    X265_FREE(m_cbf[2]);
-    X265_FREE(m_interDir);
-    X265_FREE(m_bMergeFlags);
-    X265_FREE(m_lumaIntraDir);
-    X265_FREE(m_chromaIntraDir);
-    X265_FREE(m_trIdx);
-    X265_FREE(m_transformSkip[0]);
-    X265_FREE(m_transformSkip[1]);
-    X265_FREE(m_transformSkip[2]);
-    X265_FREE(m_trCoeffY);
-    X265_FREE(m_trCoeffCb);
-    X265_FREE(m_trCoeffCr);
-    X265_FREE(m_iPCMFlags);
-    X265_FREE(m_iPCMSampleY);
-    X265_FREE(m_iPCMSampleCb);
-    X265_FREE(m_iPCMSampleCr);
-    X265_FREE(m_mvpIdx[0]);
-    X265_FREE(m_cuTransquantBypass);
-    X265_FREE(m_skipFlag);
-    X265_FREE(m_partSizes);
-    X265_FREE(m_predModes);
+    m_hChromaShift = CHROMA_H_SHIFT(csp);
+    m_vChromaShift = CHROMA_V_SHIFT(csp);
+    m_chromaFormat = csp;
 
-    m_cuMvField[0].destroy();
-    m_cuMvField[1].destroy();
-}
+    m_pic           = NULL;
+    m_slice         = NULL;
+    m_numPartitions = numPartition;
 
-// ====================================================================================================================
-// Public member functions
-// ====================================================================================================================
+    uint32_t sizeL = cuSize * cuSize;
+    uint32_t sizeC = sizeL >> (m_hChromaShift + m_vChromaShift);
 
-// --------------------------------------------------------------------------------------------------------------------
-// Initialization
-// --------------------------------------------------------------------------------------------------------------------
+    m_cuMvField[0].create(&cu->m_cuMvFieldMemPool, numPartition, index, 0);
+    m_cuMvField[1].create(&cu->m_cuMvFieldMemPool, numPartition, index, 1);
 
-/**
- - initialize top-level CU
- - internal buffers are already created
- - set values before encoding a CU
- .
- \param  pic     picture (TComPic) class pointer
- \param  iCUAddr   CU address
- */
-void TComDataCU::initCU(TComPic* pic, uint32_t cuAddr)
+    m_qp                 = cu->m_DataCUMemPool.qpMemBlock             + index * numPartition;
+    m_depth              = cu->m_DataCUMemPool.depthMemBlock          + index * numPartition;
+    m_log2CUSize         = cu->m_DataCUMemPool.log2CUSizeMemBlock     + index * numPartition;
+    m_skipFlag           = cu->m_DataCUMemPool.skipFlagMemBlock       + index * numPartition;
+    m_partSizes          = cu->m_DataCUMemPool.partSizeMemBlock       + index * numPartition;
+    m_predModes          = cu->m_DataCUMemPool.predModeMemBlock       + index * numPartition;
+    m_cuTransquantBypass = cu->m_DataCUMemPool.cuTQBypassMemBlock     + index * numPartition;
+
+    m_bMergeFlags        = cu->m_DataCUMemPool.mergeFlagMemBlock      + index * numPartition;
+    m_lumaIntraDir       = cu->m_DataCUMemPool.lumaIntraDirMemBlock   + index * numPartition;
+    m_chromaIntraDir     = cu->m_DataCUMemPool.chromaIntraDirMemBlock + index * numPartition;
+    m_interDir           = cu->m_DataCUMemPool.interDirMemBlock       + index * numPartition;
+
+    m_trIdx              = cu->m_DataCUMemPool.trIdxMemBlock          + index * numPartition;
+    m_transformSkip[0]   = cu->m_DataCUMemPool.transformSkipMemBlock  + index * numPartition * 3;
+    m_transformSkip[1]   = m_transformSkip[0]                         + numPartition;
+    m_transformSkip[2]   = m_transformSkip[0]                         + numPartition * 2;
+
+    m_cbf[0]             = cu->m_DataCUMemPool.cbfMemBlock            + index * numPartition * 3;
+    m_cbf[1]             = m_cbf[0]                                   + numPartition;
+    m_cbf[2]             = m_cbf[0]                                   + numPartition * 2;
+
+    m_mvpIdx[0]          = cu->m_DataCUMemPool.mvpIdxMemBlock         + index * numPartition * 2;
+    m_mvpIdx[1]          = m_mvpIdx[0]                                + numPartition;
+
+    m_trCoeff[0]         = cu->m_DataCUMemPool.trCoeffMemBlock        + index * (sizeL + sizeC * 2);
+    m_trCoeff[1]         = m_trCoeff[0]                               + sizeL;
+    m_trCoeff[2]         = m_trCoeff[0]                               + sizeL + sizeC;
+
+    if (isLossless)
+    {
+        m_tqBypassOrigYuv[0] = cu->m_DataCUMemPool.m_tqBypassYuvMemBlock  + index * (sizeL + sizeC * 2);
+        m_tqBypassOrigYuv[1] = m_tqBypassOrigYuv[0]                       + sizeL;
+        m_tqBypassOrigYuv[2] = m_tqBypassOrigYuv[0]                       + sizeL + sizeC;
+    }
+
+    memset(m_partSizes, SIZE_NONE, numPartition * sizeof(*m_partSizes));
+}
+
+void TComDataCU::destroy()
 {
-    m_pic              = pic;
-    m_slice            = pic->getSlice();
-    m_cuAddr           = cuAddr;
-    m_cuPelX           = (cuAddr % pic->getFrameWidthInCU()) * g_maxCUSize;
-    m_cuPelY           = (cuAddr / pic->getFrameWidthInCU()) * g_maxCUSize;
-    m_absIdxInLCU      = 0;
-    m_totalCost        = MAX_INT64;
-    m_sa8dCost         = MAX_INT64;
-    m_totalDistortion  = 0;
-    m_totalBits        = 0;
-    m_numPartitions    = pic->getNumPartInCU();
-    char* qp           = pic->getCU(getAddr())->getQP();
-    m_baseQp           = pic->getCU(getAddr())->m_baseQp;
-    for (int i = 0; i < 4; i++)
+    if (m_DataCUMemPool.qpMemBlock)
     {
-        m_avgCost[i] = 0;
-        m_count[i] = 0;
+        X265_FREE(m_DataCUMemPool.qpMemBlock);
+        m_DataCUMemPool.qpMemBlock = NULL;
     }
 
-    // CHECK_ME: why partStartIdx always negative
-    int numElements = m_numPartitions;
-    assert(numElements > 0);
-
-    {
-        memset(m_skipFlag,           false,         numElements * sizeof(*m_skipFlag));
-        memset(m_predModes,          MODE_NONE,     numElements * sizeof(*m_predModes));
-        memset(m_cuTransquantBypass, false,         numElements * sizeof(*m_cuTransquantBypass));
-        memset(m_depth,              0,             numElements * sizeof(*m_depth));
-        memset(m_trIdx,              0,             numElements * sizeof(*m_trIdx));
-        memset(m_transformSkip[0],   0,             numElements * sizeof(*m_transformSkip[0]));
-        memset(m_transformSkip[1],   0,             numElements * sizeof(*m_transformSkip[1]));
-        memset(m_transformSkip[2],   0,             numElements * sizeof(*m_transformSkip[2]));
-        memset(m_cuSize,             g_maxCUSize,   numElements * sizeof(*m_cuSize));
-        memcpy(m_qp,                 qp,            numElements * sizeof(*m_qp));
-        memset(m_bMergeFlags,        false,         numElements * sizeof(*m_bMergeFlags));
-        memset(m_lumaIntraDir,       DC_IDX,        numElements * sizeof(*m_lumaIntraDir));
-        memset(m_chromaIntraDir,     0,             numElements * sizeof(*m_chromaIntraDir));
-        memset(m_interDir,           0,             numElements * sizeof(*m_interDir));
-        memset(m_cbf[0],             0,             numElements * sizeof(*m_cbf[0]));
-        memset(m_cbf[1],             0,             numElements * sizeof(*m_cbf[1]));
-        memset(m_cbf[2],             0,             numElements * sizeof(*m_cbf[2]));
-        memset(m_iPCMFlags,          false,         numElements * sizeof(*m_iPCMFlags));
+    if (m_DataCUMemPool.depthMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.depthMemBlock);
+        m_DataCUMemPool.depthMemBlock = NULL;
     }
 
-    uint32_t y_tmp = g_maxCUSize * g_maxCUSize;
-    uint32_t c_tmp = (g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift);
+    if (m_DataCUMemPool.log2CUSizeMemBlock)
     {
-        m_cuMvField[0].clearMvField();
-        m_cuMvField[1].clearMvField();
-
-        // TODO: can be remove, but I haven't data to verify it, remove later
-        if (getSlice()->getSPS()->getUsePCM())
-        {
-            memset(m_iPCMSampleY, 0, sizeof(Pel) * y_tmp);
-            memset(m_iPCMSampleCb, 0, sizeof(Pel) * c_tmp);
-            memset(m_iPCMSampleCr, 0, sizeof(Pel) * c_tmp);
-        }
+        X265_FREE(m_DataCUMemPool.log2CUSizeMemBlock);
+        m_DataCUMemPool.log2CUSizeMemBlock = NULL;
     }
 
-    // Setting neighbor CU
-    m_cuLeft        = NULL;
-    m_cuAbove       = NULL;
-    m_cuAboveLeft   = NULL;
-    m_cuAboveRight  = NULL;
-
-    m_cuColocated[0] = NULL;
-    m_cuColocated[1] = NULL;
+    if (m_DataCUMemPool.cbfMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.cbfMemBlock);
+        m_DataCUMemPool.cbfMemBlock = NULL;
+    }
 
-    uint32_t uiWidthInCU = pic->getFrameWidthInCU();
-    if (m_cuAddr % uiWidthInCU)
+    if (m_DataCUMemPool.interDirMemBlock)
     {
-        m_cuLeft = pic->getCU(m_cuAddr - 1);
+        X265_FREE(m_DataCUMemPool.interDirMemBlock);
+        m_DataCUMemPool.interDirMemBlock = NULL;
     }
 
-    if (m_cuAddr / uiWidthInCU)
+    if (m_DataCUMemPool.mergeFlagMemBlock)
     {
-        m_cuAbove = pic->getCU(m_cuAddr - uiWidthInCU);
+        X265_FREE(m_DataCUMemPool.mergeFlagMemBlock);
+        m_DataCUMemPool.mergeFlagMemBlock = NULL;
     }
 
-    if (m_cuLeft && m_cuAbove)
+    if (m_DataCUMemPool.lumaIntraDirMemBlock)
     {
-        m_cuAboveLeft = pic->getCU(m_cuAddr - uiWidthInCU - 1);
+        X265_FREE(m_DataCUMemPool.lumaIntraDirMemBlock);
+        m_DataCUMemPool.lumaIntraDirMemBlock = NULL;
     }
 
-    if (m_cuAbove && ((m_cuAddr % uiWidthInCU) < (uiWidthInCU - 1)))
+    if(m_DataCUMemPool.chromaIntraDirMemBlock)
     {
-        m_cuAboveRight = pic->getCU(m_cuAddr - uiWidthInCU + 1);
+        X265_FREE(m_DataCUMemPool.chromaIntraDirMemBlock);
+        m_DataCUMemPool.chromaIntraDirMemBlock = NULL;
     }
 
-    if (getSlice()->getNumRefIdx(REF_PIC_LIST_0) > 0)
+    if (m_DataCUMemPool.trIdxMemBlock)
     {
-        m_cuColocated[0] = getSlice()->getRefPic(REF_PIC_LIST_0, 0)->getCU(m_cuAddr);
+        X265_FREE(m_DataCUMemPool.trIdxMemBlock);
+        m_DataCUMemPool.trIdxMemBlock = NULL;
     }
 
-    if (getSlice()->getNumRefIdx(REF_PIC_LIST_1) > 0)
+    if (m_DataCUMemPool.transformSkipMemBlock)
     {
-        m_cuColocated[1] = getSlice()->getRefPic(REF_PIC_LIST_1, 0)->getCU(m_cuAddr);
+        X265_FREE(m_DataCUMemPool.transformSkipMemBlock);
+        m_DataCUMemPool.transformSkipMemBlock = NULL;
     }
-}
 
-/** initialize prediction data with enabling sub-LCU-level delta QP
-*\param  depth  depth of the current CU
-*\param  qp     qp for the current CU
-*- set CU width and CU height according to depth
-*- set qp value according to input qp
-*- set last-coded qp value according to input last-coded qp
-*/
-void TComDataCU::initEstData(uint32_t depth, int qp)
-{
-    m_totalCost        = MAX_INT64;
-    m_sa8dCost         = MAX_INT64;
-    m_totalDistortion  = 0;
-    m_totalBits        = 0;
+    if (m_DataCUMemPool.trCoeffMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.trCoeffMemBlock);
+        m_DataCUMemPool.trCoeffMemBlock = NULL;
+    }
 
-    UChar cuSize = g_maxCUSize >> depth;
+    if (m_DataCUMemPool.mvpIdxMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.mvpIdxMemBlock);
+        m_DataCUMemPool.mvpIdxMemBlock = NULL;
+    }
 
-    for (uint32_t i = 0; i < m_numPartitions; i++)
+    if (m_DataCUMemPool.cuTQBypassMemBlock)
     {
-        m_depth[i] = depth;
-        m_cuSize[i] = cuSize;
-        m_trIdx[i] = 0;
-        m_transformSkip[0][i] = 0;
-        m_transformSkip[1][i] = 0;
-        m_transformSkip[2][i] = 0;
-        m_skipFlag[i]   = false;
-        m_partSizes[i] = SIZE_NONE;
-        m_predModes[i] = MODE_NONE;
-        m_cuTransquantBypass[i] = false;
-        m_iPCMFlags[i] = 0;
-        m_qp[i] = qp;
-        m_bMergeFlags[i] = 0;
-        m_lumaIntraDir[i] = DC_IDX;
-        m_chromaIntraDir[i] = 0;
-        m_interDir[i] = 0;
-        m_cbf[0][i] = 0;
-        m_cbf[1][i] = 0;
-        m_cbf[2][i] = 0;
+        X265_FREE(m_DataCUMemPool.cuTQBypassMemBlock);
+        m_DataCUMemPool.cuTQBypassMemBlock = NULL;
     }
 
-    m_cuMvField[0].clearMvField();
-    m_cuMvField[1].clearMvField();
-}
+    if (m_DataCUMemPool.skipFlagMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.skipFlagMemBlock);
+        m_DataCUMemPool.skipFlagMemBlock = NULL;
+    }
 
-void TComDataCU::initEstData(uint32_t depth)
-{
-    m_totalCost        = MAX_INT64;
-    m_sa8dCost         = MAX_INT64;
-    m_totalDistortion  = 0;
-    m_totalBits        = 0;
+    if (m_DataCUMemPool.partSizeMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.partSizeMemBlock);
+        m_DataCUMemPool.partSizeMemBlock = NULL;
+    }
 
-    UChar cuSize = g_maxCUSize >> depth;
+    if (m_DataCUMemPool.predModeMemBlock)
+    {
+        X265_FREE(m_DataCUMemPool.predModeMemBlock);
+        m_DataCUMemPool.predModeMemBlock = NULL;
+    }
 
-    for (uint32_t i = 0; i < m_numPartitions; i++)
+    if (m_DataCUMemPool.m_tqBypassYuvMemBlock)
     {
-        m_depth[i] = depth;
-        m_cuSize[i] = cuSize;
-        m_trIdx[i] = 0;
-        m_transformSkip[0][i] = 0;
-        m_transformSkip[1][i] = 0;
-        m_transformSkip[2][i] = 0;
-        m_skipFlag[i]   = false;
-        m_partSizes[i] = SIZE_NONE;
-        m_predModes[i] = MODE_NONE;
-        m_cuTransquantBypass[i] = false;
-        m_iPCMFlags[i] = 0;
-        m_bMergeFlags[i] = 0;
-        m_lumaIntraDir[i] = DC_IDX;
-        m_chromaIntraDir[i] = 0;
-        m_interDir[i] = 0;
-        m_cbf[0][i] = 0;
-        m_cbf[1][i] = 0;
-        m_cbf[2][i] = 0;
+        X265_FREE(m_DataCUMemPool.m_tqBypassYuvMemBlock);
+        m_DataCUMemPool.m_tqBypassYuvMemBlock = NULL;
     }
 
-    m_cuMvField[0].clearMvField();
-    m_cuMvField[1].clearMvField();
+    m_cuMvFieldMemPool.destroy();
 }
 
-// initialize Sub partition
-void TComDataCU::initSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth, int qp)
-{
-    assert(partUnitIdx < 4);
-
-    uint32_t partOffset = (cu->getTotalNumPart() >> 2) * partUnitIdx;
-
-    m_pic              = cu->getPic();
-    m_slice            = m_pic->getSlice();
-    m_cuAddr           = cu->getAddr();
-    m_absIdxInLCU      = cu->getZorderIdxInCU() + partOffset;
+// ====================================================================================================================
+// Public member functions
+// ====================================================================================================================
 
-    m_cuPelX           = cu->getCUPelX() + (g_maxCUSize >> depth) * (partUnitIdx &  1);
-    m_cuPelY           = cu->getCUPelY() + (g_maxCUSize >> depth) * (partUnitIdx >> 1);
+// --------------------------------------------------------------------------------------------------------------------
+// Initialization
+// --------------------------------------------------------------------------------------------------------------------
 
-    m_totalCost        = MAX_INT64;
+/**
+ - initialize top-level CU
+ - internal buffers are already created
+ - set values before encoding a CU
+ .
+ \param  pic     picture (TComPic) class pointer
+ \param  cuAddr  CU address
+ */
+void TComDataCU::initCU(Frame* pic, uint32_t cuAddr)
+{
+    m_pic              = pic;
+    m_slice            = pic->m_picSym->m_slice;
+    m_cuAddr           = cuAddr;
+    m_cuPelX           = (cuAddr % pic->getFrameWidthInCU()) << g_maxLog2CUSize;
+    m_cuPelY           = (cuAddr / pic->getFrameWidthInCU()) << g_maxLog2CUSize;
+    m_absIdxInLCU      = 0;
+    m_psyEnergy        = 0;
+    m_totalPsyCost     = MAX_INT64;
+    m_totalRDCost      = MAX_INT64;
     m_sa8dCost         = MAX_INT64;
     m_totalDistortion  = 0;
     m_totalBits        = 0;
-    m_numPartitions    = cu->getTotalNumPart() >> 2;
-
+    m_mvBits           = 0;
+    m_coeffBits        = 0;
+    m_numPartitions    = NUM_CU_PARTITIONS;
+    char* qp           = pic->getCU(getAddr())->getQP();
+    m_baseQp           = pic->getCU(getAddr())->m_baseQp;
     for (int i = 0; i < 4; i++)
     {
-        m_avgCost[i] = cu->m_avgCost[i];
-        m_count[i] = cu->m_count[i];
+        m_avgCost[i] = 0;
+        m_count[i] = 0;
     }
 
-    int iSizeInUchar = sizeof(UChar) * m_numPartitions;
-    int iSizeInBool  = sizeof(bool) * m_numPartitions;
+    X265_CHECK(m_numPartitions > 0, "unexpected partition count\n");
 
-    int sizeInChar = sizeof(char) * m_numPartitions;
-    memset(m_qp, qp, sizeInChar);
-
-    memset(m_bMergeFlags,     0, iSizeInBool);
-    memset(m_lumaIntraDir,    DC_IDX, iSizeInUchar);
-    memset(m_chromaIntraDir,  0, iSizeInUchar);
-    memset(m_interDir,        0, iSizeInUchar);
-    memset(m_trIdx,           0, iSizeInUchar);
-    memset(m_transformSkip[0], 0, iSizeInUchar);
-    memset(m_transformSkip[1], 0, iSizeInUchar);
-    memset(m_transformSkip[2], 0, iSizeInUchar);
-    memset(m_cbf[0],          0, iSizeInUchar);
-    memset(m_cbf[1],          0, iSizeInUchar);
-    memset(m_cbf[2],          0, iSizeInUchar);
-    memset(m_depth, depth, iSizeInUchar);
-
-    UChar cuSize = g_maxCUSize >> depth;
-    memset(m_cuSize,    cuSize,  iSizeInUchar);
-    memset(m_iPCMFlags, 0, iSizeInBool);
-    for (uint32_t i = 0; i < m_numPartitions; i++)
-    {
-        m_skipFlag[i]   = false;
-        m_partSizes[i] = SIZE_NONE;
-        m_predModes[i] = MODE_NONE;
-        m_cuTransquantBypass[i] = false;
-    }
+    memset(m_skipFlag,           false,         m_numPartitions * sizeof(*m_skipFlag));
+    memset(m_predModes,          MODE_NONE,     m_numPartitions * sizeof(*m_predModes));
+    memset(m_partSizes,          SIZE_NONE,     m_numPartitions * sizeof(*m_partSizes));
+    memset(m_cuTransquantBypass, false,         m_numPartitions * sizeof(*m_cuTransquantBypass));
+    memset(m_depth,              0,             m_numPartitions * sizeof(*m_depth));
+    memset(m_trIdx,              0,             m_numPartitions * sizeof(*m_trIdx));
+    memset(m_transformSkip[0],   0,             m_numPartitions * sizeof(*m_transformSkip[0]));
+    memset(m_transformSkip[1],   0,             m_numPartitions * sizeof(*m_transformSkip[1]));
+    memset(m_transformSkip[2],   0,             m_numPartitions * sizeof(*m_transformSkip[2]));
+    memset(m_log2CUSize,         g_maxLog2CUSize, m_numPartitions * sizeof(*m_log2CUSize));
+    memset(m_bMergeFlags,        false,         m_numPartitions * sizeof(*m_bMergeFlags));
+    memset(m_lumaIntraDir,       DC_IDX,        m_numPartitions * sizeof(*m_lumaIntraDir));
+    memset(m_chromaIntraDir,     0,             m_numPartitions * sizeof(*m_chromaIntraDir));
+    memset(m_interDir,           0,             m_numPartitions * sizeof(*m_interDir));
+    memset(m_cbf[0],             0,             m_numPartitions * sizeof(*m_cbf[0]));
+    memset(m_cbf[1],             0,             m_numPartitions * sizeof(*m_cbf[1]));
+    memset(m_cbf[2],             0,             m_numPartitions * sizeof(*m_cbf[2]));
+    if (qp != m_qp)
+        memcpy(m_qp,             qp,            m_numPartitions * sizeof(*m_qp));
 
     m_cuMvField[0].clearMvField();
     m_cuMvField[1].clearMvField();
 
-    m_cuLeft        = cu->getCULeft();
-    m_cuAbove       = cu->getCUAbove();
-    m_cuAboveLeft   = cu->getCUAboveLeft();
-    m_cuAboveRight  = cu->getCUAboveRight();
+    if (m_slice->m_pps->bTransquantBypassEnabled)
+    {
+        uint32_t y_tmp = 1 << (g_maxLog2CUSize * 2);
+        uint32_t c_tmp = 1 << (g_maxLog2CUSize * 2 - m_hChromaShift - m_vChromaShift);
+        memset(m_tqBypassOrigYuv[0], 0, sizeof(pixel) * y_tmp);
+        memset(m_tqBypassOrigYuv[1], 0, sizeof(pixel) * c_tmp);
+        memset(m_tqBypassOrigYuv[2], 0, sizeof(pixel) * c_tmp);
+    }
 
-    m_cuColocated[0] = cu->getCUColocated(REF_PIC_LIST_0);
-    m_cuColocated[1] = cu->getCUColocated(REF_PIC_LIST_1);
+    uint32_t widthInCU = pic->getFrameWidthInCU();
+    m_cuLeft = (m_cuAddr % widthInCU) ? pic->getCU(m_cuAddr - 1) : NULL;
+    m_cuAbove = (m_cuAddr / widthInCU) ? pic->getCU(m_cuAddr - widthInCU) : NULL;
+    m_cuAboveLeft = (m_cuLeft && m_cuAbove) ? pic->getCU(m_cuAddr - widthInCU - 1) : NULL;
+    m_cuAboveRight = (m_cuAbove && ((m_cuAddr % widthInCU) < (widthInCU - 1))) ? pic->getCU(m_cuAddr - widthInCU + 1) : NULL;
 }
 
-// initialize Sub partition
-void TComDataCU::initSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth)
+// initialize prediction data
+void TComDataCU::initEstData()
 {
-    assert(partUnitIdx < 4);
+    m_psyEnergy        = 0;
+    m_totalPsyCost     = MAX_INT64;
+    m_totalRDCost      = MAX_INT64;
+    m_sa8dCost         = MAX_INT64;
+    m_totalDistortion  = 0;
+    m_totalBits        = 0;
+    m_mvBits           = 0;
+    m_coeffBits        = 0;
 
+    m_cuMvField[0].clearMvField();
+    m_cuMvField[1].clearMvField();
+}
+
+// initialize Sub partition
+void TComDataCU::initSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth, int qp)
+{
+    X265_CHECK(partUnitIdx < 4, "part unit should be less than 4\n");
+    uint8_t log2CUSize = g_maxLog2CUSize - depth;
     uint32_t partOffset = (cu->getTotalNumPart() >> 2) * partUnitIdx;
 
-    m_pic              = cu->getPic();
-    m_slice            = m_pic->getSlice();
+    m_pic              = cu->m_pic;
+    m_slice            = cu->m_slice;
     m_cuAddr           = cu->getAddr();
     m_absIdxInLCU      = cu->getZorderIdxInCU() + partOffset;
 
-    m_cuPelX           = cu->getCUPelX() + (g_maxCUSize >> depth) * (partUnitIdx &  1);
-    m_cuPelY           = cu->getCUPelY() + (g_maxCUSize >> depth) * (partUnitIdx >> 1);
+    m_cuPelX           = cu->getCUPelX() + ((partUnitIdx &  1) << log2CUSize);
+    m_cuPelY           = cu->getCUPelY() + ((partUnitIdx >> 1) << log2CUSize);
 
-    m_totalCost        = MAX_INT64;
+    m_psyEnergy        = 0;
+    m_totalPsyCost     = MAX_INT64;
+    m_totalRDCost      = MAX_INT64;
     m_sa8dCost         = MAX_INT64;
     m_totalDistortion  = 0;
     m_totalBits        = 0;
+    m_mvBits           = 0;
+    m_coeffBits        = 0;
+
     m_numPartitions    = cu->getTotalNumPart() >> 2;
 
     for (int i = 0; i < 4; i++)
@@ -490,82 +418,77 @@
         m_count[i] = cu->m_count[i];
     }
 
-    int iSizeInUchar = sizeof(UChar) * m_numPartitions;
-    int iSizeInBool  = sizeof(bool) * m_numPartitions;
-
+    int sizeInBool = sizeof(bool) * m_numPartitions;
     int sizeInChar = sizeof(char) * m_numPartitions;
 
-    memcpy(m_qp, cu->getQP() + partOffset, sizeInChar);
-    memset(m_bMergeFlags,     0, iSizeInBool);
-    memset(m_lumaIntraDir,    DC_IDX, iSizeInUchar);
-    memset(m_chromaIntraDir,  0, iSizeInUchar);
-    memset(m_interDir,        0, iSizeInUchar);
-    memset(m_trIdx,           0, iSizeInUchar);
-    memset(m_transformSkip[0], 0, iSizeInUchar);
-    memset(m_transformSkip[1], 0, iSizeInUchar);
-    memset(m_transformSkip[2], 0, iSizeInUchar);
-    memset(m_cbf[0],          0, iSizeInUchar);
-    memset(m_cbf[1],          0, iSizeInUchar);
-    memset(m_cbf[2],          0, iSizeInUchar);
-    memset(m_depth, depth, iSizeInUchar);
-
-    UChar cuSize = g_maxCUSize >> depth;
-    memset(m_cuSize,    cuSize,  iSizeInUchar);
-    memset(m_iPCMFlags, 0, iSizeInBool);
-    for (uint32_t i = 0; i < m_numPartitions; i++)
-    {
-        m_skipFlag[i]   = false;
-        m_partSizes[i] = SIZE_NONE;
-        m_predModes[i] = MODE_NONE;
-        m_cuTransquantBypass[i] = false;
-    }
+    memset(m_qp,                 qp,     sizeInChar);
+    memset(m_lumaIntraDir,       DC_IDX, sizeInChar);
+    memset(m_chromaIntraDir,     0,      sizeInChar);
+    memset(m_trIdx,              0,      sizeInChar);
+    memset(m_transformSkip[0],   0,      sizeInChar);
+    memset(m_transformSkip[1],   0,      sizeInChar);
+    memset(m_transformSkip[2],   0,      sizeInChar);
+    memset(m_cbf[0],             0,      sizeInChar);
+    memset(m_cbf[1],             0,      sizeInChar);
+    memset(m_cbf[2],             0,      sizeInChar);
+    memset(m_depth,              depth,  sizeInChar);
+    memset(m_log2CUSize,         log2CUSize, sizeInChar);
+    memset(m_partSizes,          SIZE_NONE, sizeInChar);
+    memset(m_predModes,          MODE_NONE, sizeInChar);
+    memset(m_skipFlag,           false, sizeInBool);
+    memset(m_cuTransquantBypass, false, sizeInBool);
 
-    m_cuMvField[0].clearMvField();
-    m_cuMvField[1].clearMvField();
+    if (m_slice->m_sliceType != I_SLICE)
+    {
+        memset(m_bMergeFlags,      0, sizeInBool);
+        memset(m_interDir,         0, sizeInChar);
+
+        m_cuMvField[0].clearMvField();
+        m_cuMvField[1].clearMvField();
+    }
 
     m_cuLeft        = cu->getCULeft();
     m_cuAbove       = cu->getCUAbove();
     m_cuAboveLeft   = cu->getCUAboveLeft();
     m_cuAboveRight  = cu->getCUAboveRight();
-
-    m_cuColocated[0] = cu->getCUColocated(REF_PIC_LIST_0);
-    m_cuColocated[1] = cu->getCUColocated(REF_PIC_LIST_1);
 }
 
-
 void TComDataCU::copyToSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth)
 {
-    assert(partUnitIdx < 4);
+    X265_CHECK(partUnitIdx < 4, "part unit should be less than 4\n");
 
     uint32_t partOffset = (cu->getTotalNumPart() >> 2) * partUnitIdx;
 
-    m_pic              = cu->getPic();
-    m_slice            = m_pic->getSlice();
+    m_pic              = cu->m_pic;
+    m_slice            = cu->m_slice;
     m_cuAddr           = cu->getAddr();
     m_absIdxInLCU      = cu->getZorderIdxInCU() + partOffset;
 
-    m_cuPelX           = cu->getCUPelX() + (g_maxCUSize >> depth) * (partUnitIdx & 1);
-    m_cuPelY           = cu->getCUPelY() + (g_maxCUSize >> depth) * (partUnitIdx >> 1);
+    m_cuPelX           = cu->getCUPelX() + ((partUnitIdx &  1) << (g_maxLog2CUSize - depth));
+    m_cuPelY           = cu->getCUPelY() + ((partUnitIdx >> 1) << (g_maxLog2CUSize - depth));
 
-    m_totalCost        = MAX_INT64;
+    m_psyEnergy        = 0;
+    m_totalPsyCost     = MAX_INT64;
+    m_totalRDCost      = MAX_INT64;
     m_sa8dCost         = MAX_INT64;
     m_totalDistortion  = 0;
     m_totalBits        = 0;
+    m_mvBits           = 0;
+    m_coeffBits        = 0;
     m_numPartitions    = cu->getTotalNumPart() >> 2;
 
-    TComDataCU* rpcCU = m_pic->getCU(m_cuAddr);
-    int iSizeInUchar  = sizeof(UChar) * m_numPartitions;
+    TComDataCU* otherCU = m_pic->getCU(m_cuAddr);
     int sizeInChar  = sizeof(char) * m_numPartitions;
 
-    memcpy(m_skipFlag, rpcCU->getSkipFlag() + m_absIdxInLCU, sizeof(*m_skipFlag) * m_numPartitions);
-    memcpy(m_qp, rpcCU->getQP() + m_absIdxInLCU, sizeInChar);
+    memcpy(m_skipFlag, otherCU->getSkipFlag() + m_absIdxInLCU, sizeof(*m_skipFlag) * m_numPartitions);
+    memcpy(m_qp, otherCU->getQP() + m_absIdxInLCU, sizeInChar);
 
-    memcpy(m_partSizes, rpcCU->getPartitionSize() + m_absIdxInLCU, sizeof(*m_partSizes) * m_numPartitions);
-    memcpy(m_predModes, rpcCU->getPredictionMode() + m_absIdxInLCU, sizeof(*m_predModes) * m_numPartitions);
+    memcpy(m_partSizes, otherCU->getPartitionSize() + m_absIdxInLCU, sizeof(*m_partSizes) * m_numPartitions);
+    memcpy(m_predModes, otherCU->getPredictionMode() + m_absIdxInLCU, sizeof(*m_predModes) * m_numPartitions);
 
-    memcpy(m_lumaIntraDir, rpcCU->getLumaIntraDir() + m_absIdxInLCU, iSizeInUchar);
-    memcpy(m_depth, rpcCU->getDepth() + m_absIdxInLCU, iSizeInUchar);
-    memcpy(m_cuSize, rpcCU->getCUSize() + m_absIdxInLCU, iSizeInUchar);
+    memcpy(m_lumaIntraDir, otherCU->getLumaIntraDir() + m_absIdxInLCU, sizeInChar);
+    memcpy(m_depth, otherCU->getDepth() + m_absIdxInLCU, sizeInChar);
+    memcpy(m_log2CUSize, otherCU->getLog2CUSize() + m_absIdxInLCU, sizeInChar);
 }
 
 // --------------------------------------------------------------------------------------------------------------------
@@ -576,171 +499,183 @@
 // One of quarter parts overwritten by predicted sub part.
 void TComDataCU::copyPartFrom(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth, bool isRDObasedAnalysis)
 {
-    assert(partUnitIdx < 4);
+    X265_CHECK(partUnitIdx < 4, "part unit should be less than 4\n");
     if (isRDObasedAnalysis)
-        m_totalCost += cu->m_totalCost;
+    {
+        m_totalPsyCost += cu->m_totalPsyCost;
+        m_totalRDCost  += cu->m_totalRDCost;
+    }
 
+    m_psyEnergy        += cu->m_psyEnergy;
     m_totalDistortion  += cu->m_totalDistortion;
     m_totalBits        += cu->m_totalBits;
+    m_mvBits           += cu->m_mvBits;
+    m_coeffBits        += cu->m_coeffBits;
 
-    uint32_t offset         = cu->getTotalNumPart() * partUnitIdx;
-
+    uint32_t offset       = cu->getTotalNumPart() * partUnitIdx;
     uint32_t numPartition = cu->getTotalNumPart();
-    int iSizeInUchar  = sizeof(UChar) * numPartition;
-    int iSizeInBool   = sizeof(bool) * numPartition;
-
+    int sizeInBool  = sizeof(bool) * numPartition;
     int sizeInChar  = sizeof(char) * numPartition;
     memcpy(m_skipFlag  + offset, cu->getSkipFlag(),       sizeof(*m_skipFlag)   * numPartition);
     memcpy(m_qp        + offset, cu->getQP(),             sizeInChar);
     memcpy(m_partSizes + offset, cu->getPartitionSize(),  sizeof(*m_partSizes) * numPartition);
     memcpy(m_predModes + offset, cu->getPredictionMode(), sizeof(*m_predModes) * numPartition);
     memcpy(m_cuTransquantBypass + offset, cu->getCUTransquantBypass(), sizeof(*m_cuTransquantBypass) * numPartition);
-    memcpy(m_bMergeFlags      + offset, cu->getMergeFlag(),         iSizeInBool);
-    memcpy(m_lumaIntraDir     + offset, cu->getLumaIntraDir(),      iSizeInUchar);
-    memcpy(m_chromaIntraDir   + offset, cu->getChromaIntraDir(),    iSizeInUchar);
-    memcpy(m_interDir         + offset, cu->getInterDir(),          iSizeInUchar);
-    memcpy(m_trIdx            + offset, cu->getTransformIdx(),      iSizeInUchar);
-    memcpy(m_transformSkip[0] + offset, cu->getTransformSkip(TEXT_LUMA),     iSizeInUchar);
-    memcpy(m_transformSkip[1] + offset, cu->getTransformSkip(TEXT_CHROMA_U), iSizeInUchar);
-    memcpy(m_transformSkip[2] + offset, cu->getTransformSkip(TEXT_CHROMA_V), iSizeInUchar);
-
-    memcpy(m_cbf[0] + offset, cu->getCbf(TEXT_LUMA), iSizeInUchar);
-    memcpy(m_cbf[1] + offset, cu->getCbf(TEXT_CHROMA_U), iSizeInUchar);
-    memcpy(m_cbf[2] + offset, cu->getCbf(TEXT_CHROMA_V), iSizeInUchar);
+    memcpy(m_bMergeFlags      + offset, cu->getMergeFlag(),         sizeInBool);
+    memcpy(m_lumaIntraDir     + offset, cu->getLumaIntraDir(),      sizeInChar);
+    memcpy(m_chromaIntraDir   + offset, cu->getChromaIntraDir(),    sizeInChar);
+    memcpy(m_interDir         + offset, cu->getInterDir(),          sizeInChar);
+    memcpy(m_trIdx            + offset, cu->getTransformIdx(),      sizeInChar);
+    memcpy(m_transformSkip[0] + offset, cu->getTransformSkip(TEXT_LUMA),     sizeInChar);
+    memcpy(m_transformSkip[1] + offset, cu->getTransformSkip(TEXT_CHROMA_U), sizeInChar);
+    memcpy(m_transformSkip[2] + offset, cu->getTransformSkip(TEXT_CHROMA_V), sizeInChar);
+
+    memcpy(m_cbf[0] + offset, cu->getCbf(TEXT_LUMA), sizeInChar);
+    memcpy(m_cbf[1] + offset, cu->getCbf(TEXT_CHROMA_U), sizeInChar);
+    memcpy(m_cbf[2] + offset, cu->getCbf(TEXT_CHROMA_V), sizeInChar);
 
-    memcpy(m_depth  + offset, cu->getDepth(),  iSizeInUchar);
-    memcpy(m_cuSize + offset, cu->getCUSize(), iSizeInUchar);
+    memcpy(m_depth  + offset, cu->getDepth(),  sizeInChar);
+    memcpy(m_log2CUSize + offset, cu->getLog2CUSize(), sizeInChar);
 
-    memcpy(m_mvpIdx[0] + offset, cu->getMVPIdx(REF_PIC_LIST_0), iSizeInUchar);
-    memcpy(m_mvpIdx[1] + offset, cu->getMVPIdx(REF_PIC_LIST_1), iSizeInUchar);
-
-    memcpy(m_iPCMFlags + offset, cu->getIPCMFlag(), iSizeInBool);
+    memcpy(m_mvpIdx[0] + offset, cu->getMVPIdx(REF_PIC_LIST_0), sizeInChar);
+    memcpy(m_mvpIdx[1] + offset, cu->getMVPIdx(REF_PIC_LIST_1), sizeInChar);
 
     m_cuAboveLeft      = cu->getCUAboveLeft();
     m_cuAboveRight     = cu->getCUAboveRight();
     m_cuAbove          = cu->getCUAbove();
     m_cuLeft           = cu->getCULeft();
 
-    m_cuColocated[0] = cu->getCUColocated(REF_PIC_LIST_0);
-    m_cuColocated[1] = cu->getCUColocated(REF_PIC_LIST_1);
-
     m_cuMvField[0].copyFrom(cu->getCUMvField(REF_PIC_LIST_0), cu->getTotalNumPart(), offset);
     m_cuMvField[1].copyFrom(cu->getCUMvField(REF_PIC_LIST_1), cu->getTotalNumPart(), offset);
 
-    uint32_t tmp  = g_maxCUSize * g_maxCUSize >> (depth << 1);
+    uint32_t tmp  = 1 << ((g_maxLog2CUSize - depth) * 2);
     uint32_t tmp2 = partUnitIdx * tmp;
-    memcpy(m_trCoeffY  + tmp2, cu->getCoeffY(),  sizeof(TCoeff) * tmp);
-    memcpy(m_iPCMSampleY + tmp2, cu->getPCMSampleY(), sizeof(Pel) * tmp);
+    memcpy(m_trCoeff[0] + tmp2, cu->getCoeffY(), sizeof(coeff_t) * tmp);
+
+    uint32_t tmpC  = tmp  >> (m_hChromaShift + m_vChromaShift);
+    uint32_t tmpC2 = tmp2 >> (m_hChromaShift + m_vChromaShift);
+    memcpy(m_trCoeff[1] + tmpC2, cu->m_trCoeff[1], sizeof(coeff_t) * tmpC);
+    memcpy(m_trCoeff[2] + tmpC2, cu->m_trCoeff[2], sizeof(coeff_t) * tmpC);
+
+    if (m_slice->m_pps->bTransquantBypassEnabled)
+    {
+        memcpy(m_tqBypassOrigYuv[0] + tmp2, cu->getLumaOrigYuv(), sizeof(pixel) * tmp);
 
-    tmp  = (g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift) >> (depth << 1);
-    tmp2 = partUnitIdx * tmp;
-    memcpy(m_trCoeffCb + tmp2, cu->getCoeffCb(), sizeof(TCoeff) * tmp);
-    memcpy(m_trCoeffCr + tmp2, cu->getCoeffCr(), sizeof(TCoeff) * tmp);
-    memcpy(m_iPCMSampleCb + tmp2, cu->getPCMSampleCb(), sizeof(Pel) * tmp);
-    memcpy(m_iPCMSampleCr + tmp2, cu->getPCMSampleCr(), sizeof(Pel) * tmp);
+        memcpy(m_tqBypassOrigYuv[1] + tmpC2, cu->getChromaOrigYuv(1), sizeof(pixel) * tmpC);
+        memcpy(m_tqBypassOrigYuv[2] + tmpC2, cu->getChromaOrigYuv(2), sizeof(pixel) * tmpC);
+    }
 }
 
 // Copy current predicted part to a CU in picture.
 // It is used to predict for next part
-void TComDataCU::copyToPic(UChar uhDepth)
+void TComDataCU::copyToPic(uint32_t depth)
 {
-    TComDataCU* rpcCU = m_pic->getCU(m_cuAddr);
-
-    rpcCU->m_totalCost       = m_totalCost;
-    rpcCU->m_totalDistortion = m_totalDistortion;
-    rpcCU->m_totalBits       = m_totalBits;
+    TComDataCU* cu = m_pic->getCU(m_cuAddr);
 
-    int iSizeInUchar  = sizeof(UChar) * m_numPartitions;
-    int iSizeInBool   = sizeof(bool) * m_numPartitions;
+    cu->m_psyEnergy       = m_psyEnergy;
+    cu->m_totalPsyCost    = m_totalPsyCost;
+    cu->m_totalRDCost     = m_totalRDCost;
+    cu->m_totalDistortion = m_totalDistortion;
+    cu->m_totalBits       = m_totalBits;
+    cu->m_mvBits          = m_mvBits;
+    cu->m_coeffBits       = m_coeffBits;
 
+    int sizeInBool  = sizeof(bool) * m_numPartitions;
     int sizeInChar  = sizeof(char) * m_numPartitions;
 
-    memcpy(rpcCU->getSkipFlag() + m_absIdxInLCU, m_skipFlag, sizeof(*m_skipFlag) * m_numPartitions);
+    memcpy(cu->getSkipFlag() + m_absIdxInLCU, m_skipFlag, sizeof(*m_skipFlag) * m_numPartitions);
+
+    memcpy(cu->getQP() + m_absIdxInLCU, m_qp, sizeInChar);
+
+    memcpy(cu->getPartitionSize()  + m_absIdxInLCU, m_partSizes, sizeof(*m_partSizes) * m_numPartitions);
+    memcpy(cu->getPredictionMode() + m_absIdxInLCU, m_predModes, sizeof(*m_predModes) * m_numPartitions);
+    memcpy(cu->getCUTransquantBypass() + m_absIdxInLCU, m_cuTransquantBypass, sizeof(*m_cuTransquantBypass) * m_numPartitions);
+    memcpy(cu->getMergeFlag()         + m_absIdxInLCU, m_bMergeFlags,      sizeInBool);
+    memcpy(cu->getLumaIntraDir()      + m_absIdxInLCU, m_lumaIntraDir,     sizeInChar);
+    memcpy(cu->getChromaIntraDir()    + m_absIdxInLCU, m_chromaIntraDir,   sizeInChar);
+    memcpy(cu->getInterDir()          + m_absIdxInLCU, m_interDir,         sizeInChar);
+    memcpy(cu->getTransformIdx()      + m_absIdxInLCU, m_trIdx,            sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_LUMA)     + m_absIdxInLCU, m_transformSkip[0], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + m_absIdxInLCU, m_transformSkip[1], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + m_absIdxInLCU, m_transformSkip[2], sizeInChar);
+
+    memcpy(cu->getCbf(TEXT_LUMA)     + m_absIdxInLCU, m_cbf[0], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_U) + m_absIdxInLCU, m_cbf[1], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_V) + m_absIdxInLCU, m_cbf[2], sizeInChar);
+
+    memcpy(cu->getDepth()  + m_absIdxInLCU, m_depth,  sizeInChar);
+    memcpy(cu->getLog2CUSize() + m_absIdxInLCU, m_log2CUSize, sizeInChar);
+
+    memcpy(cu->getMVPIdx(REF_PIC_LIST_0) + m_absIdxInLCU, m_mvpIdx[0], sizeInChar);
+    memcpy(cu->getMVPIdx(REF_PIC_LIST_1) + m_absIdxInLCU, m_mvpIdx[1], sizeInChar);
+
+    m_cuMvField[0].copyTo(cu->getCUMvField(REF_PIC_LIST_0), m_absIdxInLCU);
+    m_cuMvField[1].copyTo(cu->getCUMvField(REF_PIC_LIST_1), m_absIdxInLCU);
+
+    uint32_t tmpY  = 1 << ((g_maxLog2CUSize - depth) * 2);
+    uint32_t tmpY2 = m_absIdxInLCU << (LOG2_UNIT_SIZE * 2);
+    memcpy(cu->getCoeffY() + tmpY2, m_trCoeff[0], sizeof(coeff_t) * tmpY);
+
+    uint32_t tmpC  = tmpY  >> (m_hChromaShift + m_vChromaShift);
+    uint32_t tmpC2 = tmpY2 >> (m_hChromaShift + m_vChromaShift);
+    memcpy(cu->m_trCoeff[1] + tmpC2, m_trCoeff[1], sizeof(coeff_t) * tmpC);
+    memcpy(cu->m_trCoeff[2] + tmpC2, m_trCoeff[2], sizeof(coeff_t) * tmpC);
+
+    if (m_slice->m_pps->bTransquantBypassEnabled)
+    {
+        uint32_t tmp  = 1 << ((g_maxLog2CUSize - depth) * 2);
+        uint32_t tmp2 = m_absIdxInLCU << (LOG2_UNIT_SIZE * 2);
+        memcpy(cu->getLumaOrigYuv() + tmp2, m_tqBypassOrigYuv[0], sizeof(pixel) * tmp);
+
+        memcpy(cu->getChromaOrigYuv(1) + tmpC2, m_tqBypassOrigYuv[1], sizeof(pixel) * tmpC);
+        memcpy(cu->getChromaOrigYuv(2) + tmpC2, m_tqBypassOrigYuv[2], sizeof(pixel) * tmpC);
+    }
+}
+
+void TComDataCU::copyCodedToPic(uint32_t depth)
+{
+    TComDataCU* cu = m_pic->getCU(m_cuAddr);
 
-    memcpy(rpcCU->getQP() + m_absIdxInLCU, m_qp, sizeInChar);
+    int sizeInChar  = sizeof(uint8_t) * m_numPartitions;
 
-    memcpy(rpcCU->getPartitionSize()  + m_absIdxInLCU, m_partSizes, sizeof(*m_partSizes) * m_numPartitions);
-    memcpy(rpcCU->getPredictionMode() + m_absIdxInLCU, m_predModes, sizeof(*m_predModes) * m_numPartitions);
-    memcpy(rpcCU->getCUTransquantBypass() + m_absIdxInLCU, m_cuTransquantBypass, sizeof(*m_cuTransquantBypass) * m_numPartitions);
-    memcpy(rpcCU->getMergeFlag()         + m_absIdxInLCU, m_bMergeFlags,         iSizeInBool);
-    memcpy(rpcCU->getLumaIntraDir()      + m_absIdxInLCU, m_lumaIntraDir,     iSizeInUchar);
-    memcpy(rpcCU->getChromaIntraDir()    + m_absIdxInLCU, m_chromaIntraDir,   iSizeInUchar);
-    memcpy(rpcCU->getInterDir()          + m_absIdxInLCU, m_interDir,         iSizeInUchar);
-    memcpy(rpcCU->getTransformIdx()      + m_absIdxInLCU, m_trIdx,            iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_LUMA)     + m_absIdxInLCU, m_transformSkip[0], iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_CHROMA_U) + m_absIdxInLCU, m_transformSkip[1], iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_CHROMA_V) + m_absIdxInLCU, m_transformSkip[2], iSizeInUchar);
-
-    memcpy(rpcCU->getCbf(TEXT_LUMA)     + m_absIdxInLCU, m_cbf[0], iSizeInUchar);
-    memcpy(rpcCU->getCbf(TEXT_CHROMA_U) + m_absIdxInLCU, m_cbf[1], iSizeInUchar);
-    memcpy(rpcCU->getCbf(TEXT_CHROMA_V) + m_absIdxInLCU, m_cbf[2], iSizeInUchar);
-
-    memcpy(rpcCU->getDepth()  + m_absIdxInLCU, m_depth,  iSizeInUchar);
-    memcpy(rpcCU->getCUSize() + m_absIdxInLCU, m_cuSize, iSizeInUchar);
-
-    memcpy(rpcCU->getMVPIdx(REF_PIC_LIST_0) + m_absIdxInLCU, m_mvpIdx[0], iSizeInUchar);
-    memcpy(rpcCU->getMVPIdx(REF_PIC_LIST_1) + m_absIdxInLCU, m_mvpIdx[1], iSizeInUchar);
-
-    m_cuMvField[0].copyTo(rpcCU->getCUMvField(REF_PIC_LIST_0), m_absIdxInLCU);
-    m_cuMvField[1].copyTo(rpcCU->getCUMvField(REF_PIC_LIST_1), m_absIdxInLCU);
-
-    memcpy(rpcCU->getIPCMFlag() + m_absIdxInLCU, m_iPCMFlags,         iSizeInBool);
-
-    uint32_t tmp  = (g_maxCUSize * g_maxCUSize) >> (uhDepth << 1);
-    uint32_t tmp2 = m_absIdxInLCU * m_pic->getMinCUSize() * m_pic->getMinCUSize();
-    memcpy(rpcCU->getCoeffY()     + tmp2, m_trCoeffY,    sizeof(TCoeff) * tmp);
-    memcpy(rpcCU->getPCMSampleY() + tmp2, m_iPCMSampleY, sizeof(Pel) * tmp);
-    tmp  = ((g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift)) >> (uhDepth << 1);
-    tmp2 = m_absIdxInLCU * (m_pic->getMinCUSize() >> m_hChromaShift) * (m_pic->getMinCUSize() >> m_vChromaShift);
-    memcpy(rpcCU->getCoeffCb() + tmp2, m_trCoeffCb, sizeof(TCoeff) * tmp);
-    memcpy(rpcCU->getCoeffCr() + tmp2, m_trCoeffCr, sizeof(TCoeff) * tmp);
-    memcpy(rpcCU->getPCMSampleCb() + tmp2, m_iPCMSampleCb, sizeof(Pel) * tmp);
-    memcpy(rpcCU->getPCMSampleCr() + tmp2, m_iPCMSampleCr, sizeof(Pel) * tmp);
-}
-
-void TComDataCU::copyCodedToPic(UChar depth)
-{
-    TComDataCU* rpcCU = m_pic->getCU(m_cuAddr);
-
-    int iSizeInUchar  = sizeof(UChar) * m_numPartitions;
-
-    memcpy(rpcCU->getSkipFlag() + m_absIdxInLCU, m_skipFlag, sizeof(*m_skipFlag) * m_numPartitions);
-    memcpy(rpcCU->getTransformIdx() + m_absIdxInLCU, m_trIdx, iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_LUMA) + m_absIdxInLCU, m_transformSkip[0], iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_CHROMA_U) + m_absIdxInLCU, m_transformSkip[1], iSizeInUchar);
-    memcpy(rpcCU->getTransformSkip(TEXT_CHROMA_V) + m_absIdxInLCU, m_transformSkip[2], iSizeInUchar);
-    memcpy(rpcCU->getChromaIntraDir() + m_absIdxInLCU, m_chromaIntraDir, iSizeInUchar);
-    memcpy(rpcCU->getQP() + m_absIdxInLCU, m_qp, sizeof(char) * m_numPartitions);
-
-    memcpy(rpcCU->getCbf(TEXT_LUMA) + m_absIdxInLCU, m_cbf[0], iSizeInUchar);
-    memcpy(rpcCU->getCbf(TEXT_CHROMA_U) + m_absIdxInLCU, m_cbf[1], iSizeInUchar);
-    memcpy(rpcCU->getCbf(TEXT_CHROMA_V) + m_absIdxInLCU, m_cbf[2], iSizeInUchar);
-
-    uint32_t tmp  = (g_maxCUSize * g_maxCUSize) >> (depth << 1);
-    uint32_t tmp2 = m_absIdxInLCU * m_pic->getMinCUSize() * m_pic->getMinCUSize();
-    memcpy(rpcCU->getCoeffY() + tmp2, m_trCoeffY, sizeof(TCoeff) * tmp);
-
-    tmp  = ((g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_hChromaShift)) >> (depth << 1);
-    tmp2 = m_absIdxInLCU * (m_pic->getMinCUSize() >> m_hChromaShift) * (m_pic->getMinCUSize() >> m_vChromaShift);
-    memcpy(rpcCU->getCoeffCb() + tmp2, m_trCoeffCb, sizeof(TCoeff) * tmp);
-    memcpy(rpcCU->getCoeffCr() + tmp2, m_trCoeffCr, sizeof(TCoeff) * tmp);
+    memcpy(cu->getSkipFlag() + m_absIdxInLCU, m_skipFlag, sizeof(*m_skipFlag) * m_numPartitions);
+    memcpy(cu->getTransformIdx() + m_absIdxInLCU, m_trIdx, sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_LUMA) + m_absIdxInLCU, m_transformSkip[0], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + m_absIdxInLCU, m_transformSkip[1], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + m_absIdxInLCU, m_transformSkip[2], sizeInChar);
+    memcpy(cu->getChromaIntraDir() + m_absIdxInLCU, m_chromaIntraDir, sizeInChar);
+    memcpy(cu->getQP() + m_absIdxInLCU, m_qp, sizeof(char) * m_numPartitions);
+
+    memcpy(cu->getCbf(TEXT_LUMA) + m_absIdxInLCU, m_cbf[0], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_U) + m_absIdxInLCU, m_cbf[1], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_V) + m_absIdxInLCU, m_cbf[2], sizeInChar);
+
+    uint32_t tmpY  = 1 << ((g_maxLog2CUSize - depth) * 2);
+    uint32_t tmpY2 = m_absIdxInLCU << (LOG2_UNIT_SIZE * 2);
+    memcpy(cu->getCoeffY() + tmpY2, m_trCoeff[0], sizeof(coeff_t) * tmpY);
+    tmpY  >>= m_hChromaShift + m_vChromaShift;
+    tmpY2 >>= m_hChromaShift + m_vChromaShift;
+    memcpy(cu->m_trCoeff[1] + tmpY2, m_trCoeff[1], sizeof(coeff_t) * tmpY);
+    memcpy(cu->m_trCoeff[2] + tmpY2, m_trCoeff[2], sizeof(coeff_t) * tmpY);
 }
 
-void TComDataCU::copyToPic(UChar depth, uint32_t partIdx, uint32_t partDepth)
+void TComDataCU::copyToPic(uint32_t depth, uint32_t partIdx, uint32_t partDepth)
 {
     TComDataCU* cu = m_pic->getCU(m_cuAddr);
     uint32_t qNumPart  = m_numPartitions >> (partDepth << 1);
 
-    uint32_t uiPartStart = partIdx * qNumPart;
-    uint32_t partOffset  = m_absIdxInLCU + uiPartStart;
+    uint32_t partStart = partIdx * qNumPart;
+    uint32_t partOffset  = m_absIdxInLCU + partStart;
 
-    cu->m_totalCost       = m_totalCost;
+    cu->m_psyEnergy       = m_psyEnergy;
+    cu->m_totalPsyCost    = m_totalPsyCost;
+    cu->m_totalRDCost     = m_totalRDCost;
     cu->m_totalDistortion = m_totalDistortion;
     cu->m_totalBits       = m_totalBits;
+    cu->m_mvBits          = m_mvBits;
+    cu->m_coeffBits       = m_coeffBits;
 
-    int sizeInUchar = sizeof(UChar) * qNumPart;
     int sizeInBool  = sizeof(bool) * qNumPart;
-
     int sizeInChar  = sizeof(char) * qNumPart;
     memcpy(cu->getSkipFlag() + partOffset, m_skipFlag,   sizeof(*m_skipFlag)   * qNumPart);
 
@@ -749,54 +684,57 @@
     memcpy(cu->getPredictionMode() + partOffset, m_predModes, sizeof(*m_predModes) * qNumPart);
     memcpy(cu->getCUTransquantBypass() + partOffset, m_cuTransquantBypass, sizeof(*m_cuTransquantBypass) * qNumPart);
     memcpy(cu->getMergeFlag()         + partOffset, m_bMergeFlags,      sizeInBool);
-    memcpy(cu->getLumaIntraDir()      + partOffset, m_lumaIntraDir,     sizeInUchar);
-    memcpy(cu->getChromaIntraDir()    + partOffset, m_chromaIntraDir,   sizeInUchar);
-    memcpy(cu->getInterDir()          + partOffset, m_interDir,         sizeInUchar);
-    memcpy(cu->getTransformIdx()      + partOffset, m_trIdx,            sizeInUchar);
-    memcpy(cu->getTransformSkip(TEXT_LUMA)     + partOffset, m_transformSkip[0], sizeInUchar);
-    memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + partOffset, m_transformSkip[1], sizeInUchar);
-    memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + partOffset, m_transformSkip[2], sizeInUchar);
-    memcpy(cu->getCbf(TEXT_LUMA)     + partOffset, m_cbf[0], sizeInUchar);
-    memcpy(cu->getCbf(TEXT_CHROMA_U) + partOffset, m_cbf[1], sizeInUchar);
-    memcpy(cu->getCbf(TEXT_CHROMA_V) + partOffset, m_cbf[2], sizeInUchar);
-
-    memcpy(cu->getDepth()  + partOffset, m_depth,  sizeInUchar);
-    memcpy(cu->getCUSize() + partOffset, m_cuSize, sizeInUchar);
-
-    memcpy(cu->getMVPIdx(REF_PIC_LIST_0) + partOffset, m_mvpIdx[0], sizeInUchar);
-    memcpy(cu->getMVPIdx(REF_PIC_LIST_1) + partOffset, m_mvpIdx[1], sizeInUchar);
-    m_cuMvField[0].copyTo(cu->getCUMvField(REF_PIC_LIST_0), m_absIdxInLCU, uiPartStart, qNumPart);
-    m_cuMvField[1].copyTo(cu->getCUMvField(REF_PIC_LIST_1), m_absIdxInLCU, uiPartStart, qNumPart);
-
-    memcpy(cu->getIPCMFlag() + partOffset, m_iPCMFlags, sizeInBool);
-
-    uint32_t tmp  = (g_maxCUSize * g_maxCUSize) >> ((depth + partDepth) << 1);
-    uint32_t tmp2 = partOffset * m_pic->getMinCUSize() * m_pic->getMinCUSize();
-    memcpy(cu->getCoeffY()  + tmp2, m_trCoeffY,  sizeof(TCoeff) * tmp);
-    memcpy(cu->getPCMSampleY() + tmp2, m_iPCMSampleY, sizeof(Pel) * tmp);
-
-    tmp  = ((g_maxCUSize >> m_hChromaShift) * (g_maxCUSize >> m_vChromaShift)) >> ((depth + partDepth) << 1);
-    tmp2 = partOffset * (m_pic->getMinCUSize() >> m_hChromaShift) * (m_pic->getMinCUSize() >> m_vChromaShift);
-    memcpy(cu->getCoeffCb() + tmp2, m_trCoeffCb, sizeof(TCoeff) * tmp);
-    memcpy(cu->getCoeffCr() + tmp2, m_trCoeffCr, sizeof(TCoeff) * tmp);
-    memcpy(cu->getPCMSampleCb() + tmp2, m_iPCMSampleCb, sizeof(Pel) * tmp);
-    memcpy(cu->getPCMSampleCr() + tmp2, m_iPCMSampleCr, sizeof(Pel) * tmp);
+    memcpy(cu->getLumaIntraDir()      + partOffset, m_lumaIntraDir,     sizeInChar);
+    memcpy(cu->getChromaIntraDir()    + partOffset, m_chromaIntraDir,   sizeInChar);
+    memcpy(cu->getInterDir()          + partOffset, m_interDir,         sizeInChar);
+    memcpy(cu->getTransformIdx()      + partOffset, m_trIdx,            sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_LUMA)     + partOffset, m_transformSkip[0], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + partOffset, m_transformSkip[1], sizeInChar);
+    memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + partOffset, m_transformSkip[2], sizeInChar);
+    memcpy(cu->getCbf(TEXT_LUMA)     + partOffset, m_cbf[0], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_U) + partOffset, m_cbf[1], sizeInChar);
+    memcpy(cu->getCbf(TEXT_CHROMA_V) + partOffset, m_cbf[2], sizeInChar);
+
+    memcpy(cu->getDepth()  + partOffset, m_depth,  sizeInChar);
+    memcpy(cu->getLog2CUSize() + partOffset, m_log2CUSize, sizeInChar);
+
+    memcpy(cu->getMVPIdx(REF_PIC_LIST_0) + partOffset, m_mvpIdx[0], sizeInChar);
+    memcpy(cu->getMVPIdx(REF_PIC_LIST_1) + partOffset, m_mvpIdx[1], sizeInChar);
+    m_cuMvField[0].copyTo(cu->getCUMvField(REF_PIC_LIST_0), m_absIdxInLCU, partStart, qNumPart);
+    m_cuMvField[1].copyTo(cu->getCUMvField(REF_PIC_LIST_1), m_absIdxInLCU, partStart, qNumPart);
+
+    uint32_t tmpY  = 1 << ((g_maxLog2CUSize - depth - partDepth) * 2);
+    uint32_t tmpY2 = partOffset << (LOG2_UNIT_SIZE * 2);
+    memcpy(cu->getCoeffY() + tmpY2, m_trCoeff[0],  sizeof(coeff_t) * tmpY);
+
+    uint32_t tmpC  = tmpY >> (m_hChromaShift + m_vChromaShift);
+    uint32_t tmpC2 = tmpY2 >> (m_hChromaShift + m_vChromaShift);
+    memcpy(cu->m_trCoeff[1] + tmpC2, m_trCoeff[1], sizeof(coeff_t) * tmpC);
+    memcpy(cu->m_trCoeff[2] + tmpC2, m_trCoeff[2], sizeof(coeff_t) * tmpC);
+
+    if (m_slice->m_pps->bTransquantBypassEnabled)
+    {
+        memcpy(cu->getLumaOrigYuv() + tmpY2, m_tqBypassOrigYuv[0], sizeof(pixel) * tmpY);
+
+        memcpy(cu->getChromaOrigYuv(1) + tmpC2, m_tqBypassOrigYuv[1], sizeof(pixel) * tmpC);
+        memcpy(cu->getChromaOrigYuv(2) + tmpC2, m_tqBypassOrigYuv[2], sizeof(pixel) * tmpC);
+    }
 }
 
 // --------------------------------------------------------------------------------------------------------------------
 // Other public functions
 // --------------------------------------------------------------------------------------------------------------------
 
-TComDataCU* TComDataCU::getPULeft(uint32_t& lPartUnitIdx, uint32_t curPartUnitIdx, bool  bEnforceSliceRestriction, bool  bEnforceTileRestriction)
+TComDataCU* TComDataCU::getPULeft(uint32_t& lPartUnitIdx, uint32_t curPartUnitIdx)
 {
     uint32_t absPartIdx       = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU];
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
+    uint32_t numPartInCUSize  = m_pic->getNumPartInCUSize();
 
-    if (!RasterAddress::isZeroCol(absPartIdx, numPartInCUWidth))
+    if (!RasterAddress::isZeroCol(absPartIdx, numPartInCUSize))
     {
+        uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU];
         lPartUnitIdx = g_rasterToZscan[absPartIdx - 1];
-        if (RasterAddress::isEqualCol(absPartIdx, absZorderCUIdx, numPartInCUWidth))
+        if (RasterAddress::isEqualCol(absPartIdx, absZorderCUIdx, numPartInCUSize))
         {
             return m_pic->getCU(getAddr());
         }
@@ -807,27 +745,20 @@
         }
     }
 
-    lPartUnitIdx = g_rasterToZscan[absPartIdx + numPartInCUWidth - 1];
-
-    if ((bEnforceSliceRestriction && (m_cuLeft == NULL || m_cuLeft->getSlice() == NULL)) ||
-        (bEnforceTileRestriction && (m_cuLeft == NULL || m_cuLeft->getSlice() == NULL)))
-    {
-        return NULL;
-    }
+    lPartUnitIdx = g_rasterToZscan[absPartIdx + numPartInCUSize - 1];
     return m_cuLeft;
 }
 
-TComDataCU* TComDataCU::getPUAbove(uint32_t& aPartUnitIdx, uint32_t  curPartUnitIdx, bool bEnforceSliceRestriction, bool planarAtLCUBoundary,
-                                   bool bEnforceTileRestriction)
+TComDataCU* TComDataCU::getPUAbove(uint32_t& aPartUnitIdx, uint32_t curPartUnitIdx, bool planarAtLCUBoundary)
 {
     uint32_t absPartIdx       = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU];
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
+    uint32_t numPartInCUSize  = m_pic->getNumPartInCUSize();
 
-    if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUWidth))
+    if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUSize))
     {
-        aPartUnitIdx = g_rasterToZscan[absPartIdx - numPartInCUWidth];
-        if (RasterAddress::isEqualRow(absPartIdx, absZorderCUIdx, numPartInCUWidth))
+        uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU];
+        aPartUnitIdx = g_rasterToZscan[absPartIdx - numPartInCUSize];
+        if (RasterAddress::isEqualRow(absPartIdx, absZorderCUIdx, numPartInCUSize))
         {
             return m_pic->getCU(getAddr());
         }
@@ -839,32 +770,24 @@
     }
 
     if (planarAtLCUBoundary)
-    {
         return NULL;
-    }
-
-    aPartUnitIdx = g_rasterToZscan[absPartIdx + m_pic->getNumPartInCU() - numPartInCUWidth];
 
-    if ((bEnforceSliceRestriction && (m_cuAbove == NULL || m_cuAbove->getSlice() == NULL)) ||
-        (bEnforceTileRestriction && (m_cuAbove == NULL || m_cuAbove->getSlice() == NULL)))
-    {
-        return NULL;
-    }
+    aPartUnitIdx = g_rasterToZscan[absPartIdx + NUM_CU_PARTITIONS - numPartInCUSize];
     return m_cuAbove;
 }
 
-TComDataCU* TComDataCU::getPUAboveLeft(uint32_t& alPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction)
+TComDataCU* TComDataCU::getPUAboveLeft(uint32_t& alPartUnitIdx, uint32_t curPartUnitIdx)
 {
-    uint32_t absPartIdx       = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU];
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
+    uint32_t absPartIdx      = g_zscanToRaster[curPartUnitIdx];
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
 
-    if (!RasterAddress::isZeroCol(absPartIdx, numPartInCUWidth))
+    if (!RasterAddress::isZeroCol(absPartIdx, numPartInCUSize))
     {
-        if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUWidth))
+        if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUSize))
         {
-            alPartUnitIdx = g_rasterToZscan[absPartIdx - numPartInCUWidth - 1];
-            if (RasterAddress::isEqualRowOrCol(absPartIdx, absZorderCUIdx, numPartInCUWidth))
+            uint32_t absZorderCUIdx  = g_zscanToRaster[m_absIdxInLCU];
+            alPartUnitIdx = g_rasterToZscan[absPartIdx - numPartInCUSize - 1];
+            if (RasterAddress::isEqualRowOrCol(absPartIdx, absZorderCUIdx, numPartInCUSize))
             {
                 return m_pic->getCU(getAddr());
             }
@@ -874,53 +797,37 @@
                 return this;
             }
         }
-        alPartUnitIdx = g_rasterToZscan[absPartIdx + getPic()->getNumPartInCU() - numPartInCUWidth - 1];
-        if ((bEnforceSliceRestriction && (m_cuAbove == NULL || m_cuAbove->getSlice() == NULL)))
-        {
-            return NULL;
-        }
+        alPartUnitIdx = g_rasterToZscan[absPartIdx + NUM_CU_PARTITIONS - numPartInCUSize - 1];
         return m_cuAbove;
     }
 
-    if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUWidth))
+    if (!RasterAddress::isZeroRow(absPartIdx, numPartInCUSize))
     {
         alPartUnitIdx = g_rasterToZscan[absPartIdx - 1];
-        if ((bEnforceSliceRestriction && (m_cuLeft == NULL || m_cuLeft->getSlice() == NULL))
-            )
-        {
-            return NULL;
-        }
         return m_cuLeft;
     }
 
-    alPartUnitIdx = g_rasterToZscan[m_pic->getNumPartInCU() - 1];
-    if ((bEnforceSliceRestriction && (m_cuAboveLeft == NULL || m_cuAboveLeft->getSlice() == NULL)))
-    {
-        return NULL;
-    }
+    alPartUnitIdx = g_rasterToZscan[NUM_CU_PARTITIONS - 1];
     return m_cuAboveLeft;
 }
 
-TComDataCU* TComDataCU::getPUAboveRight(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction)
+TComDataCU* TComDataCU::getPUAboveRight(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx)
 {
-    uint32_t absPartIdxRT     = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU] + m_cuSize[0] / m_pic->getMinCUSize() - 1;
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-
-    if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_rasterToPelX[absPartIdxRT] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicWidthInLumaSamples())
-    {
-        arPartUnitIdx = MAX_UINT;
+    if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_zscanToPelX[curPartUnitIdx] + UNIT_SIZE) >= m_slice->m_sps->picWidthInLumaSamples)
         return NULL;
-    }
 
-    if (RasterAddress::lessThanCol(absPartIdxRT, numPartInCUWidth - 1, numPartInCUWidth))
+    uint32_t absPartIdxRT    = g_zscanToRaster[curPartUnitIdx];
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+
+    if (RasterAddress::lessThanCol(absPartIdxRT, numPartInCUSize - 1, numPartInCUSize))
     {
-        if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUWidth))
+        if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUSize))
         {
-            if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - numPartInCUWidth + 1])
+            if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - numPartInCUSize + 1])
             {
-                arPartUnitIdx = g_rasterToZscan[absPartIdxRT - numPartInCUWidth + 1];
-                if (RasterAddress::isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, numPartInCUWidth))
+                uint32_t absZorderCUIdx  = g_zscanToRaster[m_absIdxInLCU] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1;
+                arPartUnitIdx = g_rasterToZscan[absPartIdxRT - numPartInCUSize + 1];
+                if (RasterAddress::isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, numPartInCUSize))
                 {
                     return m_pic->getCU(getAddr());
                 }
@@ -930,52 +837,38 @@
                     return this;
                 }
             }
-            arPartUnitIdx = MAX_UINT;
-            return NULL;
-        }
-        arPartUnitIdx = g_rasterToZscan[absPartIdxRT + m_pic->getNumPartInCU() - numPartInCUWidth + 1];
-        if ((bEnforceSliceRestriction && (m_cuAbove == NULL || m_cuAbove->getSlice() == NULL)))
-        {
             return NULL;
         }
+        arPartUnitIdx = g_rasterToZscan[absPartIdxRT + NUM_CU_PARTITIONS - numPartInCUSize + 1];
         return m_cuAbove;
     }
 
-    if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUWidth))
+    if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUSize))
     {
-        arPartUnitIdx = MAX_UINT;
         return NULL;
     }
 
-    arPartUnitIdx = g_rasterToZscan[m_pic->getNumPartInCU() - numPartInCUWidth];
-    if ((bEnforceSliceRestriction && (m_cuAboveRight == NULL || m_cuAboveRight->getSlice() == NULL ||
-                                      (m_cuAboveRight->getAddr()) > getAddr())))
-    {
-        return NULL;
-    }
+    arPartUnitIdx = g_rasterToZscan[NUM_CU_PARTITIONS - numPartInCUSize];
     return m_cuAboveRight;
 }
 
-TComDataCU* TComDataCU::getPUBelowLeft(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction)
+TComDataCU* TComDataCU::getPUBelowLeft(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx)
 {
-    uint32_t absPartIdxLB     = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInLCU] + (m_cuSize[0] / m_pic->getMinCUSize() - 1) * m_pic->getNumPartInWidth();
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-
-    if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_rasterToPelY[absPartIdxLB] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicHeightInLumaSamples())
-    {
-        blPartUnitIdx = MAX_UINT;
+    if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_zscanToPelY[curPartUnitIdx] + UNIT_SIZE) >= m_slice->m_sps->picHeightInLumaSamples)
         return NULL;
-    }
 
-    if (RasterAddress::lessThanRow(absPartIdxLB, m_pic->getNumPartInHeight() - 1, numPartInCUWidth))
+    uint32_t absPartIdxLB    = g_zscanToRaster[curPartUnitIdx];
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+
+    if (RasterAddress::lessThanRow(absPartIdxLB, numPartInCUSize - 1, numPartInCUSize))
     {
-        if (!RasterAddress::isZeroCol(absPartIdxLB, numPartInCUWidth))
+        if (!RasterAddress::isZeroCol(absPartIdxLB, numPartInCUSize))
         {
-            if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + numPartInCUWidth - 1])
+            if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + numPartInCUSize - 1])
             {
-                blPartUnitIdx = g_rasterToZscan[absPartIdxLB + numPartInCUWidth - 1];
-                if (RasterAddress::isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, numPartInCUWidth))
+                uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInLCU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1) * m_pic->getNumPartInCUSize();
+                blPartUnitIdx = g_rasterToZscan[absPartIdxLB + numPartInCUSize - 1];
+                if (RasterAddress::isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, numPartInCUSize))
                 {
                     return m_pic->getCU(getAddr());
                 }
@@ -985,42 +878,35 @@
                     return this;
                 }
             }
-            blPartUnitIdx = MAX_UINT;
-            return NULL;
-        }
-        blPartUnitIdx = g_rasterToZscan[absPartIdxLB + numPartInCUWidth * 2 - 1];
-        if ((bEnforceSliceRestriction && (m_cuLeft == NULL || m_cuLeft->getSlice() == NULL)))
-        {
             return NULL;
         }
+        blPartUnitIdx = g_rasterToZscan[absPartIdxLB + numPartInCUSize * 2 - 1];
         return m_cuLeft;
     }
 
-    blPartUnitIdx = MAX_UINT;
     return NULL;
 }
 
 TComDataCU* TComDataCU::getPUBelowLeftAdi(uint32_t& blPartUnitIdx,  uint32_t curPartUnitIdx, uint32_t partUnitOffset)
 {
-    uint32_t absPartIdxLB     = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInLCU] + ((m_cuSize[0] / m_pic->getMinCUSize()) - 1) * m_pic->getNumPartInWidth();
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-
-    if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_rasterToPelY[absPartIdxLB] + (m_pic->getPicSym()->getMinCUSize() * partUnitOffset)) >=
-        m_slice->getSPS()->getPicHeightInLumaSamples())
+    if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_zscanToPelY[curPartUnitIdx] + (partUnitOffset << LOG2_UNIT_SIZE)) >=
+        m_slice->m_sps->picHeightInLumaSamples)
     {
-        blPartUnitIdx = MAX_UINT;
         return NULL;
     }
 
-    if (RasterAddress::lessThanRow(absPartIdxLB, m_pic->getNumPartInHeight() - partUnitOffset, numPartInCUWidth))
+    uint32_t absPartIdxLB    = g_zscanToRaster[curPartUnitIdx];
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+
+    if (RasterAddress::lessThanRow(absPartIdxLB, numPartInCUSize - partUnitOffset, numPartInCUSize))
     {
-        if (!RasterAddress::isZeroCol(absPartIdxLB, numPartInCUWidth))
+        if (!RasterAddress::isZeroCol(absPartIdxLB, numPartInCUSize))
         {
-            if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + partUnitOffset * numPartInCUWidth - 1])
+            if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + partUnitOffset * numPartInCUSize - 1])
             {
-                blPartUnitIdx = g_rasterToZscan[absPartIdxLB + partUnitOffset * numPartInCUWidth - 1];
-                if (RasterAddress::isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, numPartInCUWidth))
+                uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInLCU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1) * m_pic->getNumPartInCUSize();
+                blPartUnitIdx = g_rasterToZscan[absPartIdxLB + partUnitOffset * numPartInCUSize - 1];
+                if (RasterAddress::isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, numPartInCUSize))
                 {
                     return m_pic->getCU(getAddr());
                 }
@@ -1030,42 +916,39 @@
                     return this;
                 }
             }
-            blPartUnitIdx = MAX_UINT;
             return NULL;
         }
-        blPartUnitIdx = g_rasterToZscan[absPartIdxLB + (1 + partUnitOffset) * numPartInCUWidth - 1];
-        if (m_cuLeft == NULL || m_cuLeft->getSlice() == NULL)
+        blPartUnitIdx = g_rasterToZscan[absPartIdxLB + (1 + partUnitOffset) * numPartInCUSize - 1];
+        if (m_cuLeft == NULL || m_cuLeft->m_slice == NULL)
         {
             return NULL;
         }
         return m_cuLeft;
     }
 
-    blPartUnitIdx = MAX_UINT;
     return NULL;
 }
 
 TComDataCU* TComDataCU::getPUAboveRightAdi(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx, uint32_t partUnitOffset)
 {
-    uint32_t absPartIdxRT     = g_zscanToRaster[curPartUnitIdx];
-    uint32_t absZorderCUIdx   = g_zscanToRaster[m_absIdxInLCU] + (m_cuSize[0] / m_pic->getMinCUSize()) - 1;
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-
-    if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_rasterToPelX[absPartIdxRT] + (m_pic->getPicSym()->getMinCUSize() * partUnitOffset)) >=
-        m_slice->getSPS()->getPicWidthInLumaSamples())
+    if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_zscanToPelX[curPartUnitIdx] + (partUnitOffset << LOG2_UNIT_SIZE)) >=
+        m_slice->m_sps->picWidthInLumaSamples)
     {
-        arPartUnitIdx = MAX_UINT;
         return NULL;
     }
 
-    if (RasterAddress::lessThanCol(absPartIdxRT, numPartInCUWidth - partUnitOffset, numPartInCUWidth))
+    uint32_t absPartIdxRT    = g_zscanToRaster[curPartUnitIdx];
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+
+    if (RasterAddress::lessThanCol(absPartIdxRT, numPartInCUSize - partUnitOffset, numPartInCUSize))
     {
-        if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUWidth))
+        if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUSize))
         {
-            if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - numPartInCUWidth + partUnitOffset])
+            if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - numPartInCUSize + partUnitOffset])
             {
-                arPartUnitIdx = g_rasterToZscan[absPartIdxRT - numPartInCUWidth + partUnitOffset];
-                if (RasterAddress::isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, numPartInCUWidth))
+                uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInLCU] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1;
+                arPartUnitIdx = g_rasterToZscan[absPartIdxRT - numPartInCUSize + partUnitOffset];
+                if (RasterAddress::isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, numPartInCUSize))
                 {
                     return m_pic->getCU(getAddr());
                 }
@@ -1075,25 +958,23 @@
                     return this;
                 }
             }
-            arPartUnitIdx = MAX_UINT;
             return NULL;
         }
-        arPartUnitIdx = g_rasterToZscan[absPartIdxRT + m_pic->getNumPartInCU() - numPartInCUWidth + partUnitOffset];
-        if (m_cuAbove == NULL || m_cuAbove->getSlice() == NULL)
+        arPartUnitIdx = g_rasterToZscan[absPartIdxRT + NUM_CU_PARTITIONS - numPartInCUSize + partUnitOffset];
+        if (m_cuAbove == NULL || m_cuAbove->m_slice == NULL)
         {
             return NULL;
         }
         return m_cuAbove;
     }
 
-    if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUWidth))
+    if (!RasterAddress::isZeroRow(absPartIdxRT, numPartInCUSize))
     {
-        arPartUnitIdx = MAX_UINT;
         return NULL;
     }
 
-    arPartUnitIdx = g_rasterToZscan[m_pic->getNumPartInCU() - numPartInCUWidth + partUnitOffset - 1];
-    if ((m_cuAboveRight == NULL || m_cuAboveRight->getSlice() == NULL ||
+    arPartUnitIdx = g_rasterToZscan[NUM_CU_PARTITIONS - numPartInCUSize + partUnitOffset - 1];
+    if ((m_cuAboveRight == NULL || m_cuAboveRight->m_slice == NULL ||
          (m_cuAboveRight->getAddr()) > getAddr()))
     {
         return NULL;
@@ -1102,19 +983,18 @@
 }
 
 /** Get left QpMinCu
-*\param   uiLPartUnitIdx
-*\param   uiCurrAbsIdxInLCU
+*\param   lPartUnitIdx
+*\param   curAbsIdxInLCU
 *\returns TComDataCU*   point of TComDataCU of left QpMinCu
 */
 TComDataCU* TComDataCU::getQpMinCuLeft(uint32_t& lPartUnitIdx, uint32_t curAbsIdxInLCU)
 {
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-    uint32_t absZorderQpMinCUIdx = (curAbsIdxInLCU >> ((g_maxCUDepth - getSlice()->getPPS()->getMaxCuDQPDepth()) << 1)) <<
-        ((g_maxCUDepth - getSlice()->getPPS()->getMaxCuDQPDepth()) << 1);
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+    uint32_t absZorderQpMinCUIdx = curAbsIdxInLCU & (0xFF << (g_maxFullDepth - m_slice->m_pps->maxCuDQPDepth) * 2);
     uint32_t absRorderQpMinCUIdx = g_zscanToRaster[absZorderQpMinCUIdx];
 
     // check for left LCU boundary
-    if (RasterAddress::isZeroCol(absRorderQpMinCUIdx, numPartInCUWidth))
+    if (RasterAddress::isZeroCol(absRorderQpMinCUIdx, numPartInCUSize))
     {
         return NULL;
     }
@@ -1128,32 +1008,31 @@
 
 /** Get Above QpMinCu
 *\param   aPartUnitIdx
-*\param   currAbsIdxInLCU
+*\param   curAbsIdxInLCU
 *\returns TComDataCU*   point of TComDataCU of above QpMinCu
 */
 TComDataCU* TComDataCU::getQpMinCuAbove(uint32_t& aPartUnitIdx, uint32_t curAbsIdxInLCU)
 {
-    uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-    uint32_t absZorderQpMinCUIdx = (curAbsIdxInLCU >> ((g_maxCUDepth - getSlice()->getPPS()->getMaxCuDQPDepth()) << 1)) <<
-        ((g_maxCUDepth - getSlice()->getPPS()->getMaxCuDQPDepth()) << 1);
+    uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+    uint32_t absZorderQpMinCUIdx = curAbsIdxInLCU & (0xFF << (g_maxFullDepth - m_slice->m_pps->maxCuDQPDepth) * 2);
     uint32_t absRorderQpMinCUIdx = g_zscanToRaster[absZorderQpMinCUIdx];
 
     // check for top LCU boundary
-    if (RasterAddress::isZeroRow(absRorderQpMinCUIdx, numPartInCUWidth))
+    if (RasterAddress::isZeroRow(absRorderQpMinCUIdx, numPartInCUSize))
     {
         return NULL;
     }
 
     // get index of top-CU relative to top-left corner of current quantization group
-    aPartUnitIdx = g_rasterToZscan[absRorderQpMinCUIdx - numPartInCUWidth];
+    aPartUnitIdx = g_rasterToZscan[absRorderQpMinCUIdx - numPartInCUSize];
 
     // return pointer to current LCU
     return m_pic->getCU(getAddr());
 }
 
 /** Get reference QP from left QpMinCu or latest coded QP
-*\param   uiCurrAbsIdxInLCU
-*\returns char   reference QP value
+*\param   currAbsIdxInLCU
+*\returns char reference QP value
 */
 char TComDataCU::getRefQP(uint32_t curAbsIdxInLCU)
 {
@@ -1179,7 +1058,7 @@
 
 char TComDataCU::getLastCodedQP(uint32_t absPartIdx)
 {
-    uint32_t quPartIdxMask = ~((1 << ((g_maxCUDepth - getSlice()->getPPS()->getMaxCuDQPDepth()) << 1)) - 1);
+    uint32_t quPartIdxMask = 0xFF << (g_maxFullDepth - m_slice->m_pps->maxCuDQPDepth) * 2;
     int lastValidPartIdx = getLastValidPartIdx(absPartIdx & quPartIdxMask);
 
     if (lastValidPartIdx >= 0)
@@ -1190,29 +1069,20 @@
     {
         if (getZorderIdxInCU() > 0)
         {
-            return getPic()->getCU(getAddr())->getLastCodedQP(getZorderIdxInCU());
+            return m_pic->getCU(getAddr())->getLastCodedQP(getZorderIdxInCU());
         }
-        else if (getAddr() > 0 && !(getSlice()->getPPS()->getEntropyCodingSyncEnabledFlag() &&
-                                    getAddr() % getPic()->getFrameWidthInCU() == 0))
+        else if (getAddr() > 0 && !(m_slice->m_pps->bEntropyCodingSyncEnabled &&
+                                    getAddr() % m_pic->getFrameWidthInCU() == 0))
         {
-            return getPic()->getCU(getAddr() - 1)->getLastCodedQP(getPic()->getNumPartInCU());
+            return m_pic->getCU(getAddr() - 1)->getLastCodedQP(NUM_CU_PARTITIONS);
         }
         else
         {
-            return getSlice()->getSliceQp();
+            return m_slice->m_sliceQp;
         }
     }
 }
 
-/** Check whether the CU is coded in lossless coding mode
- * \param   absPartIdx
- * \returns true if the CU is coded in lossless coding mode; false if otherwise
- */
-bool TComDataCU::isLosslessCoded(uint32_t absPartIdx)
-{
-    return getSlice()->getPPS()->getTransquantBypassEnableFlag() && getCUTransquantBypass(absPartIdx);
-}
-
 /** Get allowed chroma intra modes
 *\param   absPartIdx
 *\param   uiModeList  pointer to chroma intra modes array
@@ -1241,15 +1111,15 @@
 
 /** Get most probable intra modes
 *\param   absPartIdx
-*\param   uiIntraDirPred  pointer to the array for MPM storage
-*\param   piMode          it is set with MPM mode in case both MPM are equal. It is used to restrict RD search at encode side.
+*\param   intraDirPred  pointer to the array for MPM storage
+*\param   mode          it is set with MPM mode in case both MPM are equal. It is used to restrict RD search at encode side.
 *\returns Number of MPM
 */
-int TComDataCU::getIntraDirLumaPredictor(uint32_t absPartIdx, int32_t* intraDirPred)
+int TComDataCU::getIntraDirLumaPredictor(uint32_t absPartIdx, uint32_t* intraDirPred)
 {
     TComDataCU* tempCU;
-    uint32_t        tempPartIdx;
-    int         leftIntraDir, aboveIntraDir;
+    uint32_t    tempPartIdx;
+    uint32_t    leftIntraDir, aboveIntraDir;
 
     // Get intra direction of left PU
     tempCU = getPULeft(tempPartIdx, m_absIdxInLCU + absPartIdx);
@@ -1257,7 +1127,7 @@
     leftIntraDir  = (tempCU && tempCU->isIntra(tempPartIdx)) ? tempCU->getLumaIntraDir(tempPartIdx) : DC_IDX;
 
     // Get intra direction of above PU
-    tempCU = getPUAbove(tempPartIdx, m_absIdxInLCU + absPartIdx, true, true);
+    tempCU = getPUAbove(tempPartIdx, m_absIdxInLCU + absPartIdx, true);
 
     aboveIntraDir = (tempCU && tempCU->isIntra(tempPartIdx)) ? tempCU->getLumaIntraDir(tempPartIdx) : DC_IDX;
 
@@ -1297,8 +1167,8 @@
 uint32_t TComDataCU::getCtxSplitFlag(uint32_t absPartIdx, uint32_t depth)
 {
     TComDataCU* tempCU;
-    uint32_t        tempPartIdx;
-    uint32_t        ctx;
+    uint32_t    tempPartIdx;
+    uint32_t    ctx;
 
     // Get left split flag
     tempCU = getPULeft(tempPartIdx, m_absIdxInLCU + absPartIdx);
@@ -1311,39 +1181,25 @@
     return ctx;
 }
 
-uint32_t TComDataCU::getQuadtreeTULog2MinSizeInCU(uint32_t absPartIdx)
+void TComDataCU::getQuadtreeTULog2MinSizeInCU(uint32_t tuDepthRange[2], uint32_t absPartIdx)
 {
-    uint32_t log2CUSize = g_convertToBit[getCUSize(absPartIdx)] + 2;
-    PartSize  partSize  = getPartitionSize(absPartIdx);
-    uint32_t quadtreeTUMaxDepth = getPredictionMode(absPartIdx) == MODE_INTRA ? m_slice->getSPS()->getQuadtreeTUMaxDepthIntra() : m_slice->getSPS()->getQuadtreeTUMaxDepthInter();
-    int intraSplitFlag = (getPredictionMode(absPartIdx) == MODE_INTRA && partSize == SIZE_NxN) ? 1 : 0;
-    int interSplitFlag = ((quadtreeTUMaxDepth == 1) && (getPredictionMode(absPartIdx) == MODE_INTER) && (partSize != SIZE_2Nx2N));
+    uint32_t log2CUSize = getLog2CUSize(absPartIdx);
+    PartSize partSize   = getPartitionSize(absPartIdx);
+    uint32_t quadtreeTUMaxDepth = getPredictionMode(absPartIdx) == MODE_INTRA ? m_slice->m_sps->quadtreeTUMaxDepthIntra : m_slice->m_sps->quadtreeTUMaxDepthInter;
+    uint32_t intraSplitFlag = (getPredictionMode(absPartIdx) == MODE_INTRA && partSize == SIZE_NxN) ? 1 : 0;
+    uint32_t interSplitFlag = ((quadtreeTUMaxDepth == 1) && (getPredictionMode(absPartIdx) == MODE_INTER) && (partSize != SIZE_2Nx2N));
 
-    uint32_t log2MinTUSizeInCU = 0;
+    tuDepthRange[0] = m_slice->m_sps->quadtreeTULog2MinSize;
+    tuDepthRange[1] = m_slice->m_sps->quadtreeTULog2MaxSize;
 
-    if (log2CUSize < (m_slice->getSPS()->getQuadtreeTULog2MinSize() + quadtreeTUMaxDepth - 1 + interSplitFlag + intraSplitFlag))
-    {
-        // when fully making use of signaled TUMaxDepth + inter/intraSplitFlag, resulting luma TB size is < QuadtreeTULog2MinSize
-        log2MinTUSizeInCU = m_slice->getSPS()->getQuadtreeTULog2MinSize();
-    }
-    else
-    {
-        // when fully making use of signaled TUMaxDepth + inter/intraSplitFlag, resulting luma TB size is still >= QuadtreeTULog2MinSize
-        log2MinTUSizeInCU = log2CUSize - (quadtreeTUMaxDepth - 1 + interSplitFlag + intraSplitFlag); // stop when trafoDepth == hierarchy_depth = splitFlag
-        if (log2MinTUSizeInCU > m_slice->getSPS()->getQuadtreeTULog2MaxSize())
-        {
-            // when fully making use of signaled TUMaxDepth + inter/intraSplitFlag, resulting luma TB size is still > QuadtreeTULog2MaxSize
-            log2MinTUSizeInCU = m_slice->getSPS()->getQuadtreeTULog2MaxSize();
-        }
-    }
-    return log2MinTUSizeInCU;
+    tuDepthRange[0] = X265_MAX(tuDepthRange[0], X265_MIN(log2CUSize - (quadtreeTUMaxDepth - 1 + interSplitFlag + intraSplitFlag), tuDepthRange[1]));
 }
 
 uint32_t TComDataCU::getCtxSkipFlag(uint32_t absPartIdx)
 {
     TComDataCU* tempCU;
-    uint32_t        tempPartIdx;
-    uint32_t        ctx = 0;
+    uint32_t    tempPartIdx;
+    uint32_t    ctx = 0;
 
     // Get BCBP of left PU
     tempCU = getPULeft(tempPartIdx, m_absIdxInLCU + absPartIdx);
@@ -1356,80 +1212,68 @@
     return ctx;
 }
 
-uint32_t TComDataCU::getCtxInterDir(uint32_t absPartIdx)
+void TComDataCU::clearCbf(uint32_t absPartIdx, uint32_t depth)
 {
-    return getDepth(absPartIdx);
-}
-
-void TComDataCU::setCbfSubParts(uint32_t cbfY, uint32_t cbfU, uint32_t cbfV, uint32_t absPartIdx, uint32_t depth)
-{
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_cbf[0] + absPartIdx, cbfY, sizeof(UChar) * curPartNum);
-    memset(m_cbf[1] + absPartIdx, cbfU, sizeof(UChar) * curPartNum);
-    memset(m_cbf[2] + absPartIdx, cbfV, sizeof(UChar) * curPartNum);
+    memset(m_cbf[0] + absPartIdx, 0, sizeof(uint8_t) * curPartNum);
+    memset(m_cbf[1] + absPartIdx, 0, sizeof(uint8_t) * curPartNum);
+    memset(m_cbf[2] + absPartIdx, 0, sizeof(uint8_t) * curPartNum);
 }
 
 void TComDataCU::setCbfSubParts(uint32_t cbf, TextType ttype, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_cbf[ttype] + absPartIdx, cbf, sizeof(UChar) * curPartNum);
+    memset(m_cbf[ttype] + absPartIdx, cbf, sizeof(uint8_t) * curPartNum);
 }
 
-/** Sets a coded block flag for all sub-partitions of a partition
- * \param uiCbf The value of the coded block flag to be set
- * \param ttype
- * \param absPartIdx
- * \param partIdx
- * \param depth
- * \returns void
- */
-void TComDataCU::setCbfSubParts(uint32_t uiCbf, TextType ttype, uint32_t absPartIdx, uint32_t partIdx, uint32_t depth)
+void TComDataCU::setCbfPartRange(uint32_t cbf, TextType ttype, uint32_t absPartIdx, uint32_t coveredPartIdxes)
 {
-    setSubPart<UChar>(uiCbf, m_cbf[ttype], absPartIdx, depth, partIdx);
+    memset(m_cbf[ttype] + absPartIdx, cbf, sizeof(uint8_t) * coveredPartIdxes);
 }
 
 void TComDataCU::setDepthSubParts(uint32_t depth)
 {
     /*All 4x4 partitions in current CU have the CU depth saved*/
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
-    memset(m_depth, depth, sizeof(UChar) * curPartNum);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
+
+    memset(m_depth, depth, sizeof(uint8_t) * curPartNum);
 }
 
 bool TComDataCU::isFirstAbsZorderIdxInDepth(uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    return ((m_absIdxInLCU + absPartIdx) % curPartNum) == 0;
+    return ((m_absIdxInLCU + absPartIdx) & (curPartNum - 1)) == 0;
 }
 
 void TComDataCU::setPartSizeSubParts(PartSize mode, uint32_t absPartIdx, uint32_t depth)
 {
-    assert(sizeof(*m_partSizes) == 1);
-    memset(m_partSizes + absPartIdx, mode, m_pic->getNumPartInCU() >> (2 * depth));
+    X265_CHECK(sizeof(*m_partSizes) == 1, "size check failure\n");
+    memset(m_partSizes + absPartIdx, mode, NUM_CU_PARTITIONS >> (depth << 1));
 }
 
 void TComDataCU::setCUTransquantBypassSubParts(bool flag, uint32_t absPartIdx, uint32_t depth)
 {
-    memset(m_cuTransquantBypass + absPartIdx, flag, m_pic->getNumPartInCU() >> (2 * depth));
+    memset(m_cuTransquantBypass + absPartIdx, flag, NUM_CU_PARTITIONS >> (depth << 1));
 }
 
 void TComDataCU::setSkipFlagSubParts(bool skip, uint32_t absPartIdx, uint32_t depth)
 {
-    assert(sizeof(*m_skipFlag) == 1);
-    memset(m_skipFlag + absPartIdx, skip, m_pic->getNumPartInCU() >> (2 * depth));
+    X265_CHECK(sizeof(*m_skipFlag) == 1, "size check failure\n");
+    memset(m_skipFlag + absPartIdx, skip, NUM_CU_PARTITIONS >> (depth << 1));
 }
 
 void TComDataCU::setPredModeSubParts(PredMode eMode, uint32_t absPartIdx, uint32_t depth)
 {
-    assert(sizeof(*m_predModes) == 1);
-    memset(m_predModes + absPartIdx, eMode, m_pic->getNumPartInCU() >> (2 * depth));
+    X265_CHECK(sizeof(*m_predModes) == 1, "size check failure\n");
+    memset(m_predModes + absPartIdx, eMode, NUM_CU_PARTITIONS >> (depth << 1));
 }
 
 void TComDataCU::setQPSubCUs(int qp, TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool &foundNonZeroCbf)
 {
-    uint32_t curPartNumb = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNumb = NUM_CU_PARTITIONS >> (depth << 1);
     uint32_t curPartNumQ = curPartNumb >> 2;
 
     if (!foundNonZeroCbf)
@@ -1457,7 +1301,7 @@
 
 void TComDataCU::setQPSubParts(int qp, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
     for (uint32_t scuIdx = absPartIdx; scuIdx < absPartIdx + curPartNum; scuIdx++)
     {
@@ -1467,17 +1311,17 @@
 
 void TComDataCU::setLumaIntraDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_lumaIntraDir + absPartIdx, dir, sizeof(UChar) * curPartNum);
+    memset(m_lumaIntraDir + absPartIdx, dir, sizeof(uint8_t) * curPartNum);
 }
 
 template<typename T>
 void TComDataCU::setSubPart(T param, T* baseLCU, uint32_t cuAddr, uint32_t cuDepth, uint32_t puIdx)
 {
-    assert(sizeof(T) == 1); // Using memset() works only for types of size 1
+    X265_CHECK(sizeof(T) == 1, "size check failure\n"); // Using memset() works only for types of size 1
 
-    uint32_t curPartNumQ = (m_pic->getNumPartInCU() >> (2 * cuDepth)) >> 2;
+    uint32_t curPartNumQ = (NUM_CU_PARTITIONS >> (2 * cuDepth)) >> 2;
     switch (m_partSizes[cuAddr])
     {
     case SIZE_2Nx2N:
@@ -1506,7 +1350,7 @@
         }
         else
         {
-            assert(0);
+            X265_CHECK(0, "unexpected part unit index\n");
         }
         break;
     case SIZE_2NxnD:
@@ -1522,7 +1366,7 @@
         }
         else
         {
-            assert(0);
+            X265_CHECK(0, "unexpected part unit index\n");
         }
         break;
     case SIZE_nLx2N:
@@ -1542,7 +1386,7 @@
         }
         else
         {
-            assert(0);
+            X265_CHECK(0, "unexpected part unit index\n");
         }
         break;
     case SIZE_nRx2N:
@@ -1562,125 +1406,64 @@
         }
         else
         {
-            assert(0);
+            X265_CHECK(0, "unexpected part unit index\n");
         }
         break;
     default:
-        assert(0);
+        X265_CHECK(0, "unexpected part type\n");
     }
 }
 
 void TComDataCU::setChromIntraDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_chromaIntraDir + absPartIdx, dir, sizeof(UChar) * curPartNum);
+    memset(m_chromaIntraDir + absPartIdx, dir, sizeof(uint8_t) * curPartNum);
 }
 
 void TComDataCU::setInterDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t partIdx, uint32_t depth)
 {
-    setSubPart<UChar>(dir, m_interDir, absPartIdx, depth, partIdx);
+    setSubPart<uint8_t>(dir, m_interDir, absPartIdx, depth, partIdx);
 }
 
 void TComDataCU::setTrIdxSubParts(uint32_t trIdx, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_trIdx + absPartIdx, trIdx, sizeof(UChar) * curPartNum);
+    memset(m_trIdx + absPartIdx, trIdx, sizeof(uint8_t) * curPartNum);
 }
 
 void TComDataCU::setTransformSkipSubParts(uint32_t useTransformSkipY, uint32_t useTransformSkipU, uint32_t useTransformSkipV, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_transformSkip[0] + absPartIdx, useTransformSkipY, sizeof(UChar) * curPartNum);
-    memset(m_transformSkip[1] + absPartIdx, useTransformSkipU, sizeof(UChar) * curPartNum);
-    memset(m_transformSkip[2] + absPartIdx, useTransformSkipV, sizeof(UChar) * curPartNum);
+    memset(m_transformSkip[0] + absPartIdx, useTransformSkipY, sizeof(uint8_t) * curPartNum);
+    memset(m_transformSkip[1] + absPartIdx, useTransformSkipU, sizeof(uint8_t) * curPartNum);
+    memset(m_transformSkip[2] + absPartIdx, useTransformSkipV, sizeof(uint8_t) * curPartNum);
 }
 
 void TComDataCU::setTransformSkipSubParts(uint32_t useTransformSkip, TextType ttype, uint32_t absPartIdx, uint32_t depth)
 {
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
+    uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
 
-    memset(m_transformSkip[ttype] + absPartIdx, useTransformSkip, sizeof(UChar) * curPartNum);
+    memset(m_transformSkip[ttype] + absPartIdx, useTransformSkip, sizeof(uint8_t) * curPartNum);
 }
 
-UChar TComDataCU::getNumPartInter()
+void TComDataCU::setTransformSkipPartRange(uint32_t useTransformSkip, TextType ttype, uint32_t absPartIdx, uint32_t coveredPartIdxes)
 {
-    UChar numPart = 0;
-
-    switch (m_partSizes[0])
-    {
-    case SIZE_2Nx2N:    numPart = 1;
-        break;
-    case SIZE_2NxN:     numPart = 2;
-        break;
-    case SIZE_Nx2N:     numPart = 2;
-        break;
-    case SIZE_NxN:      numPart = 4;
-        break;
-    case SIZE_2NxnU:    numPart = 2;
-        break;
-    case SIZE_2NxnD:    numPart = 2;
-        break;
-    case SIZE_nLx2N:    numPart = 2;
-        break;
-    case SIZE_nRx2N:    numPart = 2;
-        break;
-    default:            assert(0);
-        break;
-    }
-
-    return numPart;
+    memset(m_transformSkip[ttype] + absPartIdx, useTransformSkip, sizeof(uint8_t) * coveredPartIdxes);
 }
 
 void TComDataCU::getPartIndexAndSize(uint32_t partIdx, uint32_t& outPartAddr, int& outWidth, int& outHeight)
 {
-    int cuSize = getCUSize(0);
-    switch (m_partSizes[0])
-    {
-    case SIZE_2NxN:
-        outWidth = cuSize;
-        outHeight = cuSize >> 1;
-        outPartAddr = (partIdx == 0) ? 0 : m_numPartitions >> 1;
-        break;
-    case SIZE_Nx2N:
-        outWidth = cuSize >> 1;
-        outHeight = cuSize;
-        outPartAddr = (partIdx == 0) ? 0 : m_numPartitions >> 2;
-        break;
-    case SIZE_NxN:
-        outWidth = cuSize >> 1;
-        outHeight = cuSize >> 1;
-        outPartAddr = (m_numPartitions >> 2) * partIdx;
-        break;
-    case SIZE_2NxnU:
-        outWidth     = cuSize;
-        outHeight    = (partIdx == 0) ?  cuSize >> 2 : (cuSize >> 2) + (cuSize >> 1);
-        outPartAddr = (partIdx == 0) ? 0 : m_numPartitions >> 3;
-        break;
-    case SIZE_2NxnD:
-        outWidth     = cuSize;
-        outHeight    = (partIdx == 0) ?  (cuSize >> 2) + (cuSize >> 1) : cuSize >> 2;
-        outPartAddr = (partIdx == 0) ? 0 : (m_numPartitions >> 1) + (m_numPartitions >> 3);
-        break;
-    case SIZE_nLx2N:
-        outWidth     = (partIdx == 0) ? cuSize >> 2 : (cuSize >> 2) + (cuSize >> 1);
-        outHeight    = cuSize;
-        outPartAddr = (partIdx == 0) ? 0 : m_numPartitions >> 4;
-        break;
-    case SIZE_nRx2N:
-        outWidth     = (partIdx == 0) ? (cuSize >> 2) + (cuSize >> 1) : cuSize >> 2;
-        outHeight    = cuSize;
-        outPartAddr = (partIdx == 0) ? 0 : (m_numPartitions >> 2) + (m_numPartitions >> 4);
-        break;
-    default:
-        assert(m_partSizes[0] == SIZE_2Nx2N);
-        outWidth = cuSize;
-        outHeight = cuSize;
-        outPartAddr = 0;
-        break;
-    }
+    int cuSize = 1 << getLog2CUSize(0);
+    int part_mode = m_partSizes[0];
+    int part_idx  = partIdx;
+
+    int tmp = partTable[part_mode][part_idx][0];
+    outWidth = ((tmp >> 4) * cuSize) >> 2;
+    outHeight = ((tmp & 0xF) * cuSize) >> 2;
+    outPartAddr = (partAddrTable[part_mode][part_idx] * m_numPartitions) >> 4;
 }
 
 void TComDataCU::getMvField(TComDataCU* cu, uint32_t absPartIdx, int picList, TComMvField& outMvField)
@@ -1696,119 +1479,10 @@
     outMvField.setMvField(cuMvField->getMv(absPartIdx), cuMvField->getRefIdx(absPartIdx));
 }
 
-void TComDataCU::deriveLeftRightTopIdxGeneral(uint32_t absPartIdx, uint32_t partIdx, uint32_t& outPartIdxLT, uint32_t& outPartIdxRT)
-{
-    outPartIdxLT = m_absIdxInLCU + absPartIdx;
-    uint32_t puWidth = 0;
-
-    switch (m_partSizes[absPartIdx])
-    {
-    case SIZE_2Nx2N: puWidth = m_cuSize[absPartIdx];
-        break;
-    case SIZE_2NxN:  puWidth = m_cuSize[absPartIdx];
-        break;
-    case SIZE_Nx2N:  puWidth = m_cuSize[absPartIdx]  >> 1;
-        break;
-    case SIZE_NxN:   puWidth = m_cuSize[absPartIdx]  >> 1;
-        break;
-    case SIZE_2NxnU: puWidth = m_cuSize[absPartIdx];
-        break;
-    case SIZE_2NxnD: puWidth = m_cuSize[absPartIdx];
-        break;
-    case SIZE_nLx2N:
-        if (partIdx == 0)
-        {
-            puWidth = m_cuSize[absPartIdx]  >> 2;
-        }
-        else if (partIdx == 1)
-        {
-            puWidth = (m_cuSize[absPartIdx]  >> 1) + (m_cuSize[absPartIdx]  >> 2);
-        }
-        else
-        {
-            assert(0);
-        }
-        break;
-    case SIZE_nRx2N:
-        if (partIdx == 0)
-        {
-            puWidth = (m_cuSize[absPartIdx]  >> 1) + (m_cuSize[absPartIdx]  >> 2);
-        }
-        else if (partIdx == 1)
-        {
-            puWidth = m_cuSize[absPartIdx]  >> 2;
-        }
-        else
-        {
-            assert(0);
-        }
-        break;
-    default:
-        assert(0);
-        break;
-    }
-
-    outPartIdxRT = g_rasterToZscan[g_zscanToRaster[outPartIdxLT] + puWidth / m_pic->getMinCUSize() - 1];
-}
-
-void TComDataCU::deriveLeftBottomIdxGeneral(uint32_t absPartIdx, uint32_t partIdx, uint32_t& outPartIdxLB)
-{
-    uint32_t puHeight = 0;
-
-    switch (m_partSizes[absPartIdx])
-    {
-    case SIZE_2Nx2N: puHeight = m_cuSize[absPartIdx];
-        break;
-    case SIZE_2NxN:  puHeight = m_cuSize[absPartIdx] >> 1;
-        break;
-    case SIZE_Nx2N:  puHeight = m_cuSize[absPartIdx];
-        break;
-    case SIZE_NxN:   puHeight = m_cuSize[absPartIdx] >> 1;
-        break;
-    case SIZE_2NxnU:
-        if (partIdx == 0)
-        {
-            puHeight = m_cuSize[absPartIdx] >> 2;
-        }
-        else if (partIdx == 1)
-        {
-            puHeight = (m_cuSize[absPartIdx] >> 1) + (m_cuSize[absPartIdx] >> 2);
-        }
-        else
-        {
-            assert(0);
-        }
-        break;
-    case SIZE_2NxnD:
-        if (partIdx == 0)
-        {
-            puHeight = (m_cuSize[absPartIdx] >> 1) + (m_cuSize[absPartIdx] >> 2);
-        }
-        else if (partIdx == 1)
-        {
-            puHeight = m_cuSize[absPartIdx] >> 2;
-        }
-        else
-        {
-            assert(0);
-        }
-        break;
-    case SIZE_nLx2N: puHeight = m_cuSize[absPartIdx];
-        break;
-    case SIZE_nRx2N: puHeight = m_cuSize[absPartIdx];
-        break;
-    default:
-        assert(0);
-        break;
-    }
-
-    outPartIdxLB = g_rasterToZscan[g_zscanToRaster[m_absIdxInLCU + absPartIdx] + ((puHeight / m_pic->getMinCUSize()) - 1) * m_pic->getNumPartInWidth()];
-}
-
 void TComDataCU::deriveLeftRightTopIdx(uint32_t partIdx, uint32_t& ruiPartIdxLT, uint32_t& ruiPartIdxRT)
 {
     ruiPartIdxLT = m_absIdxInLCU;
-    ruiPartIdxRT = g_rasterToZscan[g_zscanToRaster[ruiPartIdxLT] + m_cuSize[0] / m_pic->getMinCUSize() - 1];
+    ruiPartIdxRT = g_rasterToZscan[g_zscanToRaster[ruiPartIdxLT] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1];
 
     switch (m_partSizes[0])
     {
@@ -1842,14 +1516,14 @@
         ruiPartIdxRT -= (partIdx == 1) ? 0 : m_numPartitions >> 4;
         break;
     default:
-        assert(0);
+        X265_CHECK(0, "unexpected part index\n");
         break;
     }
 }
 
 void TComDataCU::deriveLeftBottomIdx(uint32_t partIdx, uint32_t& outPartIdxLB)
 {
-    outPartIdxLB = g_rasterToZscan[g_zscanToRaster[m_absIdxInLCU] + (((m_cuSize[0] / m_pic->getMinCUSize()) >> 1) - 1) * m_pic->getNumPartInWidth()];
+    outPartIdxLB = g_rasterToZscan[g_zscanToRaster[m_absIdxInLCU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE - 1)) - 1) * m_pic->getNumPartInCUSize()];
 
     switch (m_partSizes[0])
     {
@@ -1878,7 +1552,7 @@
         outPartIdxLB += (partIdx == 0) ? m_numPartitions >> 1 : (m_numPartitions >> 1) + (m_numPartitions >> 2) + (m_numPartitions >> 4);
         break;
     default:
-        assert(0);
+        X265_CHECK(0, "unexpected part index\n");
         break;
     }
 }
@@ -1890,8 +1564,9 @@
  */
 void TComDataCU::deriveRightBottomIdx(uint32_t partIdx, uint32_t& outPartIdxRB)
 {
-    outPartIdxRB = g_rasterToZscan[g_zscanToRaster[m_absIdxInLCU] + (((m_cuSize[0] / m_pic->getMinCUSize()) >> 1) - 1) *
-                                   m_pic->getNumPartInWidth() +  m_cuSize[0] / m_pic->getMinCUSize() - 1];
+    outPartIdxRB = g_rasterToZscan[g_zscanToRaster[m_absIdxInLCU] +
+                                   ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE - 1)) - 1) * m_pic->getNumPartInCUSize() +
+                                   (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1];
 
     switch (m_partSizes[0])
     {
@@ -1920,31 +1595,19 @@
         outPartIdxRB += (partIdx == 0) ? (m_numPartitions >> 2) + (m_numPartitions >> 3) + (m_numPartitions >> 4) : m_numPartitions >> 1;
         break;
     default:
-        assert(0);
+        X265_CHECK(0, "unexpected part index\n");
         break;
     }
 }
 
 void TComDataCU::deriveLeftRightTopIdxAdi(uint32_t& outPartIdxLT, uint32_t& outPartIdxRT, uint32_t partOffset, uint32_t partDepth)
 {
-    uint32_t numPartInWidth = (m_cuSize[0] / m_pic->getMinCUSize()) >> partDepth;
+    uint32_t numPartInWidth = 1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE - partDepth);
 
     outPartIdxLT = m_absIdxInLCU + partOffset;
     outPartIdxRT = g_rasterToZscan[g_zscanToRaster[outPartIdxLT] + numPartInWidth - 1];
 }
 
-void TComDataCU::deriveLeftBottomIdxAdi(uint32_t& outPartIdxLB, uint32_t partOffset, uint32_t partDepth)
-{
-    uint32_t absIdx;
-    uint32_t minCUWidth, widthInMinCUs;
-
-    minCUWidth    = getPic()->getMinCUSize();
-    widthInMinCUs = (getCUSize(0) / minCUWidth) >> partDepth;
-    absIdx        = getZorderIdxInCU() + partOffset + (m_numPartitions >> (partDepth << 1)) - 1;
-    absIdx        = g_zscanToRaster[absIdx] - (widthInMinCUs - 1);
-    outPartIdxLB    = g_rasterToZscan[absIdx];
-}
-
 bool TComDataCU::hasEqualMotion(uint32_t absPartIdx, TComDataCU* candCU, uint32_t candAbsPartIdx)
 {
     if (getInterDir(absPartIdx) != candCU->getInterDir(candAbsPartIdx))
@@ -1971,33 +1634,33 @@
  * \param absPartIdx
  * \param puIdx
  * \param depth
- * \param pcMvFieldNeighbours
- * \param puhInterDirNeighbours
- * \param numValidMergeCand
+ * \param mvFieldNeighbours
+ * \param interDirNeighbours
+ * \param maxNumMergeCand
  */
-void TComDataCU::getInterMergeCandidates(uint32_t absPartIdx, uint32_t puIdx, TComMvField* mvFieldNeighbours, UChar* interDirNeighbours,
-                                         int& numValidMergeCand)
+void TComDataCU::getInterMergeCandidates(uint32_t absPartIdx, uint32_t puIdx, TComMvField (*mvFieldNeighbours)[2], uint8_t* interDirNeighbours,
+                                         uint32_t& maxNumMergeCand)
 {
     uint32_t absPartAddr = m_absIdxInLCU + absPartIdx;
-    const uint32_t maxNumMergeCand = getSlice()->getMaxNumMergeCand();
-    const bool isInterB = getSlice()->isInterB();
+    const bool isInterB = m_slice->isInterB();
+
+    maxNumMergeCand = m_slice->m_maxNumMergeCand;
 
     for (uint32_t i = 0; i < maxNumMergeCand; ++i)
     {
-        mvFieldNeighbours[(i << 1)].refIdx = NOT_VALID;
-        mvFieldNeighbours[(i << 1) + 1].refIdx = NOT_VALID;
+        mvFieldNeighbours[i][0].refIdx = NOT_VALID;
+        mvFieldNeighbours[i][1].refIdx = NOT_VALID;
     }
 
-    numValidMergeCand = maxNumMergeCand;
     // compute the location of the current PU
     int xP, yP, nPSW, nPSH;
     this->getPartPosition(puIdx, xP, yP, nPSW, nPSH);
 
-    int count = 0;
+    uint32_t count = 0;
 
     uint32_t partIdxLT, partIdxRT, partIdxLB;
     PartSize curPS = getPartitionSize(absPartIdx);
-    deriveLeftBottomIdxGeneral(absPartIdx, puIdx, partIdxLB);
+    deriveLeftBottomIdx(puIdx, partIdxLB);
 
     //left
     uint32_t leftPartIdx = 0;
@@ -2012,20 +1675,17 @@
         // get Inter Dir
         interDirNeighbours[count] = cuLeft->getInterDir(leftPartIdx);
         // get Mv from Left
-        cuLeft->getMvField(cuLeft, leftPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count << 1]);
+        cuLeft->getMvField(cuLeft, leftPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count][0]);
         if (isInterB)
-        {
-            cuLeft->getMvField(cuLeft, leftPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[(count << 1) + 1]);
-        }
+            cuLeft->getMvField(cuLeft, leftPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[count][1]);
+
         count++;
-        // early termination
+    
         if (count == maxNumMergeCand)
-        {
             return;
-        }
     }
 
-    deriveLeftRightTopIdxGeneral(absPartIdx, puIdx, partIdxLT, partIdxRT);
+    deriveLeftRightTopIdx(puIdx, partIdxLT, partIdxRT);
 
     // above
     uint32_t abovePartIdx = 0;
@@ -2040,17 +1700,14 @@
         // get Inter Dir
         interDirNeighbours[count] = cuAbove->getInterDir(abovePartIdx);
         // get Mv from Left
-        cuAbove->getMvField(cuAbove, abovePartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count << 1]);
+        cuAbove->getMvField(cuAbove, abovePartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count][0]);
         if (isInterB)
-        {
-            cuAbove->getMvField(cuAbove, abovePartIdx, REF_PIC_LIST_1, mvFieldNeighbours[(count << 1) + 1]);
-        }
+            cuAbove->getMvField(cuAbove, abovePartIdx, REF_PIC_LIST_1, mvFieldNeighbours[count][1]);
+
         count++;
-        // early termination
+   
         if (count == maxNumMergeCand)
-        {
             return;
-        }
     }
 
     // above right
@@ -2065,20 +1722,17 @@
         // get Inter Dir
         interDirNeighbours[count] = cuAboveRight->getInterDir(aboveRightPartIdx);
         // get Mv from Left
-        cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count << 1]);
+        cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count][0]);
         if (isInterB)
-        {
-            cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[(count << 1) + 1]);
-        }
+            cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[count][1]);
+
         count++;
-        // early termination
+
         if (count == maxNumMergeCand)
-        {
             return;
-        }
     }
 
-    //left bottom
+    // left bottom
     uint32_t leftBottomPartIdx = 0;
     TComDataCU* cuLeftBottom = 0;
     cuLeftBottom = this->getPUBelowLeft(leftBottomPartIdx, partIdxLB);
@@ -2090,17 +1744,14 @@
         // get Inter Dir
         interDirNeighbours[count] = cuLeftBottom->getInterDir(leftBottomPartIdx);
         // get Mv from Left
-        cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count << 1]);
+        cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count][0]);
         if (isInterB)
-        {
-            cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[(count << 1) + 1]);
-        }
+            cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[count][1]);
+
         count++;
-        // early termination
+
         if (count == maxNumMergeCand)
-        {
             return;
-        }
     }
 
     // above left
@@ -2118,116 +1769,95 @@
             // get Inter Dir
             interDirNeighbours[count] = cuAboveLeft->getInterDir(aboveLeftPartIdx);
             // get Mv from Left
-            cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count << 1]);
+            cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, REF_PIC_LIST_0, mvFieldNeighbours[count][0]);
             if (isInterB)
-            {
-                cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[(count << 1) + 1]);
-            }
+                cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, REF_PIC_LIST_1, mvFieldNeighbours[count][1]);
+
             count++;
-            // early termination
+
             if (count == maxNumMergeCand)
-            {
                 return;
-            }
         }
     }
-    if (getSlice()->getEnableTMVPFlag())
+    // TMVP always enabled
     {
-        //>> MTK colocated-RightBottom
+        MV colmv;
         uint32_t partIdxRB;
 
         deriveRightBottomIdx(puIdx, partIdxRB);
 
-        uint32_t uiAbsPartIdxTmp = g_zscanToRaster[partIdxRB];
-        uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-
-        MV colmv;
-        int refIdx;
         int lcuIdx = -1;
 
-        if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_rasterToPelX[uiAbsPartIdxTmp] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicWidthInLumaSamples())  // image boundary check
-        {
-        }
-        else if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_rasterToPelY[uiAbsPartIdxTmp] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicHeightInLumaSamples())
-        {
-        }
-        else
-        {
-            if ((uiAbsPartIdxTmp % numPartInCUWidth < numPartInCUWidth - 1) &&        // is not at the last column of LCU
-                (uiAbsPartIdxTmp / numPartInCUWidth < m_pic->getNumPartInHeight() - 1)) // is not at the last row    of LCU
+        // image boundary check
+        if (m_pic->getCU(m_cuAddr)->getCUPelX() + g_zscanToPelX[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picWidthInLumaSamples &&
+            m_pic->getCU(m_cuAddr)->getCUPelY() + g_zscanToPelY[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picHeightInLumaSamples)
+        {
+            uint32_t absPartIdxRB = g_zscanToRaster[partIdxRB];
+            uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+            bool bNotLastCol = RasterAddress::lessThanCol(absPartIdxRB, numPartInCUSize - 1, numPartInCUSize); // is not at the last column of LCU
+            bool bNotLastRow = RasterAddress::lessThanRow(absPartIdxRB, numPartInCUSize - 1, numPartInCUSize); // is not at the last row    of LCU
+
+            if (bNotLastCol && bNotLastRow)
             {
-                absPartAddr = g_rasterToZscan[uiAbsPartIdxTmp + numPartInCUWidth + 1];
+                absPartAddr = g_rasterToZscan[absPartIdxRB + numPartInCUSize + 1];
                 lcuIdx = getAddr();
             }
-            else if (uiAbsPartIdxTmp % numPartInCUWidth < numPartInCUWidth - 1)       // is not at the last column of LCU But is last row of LCU
-            {
-                absPartAddr = g_rasterToZscan[(uiAbsPartIdxTmp + numPartInCUWidth + 1) % m_pic->getNumPartInCU()];
-            }
-            else if (uiAbsPartIdxTmp / numPartInCUWidth < m_pic->getNumPartInHeight() - 1) // is not at the last row of LCU But is last column of LCU
+            else if (bNotLastCol)
+                absPartAddr = g_rasterToZscan[(absPartIdxRB + numPartInCUSize + 1) & (numPartInCUSize - 1)];
+            else if (bNotLastRow)
             {
-                absPartAddr = g_rasterToZscan[uiAbsPartIdxTmp + 1];
+                absPartAddr = g_rasterToZscan[absPartIdxRB + 1];
                 lcuIdx = getAddr() + 1;
             }
-            else //is the right bottom corner of LCU
-            {
+            else // is the right bottom corner of LCU
                 absPartAddr = 0;
-            }
         }
 
-        refIdx = 0;
-        bool bExistMV = false;
+        int refIdx = 0;
         uint32_t partIdxCenter;
         uint32_t curLCUIdx = getAddr();
         int dir = 0;
-        uint32_t arrayAddr = count;
         xDeriveCenterIdx(puIdx, partIdxCenter);
-        bExistMV = lcuIdx >= 0 && xGetColMVP(REF_PIC_LIST_0, lcuIdx, absPartAddr, colmv, refIdx);
-        if (bExistMV == false)
-        {
+        bool bExistMV = lcuIdx >= 0 && xGetColMVP(REF_PIC_LIST_0, lcuIdx, absPartAddr, colmv, refIdx);
+        if (!bExistMV)
             bExistMV = xGetColMVP(REF_PIC_LIST_0, curLCUIdx, partIdxCenter, colmv, refIdx);
-        }
         if (bExistMV)
         {
             dir |= 1;
-            mvFieldNeighbours[2 * arrayAddr].setMvField(colmv, refIdx);
+            mvFieldNeighbours[count][0].setMvField(colmv, refIdx);
         }
 
         if (isInterB)
         {
             bExistMV = lcuIdx >= 0 && xGetColMVP(REF_PIC_LIST_1, lcuIdx, absPartAddr, colmv, refIdx);
-            if (bExistMV == false)
-            {
+            if (!bExistMV)
                 bExistMV = xGetColMVP(REF_PIC_LIST_1, curLCUIdx, partIdxCenter, colmv, refIdx);
-            }
+
             if (bExistMV)
             {
                 dir |= 2;
-                mvFieldNeighbours[2 * arrayAddr + 1].setMvField(colmv, refIdx);
+                mvFieldNeighbours[count][1].setMvField(colmv, refIdx);
             }
         }
 
         if (dir != 0)
         {
-            interDirNeighbours[arrayAddr] = dir;
+            interDirNeighbours[count] = dir;
 
             count++;
-            // early termination
+        
             if (count == maxNumMergeCand)
-            {
                 return;
-            }
         }
     }
 
-    uint32_t arrayAddr = count;
-
     if (isInterB)
     {
-        const int cutoff = count * (count - 1);
+        const uint32_t cutoff = count * (count - 1);
         uint32_t priorityList0 = 0xEDC984; // { 0, 1, 0, 2, 1, 2, 0, 3, 1, 3, 2, 3 }
         uint32_t priorityList1 = 0xB73621; // { 1, 0, 2, 0, 2, 1, 3, 0, 3, 1, 3, 2 }
 
-        for (int idx = 0; idx < cutoff; idx++)
+        for (uint32_t idx = 0; idx < cutoff; idx++)
         {
             int i = priorityList0 & 3;
             int j = priorityList1 & 3;
@@ -2237,44 +1867,42 @@
             if ((interDirNeighbours[i] & 0x1) && (interDirNeighbours[j] & 0x2))
             {
                 // get Mv from cand[i] and cand[j]
-                int refIdxL0 = mvFieldNeighbours[i << 1].refIdx;
-                int refIdxL1 = mvFieldNeighbours[(j << 1) + 1].refIdx;
-                int refPOCL0 = m_slice->getRefPOC(REF_PIC_LIST_0, refIdxL0);
-                int refPOCL1 = m_slice->getRefPOC(REF_PIC_LIST_1, refIdxL1);
-                if (!(refPOCL0 == refPOCL1 && mvFieldNeighbours[i << 1].mv == mvFieldNeighbours[(j << 1) + 1].mv))
+                int refIdxL0 = mvFieldNeighbours[i][0].refIdx;
+                int refIdxL1 = mvFieldNeighbours[j][1].refIdx;
+                int refPOCL0 = m_slice->m_refPOCList[0][refIdxL0];
+                int refPOCL1 = m_slice->m_refPOCList[1][refIdxL1];
+                if (!(refPOCL0 == refPOCL1 && mvFieldNeighbours[i][0].mv == mvFieldNeighbours[j][1].mv))
                 {
-                    mvFieldNeighbours[arrayAddr << 1].setMvField(mvFieldNeighbours[i << 1].mv, refIdxL0);
-                    mvFieldNeighbours[(arrayAddr << 1) + 1].setMvField(mvFieldNeighbours[(j << 1) + 1].mv, refIdxL1);
-                    interDirNeighbours[arrayAddr] = 3;
-
-                    arrayAddr++;
-                    // early termination
-                    if (arrayAddr == maxNumMergeCand)
-                    {
+                    mvFieldNeighbours[count][0].setMvField(mvFieldNeighbours[i][0].mv, refIdxL0);
+                    mvFieldNeighbours[count][1].setMvField(mvFieldNeighbours[j][1].mv, refIdxL1);
+                    interDirNeighbours[count] = 3;
+
+                    count++;
+
+                    if (count == maxNumMergeCand)
                         return;
-                    }
                 }
             }
         }
     }
-    int numRefIdx = (isInterB) ? X265_MIN(m_slice->getNumRefIdx(REF_PIC_LIST_0), m_slice->getNumRefIdx(REF_PIC_LIST_1)) : m_slice->getNumRefIdx(REF_PIC_LIST_0);
+    int numRefIdx = (isInterB) ? X265_MIN(m_slice->m_numRefIdx[0], m_slice->m_numRefIdx[1]) : m_slice->m_numRefIdx[0];
     int r = 0;
     int refcnt = 0;
-    while (arrayAddr < maxNumMergeCand)
+    while (count < maxNumMergeCand)
     {
-        interDirNeighbours[arrayAddr] = 1;
-        mvFieldNeighbours[arrayAddr << 1].setMvField(MV(0, 0), r);
+        interDirNeighbours[count] = 1;
+        mvFieldNeighbours[count][0].setMvField(MV(0, 0), r);
 
         if (isInterB)
         {
-            interDirNeighbours[arrayAddr] = 3;
-            mvFieldNeighbours[(arrayAddr << 1) + 1].setMvField(MV(0, 0), r);
+            interDirNeighbours[count] = 3;
+            mvFieldNeighbours[count][1].setMvField(MV(0, 0), r);
         }
-        arrayAddr++;
+
+        count++;
+
         if (refcnt == numRefIdx - 1)
-        {
             r = 0;
-        }
         else
         {
             ++r;
@@ -2290,16 +1918,12 @@
  */
 bool TComDataCU::isDiffMER(int xN, int yN, int xP, int yP)
 {
-    uint32_t plevel = this->getSlice()->getPPS()->getLog2ParallelMergeLevelMinus2() + 2;
+    uint32_t plevel = 2;
 
     if ((xN >> plevel) != (xP >> plevel))
-    {
         return true;
-    }
     if ((yN >> plevel) != (yP >> plevel))
-    {
         return true;
-    }
     return false;
 }
 
@@ -2311,63 +1935,17 @@
  */
 void TComDataCU::getPartPosition(uint32_t partIdx, int& xP, int& yP, int& nPSW, int& nPSH)
 {
-    uint32_t col = m_cuPelX;
-    uint32_t row = m_cuPelY;
-    uint32_t cuSize = getCUSize(0);
-
-    switch (m_partSizes[0])
-    {
-    case SIZE_2NxN:
-        nPSW = cuSize;
-        nPSH = cuSize >> 1;
-        xP   = col;
-        yP   = (partIdx == 0) ? row : row + nPSH;
-        break;
-    case SIZE_Nx2N:
-        nPSW = cuSize >> 1;
-        nPSH = cuSize;
-        xP   = (partIdx == 0) ? col : col + nPSW;
-        yP   = row;
-        break;
-    case SIZE_NxN:
-        nPSW = cuSize >> 1;
-        nPSH = cuSize >> 1;
-        xP   = col + (partIdx & 0x1) * nPSW;
-        yP   = row + (partIdx >> 1) * nPSH;
-        break;
-    case SIZE_2NxnU:
-        nPSW = cuSize;
-        nPSH = (partIdx == 0) ?  cuSize >> 2 : (cuSize >> 2) + (cuSize >> 1);
-        xP   = col;
-        yP   = (partIdx == 0) ? row : row + cuSize - nPSH;
-        break;
-    case SIZE_2NxnD:
-        nPSW = cuSize;
-        nPSH = (partIdx == 0) ?  (cuSize >> 2) + (cuSize >> 1) : cuSize >> 2;
-        xP   = col;
-        yP   = (partIdx == 0) ? row : row + cuSize - nPSH;
-        break;
-    case SIZE_nLx2N:
-        nPSW = (partIdx == 0) ? cuSize >> 2 : (cuSize >> 2) + (cuSize >> 1);
-        nPSH = cuSize;
-        xP   = (partIdx == 0) ? col : col + cuSize - nPSW;
-        yP   = row;
-        break;
-    case SIZE_nRx2N:
-        nPSW = (partIdx == 0) ? (cuSize >> 2) + (cuSize >> 1) : cuSize >> 2;
-        nPSH = cuSize;
-        xP   = (partIdx == 0) ? col : col + cuSize - nPSW;
-        yP   = row;
-        break;
-    default:
-        assert(m_partSizes[0] == SIZE_2Nx2N);
-        nPSW = cuSize;
-        nPSH = cuSize;
-        xP   = col;
-        yP   = row;
-
-        break;
-    }
+    int cuSize = 1 << getLog2CUSize(0);
+    int part_mode = m_partSizes[0];
+    int part_idx  = partIdx;
+
+    int tmp = partTable[part_mode][part_idx][0];
+    nPSW = ((tmp >> 4) * cuSize) >> 2;
+    nPSH = ((tmp & 0xF) * cuSize) >> 2;
+
+    tmp = partTable[part_mode][part_idx][1];
+    xP = ((tmp >> 4) * cuSize) >> 2;
+    yP = ((tmp & 0xF) * cuSize) >> 2;
 }
 
 /** Constructs a list of candidates for AMVP
@@ -2376,207 +1954,165 @@
  * \param picList
  * \param refIdx
  * \param info
+ * \param mvc
  */
-void TComDataCU::fillMvpCand(uint32_t partIdx, uint32_t partAddr, int picList, int refIdx, AMVPInfo* info)
+int TComDataCU::fillMvpCand(uint32_t partIdx, uint32_t partAddr, int picList, int refIdx, MV* amvpCand, MV* mvc)
 {
-    bool bAddedSmvp = false;
-
-    info->m_num = 0;
-    if (refIdx < 0)
-    {
-        return;
-    }
+    int num = 0;
 
     //-- Get Spatial MV
     uint32_t partIdxLT, partIdxRT, partIdxLB;
-    bool bAdded = false;
 
     deriveLeftRightTopIdx(partIdx, partIdxLT, partIdxRT);
     deriveLeftBottomIdx(partIdx, partIdxLB);
 
-    TComDataCU* tmpCU = NULL;
-    uint32_t idx;
-    tmpCU = getPUBelowLeft(idx, partIdxLB);
-    bAddedSmvp = (tmpCU != NULL) && (tmpCU->getPredictionMode(idx) != MODE_INTRA);
-
-    if (!bAddedSmvp)
-    {
-        tmpCU = getPULeft(idx, partIdxLB);
-        bAddedSmvp = (tmpCU != NULL) && (tmpCU->getPredictionMode(idx) != MODE_INTRA);
-    }
+    MV mv[MD_ABOVE_LEFT + 1];
+    MV mvOrder[MD_ABOVE_LEFT + 1];
+    bool valid[MD_ABOVE_LEFT + 1];
+    bool validOrder[MD_ABOVE_LEFT + 1];
+
+    valid[MD_BELOW_LEFT]  = xAddMVPCand(mv[MD_BELOW_LEFT], picList, refIdx, partIdxLB, MD_BELOW_LEFT);
+    valid[MD_LEFT]        = xAddMVPCand(mv[MD_LEFT], picList, refIdx, partIdxLB, MD_LEFT);
+    valid[MD_ABOVE_RIGHT] = xAddMVPCand(mv[MD_ABOVE_RIGHT], picList, refIdx, partIdxRT, MD_ABOVE_RIGHT);
+    valid[MD_ABOVE]       = xAddMVPCand(mv[MD_ABOVE], picList, refIdx, partIdxRT, MD_ABOVE);
+    valid[MD_ABOVE_LEFT]  = xAddMVPCand(mv[MD_ABOVE_LEFT], picList, refIdx, partIdxLT, MD_ABOVE_LEFT);
+
+    validOrder[MD_BELOW_LEFT]  = xAddMVPCandOrder(mvOrder[MD_BELOW_LEFT], picList, refIdx, partIdxLB, MD_BELOW_LEFT);
+    validOrder[MD_LEFT]        = xAddMVPCandOrder(mvOrder[MD_LEFT], picList, refIdx, partIdxLB, MD_LEFT);
+    validOrder[MD_ABOVE_RIGHT] = xAddMVPCandOrder(mvOrder[MD_ABOVE_RIGHT], picList, refIdx, partIdxRT, MD_ABOVE_RIGHT);
+    validOrder[MD_ABOVE]       = xAddMVPCandOrder(mvOrder[MD_ABOVE], picList, refIdx, partIdxRT, MD_ABOVE);
+    validOrder[MD_ABOVE_LEFT]  = xAddMVPCandOrder(mvOrder[MD_ABOVE_LEFT], picList, refIdx, partIdxLT, MD_ABOVE_LEFT);
 
     // Left predictor search
-    bAdded = xAddMVPCand(info, picList, refIdx, partIdxLB, MD_BELOW_LEFT);
-    if (!bAdded)
-    {
-        bAdded = xAddMVPCand(info, picList, refIdx, partIdxLB, MD_LEFT);
-    }
+    if (valid[MD_BELOW_LEFT])
+        amvpCand[num++] = mv[MD_BELOW_LEFT];
+    else if (valid[MD_LEFT])
+        amvpCand[num++] = mv[MD_LEFT];
+    else if (validOrder[MD_BELOW_LEFT])
+        amvpCand[num++] = mvOrder[MD_BELOW_LEFT];
+    else if (validOrder[MD_LEFT])
+        amvpCand[num++] = mvOrder[MD_LEFT];
+
+    bool bAddedSmvp = num > 0;
 
-    if (!bAdded)
-    {
-        bAdded = xAddMVPCandOrder(info, picList, refIdx, partIdxLB, MD_BELOW_LEFT);
-        if (!bAdded)
-        {
-            xAddMVPCandOrder(info, picList, refIdx, partIdxLB, MD_LEFT);
-        }
-    }
     // Above predictor search
-    bAdded = xAddMVPCand(info, picList, refIdx, partIdxRT, MD_ABOVE_RIGHT);
+    if (valid[MD_ABOVE_RIGHT])
+        amvpCand[num++] = mv[MD_ABOVE_RIGHT];
+    else if (valid[MD_ABOVE])
+        amvpCand[num++] = mv[MD_ABOVE];
+    else if (valid[MD_ABOVE_LEFT])
+        amvpCand[num++] = mv[MD_ABOVE_LEFT];
 
-    if (!bAdded)
+    if (!bAddedSmvp)
     {
-        bAdded = xAddMVPCand(info, picList, refIdx, partIdxRT, MD_ABOVE);
+        if (validOrder[MD_ABOVE_RIGHT])
+            amvpCand[num++] = mvOrder[MD_ABOVE_RIGHT];
+        else if (validOrder[MD_ABOVE])
+            amvpCand[num++] = mvOrder[MD_ABOVE];
+        else if (validOrder[MD_ABOVE_LEFT])
+            amvpCand[num++] = mvOrder[MD_ABOVE_LEFT];
     }
 
-    if (!bAdded)
+    int numMvc = 0;
+    for (int dir = MD_LEFT; dir <= MD_ABOVE_LEFT; dir++)
     {
-        xAddMVPCand(info, picList, refIdx, partIdxLT, MD_ABOVE_LEFT);
-    }
+        if (valid[dir] && mv[dir].notZero())
+            mvc[numMvc++] = mv[dir];
 
-    if (!bAddedSmvp)
-    {
-        bAdded = xAddMVPCandOrder(info, picList, refIdx, partIdxRT, MD_ABOVE_RIGHT);
-        if (!bAdded)
-        {
-            bAdded = xAddMVPCandOrder(info, picList, refIdx, partIdxRT, MD_ABOVE);
-        }
-
-        if (!bAdded)
-        {
-            xAddMVPCandOrder(info, picList, refIdx, partIdxLT, MD_ABOVE_LEFT);
-        }
+        if (validOrder[dir] && mvOrder[dir].notZero())
+            mvc[numMvc++] = mvOrder[dir];
     }
 
-    if (info->m_num == 2)
+    if (num == 2)
     {
-        if (info->m_mvCand[0] == info->m_mvCand[1])
-        {
-            info->m_num = 1;
-        }
+        if (amvpCand[0] == amvpCand[1])
+            num = 1;
         else
-        {
-            return;
-        }
+            /* AMVP_NUM_CANDS = 2 */
+            return numMvc;
     }
 
-    if (getSlice()->getEnableTMVPFlag())
+    // TMVP always enabled
     {
-        // Get Temporal Motion Predictor
-        int refIdxCol = refIdx;
-        MV   colmv;
+        uint32_t absPartAddr = m_absIdxInLCU + partAddr;
+        MV colmv;
         uint32_t partIdxRB;
-        uint32_t absPartIdx;
-        uint32_t absPartAddr;
 
         deriveRightBottomIdx(partIdx, partIdxRB);
-        absPartAddr = m_absIdxInLCU + partAddr;
 
         //----  co-located RightBottom Temporal Predictor (H) ---//
-        absPartIdx = g_zscanToRaster[partIdxRB];
         int lcuIdx = -1;
-        if ((m_pic->getCU(m_cuAddr)->getCUPelX() + g_rasterToPelX[absPartIdx] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicWidthInLumaSamples())  // image boundary check
-        {
-        }
-        else if ((m_pic->getCU(m_cuAddr)->getCUPelY() + g_rasterToPelY[absPartIdx] + m_pic->getMinCUSize()) >= m_slice->getSPS()->getPicHeightInLumaSamples())
-        {
-        }
-        else
-        {
-            uint32_t numPartInCUWidth = m_pic->getNumPartInWidth();
-            if ((absPartIdx % numPartInCUWidth < numPartInCUWidth - 1) &&        // is not at the last column of LCU
-                (absPartIdx / numPartInCUWidth < m_pic->getNumPartInHeight() - 1)) // is not at the last row    of LCU
+
+        // image boundary check
+        if (m_pic->getCU(m_cuAddr)->getCUPelX() + g_zscanToPelX[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picWidthInLumaSamples &&
+            m_pic->getCU(m_cuAddr)->getCUPelY() + g_zscanToPelY[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picHeightInLumaSamples)
+        {
+            uint32_t absPartIdxRB = g_zscanToRaster[partIdxRB];
+            uint32_t numPartInCUSize = m_pic->getNumPartInCUSize();
+            bool bNotLastCol = RasterAddress::lessThanCol(absPartIdxRB, numPartInCUSize - 1, numPartInCUSize); // is not at the last column of LCU
+            bool bNotLastRow = RasterAddress::lessThanRow(absPartIdxRB, numPartInCUSize - 1, numPartInCUSize); // is not at the last row    of LCU
+
+            if (bNotLastCol && bNotLastRow)
             {
-                absPartAddr = g_rasterToZscan[absPartIdx + numPartInCUWidth + 1];
+                absPartAddr = g_rasterToZscan[absPartIdxRB + numPartInCUSize + 1];
                 lcuIdx = getAddr();
             }
-            else if (absPartIdx % numPartInCUWidth < numPartInCUWidth - 1)       // is not at the last column of LCU But is last row of LCU
+            else if (bNotLastCol)
+                absPartAddr = g_rasterToZscan[(absPartIdxRB + numPartInCUSize + 1) & (numPartInCUSize - 1)];
+            else if (bNotLastRow)
             {
-                absPartAddr = g_rasterToZscan[(absPartIdx + numPartInCUWidth + 1) % m_pic->getNumPartInCU()];
-            }
-            else if (absPartIdx / numPartInCUWidth < m_pic->getNumPartInHeight() - 1) // is not at the last row of LCU But is last column of LCU
-            {
-                absPartAddr = g_rasterToZscan[absPartIdx + 1];
+                absPartAddr = g_rasterToZscan[absPartIdxRB + 1];
                 lcuIdx = getAddr() + 1;
             }
-            else //is the right bottom corner of LCU
-            {
+            else // is the right bottom corner of LCU
                 absPartAddr = 0;
-            }
         }
-        if (lcuIdx >= 0 && xGetColMVP(picList, lcuIdx, absPartAddr, colmv, refIdxCol))
+        if (lcuIdx >= 0 && xGetColMVP(picList, lcuIdx, absPartAddr, colmv, refIdx))
         {
-            info->m_mvCand[info->m_num++] = colmv;
+            amvpCand[num++] = colmv;
+            mvc[numMvc++] = colmv;
         }
         else
         {
             uint32_t partIdxCenter;
             uint32_t curLCUIdx = getAddr();
             xDeriveCenterIdx(partIdx, partIdxCenter);
-            if (xGetColMVP(picList, curLCUIdx, partIdxCenter, colmv, refIdxCol))
+            if (xGetColMVP(picList, curLCUIdx, partIdxCenter, colmv, refIdx))
             {
-                info->m_mvCand[info->m_num++] = colmv;
+                amvpCand[num++] = colmv;
+                mvc[numMvc++] = colmv;
             }
         }
         //----  co-located RightBottom Temporal Predictor  ---//
     }
 
-    while (info->m_num < AMVP_MAX_NUM_CANDS)
+    while (num < AMVP_NUM_CANDS)
     {
-        info->m_mvCand[info->m_num] = 0;
-        info->m_num++;
+        amvpCand[num++] = 0;
     }
-}
 
-bool TComDataCU::isBipredRestriction()
-{
-    return getCUSize(0) == 8 && getPartitionSize(0) != SIZE_2Nx2N;
+    return numMvc;
 }
 
 void TComDataCU::clipMv(MV& outMV)
 {
     int mvshift = 2;
     int offset = 8;
-    int xmax = (m_slice->getSPS()->getPicWidthInLumaSamples() + offset - m_cuPelX - 1) << mvshift;
+    int xmax = (m_slice->m_sps->picWidthInLumaSamples + offset - m_cuPelX - 1) << mvshift;
     int xmin = (-(int)g_maxCUSize - offset - (int)m_cuPelX + 1) << mvshift;
 
-    int ymax = (m_slice->getSPS()->getPicHeightInLumaSamples() + offset - m_cuPelY - 1) << mvshift;
+    int ymax = (m_slice->m_sps->picHeightInLumaSamples + offset - m_cuPelY - 1) << mvshift;
     int ymin = (-(int)g_maxCUSize - offset - (int)m_cuPelY + 1) << mvshift;
 
     outMV.x = X265_MIN(xmax, X265_MAX(xmin, (int)outMV.x));
     outMV.y = X265_MIN(ymax, X265_MAX(ymin, (int)outMV.y));
 }
 
-/** Set a I_PCM flag for all sub-partitions of a partition.
- * \param bIpcmFlag I_PCM flag
- * \param absPartIdx partition index
- * \param depth CU depth
- * \returns void
- */
-void TComDataCU::setIPCMFlagSubParts(bool bIpcmFlag, uint32_t absPartIdx, uint32_t depth)
-{
-    uint32_t curPartNum = m_pic->getNumPartInCU() >> (depth << 1);
-
-    memset(m_iPCMFlags + absPartIdx, bIpcmFlag, sizeof(bool) * curPartNum);
-}
-
-/** Test whether the current block is skipped
- * \param partIdx Block index
- * \returns Flag indicating whether the block is skipped
- */
-bool TComDataCU::isSkipped(uint32_t partIdx)
-{
-    return getSkipFlag(partIdx);
-}
-
 // ====================================================================================================================
 // Protected member functions
 // ====================================================================================================================
 
-#if _MSC_VER
-#pragma warning(disable: 4701) // potentially uninitialized local variables
-#endif
-
-bool TComDataCU::xAddMVPCand(AMVPInfo* info, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir)
+bool TComDataCU::xAddMVPCand(MV& mvp, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir)
 {
     TComDataCU* tmpCU = NULL;
     uint32_t idx = 0;
@@ -2584,68 +2120,51 @@
     switch (dir)
     {
     case MD_LEFT:
-    {
         tmpCU = getPULeft(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE:
-    {
         tmpCU = getPUAbove(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE_RIGHT:
-    {
         tmpCU = getPUAboveRight(idx, partUnitIdx);
         break;
-    }
     case MD_BELOW_LEFT:
-    {
         tmpCU = getPUBelowLeft(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE_LEFT:
-    {
         tmpCU = getPUAboveLeft(idx, partUnitIdx);
         break;
-    }
     default:
-    {
-        break;
-    }
+        return false;
     }
 
-    if (tmpCU == NULL)
-    {
+    if (!tmpCU)
         return false;
-    }
 
-    if (tmpCU->getCUMvField(picList)->getRefIdx(idx) >= 0 && m_slice->getRefPic(picList, refIdx)->getPOC() == tmpCU->getSlice()->getRefPOC(picList, tmpCU->getCUMvField(picList)->getRefIdx(idx)))
+    int refPOC = m_slice->m_refPOCList[picList][refIdx];
+    int otherPOC = tmpCU->m_slice->m_refPOCList[picList][tmpCU->getCUMvField(picList)->getRefIdx(idx)];
+    if (tmpCU->getCUMvField(picList)->getRefIdx(idx) >= 0 && refPOC == otherPOC)
     {
-        MV mvp = tmpCU->getCUMvField(picList)->getMv(idx);
-        info->m_mvCand[info->m_num++] = mvp;
+        mvp = tmpCU->getCUMvField(picList)->getMv(idx);
         return true;
     }
 
-    int refPicList2nd = REF_PIC_LIST_0;
-    if (picList == REF_PIC_LIST_0)
-    {
-        refPicList2nd = REF_PIC_LIST_1;
-    }
-    else if (picList == REF_PIC_LIST_1)
-    {
-        refPicList2nd = REF_PIC_LIST_0;
-    }
+    int refPicList2nd = 0;
+    if (picList == 0)
+        refPicList2nd = 1;
+    else if (picList == 1)
+        refPicList2nd = 0;
 
-    int curRefPOC = m_slice->getRefPic(picList, refIdx)->getPOC();
+    int curRefPOC = m_slice->m_refPOCList[picList][refIdx];
     int neibRefPOC;
 
     if (tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx) >= 0)
     {
-        neibRefPOC = tmpCU->getSlice()->getRefPOC(refPicList2nd, tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx));
-        if (neibRefPOC == curRefPOC) // Same Reference Frame But Diff List//
+        neibRefPOC = tmpCU->m_slice->m_refPOCList[refPicList2nd][tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx)];
+        if (neibRefPOC == curRefPOC)
         {
-            MV mvp = tmpCU->getCUMvField(refPicList2nd)->getMv(idx);
-            info->m_mvCand[info->m_num++] = mvp;
+            // Same reference frame but different list
+            mvp = tmpCU->getCUMvField(refPicList2nd)->getMv(idx);
             return true;
         }
     }
@@ -2660,7 +2179,7 @@
  * \param dir
  * \returns bool
  */
-bool TComDataCU::xAddMVPCandOrder(AMVPInfo* info, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir)
+bool TComDataCU::xAddMVPCandOrder(MV& outMV, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir)
 {
     TComDataCU* tmpCU = NULL;
     uint32_t idx = 0;
@@ -2668,120 +2187,68 @@
     switch (dir)
     {
     case MD_LEFT:
-    {
         tmpCU = getPULeft(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE:
-    {
         tmpCU = getPUAbove(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE_RIGHT:
-    {
         tmpCU = getPUAboveRight(idx, partUnitIdx);
         break;
-    }
     case MD_BELOW_LEFT:
-    {
         tmpCU = getPUBelowLeft(idx, partUnitIdx);
         break;
-    }
     case MD_ABOVE_LEFT:
-    {
         tmpCU = getPUAboveLeft(idx, partUnitIdx);
         break;
-    }
     default:
-    {
-        break;
-    }
+        return false;
     }
 
-    if (tmpCU == NULL)
-    {
+    if (!tmpCU)
         return false;
-    }
 
     int refPicList2nd = REF_PIC_LIST_0;
     if (picList == REF_PIC_LIST_0)
-    {
         refPicList2nd = REF_PIC_LIST_1;
-    }
     else if (picList == REF_PIC_LIST_1)
-    {
         refPicList2nd = REF_PIC_LIST_0;
-    }
 
-    int curPOC = m_slice->getPOC();
-    int curRefPOC = m_slice->getRefPic(picList, refIdx)->getPOC();
+    int curPOC = m_slice->m_poc;
+    int curRefPOC = m_slice->m_refPOCList[picList][refIdx];
     int neibPOC = curPOC;
     int neibRefPOC;
 
-    bool bIsCurrRefLongTerm = m_slice->getRefPic(picList, refIdx)->getIsLongTerm();
-    bool bIsNeibRefLongTerm = false;
-
     //---------------  V1 (END) ------------------//
     if (tmpCU->getCUMvField(picList)->getRefIdx(idx) >= 0)
     {
-        neibRefPOC = tmpCU->getSlice()->getRefPOC(picList, tmpCU->getCUMvField(picList)->getRefIdx(idx));
+        neibRefPOC = tmpCU->m_slice->m_refPOCList[picList][tmpCU->getCUMvField(picList)->getRefIdx(idx)];
         MV mvp = tmpCU->getCUMvField(picList)->getMv(idx);
-        MV outMV;
 
-        bIsNeibRefLongTerm = tmpCU->getSlice()->getRefPic(picList, tmpCU->getCUMvField(picList)->getRefIdx(idx))->getIsLongTerm();
-        if (bIsCurrRefLongTerm == bIsNeibRefLongTerm)
-        {
-            if (bIsCurrRefLongTerm || bIsNeibRefLongTerm)
-            {
-                outMV = mvp;
-            }
-            else
-            {
-                int iScale = xGetDistScaleFactor(curPOC, curRefPOC, neibPOC, neibRefPOC);
-                if (iScale == 4096)
-                {
-                    outMV = mvp;
-                }
-                else
-                {
-                    outMV = scaleMv(mvp, iScale);
-                }
-            }
-            info->m_mvCand[info->m_num++] = outMV;
-            return true;
-        }
+        int scale = xGetDistScaleFactor(curPOC, curRefPOC, neibPOC, neibRefPOC);
+        if (scale == 4096)
+            outMV = mvp;
+        else
+            outMV = scaleMv(mvp, scale);
+
+        return true;
     }
 
     //---------------------- V2(END) --------------------//
     if (tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx) >= 0)
     {
-        neibRefPOC = tmpCU->getSlice()->getRefPOC(refPicList2nd, tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx));
+        neibRefPOC = tmpCU->m_slice->m_refPOCList[refPicList2nd][tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx)];
         MV mvp = tmpCU->getCUMvField(refPicList2nd)->getMv(idx);
-        MV outMV;
 
-        bIsNeibRefLongTerm = tmpCU->getSlice()->getRefPic(refPicList2nd, tmpCU->getCUMvField(refPicList2nd)->getRefIdx(idx))->getIsLongTerm();
-        if (bIsCurrRefLongTerm == bIsNeibRefLongTerm)
-        {
-            if (bIsCurrRefLongTerm || bIsNeibRefLongTerm)
-            {
-                outMV = mvp;
-            }
-            else
-            {
-                int iScale = xGetDistScaleFactor(curPOC, curRefPOC, neibPOC, neibRefPOC);
-                if (iScale == 4096)
-                {
-                    outMV = mvp;
-                }
-                else
-                {
-                    outMV = scaleMv(mvp, iScale);
-                }
-            }
-            info->m_mvCand[info->m_num++] = outMV;
-            return true;
-        }
+        int scale = xGetDistScaleFactor(curPOC, curRefPOC, neibPOC, neibRefPOC);
+        if (scale == 4096)
+            outMV = mvp;
+        else
+            outMV = scaleMv(mvp, scale);
+
+        return true;
     }
+
     //---------------------- V3(END) --------------------//
     return false;
 }
@@ -2795,28 +2262,26 @@
  */
 bool TComDataCU::xGetColMVP(int picList, int cuAddr, int partUnitIdx, MV& outMV, int& outRefIdx)
 {
-    uint32_t absPartAddr = partUnitIdx & m_unitMask;
+    uint32_t absPartAddr = partUnitIdx & TMVP_UNIT_MASK;
 
     int colRefPicList;
     int colPOC, colRefPOC, curPOC, curRefPOC, scale;
     MV colmv;
 
     // use coldir.
-    TComPic *colPic = getSlice()->getRefPic(getSlice()->isInterB() ? 1 - getSlice()->getColFromL0Flag() : 0, getSlice()->getColRefIdx());
+    Frame *colPic = m_slice->m_refPicList[m_slice->isInterB() ? 1 - m_slice->m_colFromL0Flag : 0][m_slice->m_colRefIdx];
     TComDataCU *colCU = colPic->getCU(cuAddr);
 
-    if (colCU->getPic() == 0 || colCU->getPartitionSize(partUnitIdx) == SIZE_NONE)
-    {
+    if (colCU->m_pic == 0 || colCU->getPartitionSize(partUnitIdx) == SIZE_NONE)
         return false;
-    }
-    curPOC = m_slice->getPOC();
-    colPOC = colCU->getSlice()->getPOC();
+
+    curPOC = m_slice->m_poc;
+    colPOC = colCU->m_slice->m_poc;
 
     if (colCU->isIntra(absPartAddr))
-    {
         return false;
-    }
-    colRefPicList = getSlice()->getCheckLDC() ? picList : getSlice()->getColFromL0Flag();
+
+    colRefPicList = m_slice->m_bCheckLDC ? picList : m_slice->m_colFromL0Flag;
 
     int colRefIdx = colCU->getCUMvField(colRefPicList)->getRefIdx(absPartAddr);
 
@@ -2826,40 +2291,21 @@
         colRefIdx = colCU->getCUMvField(colRefPicList)->getRefIdx(absPartAddr);
 
         if (colRefIdx < 0)
-        {
             return false;
-        }
     }
 
     // Scale the vector.
-    colRefPOC = colCU->getSlice()->getRefPOC(colRefPicList, colRefIdx);
+    colRefPOC = colCU->m_slice->m_refPOCList[colRefPicList][colRefIdx];
     colmv = colCU->getCUMvField(colRefPicList)->getMv(absPartAddr);
 
-    curRefPOC = m_slice->getRefPic(picList, outRefIdx)->getPOC();
-    bool bIsCurrRefLongTerm = m_slice->getRefPic(picList, outRefIdx)->getIsLongTerm();
-    bool bIsColRefLongTerm = colCU->getSlice()->getIsUsedAsLongTerm(colRefPicList, colRefIdx);
+    curRefPOC = m_slice->m_refPOCList[picList][outRefIdx];
 
-    if (bIsCurrRefLongTerm != bIsColRefLongTerm)
-    {
-        return false;
-    }
-
-    if (bIsCurrRefLongTerm || bIsColRefLongTerm)
-    {
+    scale = xGetDistScaleFactor(curPOC, curRefPOC, colPOC, colRefPOC);
+    if (scale == 4096)
         outMV = colmv;
-    }
     else
-    {
-        scale = xGetDistScaleFactor(curPOC, curRefPOC, colPOC, colRefPOC);
-        if (scale == 4096)
-        {
-            outMV = colmv;
-        }
-        else
-        {
-            outMV = scaleMv(colmv, scale);
-        }
-    }
+        outMV = scaleMv(colmv, scale);
+
     return true;
 }
 
@@ -2891,59 +2337,74 @@
 void TComDataCU::xDeriveCenterIdx(uint32_t partIdx, uint32_t& outPartIdxCenter)
 {
     uint32_t partAddr;
-    int  partWidth;
-    int  partHeight;
+    int partWidth;
+    int partHeight;
 
     getPartIndexAndSize(partIdx, partAddr, partWidth, partHeight);
 
     outPartIdxCenter = m_absIdxInLCU + partAddr; // partition origin.
     outPartIdxCenter = g_rasterToZscan[g_zscanToRaster[outPartIdxCenter]
-                                       + (partHeight / m_pic->getMinCUSize()) / 2 * m_pic->getNumPartInWidth()
-                                       + (partWidth / m_pic->getMinCUSize()) / 2];
+                                       + (partHeight >> (LOG2_UNIT_SIZE + 1)) * m_pic->getNumPartInCUSize()
+                                       + (partWidth  >> (LOG2_UNIT_SIZE + 1))];
 }
 
-uint32_t TComDataCU::getCoefScanIdx(uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma, bool bIsIntra)
+ScanType TComDataCU::getCoefScanIdx(uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma, bool bIsIntra)
 {
-    uint32_t scanIdx;
     uint32_t dirMode;
 
     if (!bIsIntra)
-    {
         return SCAN_DIAG;
-    }
+
     //check that MDCS can be used for this TU
 
     if (bIsLuma)
     {
-        if (log2TrSize > MDCS_LOG2_MAX_SIZE) return SCAN_DIAG;
+        if (log2TrSize > MDCS_LOG2_MAX_SIZE)
+            return SCAN_DIAG;
 
         dirMode = getLumaIntraDir(absPartIdx);
     }
     else
     {
-        if (log2TrSize > (MDCS_LOG2_MAX_SIZE - m_hChromaShift)) return SCAN_DIAG;
+        if (log2TrSize > (MDCS_LOG2_MAX_SIZE - m_hChromaShift))
+            return SCAN_DIAG;
 
-        dirMode  = getChromaIntraDir(absPartIdx);
+        dirMode = getChromaIntraDir(absPartIdx);
         if (dirMode == DM_CHROMA_IDX)
         {
-            dirMode = getLumaIntraDir(absPartIdx);
+            dirMode = getLumaIntraDir((m_chromaFormat == X265_CSP_I444) ? absPartIdx : absPartIdx & 0xFC);
+            dirMode = (m_chromaFormat == X265_CSP_I422) ? g_chroma422IntraAngleMappingTable[dirMode] : dirMode;
         }
-        // TODO: 4:2:2
     }
 
     if (abs((int)dirMode - VER_IDX) <= MDCS_ANGLE_LIMIT)
-        scanIdx = SCAN_HOR;
+        return SCAN_HOR;
     else if (abs((int)dirMode - HOR_IDX) <= MDCS_ANGLE_LIMIT)
-        scanIdx = SCAN_VER;
+        return SCAN_VER;
     else
-        scanIdx = SCAN_DIAG;
-
-    return scanIdx;
+        return SCAN_DIAG;
 }
 
-uint32_t TComDataCU::getSCUAddr()
+void TComDataCU::getTUEntropyCodingParameters(TUEntropyCodingParameters &result, uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma)
 {
-    return (m_cuAddr) * (1 << (m_slice->getSPS()->getMaxCUDepth() << 1)) + m_absIdxInLCU;
+    // set the group layout
+    result.log2TrSizeCG = log2TrSize - 2;
+
+    // set the scan orders
+    result.scanType = getCoefScanIdx(absPartIdx, log2TrSize, bIsLuma, isIntra(absPartIdx));
+    result.scan     = g_scanOrder[result.scanType][log2TrSize - 2];
+    result.scanCG   = g_scanOrderCG[result.scanType][result.log2TrSizeCG];
+
+    if (log2TrSize == 2)
+        result.firstSignificanceMapContext = 0;
+    else if (log2TrSize == 3)
+    {
+        result.firstSignificanceMapContext = 9;
+        if (result.scanType != SCAN_DIAG && bIsLuma)
+            result.firstSignificanceMapContext += 6;
+    }
+    else
+        result.firstSignificanceMapContext = bIsLuma ? 21 : 12;
 }
 
 //! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComDataCU.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComDataCU.h Changed

@@ -39,17 +39,27 @@
 #ifndef X265_TCOMDATACU_H
 #define X265_TCOMDATACU_H
 
-#include "CommonDef.h"
+#include "common.h"
+#include "slice.h"
 #include "TComMotionInfo.h"
-#include "TComSlice.h"
-#include "TComRdCost.h"
 #include "TComPattern.h"
 
-#include <assert.h>
-
 namespace x265 {
 // private namespace
 
+// TU settings for entropy encoding
+struct TUEntropyCodingParameters
+{
+    const uint16_t *scan;
+    const uint16_t *scanCG;
+    ScanType        scanType;
+    uint32_t        log2TrSizeCG;
+    uint32_t        firstSignificanceMapContext;
+};
+
+class Frame;
+class Slice;
+
 //! \ingroup TLibCommon
 //! \{
 
@@ -71,6 +81,78 @@
     NUM_SGU_BORDER
 };
 
+struct DataCUMemPool
+{
+    char*    qpMemBlock;
+    uint8_t* depthMemBlock;
+    uint8_t* log2CUSizeMemBlock;
+    bool*    skipFlagMemBlock;
+    char*    partSizeMemBlock;
+    char*    predModeMemBlock;
+    bool*    cuTQBypassMemBlock;
+    bool*    mergeFlagMemBlock;
+    uint8_t* lumaIntraDirMemBlock;
+    uint8_t* chromaIntraDirMemBlock;
+    uint8_t* interDirMemBlock;
+    uint8_t* trIdxMemBlock;
+    uint8_t* transformSkipMemBlock;
+    uint8_t* cbfMemBlock;
+    uint8_t* mvpIdxMemBlock;
+    coeff_t* trCoeffMemBlock;
+    pixel*   m_tqBypassYuvMemBlock;
+};
+
+struct CU
+{
+    enum {
+        INTRA           = 1<<0, // CU is intra predicted
+        PRESENT         = 1<<1, // CU is not completely outside the frame
+        SPLIT_MANDATORY = 1<<2, // CU split is mandatory if CU is inside frame and can be splitted
+        LEAF            = 1<<3, // CU is a leaf node of the CTU
+        SPLIT           = 1<<4, // CU is currently split in four child CUs.
+    };
+    uint32_t log2CUSize; // Log of the CU size.
+    uint32_t childIdx;   // Index of the first child CU
+    uint32_t encodeIdx;  // Encoding index of this CU in terms of 8x8 blocks.
+    uint32_t offset[2];  // Offset of the luma CU in the X, Y direction in terms of pixels from the CTU origin
+    uint32_t flags;      // CU flags.
+};
+
+// Partition count table, index represents partitioning mode.
+const uint8_t nbPartsTable[8] = { 1, 2, 2, 4, 2, 2, 2, 2 };
+
+// Partition table.
+// First index is partitioning mode. Second index is partition index.
+// Third index is 0 for partition sizes, 1 for partition offsets. The 
+// sizes and offsets are encoded as two packed 4-bit values (X,Y). 
+// X and Y represent 1/4 fractions of the block size.
+const uint8_t partTable[8][4][2] =
+{
+//        XY
+    { { 0x44, 0x00 }, { 0x00, 0x00 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_2Nx2N.
+    { { 0x42, 0x00 }, { 0x42, 0x02 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_2NxN.
+    { { 0x24, 0x00 }, { 0x24, 0x20 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_Nx2N.
+    { { 0x22, 0x00 }, { 0x22, 0x20 }, { 0x22, 0x02 }, { 0x22, 0x22 } }, // SIZE_NxN.
+    { { 0x41, 0x00 }, { 0x43, 0x01 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_2NxnU.
+    { { 0x43, 0x00 }, { 0x41, 0x03 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_2NxnD.
+    { { 0x14, 0x00 }, { 0x34, 0x10 }, { 0x00, 0x00 }, { 0x00, 0x00 } }, // SIZE_nLx2N.
+    { { 0x34, 0x00 }, { 0x14, 0x30 }, { 0x00, 0x00 }, { 0x00, 0x00 } }  // SIZE_nRx2N.
+};
+
+// Partition Address table.
+// First index is partitioning mode. Second index is partition address.
+const uint8_t partAddrTable[8][4] =
+{
+    { 0x00, 0x00, 0x00, 0x00 }, // SIZE_2Nx2N.
+    { 0x00, 0x08, 0x08, 0x08 }, // SIZE_2NxN.
+    { 0x00, 0x04, 0x04, 0x04 }, // SIZE_Nx2N.
+    { 0x00, 0x04, 0x08, 0x0C }, // SIZE_NxN.
+    { 0x00, 0x02, 0x02, 0x02 }, // SIZE_2NxnU.
+    { 0x00, 0x0A, 0x0A, 0x0A }, // SIZE_2NxnD.
+    { 0x00, 0x01, 0x01, 0x01 }, // SIZE_nLx2N.
+    { 0x00, 0x05, 0x05, 0x05 }  // SIZE_nRx2N.
+};
+
 // ====================================================================================================================
 // Class definition
 // ====================================================================================================================
@@ -78,14 +160,14 @@
 /// CU data structure class
 class TComDataCU
 {
-private:
+public:
 
     // -------------------------------------------------------------------------------------------------------------------
     // class pointers
     // -------------------------------------------------------------------------------------------------------------------
 
-    TComPic*      m_pic;            ///< picture class pointer
-    TComSlice*    m_slice;          ///< slice header pointer
+    Frame*        m_pic;            ///< picture class pointer
+    Slice*        m_slice;          ///< slice header pointer
 
     // -------------------------------------------------------------------------------------------------------------------
     // CU description
@@ -96,13 +178,11 @@
     uint32_t      m_cuPelX;          ///< CU position in a pixel (X)
     uint32_t      m_cuPelY;          ///< CU position in a pixel (Y)
     uint32_t      m_numPartitions;   ///< total number of minimum partitions in a CU
-    uint8_t*      m_cuSize;          ///< array of cu width/height
-    UChar*        m_depth;           ///< array of depths
+    uint8_t*      m_log2CUSize;      ///< array of cu width/height
+    uint8_t*      m_depth;           ///< array of depths
     int           m_chromaFormat;
     int           m_hChromaShift;
     int           m_vChromaShift;
-    int           m_unitSize;        ///< size of a "minimum partition"
-    uint32_t      m_unitMask;        ///< mask for mapping index to CompressMV field
 
     // -------------------------------------------------------------------------------------------------------------------
     // CU data
@@ -112,17 +192,12 @@
     char*         m_predModes;          ///< array of prediction modes
     bool*         m_cuTransquantBypass; ///< array of cu_transquant_bypass flags
     char*         m_qp;                 ///< array of QP values
-    UChar*        m_trIdx;              ///< array of transform indices
-    UChar*        m_transformSkip[3];   ///< array of transform skipping flags
-    UChar*        m_cbf[3];             ///< array of coded block flags (CBF)
+    uint8_t*      m_trIdx;              ///< array of transform indices
+    uint8_t*      m_transformSkip[3];   ///< array of transform skipping flags
+    uint8_t*      m_cbf[3];             ///< array of coded block flags (CBF)
     TComCUMvField m_cuMvField[2];       ///< array of motion vectors
-    TCoeff*       m_trCoeffY;           ///< transformed coefficient buffer (Y)
-    TCoeff*       m_trCoeffCb;          ///< transformed coefficient buffer (Cb)
-    TCoeff*       m_trCoeffCr;          ///< transformed coefficient buffer (Cr)
-
-    Pel*          m_iPCMSampleY;        ///< PCM sample buffer (Y)
-    Pel*          m_iPCMSampleCb;       ///< PCM sample buffer (Cb)
-    Pel*          m_iPCMSampleCr;       ///< PCM sample buffer (Cr)
+    coeff_t*      m_trCoeff[3];         ///< transformed coefficient buffer
+    pixel*        m_tqBypassOrigYuv[3]; ///< Original Lossless YUV buffer (Y/Cb/Cr)
 
     // -------------------------------------------------------------------------------------------------------------------
     // neighbor access variables
@@ -132,92 +207,99 @@
     TComDataCU*   m_cuAboveRight;    ///< pointer of above-right CU
     TComDataCU*   m_cuAbove;         ///< pointer of above CU
     TComDataCU*   m_cuLeft;          ///< pointer of left CU
-    TComDataCU*   m_cuColocated[2];  ///< pointer of temporally colocated CU's for both directions
 
     // -------------------------------------------------------------------------------------------------------------------
     // coding tool information
     // -------------------------------------------------------------------------------------------------------------------
 
     bool*         m_bMergeFlags;      ///< array of merge flags
-    UChar*        m_lumaIntraDir;     ///< array of intra directions (luma)
-    UChar*        m_chromaIntraDir;   ///< array of intra directions (chroma)
-    UChar*        m_interDir;         ///< array of inter directions
+    uint8_t*      m_lumaIntraDir;     ///< array of intra directions (luma)
+    uint8_t*      m_chromaIntraDir;   ///< array of intra directions (chroma)
+    uint8_t*      m_interDir;         ///< array of inter directions
     uint8_t*      m_mvpIdx[2];        ///< array of motion vector predictor candidates or merge candidate indices [0]
-    bool*         m_iPCMFlags;        ///< array of intra_pcm flags
 
     // -------------------------------------------------------------------------------------------------------------------
     // misc. variables
     // -------------------------------------------------------------------------------------------------------------------
 
-protected:
+    DataCUMemPool m_DataCUMemPool;
+    TComCUMvField m_cuMvFieldMemPool;
 
+    // CU data. Index is the CU index. Neighbour CUs (top-left, top, top-right, left) are appended to the end,
+    // required for prediction of current CU.
+    // (1 + 4 + 16 + 64) + (1 + 8 + 1 + 8 + 1) = 104.
+    CU m_CULocalData[104]; 
+
+protected:
     /// add possible motion vector predictor candidates
-    bool          xAddMVPCand(AMVPInfo* info, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir);
+    bool xAddMVPCand(MV& mvp, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir);
 
-    bool          xAddMVPCandOrder(AMVPInfo* info, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir);
+    bool xAddMVPCandOrder(MV& mvp, int picList, int refIdx, uint32_t partUnitIdx, MVP_DIR dir);
 
-    void          deriveRightBottomIdx(uint32_t partIdx, uint32_t& outPartIdxRB);
+    void deriveRightBottomIdx(uint32_t partIdx, uint32_t& outPartIdxRB);
 
-    bool          xGetColMVP(int picList, int cuAddr, int partUnitIdx, MV& outMV, int& outRefIdx);
+    bool xGetColMVP(int picList, int cuAddr, int partUnitIdx, MV& outMV, int& outRefIdx);
 
     /// compute scaling factor from POC difference
-    int           xGetDistScaleFactor(int curPOC, int curRefPOC, int colPOC, int colRefPOC);
+    int  xGetDistScaleFactor(int curPOC, int curRefPOC, int colPOC, int colRefPOC);
 
     void xDeriveCenterIdx(uint32_t partIdx, uint32_t& outPartIdxCenter);
 
 public:
 
     TComDataCU();
-    virtual ~TComDataCU();
+    ~TComDataCU() {}
 
-    uint64_t      m_totalCost;       ///< sum of partition RD costs
+    uint32_t      m_psyEnergy;
+    uint64_t      m_totalPsyCost;
+    uint64_t      m_totalRDCost;       ///< sum of partition RD costs
     uint32_t      m_totalDistortion; ///< sum of partition distortion
     uint32_t      m_totalBits;       ///< sum of partition signal bits
     uint64_t      m_avgCost[4];      // stores the avg cost of CU's in frame for each depth
     uint32_t      m_count[4];
     uint64_t      m_sa8dCost;
     double        m_baseQp;          //Qp of Cu set from RateControl/Vbv.
+    uint32_t      m_mvBits;         // Mv bits + Ref + block type
+    uint32_t      m_coeffBits;        // Texture bits (DCT Coeffs)
+
     // -------------------------------------------------------------------------------------------------------------------
     // create / destroy / initialize / copy
     // -------------------------------------------------------------------------------------------------------------------
+    void          create(TComDataCU *p, uint32_t numPartition, uint32_t cuSize, int csp, int index, bool isLossLess);
+
+    bool          initialize(uint32_t numPartition, uint32_t sizeL, uint32_t sizeC, uint32_t numBlocks, bool isLossless);
 
-    bool          create(uint32_t numPartition, uint32_t width, uint32_t height, int unitSize, int csp);
     void          destroy();
 
-    void          initCU(TComPic* pic, uint32_t cuAddr);
-    void          initEstData(uint32_t depth);
-    void          initEstData(uint32_t depth, int qp);
-    void          initSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth);
+    void          initCU(Frame* pic, uint32_t cuAddr);
+    void          initEstData();
     void          initSubCU(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth, int qp);
 
     void          copyToSubCU(TComDataCU* lcu, uint32_t partUnitIdx, uint32_t depth);
     void          copyPartFrom(TComDataCU* cu, uint32_t partUnitIdx, uint32_t depth, bool isRDObasedAnalysis = true);
 
-    void          copyToPic(UChar depth);
-    void          copyToPic(UChar depth, uint32_t partIdx, uint32_t partDepth);
-    void          copyCodedToPic(UChar depth);
+    void          copyToPic(uint32_t depth);
+    void          copyToPic(uint32_t depth, uint32_t partIdx, uint32_t partDepth);
+    void          copyCodedToPic(uint32_t depth);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for CU description
     // -------------------------------------------------------------------------------------------------------------------
 
-    TComPic*      getPic()                         { return m_pic; }
-
-    TComSlice*    getSlice()                       { return m_slice; }
-
     uint32_t&     getAddr()                        { return m_cuAddr; }
 
     uint32_t&     getZorderIdxInCU()               { return m_absIdxInLCU; }
 
-    uint32_t      getSCUAddr();
+    uint32_t      getSCUAddr() const               { return (m_cuAddr << g_maxFullDepth * 2) + m_absIdxInLCU; }
+
 
     uint32_t      getCUPelX()                      { return m_cuPelX; }
 
     uint32_t      getCUPelY()                      { return m_cuPelY; }
 
-    UChar*        getDepth()                       { return m_depth; }
+    uint8_t*      getDepth()                       { return m_depth; }
 
-    UChar         getDepth(uint32_t idx)           { return m_depth[idx]; }
+    uint8_t       getDepth(uint32_t idx)           { return m_depth[idx]; }
 
     void          setDepthSubParts(uint32_t depth);
 
@@ -227,86 +309,87 @@
 
     char*         getPartitionSize()                      { return m_partSizes; }
 
-    PartSize      getPartitionSize(uint32_t idx)              { return static_cast<PartSize>(m_partSizes[idx]); }
+    PartSize      getPartitionSize(uint32_t idx)          { return static_cast<PartSize>(m_partSizes[idx]); }
 
     void          setPartSizeSubParts(PartSize eMode, uint32_t absPartIdx, uint32_t depth);
     void          setCUTransquantBypassSubParts(bool flag, uint32_t absPartIdx, uint32_t depth);
 
-    bool*        getSkipFlag()                        { return m_skipFlag; }
+    bool*         getSkipFlag()                        { return m_skipFlag; }
 
-    bool         getSkipFlag(uint32_t idx)            { return m_skipFlag[idx]; }
+    bool          getSkipFlag(uint32_t idx)            { return m_skipFlag[idx]; }
 
-    void         setSkipFlagSubParts(bool skip, uint32_t absPartIdx, uint32_t depth);
+    void          setSkipFlagSubParts(bool skip, uint32_t absPartIdx, uint32_t depth);
 
     char*         getPredictionMode()                 { return m_predModes; }
 
-    PredMode      getPredictionMode(uint32_t idx)         { return static_cast<PredMode>(m_predModes[idx]); }
+    PredMode      getPredictionMode(uint32_t idx)     { return static_cast<PredMode>(m_predModes[idx]); }
 
     bool*         getCUTransquantBypass()             { return m_cuTransquantBypass; }
 
-    bool          getCUTransquantBypass(uint32_t idx)     { return m_cuTransquantBypass[idx]; }
+    bool          getCUTransquantBypass(uint32_t idx) { return m_cuTransquantBypass[idx]; }
 
     void          setPredModeSubParts(PredMode eMode, uint32_t absPartIdx, uint32_t depth);
 
-    uint8_t*      getCUSize()                     { return m_cuSize; }
+    uint8_t*      getLog2CUSize()                     { return m_log2CUSize; }
 
-    uint8_t       getCUSize(uint32_t idx)            { return m_cuSize[idx]; }
+    uint8_t       getLog2CUSize(uint32_t idx) const   { return m_log2CUSize[idx]; }
 
-    char*         getQP()                        { return m_qp; }
+    char*         getQP()                         { return m_qp; }
 
-    char          getQP(uint32_t idx)                { return m_qp[idx]; }
+    char          getQP(uint32_t idx)             { return m_qp[idx]; }
 
-    void          setQP(uint32_t idx, char value)    { m_qp[idx] =  value; }
+    void          setQP(uint32_t idx, char value) { m_qp[idx] =  value; }
 
     void          setQPSubParts(int qp,   uint32_t absPartIdx, uint32_t depth);
     int           getLastValidPartIdx(int absPartIdx);
     char          getLastCodedQP(uint32_t absPartIdx);
     void          setQPSubCUs(int qp, TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool &foundNonZeroCbf);
 
-    bool          isLosslessCoded(uint32_t absPartIdx);
+    bool          isLosslessCoded(uint32_t idx) const { return m_cuTransquantBypass[idx] && m_slice->m_pps->bTransquantBypassEnabled; }
 
-    UChar*        getTransformIdx()                    { return m_trIdx; }
+    uint8_t*      getTransformIdx()                    { return m_trIdx; }
 
-    UChar         getTransformIdx(uint32_t idx)            { return m_trIdx[idx]; }
+    uint8_t       getTransformIdx(uint32_t idx)        { return m_trIdx[idx]; }
 
     void          setTrIdxSubParts(uint32_t uiTrIdx, uint32_t absPartIdx, uint32_t depth);
 
-    UChar*        getTransformSkip(TextType ttype)     { return m_transformSkip[ttype]; }
+    uint8_t*      getTransformSkip(TextType ttype)     { return m_transformSkip[ttype]; }
 
-    UChar         getTransformSkip(uint32_t idx, TextType ttype)    { return m_transformSkip[ttype][idx]; }
+    uint8_t       getTransformSkip(uint32_t idx, TextType ttype)    { return m_transformSkip[ttype][idx]; }
 
     void          setTransformSkipSubParts(uint32_t useTransformSkip, TextType ttype, uint32_t absPartIdx, uint32_t depth);
     void          setTransformSkipSubParts(uint32_t useTransformSkipY, uint32_t useTransformSkipU, uint32_t useTransformSkipV, uint32_t absPartIdx, uint32_t depth);
 
-    uint32_t          getQuadtreeTULog2MinSizeInCU(uint32_t absPartIdx);
+    void          getQuadtreeTULog2MinSizeInCU(uint32_t tuDepthRange[2], uint32_t absPartIdx);
 
     TComCUMvField* getCUMvField(int e)        { return &m_cuMvField[e]; }
 
-    TCoeff*&      getCoeffY()                 { return m_trCoeffY; }
+    coeff_t*      getCoeffY()                 { return m_trCoeff[0]; }
 
-    TCoeff*&      getCoeffCb()                { return m_trCoeffCb; }
+    coeff_t*      getCoeffCb()                { return m_trCoeff[1]; }
 
-    TCoeff*&      getCoeffCr()                { return m_trCoeffCr; }
+    coeff_t*      getCoeffCr()                { return m_trCoeff[2]; }
 
-    Pel*&         getPCMSampleY()             { return m_iPCMSampleY; }
+    coeff_t*      getCoeff(TextType ttype)    { return m_trCoeff[ttype]; }
 
-    Pel*&         getPCMSampleCb()            { return m_iPCMSampleCb; }
+    pixel*&       getLumaOrigYuv()             { return m_tqBypassOrigYuv[0]; }
 
-    Pel*&         getPCMSampleCr()            { return m_iPCMSampleCr; }
+    pixel*&       getChromaOrigYuv(uint32_t chromaId) { return m_tqBypassOrigYuv[chromaId]; }
 
-    UChar         getCbf(uint32_t idx, TextType ttype) { return m_cbf[ttype][idx]; }
+    uint8_t       getCbf(uint32_t idx, TextType ttype) { return m_cbf[ttype][idx]; }
 
-    UChar*        getCbf(TextType ttype)           { return m_cbf[ttype]; }
+    uint8_t*      getCbf(TextType ttype)      { return m_cbf[ttype]; }
 
-    UChar         getCbf(uint32_t idx, TextType ttype, uint32_t trDepth) { return (getCbf(idx, ttype) >> trDepth) & 0x1; }
+    uint8_t       getCbf(uint32_t idx, TextType ttype, uint32_t trDepth) { return (getCbf(idx, ttype) >> trDepth) & 0x1; }
 
-    void          setCbf(uint32_t idx, TextType ttype, UChar uh)     { m_cbf[ttype][idx] = uh; }
+    void          setCbf(uint32_t idx, TextType ttype, uint8_t uh)       { m_cbf[ttype][idx] = uh; }
 
-    UChar         getQtRootCbf(uint32_t idx)           { return getCbf(idx, TEXT_LUMA) || getCbf(idx, TEXT_CHROMA_U) || getCbf(idx, TEXT_CHROMA_V); }
+    uint8_t       getQtRootCbf(uint32_t idx)  { return getCbf(idx, TEXT_LUMA) || getCbf(idx, TEXT_CHROMA_U) || getCbf(idx, TEXT_CHROMA_V); }
 
-    void          setCbfSubParts(uint32_t cbfY, uint32_t cbfU, uint32_t cbfV, uint32_t absPartIdx, uint32_t depth);
+    void          clearCbf(uint32_t absPartIdx, uint32_t depth);
     void          setCbfSubParts(uint32_t cbf, TextType ttype, uint32_t absPartIdx, uint32_t depth);
-    void          setCbfSubParts(uint32_t cbf, TextType ttype, uint32_t absPartIdx, uint32_t partIdx, uint32_t depth);
+    void          setCbfPartRange(uint32_t cbf, TextType ttype, uint32_t absPartIdx, uint32_t coveredPartIdxes);
+    void          setTransformSkipPartRange(uint32_t useTransformSkip, TextType ttype, uint32_t absPartIdx, uint32_t coveredPartIdxes);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for coding tool information
@@ -314,50 +397,43 @@
 
     bool*         getMergeFlag()                    { return m_bMergeFlags; }
 
-    bool          getMergeFlag(uint32_t idx)            { return m_bMergeFlags[idx]; }
+    bool          getMergeFlag(uint32_t idx)        { return m_bMergeFlags[idx]; }
 
     void          setMergeFlag(uint32_t idx, bool bMergeFlag) { m_bMergeFlags[idx] = bMergeFlag; }
 
     uint8_t*      getMergeIndex()                   { return m_mvpIdx[0]; }
 
-    uint8_t       getMergeIndex(uint32_t idx)           { return m_mvpIdx[0][idx]; }
+    uint8_t       getMergeIndex(uint32_t idx)       { return m_mvpIdx[0][idx]; }
 
     void          setMergeIndex(uint32_t idx, int mergeIndex) { m_mvpIdx[0][idx] = (uint8_t)mergeIndex; }
 
     template<typename T>
     void          setSubPart(T bParameter, T* pbBaseLCU, uint32_t cuAddr, uint32_t cuDepth, uint32_t puIdx);
 
-    UChar*        getLumaIntraDir()         { return m_lumaIntraDir; }
+    uint8_t*      getLumaIntraDir()         { return m_lumaIntraDir; }
 
-    UChar         getLumaIntraDir(uint32_t idx) { return m_lumaIntraDir[idx]; }
+    uint8_t       getLumaIntraDir(uint32_t idx) { return m_lumaIntraDir[idx]; }
 
     void          setLumaIntraDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t depth);
 
-    UChar*        getChromaIntraDir()                 { return m_chromaIntraDir; }
+    uint8_t*      getChromaIntraDir()       { return m_chromaIntraDir; }
 
-    UChar         getChromaIntraDir(uint32_t idx)         { return m_chromaIntraDir[idx]; }
+    uint8_t       getChromaIntraDir(uint32_t idx) { return m_chromaIntraDir[idx]; }
 
     void          setChromIntraDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t depth);
 
-    UChar*        getInterDir()                    { return m_interDir; }
-
-    UChar         getInterDir(uint32_t idx)            { return m_interDir[idx]; }
-
-    void          setInterDirSubParts(uint32_t dir,  uint32_t absPartIdx, uint32_t partIdx, uint32_t depth);
-    bool*         getIPCMFlag()                     { return m_iPCMFlags; }
-
-    bool          getIPCMFlag(uint32_t idx)             { return m_iPCMFlags[idx]; }
+    uint8_t*      getInterDir()             { return m_interDir; }
 
-    void          setIPCMFlag(uint32_t idx, bool b)     { m_iPCMFlags[idx] = b; }
+    uint8_t       getInterDir(uint32_t idx) { return m_interDir[idx]; }
 
-    void          setIPCMFlagSubParts(bool bIpcmFlag, uint32_t absPartIdx, uint32_t depth);
+    void          setInterDirSubParts(uint32_t dir, uint32_t absPartIdx, uint32_t partIdx, uint32_t depth);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for accessing partition information
     // -------------------------------------------------------------------------------------------------------------------
 
     void          getPartIndexAndSize(uint32_t partIdx, uint32_t& ruiPartAddr, int& riWidth, int& riHeight);
-    UChar         getNumPartInter();
+    uint8_t       getNumPartInter() { return nbPartsTable[(int)m_partSizes[0]]; }
     bool          isFirstAbsZorderIdxInDepth(uint32_t absPartIdx, uint32_t depth);
 
     // -------------------------------------------------------------------------------------------------------------------
@@ -366,12 +442,12 @@
 
     void          getMvField(TComDataCU* cu, uint32_t absPartIdx, int picList, TComMvField& rcMvField);
 
-    void          fillMvpCand(uint32_t partIdx, uint32_t partAddr, int picList, int refIdx, AMVPInfo* info);
+    int           fillMvpCand(uint32_t partIdx, uint32_t partAddr, int picList, int refIdx, MV* amvpCand, MV* mvc);
     bool          isDiffMER(int xN, int yN, int xP, int yP);
     void          getPartPosition(uint32_t partIdx, int& xP, int& yP, int& nPSW, int& nPSH);
     void          setMVPIdx(int picList, uint32_t idx, int mvpIdx) { m_mvpIdx[picList][idx] = (uint8_t)mvpIdx; }
 
-    uint8_t       getMVPIdx(int picList, uint32_t idx)             { return m_mvpIdx[picList][idx]; }
+    uint8_t       getMVPIdx(int picList, uint32_t idx)         { return m_mvpIdx[picList][idx]; }
 
     uint8_t*      getMVPIdx(int picList)                       { return m_mvpIdx[picList]; }
 
@@ -389,72 +465,57 @@
 
     TComDataCU*   getCUAboveRight() { return m_cuAboveRight; }
 
-    TComDataCU*   getCUColocated(int picList) { return m_cuColocated[picList]; }
-
-    TComDataCU*   getPULeft(uint32_t& lPartUnitIdx,
-                            uint32_t  curPartUnitIdx,
-                            bool      bEnforceSliceRestriction = true,
-                            bool      bEnforceTileRestriction = true);
-    TComDataCU*   getPUAbove(uint32_t& uiAPartUnitIdx,
-                             uint32_t  curPartUnitIdx,
-                             bool      bEnforceSliceRestriction = true,
-                             bool      planarAtLCUBoundary = false,
-                             bool      bEnforceTileRestriction = true);
-    TComDataCU*   getPUAboveLeft(uint32_t& alPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction = true);
-    TComDataCU*   getPUAboveRight(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction = true);
-    TComDataCU*   getPUBelowLeft(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx, bool bEnforceSliceRestriction = true);
-
-    TComDataCU*   getQpMinCuLeft(uint32_t& lPartUnitIdx, uint32_t uiCurrAbsIdxInLCU);
+    TComDataCU*   getPULeft(uint32_t& lPartUnitIdx, uint32_t curPartUnitIdx);
+    TComDataCU*   getPUAbove(uint32_t& aPartUnitIdx, uint32_t curPartUnitIdx, bool planarAtLCUBoundary = false);
+    TComDataCU*   getPUAboveLeft(uint32_t& alPartUnitIdx, uint32_t curPartUnitIdx);
+    TComDataCU*   getPUAboveRight(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx);
+    TComDataCU*   getPUBelowLeft(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx);
+
+    TComDataCU*   getQpMinCuLeft(uint32_t& lPartUnitIdx, uint32_t currAbsIdxInLCU);
     TComDataCU*   getQpMinCuAbove(uint32_t& aPartUnitIdx, uint32_t currAbsIdxInLCU);
-    char          getRefQP(uint32_t uiCurrAbsIdxInLCU);
+    char          getRefQP(uint32_t currAbsIdxInLCU);
 
     TComDataCU*   getPUAboveRightAdi(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx, uint32_t partUnitOffset = 1);
     TComDataCU*   getPUBelowLeftAdi(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx, uint32_t partUnitOffset = 1);
 
     void          deriveLeftRightTopIdx(uint32_t partIdx, uint32_t& partIdxLT, uint32_t& partIdxRT);
     void          deriveLeftBottomIdx(uint32_t partIdx, uint32_t& partIdxLB);
-
     void          deriveLeftRightTopIdxAdi(uint32_t& partIdxLT, uint32_t& partIdxRT, uint32_t partOffset, uint32_t partDepth);
-    void          deriveLeftBottomIdxAdi(uint32_t& partIdxLB, uint32_t partOffset, uint32_t partDepth);
 
     bool          hasEqualMotion(uint32_t absPartIdx, TComDataCU* candCU, uint32_t candAbsPartIdx);
-    void          getInterMergeCandidates(uint32_t absPartIdx, uint32_t puIdx, TComMvField* mFieldNeighbours, UChar* interDirNeighbours, int& numValidMergeCand);
-    void          deriveLeftRightTopIdxGeneral(uint32_t absPartIdx, uint32_t partIdx, uint32_t& partIdxLT, uint32_t& partIdxRT);
-    void          deriveLeftBottomIdxGeneral(uint32_t absPartIdx, uint32_t partIdx, uint32_t& partIdxLB);
+    void          getInterMergeCandidates(uint32_t absPartIdx, uint32_t puIdx, TComMvField (*mvFieldNeighbours)[2], uint8_t* interDirNeighbours, uint32_t& maxNumMergeCand);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for modes
     // -------------------------------------------------------------------------------------------------------------------
 
-    bool          isIntra(uint32_t partIdx)  { return m_predModes[partIdx] == MODE_INTRA; }
-
-    bool          isSkipped(uint32_t partIdx); ///< SKIP (no residual)
-    bool          isBipredRestriction();
+    bool          isIntra(uint32_t partIdx) const { return m_predModes[partIdx] == MODE_INTRA; }
+    bool          isSkipped(uint32_t idx) const { return m_skipFlag[idx]; }
+    bool          isBipredRestriction() const { return m_log2CUSize[0] == 3 && m_partSizes[0] != SIZE_2Nx2N; }
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for symbol prediction (most probable / mode conversion)
     // -------------------------------------------------------------------------------------------------------------------
 
     void          getAllowedChromaDir(uint32_t absPartIdx, uint32_t* modeList);
-    int           getIntraDirLumaPredictor(uint32_t absPartIdx, int32_t* intraDirPred);
+    int           getIntraDirLumaPredictor(uint32_t absPartIdx, uint32_t* intraDirPred);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for SBAC context
     // -------------------------------------------------------------------------------------------------------------------
 
     uint32_t      getCtxSplitFlag(uint32_t absPartIdx, uint32_t depth);
-    uint32_t      getCtxQtCbf(TextType ttype, uint32_t trDepth) { return ttype == TEXT_LUMA ? (trDepth == 0 ? 1 : 0) : trDepth + 2; }
-
     uint32_t      getCtxSkipFlag(uint32_t absPartIdx);
-    uint32_t      getCtxInterDir(uint32_t absPartIdx);
+    uint32_t      getCtxInterDir(uint32_t idx) const { return m_depth[idx]; }
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions for RD cost storage
     // -------------------------------------------------------------------------------------------------------------------
 
-    uint32_t&     getTotalNumPart()               { return m_numPartitions; }
+    uint32_t&     getTotalNumPart()     { return m_numPartitions; }
 
-    uint32_t      getCoefScanIdx(uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma, bool bIsIntra);
+    ScanType      getCoefScanIdx(uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma, bool bIsIntra);
+    void          getTUEntropyCodingParameters(TUEntropyCodingParameters &result, uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma);
 
     // -------------------------------------------------------------------------------------------------------------------
     // member functions to support multiple color space formats

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComMotionInfo.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComMotionInfo.cpp Changed

@@ -51,28 +51,36 @@
 // Create / destroy
 // --------------------------------------------------------------------------------------------------------------------
 
-bool TComCUMvField::create(uint32_t numPartition)
+bool TComCUMvField::initialize(uint32_t numPartition, uint32_t numBlocks)
 {
-    CHECKED_MALLOC(m_mv, MV, numPartition);
-    CHECKED_MALLOC(m_mvd, MV, numPartition);
-    CHECKED_MALLOC(m_refIdx, char, numPartition);
-
-    m_numPartitions = numPartition;
+    CHECKED_MALLOC(m_MVFieldMemPool.m_mvMemBlock,     MV,   numPartition * 2 * numBlocks);
+    CHECKED_MALLOC(m_MVFieldMemPool.m_mvdMemBlock,    MV,   numPartition * 2 * numBlocks);
+    CHECKED_MALLOC(m_MVFieldMemPool.m_refIdxMemBlock, char, numPartition * 2 * numBlocks);
 
     return true;
+
 fail:
     return false;
 }
 
+void TComCUMvField::create(TComCUMvField *p, uint32_t numPartition, int index, int idx)
+{
+    m_mv     = p->m_MVFieldMemPool.m_mvMemBlock     + (index * 2 + idx) * numPartition;
+    m_mvd    = p->m_MVFieldMemPool.m_mvdMemBlock    + (index * 2 + idx) * numPartition;
+    m_refIdx = p->m_MVFieldMemPool.m_refIdxMemBlock + (index * 2 + idx) * numPartition;
+
+    m_numPartitions = numPartition;
+}
+
 void TComCUMvField::destroy()
 {
-    X265_FREE(m_mv);
-    X265_FREE(m_mvd);
-    X265_FREE(m_refIdx);
+    X265_FREE(m_MVFieldMemPool.m_mvMemBlock);
+    X265_FREE(m_MVFieldMemPool.m_mvdMemBlock);
+    X265_FREE(m_MVFieldMemPool.m_refIdxMemBlock);
 
-    m_mv     = NULL;
-    m_mvd    = NULL;
-    m_refIdx = NULL;
+    m_MVFieldMemPool.m_mvMemBlock     = NULL;
+    m_MVFieldMemPool.m_mvdMemBlock    = NULL;
+    m_MVFieldMemPool.m_refIdxMemBlock = NULL;
 
     m_numPartitions = 0;
 }
@@ -83,7 +91,7 @@
 
 void TComCUMvField::clearMvField()
 {
-    assert(sizeof(*m_refIdx) == 1);
+    X265_CHECK(sizeof(*m_refIdx) == 1, "size check\n");
     memset(m_refIdx, NOT_VALID, m_numPartitions * sizeof(*m_refIdx));
 }
 
@@ -294,7 +302,7 @@
         break;
     }
     default:
-        assert(0);
+        X265_CHECK(0, "unknown partition type\n");
         break;
     }
 }

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComMotionInfo.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComMotionInfo.h Changed

@@ -39,7 +39,6 @@
 #ifndef X265_TCOMMOTIONINFO_H
 #define X265_TCOMMOTIONINFO_H
 
-#include "CommonDef.h"
 #include "common.h"
 #include "mv.h"
 
@@ -53,12 +52,12 @@
 // Type definition
 // ====================================================================================================================
 
-/// parameters for AMVP
-typedef struct _AMVPInfo
+typedef struct
 {
-    MV m_mvCand[AMVP_MAX_NUM_CANDS_MEM];  ///< array of motion vector predictor candidates
-    int      m_num;                             ///< number of motion vector predictor candidates
-} AMVPInfo;
+    MV*   m_mvMemBlock;
+    MV*   m_mvdMemBlock;
+    char* m_refIdxMemBlock;
+} MVFieldMemPool;
 
 // ====================================================================================================================
 // Class definition
@@ -89,7 +88,7 @@
     MV* m_mv;
     MV* m_mvd;
     char*     m_refIdx;
-    uint32_t      m_numPartitions;
+    uint32_t  m_numPartitions;
 
     template<typename T>
     void setAll(T *p, T const & val, PartSize cuMode, int partAddr, uint32_t depth, int partIdx);
@@ -99,10 +98,13 @@
     ~TComCUMvField() {}
 
     // ------------------------------------------------------------------------------------------------------------------
-    // create / destroy
+    // initialize / create / destroy
     // ------------------------------------------------------------------------------------------------------------------
 
-    bool create(uint32_t numPartition);
+    MVFieldMemPool m_MVFieldMemPool;
+
+    bool initialize(uint32_t numPartition, uint32_t numBlocks);
+    void create(TComCUMvField *p, uint32_t numPartition, int index, int idx);
     void destroy();
 
     // ------------------------------------------------------------------------------------------------------------------
@@ -129,11 +131,10 @@
     // set
     // ------------------------------------------------------------------------------------------------------------------
 
-    void    setAllMv(const MV& mv,              PartSize cuMode, int partAddr, uint32_t depth, int partIdx = 0);
+    void    setAllMv(const MV& mv,                    PartSize cuMode, int partAddr, uint32_t depth, int partIdx = 0);
     void    setAllRefIdx(int refIdx,                  PartSize mbMode, int partAddr, uint32_t depth, int partIdx = 0);
     void    setAllMvField(const TComMvField& mvField, PartSize mbMode, int partAddr, uint32_t depth, int partIdx = 0);
     void    setMvd(int idx, const MV& mvd) { m_mvd[idx] = mvd; }
-
 };
 }

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPattern.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPattern.cpp Changed

@@ -35,9 +35,10 @@
     \brief    neighboring pixel access classes
 */
 
-#include "TComPic.h"
+#include "frame.h"
 #include "TComPattern.h"
 #include "TComDataCU.h"
+#include "TComRom.h"
 
 using namespace x265;
 
@@ -49,385 +50,326 @@
 // ====================================================================================================================
 
 void TComPattern::initAdiPattern(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf,
-                                 int strideOrig, int heightOrig)
+                                 pixel* refAbove, pixel* refLeft, pixel* refAboveFlt, pixel* refLeftFlt, int dirMode)
 {
-    pixel* roiOrigin;
-    pixel* adiTemp;
-    uint32_t cuWidth = cu->getCUSize(0) >> partDepth;
-    uint32_t cuHeight = cu->getCUSize(0) >> partDepth;
-    uint32_t cuWidth2  = cuWidth << 1;
-    uint32_t cuHeight2 = cuHeight << 1;
-    uint32_t width;
-    uint32_t height;
-    int  picStride = cu->getPic()->getStride();
-    int  unitSize = 0;
-    int  numUnitsInCU = 0;
-    int  totalUnits = 0;
-    bool bNeighborFlags[4 * MAX_NUM_SPU_W + 1];
-    int  numIntraNeighbor = 0;
+    IntraNeighbors intraNeighbors;
 
-    uint32_t partIdxLT, partIdxRT, partIdxLB;
+    initIntraNeighbors(cu, zOrderIdxInPart, partDepth, true, &intraNeighbors);
+    uint32_t tuSize = intraNeighbors.tuSize;
+    uint32_t tuSize2 = tuSize << 1;
 
-    cu->deriveLeftRightTopIdxAdi(partIdxLT, partIdxRT, zOrderIdxInPart, partDepth);
-    cu->deriveLeftBottomIdxAdi(partIdxLB,              zOrderIdxInPart, partDepth);
+    pixel* adiOrigin = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + zOrderIdxInPart);
+    int picStride = cu->m_pic->getStride();
 
-    unitSize      = g_maxCUSize >> g_maxCUDepth;
-    numUnitsInCU  = cuWidth / unitSize;
-    totalUnits    = (numUnitsInCU << 2) + 1;
+    fillReferenceSamples(adiOrigin, picStride, adiBuf, intraNeighbors);
 
-    bNeighborFlags[numUnitsInCU * 2] = isAboveLeftAvailable(cu, partIdxLT);
-    numIntraNeighbor  += (int)(bNeighborFlags[numUnitsInCU * 2]);
-    numIntraNeighbor  += isAboveAvailable(cu, partIdxLT, partIdxRT, bNeighborFlags + (numUnitsInCU * 2) + 1);
-    numIntraNeighbor  += isAboveRightAvailable(cu, partIdxLT, partIdxRT, bNeighborFlags + (numUnitsInCU * 3) + 1);
-    numIntraNeighbor  += isLeftAvailable(cu, partIdxLT, partIdxLB, bNeighborFlags + (numUnitsInCU * 2) - 1);
-    numIntraNeighbor  += isBelowLeftAvailable(cu, partIdxLT, partIdxLB, bNeighborFlags + numUnitsInCU   - 1);
+    // initialization of ADI buffers
+    const int bufOffset = tuSize - 1;
+    refAbove += bufOffset;
+    refLeft += bufOffset;
 
-    width = cuWidth2 + 1;
-    height = cuHeight2 + 1;
+    //  ADI_BUF_STRIDE * (2 * tuSize + 1);
+    memcpy(refAbove, adiBuf, (tuSize2 + 1) * sizeof(pixel));
+    for (int k = 0; k < tuSize2 + 1; k++)
+        refLeft[k] = adiBuf[k * ADI_BUF_STRIDE];
+    
+    bool bUseFilteredPredictions = (dirMode == ALL_IDX ? (8 | 16 | 32) & tuSize : g_intraFilterFlags[dirMode] & tuSize);
 
-    if (((width << 2) > strideOrig) || ((height << 2) > heightOrig))
+    if (bUseFilteredPredictions)
     {
-        return;
-    }
+        // generate filtered intra prediction samples
+        refAboveFlt += bufOffset;
+        refLeftFlt += bufOffset;
 
-    roiOrigin = cu->getPic()->getPicYuvRec()->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + zOrderIdxInPart);
-    adiTemp   = adiBuf;
+        bool bStrongSmoothing = (tuSize == 32 && cu->m_slice->m_sps->bUseStrongIntraSmoothing);
 
-    fillReferenceSamples(roiOrigin, adiTemp, bNeighborFlags, numIntraNeighbor, unitSize, numUnitsInCU, totalUnits, cuWidth, cuHeight, width, height, picStride);
-
-    // generate filtered intra prediction samples
-    // left and left above border + above and above right border + top left corner = length of 3. filter buffer
-    int bufSize = cuHeight2 + cuWidth2 + 1;
-    uint32_t wh = ADI_BUF_STRIDE * height;         // number of elements in one buffer
-
-    pixel* filteredBuf1 = adiBuf + wh;         // 1. filter buffer
-    pixel* filteredBuf2 = filteredBuf1 + wh; // 2. filter buffer
-    pixel* filterBuf = filteredBuf2 + wh;    // buffer for 2. filtering (sequential)
-    pixel* filterBufN = filterBuf + bufSize; // buffer for 1. filtering (sequential)
-
-    int l = 0;
-    // left border from bottom to top
-    for (int i = 0; i < cuHeight2; i++)
-    {
-        filterBuf[l++] = adiTemp[ADI_BUF_STRIDE * (cuHeight2 - i)];
-    }
+        if (bStrongSmoothing)
+        {
+            const int trSize  = 32;
+            const int trSize2 = 32 * 2;
+            const int threshold = 1 << (X265_DEPTH - 5);
+            int refBL = refLeft[trSize2];
+            int refTL = refAbove[0];
+            int refTR = refAbove[trSize2];
+            bStrongSmoothing = (abs(refBL + refTL - 2 * refLeft[trSize])  < threshold &&
+                                abs(refTL + refTR - 2 * refAbove[trSize]) < threshold);
+
+            if (bStrongSmoothing)
+            {
+                // bilinear interpolation
+                const int shift = 5 + 1; // intraNeighbors.log2TrSize + 1;
+                int init = (refTL << shift) + tuSize;
+                int delta;
 
-    // top left corner
-    filterBuf[l++] = adiTemp[0];
+                refLeftFlt[0] = refAboveFlt[0] = refAbove[0];
 
-    // above border from left to right
-    memcpy(&filterBuf[l], &adiTemp[1], cuWidth2 * sizeof(*filterBuf));
+                //TODO: Performance Primitive???
+                delta = refBL - refTL;
+                for (int i = 1; i < trSize2; i++)
+                    refLeftFlt[i] = (init + delta * i) >> shift;
+                refLeftFlt[trSize2] = refLeft[trSize2];
 
-    if (cu->getSlice()->getSPS()->getUseStrongIntraSmoothing())
-    {
-        int blkSize = 32;
-        int bottomLeft = filterBuf[0];
-        int topLeft = filterBuf[cuHeight2];
-        int topRight = filterBuf[bufSize - 1];
-        int threshold = 1 << (X265_DEPTH - 5);
-        bool bilinearLeft = abs(bottomLeft + topLeft - 2 * filterBuf[cuHeight]) < threshold;
-        bool bilinearAbove  = abs(topLeft + topRight - 2 * filterBuf[cuHeight2 + cuHeight]) < threshold;
-
-        if (cuWidth >= blkSize && (bilinearLeft && bilinearAbove))
-        {
-            int shift = g_convertToBit[cuWidth] + 3; // log2(uiCuHeight2)
-            filterBufN[0] = filterBuf[0];
-            filterBufN[cuHeight2] = filterBuf[cuHeight2];
-            filterBufN[bufSize - 1] = filterBuf[bufSize - 1];
-            //TODO: Performance Primitive???
-            for (int i = 1; i < cuHeight2; i++)
-            {
-                filterBufN[i] = ((cuHeight2 - i) * bottomLeft + i * topLeft + cuHeight) >> shift;
-            }
+                delta = refTR - refTL;
+                for (int i = 1; i < trSize2; i++)
+                    refAboveFlt[i] = (init + delta * i) >> shift;
+                refAboveFlt[trSize2] = refAbove[trSize2];
 
-            for (int i = 1; i < cuWidth2; i++)
-            {
-                filterBufN[cuHeight2 + i] = ((cuWidth2 - i) * topLeft + i * topRight + cuWidth) >> shift;
-            }
-        }
-        else
-        {
-            // 1. filtering with [1 2 1]
-            filterBufN[0] = filterBuf[0];
-            filterBufN[bufSize - 1] = filterBuf[bufSize - 1];
-            for (int i = 1; i < bufSize - 1; i++)
-            {
-                filterBufN[i] = (filterBuf[i - 1] + 2 * filterBuf[i] + filterBuf[i + 1] + 2) >> 2;
+                return;
             }
         }
-    }
-    else
-    {
-        // 1. filtering with [1 2 1]
-        filterBufN[0] = filterBuf[0];
-        filterBufN[bufSize - 1] = filterBuf[bufSize - 1];
-        for (int i = 1; i < bufSize - 1; i++)
-        {
-            filterBufN[i] = (filterBuf[i - 1] + 2 * filterBuf[i] + filterBuf[i + 1] + 2) >> 2;
-        }
-    }
 
-    // fill 1. filter buffer with filtered values
-    l = 0;
-    for (int i = 0; i < cuHeight2; i++)
-    {
-        filteredBuf1[ADI_BUF_STRIDE * (cuHeight2 - i)] = filterBufN[l++];
+        refLeft[-1] = refAbove[1];
+        for (int i = 0; i < tuSize2; i++)
+            refLeftFlt[i] = (refLeft[i - 1] + 2 * refLeft[i] + refLeft[i + 1] + 2) >> 2;
+        refLeftFlt[tuSize2] = refLeft[tuSize2];
+
+        refAboveFlt[0] = refLeftFlt[0];
+        for (int i = 1; i < tuSize2; i++)
+            refAboveFlt[i] = (refAbove[i - 1] + 2 * refAbove[i] + refAbove[i + 1] + 2) >> 2;
+        refAboveFlt[tuSize2] = refAbove[tuSize2];
     }
+}
 
-    filteredBuf1[0] = filterBufN[l++];
-    memcpy(&filteredBuf1[1], &filterBufN[l], cuWidth2 * sizeof(*filteredBuf1));
+void TComPattern::initAdiPatternChroma(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf, uint32_t chromaId)
+{
+    IntraNeighbors intraNeighbors;
+
+    initIntraNeighbors(cu, zOrderIdxInPart, partDepth, false, &intraNeighbors);
+    uint32_t tuSize = intraNeighbors.tuSize;
+
+    pixel* adiOrigin = cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), cu->getZorderIdxInCU() + zOrderIdxInPart);
+    int picStride = cu->m_pic->getCStride();
+    pixel* adiRef = getAdiChromaBuf(chromaId, tuSize, adiBuf);
+
+    fillReferenceSamples(adiOrigin, picStride, adiRef, intraNeighbors);
 }
 
-// Overloaded initialization of ADI buffers to support buffered references for xpredIntraAngBufRef
-void TComPattern::initAdiPattern(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf, int strideOrig, int heightOrig,
-                                 pixel* refAbove, pixel* refLeft, pixel* refAboveFlt, pixel* refLeftFlt)
+void TComPattern::initIntraNeighbors(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, bool isLuma, IntraNeighbors *intraNeighbors)
 {
-    initAdiPattern(cu, zOrderIdxInPart, partDepth, adiBuf, strideOrig, heightOrig);
-    uint32_t cuWidth   = cu->getCUSize(0) >> partDepth;
-    uint32_t cuHeight  = cu->getCUSize(0) >> partDepth;
-    uint32_t cuWidth2  = cuWidth << 1;
-    uint32_t cuHeight2 = cuHeight << 1;
-
-    refAbove += cuWidth - 1;
-    refAboveFlt += cuWidth - 1;
-    refLeft += cuWidth - 1;
-    refLeftFlt += cuWidth - 1;
-
-    //  ADI_BUF_STRIDE * (2 * height + 1);
-    memcpy(refAbove, adiBuf, (cuWidth2 + 1) * sizeof(pixel));
-    memcpy(refAboveFlt, adiBuf + ADI_BUF_STRIDE * (2 * cuHeight + 1), (cuWidth2 + 1) * sizeof(pixel));
-
-    for (int k = 0; k < cuHeight2 + 1; k++)
+    uint32_t log2TrSize = cu->getLog2CUSize(0) - partDepth;
+    int log2UnitWidth  = LOG2_UNIT_SIZE;
+    int log2UnitHeight = LOG2_UNIT_SIZE;
+
+    if (!isLuma)
     {
-        refLeft[k] = adiBuf[k * ADI_BUF_STRIDE];
-        refLeftFlt[k] = (adiBuf + ADI_BUF_STRIDE * (cuHeight2 + 1))[k * ADI_BUF_STRIDE];   // Smoothened
+        log2TrSize     -= cu->getHorzChromaShift();
+        log2UnitWidth  -= cu->getHorzChromaShift();
+        log2UnitHeight -= cu->getVertChromaShift();
     }
-}
 
-void TComPattern::initAdiPatternChroma(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf, int strideOrig, int heightOrig, int chromaId)
-{
-    pixel*  roiOrigin;
-    pixel*  adiTemp;
-    uint32_t  cuWidth  = cu->getCUSize(0) >> partDepth;
-    uint32_t  cuHeight = cu->getCUSize(0) >> partDepth;
-    uint32_t  width;
-    uint32_t  height;
-    int   picStride = cu->getPic()->getCStride();
-
-    int   unitSize = 0;
-    int   numUnitsInCU = 0;
-    int   totalUnits = 0;
-    bool  bNeighborFlags[4 * MAX_NUM_SPU_W + 1];
     int   numIntraNeighbor = 0;
+    bool *bNeighborFlags = intraNeighbors->bNeighborFlags;
 
     uint32_t partIdxLT, partIdxRT, partIdxLB;
 
     cu->deriveLeftRightTopIdxAdi(partIdxLT, partIdxRT, zOrderIdxInPart, partDepth);
-    cu->deriveLeftBottomIdxAdi(partIdxLB,              zOrderIdxInPart, partDepth);
 
-    unitSize      = (g_maxCUSize >> g_maxCUDepth) >> cu->getHorzChromaShift(); // for chroma
-    numUnitsInCU  = (cuWidth / unitSize) >> cu->getHorzChromaShift();           // for chroma
-    totalUnits    = (numUnitsInCU << 2) + 1;
-
-    bNeighborFlags[numUnitsInCU * 2] = isAboveLeftAvailable(cu, partIdxLT);
-    numIntraNeighbor += (int)(bNeighborFlags[numUnitsInCU * 2]);
-    numIntraNeighbor += isAboveAvailable(cu, partIdxLT, partIdxRT, bNeighborFlags + (numUnitsInCU * 2) + 1);
-    numIntraNeighbor += isAboveRightAvailable(cu, partIdxLT, partIdxRT, bNeighborFlags + (numUnitsInCU * 3) + 1);
-    numIntraNeighbor += isLeftAvailable(cu, partIdxLT, partIdxLB, bNeighborFlags + (numUnitsInCU * 2) - 1);
-    numIntraNeighbor += isBelowLeftAvailable(cu, partIdxLT, partIdxLB, bNeighborFlags + numUnitsInCU   - 1);
-
-    cuWidth = cuWidth >> cu->getHorzChromaShift(); // for chroma
-    cuHeight = cuHeight >> cu->getVertChromaShift(); // for chroma
-
-    width = cuWidth * 2 + 1;
-    height = cuHeight * 2 + 1;
-
-    if ((4 * width > strideOrig) || (4 * height > heightOrig))
-    {
-        return;
-    }
-    roiOrigin = chromaId > 0 ? cu->getPic()->getPicYuvRec()->getCrAddr(cu->getAddr(), cu->getZorderIdxInCU() + zOrderIdxInPart) : cu->getPic()->getPicYuvRec()->getCbAddr(cu->getAddr(), cu->getZorderIdxInCU() + zOrderIdxInPart);
-    adiTemp   = chromaId > 0 ? (adiBuf + 2 * ADI_BUF_STRIDE * height) : adiBuf;
-    fillReferenceSamples(roiOrigin, adiTemp, bNeighborFlags, numIntraNeighbor, unitSize, numUnitsInCU, totalUnits,
-                         cuWidth, cuHeight, width, height, picStride);
+    uint32_t tuSize  = 1 << log2TrSize;
+    int  tuWidthInUnits  = tuSize >> log2UnitWidth;
+    int  tuHeightInUnits = tuSize >> log2UnitHeight;
+    int  aboveUnits      = tuWidthInUnits << 1;
+    int  leftUnits       = tuHeightInUnits << 1;
+    int  partIdxStride   = cu->m_pic->getNumPartInCUSize();
+    partIdxLB            = g_rasterToZscan[g_zscanToRaster[partIdxLT] + ((tuHeightInUnits - 1) * partIdxStride)];
+
+    bNeighborFlags[leftUnits] = isAboveLeftAvailable(cu, partIdxLT);
+    numIntraNeighbor += (int)(bNeighborFlags[leftUnits]);
+    numIntraNeighbor += isAboveAvailable(cu, partIdxLT, partIdxRT, (bNeighborFlags + leftUnits + 1));
+    numIntraNeighbor += isAboveRightAvailable(cu, partIdxLT, partIdxRT, (bNeighborFlags + leftUnits + 1 + tuWidthInUnits));
+    numIntraNeighbor += isLeftAvailable(cu, partIdxLT, partIdxLB, (bNeighborFlags + leftUnits - 1));
+    numIntraNeighbor += isBelowLeftAvailable(cu, partIdxLT, partIdxLB, (bNeighborFlags + leftUnits   - 1 - tuHeightInUnits));
+
+    intraNeighbors->numIntraNeighbor = numIntraNeighbor;
+    intraNeighbors->totalUnits       = aboveUnits + leftUnits + 1;
+    intraNeighbors->aboveUnits       = aboveUnits;
+    intraNeighbors->leftUnits        = leftUnits;
+    intraNeighbors->unitWidth        = 1 << log2UnitWidth;
+    intraNeighbors->unitHeight       = 1 << log2UnitHeight;
+    intraNeighbors->tuSize           = tuSize;
+    intraNeighbors->log2TrSize       = log2TrSize;
 }
 
-void TComPattern::fillReferenceSamples(pixel* roiOrigin, pixel* adiTemp, bool* bNeighborFlags, int numIntraNeighbor, int unitSize, int numUnitsInCU, int totalUnits, uint32_t cuWidth, uint32_t cuHeight, uint32_t width, uint32_t height, int picStride)
+void TComPattern::fillReferenceSamples(pixel* adiOrigin, int picStride, pixel* adiRef, const IntraNeighbors& intraNeighbors)
 {
-    pixel* roiTemp;
-    int i, j;
-    int dcValue = 1 << (X265_DEPTH - 1);
+    int numIntraNeighbor = intraNeighbors.numIntraNeighbor;
+    int totalUnits       = intraNeighbors.totalUnits;
+    uint32_t tuSize      = intraNeighbors.tuSize;
+
+    uint32_t refSize = tuSize * 2 + 1;
+    int  i, j;
+    int  dcValue = 1 << (X265_DEPTH - 1);
 
     if (numIntraNeighbor == 0)
     {
         // Fill border with DC value
-        for (i = 0; i < width; i++)
-        {
-            adiTemp[i] = dcValue;
-        }
+        for (i = 0; i < refSize; i++)
+            adiRef[i] = dcValue;
 
-        for (i = 1; i < height; i++)
-        {
-            adiTemp[i * ADI_BUF_STRIDE] = dcValue;
-        }
+        for (i = 1; i < refSize; i++)
+            adiRef[i * ADI_BUF_STRIDE] = dcValue;
     }
     else if (numIntraNeighbor == totalUnits)
     {
-        // Fill top-left border with rec. samples
-        roiTemp = roiOrigin - picStride - 1;
-        adiTemp[0] = roiTemp[0];
+        // Fill top border with rec. samples
+        pixel* adiTemp = adiOrigin - picStride - 1;
+        memcpy(adiRef, adiTemp, refSize * sizeof(*adiRef));
 
         // Fill left border with rec. samples
-        // Fill below left border with rec. samples
-        roiTemp = roiOrigin - 1;
-
-        for (i = 0; i < 2 * cuHeight; i++)
+        adiTemp = adiOrigin - 1;
+        for (i = 1; i < refSize; i++)
         {
-            adiTemp[(1 + i) * ADI_BUF_STRIDE] = roiTemp[0];
-            roiTemp += picStride;
+            adiRef[i * ADI_BUF_STRIDE] = adiTemp[0];
+            adiTemp += picStride;
         }
-
-        // Fill top border with rec. samples
-        // Fill top right border with rec. samples
-        roiTemp = roiOrigin - picStride;
-        memcpy(&adiTemp[1], roiTemp, 2 * cuWidth * sizeof(*adiTemp));
     }
     else // reference samples are partially available
     {
-        int    numUnits2 = numUnitsInCU << 1;
-        int    totalSamples = totalUnits * unitSize;
-        pixel  adiLine[5 * MAX_CU_SIZE];
-        pixel* adiLineTemp;
-        bool*  neighborFlagPtr;
-        int    next, curr;
-        pixel  iRef = 0;
+        const bool *bNeighborFlags = intraNeighbors.bNeighborFlags;
+        int aboveUnits       = intraNeighbors.aboveUnits;
+        int leftUnits        = intraNeighbors.leftUnits;
+        int unitWidth        = intraNeighbors.unitWidth;
+        int unitHeight       = intraNeighbors.unitHeight;
+        int  totalSamples = (leftUnits * unitHeight) + ((aboveUnits + 1) * unitWidth);
+        pixel pAdiLine[5 * MAX_CU_SIZE];
+        pixel *pAdiLineTemp;
+        const bool  *pNeighborFlags;
+        int   next, curr;
 
         // Initialize
         for (i = 0; i < totalSamples; i++)
         {
-            adiLine[i] = dcValue;
+            pAdiLine[i] = dcValue;
         }
 
         // Fill top-left sample
-        roiTemp = roiOrigin - picStride - 1;
-        adiLineTemp = adiLine + (numUnits2 * unitSize);
-        neighborFlagPtr = bNeighborFlags + numUnits2;
-        if (*neighborFlagPtr)
+        pixel* adiTemp =  adiOrigin - picStride - 1;
+        pAdiLineTemp = pAdiLine + (leftUnits * unitHeight);
+        pNeighborFlags = bNeighborFlags + leftUnits;
+        if (*pNeighborFlags)
         {
-            adiLineTemp[0] = roiTemp[0];
-            for (i = 1; i < unitSize; i++)
+            pixel topLeftVal = adiTemp[0];
+            for (i = 0; i < unitWidth; i++)
             {
-                adiLineTemp[i] = adiLineTemp[0];
+                pAdiLineTemp[i] = topLeftVal;
             }
         }
 
         // Fill left & below-left samples
-        roiTemp += picStride;
-        adiLineTemp--;
-        neighborFlagPtr--;
-        for (j = 0; j < numUnits2; j++)
+        adiTemp += picStride;
+        pAdiLineTemp--;
+        pNeighborFlags--;
+        for (j = 0; j < leftUnits; j++)
         {
-            if (*neighborFlagPtr)
+            if (*pNeighborFlags)
             {
-                for (i = 0; i < unitSize; i++)
+                for (i = 0; i < unitHeight; i++)
                 {
-                    adiLineTemp[-i] = roiTemp[i * picStride];
+                    pAdiLineTemp[-i] = adiTemp[i * picStride];
                 }
             }
-            roiTemp += unitSize * picStride;
-            adiLineTemp -= unitSize;
-            neighborFlagPtr--;
+            adiTemp += unitHeight * picStride;
+            pAdiLineTemp -= unitHeight;
+            pNeighborFlags--;
         }
 
         // Fill above & above-right samples
-        roiTemp = roiOrigin - picStride;
-        adiLineTemp = adiLine + ((numUnits2 + 1) * unitSize);
-        neighborFlagPtr = bNeighborFlags + numUnits2 + 1;
-        for (j = 0; j < numUnits2; j++)
+        adiTemp = adiOrigin - picStride;
+        pAdiLineTemp = pAdiLine + (leftUnits * unitHeight) + unitWidth;
+        pNeighborFlags = bNeighborFlags + leftUnits + 1;
+        for (j = 0; j < aboveUnits; j++)
         {
-            if (*neighborFlagPtr)
-            {
-                memcpy(adiLineTemp, roiTemp, unitSize * sizeof(*adiTemp));
-            }
-            roiTemp += unitSize;
-            adiLineTemp += unitSize;
-            neighborFlagPtr++;
+            if (*pNeighborFlags)
+                memcpy(pAdiLineTemp, adiTemp, unitWidth * sizeof(*adiTemp));
+            adiTemp += unitWidth;
+            pAdiLineTemp += unitWidth;
+            pNeighborFlags++;
         }
 
         // Pad reference samples when necessary
         curr = 0;
         next = 1;
-        adiLineTemp = adiLine;
-        while (curr < totalUnits)
+        pAdiLineTemp = pAdiLine;
+        int pAdiLineTopRowOffset = leftUnits * (unitHeight - unitWidth);
+        if (!bNeighborFlags[0])
         {
-            if (!bNeighborFlags[curr])
+            // very bottom unit of bottom-left; at least one unit will be valid.
+            while (next < totalUnits && !bNeighborFlags[next])
+            {
+                next++;
+            }
+
+            pixel *pAdiLineNext = pAdiLine + ((next < leftUnits) ? (next * unitHeight) : (pAdiLineTopRowOffset + (next * unitWidth)));
+            const pixel refSample = *pAdiLineNext;
+            // Pad unavailable samples with new value
+            int nextOrTop = std::min<int>(next, leftUnits);
+            // fill left column
+            while (curr < nextOrTop)
+            {
+                for (i = 0; i < unitHeight; i++)
+                {
+                    pAdiLineTemp[i] = refSample;
+                }
+
+                pAdiLineTemp += unitHeight;
+                curr++;
+            }
+
+            // fill top row
+            while (curr < next)
             {
-                if (curr == 0)
+                for (i = 0; i < unitWidth; i++)
                 {
-                    while (next < totalUnits && !bNeighborFlags[next])
-                    {
-                        next++;
-                    }
-
-                    iRef = adiLine[next * unitSize];
-                    // Pad unavailable samples with new value
-                    while (curr < next)
-                    {
-                        for (i = 0; i < unitSize; i++)
-                        {
-                            adiLineTemp[i] = iRef;
-                        }
-
-                        adiLineTemp += unitSize;
-                        curr++;
-                    }
+                    pAdiLineTemp[i] = refSample;
                 }
-                else
+
+                pAdiLineTemp += unitWidth;
+                curr++;
+            }
+        }
+
+        // pad all other reference samples.
+        while (curr < totalUnits)
+        {
+            if (!bNeighborFlags[curr]) // samples not available
+            {
+                int numSamplesInCurrUnit = (curr >= leftUnits) ? unitWidth : unitHeight;
+                const pixel refSample = *(pAdiLineTemp - 1);
+                for (i = 0; i < numSamplesInCurrUnit; i++)
                 {
-                    iRef = adiLine[curr * unitSize - 1];
-                    for (i = 0; i < unitSize; i++)
-                    {
-                        adiLineTemp[i] = iRef;
-                    }
-
-                    adiLineTemp += unitSize;
-                    curr++;
+                    pAdiLineTemp[i] = refSample;
                 }
+
+                pAdiLineTemp += numSamplesInCurrUnit;
+                curr++;
             }
             else
             {
-                adiLineTemp += unitSize;
+                pAdiLineTemp += (curr >= leftUnits) ? unitWidth : unitHeight;
                 curr++;
             }
         }
 
         // Copy processed samples
-        adiLineTemp = adiLine + height + unitSize - 2;
-        memcpy(adiTemp, adiLineTemp, width * sizeof(*adiTemp));
+        pAdiLineTemp = pAdiLine + refSize + unitWidth - 2;
+        memcpy(adiRef, pAdiLineTemp, refSize * sizeof(*adiRef));
 
-        adiLineTemp = adiLine + height - 1;
-        for (i = 1; i < height; i++)
+        pAdiLineTemp = pAdiLine + refSize - 1;
+        for (i = 1; i < refSize; i++)
         {
-            adiTemp[i * ADI_BUF_STRIDE] = adiLineTemp[-i];
+            adiRef[i * ADI_BUF_STRIDE] = pAdiLineTemp[-i];
         }
     }
 }
 
 bool TComPattern::isAboveLeftAvailable(TComDataCU* cu, uint32_t partIdxLT)
 {
-    bool bAboveLeftFlag;
     uint32_t partAboveLeft;
     TComDataCU* pcCUAboveLeft = cu->getPUAboveLeft(partAboveLeft, partIdxLT);
 
-    if (cu->getSlice()->getPPS()->getConstrainedIntraPred())
-    {
-        bAboveLeftFlag = (pcCUAboveLeft && pcCUAboveLeft->getPredictionMode(partAboveLeft) == MODE_INTRA);
-    }
+    if (!cu->m_slice->m_pps->bConstrainedIntraPred)
+        return pcCUAboveLeft ? true : false;
     else
-    {
-        bAboveLeftFlag = (pcCUAboveLeft ? true : false);
-    }
-    return bAboveLeftFlag;
+        return pcCUAboveLeft && pcCUAboveLeft->isIntra(partAboveLeft);
 }
 
 int TComPattern::isAboveAvailable(TComDataCU* cu, uint32_t partIdxLT, uint32_t partIdxRT, bool *bValidFlags)
@@ -442,29 +384,14 @@
     {
         uint32_t uiPartAbove;
         TComDataCU* pcCUAbove = cu->getPUAbove(uiPartAbove, g_rasterToZscan[rasterPart]);
-        if (cu->getSlice()->getPPS()->getConstrainedIntraPred())
+        if (pcCUAbove && (!cu->m_slice->m_pps->bConstrainedIntraPred || pcCUAbove->isIntra(uiPartAbove)))
         {
-            if (pcCUAbove && pcCUAbove->getPredictionMode(uiPartAbove) == MODE_INTRA)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            numIntra++;
+            *validFlagPtr = true;
         }
         else
         {
-            if (pcCUAbove)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            *validFlagPtr = false;
         }
         validFlagPtr++;
     }
@@ -476,7 +403,7 @@
 {
     const uint32_t rasterPartBegin = g_zscanToRaster[partIdxLT];
     const uint32_t rasterPartEnd = g_zscanToRaster[partIdxLB] + 1;
-    const uint32_t idxStep = cu->getPic()->getNumPartInWidth();
+    const uint32_t idxStep = cu->m_pic->getNumPartInCUSize();
     bool *validFlagPtr = bValidFlags;
     int numIntra = 0;
 
@@ -484,29 +411,14 @@
     {
         uint32_t partLeft;
         TComDataCU* pcCULeft = cu->getPULeft(partLeft, g_rasterToZscan[rasterPart]);
-        if (cu->getSlice()->getPPS()->getConstrainedIntraPred())
+        if (pcCULeft && (!cu->m_slice->m_pps->bConstrainedIntraPred || pcCULeft->isIntra(partLeft)))
         {
-            if (pcCULeft && pcCULeft->getPredictionMode(partLeft) == MODE_INTRA)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            numIntra++;
+            *validFlagPtr = true;
         }
         else
         {
-            if (pcCULeft)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            *validFlagPtr = false;
         }
         validFlagPtr--; // opposite direction
     }
@@ -524,29 +436,14 @@
     {
         uint32_t uiPartAboveRight;
         TComDataCU* pcCUAboveRight = cu->getPUAboveRightAdi(uiPartAboveRight, partIdxRT, offset);
-        if (cu->getSlice()->getPPS()->getConstrainedIntraPred())
+        if (pcCUAboveRight && (!cu->m_slice->m_pps->bConstrainedIntraPred || pcCUAboveRight->isIntra(uiPartAboveRight)))
         {
-            if (pcCUAboveRight && pcCUAboveRight->getPredictionMode(uiPartAboveRight) == MODE_INTRA)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            numIntra++;
+            *validFlagPtr = true;
         }
         else
         {
-            if (pcCUAboveRight)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            *validFlagPtr = false;
         }
         validFlagPtr++;
     }
@@ -556,7 +453,7 @@
 
 int TComPattern::isBelowLeftAvailable(TComDataCU* cu, uint32_t partIdxLT, uint32_t partIdxLB, bool *bValidFlags)
 {
-    const uint32_t numUnitsInPU = (g_zscanToRaster[partIdxLB] - g_zscanToRaster[partIdxLT]) / cu->getPic()->getNumPartInWidth() + 1;
+    const uint32_t numUnitsInPU = (g_zscanToRaster[partIdxLB] - g_zscanToRaster[partIdxLT]) / cu->m_pic->getNumPartInCUSize() + 1;
     bool *validFlagPtr = bValidFlags;
     int numIntra = 0;
 
@@ -564,29 +461,14 @@
     {
         uint32_t uiPartBelowLeft;
         TComDataCU* pcCUBelowLeft = cu->getPUBelowLeftAdi(uiPartBelowLeft, partIdxLB, offset);
-        if (cu->getSlice()->getPPS()->getConstrainedIntraPred())
+        if (pcCUBelowLeft && (!cu->m_slice->m_pps->bConstrainedIntraPred || pcCUBelowLeft->isIntra(uiPartBelowLeft)))
         {
-            if (pcCUBelowLeft && pcCUBelowLeft->getPredictionMode(uiPartBelowLeft) == MODE_INTRA)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            numIntra++;
+            *validFlagPtr = true;
         }
         else
         {
-            if (pcCUBelowLeft)
-            {
-                numIntra++;
-                *validFlagPtr = true;
-            }
-            else
-            {
-                *validFlagPtr = false;
-            }
+            *validFlagPtr = false;
         }
         validFlagPtr--; // opposite direction
     }

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPattern.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPattern.h Changed

@@ -38,7 +38,8 @@
 #ifndef X265_TCOMPATTERN_H
 #define X265_TCOMPATTERN_H
 
-#include "CommonDef.h"
+#include "common.h"
+#include "TComRom.h"
 
 //! \ingroup TLibCommon
 //! \{
@@ -52,15 +53,28 @@
 
 class TComDataCU;
 
+struct IntraNeighbors
+{
+    int  numIntraNeighbor;
+    int  totalUnits;
+    int  aboveUnits;
+    int  leftUnits;
+    int  unitWidth;
+    int  unitHeight;
+    int  tuSize;
+    uint32_t log2TrSize;
+    bool bNeighborFlags[4 * MAX_NUM_SPU_W + 1];
+};
+
 /// neighboring pixel access class for all components
 class TComPattern
 {
 public:
 
     // access functions of ADI buffers
-    static pixel* getAdiChromaBuf(int chromaId, int cuHeight, pixel* adiBuf)
+    static pixel* getAdiChromaBuf(uint32_t chromaId, int tuSize, pixel* adiBuf)
     {
-        return adiBuf + (chromaId == 0 ? 0 : 2 * ADI_BUF_STRIDE * (cuHeight * 2 + 1));
+        return adiBuf + (chromaId == 1 ? 0 : 2 * ADI_BUF_STRIDE * (tuSize * 2 + 1));
     }
 
     // -------------------------------------------------------------------------------------------------------------------
@@ -69,21 +83,19 @@
 
     /// set parameters from pixel buffers for accessing neighboring pixels
     static void initAdiPattern(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf,
-                               int strideOrig, int heightOrig, pixel* refAbove, pixel* refLeft,
-                               pixel* refAboveFlt, pixel* refLeftFlt);
-
-    /// set luma parameters from CU data for accessing ADI data
-    static void  initAdiPattern(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, pixel* adiBuf,
-                                int strideOrig, int heightOrig);
+                               pixel* refAbove, pixel* refLeft,
+                               pixel* refAboveFlt, pixel* refLeftFlt, int dirMode);
 
     /// set chroma parameters from CU data for accessing ADI data
-    static void  initAdiPatternChroma(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth,
-                                      pixel* adiBuf, int strideOrig, int heightOrig, int chromaId);
+    static void initAdiPatternChroma(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth,
+                                     pixel* adiBuf, uint32_t chromaId);
+
+    static void initIntraNeighbors(TComDataCU* cu, uint32_t zOrderIdxInPart, uint32_t partDepth, bool isLuma, IntraNeighbors *IntraNeighbors);
 
 private:
 
     /// padding of unavailable reference samples for intra prediction
-    static void fillReferenceSamples(pixel* roiOrigin, pixel* adiTemp, bool* bNeighborFlags, int numIntraNeighbor, int unitSize, int numUnitsInCU, int totalUnits, uint32_t cuWidth, uint32_t cuHeight, uint32_t width, uint32_t height, int picStride);
+    static void fillReferenceSamples(pixel* adiOrigin, int picStride, pixel* adiRef, const IntraNeighbors& intraNeighbors);
 
     /// constrained intra prediction
     static bool  isAboveLeftAvailable(TComDataCU* cu, uint32_t partIdxLT);

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPicSym.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPicSym.cpp Changed

@@ -36,63 +36,46 @@
 */
 
 #include "TComPicSym.h"
-#include "TComSampleAdaptiveOffset.h"
+#include "TComPicYuv.h"
 
 using namespace x265;
 
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Constructor / destructor / create / destroy
-// ====================================================================================================================
-
 TComPicSym::TComPicSym()
     : m_widthInCU(0)
     , m_heightInCU(0)
-    , m_maxCUSize(0)
-    , m_minCUSize(0)
-    , m_totalDepth(0)
     , m_numPartitions(0)
-    , m_numPartInWidth(0)
-    , m_numPartInHeight(0)
+    , m_numPartInCUSize(0)
     , m_numCUsInFrame(0)
     , m_slice(NULL)
     , m_cuData(NULL)
-{}
+{
+    m_saoParam = NULL;
+}
 
-bool TComPicSym::create(int picWidth, int picHeight, int picCsp, uint32_t maxCUSize, uint32_t maxDepth)
+bool TComPicSym::create(x265_param *param)
 {
     uint32_t i;
 
-    m_saoParam        = NULL;
-    m_totalDepth      = maxDepth;
-    m_numPartitions   = 1 << (m_totalDepth << 1);
+    m_numPartitions   = 1 << (g_maxFullDepth * 2);
+    m_numPartInCUSize = 1 << g_maxFullDepth;
 
-    m_maxCUSize       = maxCUSize;
-
-    m_minCUSize       = maxCUSize  >> m_totalDepth;
-
-    m_numPartInWidth  = m_maxCUSize / m_minCUSize;
-    m_numPartInHeight = m_maxCUSize / m_minCUSize;
-
-    m_widthInCU       = (picWidth % m_maxCUSize) ? picWidth / m_maxCUSize  + 1 : picWidth / m_maxCUSize;
-    m_heightInCU      = (picHeight % m_maxCUSize) ? picHeight / m_maxCUSize + 1 : picHeight / m_maxCUSize;
+    m_widthInCU       = (param->sourceWidth  + g_maxCUSize - 1) >> g_maxLog2CUSize;
+    m_heightInCU      = (param->sourceHeight + g_maxCUSize - 1) >> g_maxLog2CUSize;
 
     m_numCUsInFrame   = m_widthInCU * m_heightInCU;
 
-    m_slice = new TComSlice;
-    m_cuData = new TComDataCU*[m_numCUsInFrame];
-    if (!m_slice || !m_cuData)
-        return false;
+    m_slice = new Slice;
+    m_cuData = new TComDataCU[m_numCUsInFrame];
 
+    bool tqBypass = param->bCULossless || param->bLossless;
     for (i = 0; i < m_numCUsInFrame; i++)
     {
-        m_cuData[i] = new TComDataCU;
-        if (!m_cuData[i])
-            return false;
-        if (!m_cuData[i]->create(m_numPartitions, m_maxCUSize, m_maxCUSize, m_maxCUSize >> m_totalDepth, picCsp))
+        uint32_t sizeL = 1 << (g_maxLog2CUSize * 2);
+        uint32_t sizeC = sizeL >> (CHROMA_H_SHIFT(param->internalCsp) + CHROMA_V_SHIFT(param->internalCsp));
+        if (!m_cuData[i].initialize(m_numPartitions, sizeL, sizeC, 1, tqBypass))
             return false;
+
+        m_cuData[i].create(&m_cuData[i], m_numPartitions, g_maxCUSize, param->internalCsp, 0, tqBypass);
     }
 
     return true;
@@ -101,29 +84,13 @@
 void TComPicSym::destroy()
 {
     delete m_slice;
-    m_slice = NULL;
-
-    for (int i = 0; i < m_numCUsInFrame; i++)
-    {
-        m_cuData[i]->destroy();
-        delete m_cuData[i];
-    }
 
-    delete [] m_cuData;
-    m_cuData = NULL;
-
-    if (m_saoParam)
+    if (m_cuData)
     {
-        TComSampleAdaptiveOffset::freeSaoParam(m_saoParam);
-        delete m_saoParam;
-        m_saoParam = NULL;
+        for (int i = 0; i < m_numCUsInFrame; i++)
+            m_cuData[i].destroy();
+        delete[] m_cuData;
     }
-}
 
-void TComPicSym::allocSaoParam(TComSampleAdaptiveOffset *sao)
-{
-    m_saoParam = new SAOParam;
-    sao->allocSaoParam(m_saoParam);
+    delete m_saoParam;
 }
-
-//! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPicSym.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPicSym.h Changed

@@ -39,75 +39,53 @@
 #define X265_TCOMPICSYM_H
 
 // Include files
-#include "CommonDef.h"
-#include "TComSlice.h"
+#include "common.h"
+#include "slice.h"
 #include "TComDataCU.h"
 
 namespace x265 {
 // private namespace
 
-struct SAOParam;
-class TComSampleAdaptiveOffset;
+class TComPicYuv;
 
-//! \ingroup TLibCommon
-//! \{
-
-// ====================================================================================================================
-// Class definition
-// ====================================================================================================================
-
-/// picture symbol class
 class TComPicSym
 {
-private:
+public:
 
     uint32_t      m_widthInCU;
     uint32_t      m_heightInCU;
 
-    uint32_t      m_maxCUSize;
-    uint32_t      m_minCUSize;
-
-    UChar         m_totalDepth;
     uint32_t      m_numPartitions;
-    uint32_t      m_numPartInWidth;
-    uint32_t      m_numPartInHeight;
+    uint32_t      m_numPartInCUSize;
     uint32_t      m_numCUsInFrame;
 
-    TComSlice*    m_slice;
-    TComDataCU**  m_cuData;
+    Slice*        m_slice;
+    TComDataCU*   m_cuData;
 
     SAOParam*     m_saoParam;
 
-public:
+    TComPicSym*   m_freeListNext;
+    TComPicYuv*   m_reconPicYuv;
 
-    bool        create(int picWidth, int picHeight, int picCsp, uint32_t maxCUSize, uint32_t maxDepth);
-    void        destroy();
+    bool          m_bHasReferences; // used during DPB/RPS updates
 
     TComPicSym();
 
-    TComSlice*  getSlice()                { return m_slice; }
-
-    uint32_t    getFrameWidthInCU()       { return m_widthInCU; }
-
-    uint32_t    getFrameHeightInCU()      { return m_heightInCU; }
-
-    uint32_t    getMinCUSize()            { return m_minCUSize; }
-
-    uint32_t    getNumberOfCUsInFrame()   { return m_numCUsInFrame; }
+    bool        create(x265_param *param);
+    void        destroy();
 
-    TComDataCU*&  getCU(uint32_t cuAddr)  { return m_cuData[cuAddr]; }
+    uint32_t    getFrameWidthInCU() const { return m_widthInCU; }
 
-    uint32_t    getNumPartition()         { return m_numPartitions; }
+    uint32_t    getFrameHeightInCU() const { return m_heightInCU; }
 
-    uint32_t    getNumPartInWidth()       { return m_numPartInWidth; }
+    uint32_t    getNumberOfCUsInFrame() const { return m_numCUsInFrame; }
 
-    uint32_t    getNumPartInHeight()      { return m_numPartInHeight; }
+    TComDataCU* getCU(uint32_t cuAddr)    { return &m_cuData[cuAddr]; }
 
-    void allocSaoParam(TComSampleAdaptiveOffset *sao);
+    uint32_t    getNumPartition() const   { return m_numPartitions; }
 
-    SAOParam *getSaoParam()               { return m_saoParam; }
-}; // END CLASS DEFINITION TComPicSym
+    uint32_t    getNumPartInCUSize() const { return m_numPartInCUSize; }
+};
 }
-//! \}
 
 #endif // ifndef X265_TCOMPICSYM_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPicYuv.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPicYuv.cpp Changed

@@ -46,13 +46,13 @@
 
 TComPicYuv::TComPicYuv()
 {
-    m_picBufY = NULL; // Buffer (including margin)
-    m_picBufU = NULL;
-    m_picBufV = NULL;
+    m_picBuf[0] = NULL; // Buffer (including margin)
+    m_picBuf[1] = NULL;
+    m_picBuf[2] = NULL;
 
-    m_picOrgY = NULL;  // m_apiPicBufY + m_iMarginLuma*getStride() + m_iMarginLuma
-    m_picOrgU = NULL;
-    m_picOrgV = NULL;
+    m_picOrg[0] = NULL;  // m_apiPicBufY + m_iMarginLuma*getStride() + m_iMarginLuma
+    m_picOrg[1] = NULL;
+    m_picOrg[2] = NULL;
 
     m_cuOffsetY = NULL;
     m_cuOffsetC = NULL;
@@ -60,11 +60,7 @@
     m_buOffsetC = NULL;
 }
 
-TComPicYuv::~TComPicYuv()
-{
-}
-
-bool TComPicYuv::create(int picWidth, int picHeight, int picCsp, uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxCUDepth)
+bool TComPicYuv::create(int picWidth, int picHeight, int picCsp, uint32_t maxCUSize, uint32_t maxFullDepth)
 {
     m_picWidth  = picWidth;
     m_picHeight = picHeight;
@@ -73,11 +69,10 @@
     m_picCsp = picCsp;
 
     // --> After config finished!
-    m_cuWidth  = maxCUWidth;
-    m_cuHeight = maxCUHeight;
+    m_cuSize = maxCUSize;
 
-    m_numCuInWidth = (m_picWidth + m_cuWidth - 1)  / m_cuWidth;
-    m_numCuInHeight = (m_picHeight + m_cuHeight - 1) / m_cuHeight;
+    m_numCuInWidth = (m_picWidth + m_cuSize - 1)  / m_cuSize;
+    m_numCuInHeight = (m_picHeight + m_cuSize - 1) / m_cuSize;
 
     m_lumaMarginX = g_maxCUSize + 32; // search margin and 8-tap filter half-length, padded for 32-byte alignment
     m_lumaMarginY = g_maxCUSize + 16; // margin for 8-tap filter and infinite padding
@@ -88,14 +83,15 @@
 
     m_strideC = ((m_numCuInWidth * g_maxCUSize) >> m_hChromaShift) + (m_chromaMarginX * 2);
     int maxHeight = m_numCuInHeight * g_maxCUSize;
+    uint32_t numPartitions = 1 << (maxFullDepth * 2);
 
-    CHECKED_MALLOC(m_picBufY, pixel, m_stride * (maxHeight + (m_lumaMarginY * 2)));
-    CHECKED_MALLOC(m_picBufU, pixel, m_strideC * ((maxHeight >> m_vChromaShift) + (m_chromaMarginY * 2)));
-    CHECKED_MALLOC(m_picBufV, pixel, m_strideC * ((maxHeight >> m_vChromaShift) + (m_chromaMarginY * 2)));
+    CHECKED_MALLOC(m_picBuf[0], pixel, m_stride * (maxHeight + (m_lumaMarginY * 2)));
+    CHECKED_MALLOC(m_picBuf[1], pixel, m_strideC * ((maxHeight >> m_vChromaShift) + (m_chromaMarginY * 2)));
+    CHECKED_MALLOC(m_picBuf[2], pixel, m_strideC * ((maxHeight >> m_vChromaShift) + (m_chromaMarginY * 2)));
 
-    m_picOrgY = m_picBufY + m_lumaMarginY   * getStride()  + m_lumaMarginX;
-    m_picOrgU = m_picBufU + m_chromaMarginY * getCStride() + m_chromaMarginX;
-    m_picOrgV = m_picBufV + m_chromaMarginY * getCStride() + m_chromaMarginX;
+    m_picOrg[0] = m_picBuf[0] + m_lumaMarginY   * getStride()  + m_lumaMarginX;
+    m_picOrg[1] = m_picBuf[1] + m_chromaMarginY * getCStride() + m_chromaMarginX;
+    m_picOrg[2] = m_picBuf[2] + m_chromaMarginY * getCStride() + m_chromaMarginX;
 
     /* TODO: these four buffers are the same for every TComPicYuv in the encoder */
     CHECKED_MALLOC(m_cuOffsetY, int, m_numCuInWidth * m_numCuInHeight);
@@ -104,20 +100,19 @@
     {
         for (int cuCol = 0; cuCol < m_numCuInWidth; cuCol++)
         {
-            m_cuOffsetY[cuRow * m_numCuInWidth + cuCol] = getStride() * cuRow * m_cuHeight + cuCol * m_cuWidth;
-            m_cuOffsetC[cuRow * m_numCuInWidth + cuCol] = getCStride() * cuRow * (m_cuHeight >> m_vChromaShift) + cuCol * (m_cuWidth >> m_hChromaShift);
+            m_cuOffsetY[cuRow * m_numCuInWidth + cuCol] = getStride() * cuRow * m_cuSize + cuCol * m_cuSize;
+            m_cuOffsetC[cuRow * m_numCuInWidth + cuCol] = getCStride() * cuRow * (m_cuSize >> m_vChromaShift) + cuCol * (m_cuSize >> m_hChromaShift);
         }
     }
 
-    CHECKED_MALLOC(m_buOffsetY, int, (size_t)1 << (2 * maxCUDepth));
-    CHECKED_MALLOC(m_buOffsetC, int, (size_t)1 << (2 * maxCUDepth));
-    for (int buRow = 0; buRow < (1 << maxCUDepth); buRow++)
+    CHECKED_MALLOC(m_buOffsetY, int, (size_t)numPartitions);
+    CHECKED_MALLOC(m_buOffsetC, int, (size_t)numPartitions);
+    for (uint32_t idx = 0; idx < numPartitions; ++idx)
     {
-        for (int buCol = 0; buCol < (1 << maxCUDepth); buCol++)
-        {
-            m_buOffsetY[(buRow << maxCUDepth) + buCol] = getStride() * buRow * (maxCUHeight >> maxCUDepth) + buCol * (maxCUWidth  >> maxCUDepth);
-            m_buOffsetC[(buRow << maxCUDepth) + buCol] = getCStride() * buRow * ((maxCUHeight >> m_vChromaShift) >> maxCUDepth) + buCol * ((maxCUWidth >> m_hChromaShift) >> maxCUDepth);
-        }
+        int x = g_zscanToPelX[idx];
+        int y = g_zscanToPelY[idx];
+        m_buOffsetY[idx] = getStride() * y + x;
+        m_buOffsetC[idx] = getCStride() * (y >> m_vChromaShift) + (x >> m_hChromaShift);
     }
 
     return true;
@@ -128,9 +123,9 @@
 
 void TComPicYuv::destroy()
 {
-    X265_FREE(m_picBufY);
-    X265_FREE(m_picBufU);
-    X265_FREE(m_picBufV);
+    X265_FREE(m_picBuf[0]);
+    X265_FREE(m_picBuf[1]);
+    X265_FREE(m_picBuf[2]);
     X265_FREE(m_cuOffsetY);
     X265_FREE(m_cuOffsetC);
     X265_FREE(m_buOffsetY);
@@ -150,7 +145,7 @@
 
 /* Copy pixels from an x265_picture into internal TComPicYuv instance.
  * Shift pixels as necessary, mask off bits above X265_DEPTH for safety. */
-void TComPicYuv::copyFromPicture(const x265_picture& pic, int32_t *pad)
+void TComPicYuv::copyFromPicture(const x265_picture& pic, int padx, int pady)
 {
     /* m_picWidth is the width that is being encoded, padx indicates how many
      * of those pixels are padding to reach multiple of MinCU(4) size.
@@ -161,9 +156,6 @@
      *
      * The same applies to m_picHeight and pady */
 
-    int padx = pad[0];
-    int pady = pad[1];
-
     /* width and height - without padsize (input picture raw width and height) */
     int width = m_picWidth - padx;
     int height = m_picHeight - pady;
@@ -182,9 +174,6 @@
 
     if (pic.bitDepth < X265_DEPTH)
     {
-        /* 8bit input, 10bit internal depth. Do a simple up-shift of 2 bits */
-        assert(X265_DEPTH == 10);
-
         pixel *yPixel = getLumaAddr();
         pixel *uPixel = getCbAddr();
         pixel *vPixel = getCrAddr();
@@ -192,31 +181,11 @@
         uint8_t *yChar = (uint8_t*)pic.planes[0];
         uint8_t *uChar = (uint8_t*)pic.planes[1];
         uint8_t *vChar = (uint8_t*)pic.planes[2];
+        int shift = X265_MAX(0, X265_DEPTH - pic.bitDepth);
 
-        for (int r = 0; r < height; r++)
-        {
-            for (int c = 0; c < width; c++)
-            {
-                yPixel[c] = ((pixel)yChar[c]) << 2;
-            }
-
-            yPixel += getStride();
-            yChar += pic.stride[0] / sizeof(*yChar);
-        }
-
-        for (int r = 0; r < height >> m_vChromaShift; r++)
-        {
-            for (int c = 0; c < width >> m_hChromaShift; c++)
-            {
-                uPixel[c] = ((pixel)uChar[c]) << 2;
-                vPixel[c] = ((pixel)vChar[c]) << 2;
-            }
-
-            uPixel += getCStride();
-            vPixel += getCStride();
-            uChar += pic.stride[1] / sizeof(*uChar);
-            vChar += pic.stride[2] / sizeof(*vChar);
-        }
+        primitives.planecopy_cp(yChar, pic.stride[0] / sizeof(*yChar), yPixel, getStride(), width, height, shift);
+        primitives.planecopy_cp(uChar, pic.stride[1] / sizeof(*uChar), uPixel, getCStride(), width >> m_hChromaShift, height >> m_vChromaShift, shift);
+        primitives.planecopy_cp(vChar, pic.stride[2] / sizeof(*vChar), vPixel, getCStride(), width >> m_hChromaShift, height >> m_vChromaShift, shift);
     }
     else if (pic.bitDepth == 8)
     {
@@ -268,30 +237,10 @@
         int shift = X265_MAX(0, pic.bitDepth - X265_DEPTH);
 
         /* shift and mask pixels to final size */
-        for (int r = 0; r < height; r++)
-        {
-            for (int c = 0; c < width; c++)
-            {
-                yPixel[c] = (pixel)((yShort[c] >> shift) & mask);
-            }
-
-            yPixel += getStride();
-            yShort += pic.stride[0] / sizeof(*yShort);
-        }
 
-        for (int r = 0; r < height >> m_vChromaShift; r++)
-        {
-            for (int c = 0; c < width >> m_hChromaShift; c++)
-            {
-                uPixel[c] = (pixel)((uShort[c] >> shift) & mask);
-                vPixel[c] = (pixel)((vShort[c] >> shift) & mask);
-            }
-
-            uPixel += getCStride();
-            vPixel += getCStride();
-            uShort += pic.stride[1] / sizeof(*uShort);
-            vShort += pic.stride[2] / sizeof(*vShort);
-        }
+        primitives.planecopy_sp(yShort, pic.stride[0] / sizeof(*yShort), yPixel, getStride(), width, height, shift, mask);
+        primitives.planecopy_sp(uShort, pic.stride[1] / sizeof(*uShort), uPixel, getCStride(), width >> m_hChromaShift, height >> m_vChromaShift, shift, mask);
+        primitives.planecopy_sp(vShort, pic.stride[2] / sizeof(*vShort), vPixel, getCStride(), width >> m_hChromaShift, height >> m_vChromaShift, shift, mask);
     }
 
     /* extend the right edge if width was not multiple of the minimum CU size */

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPicYuv.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPicYuv.h Changed

@@ -38,7 +38,7 @@
 #ifndef X265_TCOMPICYUV_H
 #define X265_TCOMPICYUV_H
 
-#include "CommonDef.h"
+#include "common.h"
 #include "TComRom.h"
 #include "x265.h"
 #include "md5.h"
@@ -64,13 +64,9 @@
     //  YUV buffer
     // ------------------------------------------------------------------------------------------------
 
-    Pel*  m_picBufY;         ///< Buffer (including margin)
-    Pel*  m_picBufU;
-    Pel*  m_picBufV;
+    pixel*  m_picBuf[3];        ///< Buffer (including margin)
 
-    Pel*  m_picOrgY;          ///< m_apiPicBufY + m_iMarginLuma*getStride() + m_iMarginLuma
-    Pel*  m_picOrgU;
-    Pel*  m_picOrgV;
+    pixel*  m_picOrg[3];        ///< m_apiPicBufY + m_iMarginLuma*getStride() + m_iMarginLuma
 
     // ------------------------------------------------------------------------------------------------
     //  Parameter for general YUV buffer usage
@@ -81,8 +77,7 @@
     int   m_hChromaShift;
     int   m_vChromaShift;
 
-    int   m_cuWidth;           ///< Width of Coding Unit (CU)
-    int   m_cuHeight;          ///< Height of Coding Unit (CU)
+    int   m_cuSize;            ///< Width/Height of Coding Unit (CU)
     int32_t*  m_cuOffsetY;
     int32_t*  m_cuOffsetC;
     int32_t*  m_buOffsetY;
@@ -99,13 +94,13 @@
     int   m_numCuInHeight;
 
     TComPicYuv();
-    virtual ~TComPicYuv();
+    ~TComPicYuv() {}
 
     // ------------------------------------------------------------------------------------------------
     //  Memory management
     // ------------------------------------------------------------------------------------------------
 
-    bool  create(int picWidth, int picHeight, int csp, uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxCUDepth);
+    bool  create(int picWidth, int picHeight, int csp, uint32_t maxCUSize, uint32_t maxFullDepth);
     void  destroy();
 
     // ------------------------------------------------------------------------------------------------
@@ -132,43 +127,46 @@
     //  Access function for picture buffer
     // ------------------------------------------------------------------------------------------------
 
-    //  Access starting position of picture buffer with margin
-    Pel*  getBufY()     { return m_picBufY; }
+    //  Access starting position of original picture
+    pixel*  getLumaAddr()   { return m_picOrg[0]; }
 
-    Pel*  getBufU()     { return m_picBufU; }
+    pixel*  getCbAddr()     { return m_picOrg[1]; }
 
-    Pel*  getBufV()     { return m_picBufV; }
+    pixel*  getCrAddr()     { return m_picOrg[2]; }
 
-    //  Access starting position of original picture
-    Pel*  getLumaAddr()   { return m_picOrgY; }
+    pixel*  getChromaAddr(uint32_t chromaId)     { return m_picOrg[chromaId]; }
+
+    //  Access starting position of original picture for specific coding unit (CU) or partition unit (PU)
+    pixel*  getLumaAddr(int cuAddr) { return m_picOrg[0] + m_cuOffsetY[cuAddr]; }
 
-    Pel*  getCbAddr()     { return m_picOrgU; }
+    pixel*  getPlaneAddr(int plane, int cuAddr) { return m_picOrg[plane] + (plane ? m_cuOffsetC[cuAddr] : m_cuOffsetY[cuAddr]); }
 
-    Pel*  getCrAddr()     { return m_picOrgV; }
+    pixel*  getCbAddr(int cuAddr) { return m_picOrg[1] + m_cuOffsetC[cuAddr]; }
 
-    //  Access starting position of original picture for specific coding unit (CU) or partition unit (PU)
-    Pel*  getLumaAddr(int cuAddr) { return m_picOrgY + m_cuOffsetY[cuAddr]; }
+    pixel*  getCrAddr(int cuAddr) { return m_picOrg[2] + m_cuOffsetC[cuAddr]; }
+
+    pixel*  getChromaAddr(uint32_t chromaId, int cuAddr) { return m_picOrg[chromaId] + m_cuOffsetC[cuAddr]; }
 
-    Pel*  getCbAddr(int cuAddr) { return m_picOrgU + m_cuOffsetC[cuAddr]; }
+    pixel*  getLumaAddr(int cuAddr, int absZOrderIdx) { return m_picOrg[0] + m_cuOffsetY[cuAddr] + m_buOffsetY[absZOrderIdx]; }
 
-    Pel*  getCrAddr(int cuAddr) { return m_picOrgV + m_cuOffsetC[cuAddr]; }
+    pixel*  getCbAddr(int cuAddr, int absZOrderIdx) { return m_picOrg[1] + m_cuOffsetC[cuAddr] + m_buOffsetC[absZOrderIdx]; }
 
-    Pel*  getLumaAddr(int cuAddr, int absZOrderIdx) { return m_picOrgY + m_cuOffsetY[cuAddr] + m_buOffsetY[g_zscanToRaster[absZOrderIdx]]; }
+    pixel*  getCrAddr(int cuAddr, int absZOrderIdx) { return m_picOrg[2] + m_cuOffsetC[cuAddr] + m_buOffsetC[absZOrderIdx]; }
 
-    Pel*  getCbAddr(int cuAddr, int absZOrderIdx) { return m_picOrgU + m_cuOffsetC[cuAddr] + m_buOffsetC[g_zscanToRaster[absZOrderIdx]]; }
+    pixel*  getChromaAddr(uint32_t chromaId, int cuAddr, int absZOrderIdx) { return m_picOrg[chromaId] + m_cuOffsetC[cuAddr] + m_buOffsetC[absZOrderIdx]; }
 
-    Pel*  getCrAddr(int cuAddr, int absZOrderIdx) { return m_picOrgV + m_cuOffsetC[cuAddr] + m_buOffsetC[g_zscanToRaster[absZOrderIdx]]; }
+    int32_t getChromaAddrOffset(int cuAddr, int absZOrderIdx) { return m_cuOffsetC[cuAddr] + m_buOffsetC[absZOrderIdx]; }
 
     uint32_t getCUHeight(int rowNum);
 
-    void  copyFromPicture(const x265_picture&, int32_t *pad);
+    void  copyFromPicture(const x265_picture&, int padx, int pady);
 }; // END CLASS DEFINITION TComPicYuv
 
-void updateChecksum(const Pel* plane, uint32_t& checksumVal, uint32_t height, uint32_t width, uint32_t stride, int row, uint32_t cuHeight);
-void updateCRC(const Pel* plane, uint32_t& crcVal, uint32_t height, uint32_t width, uint32_t stride);
-void crcFinish(uint32_t & crc, UChar digest[16]);
-void checksumFinish(uint32_t & checksum, UChar digest[16]);
-void updateMD5Plane(MD5Context& md5, const Pel* plane, uint32_t width, uint32_t height, uint32_t stride);
+void updateChecksum(const pixel* plane, uint32_t& checksumVal, uint32_t height, uint32_t width, uint32_t stride, int row, uint32_t cuHeight);
+void updateCRC(const pixel* plane, uint32_t& crcVal, uint32_t height, uint32_t width, uint32_t stride);
+void crcFinish(uint32_t & crc, uint8_t digest[16]);
+void checksumFinish(uint32_t checksum, uint8_t digest[16]);
+void updateMD5Plane(MD5Context& md5, const pixel* plane, uint32_t width, uint32_t height, uint32_t stride);
 }
 //! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComPicYuvMD5.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComPicYuvMD5.cpp Changed

@@ -43,14 +43,14 @@
  * OUTBIT_BITDEPTH_DIV8.
  */
 template<uint32_t OUTPUT_BITDEPTH_DIV8>
-static void md5_block(MD5Context& md5, const Pel* plane, uint32_t n)
+static void md5_block(MD5Context& md5, const pixel* plane, uint32_t n)
 {
-    /* create a 64 byte buffer for packing Pel's into */
-    UChar buf[64 / OUTPUT_BITDEPTH_DIV8][OUTPUT_BITDEPTH_DIV8];
+    /* create a 64 byte buffer for packing pixel's into */
+    uint8_t buf[64 / OUTPUT_BITDEPTH_DIV8][OUTPUT_BITDEPTH_DIV8];
 
     for (uint32_t i = 0; i < n; i++)
     {
-        Pel pel = plane[i];
+        pixel pel = plane[i];
         /* perform bitdepth and endian conversion */
         for (uint32_t d = 0; d < OUTPUT_BITDEPTH_DIV8; d++)
         {
@@ -58,7 +58,7 @@
         }
     }
 
-    MD5Update(&md5, (UChar*)buf, n * OUTPUT_BITDEPTH_DIV8);
+    MD5Update(&md5, (uint8_t*)buf, n * OUTPUT_BITDEPTH_DIV8);
 }
 
 /**
@@ -66,7 +66,7 @@
  * is adjusted to OUTBIT_BITDEPTH_DIV8.
  */
 template<uint32_t OUTPUT_BITDEPTH_DIV8>
-static void md5_plane(MD5Context& md5, const Pel* plane, uint32_t width, uint32_t height, uint32_t stride)
+static void md5_plane(MD5Context& md5, const pixel* plane, uint32_t width, uint32_t height, uint32_t stride)
 {
     /* N is the number of samples to process per md5 update.
      * All N samples must fit in buf */
@@ -88,7 +88,7 @@
     }
 }
 
-void updateCRC(const Pel* plane, uint32_t& crcVal, uint32_t height, uint32_t width, uint32_t stride)
+void updateCRC(const pixel* plane, uint32_t& crcVal, uint32_t height, uint32_t width, uint32_t stride)
 {
     uint32_t crcMsb;
     uint32_t bitVal;
@@ -123,7 +123,7 @@
     }
 }
 
-void crcFinish(uint32_t& crcVal, UChar digest[16])
+void crcFinish(uint32_t& crcVal, uint8_t digest[16])
 {
     uint32_t crcMsb;
 
@@ -137,9 +137,9 @@
     digest[1] =  crcVal        & 0xff;
 }
 
-void updateChecksum(const Pel* plane, uint32_t& checksumVal, uint32_t height, uint32_t width, uint32_t stride, int row, uint32_t cuHeight)
+void updateChecksum(const pixel* plane, uint32_t& checksumVal, uint32_t height, uint32_t width, uint32_t stride, int row, uint32_t cuHeight)
 {
-    UChar xor_mask;
+    uint8_t xor_mask;
 
     for (uint32_t y = row * cuHeight; y < ((row * cuHeight) + height); y++)
     {
@@ -156,7 +156,7 @@
     }
 }
 
-void checksumFinish(uint32_t& checksum, UChar digest[16])
+void checksumFinish(uint32_t checksum, uint8_t digest[16])
 {
     digest[0] = (checksum >> 24) & 0xff;
     digest[1] = (checksum >> 16) & 0xff;
@@ -164,10 +164,10 @@
     digest[3] =  checksum        & 0xff;
 }
 
-void updateMD5Plane(MD5Context& md5, const Pel* plane, uint32_t width, uint32_t height, uint32_t stride)
+void updateMD5Plane(MD5Context& md5, const pixel* plane, uint32_t width, uint32_t height, uint32_t stride)
 {
     /* choose an md5_plane packing function based on the system bitdepth */
-    typedef void (*MD5PlaneFunc)(MD5Context&, const Pel*, uint32_t, uint32_t, uint32_t);
+    typedef void (*MD5PlaneFunc)(MD5Context&, const pixel*, uint32_t, uint32_t, uint32_t);
     MD5PlaneFunc md5_plane_func;
     md5_plane_func = X265_DEPTH <= 8 ? (MD5PlaneFunc)md5_plane<1> : (MD5PlaneFunc)md5_plane<2>;

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComRom.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComRom.cpp Changed

@@ -42,88 +42,54 @@
 namespace x265 {
 //! \ingroup TLibCommon
 //! \{
-// scanning order table
-uint16_t* g_scanOrder[SCAN_NUMBER_OF_GROUP_TYPES][SCAN_NUMBER_OF_TYPES][MAX_CU_DEPTH];
 
-class ScanGenerator
+// lambda = pow(2, (double)q / 6 - 2);
+double x265_lambda_tab[QP_MAX_MAX + 1] =
 {
-private:
-
-    uint32_t m_line, m_column;
-    uint32_t m_blockWidth, m_blockHeight;
-    uint32_t m_stride;
-    COEFF_SCAN_TYPE m_scanType;
-
-public:
-
-    ScanGenerator(uint32_t blockWidth, uint32_t blockHeight, uint32_t stride, COEFF_SCAN_TYPE scanType)
-        : m_line(0), m_column(0), m_blockWidth(blockWidth), m_blockHeight(blockHeight), m_stride(stride), m_scanType(scanType)
-    { }
-
-    uint32_t GetCurrentX() const { return m_column; }
+    0.2500, 0.2806, 0.3150, 0.3536, 0.3969,
+    0.4454, 0.5000, 0.5612, 0.6300, 0.7071,
+    0.7937, 0.8909, 1.0000, 1.1225, 1.2599,
+    1.4142, 1.5874, 1.7818, 2.0000, 2.2449,
+    2.5198, 2.8284, 3.1748, 3.5636, 4.0000,
+    4.4898, 5.0397, 5.6569, 6.3496, 7.1272,
+    8.0000, 8.9797, 10.0794, 11.3137, 12.6992,
+    14.2544, 16.0000, 17.9594, 20.1587, 22.6274,
+    25.3984, 28.5088, 32.0000, 35.9188, 40.3175,
+    45.2548, 50.7968, 57.0175, 64.0000, 71.8376,
+    80.6349, 90.5097, 101.5937, 114.0350, 128.0000,
+    143.6751, 161.2699, 181.0193, 203.1873, 228.0701,
+    256.0000, 287.3503, 322.5398, 362.0387, 406.3747,
+    456.1401, 512.0000, 574.7006, 645.0796, 724.0773
+};
 
-    uint32_t GetCurrentY() const { return m_line; }
+// lambda2 = pow(lambda, 2) * scale (0.85);
+double x265_lambda2_tab[QP_MAX_MAX + 1] =
+{
+    0.0531, 0.0669, 0.0843, 0.1063, 0.1339,
+    0.1687, 0.2125, 0.2677, 0.3373, 0.4250,
+    0.5355, 0.6746, 0.8500, 1.0709, 1.3493,
+    1.7000, 2.1419, 2.6986, 3.4000, 4.2837,
+    5.3970, 6.8000, 8.5675, 10.7943, 13.6000,
+    17.1345, 21.5887, 27.2004, 34.2699, 43.1773,
+    54.4000, 68.5397, 86.3551, 108.7998, 137.0792,
+    172.7097, 217.6000, 274.1590, 345.4172, 435.1993,
+    548.3169, 690.8389, 870.4000, 1096.6362, 1381.6757,
+    1740.7974, 2193.2676, 2763.3460, 3481.6000, 4386.5446,
+    5526.6890, 6963.2049, 8773.0879, 11053.3840, 13926.4000,
+    17546.1542, 22106.7835, 27852.7889, 35092.3170, 44213.5749,
+    55705.6000, 70184.6657, 88427.1342, 111411.2172, 140369.3373,
+    176854.2222, 222822.4000, 280738.6627, 353708.5368, 445644.7459
+};
 
-    uint32_t GetNextIndex(uint32_t blockOffsetX, uint32_t blockOffsetY)
-    {
-        int rtn = ((m_line + blockOffsetY) * m_stride) + m_column + blockOffsetX;
-
-        //advance line and column to the next position
-        switch (m_scanType)
-        {
-        case SCAN_DIAG:
-        {
-            if ((m_column == (m_blockWidth - 1)) || (m_line == 0))     //if we reach the end of a rank, go diagonally down to the next one
-            {
-                m_line   += m_column + 1;
-                m_column  = 0;
-
-                if (m_line >= m_blockHeight)     //if that takes us outside the block, adjust so that we are back on the bottom row
-                {
-                    m_column += m_line - (m_blockHeight - 1);
-                    m_line    = m_blockHeight - 1;
-                }
-            }
-            else
-            {
-                m_column++;
-                m_line--;
-            }
-        }
-        break;
-
-        case SCAN_HOR:
-        {
-            if (m_column == (m_blockWidth - 1))
-            {
-                m_line++;
-                m_column = 0;
-            }
-            else m_column++;
-        }
-        break;
-
-        case SCAN_VER:
-        {
-            if (m_line == (m_blockHeight - 1))
-            {
-                m_column++;
-                m_line = 0;
-            }
-            else m_line++;
-        }
-        break;
-
-        default:
-        {
-            std::cerr << "ERROR: Unknown scan type \"" << m_scanType << "\"in ScanGenerator::GetNextIndex" << std::endl;
-            exit(1);
-        }
-        break;
-        }
-
-        return rtn;
-    }
+const uint16_t x265_chroma_lambda2_offset_tab[MAX_CHROMA_LAMBDA_OFFSET+1] =
+{
+       16,    20,    25,    32,    40,    50,
+       64,    80,   101,   128,   161,   203,
+      256,   322,   406,   512,   645,   812,
+     1024,  1290,  1625,  2048,  2580,  3250,
+     4096,  5160,  6501,  8192, 10321, 13003,
+    16384, 20642, 26007, 32768, 41285, 52015,
+    65535
 };
 
 static int initialized /* = 0 */;
@@ -133,113 +99,72 @@
 {
     if (ATOMIC_CAS32(&initialized, 0, 1) == 1)
         return;
-
-    int i, c;
-
-    // g_aucConvertToBit[ x ]: log2(x/4), if x=4 -> 0, x=8 -> 1, x=16 -> 2, ...
-    ::memset(g_convertToBit, -1, sizeof(g_convertToBit));
-    c = 0;
-    for (i = 4; i <= MAX_CU_SIZE; i *= 2)
-    {
-        g_convertToBit[i] = c;
-        c++;
-    }
-
-    // initialise scan orders
-    for (uint32_t log2BlockSize = 0; log2BlockSize < MAX_CU_DEPTH; log2BlockSize++)
-    {
-        const uint32_t blockWidth  = 1 << log2BlockSize;
-        const uint32_t blockHeight = 1 << log2BlockSize;
-        const uint32_t totalValues = blockWidth * blockHeight;
-        //non-grouped scan orders
-        for (uint32_t scanTypeIndex = 0; scanTypeIndex < SCAN_NUMBER_OF_TYPES; scanTypeIndex++)
-        {
-            const COEFF_SCAN_TYPE scanType = COEFF_SCAN_TYPE(scanTypeIndex);
-            g_scanOrder[SCAN_UNGROUPED][scanType][log2BlockSize] = X265_MALLOC(uint16_t, totalValues);
-            ScanGenerator fullBlockScan(blockWidth, blockHeight, blockWidth, scanType);
-
-            for (uint32_t scanPosition = 0; scanPosition < totalValues; scanPosition++)
-            {
-                g_scanOrder[SCAN_UNGROUPED][scanType][log2BlockSize][scanPosition] = fullBlockScan.GetNextIndex(0, 0);
-            }
-        }
-
-        //grouped scan orders
-        const uint32_t  groupWidth           = 1 << MLS_CG_LOG2_SIZE;
-        const uint32_t  groupHeight          = 1 << MLS_CG_LOG2_SIZE;
-        const uint32_t  widthInGroups        = blockWidth  >> MLS_CG_LOG2_SIZE;
-        const uint32_t  heightInGroups       = blockHeight >> MLS_CG_LOG2_SIZE;
-
-        const uint32_t  groupSize            = groupWidth    * groupHeight;
-        const uint32_t  totalGroups          = widthInGroups * heightInGroups;
-
-        for (uint32_t scanTypeIndex = 0; scanTypeIndex < SCAN_NUMBER_OF_TYPES; scanTypeIndex++)
-        {
-            const COEFF_SCAN_TYPE scanType = COEFF_SCAN_TYPE(scanTypeIndex);
-
-            g_scanOrder[SCAN_GROUPED_4x4][scanType][log2BlockSize] = X265_MALLOC(uint16_t, totalValues);
-
-            ScanGenerator fullBlockScan(widthInGroups, heightInGroups, groupWidth, scanType);
-
-            for (uint32_t groupIndex = 0; groupIndex < totalGroups; groupIndex++)
-            {
-                const uint32_t groupPositionY  = fullBlockScan.GetCurrentY();
-                const uint32_t groupPositionX  = fullBlockScan.GetCurrentX();
-                const uint32_t groupOffsetX    = groupPositionX * groupWidth;
-                const uint32_t groupOffsetY    = groupPositionY * groupHeight;
-                const uint32_t groupOffsetScan = groupIndex     * groupSize;
-
-                ScanGenerator groupScan(groupWidth, groupHeight, blockWidth, scanType);
-
-                for (uint32_t scanPosition = 0; scanPosition < groupSize; scanPosition++)
-                {
-                    g_scanOrder[SCAN_GROUPED_4x4][scanType][log2BlockSize][groupOffsetScan + scanPosition] = groupScan.GetNextIndex(groupOffsetX, groupOffsetY);
-                }
-
-                fullBlockScan.GetNextIndex(0, 0);
-            }
-        }
-
-        //--------------------------------------------------------------------------------------------------
-    }
 }
 
 void destroyROM()
 {
     if (ATOMIC_CAS32(&initialized, 1, 0) == 0)
         return;
-
-    for (uint32_t groupTypeIndex = 0; groupTypeIndex < SCAN_NUMBER_OF_GROUP_TYPES; groupTypeIndex++)
-    {
-        for (uint32_t scanOrderIndex = 0; scanOrderIndex < SCAN_NUMBER_OF_TYPES; scanOrderIndex++)
-        {
-            for (uint32_t log2BlockSize = 0; log2BlockSize < MAX_CU_DEPTH; log2BlockSize++)
-            {
-                X265_FREE(g_scanOrder[groupTypeIndex][scanOrderIndex][log2BlockSize]);
-            }
-        }
-    }
 }
 
 // ====================================================================================================================
 // Data structure related table & variable
 // ====================================================================================================================
 
-uint32_t g_maxCUSize   = MAX_CU_SIZE;
-uint32_t g_maxCUDepth  = MAX_CU_DEPTH;
-uint32_t g_addCUDepth  = 0;
-uint32_t g_zscanToRaster[MAX_NUM_SPU_W * MAX_NUM_SPU_W] = { 0, };
-uint32_t g_rasterToZscan[MAX_NUM_SPU_W * MAX_NUM_SPU_W] = { 0, };
-uint32_t g_rasterToPelX[MAX_NUM_SPU_W * MAX_NUM_SPU_W] = { 0, };
-uint32_t g_rasterToPelY[MAX_NUM_SPU_W * MAX_NUM_SPU_W] = { 0, };
+uint32_t g_maxLog2CUSize = MAX_LOG2_CU_SIZE;
+uint32_t g_maxCUSize     = MAX_CU_SIZE;
+uint32_t g_maxFullDepth  = NUM_FULL_DEPTH - 1;
+uint32_t g_maxCUDepth    = NUM_CU_DEPTH - 1;
+uint32_t g_zscanToRaster[MAX_NUM_PARTITIONS] = { 0, };
+uint32_t g_rasterToZscan[MAX_NUM_PARTITIONS] = { 0, };
+
+const uint8_t g_zscanToPelX[MAX_NUM_PARTITIONS] =
+{
+    0, 4, 0, 4, 8, 12, 8, 12, 0, 4, 0, 4, 8, 12, 8, 12,
+    16, 20, 16, 20, 24, 28, 24, 28, 16, 20, 16, 20, 24, 28, 24, 28,
+    0, 4, 0, 4, 8, 12, 8, 12, 0, 4, 0, 4, 8, 12, 8, 12,
+    16, 20, 16, 20, 24, 28, 24, 28, 16, 20, 16, 20, 24, 28, 24, 28,
+    32, 36, 32, 36, 40, 44, 40, 44, 32, 36, 32, 36, 40, 44, 40, 44,
+    48, 52, 48, 52, 56, 60, 56, 60, 48, 52, 48, 52, 56, 60, 56, 60,
+    32, 36, 32, 36, 40, 44, 40, 44, 32, 36, 32, 36, 40, 44, 40, 44,
+    48, 52, 48, 52, 56, 60, 56, 60, 48, 52, 48, 52, 56, 60, 56, 60,
+    0, 4, 0, 4, 8, 12, 8, 12, 0, 4, 0, 4, 8, 12, 8, 12,
+    16, 20, 16, 20, 24, 28, 24, 28, 16, 20, 16, 20, 24, 28, 24, 28,
+    0, 4, 0, 4, 8, 12, 8, 12, 0, 4, 0, 4, 8, 12, 8, 12,
+    16, 20, 16, 20, 24, 28, 24, 28, 16, 20, 16, 20, 24, 28, 24, 28,
+    32, 36, 32, 36, 40, 44, 40, 44, 32, 36, 32, 36, 40, 44, 40, 44,
+    48, 52, 48, 52, 56, 60, 56, 60, 48, 52, 48, 52, 56, 60, 56, 60,
+    32, 36, 32, 36, 40, 44, 40, 44, 32, 36, 32, 36, 40, 44, 40, 44,
+    48, 52, 48, 52, 56, 60, 56, 60, 48, 52, 48, 52, 56, 60, 56, 60
+};
+
+const uint8_t g_zscanToPelY[MAX_NUM_PARTITIONS] =
+{
+    0, 0, 4, 4, 0, 0, 4, 4, 8, 8, 12, 12, 8, 8, 12, 12,
+    0, 0, 4, 4, 0, 0, 4, 4, 8, 8, 12, 12, 8, 8, 12, 12,
+    16, 16, 20, 20, 16, 16, 20, 20, 24, 24, 28, 28, 24, 24, 28, 28,
+    16, 16, 20, 20, 16, 16, 20, 20, 24, 24, 28, 28, 24, 24, 28, 28,
+    0, 0, 4, 4, 0, 0, 4, 4, 8, 8, 12, 12, 8, 8, 12, 12,
+    0, 0, 4, 4, 0, 0, 4, 4, 8, 8, 12, 12, 8, 8, 12, 12,
+    16, 16, 20, 20, 16, 16, 20, 20, 24, 24, 28, 28, 24, 24, 28, 28,
+    16, 16, 20, 20, 16, 16, 20, 20, 24, 24, 28, 28, 24, 24, 28, 28,
+    32, 32, 36, 36, 32, 32, 36, 36, 40, 40, 44, 44, 40, 40, 44, 44,
+    32, 32, 36, 36, 32, 32, 36, 36, 40, 40, 44, 44, 40, 40, 44, 44,
+    48, 48, 52, 52, 48, 48, 52, 52, 56, 56, 60, 60, 56, 56, 60, 60,
+    48, 48, 52, 52, 48, 48, 52, 52, 56, 56, 60, 60, 56, 56, 60, 60,
+    32, 32, 36, 36, 32, 32, 36, 36, 40, 40, 44, 44, 40, 40, 44, 44,
+    32, 32, 36, 36, 32, 32, 36, 36, 40, 40, 44, 44, 40, 40, 44, 44,
+    48, 48, 52, 52, 48, 48, 52, 52, 56, 56, 60, 60, 56, 56, 60, 60,
+    48, 48, 52, 52, 48, 48, 52, 52, 56, 56, 60, 60, 56, 56, 60, 60
+};
 
 const uint32_t g_puOffset[8] = { 0, 8, 4, 4, 2, 10, 1, 5 };
 
-void initZscanToRaster(int maxDepth, int depth, uint32_t startVal, uint32_t*& curIdx)
+void initZscanToRaster(uint32_t maxFullDepth, uint32_t depth, uint32_t startVal, uint32_t*& curIdx)
 {
-    int stride = 1 << (maxDepth - 1);
+    uint32_t stride = 1 << maxFullDepth;
 
-    if (depth == maxDepth)
+    if (depth > maxFullDepth)
     {
         curIdx[0] = startVal;
         curIdx++;
@@ -247,60 +172,23 @@
     else
     {
         int step = stride >> depth;
-        initZscanToRaster(maxDepth, depth + 1, startVal,                        curIdx);
-        initZscanToRaster(maxDepth, depth + 1, startVal + step,                 curIdx);
-        initZscanToRaster(maxDepth, depth + 1, startVal + step * stride,        curIdx);
-        initZscanToRaster(maxDepth, depth + 1, startVal + step * stride + step, curIdx);
+        initZscanToRaster(maxFullDepth, depth + 1, startVal,                        curIdx);
+        initZscanToRaster(maxFullDepth, depth + 1, startVal + step,                 curIdx);
+        initZscanToRaster(maxFullDepth, depth + 1, startVal + step * stride,        curIdx);
+        initZscanToRaster(maxFullDepth, depth + 1, startVal + step * stride + step, curIdx);
     }
 }
 
-void initRasterToZscan(uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxDepth)
+void initRasterToZscan(uint32_t maxFullDepth)
 {
-    uint32_t  minWidth  = maxCUWidth  >> (maxDepth - 1);
-    uint32_t  minHeight = maxCUHeight >> (maxDepth - 1);
-
-    uint32_t  numPartInWidth  = (uint32_t)maxCUWidth  / minWidth;
-    uint32_t  numPartInHeight = (uint32_t)maxCUHeight / minHeight;
+    uint32_t numPartitions = 1 << (maxFullDepth * 2);
 
-    for (uint32_t i = 0; i < numPartInWidth * numPartInHeight; i++)
+    for (uint32_t i = 0; i < numPartitions; i++)
     {
         g_rasterToZscan[g_zscanToRaster[i]] = i;
     }
 }
 
-void initRasterToPelXY(uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxDepth)
-{
-    uint32_t i;
-
-    uint32_t* tempX = &g_rasterToPelX[0];
-    uint32_t* tempY = &g_rasterToPelY[0];
-
-    uint32_t  minWidth  = maxCUWidth  >> (maxDepth - 1);
-    uint32_t  minHeight = maxCUHeight >> (maxDepth - 1);
-
-    uint32_t  numPartInWidth  = maxCUWidth  / minWidth;
-    uint32_t  numPartInHeight = maxCUHeight / minHeight;
-
-    tempX[0] = 0;
-    tempX++;
-    for (i = 1; i < numPartInWidth; i++)
-    {
-        tempX[0] = tempX[-1] + minWidth;
-        tempX++;
-    }
-
-    for (i = 1; i < numPartInHeight; i++)
-    {
-        memcpy(tempX, tempX - numPartInWidth, sizeof(uint32_t) * numPartInWidth);
-        tempX += numPartInWidth;
-    }
-
-    for (i = 1; i < numPartInWidth * numPartInHeight; i++)
-    {
-        tempY[i] = (i / numPartInWidth) * minWidth;
-    }
-}
-
 const int16_t g_lumaFilter[4][NTAPS_LUMA] =
 {
     {  0, 0,   0, 64,  0,   0, 0,  0 },
@@ -321,16 +209,6 @@
     { -2, 10, 58, -2 }
 };
 
-const int g_quantScales[6] =
-{
-    26214, 23302, 20560, 18396, 16384, 14564
-};
-
-const int g_invQuantScales[6] =
-{
-    40, 45, 51, 57, 64, 72
-};
-
 const int16_t g_t4[4][4] =
 {
     { 64, 64, 64, 64 },
@@ -406,105 +284,142 @@
     {  9, -25, 43, -57, 70, -80, 87, -90, 90, -87, 80, -70, 57, -43, 25, -9, -9, 25, -43, 57, -70, 80, -87, 90, -90, 87, -80, 70, -57, 43, -25,  9 },
     {  4, -13, 22, -31, 38, -46, 54, -61, 67, -73, 78, -82, 85, -88, 90, -90, 90, -90, 88, -85, 82, -78, 73, -67, 61, -54, 46, -38, 31, -22, 13, -4 }
 };
-const uint8_t g_chromaScale[NUM_CHROMA_FORMAT][chromaQPMappingTableSize] =
+
+const uint8_t g_chromaScale[chromaQPMappingTableSize] =
 {
-    { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 29, 30, 31, 32, 33, 33, 34, 34, 35, 35, 36, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51 },
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 40, 41, 42, 42, 43, 44, 44, 45, 45, 46, 47, 48, 49, 50, 51 },
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 51, 51, 51, 51, 51, 51 }
+    0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 29, 30, 31, 32, 33, 33, 34, 34, 35, 35, 36, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,
+    51, 51, 51, 51, 51, 51, 51, 51, 51, 51, 51, 51
 };
 
-// ====================================================================================================================
-// Misc.
-// ====================================================================================================================
-
-uint8_t g_convertToBit[MAX_CU_SIZE + 1];
+const uint8_t g_chroma422IntraAngleMappingTable[36] =
+{ 0, 1, 2, 2, 2, 2, 3, 5, 7, 8, 10, 12, 13, 15, 17, 18, 19, 20, 21, 22, 23, 23, 24, 24, 25, 25, 26, 27, 27, 28, 28, 29, 29, 30, 31, DM_CHROMA_IDX };
 
-#if ENC_DEC_TRACE
-FILE*  g_hTrace = NULL;
-const bool g_bEncDecTraceEnable  = true;
-const bool g_bEncDecTraceDisable = false;
-bool   g_HLSTraceEnable = true;
-bool   g_bJustDoIt = false;
-uint64_t g_nSymbolCounter = 0;
-#endif
+const uint8_t g_log2Size[MAX_CU_SIZE + 1] =
+{
+    0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3,
+    4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
+    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
+    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
+    6
+};
 
 // ====================================================================================================================
 // Scanning order & context model mapping
 // ====================================================================================================================
 
-const uint32_t g_minInGroup[10] = { 0, 1, 2, 3, 4, 6, 8, 12, 16, 24 };
-const uint8_t g_groupIdx[32]   = { 0, 1, 2, 3, 4, 4, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9 };
-
-// Rice parameters for absolute transform levels
-const uint8_t g_goRiceRange[5] = { 7, 14, 26, 46, 78 };
+const uint16_t g_scan2x2[][2*2] =
+{
+    { 0, 2, 1, 3 },
+    { 0, 1, 2, 3 },
+};
 
-//const uint8_t g_goRicePrefixLen[5] = { 8, 7, 6, 5, 4 };
+const uint16_t g_scan8x8[NUM_SCAN_TYPE][8 * 8] =
+{
+    { 0,   8,  1, 16,  9,  2, 24, 17, 10,  3, 25, 18, 11, 26, 19, 27, 32, 40, 33, 48, 41, 34, 56, 49, 42, 35, 57, 50, 43, 58, 51, 59,
+      4,  12,  5, 20, 13,  6, 28, 21, 14,  7, 29, 22, 15, 30, 23, 31, 36, 44, 37, 52, 45, 38, 60, 53, 46, 39, 61, 54, 47, 62, 55, 63 },
+    { 0,   1,  2,  3,  8,  9, 10, 11, 16, 17, 18, 19, 24, 25, 26, 27,  4,  5,  6,  7, 12, 13, 14, 15, 20, 21, 22, 23, 28, 29, 30, 31,
+      32, 33, 34, 35, 40, 41, 42, 43, 48, 49, 50, 51, 56, 57, 58, 59, 36, 37, 38, 39, 44, 45, 46, 47, 52, 53, 54, 55, 60, 61, 62, 63 },
+    { 0,   8, 16, 24,  1,  9, 17, 25,  2, 10, 18, 26,  3, 11, 19, 27, 32, 40, 48, 56, 33, 41, 49, 57, 34, 42, 50, 58, 35, 43, 51, 59,
+      4,  12, 20, 28,  5, 13, 21, 29,  6, 14, 22, 30,  7, 15, 23, 31, 36, 44, 52, 60, 37, 45, 53, 61, 38, 46, 54, 62, 39, 47, 55, 63 }
+};
 
-int g_quantTSDefault4x4[16] =
+const uint16_t g_scan4x4[NUM_SCAN_TYPE][4 * 4] =
 {
-    16, 16, 16, 16,
-    16, 16, 16, 16,
-    16, 16, 16, 16,
-    16, 16, 16, 16
+    { 0,  4,  1,  8,  5,  2, 12,  9,  6,  3, 13, 10,  7, 14, 11, 15 },
+    { 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15 },
+    { 0,  4,  8, 12,  1,  5,  9, 13,  2,  6, 10, 14,  3,  7, 11, 15 }
 };
 
-int g_quantIntraDefault8x8[64] =
+const uint16_t g_scan16x16[16 * 16] =
 {
-    16, 16, 16, 16, 17, 18, 21, 24,
-    16, 16, 16, 16, 17, 19, 22, 25,
-    16, 16, 17, 18, 20, 22, 25, 29,
-    16, 16, 18, 21, 24, 27, 31, 36,
-    17, 17, 20, 24, 30, 35, 41, 47,
-    18, 19, 22, 27, 35, 44, 54, 65,
-    21, 22, 25, 31, 41, 54, 70, 88,
-    24, 25, 29, 36, 47, 65, 88, 115
+    0, 16, 1, 32, 17, 2, 48, 33, 18, 3, 49, 34, 19, 50, 35, 51,
+    64, 80, 65, 96, 81, 66, 112, 97, 82, 67, 113, 98, 83, 114, 99, 115,
+    4, 20, 5, 36, 21, 6, 52, 37, 22, 7, 53, 38, 23, 54, 39, 55,
+    128, 144, 129, 160, 145, 130, 176, 161, 146, 131, 177, 162, 147, 178, 163, 179,
+    68, 84, 69, 100, 85, 70, 116, 101, 86, 71, 117, 102, 87, 118, 103, 119,
+    8, 24, 9, 40, 25, 10, 56, 41, 26, 11, 57, 42, 27, 58, 43, 59,
+    192,208, 193,224,209, 194,240,225,210, 195,241,226,211,242,227,243,
+    132, 148, 133, 164, 149, 134, 180, 165, 150, 135, 181, 166, 151, 182, 167, 183,
+    72, 88, 73, 104, 89, 74, 120, 105, 90, 75, 121, 106, 91, 122, 107, 123,
+    12, 28, 13, 44, 29, 14, 60, 45, 30, 15, 61, 46, 31, 62, 47, 63,
+    196,212, 197,228,213, 198,244,229,214, 199,245,230,215,246,231,247,
+    136, 152, 137, 168, 153, 138, 184, 169, 154, 139, 185, 170, 155, 186, 171, 187,
+    76, 92, 77, 108, 93, 78, 124, 109, 94, 79, 125, 110, 95, 126, 111, 127,
+    200,216,201,232,217,202,248,233,218,203,249,234,219,250,235,251,
+    140, 156, 141, 172, 157, 142, 188, 173, 158, 143, 189, 174, 159, 190, 175, 191,
+    204,220,205,236,221,206,252,237,222,207,253,238,223,254,239,255
 };
 
-int g_quantInterDefault8x8[64] =
+const uint16_t g_scan8x8diag[8 * 8] =
 {
-    16, 16, 16, 16, 17, 18, 20, 24,
-    16, 16, 16, 17, 18, 20, 24, 25,
-    16, 16, 17, 18, 20, 24, 25, 28,
-    16, 17, 18, 20, 24, 25, 28, 33,
-    17, 18, 20, 24, 25, 28, 33, 41,
-    18, 20, 24, 25, 28, 33, 41, 54,
-    20, 24, 25, 28, 33, 41, 54, 71,
-    24, 25, 28, 33, 41, 54, 71, 91
+    0,   8,  1, 16,  9,  2, 24, 17,
+    10,  3, 32, 25, 18, 11,  4, 40,
+    33, 26, 19, 12,  5, 48, 41, 34,
+    27, 20, 13,  6, 56, 49, 42, 35,
+    28, 21, 14,  7, 57, 50, 43, 36,
+    29, 22, 15, 58, 51, 44, 37, 30,
+    23, 59, 52, 45, 38, 31, 60, 53,
+    46, 39, 61, 54, 47, 62, 55, 63
 };
-const uint32_t g_scalingListSize[4] = { 16, 64, 256, 1024 };
-const uint32_t g_scalingListSizeX[4] = { 4, 8, 16,  32 };
-const uint32_t g_scalingListNum[SCALING_LIST_SIZE_NUM] = { 6, 6, 6, 6 };
 
-const int g_winUnitX[] = { 1, 2, 2, 1 };
-const int g_winUnitY[] = { 1, 2, 1, 1 };
+const uint16_t g_scan32x32[32 * 32] =
+{
+    0,32,1,64,33,2,96,65,34,3,97,66,35,98,67,99,128,160,129,192,161,130,224,193,162,131,225,194,163,226,195,227,
+    4,36,5,68,37,6,100,69,38,7,101,70,39,102,71,103,256,288,257,320,289,258,352,321,290,259,353,322,291,354,323,355,
+    132,164,133,196,165,134,228,197,166,135,229,198,167,230,199,231,8,40,9,72,41,10,104,73,42,11,105,74,43,106,75,107,
+    384,416,385,448,417,386,480,449,418,387,481,450,419,482,451,483,260,292,261,324,293,262,356,325,294,263,357,326,295,358,327,359,
+    136,168,137,200,169,138,232,201,170,139,233,202,171,234,203,235,12,44,13,76,45,14,108,77,46,15,109,78,47,110,79,111,
+    512,544,513,576,545,514,608,577,546,515,609,578,547,610,579,611,388,420,389,452,421,390,484,453,422,391,485,454,423,486,455,487,
+    264,296,265,328,297,266,360,329,298,267,361,330,299,362,331,363,140,172,141,204,173,142,236,205,174,143,237,206,175,238,207,239,
+    16,48,17,80,49,18,112,81,50,19,113,82,51,114,83,115,640,672,641,704,673,642,736,705,674,643,737,706,675,738,707,739,
+    516,548,517,580,549,518,612,581,550,519,613,582,551,614,583,615,392,424,393,456,425,394,488,457,426,395,489,458,427,490,459,491,
+    268,300,269,332,301,270,364,333,302,271,365,334,303,366,335,367,144,176,145,208,177,146,240,209,178,147,241,210,179,242,211,243,
+    20,52,21,84,53,22,116,85,54,23,117,86,55,118,87,119,768,800,769,832,801,770,864,833,802,771,865,834,803,866,835,867,
+    644,676,645,708,677,646,740,709,678,647,741,710,679,742,711,743,520,552,521,584,553,522,616,585,554,523,617,586,555,618,587,619,
+    396,428,397,460,429,398,492,461,430,399,493,462,431,494,463,495,272,304,273,336,305,274,368,337,306,275,369,338,307,370,339,371,
+    148,180,149,212,181,150,244,213,182,151,245,214,183,246,215,247,24,56,25,88,57,26,120,89,58,27,121,90,59,122,91,123,
+    896,928,897,960,929,898,992,961,930,899,993,962,931,994,963,995,772,804,773,836,805,774,868,837,806,775,869,838,807,870,839,871,
+    648,680,649,712,681,650,744,713,682,651,745,714,683,746,715,747,524,556,525,588,557,526,620,589,558,527,621,590,559,622,591,623,
+    400,432,401,464,433,402,496,465,434,403,497,466,435,498,467,499,276,308,277,340,309,278,372,341,310,279,373,342,311,374,343,375,
+    152,184,153,216,185,154,248,217,186,155,249,218,187,250,219,251,28,60,29,92,61,30,124,93,62,31,125,94,63,126,95,127,
+    900,932,901,964,933,902,996,965,934,903,997,966,935,998,967,999,776,808,777,840,809,778,872,841,810,779,873,842,811,874,843,875,
+    652,684,653,716,685,654,748,717,686,655,749,718,687,750,719,751,528,560,529,592,561,530,624,593,562,531,625,594,563,626,595,627,
+    404,436,405,468,437,406,500,469,438,407,501,470,439,502,471,503,280,312,281,344,313,282,376,345,314,283,377,346,315,378,347,379,
+    156,188,157,220,189,158,252,221,190,159,253,222,191,254,223,255,904,936,905,968,937,906,1000,969,938,907,1001,970,939,1002,971,1003,
+    780,812,781,844,813,782,876,845,814,783,877,846,815,878,847,879,656,688,657,720,689,658,752,721,690,659,753,722,691,754,723,755,
+    532,564,533,596,565,534,628,597,566,535,629,598,567,630,599,631,408,440,409,472,441,410,504,473,442,411,505,474,443,506,475,507,
+    284,316,285,348,317,286,380,349,318,287,381,350,319,382,351,383,908,940,909,972,941,910,1004,973,942,911,1005,974,943,1006,975,1007,
+    784,816,785,848,817,786,880,849,818,787,881,850,819,882,851,883,660,692,661,724,693,662,756,725,694,663,757,726,695,758,727,759,
+    536,568,537,600,569,538,632,601,570,539,633,602,571,634,603,635,412,444,413,476,445,414,508,477,446,415,509,478,447,510,479,511,
+    912,944,913,976,945,914,1008,977,946,915,1009,978,947,1010,979,1011,788,820,789,852,821,790,884,853,822,791,885,854,823,886,855,887,
+    664,696,665,728,697,666,760,729,698,667,761,730,699,762,731,763,540,572,541,604,573,542,636,605,574,543,637,606,575,638,607,639,
+    916,948,917,980,949,918,1012,981,950,919,1013,982,951,1014,983,1015,792,824,793,856,825,794,888,857,826,795,889,858,827,890,859,891,
+    668,700,669,732,701,670,764,733,702,671,765,734,703,766,735,767,920,952,921,984,953,922,1016,985,954,923,1017,986,955,1018,987,1019,
+    796,828,797,860,829,798,892,861,830,799,893,862,831,894,863,895,924,956,925,988,957,926,1020,989,958,927,1021,990,959,1022,991,1023
+};
 
-const double x265_lambda2_tab_I[MAX_MAX_QP + 1] =
+const uint16_t* const g_scanOrder[NUM_SCAN_TYPE][NUM_SCAN_SIZE] =
 {
-    0.012265625, 0.022265625, 0.028052813, 0.035344375, 0.04453125, 0.056105938, 0.070689063, 0.0890625,
-    0.112211563, 0.141377813, 0.178125, 0.224423438, 0.282755938, 0.35625, 0.448846875,
-    0.565511563, 0.7125, 0.89769375, 1.131023125, 1.425, 1.7953875, 2.262046563, 2.85,
-    3.590775, 4.524093125, 5.7, 7.7800125, 10.55621688, 14.25, 19.1508, 25.63652688, 34.2,
-    45.48315, 60.32124, 79.8, 105.3293997, 138.7388519, 182.4, 229.8095994, 289.5419519, 364.8,
-    459.6191991, 579.0839038, 729.6, 919.2383981, 1158.167808, 1459.2, 1838.476796, 2316.335615,
-    2918.4, 3676.953592, 4632.67123, 5836.799769, 7353.906601, 9265.341359, 11673.59815, 14707.81146,
-    18530.68052, 23347.19353, 29415.61942, 37061.35663, 46694.38151, 58831.23184, 74122.70446,
-    93388.75192, 117662.4497, 148245.3913, 186777.4817, 235324.8715, 296490.7474
+    { g_scan4x4[0], g_scan8x8[0], g_scan16x16, g_scan32x32 },
+    { g_scan4x4[1], g_scan8x8[1], g_scan16x16, g_scan32x32 },
+    { g_scan4x4[2], g_scan8x8[2], g_scan16x16, g_scan32x32 }
 };
 
-const double x265_lambda2_non_I[MAX_MAX_QP + 1] =
+const uint16_t* const g_scanOrderCG[NUM_SCAN_TYPE][NUM_SCAN_SIZE] =
 {
-    0.05231, 0.060686, 0.07646, 0.096333333, 0.151715667, 0.15292, 0.192666667, 0.242745, 0.382299,
-    0.385333333, 0.485489333, 0.611678333, 0.963333333, 0.970979, 1.223356667, 1.541333333, 2.427447667,
-    2.446714, 3.082666667, 3.883916667, 6.116785, 6.165333333, 7.767833333, 9.786856667, 15.41333333,
-    16.57137733, 22.183542, 29.5936, 39.357022, 52.19656867, 69.05173333, 91.14257667, 150.0651347,
-    157.8325333, 207.1422197, 271.4221573, 443.904, 447.4271947, 563.722941, 710.2464, 1118.567987,
-    1127.445883, 1420.4928, 1789.70878, 2818.614706, 2840.9856, 3579.41756, 4509.78353, 7102.464,
-    7158.83512, 9019.56706, 11363.9424, 14317.66967, 18039.13269, 22727.8821, 28635.33594, 36078.2611,
-    45455.7588, 57270.66508, 72156.51362, 90911.5068, 114541.3166, 144313.0101, 181822.992, 229082.6059,
-    288625.9859, 363645.9408, 458165.1574, 577251.9032, 727291.7952
+    { g_scan4x4[0], g_scan2x2[0], g_scan4x4[0], g_scan8x8diag },
+    { g_scan4x4[1], g_scan2x2[1], g_scan4x4[0], g_scan8x8diag },
+    { g_scan4x4[2], g_scan2x2[0], g_scan4x4[0], g_scan8x8diag }
 };
 
+const uint8_t g_minInGroup[10] = { 0, 1, 2, 3, 4, 6, 8, 12, 16, 24 };
+
+// Rice parameters for absolute transform levels
+const uint8_t g_goRiceRange[5] = { 7, 14, 26, 46, 78 };
+
+const int g_winUnitX[] = { 1, 2, 2, 1 };
+const int g_winUnitY[] = { 1, 2, 1, 1 };
+
 const uint8_t g_lpsTable[64][4] =
 {
     { 128, 176, 208, 240 },
@@ -580,5 +495,27 @@
     106,  110,  114,  118,  122,  126,  130,  135,  139,  143,  147,  152,  156,  161,  165,  170,
     175,  179,  184,  189,  194,  198,  203,  208,  214,  219,  224,  229,  234,  240,  245,  250
 };
+
+/* g_intraFilterFlags[dir] & trSize */
+const uint8_t g_intraFilterFlags[NUM_INTRA_MODE] =
+{
+    0x38, 0x00,
+    0x38, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x20, 0x00, 0x20, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30,
+    0x38, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30, 0x20, 0x00, 0x20, 0x30, 0x30, 0x30, 0x30, 0x30, 0x30,
+    0x38, 
+};
+
+/* Contains how much to increment shared depth buffer for different ctu sizes to get next best depth
+ * here, depth 0 = 64x64, depth 1 = 32x32, depth 2 = 16x16 and depth 3 = 8x8
+ * if ctu = 64, depth buffer size is 256 combination of depth values 0, 1, 2, 3
+ * if ctu = 32, depth buffer size is 64 combination of depth values 1, 2, 3
+ * if ctu = 16, depth buffer size is 16 combination of depth values 2, 3 */
+const uint32_t g_depthInc[3][4] =
+{
+    { 16,  4,  0, 0},
+    { 64, 16,  4, 1},
+    {256, 64, 16, 4}
+};
+
 }
 //! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComRom.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComRom.h Changed

@@ -38,82 +38,69 @@
 #ifndef X265_TCOMROM_H
 #define X265_TCOMROM_H
 
-#include "CommonDef.h"
-
-#include <iostream>
+#include "common.h"
 
 namespace x265 {
 // private namespace
 
-//! \ingroup TLibCommon
-//! \{
+#define NUM_CU_DEPTH            4                           // maximun number of CU depths
+#define NUM_FULL_DEPTH          5                           // maximun number of full depths
+#define MIN_LOG2_CU_SIZE        3                           // log2(minCUSize)
+#define MAX_LOG2_CU_SIZE        6                           // log2(maxCUSize)
+#define MIN_CU_SIZE             (1 << MIN_LOG2_CU_SIZE)     // minimum allowable size of CU
+#define MAX_CU_SIZE             (1 << MAX_LOG2_CU_SIZE)     // maximum allowable size of CU
+
+#define LOG2_UNIT_SIZE          2                           // log2(unitSize)
+#define UNIT_SIZE               (1 << LOG2_UNIT_SIZE)       // unit size of CU partition
+#define TMVP_UNIT_MASK          0xF0                        // mask for mapping index to CompressMV field
 
-// ====================================================================================================================
-// Macros
-// ====================================================================================================================
+#define MAX_NUM_PARTITIONS      256
 
-#define MAX_CU_DEPTH            6                           // log2(LCUSize)
-#define MAX_CU_SIZE             (1 << (MAX_CU_DEPTH))       // maximum allowable size of CU
 #define MIN_PU_SIZE             4
 #define MIN_TU_SIZE             4
 #define MAX_NUM_SPU_W           (MAX_CU_SIZE / MIN_PU_SIZE) // maximum number of SPU in horizontal line
 #define ADI_BUF_STRIDE          (2 * MAX_CU_SIZE + 1 + 15)  // alignment to 16 bytes
 
-// ====================================================================================================================
-// Initialize / destroy functions
-// ====================================================================================================================
+#define MAX_LOG2_TR_SIZE 5
+#define MAX_LOG2_TS_SIZE 2 // TODO: RExt
+#define MAX_TR_SIZE (1 << MAX_LOG2_TR_SIZE)
+#define MAX_TS_SIZE (1 << MAX_LOG2_TS_SIZE)
+
+#define SLFASE_CONSTANT 0x5f4e4a53
 
 void initROM();
 void destroyROM();
 
-// ====================================================================================================================
-static const int chromaQPMappingTableSize = 58;
+static const int chromaQPMappingTableSize = 70;
 
-extern const uint8_t g_chromaScale[NUM_CHROMA_FORMAT][chromaQPMappingTableSize];
-// Data structure related table & variable
-// ====================================================================================================================
+extern const uint8_t g_chromaScale[chromaQPMappingTableSize];
+extern const uint8_t g_chroma422IntraAngleMappingTable[36];
 
 // flexible conversion from relative to absolute index
-extern uint32_t g_zscanToRaster[MAX_NUM_SPU_W * MAX_NUM_SPU_W];
-extern uint32_t g_rasterToZscan[MAX_NUM_SPU_W * MAX_NUM_SPU_W];
-extern uint16_t*  g_scanOrder[SCAN_NUMBER_OF_GROUP_TYPES][SCAN_NUMBER_OF_TYPES][MAX_CU_DEPTH];
-void initZscanToRaster(int maxDepth, int depth, uint32_t startVal, uint32_t*& curIdx);
-void initRasterToZscan(uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxCUDepth);
+extern uint32_t g_zscanToRaster[MAX_NUM_PARTITIONS];
+extern uint32_t g_rasterToZscan[MAX_NUM_PARTITIONS];
 
-// conversion of partition index to picture pel position
-extern uint32_t g_rasterToPelX[MAX_NUM_SPU_W * MAX_NUM_SPU_W];
-extern uint32_t g_rasterToPelY[MAX_NUM_SPU_W * MAX_NUM_SPU_W];
+void initZscanToRaster(uint32_t maxFullDepth, uint32_t depth, uint32_t startVal, uint32_t*& curIdx);
+void initRasterToZscan(uint32_t maxFullDepth);
 
-void initRasterToPelXY(uint32_t maxCUWidth, uint32_t maxCUHeight, uint32_t maxCUDepth);
+// conversion of partition index to picture pel position
+extern const uint8_t g_zscanToPelX[MAX_NUM_PARTITIONS];
+extern const uint8_t g_zscanToPelY[MAX_NUM_PARTITIONS];
 
 // global variable (LCU width/height, max. CU depth)
+extern uint32_t g_maxLog2CUSize;
 extern uint32_t g_maxCUSize;
 extern uint32_t g_maxCUDepth;
-extern uint32_t g_addCUDepth;
-
-#define MAX_TS_WIDTH  4
-#define MAX_TS_HEIGHT 4
+extern uint32_t g_maxFullDepth;
 
 extern const uint32_t g_puOffset[8];
 
-#define QUANT_IQUANT_SHIFT    20 // Q(QP%6) * IQ(QP%6) = 2^20
-#define QUANT_SHIFT           14 // Q(4) = 2^14
-#define SCALE_BITS            15 // Inherited from TMuC, presumably for fractional bit estimates in RDOQ
-#define MAX_TR_DYNAMIC_RANGE  15 // Maximum transform dynamic range (excluding sign bit)
-
-#define SHIFT_INV_1ST          7 // Shift after first inverse transform stage
-#define SHIFT_INV_2ND         12 // Shift after second inverse transform stage
-
-extern const int g_quantScales[6];     // Q(QP%6)
-extern const int g_invQuantScales[6];  // IQ(QP%6)
 extern const int16_t g_t4[4][4];
 extern const int16_t g_t8[8][8];
 extern const int16_t g_t16[16][16];
 extern const int16_t g_t32[32][32];
 
-// ====================================================================================================================
 // Subpel interpolation defines and constants
-// ====================================================================================================================
 
 #define NTAPS_LUMA        8                            ///< Number of taps for luma
 #define NTAPS_CHROMA      4                            ///< Number of taps for chroma
@@ -124,155 +111,54 @@
 extern const int16_t g_lumaFilter[4][NTAPS_LUMA];     ///< Luma filter taps
 extern const int16_t g_chromaFilter[8][NTAPS_CHROMA]; ///< Chroma filter taps
 
-// ====================================================================================================================
 // Scanning order & context mapping table
-// ====================================================================================================================
-
-extern const uint8_t g_groupIdx[32];
-extern const uint32_t g_minInGroup[10];
-
-extern const uint8_t g_goRiceRange[5];      //!< maximum value coded with Rice codes
-//extern const uint8_t g_goRicePrefixLen[5];  //!< prefix length for each maximum value
-
-// ====================================================================================================================
-// Bit-depth
-// ====================================================================================================================
-
-/** clip x, such that 0 <= x <= #g_maxLumaVal */
-template<typename T>
-inline T ClipY(T x) { return std::min<T>(T((1 << X265_DEPTH) - 1), std::max<T>(T(0), x)); }
-
-template<typename T>
-inline T ClipC(T x) { return std::min<T>(T((1 << X265_DEPTH) - 1), std::max<T>(T(0), x)); }
-
-/** clip a, such that minVal <= a <= maxVal */
-template<typename T>
-inline T Clip3(T minVal, T maxVal, T a) { return std::min<T>(std::max<T>(minVal, a), maxVal); } ///< general min/max clip
-
-// ====================================================================================================================
-// Misc.
-// ====================================================================================================================
-
-extern uint8_t g_convertToBit[MAX_CU_SIZE + 1]; // from width to log2(width)-2
-
-#ifndef ENC_DEC_TRACE
-# define ENC_DEC_TRACE 0
-#endif
-
-#if ENC_DEC_TRACE
-extern FILE*  g_hTrace;
-extern bool   g_bJustDoIt;
-extern const bool g_bEncDecTraceEnable;
-extern const bool g_bEncDecTraceDisable;
-extern bool   g_HLSTraceEnable;
-extern uint64_t g_nSymbolCounter;
-
-#define COUNTER_START    1
-#define COUNTER_END      0 //( uint64_t(1) << 63 )
-
-#define DTRACE_CABAC_F(x)     if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "%f", x);
-#define DTRACE_CABAC_V(x)     if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "%d", x);
-#define DTRACE_CABAC_VL(x)    if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "%ld", x);
-#define DTRACE_CABAC_T(x)     if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "%s", x);
-#define DTRACE_CABAC_X(x)     if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "%x", x);
-#define DTRACE_CABAC_R(x, y)  if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, x,    y);
-#define DTRACE_CABAC_N        if ((g_nSymbolCounter >= COUNTER_START && g_nSymbolCounter <= COUNTER_END) || g_bJustDoIt) fprintf(g_hTrace, "\n");
-#else // if ENC_DEC_TRACE
-
-#define DTRACE_CABAC_F(x)
-#define DTRACE_CABAC_V(x)
-#define DTRACE_CABAC_VL(x)
-#define DTRACE_CABAC_T(x)
-#define DTRACE_CABAC_X(x)
-#define DTRACE_CABAC_R(x, y)
-#define DTRACE_CABAC_N
-
-#endif // if ENC_DEC_TRACE
-
-#define SCALING_LIST_NUM 6         ///< list number for quantization matrix
-#define SCALING_LIST_NUM_32x32 2   ///< list number for quantization matrix 32x32
-#define SCALING_LIST_REM_NUM 6     ///< remainder of QP/6
-#define SCALING_LIST_START_VALUE 8 ///< start value for dpcm mode
-#define MAX_MATRIX_COEF_NUM 64     ///< max coefficient number for quantization matrix
-#define MAX_MATRIX_SIZE_NUM 8      ///< max size number for quantization matrix
-#define SCALING_LIST_DC 16         ///< default DC value
-enum ScalingListSize
-{
-    SCALING_LIST_4x4 = 0,
-    SCALING_LIST_8x8,
-    SCALING_LIST_16x16,
-    SCALING_LIST_32x32,
-    SCALING_LIST_SIZE_NUM
-};
 
-static const char MatrixType[4][6][20] =
+// coefficient scanning type used in ACS
+enum ScanType
 {
-    {
-        "INTRA4X4_LUMA",
-        "INTRA4X4_CHROMAU",
-        "INTRA4X4_CHROMAV",
-        "INTER4X4_LUMA",
-        "INTER4X4_CHROMAU",
-        "INTER4X4_CHROMAV"
-    },
-    {
-        "INTRA8X8_LUMA",
-        "INTRA8X8_CHROMAU",
-        "INTRA8X8_CHROMAV",
-        "INTER8X8_LUMA",
-        "INTER8X8_CHROMAU",
-        "INTER8X8_CHROMAV"
-    },
-    {
-        "INTRA16X16_LUMA",
-        "INTRA16X16_CHROMAU",
-        "INTRA16X16_CHROMAV",
-        "INTER16X16_LUMA",
-        "INTER16X16_CHROMAU",
-        "INTER16X16_CHROMAV"
-    },
-    {
-        "INTRA32X32_LUMA",
-        "INTER32X32_LUMA",
-    },
+    SCAN_DIAG = 0,     // up-right diagonal scan
+    SCAN_HOR = 1,      // horizontal first scan
+    SCAN_VER = 2,      // vertical first scan
+    NUM_SCAN_TYPE = 3
 };
-static const char MatrixType_DC[4][12][22] =
+
+enum SignificanceMapContextType
 {
-    {},
-    {},
-    {
-        "INTRA16X16_LUMA_DC",
-        "INTRA16X16_CHROMAU_DC",
-        "INTRA16X16_CHROMAV_DC",
-        "INTER16X16_LUMA_DC",
-        "INTER16X16_CHROMAU_DC",
-        "INTER16X16_CHROMAV_DC"
-    },
-    {
-        "INTRA32X32_LUMA_DC",
-        "INTER32X32_LUMA_DC",
-    },
+    CONTEXT_TYPE_4x4 = 0,
+    CONTEXT_TYPE_8x8 = 1,
+    CONTEXT_TYPE_NxN = 2,
+    CONTEXT_NUMBER_OF_TYPES = 3
 };
-extern int g_quantIntraDefault8x8[64];
-extern int g_quantIntraDefault16x16[256];
-extern int g_quantIntraDefault32x32[1024];
-extern int g_quantInterDefault8x8[64];
-extern int g_quantInterDefault16x16[256];
-extern int g_quantInterDefault32x32[1024];
-extern int g_quantTSDefault4x4[16];
-extern const uint32_t g_scalingListSize[SCALING_LIST_SIZE_NUM];
-extern const uint32_t g_scalingListSizeX[SCALING_LIST_SIZE_NUM];
-extern const uint32_t g_scalingListNum[SCALING_LIST_SIZE_NUM];
-//! \}
+
+#define NUM_SCAN_SIZE 4
+
+extern const uint16_t* const g_scanOrder[NUM_SCAN_TYPE][NUM_SCAN_SIZE];
+extern const uint16_t* const g_scanOrderCG[NUM_SCAN_TYPE][NUM_SCAN_SIZE];
+extern const uint16_t g_scan8x8diag[8 * 8];
+extern const uint16_t g_scan4x4[NUM_SCAN_TYPE][4 * 4];
+
+extern const uint8_t g_minInGroup[10];
+extern const uint8_t g_goRiceRange[5]; // maximum value coded with Rice codes
+
+extern const uint8_t g_log2Size[MAX_CU_SIZE + 1]; // from size to log2(size)
 
 // Map Luma samples to chroma samples
 extern const int g_winUnitX[MAX_CHROMA_FORMAT_IDC + 1];
 extern const int g_winUnitY[MAX_CHROMA_FORMAT_IDC + 1];
-extern const double x265_lambda2_tab_I[MAX_MAX_QP + 1];
-extern const double x265_lambda2_non_I[MAX_MAX_QP + 1];
+
+extern double x265_lambda_tab[QP_MAX_MAX + 1];
+extern double x265_lambda2_tab[QP_MAX_MAX + 1];
+extern const uint16_t x265_chroma_lambda2_offset_tab[MAX_CHROMA_LAMBDA_OFFSET+1];
+
 // CABAC tables
 extern const uint8_t g_lpsTable[64][4];
 extern const uint8_t x265_exp2_lut[64];
+
+// Intra tables
+extern const uint8_t g_intraFilterFlags[NUM_INTRA_MODE];
+
+extern const uint32_t g_depthInc[3][4];
+
 }
 
 #endif  //ifndef X265_TCOMROM_H

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComYuv.cpp -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComYuv.cpp Changed

@@ -36,9 +36,9 @@
     \todo     this should be merged with TComPicYuv
 */
 
-#include "CommonDef.h"
+#include "common.h"
 #include "TComYuv.h"
-#include "TComPrediction.h"
+#include "TComPicYuv.h"
 #include "shortyuv.h"
 #include "primitives.h"
 
@@ -49,13 +49,15 @@
 
 TComYuv::TComYuv()
 {
-    m_bufY = NULL;
-    m_bufU = NULL;
-    m_bufV = NULL;
+    m_buf[0] = NULL;
+    m_buf[1] = NULL;
+    m_buf[2] = NULL;
 }
 
 TComYuv::~TComYuv()
-{}
+{
+    destroy();
+}
 
 bool TComYuv::create(uint32_t width, uint32_t height, int csp)
 {
@@ -72,10 +74,13 @@
     m_csp = csp;
     m_part = partitionFromSizes(m_width, m_height);
 
+    uint32_t sizeL = width * height;
+    uint32_t sizeC = m_cwidth * m_cheight;
+    X265_CHECK((sizeC & 15) == 0, "invalid size");
     // memory allocation (padded for SIMD reads)
-    CHECKED_MALLOC(m_bufY, pixel, width * height);
-    CHECKED_MALLOC(m_bufU, pixel, m_cwidth * m_cheight + 8);
-    CHECKED_MALLOC(m_bufV, pixel, m_cwidth * m_cheight + 8);
+    CHECKED_MALLOC(m_buf[0], pixel, sizeL + sizeC * 2 + 8);
+    m_buf[1] = m_buf[0] + sizeL;
+    m_buf[2] = m_buf[0] + sizeL + sizeC;
     return true;
 
 fail:
@@ -85,61 +90,62 @@
 void TComYuv::destroy()
 {
     // memory free
-    X265_FREE(m_bufY);
-    m_bufY = NULL;
-    X265_FREE(m_bufU);
-    m_bufU = NULL;
-    X265_FREE(m_bufV);
-    m_bufV = NULL;
+    X265_FREE(m_buf[0]);
+    m_buf[0] = NULL;
+    m_buf[1] = NULL;
+    m_buf[2] = NULL;
 }
 
 void TComYuv::clear()
 {
-    ::memset(m_bufY, 0, (m_width  * m_height) * sizeof(pixel));
-    ::memset(m_bufU, 0, (m_cwidth * m_cheight) * sizeof(pixel));
-    ::memset(m_bufV, 0, (m_cwidth * m_cheight) * sizeof(pixel));
+    ::memset(m_buf[0], 0, (m_width  * m_height) * sizeof(pixel));
+    ::memset(m_buf[1], 0, (m_cwidth * m_cheight) * sizeof(pixel));
+    ::memset(m_buf[2], 0, (m_cwidth * m_cheight) * sizeof(pixel));
 }
 
-void TComYuv::copyToPicYuv(TComPicYuv* destPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx, uint32_t depth, uint32_t partIdx)
+void TComYuv::copyToPicYuv(TComPicYuv* destPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx)
 {
-    int width = m_width >> depth;
-    int part = partitionFromSizes(width, m_height >> depth);
-    pixel* srcY = getLumaAddr(partIdx, width);
     pixel* dstY = destPicYuv->getLumaAddr(cuAddr, absZOrderIdx);
 
-    primitives.luma_copy_pp[part](dstY, destPicYuv->getStride(), srcY, getStride());
+    primitives.luma_copy_pp[m_part](dstY, destPicYuv->getStride(), m_buf[0], getStride());
 
-    width = m_cwidth >> depth;
-    pixel* srcU = getCbAddr(partIdx, width);
-    pixel* srcV = getCrAddr(partIdx, width);
     pixel* dstU = destPicYuv->getCbAddr(cuAddr, absZOrderIdx);
     pixel* dstV = destPicYuv->getCrAddr(cuAddr, absZOrderIdx);
-    primitives.chroma[m_csp].copy_pp[part](dstU, destPicYuv->getCStride(), srcU, getCStride());
-    primitives.chroma[m_csp].copy_pp[part](dstV, destPicYuv->getCStride(), srcV, getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](dstU, destPicYuv->getCStride(), m_buf[1], getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](dstV, destPicYuv->getCStride(), m_buf[2], getCStride());
 }
 
 void TComYuv::copyFromPicYuv(TComPicYuv* srcPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx)
 {
     pixel* srcY = srcPicYuv->getLumaAddr(cuAddr, absZOrderIdx);
 
-    primitives.luma_copy_pp[m_part](m_bufY, getStride(), srcY, srcPicYuv->getStride());
+    primitives.luma_copy_pp[m_part](m_buf[0], getStride(), srcY, srcPicYuv->getStride());
 
     pixel* srcU = srcPicYuv->getCbAddr(cuAddr, absZOrderIdx);
     pixel* srcV = srcPicYuv->getCrAddr(cuAddr, absZOrderIdx);
-    primitives.chroma[m_csp].copy_pp[m_part](m_bufU, getCStride(), srcU, srcPicYuv->getCStride());
-    primitives.chroma[m_csp].copy_pp[m_part](m_bufV, getCStride(), srcV, srcPicYuv->getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](m_buf[1], getCStride(), srcU, srcPicYuv->getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](m_buf[2], getCStride(), srcV, srcPicYuv->getCStride());
+}
+
+void TComYuv::copyFromYuv(TComYuv* srcYuv)
+{
+    X265_CHECK(m_width <= srcYuv->m_width && m_height <= srcYuv->m_height, "invalid size\n");
+
+    primitives.luma_copy_pp[m_part](m_buf[0], m_width, srcYuv->m_buf[0], srcYuv->m_width);
+    primitives.chroma[m_csp].copy_pp[m_part](m_buf[1], m_cwidth, srcYuv->m_buf[1], srcYuv->m_cwidth);
+    primitives.chroma[m_csp].copy_pp[m_part](m_buf[2], m_cwidth, srcYuv->m_buf[2], srcYuv->m_cwidth);
 }
 
 void TComYuv::copyToPartYuv(TComYuv* dstPicYuv, uint32_t partIdx)
 {
     pixel* dstY = dstPicYuv->getLumaAddr(partIdx);
 
-    primitives.luma_copy_pp[m_part](dstY, dstPicYuv->getStride(), m_bufY, getStride());
+    primitives.luma_copy_pp[m_part](dstY, dstPicYuv->getStride(), m_buf[0], getStride());
 
     pixel* dstU = dstPicYuv->getCbAddr(partIdx);
     pixel* dstV = dstPicYuv->getCrAddr(partIdx);
-    primitives.chroma[m_csp].copy_pp[m_part](dstU, dstPicYuv->getCStride(), m_bufU, getCStride());
-    primitives.chroma[m_csp].copy_pp[m_part](dstV, dstPicYuv->getCStride(), m_bufV, getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](dstU, dstPicYuv->getCStride(), m_buf[1], getCStride());
+    primitives.chroma[m_csp].copy_pp[m_part](dstV, dstPicYuv->getCStride(), m_buf[2], getCStride());
 }
 
 void TComYuv::copyPartToYuv(TComYuv* dstPicYuv, uint32_t partIdx)
@@ -148,6 +154,7 @@
     pixel* dstY = dstPicYuv->getLumaAddr();
 
     int part = dstPicYuv->m_part;
+
     primitives.luma_copy_pp[part](dstY, dstPicYuv->getStride(), srcY, getStride());
 
     pixel* srcU = getCbAddr(partIdx);
@@ -158,97 +165,13 @@
     primitives.chroma[m_csp].copy_pp[part](dstV, dstPicYuv->getCStride(), srcV, getCStride());
 }
 
-void TComYuv::copyPartToPartYuv(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma)
+void TComYuv::addClip(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size)
 {
-    int part = partitionFromSizes(width, height);
-
-    assert(width != 4 || height != 4);
-
-    if (bLuma)
-    {
-        pixel* src = getLumaAddr(partIdx);
-        pixel* dst = dstPicYuv->getLumaAddr(partIdx);
-
-        uint32_t srcstride = getStride();
-        uint32_t dststride = dstPicYuv->getStride();
-
-        primitives.luma_copy_pp[part](dst, dststride, src, srcstride);
-    }
-    if (bChroma)
-    {
-        pixel* srcU = getCbAddr(partIdx);
-        pixel* srcV = getCrAddr(partIdx);
-        pixel* dstU = dstPicYuv->getCbAddr(partIdx);
-        pixel* dstV = dstPicYuv->getCrAddr(partIdx);
-
-        uint32_t srcstride = getCStride();
-        uint32_t dststride = dstPicYuv->getCStride();
-
-        primitives.chroma[m_csp].copy_pp[part](dstU, dststride, srcU, srcstride);
-        primitives.chroma[m_csp].copy_pp[part](dstV, dststride, srcV, srcstride);
-    }
-}
-
-void TComYuv::copyPartToPartLuma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize)
-{
-    int part = partitionFromSizes(lumaSize, lumaSize);
-
-    int16_t* dst = dstPicYuv->getLumaAddr(partIdx);
-    uint32_t dststride = dstPicYuv->m_width;
-
-    primitives.luma_copy_ps[part](dst, dststride, getLumaAddr(partIdx), getStride());
+    addClipLuma(srcYuv0, srcYuv1, log2Size);
+    addClipChroma(srcYuv0, srcYuv1, log2Size);
 }
 
-void TComYuv::copyPartToPartChroma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, uint32_t chromaId)
-{
-    int part = partitionFromSizes(lumaSize, lumaSize);
-
-    assert(lumaSize != 4);
-
-    if (chromaId == 0)
-    {
-        pixel*   srcU = getCbAddr(partIdx);
-        int16_t* dstU = dstPicYuv->getCbAddr(partIdx);
-
-        uint32_t srcstride = getCStride();
-        uint32_t dststride = dstPicYuv->m_cwidth;
-
-        primitives.chroma[m_csp].copy_ps[part](dstU, dststride, srcU, srcstride);
-    }
-    else if (chromaId == 1)
-    {
-        pixel*   srcV = getCrAddr(partIdx);
-        int16_t* dstV = dstPicYuv->getCrAddr(partIdx);
-
-        uint32_t srcstride = getCStride();
-        uint32_t dststride = dstPicYuv->m_cwidth;
-
-        primitives.chroma[m_csp].copy_ps[part](dstV, dststride, srcV, srcstride);
-    }
-    else
-    {
-        pixel*   srcU = getCbAddr(partIdx);
-        pixel*   srcV = getCrAddr(partIdx);
-        int16_t* dstU = dstPicYuv->getCbAddr(partIdx);
-        int16_t* dstV = dstPicYuv->getCrAddr(partIdx);
-
-        uint32_t srcstride = getCStride();
-        uint32_t dststride = dstPicYuv->m_cwidth;
-
-        primitives.chroma[m_csp].copy_ps[part](dstU, dststride, srcU, srcstride);
-        primitives.chroma[m_csp].copy_ps[part](dstV, dststride, srcV, srcstride);
-    }
-}
-
-void TComYuv::addClip(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partSize)
-{
-    int part = partitionFromSizes(partSize, partSize);
-
-    addClipLuma(srcYuv0, srcYuv1, part);
-    addClipChroma(srcYuv0, srcYuv1, part);
-}
-
-void TComYuv::addClipLuma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t part)
+void TComYuv::addClipLuma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size)
 {
     pixel* src0 = srcYuv0->getLumaAddr();
     int16_t* src1 = srcYuv1->getLumaAddr();
@@ -258,14 +181,14 @@
     uint32_t src1Stride = srcYuv1->m_width;
     uint32_t dststride  = getStride();
 
-    primitives.luma_add_ps[part](dst, dststride, src0, src1, src0Stride, src1Stride);
+    primitives.luma_add_ps[log2Size - 2](dst, dststride, src0, src1, src0Stride, src1Stride);
 }
 
-void TComYuv::addClipChroma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t part)
+void TComYuv::addClipChroma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size)
 {
     pixel* srcU0 = srcYuv0->getCbAddr();
-    int16_t* srcU1 = srcYuv1->getCbAddr();
     pixel* srcV0 = srcYuv0->getCrAddr();
+    int16_t* srcU1 = srcYuv1->getCbAddr();
     int16_t* srcV1 = srcYuv1->getCrAddr();
     pixel* dstU = getCbAddr();
     pixel* dstV = getCrAddr();
@@ -274,117 +197,36 @@
     uint32_t src1Stride = srcYuv1->m_cwidth;
     uint32_t dststride  = getCStride();
 
-    primitives.chroma[m_csp].add_ps[part](dstU, dststride, srcU0, srcU1, src0Stride, src1Stride);
-    primitives.chroma[m_csp].add_ps[part](dstV, dststride, srcV0, srcV1, src0Stride, src1Stride);
-}
-
-void TComYuv::addAvg(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma)
-{
-    int x, y;
-    uint32_t src0Stride, src1Stride, dststride;
-    int shiftNum, offset;
-
-    pixel* srcY0 = srcYuv0->getLumaAddr(partUnitIdx);
-    pixel* srcU0 = srcYuv0->getCbAddr(partUnitIdx);
-    pixel* srcV0 = srcYuv0->getCrAddr(partUnitIdx);
-
-    pixel* srcY1 = srcYuv1->getLumaAddr(partUnitIdx);
-    pixel* srcU1 = srcYuv1->getCbAddr(partUnitIdx);
-    pixel* srcV1 = srcYuv1->getCrAddr(partUnitIdx);
-
-    pixel* dstY  = getLumaAddr(partUnitIdx);
-    pixel* dstU  = getCbAddr(partUnitIdx);
-    pixel* dstV  = getCrAddr(partUnitIdx);
-
-    if (bLuma)
-    {
-        src0Stride = srcYuv0->getStride();
-        src1Stride = srcYuv1->getStride();
-        dststride  = getStride();
-        shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
-        offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
-
-        for (y = 0; y < height; y++)
-        {
-            for (x = 0; x < width; x += 4)
-            {
-                dstY[x + 0] = ClipY((srcY0[x + 0] + srcY1[x + 0] + offset) >> shiftNum);
-                dstY[x + 1] = ClipY((srcY0[x + 1] + srcY1[x + 1] + offset) >> shiftNum);
-                dstY[x + 2] = ClipY((srcY0[x + 2] + srcY1[x + 2] + offset) >> shiftNum);
-                dstY[x + 3] = ClipY((srcY0[x + 3] + srcY1[x + 3] + offset) >> shiftNum);
-            }
-
-            srcY0 += src0Stride;
-            srcY1 += src1Stride;
-            dstY  += dststride;
-        }
-    }
-    if (bChroma)
-    {
-        shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
-        offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
-
-        src0Stride = srcYuv0->getCStride();
-        src1Stride = srcYuv1->getCStride();
-        dststride  = getCStride();
-
-        width  >>= m_hChromaShift;
-        height >>= m_vChromaShift;
-
-        for (y = height - 1; y >= 0; y--)
-        {
-            for (x = width - 1; x >= 0; )
-            {
-                // note: chroma min width is 2
-                dstU[x] = ClipC((srcU0[x] + srcU1[x] + offset) >> shiftNum);
-                dstV[x] = ClipC((srcV0[x] + srcV1[x] + offset) >> shiftNum);
-                x--;
-                dstU[x] = ClipC((srcU0[x] + srcU1[x] + offset) >> shiftNum);
-                dstV[x] = ClipC((srcV0[x] + srcV1[x] + offset) >> shiftNum);
-                x--;
-            }
-
-            srcU0 += src0Stride;
-            srcU1 += src1Stride;
-            srcV0 += src0Stride;
-            srcV1 += src1Stride;
-            dstU  += dststride;
-            dstV  += dststride;
-        }
-    }
+    primitives.chroma[m_csp].add_ps[log2Size - 2](dstU, dststride, srcU0, srcU1, src0Stride, src1Stride);
+    primitives.chroma[m_csp].add_ps[log2Size - 2](dstV, dststride, srcV0, srcV1, src0Stride, src1Stride);
 }
 
 void TComYuv::addAvg(ShortYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma)
 {
-    uint32_t src0Stride, src1Stride, dststride;
-
-    int16_t* srcY0 = srcYuv0->getLumaAddr(partUnitIdx);
-    int16_t* srcU0 = srcYuv0->getCbAddr(partUnitIdx);
-    int16_t* srcV0 = srcYuv0->getCrAddr(partUnitIdx);
-
-    int16_t* srcY1 = srcYuv1->getLumaAddr(partUnitIdx);
-    int16_t* srcU1 = srcYuv1->getCbAddr(partUnitIdx);
-    int16_t* srcV1 = srcYuv1->getCrAddr(partUnitIdx);
-
-    pixel* dstY = getLumaAddr(partUnitIdx);
-    pixel* dstU = getCbAddr(partUnitIdx);
-    pixel* dstV = getCrAddr(partUnitIdx);
-
     int part = partitionFromSizes(width, height);
 
     if (bLuma)
     {
-        src0Stride = srcYuv0->m_width;
-        src1Stride = srcYuv1->m_width;
-        dststride  = getStride();
+        int16_t* srcY0 = srcYuv0->getLumaAddr(partUnitIdx);
+        int16_t* srcY1 = srcYuv1->getLumaAddr(partUnitIdx);
+        pixel* dstY = getLumaAddr(partUnitIdx);
+        uint32_t src0Stride = srcYuv0->m_width;
+        uint32_t src1Stride = srcYuv1->m_width;
+        uint32_t dststride  = getStride();
 
         primitives.luma_addAvg[part](srcY0, srcY1, dstY, src0Stride, src1Stride, dststride);
     }
     if (bChroma)
     {
-        src0Stride = srcYuv0->m_cwidth;
-        src1Stride = srcYuv1->m_cwidth;
-        dststride  = getCStride();
+        int16_t* srcU0 = srcYuv0->getCbAddr(partUnitIdx);
+        int16_t* srcV0 = srcYuv0->getCrAddr(partUnitIdx);
+        int16_t* srcU1 = srcYuv1->getCbAddr(partUnitIdx);
+        int16_t* srcV1 = srcYuv1->getCrAddr(partUnitIdx);
+        pixel* dstU = getCbAddr(partUnitIdx);
+        pixel* dstV = getCrAddr(partUnitIdx);
+        uint32_t src0Stride = srcYuv0->m_cwidth;
+        uint32_t src1Stride = srcYuv1->m_cwidth;
+        uint32_t dststride  = getCStride();
 
         primitives.chroma[m_csp].addAvg[part](srcU0, srcU1, dstU, src0Stride, src1Stride, dststride);
         primitives.chroma[m_csp].addAvg[part](srcV0, srcV1, dstV, src0Stride, src1Stride, dststride);

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TComYuv.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TComYuv.h Changed

@@ -39,7 +39,7 @@
 #ifndef X265_TCOMYUV_H
 #define X265_TCOMYUV_H
 
-#include "CommonDef.h"
+#include "common.h"
 #include "TComRom.h"
 #include "primitives.h"
 
@@ -65,9 +65,7 @@
     //  YUV buffer
     // ------------------------------------------------------------------------------------------------------------------
 
-    pixel* m_bufY;
-    pixel* m_bufU;
-    pixel* m_bufV;
+    pixel* m_buf[3];
 
     // ------------------------------------------------------------------------------------------------------------------
     //  Parameter for general YUV buffer usage
@@ -82,20 +80,20 @@
     int m_vChromaShift;
     int m_csp;
 
-    static int getAddrOffset(uint32_t partUnitIdx, uint32_t width)
+    int getChromaAddrOffset(uint32_t idx, uint32_t width)
     {
-        int blkX = g_rasterToPelX[g_zscanToRaster[partUnitIdx]];
-        int blkY = g_rasterToPelY[g_zscanToRaster[partUnitIdx]];
+        int blkX = g_zscanToPelX[idx] >> m_hChromaShift;
+        int blkY = g_zscanToPelY[idx] >> m_vChromaShift;
 
         return blkX + blkY * width;
     }
 
-    static int getAddrOffset(uint32_t unitIdx, uint32_t size, uint32_t width)
+    static int getAddrOffset(uint32_t idx, uint32_t width)
     {
-        int blkX = (unitIdx * size) &  (width - 1);
-        int blkY = (unitIdx * size) & ~(width - 1);
+        int blkX = g_zscanToPelX[idx];
+        int blkY = g_zscanToPelY[idx];
 
-        return blkX + blkY * size;
+        return blkX + blkY * width;
     }
 
 public:
@@ -103,7 +101,7 @@
     int m_part; // partitionFromSizes(m_width, m_height)
 
     TComYuv();
-    virtual ~TComYuv();
+    ~TComYuv();
 
     // ------------------------------------------------------------------------------------------------------------------
     //  Memory management
@@ -118,34 +116,30 @@
     // ------------------------------------------------------------------------------------------------------------------
 
     //  Copy YUV buffer to picture buffer
-    void    copyToPicYuv(TComPicYuv* destPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx, uint32_t depth, uint32_t partIdx);
+    void    copyToPicYuv(TComPicYuv* destPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx);
 
     //  Copy YUV buffer from picture buffer
     void    copyFromPicYuv(TComPicYuv* srcPicYuv, uint32_t cuAddr, uint32_t absZOrderIdx);
 
+    //  Copy from same size YUV buffer
+    void    copyFromYuv(TComYuv* srcYuv);
+
     //  Copy Small YUV buffer to the part of other Big YUV buffer
     void    copyToPartYuv(TComYuv* dstPicYuv, uint32_t partIdx);
 
     //  Copy the part of Big YUV buffer to other Small YUV buffer
     void    copyPartToYuv(TComYuv* dstPicYuv, uint32_t srcPartIdx);
 
-    //  Copy YUV partition buffer to other YUV partition buffer
-    void    copyPartToPartYuv(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma);
-
-    void    copyPartToPartLuma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize);
-    void    copyPartToPartChroma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, uint32_t chromaId);
-
     // ------------------------------------------------------------------------------------------------------------------
     //  Algebraic operation for YUV buffer
     // ------------------------------------------------------------------------------------------------------------------
 
     //  Clip(srcYuv0 + srcYuv1) -> m_apiBuf
-    void    addClip(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partSize);
-    void    addClipLuma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t part);
-    void    addClipChroma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t part);
+    void    addClip(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size);
+    void    addClipLuma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size);
+    void    addClipChroma(TComYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t log2Size);
 
     //  (srcYuv0 + srcYuv1)/2 for YUV partition
-    void    addAvg(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma);
     void    addAvg(ShortYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partUnitIdx, uint32_t width, uint32_t height, bool bLuma, bool bChroma);
 
     // ------------------------------------------------------------------------------------------------------------------
@@ -153,25 +147,22 @@
     // ------------------------------------------------------------------------------------------------------------------
 
     //  Access starting position of YUV buffer
-    pixel* getLumaAddr()  { return m_bufY; }
-
-    pixel* getCbAddr()    { return m_bufU; }
+    pixel* getLumaAddr()  { return m_buf[0]; }
 
-    pixel* getCrAddr()    { return m_bufV; }
+    pixel* getCbAddr()    { return m_buf[1]; }
 
-    //  Access starting position of YUV partition unit buffer
-    pixel* getLumaAddr(uint32_t partUnitIdx) { return m_bufY + getAddrOffset(partUnitIdx, m_width); }
+    pixel* getCrAddr()    { return m_buf[2]; }
 
-    pixel* getCbAddr(uint32_t partUnitIdx) { return m_bufU + (getAddrOffset(partUnitIdx, m_cwidth) >> m_hChromaShift); }
+    pixel* getChromaAddr(uint32_t chromaId)    { return m_buf[chromaId]; }
 
-    pixel* getCrAddr(uint32_t partUnitIdx) { return m_bufV + (getAddrOffset(partUnitIdx, m_cwidth) >> m_hChromaShift); }
+    //  Access starting position of YUV partition unit buffer
+    pixel* getLumaAddr(uint32_t partUnitIdx) { return m_buf[0] + getAddrOffset(partUnitIdx, m_width); }
 
-    //  Access starting position of YUV transform unit buffer
-    pixel* getLumaAddr(uint32_t transUnitIdx, uint32_t blkSize) { return m_bufY + getAddrOffset(transUnitIdx, blkSize, m_width); }
+    pixel* getCbAddr(uint32_t partUnitIdx) { return m_buf[1] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
-    pixel* getCbAddr(uint32_t transUnitIdx, uint32_t blkSize) { return m_bufU + getAddrOffset(transUnitIdx, blkSize, m_cwidth); }
+    pixel* getCrAddr(uint32_t partUnitIdx) { return m_buf[2] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
-    pixel* getCrAddr(uint32_t transUnitIdx, uint32_t blkSize) { return m_bufV + getAddrOffset(transUnitIdx, blkSize, m_cwidth); }
+    pixel* getChromaAddr(uint32_t chromaId, uint32_t partUnitIdx) { return m_buf[chromaId] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
     //  Get stride value of YUV buffer
     uint32_t getStride()    { return m_width;   }
@@ -185,6 +176,14 @@
     uint32_t getCHeight()   { return m_cheight; }
 
     uint32_t getCWidth()    { return m_cwidth;  }
+
+    // -------------------------------------------------------------------------------------------------------------------
+    // member functions to support multiple color space formats
+    // -------------------------------------------------------------------------------------------------------------------
+
+    int  getHorzChromaShift()  { return m_hChromaShift; }
+
+    int  getVertChromaShift()  { return m_vChromaShift; }
 };
 }
 //! \}

x265-hg20140324.tar.bz2/source/Lib/TLibCommon/TypeDef.h -> x265-hg20140928.tar.bz2/source/Lib/TLibCommon/TypeDef.h Changed

@@ -31,10 +31,6 @@
  * THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-/** \file     TypeDef.h
-    \brief    Define basic types, new types and enumerations
-*/
-
 #ifndef X265_TYPEDEF_H
 #define X265_TYPEDEF_H
 
@@ -43,36 +39,7 @@
 namespace x265 {
 // private namespace
 
-#define REF_PIC_LIST_0 0
-#define REF_PIC_LIST_1 1
-#define REF_PIC_LIST_X 100
-
-// ====================================================================================================================
-// Basic type redefinition
-// ====================================================================================================================
-
-typedef unsigned char  UChar;
-
-// ====================================================================================================================
-// Type definition
-// ====================================================================================================================
-
-#if HIGH_BIT_DEPTH
-typedef uint16_t Pel;          // 16-bit pixel type
-#define X265_DEPTH 10          // compile time configurable bit depth
-#else
-typedef uint8_t  Pel;          // 8-bit pixel type
-#define X265_DEPTH 8           // compile time configurable bit depth
-#endif
-typedef int32_t  TCoeff;       // transform coefficient
-
-// ====================================================================================================================
-// Enumeration
-// ====================================================================================================================
-#define MDCS_ANGLE_LIMIT                                  4         ///< (default 4) 0 = Horizontal/vertical only, 1 = Horizontal/vertical +/- 1, 2 = Horizontal/vertical +/- 2 etc...
-#define MDCS_LOG2_MAX_SIZE                                3         ///< (default 3) TUs with log2 of size greater than this can only use diagonal scan
-
-/// supported slice type
+// supported slice types
 enum SliceType
 {
     B_SLICE,
@@ -80,144 +47,59 @@
     I_SLICE
 };
 
-/// chroma formats (according to semantics of chroma_format_idc)
-enum ChromaFormat
-{
-    CHROMA_400  = 0,
-    CHROMA_420  = 1,
-    CHROMA_422  = 2,
-    CHROMA_444  = 3,
-    NUM_CHROMA_FORMAT = 4
-};
-
-#define CHROMA_H_SHIFT(x) (x == X265_CSP_I420 || x == X265_CSP_I422)
-#define CHROMA_V_SHIFT(x) (x == X265_CSP_I420)
-
-/// supported partition shape
+// supported partition shape
 enum PartSize
 {
-    SIZE_2Nx2N,         ///< symmetric motion partition,  2Nx2N
-    SIZE_2NxN,          ///< symmetric motion partition,  2Nx N
-    SIZE_Nx2N,          ///< symmetric motion partition,   Nx2N
-    SIZE_NxN,           ///< symmetric motion partition,   Nx N
-    SIZE_2NxnU,         ///< asymmetric motion partition, 2Nx( N/2) + 2Nx(3N/2)
-    SIZE_2NxnD,         ///< asymmetric motion partition, 2Nx(3N/2) + 2Nx( N/2)
-    SIZE_nLx2N,         ///< asymmetric motion partition, ( N/2)x2N + (3N/2)x2N
-    SIZE_nRx2N,         ///< asymmetric motion partition, (3N/2)x2N + ( N/2)x2N
+    SIZE_2Nx2N,         // symmetric motion partition,  2Nx2N
+    SIZE_2NxN,          // symmetric motion partition,  2Nx N
+    SIZE_Nx2N,          // symmetric motion partition,   Nx2N
+    SIZE_NxN,           // symmetric motion partition,   Nx N
+    SIZE_2NxnU,         // asymmetric motion partition, 2Nx( N/2) + 2Nx(3N/2)
+    SIZE_2NxnD,         // asymmetric motion partition, 2Nx(3N/2) + 2Nx( N/2)
+    SIZE_nLx2N,         // asymmetric motion partition, ( N/2)x2N + (3N/2)x2N
+    SIZE_nRx2N,         // asymmetric motion partition, (3N/2)x2N + ( N/2)x2N
     SIZE_NONE = 15
 };
 
-/// supported prediction type
+// supported prediction type
 enum PredMode
 {
-    MODE_INTER,         ///< inter-prediction mode
-    MODE_INTRA,         ///< intra-prediction mode
+    MODE_INTER,         // inter-prediction mode
+    MODE_INTRA,         // intra-prediction mode
     MODE_NONE = 15
 };
 
-/// texture component type
+// texture component type
 enum TextType
 {
-    TEXT_LUMA     = 0,  ///< luma
-    TEXT_CHROMA   = 1,  ///< chroma (U+V)
-    TEXT_CHROMA_U = 1,  ///< chroma U
-    TEXT_CHROMA_V = 2,  ///< chroma V
+    TEXT_LUMA     = 0,  // luma
+    TEXT_CHROMA_U = 1,  // chroma U
+    TEXT_CHROMA_V = 2,  // chroma V
+    MAX_NUM_COMPONENT = 3
 };
 
-/// index for SBAC based RD optimization
+// index for SBAC based RD optimization
 enum CI_IDX
 {
-    CI_CURR_BEST = 0,   ///< best mode index
-    CI_NEXT_BEST,       ///< next best index
-    CI_TEMP_BEST,       ///< temporal index
+    CI_CURR_BEST = 0,   // best mode index
+    CI_NEXT_BEST,       // next best index
+    CI_TEMP_BEST,       // temporal index
     CI_QT_TRAFO_TEST,
     CI_QT_TRAFO_ROOT,
-    CI_NUM,             ///< total number
+    CI_NUM,             // total number
     CI_NUM_SAO   = 3,
 };
 
-/// motion vector predictor direction used in AMVP
+// motion vector predictor direction used in AMVP
 enum MVP_DIR
 {
-    MD_LEFT = 0,        ///< MVP of left block
-    MD_ABOVE,           ///< MVP of above block
-    MD_ABOVE_RIGHT,     ///< MVP of above right block
-    MD_BELOW_LEFT,      ///< MVP of below left block
-    MD_ABOVE_LEFT       ///< MVP of above left block
-};
-
-/// coefficient scanning type used in ACS
-enum COEFF_SCAN_TYPE
-{
-    SCAN_DIAG = 0,      ///< up-right diagonal scan
-    SCAN_HOR  = 1,      ///< horizontal first scan
-    SCAN_VER  = 2,      ///< vertical first scan
-    SCAN_NUMBER_OF_TYPES = 3
-};
-
-enum SignificanceMapContextType
-{
-    CONTEXT_TYPE_4x4    = 0,
-    CONTEXT_TYPE_8x8    = 1,
-    CONTEXT_TYPE_NxN    = 2,
-    CONTEXT_NUMBER_OF_TYPES = 3
+    MD_LEFT = 0,        // MVP of left block
+    MD_ABOVE,           // MVP of above block
+    MD_ABOVE_RIGHT,     // MVP of above right block
+    MD_BELOW_LEFT,      // MVP of below left block
+    MD_ABOVE_LEFT       // MVP of above left block
 };
 
-enum COEFF_SCAN_GROUP_TYPE
-{
-    SCAN_UNGROUPED   = 0,
-    SCAN_GROUPED_4x4 = 1,
-    SCAN_NUMBER_OF_GROUP_TYPES = 2
-};
-
-//TU settings for entropy encoding
-struct TUEntropyCodingParameters
-{
-    const uint16_t            *scan;
-    const uint16_t            *scanCG;
-    COEFF_SCAN_TYPE      scanType;
-    TextType             ctype;
-    uint32_t             log2TrSize;
-    uint32_t             log2TrSizeCG;
-    uint32_t             firstSignificanceMapContext;
-};
-
-namespace Profile {
-enum Name
-{
-    NONE = 0,
-    MAIN = 1,
-    MAIN10 = 2,
-    MAINSTILLPICTURE = 3,
-};
-}
-
-namespace Level {
-enum Tier
-{
-    MAIN = 0,
-    HIGH = 1,
-};
-
-enum Name
-{
-    NONE     = 0,
-    LEVEL1   = 30,
-    LEVEL2   = 60,
-    LEVEL2_1 = 63,
-    LEVEL3   = 90,
-    LEVEL3_1 = 93,
-    LEVEL4   = 120,
-    LEVEL4_1 = 123,
-    LEVEL5   = 150,
-    LEVEL5_1 = 153,
-    LEVEL5_2 = 156,
-    LEVEL6   = 180,
-    LEVEL6_1 = 183,
-    LEVEL6_2 = 186,
-};
-}
 }
-//! \}
 
 #endif // ifndef X265_TYPEDEF_H

x265-hg20140324.tar.bz2/source/PPA/ppa.cpp -> x265-hg20140928.tar.bz2/source/PPA/ppa.cpp Changed

x265-hg20140324.tar.bz2/source/PPA/ppa.h -> x265-hg20140928.tar.bz2/source/PPA/ppa.h Changed

x265-hg20140324.tar.bz2/source/PPA/ppaApi.h -> x265-hg20140928.tar.bz2/source/PPA/ppaApi.h Changed

x265-hg20140324.tar.bz2/source/PPA/ppaCPUEvents.h -> x265-hg20140928.tar.bz2/source/PPA/ppaCPUEvents.h Changed

@@ -1,34 +1,24 @@
 PPA_REGISTER_CPU_EVENT(encode_block)
-PPA_REGISTER_CPU_EVENT(read_yuv)
-PPA_REGISTER_CPU_EVENT(write_yuv)
 PPA_REGISTER_CPU_EVENT(bitstream_write)
 PPA_REGISTER_CPU_EVENT(DPB_prepareEncode)
 PPA_REGISTER_CPU_EVENT(FrameEncoder_compressFrame)
 PPA_REGISTER_CPU_EVENT(FrameEncoder_compressRows)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressCU)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressCU_Depth1)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressCU_Depth2)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressCU_Depth3)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressCU_Depth4)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressIntraCU)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressIntraCU_Depth1)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressIntraCU_Depth2)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressIntraCU_Depth3)
-PPA_REGISTER_CPU_EVENT(TEncCu_xCompressIntraCU_Depth4)
-PPA_REGISTER_CPU_EVENT(TEncSearch_xMotionEstimation)
-PPA_REGISTER_CPU_EVENT(TEncSearch_xMotionEstimation_Depth1)
-PPA_REGISTER_CPU_EVENT(TEncSearch_xMotionEstimation_Depth2)
-PPA_REGISTER_CPU_EVENT(TEncSearch_xMotionEstimation_Depth3)
-PPA_REGISTER_CPU_EVENT(TEncSearch_xMotionEstimation_Depth4)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckRDCostIntra)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckRDCostIntra_Depth1)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckRDCostIntra_Depth2)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckRDCostIntra_Depth3)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckRDCostIntra_Depth4)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCalcRDCostIntra)
-PPA_REGISTER_CPU_EVENT(TEncCU_xCheckIntraPCM)
-PPA_REGISTER_CPU_EVENT(GenerateIntermediates)
-PPA_REGISTER_CPU_EVENT(GenerateReferencePlanes)
+PPA_REGISTER_CPU_EVENT(CompressCU)
+PPA_REGISTER_CPU_EVENT(CompressCU_Depth1)
+PPA_REGISTER_CPU_EVENT(CompressCU_Depth2)
+PPA_REGISTER_CPU_EVENT(CompressCU_Depth3)
+PPA_REGISTER_CPU_EVENT(CompressCU_Depth4)
+PPA_REGISTER_CPU_EVENT(CompressIntraCU)
+PPA_REGISTER_CPU_EVENT(CompressIntraCU_Depth1)
+PPA_REGISTER_CPU_EVENT(CompressIntraCU_Depth2)
+PPA_REGISTER_CPU_EVENT(CompressIntraCU_Depth3)
+PPA_REGISTER_CPU_EVENT(CompressIntraCU_Depth4)
+PPA_REGISTER_CPU_EVENT(CheckRDCostIntra)
+PPA_REGISTER_CPU_EVENT(CheckRDCostIntra_Depth1)
+PPA_REGISTER_CPU_EVENT(CheckRDCostIntra_Depth2)
+PPA_REGISTER_CPU_EVENT(CheckRDCostIntra_Depth3)
+PPA_REGISTER_CPU_EVENT(CheckRDCostIntra_Depth4)
+PPA_REGISTER_CPU_EVENT(CalcRDCostIntra)
 PPA_REGISTER_CPU_EVENT(Thread_ProcessRow)
 PPA_REGISTER_CPU_EVENT(Thread_compressCU)
 PPA_REGISTER_CPU_EVENT(Thread_encodeCU)

x265-hg20140324.tar.bz2/source/cmake/CMakeASM_YASMInformation.cmake -> x265-hg20140928.tar.bz2/source/cmake/CMakeASM_YASMInformation.cmake Changed

@@ -2,43 +2,46 @@
 set(CMAKE_ASM${ASM_DIALECT}_SOURCE_FILE_EXTENSIONS asm)
 
 if(X64)
-    set(ASM_FLAGS "-DARCH_X86_64=1 -m amd64 -DPIC")
+    list(APPEND ASM_FLAGS -DARCH_X86_64=1 -DPIC)
     if(APPLE)
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f macho64 -DPREFIX")
+        set(ARGS -f macho64 -m amd64 -DPREFIX)
     elseif(UNIX AND NOT CYGWIN)
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f elf64")
+        set(ARGS -f elf64 -m amd64)
     else()
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f win64")
+        set(ARGS -f win64 -m amd64)
     endif()
 else()
-    set(ASM_FLAGS -DARCH_X86_64=0)
+    list(APPEND ASM_FLAGS -DARCH_X86_64=0)
     if(APPLE)
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f macho")
+        set(ARGS -f macho -DPREFIX)
     elseif(UNIX AND NOT CYGWIN)
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f elf32")
+        set(ARGS -f elf32)
     else()
-        set(CMAKE_ASM${ASM_DIALECT}_COMPILER_ARG1 "-f win32 -DPREFIX")
+        set(ARGS -f win32 -DPREFIX)
     endif()
 endif()
 
-if (GCC)
-    set(ASM_FLAGS "${ASM_FLAGS} -DHAVE_ALIGNED_STACK=1")
+if(GCC)
+    list(APPEND ASM_FLAGS -DHAVE_ALIGNED_STACK=1)
 else()
-    set(ASM_FLAGS "${ASM_FLAGS} -DHAVE_ALIGNED_STACK=0")
+    list(APPEND ASM_FLAGS -DHAVE_ALIGNED_STACK=0)
 endif()
 
-if (HIGH_BIT_DEPTH)
-    set(ASM_FLAGS "${ASM_FLAGS} -DHIGH_BIT_DEPTH=1 -DBIT_DEPTH=10")
+if(HIGH_BIT_DEPTH)
+    list(APPEND ASM_FLAGS -DHIGH_BIT_DEPTH=1 -DBIT_DEPTH=10)
 else()
-    set(ASM_FLAGS "${ASM_FLAGS} -DHIGH_BIT_DEPTH=0 -DBIT_DEPTH=8")
+    list(APPEND ASM_FLAGS -DHIGH_BIT_DEPTH=0 -DBIT_DEPTH=8)
 endif()
+set(YASM_FLAGS ${ARGS} ${ASM_FLAGS} PARENT_SCOPE)
+string(REPLACE ";" " " CMAKE_ASM_YASM_COMPILER_ARG1 "${ARGS}")
 
 # This section exists to override the one in CMakeASMInformation.cmake
 # (the default Information file). This removes the <FLAGS>
 # thing so that your C compiler flags that have been set via
 # set_target_properties don't get passed to yasm and confuse it.
 if(NOT CMAKE_ASM${ASM_DIALECT}_COMPILE_OBJECT)
-    set(CMAKE_ASM${ASM_DIALECT}_COMPILE_OBJECT "<CMAKE_ASM${ASM_DIALECT}_COMPILER> ${ASM_FLAGS} -o <OBJECT> <SOURCE>")
+    string(REPLACE ";" " " STR_ASM_FLAGS "${ASM_FLAGS}")
+    set(CMAKE_ASM${ASM_DIALECT}_COMPILE_OBJECT "<CMAKE_ASM${ASM_DIALECT}_COMPILER> ${STR_ASM_FLAGS} -o <OBJECT> <SOURCE>")
 endif()
 
 include(CMakeASMInformation)

x265-hg20140928.tar.bz2/source/cmake/clean-generated.cmake Added

x265-hg20140324.tar.bz2/source/cmake/version.cmake -> x265-hg20140928.tar.bz2/source/cmake/version.cmake Changed

x265-hg20140324.tar.bz2/source/common/CMakeLists.txt -> x265-hg20140928.tar.bz2/source/common/CMakeLists.txt Changed

@@ -1,46 +1,26 @@
 # vim: syntax=cmake
-set(LIBCOMMON_HDR
-    ../Lib/TLibCommon/CommonDef.h
-    ../Lib/TLibCommon/ContextTables.h
-    ../Lib/TLibCommon/NAL.h
-    ../Lib/TLibCommon/SEI.h
-    ../Lib/TLibCommon/TComBitCounter.h
-    ../Lib/TLibCommon/TComBitStream.h
-    ../Lib/TLibCommon/TComDataCU.h
-    ../Lib/TLibCommon/TComLoopFilter.h
-    ../Lib/TLibCommon/TComMotionInfo.h
-    ../Lib/TLibCommon/TComPattern.h
-    ../Lib/TLibCommon/TComPic.h
-    ../Lib/TLibCommon/TComPicSym.h
-    ../Lib/TLibCommon/TComPicYuv.h
-    ../Lib/TLibCommon/TComPrediction.h
-    ../Lib/TLibCommon/TComRdCost.h
-    ../Lib/TLibCommon/TComRom.h
-    ../Lib/TLibCommon/TComSampleAdaptiveOffset.h
-    ../Lib/TLibCommon/TComSlice.h
-    ../Lib/TLibCommon/TComTrQuant.h
-    ../Lib/TLibCommon/TComWeightPrediction.h
-    ../Lib/TLibCommon/TComYuv.h
-    ../Lib/TLibCommon/TypeDef.h)
 set(LIBCOMMON_SRC
-    ../Lib/TLibCommon/TComBitStream.cpp
     ../Lib/TLibCommon/TComDataCU.cpp
-    ../Lib/TLibCommon/TComLoopFilter.cpp
     ../Lib/TLibCommon/TComMotionInfo.cpp
     ../Lib/TLibCommon/TComPattern.cpp
-    ../Lib/TLibCommon/TComPic.cpp
     ../Lib/TLibCommon/TComPicSym.cpp
     ../Lib/TLibCommon/TComPicYuv.cpp
     ../Lib/TLibCommon/TComPicYuvMD5.cpp
-    ../Lib/TLibCommon/TComPrediction.cpp
     ../Lib/TLibCommon/TComRom.cpp
-    ../Lib/TLibCommon/TComSampleAdaptiveOffset.cpp
-    ../Lib/TLibCommon/TComSlice.cpp
-    ../Lib/TLibCommon/TComTrQuant.cpp
-    ../Lib/TLibCommon/TComWeightPrediction.cpp
     ../Lib/TLibCommon/TComYuv.cpp)
-source_group(TLibCommon FILES ${LIBCOMMON_SRC})
-source_group(TLibCommonH FILES ${LIBCOMMON_HDR})
+set(LIBCOMMON_HDR
+    ../Lib/TLibCommon/CommonDef.h
+    ../Lib/TLibCommon/TypeDef.h
+    ../Lib/TLibCommon/ContextTables.h
+    ../Lib/TLibCommon/TComDataCU.h
+    ../Lib/TLibCommon/TComMotionInfo.h
+    ../Lib/TLibCommon/TComPattern.h
+    ../Lib/TLibCommon/TComPicSym.h
+    ../Lib/TLibCommon/TComPicYuv.h
+    ../Lib/TLibCommon/TComRom.h
+    ../Lib/TLibCommon/TComYuv.h)
+
+source_group(TLibCommon FILES ${LIBCOMMON_SRC} ${LIBCOMMON_HDR})
 if(GCC)
     set_source_files_properties(${LIBCOMMON_SRC} PROPERTIES COMPILE_FLAGS 
         "-Wno-sign-compare")
@@ -50,7 +30,7 @@
     # /wd4244 type conversion, possible loss of data
     # /wd4389 signed/unsigned mismatch
     # /wd4018 '<' signed/unsigned mismatch
-    # /wd4800 performance warning: bool coersion
+    # /wd4800 forcing value to bool 'true' or 'false' (performance warning)
     set_source_files_properties(${LIBCOMMON_SRC} PROPERTIES COMPILE_FLAGS 
         "/wd4244 /wd4389 /wd4018 /wd4800")
     if(INTEL_CXX)
@@ -58,16 +38,13 @@
     endif()
 endif(MSVC)
 
-set(SSE3  vec/dct-sse3.cpp  vec/blockcopy-sse3.cpp)
+set(SSE3  vec/dct-sse3.cpp)
 set(SSSE3 vec/dct-ssse3.cpp)
 set(SSE41 vec/dct-sse41.cpp)
 
 if(MSVC AND X86)
     set(PRIMITIVES ${SSE3} ${SSSE3} ${SSE41})
-    # wd4127 conditional expression is constant
-    # wd4244 'argument' : conversion from 'int' to 'char', possible loss of data
-    # wd4100 unreferenced formal parameter
-    set(WARNDISABLE "/wd4244 /wd4127 /wd4100")
+    set(WARNDISABLE "/wd4100") # unreferenced formal parameter
     if(INTEL_CXX)
         add_definitions(/Qwd111) # statement is unreachable
         add_definitions(/Qwd128) # loop is unreachable
@@ -89,7 +66,7 @@
     else()
         set(WARNDISABLE "-Wno-unused-parameter")
     endif()
-    if(INTEL_CXX OR CLANG OR (NOT GCC_VERSION VERSION_LESS 4.3))
+    if(INTEL_CXX OR CLANG OR (NOT CC_VERSION VERSION_LESS 4.3))
         set(PRIMITIVES ${SSE3} ${SSSE3} ${SSE41})
         set_source_files_properties(${SSE3}  PROPERTIES COMPILE_FLAGS "${WARNDISABLE} -msse3")
         set_source_files_properties(${SSSE3} PROPERTIES COMPILE_FLAGS "${WARNDISABLE} -mssse3")
@@ -119,10 +96,10 @@
         # MSVC requires custom build rules in the main cmake script for yasm
         set(MSVC_ASMS "${A_SRCS}" CACHE INTERNAL "yasm sources")
         set(A_SRCS)
-    else()
-        enable_language(ASM_YASM)
     endif()
 
+    enable_language(ASM_YASM)
+
     foreach(SRC ${A_SRCS} ${C_SRCS})
         set(ASM_PRIMITIVES ${ASM_PRIMITIVES} x86/${SRC})
     endforeach()
@@ -130,28 +107,38 @@
 endif(ENABLE_ASSEMBLY)
 
 check_symbol_exists(strtok_r "string.h" HAVE_STRTOK_R)
-set_source_files_properties(param.cpp PROPERTIES COMPILE_FLAGS -DHAVE_STRTOK_R=${HAVE_STRTOK_R})
+if(HAVE_STRTOK_R)
+    set_source_files_properties(param.cpp PROPERTIES COMPILE_FLAGS -DHAVE_STRTOK_R=1)
+endif()
 
-if(GCC AND GCC_HAS_NO_NARROWING)
+if(GCC AND CC_HAS_NO_NARROWING)
     set_source_files_properties(cpu.cpp PROPERTIES COMPILE_FLAGS -Wno-narrowing)
 endif()
+if(WIN32)
+    set(WINXP winxp.h winxp.cpp)
+endif(WIN32)
 
 set_source_files_properties(version.cpp PROPERTIES COMPILE_FLAGS -DX265_VERSION=${X265_VERSION})
 
 add_library(common OBJECT
     ${ASM_PRIMITIVES} ${VEC_PRIMITIVES}
-    ${LIBCOMMON_SRC} ${LIBCOMMON_HDR}
+    ${LIBCOMMON_SRC} ${LIBCOMMON_HDR} ${WINXP}
     primitives.cpp primitives.h
-    pixel.cpp dct.cpp ipfilter.cpp intrapred.cpp
+    pixel.cpp dct.cpp ipfilter.cpp intrapred.cpp loopfilter.cpp
     cpu.cpp cpu.h version.cpp
     threading.cpp threading.h
     threadpool.cpp threadpool.h
     wavefront.h wavefront.cpp
     md5.cpp md5.h
-    bitstream.h mv.h
+    bitstream.h bitstream.cpp
+    mv.h
     shortyuv.cpp shortyuv.h
     common.cpp common.h
     param.cpp param.h
+    frame.cpp frame.h
+    slice.cpp slice.h
     lowres.cpp lowres.h
     piclist.cpp piclist.h
-    loopfilter.cpp)
+    scalinglist.cpp scalinglist.h
+    quant.cpp quant.h
+    deblock.cpp deblock.h)

x265-hg20140928.tar.bz2/source/common/bitstream.cpp Added

@@ -0,0 +1,125 @@
+#include "common.h"
+#include "bitstream.h"
+
+using namespace x265;
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4244)
+#endif
+
+#define MIN_FIFO_SIZE 1000
+
+Bitstream::Bitstream()
+{
+    m_fifo = X265_MALLOC(uint8_t, MIN_FIFO_SIZE);
+    m_byteAlloc = MIN_FIFO_SIZE;
+    resetBits();
+}
+
+void Bitstream::push_back(uint8_t val)
+{
+    if (!m_fifo)
+        return;
+
+    if (m_byteOccupancy >= m_byteAlloc)
+    {
+        /** reallocate buffer with doubled size */
+        uint8_t *temp = X265_MALLOC(uint8_t, m_byteAlloc * 2);
+        if (temp)
+        {
+            ::memcpy(temp, m_fifo, m_byteOccupancy);
+            X265_FREE(m_fifo);
+            m_fifo = temp;
+            m_byteAlloc *= 2;
+        }
+        else
+        {
+            x265_log(NULL, X265_LOG_ERROR, "Unable to realloc bitstream buffer");
+            return;
+        }
+    }
+    m_fifo[m_byteOccupancy++] = val;
+}
+
+void Bitstream::write(uint32_t val, uint32_t numBits)
+{
+    X265_CHECK(numBits <= 32, "numBits out of range\n");
+    X265_CHECK(numBits == 32 || ((val & (~0 << numBits)) == 0), "numBits & val out of range\n");
+
+    uint32_t totalPartialBits = m_partialByteBits + numBits;
+    uint32_t nextPartialBits = totalPartialBits & 7;
+    uint8_t  nextHeldByte = val << (8 - nextPartialBits);
+    uint32_t writeBytes = totalPartialBits >> 3;
+
+    if (writeBytes)
+    {
+        /* topword aligns m_partialByte with the msb of val */
+        uint32_t topword = (numBits - nextPartialBits) & ~7;
+        uint32_t write_bits = (m_partialByte << topword) | (val >> nextPartialBits);
+
+        switch (writeBytes)
+        {
+        case 4: push_back(write_bits >> 24);
+        case 3: push_back(write_bits >> 16);
+        case 2: push_back(write_bits >> 8);
+        case 1: push_back(write_bits);
+        }
+
+        m_partialByte = nextHeldByte;
+        m_partialByteBits = nextPartialBits;
+    }
+    else
+    {
+        m_partialByte |= nextHeldByte;
+        m_partialByteBits = nextPartialBits;
+    }
+}
+
+void Bitstream::writeByte(uint32_t val)
+{
+    // Only CABAC will call writeByte, the fifo must be byte aligned
+    X265_CHECK(!m_partialByteBits, "expecting m_partialByteBits = 0\n");
+
+    push_back(val);
+}
+
+void Bitstream::writeAlignOne()
+{
+    uint32_t numBits = (8 - m_partialByteBits) & 0x7;
+
+    write((1 << numBits) - 1, numBits);
+}
+
+void Bitstream::writeAlignZero()
+{
+    if (m_partialByteBits)
+    {
+        push_back(m_partialByte);
+        m_partialByte = 0;
+        m_partialByteBits = 0;
+    }
+}
+
+void Bitstream::writeByteAlignment()
+{
+    write(1, 1);
+    writeAlignZero();
+}
+
+void SyntaxElementWriter::writeUvlc(uint32_t code)
+{
+    uint32_t length = 1;
+    uint32_t temp = ++code;
+
+    X265_CHECK(temp, "writing -1 code, will cause infinite loop\n");
+
+    while (1 != temp)
+    {
+        temp >>= 1;
+        length += 2;
+    }
+
+    // Take care of cases where length > 32
+    m_bitIf->write(0, length >> 1);
+    m_bitIf->write(code, (length + 1) >> 1);
+}

x265-hg20140324.tar.bz2/source/common/bitstream.h -> x265-hg20140928.tar.bz2/source/common/bitstream.h Changed

@@ -18,53 +18,141 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_BITSTREAM_H
 #define X265_BITSTREAM_H 1
 
+namespace x265 {
+// private namespace
+
+class BitInterface
+{
+public:
+
+    virtual void     write(uint32_t val, uint32_t numBits)  = 0;
+    virtual void     writeByte(uint32_t val)                = 0;
+    virtual void     resetBits()                            = 0;
+    virtual uint32_t getNumberOfWrittenBits() const         = 0;
+    virtual void     writeAlignOne()                        = 0;
+    virtual void     writeAlignZero()                       = 0;
+    virtual ~BitInterface() {}
+};
+
+class BitCounter : public BitInterface
+{
+protected:
+
+    uint32_t  m_bitCounter;
+
+public:
+
+    BitCounter() : m_bitCounter(0) {}
+
+    void     write(uint32_t, uint32_t num)  { m_bitCounter += num; }
+    void     writeByte(uint32_t)            { m_bitCounter += 8;   }
+    void     resetBits()                    { m_bitCounter = 0;    }
+    uint32_t getNumberOfWrittenBits() const { return m_bitCounter; }
+    void     writeAlignOne()                { }
+    void     writeAlignZero()               { }
+};
+
+
+class Bitstream : public BitInterface
+{
+public:
+
+    Bitstream();
+    ~Bitstream()                             { X265_FREE(m_fifo); }
+
+    void     resetBits()                     { m_partialByteBits = m_byteOccupancy = 0; m_partialByte = 0; }
+    uint32_t getNumberOfWrittenBytes() const { return m_byteOccupancy; }
+    uint32_t getNumberOfWrittenBits()  const { return m_byteOccupancy * 8 + m_partialByteBits; }
+    const uint8_t* getFIFO() const           { return m_fifo; }
+
+    void     write(uint32_t val, uint32_t numBits);
+    void     writeByte(uint32_t val);
+
+    void     writeAlignOne();      // insert one bits until the bitstream is byte-aligned
+    void     writeAlignZero();     // insert zero bits until the bitstream is byte-aligned
+    void     writeByteAlignment(); // insert 1 bit, then pad to byte-align with zero
+
+private:
+
+    uint8_t *m_fifo;
+    uint32_t m_byteAlloc;
+    uint32_t m_byteOccupancy;
+    uint32_t m_partialByteBits;
+    uint8_t  m_partialByte;
+
+    void     push_back(uint8_t val);
+};
+
 static const uint8_t bitSize[256] =
 {
-     1, 1, 3, 3, 5, 5, 5, 5, 7, 7, 7, 7, 7, 7, 7, 7,
-     9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
-    11,11,11,11,11,11,11,11,11,11,11,11,11,11,11,11,
-    11,11,11,11,11,11,11,11,11,11,11,11,11,11,11,11,
-    13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,
-    13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,
-    13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,
-    13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,13,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
-    15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,
+    1, 1, 3, 3, 5, 5, 5, 5, 7, 7, 7, 7, 7, 7, 7, 7,
+    9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9,
+    11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+    11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
+    13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
+    13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
+    13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
+    13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
+    15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15,
 };
 
-static inline int bs_size_ue( unsigned int val )
+static inline int bs_size_ue(unsigned int val)
 {
-    return bitSize[val+1];
+    return bitSize[val + 1];
 }
 
-static inline int bs_size_ue_big( unsigned int val )
+static inline int bs_size_ue_big(unsigned int val)
 {
-    if( val < 255 )
-        return bitSize[val+1];
+    if (val < 255)
+        return bitSize[val + 1];
     else
-        return bitSize[(val+1)>>8] + 16;
+        return bitSize[(val + 1) >> 8] + 16;
 }
 
-static inline int bs_size_se( int val )
+static inline int bs_size_se(int val)
 {
-    int tmp = 1 - val*2;
-    if( tmp < 0 ) tmp = val*2;
-    if( tmp < 256 )
+    int tmp = 1 - val * 2;
+
+    if (tmp < 0) tmp = val * 2;
+    if (tmp < 256)
         return bitSize[tmp];
     else
-        return bitSize[tmp>>8]+16;
+        return bitSize[tmp >> 8] + 16;
+}
+
+class SyntaxElementWriter
+{
+public:
+
+    BitInterface* m_bitIf;
+
+    SyntaxElementWriter() : m_bitIf(NULL) {}
+
+    /* silently discard the name of the syntax element */
+    inline void WRITE_CODE(uint32_t code, uint32_t length, const char *) { writeCode(code, length); }
+    inline void WRITE_UVLC(uint32_t code,                  const char *) { writeUvlc(code); }
+    inline void WRITE_SVLC(int32_t  code,                  const char *) { writeSvlc(code); }
+    inline void WRITE_FLAG(bool flag,                      const char *) { writeFlag(flag); }
+
+    void writeCode(uint32_t code, uint32_t length) { m_bitIf->write(code, length); }
+    void writeUvlc(uint32_t code);
+    void writeSvlc(int32_t code)                   { uint32_t ucode = (code <= 0) ? -code << 1 : (code << 1) - 1; writeUvlc(ucode); }
+    void writeFlag(bool code)                      { m_bitIf->write(code, 1); }
+};
+
 }
 
-#endif
+#endif // ifndef X265_BITSTREAM_H

x265-hg20140324.tar.bz2/source/common/common.cpp -> x265-hg20140928.tar.bz2/source/common/common.cpp Changed

@@ -18,15 +18,14 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/CommonDef.h"
+#include "common.h"
+#include "slice.h"
+#include "threading.h"
 #include "TLibCommon/TComRom.h"
-#include "TLibCommon/TComSlice.h"
 #include "x265.h"
-#include "threading.h"
-#include "common.h"
 
 #if _WIN32
 #include <sys/types.h>
@@ -98,7 +97,7 @@
     return (x265_exp2_lut[i & 63] + 256) << (i >> 6) >> 8;
 }
 
-void x265_log(x265_param *param, int level, const char *fmt, ...)
+void x265_log(const x265_param *param, int level, const char *fmt, ...)
 {
     if (param && level > param->logLevel)
         return;
@@ -153,3 +152,58 @@
 {
     return 0.85 * pow(2.0, (qp - 12.0) / 6.0);
 }
+
+uint32_t x265_picturePlaneSize(int csp, int width, int height, int plane)
+{
+    uint32_t size = (uint32_t)(width >> x265_cli_csps[csp].width[plane]) * (height >> x265_cli_csps[csp].height[plane]);
+
+    return size;
+}
+
+char* x265_slurp_file(const char *filename)
+{
+    if (!filename)
+        return NULL;
+
+    int bError = 0;
+    size_t fSize;
+    char *buf = NULL;
+
+    FILE *fh = fopen(filename, "rb");
+    if (!fh)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "unable to open file %s\n", filename);
+        return NULL;
+    }
+
+    bError |= fseek(fh, 0, SEEK_END) < 0;
+    bError |= (fSize = ftell(fh)) <= 0;
+    bError |= fseek(fh, 0, SEEK_SET) < 0;
+    if (bError)
+        goto error;
+
+    buf = X265_MALLOC(char, fSize + 2);
+    if (!buf)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "unable to allocate memory\n");
+        goto error;
+    }
+
+    bError |= fread(buf, 1, fSize, fh) != fSize;
+    if (buf[fSize - 1] != '\n')
+        buf[fSize++] = '\n';
+    buf[fSize] = 0;
+    fclose(fh);
+
+    if (bError)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "unable to read the file\n");
+        X265_FREE(buf);
+        buf = NULL;
+    }
+    return buf;
+
+error:
+    fclose(fh);
+    return NULL;
+}

x265-hg20140324.tar.bz2/source/common/common.h -> x265-hg20140928.tar.bz2/source/common/common.h Changed

x265-hg20140324.tar.bz2/source/common/cpu.cpp -> x265-hg20140928.tar.bz2/source/common/cpu.cpp Changed

@@ -3,7 +3,7 @@
  *
  * Authors: Loren Merritt <lorenm@u.washington.edu>
  *          Laurent Aimar <fenrir@via.ecp.fr>
- *          Jason Garrett-Glaser <darkshikari@gmail.com>
+ *          Fiona Glaser <fiona@x264.com>
  *          Steve Borho <steve@borho.org>
  *
  * This program is free software; you can redistribute it and/or modify
@@ -21,7 +21,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "cpu.h"
@@ -244,7 +244,7 @@
             if (model == 9 || model == 13 || model == 14)
             {
                 cpu &= ~(X265_CPU_SSE2 | X265_CPU_SSE3);
-                assert(!(cpu & (X265_CPU_SSSE3 | X265_CPU_SSE4)));
+                X265_CHECK(!(cpu & (X265_CPU_SSSE3 | X265_CPU_SSE4)), "unexpected CPU ID %d\n", cpu);
             }
             /* Detect Atom CPU */
             else if (model == 28)

x265-hg20140324.tar.bz2/source/common/cpu.h -> x265-hg20140928.tar.bz2/source/common/cpu.h Changed

x265-hg20140324.tar.bz2/source/common/dct.cpp -> x265-hg20140928.tar.bz2/source/common/dct.cpp Changed

@@ -24,7 +24,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "common.h"
@@ -718,36 +718,35 @@
     }
 }
 
-void dequant_normal_c(const int32_t* quantCoef, int32_t* coef, int num, int scale, int shift)
+void dequant_normal_c(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift)
 {
-#if !HIGH_BIT_DEPTH
+#if HIGH_BIT_DEPTH
+    X265_CHECK(scale < 32768 || ((scale & 3) == 0 && shift > 2), "dequant invalid scale %d\n", scale);
+#else
     // NOTE: maximum of scale is (72 * 256)
-    assert(scale < 32768);
+    X265_CHECK(scale < 32768, "dequant invalid scale %d\n", scale);
 #endif
-    assert(num <= 32 * 32);
-    assert((num % 8) == 0);
-    assert(shift <= 10);
+    X265_CHECK(num <= 32 * 32, "dequant num %d too large\n", num);
+    X265_CHECK((num % 8) == 0, "dequant num %d not multiple of 8\n", num);
+    X265_CHECK(shift <= 10, "shift too large %d\n", shift);
+    X265_CHECK(((intptr_t)coef & 31) == 0, "dequant coef buffer not aligned\n");
 
     int add, coeffQ;
 
-    int clipQCoef;
-
     add = 1 << (shift - 1);
 
     for (int n = 0; n < num; n++)
     {
-        clipQCoef = Clip3(-32768, 32767, quantCoef[n]);
-        coeffQ = (clipQCoef * scale + add) >> shift;
+        coeffQ = (quantCoef[n] * scale + add) >> shift;
         coef[n] = Clip3(-32768, 32767, coeffQ);
     }
 }
 
-void dequant_scaling_c(const int32_t* quantCoef, const int32_t *deQuantCoef, int32_t* coef, int num, int per, int shift)
+void dequant_scaling_c(const int16_t* quantCoef, const int32_t *deQuantCoef, int32_t* coef, int num, int per, int shift)
 {
-    assert(num <= 32 * 32);
+    X265_CHECK(num <= 32 * 32, "dequant num %d too large\n", num);
 
     int add, coeffQ;
-    int clipQCoef;
 
     shift += 4;
 
@@ -757,8 +756,7 @@
 
         for (int n = 0; n < num; n++)
         {
-            clipQCoef = Clip3(-32768, 32767, quantCoef[n]);
-            coeffQ = ((clipQCoef * deQuantCoef[n]) + add) >> (shift - per);
+            coeffQ = ((quantCoef[n] * deQuantCoef[n]) + add) >> (shift - per);
             coef[n] = Clip3(-32768, 32767, coeffQ);
         }
     }
@@ -766,42 +764,64 @@
     {
         for (int n = 0; n < num; n++)
         {
-            clipQCoef = Clip3(-32768, 32767, quantCoef[n]);
-            coeffQ   = Clip3(-32768, 32767, clipQCoef * deQuantCoef[n]);
+            coeffQ   = Clip3(-32768, 32767, quantCoef[n] * deQuantCoef[n]);
             coef[n] = Clip3(-32768, 32767, coeffQ << (per - shift));
         }
     }
 }
 
-uint32_t quant_c(int32_t* coef, int32_t* quantCoeff, int32_t* deltaU, int32_t* qCoef, int qBits, int add, int numCoeff, int32_t* lastPos)
+uint32_t quant_c(int32_t* coef, int32_t* quantCoeff, int32_t* deltaU, int16_t* qCoef, int qBits, int add, int numCoeff)
 {
+    X265_CHECK(qBits >= 8, "qBits less than 8\n");
+    X265_CHECK((numCoeff % 16) == 0, "numCoeff must be multiple of 16\n");
     int qBits8 = qBits - 8;
-    uint32_t acSum = 0;
+    uint32_t numSig = 0;
 
     for (int blockpos = 0; blockpos < numCoeff; blockpos++)
     {
-        int level;
-        int sign;
-        level = coef[blockpos];
-        sign  = (level < 0 ? -1 : 1);
+        int level = coef[blockpos];
+        int sign  = (level < 0 ? -1 : 1);
 
         int tmplevel = abs(level) * quantCoeff[blockpos];
         level = ((tmplevel + add) >> qBits);
-        if (level)
-            *lastPos = blockpos;
         deltaU[blockpos] = ((tmplevel - (level << qBits)) >> qBits8);
-        acSum += level;
+        if (level)
+            ++numSig;
+        level *= sign;
+        qCoef[blockpos] = (int16_t)Clip3(-32768, 32767, level);
+    }
+
+    return numSig;
+}
+
+uint32_t nquant_c(int32_t* coef, int32_t* quantCoeff, int16_t* qCoef, int qBits, int add, int numCoeff)
+{
+    X265_CHECK((numCoeff % 16) == 0, "number of quant coeff is not multiple of 4x4\n");
+    X265_CHECK((uint32_t)add < ((uint32_t)1 << qBits), "2 ^ qBits less than add\n");
+    X265_CHECK(((intptr_t)quantCoeff & 31) == 0, "quantCoeff buffer not aligned\n");
+
+    uint32_t numSig = 0;
+
+    for (int blockpos = 0; blockpos < numCoeff; blockpos++)
+    {
+        int level = coef[blockpos];
+        int sign  = (level < 0 ? -1 : 1);
+
+        int tmplevel = abs(level) * quantCoeff[blockpos];
+        level = ((tmplevel + add) >> qBits);
+        if (level)
+            ++numSig;
         level *= sign;
-        qCoef[blockpos] = Clip3(-32768, 32767, level);
+        qCoef[blockpos] = (int16_t)Clip3(-32768, 32767, level);
     }
 
-    return acSum;
+    return numSig;
 }
 
-int  count_nonzero_c(const int32_t *quantCoeff, int numCoeff)
+int  count_nonzero_c(const int16_t *quantCoeff, int numCoeff)
 {
-    assert(((intptr_t)quantCoeff & 15) == 0);
-    assert(numCoeff > 0 && (numCoeff & 15) == 0);
+    X265_CHECK(((intptr_t)quantCoeff & 15) == 0, "quant buffer not aligned\n");
+    X265_CHECK(numCoeff > 0 && (numCoeff & 15) == 0, "numCoeff invalid %d\n", numCoeff);
 
     int count = 0;
 
@@ -812,6 +832,36 @@
 
     return count;
 }
+
+template<int trSize>
+uint32_t copy_count(int16_t* coeff, int16_t* residual, intptr_t stride)
+{
+    uint32_t numSig = 0;
+    for (int k = 0; k < trSize; k++)
+    {
+        for (int j = 0; j < trSize; j++)
+        {
+            coeff[k * trSize + j] = residual[k * stride + j];
+            numSig += (residual[k * stride + j] != 0);
+        }
+    }
+
+    return numSig;
+}
+
+void denoiseDct_c(int32_t* dctCoef, uint32_t* resSum, uint16_t* offset, int numCoeff)
+{
+    for (int i = 0; i < numCoeff; i++)
+    {
+        int level = dctCoef[i];
+        int sign = level >> 31;
+        level = (level + sign) ^ sign;
+        resSum[i] += level;
+        level -= offset[i];
+        dctCoef[i] = level < 0 ? 0 : (level ^ sign) - sign;
+    }
+}
+
 }  // closing - anonymous file-static namespace
 
 namespace x265 {
@@ -822,6 +872,7 @@
     p.dequant_scaling = dequant_scaling_c;
     p.dequant_normal = dequant_normal_c;
     p.quant = quant_c;
+    p.nquant = nquant_c;
     p.dct[DST_4x4] = dst4_c;
     p.dct[DCT_4x4] = dct4_c;
     p.dct[DCT_8x8] = dct8_c;
@@ -833,5 +884,11 @@
     p.idct[IDCT_16x16] = idct16_c;
     p.idct[IDCT_32x32] = idct32_c;
     p.count_nonzero = count_nonzero_c;
+    p.denoiseDct = denoiseDct_c;
+
+    p.copy_cnt[BLOCK_4x4] = copy_count<4>;
+    p.copy_cnt[BLOCK_8x8] = copy_count<8>;
+    p.copy_cnt[BLOCK_16x16] = copy_count<16>;
+    p.copy_cnt[BLOCK_32x32] = copy_count<32>;
 }
 }

x265-hg20140928.tar.bz2/source/common/deblock.cpp Added

@@ -0,0 +1,646 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Author: Gopu Govindaswamy <gopu@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "common.h"
+#include "deblock.h"
+#include "frame.h"
+#include "slice.h"
+#include "mv.h"
+
+using namespace x265;
+
+#define DEBLOCK_SMALLEST_BLOCK  8
+#define DEFAULT_INTRA_TC_OFFSET 2
+
+void Deblock::deblockCTU(TComDataCU* cu, int32_t dir)
+{
+    uint8_t blockingStrength[MAX_NUM_PARTITIONS];
+
+    memset(blockingStrength, 0, sizeof(uint8_t) * m_numPartitions);
+
+    deblockCU(cu, 0, 0, dir, blockingStrength);
+}
+
+/* Deblocking filter process in CU-based (the same function as conventional's)
+ * param Edge the direction of the edge in block boundary (horizonta/vertical), which is added newly */
+void Deblock::deblockCU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, const int32_t dir, uint8_t blockingStrength[])
+{
+    if (!cu->m_pic || cu->getPartitionSize(absZOrderIdx) == SIZE_NONE)
+        return;
+
+    Frame* pic = cu->m_pic;
+    uint32_t curNumParts = NUM_CU_PARTITIONS >> (depth << 1);
+
+    if (cu->getDepth(absZOrderIdx) > depth)
+    {
+        uint32_t qNumParts   = curNumParts >> 2;
+        uint32_t xmax = cu->m_slice->m_sps->picWidthInLumaSamples  - cu->getCUPelX();
+        uint32_t ymax = cu->m_slice->m_sps->picHeightInLumaSamples - cu->getCUPelY();
+        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
+            if (g_zscanToPelX[absZOrderIdx] < xmax && g_zscanToPelY[absZOrderIdx] < ymax)
+                deblockCU(cu, absZOrderIdx, depth + 1, dir, blockingStrength);
+        return;
+    }
+
+    const uint32_t widthInBaseUnits  = cu->m_pic->getNumPartInCUSize() >> depth;
+    Param params;
+    setLoopfilterParam(cu, absZOrderIdx, &params);
+    setEdgefilterPU(cu, absZOrderIdx, dir, blockingStrength, widthInBaseUnits);
+    setEdgefilterTU(cu, absZOrderIdx, depth, dir, blockingStrength);
+    setEdgefilterMultiple(cu, absZOrderIdx, dir, 0, (dir == EDGE_VER ? params.leftEdge : params.topEdge), blockingStrength, widthInBaseUnits);
+
+    for (uint32_t partIdx = absZOrderIdx; partIdx < absZOrderIdx + curNumParts; partIdx++)
+    {
+        uint32_t bsCheck = !(partIdx & (1 << dir));
+
+        if (bsCheck && blockingStrength[partIdx])
+            getBoundaryStrengthSingle(cu, dir, partIdx, blockingStrength);
+    }
+
+    const uint32_t partIdxIncr = DEBLOCK_SMALLEST_BLOCK >> LOG2_UNIT_SIZE;
+    uint32_t sizeInPU = pic->getNumPartInCUSize() >> depth;
+    uint32_t shiftFactor = (dir == EDGE_VER) ? cu->getHorzChromaShift() : cu->getVertChromaShift();
+    uint32_t chromaMask = ((DEBLOCK_SMALLEST_BLOCK << shiftFactor) >> LOG2_UNIT_SIZE) - 1;
+    uint32_t e0 = (dir == EDGE_VER ? g_zscanToPelX[absZOrderIdx] : g_zscanToPelY[absZOrderIdx]) >> LOG2_UNIT_SIZE;
+        
+    for (uint32_t e = 0; e < sizeInPU; e += partIdxIncr)
+    {
+        edgeFilterLuma(cu, absZOrderIdx, depth, dir, e, blockingStrength);
+        if (!((e0 + e) & chromaMask))
+            edgeFilterChroma(cu, absZOrderIdx, depth, dir, e, blockingStrength);
+    }
+}
+
+static inline uint32_t calcBsIdx(TComDataCU* cu, uint32_t absZOrderIdx, int32_t dir, int32_t edgeIdx, int32_t baseUnitIdx)
+{
+    uint32_t lcuWidthInBaseUnits = cu->m_pic->getNumPartInCUSize();
+
+    if (dir)
+        return g_rasterToZscan[g_zscanToRaster[absZOrderIdx] + edgeIdx * lcuWidthInBaseUnits + baseUnitIdx];
+    else
+        return g_rasterToZscan[g_zscanToRaster[absZOrderIdx] + baseUnitIdx * lcuWidthInBaseUnits + edgeIdx];
+}
+
+void Deblock::setEdgefilterMultiple(TComDataCU* cu, uint32_t scanIdx, int32_t dir, int32_t edgeIdx, uint8_t value, uint8_t blockingStrength[], uint32_t widthInBaseUnits)
+{
+    const uint32_t numElem = widthInBaseUnits;
+    X265_CHECK(numElem > 0, "numElem edge filter check\n");
+    for (uint32_t i = 0; i < numElem; i++)
+    {
+        const uint32_t bsidx = calcBsIdx(cu, scanIdx, dir, edgeIdx, i);
+        blockingStrength[bsidx] = value;
+    }
+}
+
+void Deblock::setEdgefilterTU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, uint8_t blockingStrength[])
+{
+    if (cu->getTransformIdx(absZOrderIdx) + cu->getDepth(absZOrderIdx) > (uint8_t)depth)
+    {
+        const uint32_t curNumParts = NUM_CU_PARTITIONS >> (depth << 1);
+        const uint32_t qNumParts   = curNumParts >> 2;
+
+        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
+            setEdgefilterTU(cu, absZOrderIdx, depth + 1, dir, blockingStrength);
+        return;
+    }
+
+    uint32_t widthInBaseUnits  = 1 << (cu->getLog2CUSize(absZOrderIdx) - cu->getTransformIdx(absZOrderIdx) - LOG2_UNIT_SIZE);
+    setEdgefilterMultiple(cu, absZOrderIdx, dir, 0, 2, blockingStrength, widthInBaseUnits);
+}
+
+void Deblock::setEdgefilterPU(TComDataCU* cu, uint32_t absZOrderIdx, int32_t dir, uint8_t blockingStrength[], uint32_t widthInBaseUnits)
+{
+    const uint32_t hWidthInBaseUnits = widthInBaseUnits >> 1;
+    const uint32_t qWidthInBaseUnits = widthInBaseUnits >> 2;
+
+    switch (cu->getPartitionSize(absZOrderIdx))
+    {
+    case SIZE_2NxN:
+        if (EDGE_HOR == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, hWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_Nx2N:
+        if (EDGE_VER == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, hWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_NxN:
+        setEdgefilterMultiple(cu, absZOrderIdx, dir, hWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_2NxnU:
+        if (EDGE_HOR == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, qWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_nLx2N:
+        if (EDGE_VER == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, qWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_2NxnD:
+        if (EDGE_HOR == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, widthInBaseUnits - qWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+    case SIZE_nRx2N:
+        if (EDGE_VER == dir)
+            setEdgefilterMultiple(cu, absZOrderIdx, dir, widthInBaseUnits - qWidthInBaseUnits, 1, blockingStrength, widthInBaseUnits);
+        break;
+
+    case SIZE_2Nx2N:
+    default:
+        break;
+    }
+}
+
+void Deblock::setLoopfilterParam(TComDataCU* cu, uint32_t absZOrderIdx, Param *params)
+{
+    uint32_t x = cu->getCUPelX() + g_zscanToPelX[absZOrderIdx];
+    uint32_t y = cu->getCUPelY() + g_zscanToPelY[absZOrderIdx];
+
+    TComDataCU* tempCU;
+    uint32_t    tempPartIdx;
+
+    if (!x)
+        params->leftEdge = 0;
+    else
+    {
+        tempCU = cu->getPULeft(tempPartIdx, absZOrderIdx);
+        if (tempCU)
+            params->leftEdge = 2;
+        else
+            params->leftEdge = 0;
+    }
+
+    if (!y)
+        params->topEdge = 0;
+    else
+    {
+        tempCU = cu->getPUAbove(tempPartIdx, absZOrderIdx);
+        if (tempCU)
+            params->topEdge = 2;
+        else
+            params->topEdge = 0;
+    }
+}
+
+void Deblock::getBoundaryStrengthSingle(TComDataCU* cu, int32_t dir, uint32_t absPartIdx, uint8_t blockingStrength[])
+{
+    Slice* const slice = cu->m_slice;
+    const uint32_t partQ = absPartIdx;
+    TComDataCU* const cuQ = cu;
+
+    uint32_t partP;
+    TComDataCU* cuP;
+    uint8_t bs = 0;
+
+    // Calculate block index
+    if (dir == EDGE_VER)
+        cuP = cuQ->getPULeft(partP, partQ);
+    else // (dir == EDGE_HOR)
+        cuP = cuQ->getPUAbove(partP, partQ);
+
+    // Set BS for Intra MB : BS = 4 or 3
+    if (cuP->isIntra(partP) || cuQ->isIntra(partQ))
+        bs = 2;
+
+    // Set BS for not Intra MB : BS = 2 or 1 or 0
+    if (!cuP->isIntra(partP) && !cuQ->isIntra(partQ))
+    {
+        uint32_t nsPartQ = partQ;
+        uint32_t nsPartP = partP;
+
+        if (blockingStrength[absPartIdx] > 1 &&
+            (cuQ->getCbf(nsPartQ, TEXT_LUMA, cuQ->getTransformIdx(nsPartQ)) ||
+             cuP->getCbf(nsPartP, TEXT_LUMA, cuP->getTransformIdx(nsPartP))))
+            bs = 1;
+        else
+        {
+            if (dir == EDGE_HOR)
+                cuP = cuQ->getPUAbove(partP, partQ);
+
+            if (slice->isInterB() || cuP->m_slice->isInterB())
+            {
+                int32_t refIdx;
+                Frame *refP0, *refP1, *refQ0, *refQ1;
+                refIdx = cuP->getCUMvField(0)->getRefIdx(partP);
+                refP0 = (refIdx < 0) ? NULL : cuP->m_slice->m_refPicList[0][refIdx];
+                refIdx = cuP->getCUMvField(1)->getRefIdx(partP);
+                refP1 = (refIdx < 0) ? NULL : cuP->m_slice->m_refPicList[1][refIdx];
+                refIdx = cuQ->getCUMvField(0)->getRefIdx(partQ);
+                refQ0 = (refIdx < 0) ? NULL : slice->m_refPicList[0][refIdx];
+                refIdx = cuQ->getCUMvField(1)->getRefIdx(partQ);
+                refQ1 = (refIdx < 0) ? NULL : slice->m_refPicList[1][refIdx];
+
+                MV mvp0 = cuP->getCUMvField(0)->getMv(partP);
+                MV mvp1 = cuP->getCUMvField(1)->getMv(partP);
+                MV mvq0 = cuQ->getCUMvField(0)->getMv(partQ);
+                MV mvq1 = cuQ->getCUMvField(1)->getMv(partQ);
+
+                if (!refP0) mvp0 = 0;
+                if (!refP1) mvp1 = 0;
+                if (!refQ0) mvq0 = 0;
+                if (!refQ1) mvq1 = 0;
+
+                if (((refP0 == refQ0) && (refP1 == refQ1)) || ((refP0 == refQ1) && (refP1 == refQ0)))
+                {
+                    if (refP0 != refP1) // Different L0 & L1
+                    {
+                        if (refP0 == refQ0)
+                        {
+                            bs  = ((abs(mvq0.x - mvp0.x) >= 4) ||
+                                   (abs(mvq0.y - mvp0.y) >= 4) ||
+                                   (abs(mvq1.x - mvp1.x) >= 4) ||
+                                   (abs(mvq1.y - mvp1.y) >= 4)) ? 1 : 0;
+                        }
+                        else
+                        {
+                            bs  = ((abs(mvq1.x - mvp0.x) >= 4) ||
+                                   (abs(mvq1.y - mvp0.y) >= 4) ||
+                                   (abs(mvq0.x - mvp1.x) >= 4) ||
+                                   (abs(mvq0.y - mvp1.y) >= 4)) ? 1 : 0;
+                        }
+                    }
+                    else // Same L0 & L1
+                    {
+                        bs  = ((abs(mvq0.x - mvp0.x) >= 4) ||
+                               (abs(mvq0.y - mvp0.y) >= 4) ||
+                               (abs(mvq1.x - mvp1.x) >= 4) ||
+                               (abs(mvq1.y - mvp1.y) >= 4)) &&
+                              ((abs(mvq1.x - mvp0.x) >= 4) ||
+                               (abs(mvq1.y - mvp0.y) >= 4) ||
+                               (abs(mvq0.x - mvp1.x) >= 4) ||
+                               (abs(mvq0.y - mvp1.y) >= 4)) ? 1 : 0;
+                    }
+                }
+                else // for all different Ref_Idx
+                    bs = 1;
+            }
+            else // slice->isInterP()
+            {
+                int32_t refIdx;
+                Frame *refp0, *refq0;
+                refIdx = cuP->getCUMvField(0)->getRefIdx(partP);
+                refp0 = (refIdx < 0) ? NULL : cuP->m_slice->m_refPicList[0][refIdx];
+                refIdx = cuQ->getCUMvField(0)->getRefIdx(partQ);
+                refq0 = (refIdx < 0) ? NULL : slice->m_refPicList[0][refIdx];
+                MV mvp0 = cuP->getCUMvField(0)->getMv(partP);
+                MV mvq0 = cuQ->getCUMvField(0)->getMv(partQ);
+
+                if (!refp0) mvp0 = 0;
+                if (!refq0) mvq0 = 0;
+
+                bs = ((refp0 != refq0) ||
+                      (abs(mvq0.x - mvp0.x) >= 4) ||
+                      (abs(mvq0.y - mvp0.y) >= 4)) ? 1 : 0;
+            }
+        }
+    }
+
+    blockingStrength[absPartIdx] = bs;
+}
+
+static inline int32_t calcDP(pixel* src, int32_t offset)
+{
+    return abs(static_cast<int32_t>(src[-offset * 3]) - 2 * src[-offset * 2] + src[-offset]);
+}
+
+static inline int32_t calcDQ(pixel* src, int32_t offset)
+{
+    return abs(static_cast<int32_t>(src[0]) - 2 * src[offset] + src[offset * 2]);
+}
+
+static inline bool useStrongFiltering(int32_t offset, int32_t beta, int32_t tc, pixel* src)
+{
+    int16_t m4     = (int16_t)src[0];
+    int16_t m3     = (int16_t)src[-offset];
+    int16_t m7     = (int16_t)src[offset * 3];
+    int16_t m0     = (int16_t)src[-offset * 4];
+    int32_t strong = abs(m0 - m3) + abs(m7 - m4);
+
+    return (strong < (beta >> 3)) && (abs(m3 - m4) < ((tc * 5 + 1) >> 1));
+}
+
+/* Deblocking for the luminance component with strong or weak filter
+ * \param src            pointer to picture data
+ * \param offset         offset value for picture data
+ * \param tc             tc value
+ * \param partPNoFilter  indicator to disable filtering on partP
+ * \param partQNoFilter  indicator to disable filtering on partQ
+ * \param filterSecondP  decision weak filter/no filter for partP
+ * \param filterSecondQ  decision weak filter/no filter for partQ */
+static inline void pelFilterLumaStrong(pixel* src, int32_t srcStep, int32_t offset, int32_t tc, bool partPNoFilter, bool partQNoFilter)
+{
+    for (int32_t i = 0; i < UNIT_SIZE; i++, src += srcStep)
+    {
+        int16_t m4  = (int16_t)src[0];
+        int16_t m3  = (int16_t)src[-offset];
+        int16_t m5  = (int16_t)src[offset];
+        int16_t m2  = (int16_t)src[-offset * 2];
+        int32_t tc2 = 2 * tc;
+        if (!partPNoFilter)
+        {
+            int16_t m1  = (int16_t)src[-offset * 3];
+            int16_t m0  = (int16_t)src[-offset * 4];
+            src[-offset * 3] = (pixel)(Clip3(-tc2, tc2, ((2 * m0 + 3 * m1 + m2 + m3 + m4 + 4) >> 3) - m1) + m1);
+            src[-offset * 2] = (pixel)(Clip3(-tc2, tc2, ((m1 + m2 + m3 + m4 + 2) >> 2) - m2) + m2);
+            src[-offset]     = (pixel)(Clip3(-tc2, tc2, ((m1 + 2 * m2 + 2 * m3 + 2 * m4 + m5 + 4) >> 3) - m3) + m3);
+        }
+        if (!partQNoFilter)
+        {
+            int16_t m6  = (int16_t)src[offset * 2];
+            int16_t m7  = (int16_t)src[offset * 3];
+            src[0]           = (pixel)(Clip3(-tc2, tc2, ((m2 + 2 * m3 + 2 * m4 + 2 * m5 + m6 + 4) >> 3) - m4) + m4);
+            src[offset]      = (pixel)(Clip3(-tc2, tc2, ((m3 + m4 + m5 + m6 + 2) >> 2) - m5) + m5);
+            src[offset * 2]  = (pixel)(Clip3(-tc2, tc2, ((m3 + m4 + m5 + 3 * m6 + 2 * m7 + 4) >> 3) - m6) + m6);
+        }
+    }
+}
+
+/* Weak filter */
+static inline void pelFilterLuma(pixel* src, int32_t srcStep, int32_t offset, int32_t tc, bool partPNoFilter, bool partQNoFilter,
+                                 bool filterSecondP, bool filterSecondQ)
+{
+    int32_t thrCut = tc * 10;
+
+    for (int32_t i = 0; i < UNIT_SIZE; i++, src += srcStep)
+    {
+        int16_t m4  = (int16_t)src[0];
+        int16_t m3  = (int16_t)src[-offset];
+        int16_t m5  = (int16_t)src[offset];
+        int16_t m2  = (int16_t)src[-offset * 2];
+
+        int32_t delta = (9 * (m4 - m3) - 3 * (m5 - m2) + 8) >> 4;
+
+        if (abs(delta) < thrCut)
+        {
+            delta = Clip3(-tc, tc, delta);
+
+            int32_t tc2 = tc >> 1;
+            if (!partPNoFilter)
+            {
+                src[-offset] = Clip(m3 + delta);
+                if (filterSecondP)
+                {
+                    int16_t m1  = (int16_t)src[-offset * 3];
+                    int32_t delta1 = Clip3(-tc2, tc2, ((((m1 + m3 + 1) >> 1) - m2 + delta) >> 1));
+                    src[-offset * 2] = Clip(m2 + delta1);
+                }
+            }
+            if (!partQNoFilter)
+            {
+                src[0] = Clip(m4 - delta);
+                if (filterSecondQ)
+                {
+                    int16_t m6  = (int16_t)src[offset * 2];
+                    int32_t delta2 = Clip3(-tc2, tc2, ((((m6 + m4 + 1) >> 1) - m5 - delta) >> 1));
+                    src[offset] = Clip(m5 + delta2);
+                }
+            }
+        }
+    }
+}
+
+/* Deblocking of one line/column for the chrominance component
+ * \param src            pointer to picture data
+ * \param offset         offset value for picture data
+ * \param tc             tc value
+ * \param partPNoFilter  indicator to disable filtering on partP
+ * \param partQNoFilter  indicator to disable filtering on partQ */
+static inline void pelFilterChroma(pixel* src, int32_t srcStep, int32_t offset, int32_t tc, bool partPNoFilter, bool partQNoFilter)
+{
+    for (int32_t i = 0; i < UNIT_SIZE; i++, src += srcStep)
+    {
+        int16_t m4  = (int16_t)src[0];
+        int16_t m3  = (int16_t)src[-offset];
+        int16_t m5  = (int16_t)src[offset];
+        int16_t m2  = (int16_t)src[-offset * 2];
+
+        int32_t delta = Clip3(-tc, tc, ((((m4 - m3) << 2) + m2 - m5 + 4) >> 3));
+        if (!partPNoFilter)
+            src[-offset] = Clip(m3 + delta);
+        if (!partQNoFilter)
+            src[0] = Clip(m4 - delta);
+    }
+}
+
+void Deblock::edgeFilterLuma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, int32_t edge, const uint8_t blockingStrength[])
+{
+    TComPicYuv* reconYuv = cu->m_pic->getPicYuvRec();
+    pixel* src = reconYuv->getLumaAddr(cu->getAddr(), absZOrderIdx);
+
+    int32_t stride = reconYuv->getStride();
+    uint32_t numParts = cu->m_pic->getNumPartInCUSize() >> depth;
+
+    int32_t offset, srcStep;
+
+    bool  partPNoFilter = false;
+    bool  partQNoFilter = false;
+    uint32_t  partP = 0;
+    uint32_t  partQ = 0;
+    TComDataCU* cuP = cu;
+    TComDataCU* cuQ = cu;
+    int32_t betaOffset = cuQ->m_slice->m_pps->deblockingFilterBetaOffsetDiv2 << 1;
+    int32_t tcOffset = cuQ->m_slice->m_pps->deblockingFilterTcOffsetDiv2 << 1;
+
+    if (dir == EDGE_VER)
+    {
+        offset = 1;
+        srcStep = stride;
+        src += (edge << LOG2_UNIT_SIZE);
+    }
+    else // (dir == EDGE_HOR)
+    {
+        offset = stride;
+        srcStep = 1;
+        src += (edge << LOG2_UNIT_SIZE) * stride;
+    }
+
+    for (uint32_t idx = 0; idx < numParts; idx++)
+    {
+        uint32_t unitOffset = idx << LOG2_UNIT_SIZE;
+        uint32_t bsAbsIdx = calcBsIdx(cu, absZOrderIdx, dir, edge, idx);
+        uint32_t bs = blockingStrength[bsAbsIdx];
+        if (bs)
+        {
+            int32_t qpQ = cu->getQP(bsAbsIdx);
+            partQ = bsAbsIdx;
+
+            // Derive neighboring PU index
+            if (dir == EDGE_VER)
+                cuP = cuQ->getPULeft(partP, partQ);
+            else // (dir == EDGE_HOR)
+                cuP = cuQ->getPUAbove(partP, partQ);
+
+            int32_t qpP = cuP->getQP(partP);
+            int32_t qp = (qpP + qpQ + 1) >> 1;
+
+            int32_t indexB = Clip3(0, QP_MAX_SPEC, qp + betaOffset);
+
+            const int32_t bitdepthShift = X265_DEPTH - 8;
+            int32_t beta = s_betaTable[indexB] << bitdepthShift;
+
+            int32_t dp0 = calcDP(src + srcStep * (unitOffset + 0), offset);
+            int32_t dq0 = calcDQ(src + srcStep * (unitOffset + 0), offset);
+            int32_t dp3 = calcDP(src + srcStep * (unitOffset + 3), offset);
+            int32_t dq3 = calcDQ(src + srcStep * (unitOffset + 3), offset);
+            int32_t d0 = dp0 + dq0;
+            int32_t d3 = dp3 + dq3;
+
+            int32_t d =  d0 + d3;
+
+            if (d < beta)
+            {
+                if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+                {
+                    // check if each of PUs is lossless coded
+                    partPNoFilter = cuP->getCUTransquantBypass(partP);
+                    partQNoFilter = cuQ->getCUTransquantBypass(partQ);
+                }
+
+                int32_t indexTC = Clip3(0, QP_MAX_SPEC + DEFAULT_INTRA_TC_OFFSET, int32_t(qp + DEFAULT_INTRA_TC_OFFSET * (bs - 1) + tcOffset));
+                int32_t tc = s_tcTable[indexTC] << bitdepthShift;
+
+                bool sw = (2 * d0 < (beta >> 2) &&
+                           2 * d3 < (beta >> 2) &&
+                           useStrongFiltering(offset, beta, tc, src + srcStep * (unitOffset + 0)) &&
+                           useStrongFiltering(offset, beta, tc, src + srcStep * (unitOffset + 3)));
+
+                if (sw)
+                    pelFilterLumaStrong(src + srcStep * unitOffset, srcStep, offset, tc, partPNoFilter, partQNoFilter);
+                else
+                {
+                    int32_t sideThreshold = (beta + (beta >> 1)) >> 3;
+                    int32_t dp = dp0 + dp3;
+                    int32_t dq = dq0 + dq3;
+                    bool filterP = (dp < sideThreshold);
+                    bool filterQ = (dq < sideThreshold);
+
+                    pelFilterLuma(src + srcStep * unitOffset, srcStep, offset, tc, partPNoFilter, partQNoFilter, filterP, filterQ);
+                }
+            }
+        }
+    }
+}
+
+void Deblock::edgeFilterChroma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, int32_t edge, const uint8_t blockingStrength[])
+{
+    int32_t chFmt = cu->getChromaFormat();
+    int32_t offset, srcStep, chromaShift;
+
+    bool partPNoFilter = false;
+    bool partQNoFilter = false;
+    uint32_t partP;
+    uint32_t partQ;
+    TComDataCU* cuP;
+    TComDataCU* cuQ = cu;
+    int32_t tcOffset = cu->m_slice->m_pps->deblockingFilterTcOffsetDiv2 << 1;
+
+    X265_CHECK(((dir == EDGE_VER)
+                ? ((g_zscanToPelX[absZOrderIdx] + edge * UNIT_SIZE) >> cu->getHorzChromaShift())
+                : ((g_zscanToPelY[absZOrderIdx] + edge * UNIT_SIZE) >> cu->getVertChromaShift())) % DEBLOCK_SMALLEST_BLOCK == 0,
+               "invalid edge\n");
+
+
+    TComPicYuv* reconYuv = cu->m_pic->getPicYuvRec();
+    int32_t stride = reconYuv->getCStride();
+    int32_t srcOffset = reconYuv->getChromaAddrOffset(cu->getAddr(), absZOrderIdx);
+
+    if (dir == EDGE_VER)
+    {
+        chromaShift = cu->getVertChromaShift();
+        srcOffset += (edge << (LOG2_UNIT_SIZE - cu->getHorzChromaShift()));
+        offset     = 1;
+        srcStep    = stride;
+    }
+    else // (dir == EDGE_HOR)
+    {
+        chromaShift = cu->getHorzChromaShift();
+        srcOffset += edge * stride << (LOG2_UNIT_SIZE - cu->getVertChromaShift());
+        offset     = stride;
+        srcStep    = 1;
+    }
+
+    pixel* srcChroma[2];
+    srcChroma[0] = reconYuv->getCbAddr() + srcOffset;
+    srcChroma[1] = reconYuv->getCrAddr() + srcOffset;
+
+    uint32_t numUnits = cu->m_pic->getNumPartInCUSize() >> (depth + chromaShift);
+
+    for (uint32_t idx = 0; idx < numUnits; idx++)
+    {
+        uint32_t unitOffset = idx << LOG2_UNIT_SIZE;
+        uint32_t bsAbsIdx = calcBsIdx(cu, absZOrderIdx, dir, edge, idx << chromaShift);
+        uint32_t bs = blockingStrength[bsAbsIdx];
+
+        if (bs > 1)
+        {
+            int32_t qpQ = cu->getQP(bsAbsIdx);
+            partQ = bsAbsIdx;
+
+            // Derive neighboring PU index
+            if (dir == EDGE_VER)
+                cuP = cuQ->getPULeft(partP, partQ);
+            else // (dir == EDGE_HOR)
+                cuP = cuQ->getPUAbove(partP, partQ);
+
+            int32_t qpP = cuP->getQP(partP);
+
+            if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+            {
+                // check if each of PUs is lossless coded
+                partPNoFilter = cuP->getCUTransquantBypass(partP);
+                partQNoFilter = cuQ->getCUTransquantBypass(partQ);
+            }
+
+            for (uint32_t chromaIdx = 0; chromaIdx < 2; chromaIdx++)
+            {
+                int32_t chromaQPOffset  = !chromaIdx ? cu->m_slice->m_pps->chromaCbQpOffset : cu->m_slice->m_pps->chromaCrQpOffset;
+                int32_t qp = ((qpP + qpQ + 1) >> 1) + chromaQPOffset;
+                if (qp >= 30)
+                {
+                    if (chFmt == X265_CSP_I420)
+                        qp = g_chromaScale[qp];
+                    else
+                        qp = X265_MIN(qp, 51);
+                }
+
+                int32_t indexTC = Clip3(0, QP_MAX_SPEC + DEFAULT_INTRA_TC_OFFSET, int32_t(qp + DEFAULT_INTRA_TC_OFFSET + tcOffset));
+                const int32_t bitdepthShift = X265_DEPTH - 8;
+                int32_t tc = s_tcTable[indexTC] << bitdepthShift;
+                pixel* srcC = srcChroma[chromaIdx];
+
+                pelFilterChroma(srcC + srcStep * unitOffset, srcStep, offset, tc, partPNoFilter, partQNoFilter);
+            }
+        }
+    }
+}
+
+const uint8_t Deblock::s_tcTable[54] =
+{
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2,
+    2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 5, 5, 6, 6, 7, 8, 9, 10, 11, 13, 14, 16, 18, 20, 22, 24
+};
+
+const uint8_t Deblock::s_betaTable[52] =
+{
+    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
+    18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42, 44, 46, 48, 50, 52, 54, 56, 58, 60, 62, 64
+};
+

x265-hg20140928.tar.bz2/source/common/deblock.h Added

@@ -0,0 +1,76 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Author: Gopu Govindaswamy <gopu@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_DEBLOCK_H
+#define X265_DEBLOCK_H
+
+#include "common.h"
+#include "TLibCommon/TComRom.h"
+
+namespace x265 {
+// private namespace
+
+class TComDataCU;
+
+class Deblock
+{
+public:
+    enum { EDGE_VER, EDGE_HOR };
+
+    uint32_t m_numPartitions;
+
+    Deblock() : m_numPartitions(0) {}
+
+    void init() { m_numPartitions = 1 << (g_maxFullDepth * 2); }
+
+    void deblockCTU(TComDataCU* cu, int32_t dir);
+
+protected:
+
+    // CU-level deblocking function
+    void deblockCU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, const int32_t Edge, uint8_t blockingStrength[]);
+
+    struct Param
+    {
+        uint8_t leftEdge;
+        uint8_t topEdge;
+    };
+
+    // set filtering functions
+    void setLoopfilterParam(TComDataCU* cu, uint32_t absZOrderIdx, Param *params);
+    void setEdgefilterTU(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, uint8_t blockingStrength[]);
+    void setEdgefilterPU(TComDataCU* cu, uint32_t absZOrderIdx, int32_t dir, uint8_t blockingStrength[], uint32_t widthInBaseUnits);
+    void setEdgefilterMultiple(TComDataCU* cu, uint32_t absZOrderIdx, int32_t dir, int32_t edgeIdx, uint8_t value, uint8_t blockingStrength[], uint32_t widthInBaseUnits);
+
+    // get filtering functions
+    void getBoundaryStrengthSingle(TComDataCU* cu, int32_t dir, uint32_t partIdx, uint8_t blockingStrength[]);
+
+    // filter luma/chroma functions
+    void edgeFilterLuma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, int32_t edge, const uint8_t blockingStrength[]);
+    void edgeFilterChroma(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth, int32_t dir, int32_t edge, const uint8_t blockingStrength[]);
+
+    static const uint8_t s_tcTable[54];
+    static const uint8_t s_betaTable[52];
+};
+}
+#endif // ifndef X265_DEBLOCK_H

x265-hg20140928.tar.bz2/source/common/frame.cpp Added

@@ -0,0 +1,176 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Author: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "common.h"
+#include "frame.h"
+#include "mv.h"
+
+using namespace x265;
+
+Frame::Frame()
+    : m_origPicYuv(NULL)
+    , m_reconPicYuv(NULL)
+    , m_rowDiagQp(NULL)
+    , m_rowDiagQScale(NULL)
+    , m_rowDiagSatd(NULL)
+    , m_rowDiagIntraSatd(NULL)
+    , m_rowEncodedBits(NULL)
+    , m_numEncodedCusPerRow(NULL)
+    , m_rowSatdForVbv(NULL)
+    , m_cuCostsForVbv(NULL)
+    , m_intraCuCostsForVbv(NULL)
+{
+    m_picSym = NULL;
+    m_reconRowCount.set(0);
+    m_countRefEncoders = 0;
+    memset(&m_lowres, 0, sizeof(m_lowres));
+    m_next = NULL;
+    m_prev = NULL;
+    m_qpaAq = NULL;
+    m_qpaRc = NULL;
+    m_avgQpRc = 0;
+    m_avgQpAq = 0;
+    m_bChromaPlanesExtended = false;
+    m_intraData = NULL;
+    m_interData = NULL;
+}
+
+bool Frame::create(x265_param *param, Window& display, Window& conformance)
+{
+    m_conformanceWindow = conformance;
+    m_defaultDisplayWindow = display;
+
+    m_origPicYuv = new TComPicYuv;
+
+    bool ok = true;
+    ok &= m_origPicYuv->create(param->sourceWidth, param->sourceHeight, param->internalCsp, g_maxCUSize, g_maxFullDepth);
+    ok &= m_lowres.create(m_origPicYuv, param->bframes, !!param->rc.aqMode);
+
+    bool isVbv = param->rc.vbvBufferSize > 0 && param->rc.vbvMaxBitrate > 0;
+    if (ok && (isVbv || param->rc.aqMode))
+    {
+        int numCols = (param->sourceWidth + g_maxCUSize - 1) >> g_maxLog2CUSize;
+        int numRows = (param->sourceHeight + g_maxCUSize - 1) >> g_maxLog2CUSize;
+
+        if (param->rc.aqMode)
+            CHECKED_MALLOC(m_qpaAq, double, numRows);
+        if (isVbv)
+        {
+            CHECKED_MALLOC(m_rowDiagQp, double, numRows);
+            CHECKED_MALLOC(m_rowDiagQScale, double, numRows);
+            CHECKED_MALLOC(m_rowDiagSatd, uint32_t, numRows);
+            CHECKED_MALLOC(m_rowDiagIntraSatd, uint32_t, numRows);
+            CHECKED_MALLOC(m_rowEncodedBits, uint32_t, numRows);
+            CHECKED_MALLOC(m_numEncodedCusPerRow, uint32_t, numRows);
+            CHECKED_MALLOC(m_rowSatdForVbv, uint32_t, numRows);
+            CHECKED_MALLOC(m_cuCostsForVbv, uint32_t, numRows * numCols);
+            CHECKED_MALLOC(m_intraCuCostsForVbv, uint32_t, numRows * numCols);
+            CHECKED_MALLOC(m_qpaRc, double, numRows);
+        }
+        reinit(param);
+    }
+
+    return ok;
+
+fail:
+    ok = false;
+    return ok;
+}
+
+bool Frame::allocPicSym(x265_param *param)
+{
+    m_picSym = new TComPicSym;
+    m_reconPicYuv = new TComPicYuv;
+    m_picSym->m_reconPicYuv = m_reconPicYuv;
+    bool ok = m_picSym->create(param) &&
+            m_reconPicYuv->create(param->sourceWidth, param->sourceHeight, param->internalCsp, g_maxCUSize, g_maxFullDepth);
+    if (ok)
+    {
+        // initialize m_reconpicYuv as SAO may read beyond the end of the picture accessing uninitialized pixels
+        int maxHeight = m_reconPicYuv->m_numCuInHeight * g_maxCUSize;
+        memset(m_reconPicYuv->m_picOrg[0], 0, m_reconPicYuv->m_stride * maxHeight);
+        memset(m_reconPicYuv->m_picOrg[1], 0, m_reconPicYuv->m_strideC * (maxHeight >> m_reconPicYuv->m_vChromaShift));
+        memset(m_reconPicYuv->m_picOrg[2], 0, m_reconPicYuv->m_strideC * (maxHeight >> m_reconPicYuv->m_vChromaShift));
+    }
+    return ok;
+}
+
+void Frame::reinit(x265_param *param)
+{
+    int numCols = (param->sourceWidth + g_maxCUSize - 1) >> g_maxLog2CUSize;
+    int numRows = (param->sourceHeight + g_maxCUSize - 1) >> g_maxLog2CUSize;
+    if (param->rc.vbvBufferSize > 0 && param->rc.vbvMaxBitrate > 0)
+    {
+        memset(m_rowDiagQp, 0, numRows * sizeof(double));
+        memset(m_rowDiagQScale, 0, numRows * sizeof(double));
+        memset(m_rowDiagSatd, 0, numRows * sizeof(uint32_t));
+        memset(m_rowDiagIntraSatd, 0, numRows * sizeof(uint32_t));
+        memset(m_rowEncodedBits, 0, numRows * sizeof(uint32_t));
+        memset(m_numEncodedCusPerRow, 0, numRows * sizeof(uint32_t));
+        memset(m_rowSatdForVbv, 0, numRows * sizeof(uint32_t));
+        memset(m_cuCostsForVbv, 0,  numRows * numCols * sizeof(uint32_t));
+        memset(m_intraCuCostsForVbv, 0, numRows * numCols * sizeof(uint32_t));
+        memset(m_qpaRc, 0, numRows * sizeof(double));
+    }
+    if (param->rc.aqMode)
+        memset(m_qpaAq, 0, numRows * sizeof(double));
+}
+
+void Frame::destroy()
+{
+    if (m_picSym)
+    {
+        m_picSym->destroy();
+        delete m_picSym;
+        m_picSym = NULL;
+    }
+
+    if (m_origPicYuv)
+    {
+        m_origPicYuv->destroy();
+        delete m_origPicYuv;
+        m_origPicYuv = NULL;
+    }
+
+    if (m_reconPicYuv)
+    {
+        m_reconPicYuv->destroy();
+        delete m_reconPicYuv;
+        m_reconPicYuv = NULL;
+    }
+    m_lowres.destroy();
+
+    X265_FREE(m_rowDiagQp);
+    X265_FREE(m_rowDiagQScale);
+    X265_FREE(m_rowDiagSatd);
+    X265_FREE(m_rowDiagIntraSatd);
+    X265_FREE(m_rowEncodedBits);
+    X265_FREE(m_numEncodedCusPerRow);
+    X265_FREE(m_rowSatdForVbv);
+    X265_FREE(m_cuCostsForVbv);
+    X265_FREE(m_intraCuCostsForVbv);
+    X265_FREE(m_qpaAq);
+    X265_FREE(m_qpaRc);
+}
+
+//! \}

x265-hg20140928.tar.bz2/source/common/frame.h Added

@@ -0,0 +1,127 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Author: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_FRAME_H
+#define X265_FRAME_H
+#define NUM_CU_PARTITIONS (1U << (g_maxFullDepth << 1))
+
+#include "common.h"
+#include "TLibCommon/TComPicSym.h"
+#include "TLibCommon/TComPicYuv.h"
+#include "lowres.h"
+#include "threading.h"
+#include "md5.h"
+#include "sei.h"
+
+namespace x265 {
+// private namespace
+
+class Encoder;
+
+class Frame
+{
+public:
+
+    TComPicYuv*       m_origPicYuv;
+
+    Window            m_conformanceWindow;
+    Window            m_defaultDisplayWindow;
+
+    TComPicSym*       m_picSym;
+    TComPicYuv*       m_reconPicYuv;
+    int               m_POC;
+
+    //** Frame Parallelism - notification between FrameEncoders of available motion reference rows **
+    ThreadSafeInteger m_reconRowCount;      // count of CTU rows completely reconstructed and extended for motion reference
+    volatile uint32_t m_countRefEncoders;   // count of FrameEncoder threads monitoring m_reconRowCount
+    void*             m_userData;           // user provided pointer passed in with this picture
+
+    int64_t           m_pts;                // user provided presentation time stamp
+    int64_t           m_reorderedPts;
+    int64_t           m_dts;
+
+    Lowres            m_lowres;
+
+    Frame*            m_next;               // PicList doubly linked list pointers
+    Frame*            m_prev;
+
+    bool              m_bChromaPlanesExtended; // orig chroma planes motion extended for weightp analysis
+
+    /* TODO: much of this data can be moved to RCE */
+    double*           m_rowDiagQp;
+    double*           m_rowDiagQScale;
+    uint32_t*         m_rowDiagSatd;
+    uint32_t*         m_rowDiagIntraSatd;
+    uint32_t*         m_rowEncodedBits;
+    uint32_t*         m_numEncodedCusPerRow;
+    uint32_t*         m_rowSatdForVbv;
+    uint32_t*         m_cuCostsForVbv;
+    uint32_t*         m_intraCuCostsForVbv;
+    double*           m_qpaAq;
+    double*           m_qpaRc;
+    double            m_avgQpRc;    // avg QP as decided by ratecontrol
+    double            m_avgQpAq;    // avg QP as decided by AQ in addition to ratecontrol
+    double            m_rateFactor; // calculated based on the Frame QP
+    int32_t           m_forceqp;    // Force to use the qp specified in qp file
+
+    x265_intra_data*  m_intraData;  // intra analysis information
+    x265_inter_data*  m_interData;  // inter analysis information
+
+    Frame();
+    ~Frame() {}
+
+    bool        create(x265_param *param, Window& display, Window& conformance);
+    bool        allocPicSym(x265_param *param);
+    void        reinit(x265_param *param);
+    void        destroy();
+
+    int         getPOC()                   { return m_POC; }
+
+    Window&     getConformanceWindow()     { return m_conformanceWindow; }
+
+    Window&     getDefDisplayWindow()      { return m_defaultDisplayWindow; }
+
+    TComPicYuv* getPicYuvOrg()             { return m_origPicYuv; }
+
+    TComPicYuv* getPicYuvRec()             { return m_reconPicYuv; }
+
+    int         getStride()                { return m_origPicYuv->getStride(); }
+
+    int         getCStride()               { return m_origPicYuv->getCStride(); }
+
+    /* Reflector methods for data stored in m_picSym */
+    TComPicSym* getPicSym()                { return m_picSym; }
+
+    TComDataCU* getCU(uint32_t cuAddr)     { return m_picSym->getCU(cuAddr); }
+
+    uint32_t    getNumCUsInFrame() const   { return m_picSym->getNumberOfCUsInFrame(); }
+
+    uint32_t    getNumPartInCUSize() const { return m_picSym->getNumPartInCUSize(); }
+
+    uint32_t    getFrameWidthInCU() const  { return m_picSym->getFrameWidthInCU(); }
+
+    uint32_t    getFrameHeightInCU() const { return m_picSym->getFrameHeightInCU(); }
+};
+}
+
+#endif // ifndef X265_FRAME_H

x265-hg20140324.tar.bz2/source/common/intrapred.cpp -> x265-hg20140928.tar.bz2/source/common/intrapred.cpp Changed

@@ -18,25 +18,12 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComPrediction.h"
 #include "TLibCommon/TComRom.h"
 #include "primitives.h"
 
-namespace x265 {
-unsigned char IntraFilterType[][35] =
-{
-    //  Index:    0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34
-    /*  4x4  */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
-    /*  8x8  */ { 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 },
-    /* 16x16 */ { 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1 },
-    /* 32x32 */ { 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1 },
-    /* 64x64 */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
-};
-}
-
 using namespace x265;
 
 namespace {
@@ -99,8 +86,8 @@
     }
 }
 
-template<int width>
-void planad_pred_c(pixel* dst, intptr_t dstStride, pixel* left, pixel* above, int /*dirMode*/, int /*bFilter*/)
+template<int log2Size>
+void planar_pred_c(pixel* dst, intptr_t dstStride, pixel* left, pixel* above, int /*dirMode*/, int /*bFilter*/)
 {
     above += 1;
     left  += 1;
@@ -110,10 +97,10 @@
     int32_t leftColumn[MAX_CU_SIZE + 1], topRow[MAX_CU_SIZE + 1];
     // CHECK_ME: dynamic range is 9 bits or 15 bits(I assume max input bit_depth is 14 bits)
     int16_t bottomRow[MAX_CU_SIZE], rightColumn[MAX_CU_SIZE];
-    int blkSize = width;
-    int offset2D = width;
-    int shift1D = g_convertToBit[width] + 2;
-    int shift2D = shift1D + 1;
+    const int blkSize = 1 << log2Size;
+    const int offset2D = blkSize;
+    const int shift1D = log2Size;
+    const int shift2D = shift1D + 1;
 
     // Get left and above reference column and row
     for (k = 0; k < blkSize + 1; k++)
@@ -257,14 +244,15 @@
     }
 }
 
-template<int size>
-void all_angs_pred_c(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma)
+template<int log2Size>
+void all_angs_pred_c(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma)
 {
+    const int size = 1 << log2Size;
     for (int mode = 2; mode <= 34; mode++)
     {
-        pixel *left = (IntraFilterType[(int)g_convertToBit[size]][mode] ? left1 : left0);
-        pixel *above = (IntraFilterType[(int)g_convertToBit[size]][mode] ? above1 : above0);
-        pixel *out = dest + (mode - 2) * (size * size);
+        pixel *left  = (g_intraFilterFlags[mode] & size ? left1  : left0);
+        pixel *above = (g_intraFilterFlags[mode] & size ? above1 : above0);
+        pixel *out = dest + ((mode - 2) << (log2Size * 2));
 
         intra_pred_ang_c<size>(out, size, left, above, mode, bLuma);
 
@@ -293,27 +281,27 @@
 
 void Setup_C_IPredPrimitives(EncoderPrimitives& p)
 {
-    p.intra_pred[BLOCK_4x4][0] = planad_pred_c<4>;
-    p.intra_pred[BLOCK_8x8][0] = planad_pred_c<8>;
-    p.intra_pred[BLOCK_16x16][0] = planad_pred_c<16>;
-    p.intra_pred[BLOCK_32x32][0] = planad_pred_c<32>;
+    p.intra_pred[0][BLOCK_4x4] = planar_pred_c<2>;
+    p.intra_pred[0][BLOCK_8x8] = planar_pred_c<3>;
+    p.intra_pred[0][BLOCK_16x16] = planar_pred_c<4>;
+    p.intra_pred[0][BLOCK_32x32] = planar_pred_c<5>;
 
     // Intra Prediction DC
-    p.intra_pred[BLOCK_4x4][1] = intra_pred_dc_c<4>;
-    p.intra_pred[BLOCK_8x8][1] = intra_pred_dc_c<8>;
-    p.intra_pred[BLOCK_16x16][1] = intra_pred_dc_c<16>;
-    p.intra_pred[BLOCK_32x32][1] = intra_pred_dc_c<32>;
+    p.intra_pred[1][BLOCK_4x4] = intra_pred_dc_c<4>;
+    p.intra_pred[1][BLOCK_8x8] = intra_pred_dc_c<8>;
+    p.intra_pred[1][BLOCK_16x16] = intra_pred_dc_c<16>;
+    p.intra_pred[1][BLOCK_32x32] = intra_pred_dc_c<32>;
     for (int i = 2; i < NUM_INTRA_MODE; i++)
     {
-        p.intra_pred[BLOCK_4x4][i] = intra_pred_ang_c<4>;
-        p.intra_pred[BLOCK_8x8][i] = intra_pred_ang_c<8>;
-        p.intra_pred[BLOCK_16x16][i] = intra_pred_ang_c<16>;
-        p.intra_pred[BLOCK_32x32][i] = intra_pred_ang_c<32>;
+        p.intra_pred[i][BLOCK_4x4] = intra_pred_ang_c<4>;
+        p.intra_pred[i][BLOCK_8x8] = intra_pred_ang_c<8>;
+        p.intra_pred[i][BLOCK_16x16] = intra_pred_ang_c<16>;
+        p.intra_pred[i][BLOCK_32x32] = intra_pred_ang_c<32>;
     }
 
-    p.intra_pred_allangs[BLOCK_4x4] = all_angs_pred_c<4>;
-    p.intra_pred_allangs[BLOCK_8x8] = all_angs_pred_c<8>;
-    p.intra_pred_allangs[BLOCK_16x16] = all_angs_pred_c<16>;
-    p.intra_pred_allangs[BLOCK_32x32] = all_angs_pred_c<32>;
+    p.intra_pred_allangs[BLOCK_4x4] = all_angs_pred_c<2>;
+    p.intra_pred_allangs[BLOCK_8x8] = all_angs_pred_c<3>;
+    p.intra_pred_allangs[BLOCK_16x16] = all_angs_pred_c<4>;
+    p.intra_pred_allangs[BLOCK_32x32] = all_angs_pred_c<5>;
 }
 }

x265-hg20140324.tar.bz2/source/common/ipfilter.cpp -> x265-hg20140928.tar.bz2/source/common/ipfilter.cpp Changed

x265-hg20140324.tar.bz2/source/common/loopfilter.cpp -> x265-hg20140928.tar.bz2/source/common/loopfilter.cpp Changed

@@ -3,6 +3,7 @@
 *
 * Authors: Praveen Kumar Tiwari <praveen@multicorewareinc.com>
 *          Dnyaneshwar Gorade <dnyaneshwar@multicorewareinc.com>
+*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
@@ -18,7 +19,7 @@
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 *
 * This program is also available under a commercial proprietary license.
-* For more information, contact us at licensing@multicorewareinc.com.
+* For more information, contact us at license @ x265.com.
 *****************************************************************************/
 
 #include "TLibCommon/TypeDef.h"
@@ -35,7 +36,7 @@
 
     for (x = 0; x < lcuWidth; x++)
     {
-        signRight = ((rec[x] - rec[x+1]) < 0) ? -1 : ((rec[x] - rec[x+1]) > 0) ? 1 : 0;
+        signRight = ((rec[x] - rec[x + 1]) < 0) ? -1 : ((rec[x] - rec[x + 1]) > 0) ? 1 : 0;
         edgeType = signRight + signLeft + 2;
         signLeft  = -signRight;

x265-hg20140324.tar.bz2/source/common/lowres.cpp -> x265-hg20140928.tar.bz2/source/common/lowres.cpp Changed

@@ -18,10 +18,10 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComPic.h"
+#include "TLibCommon/TComPicYuv.h"
 #include "lowres.h"
 #include "mv.h"
 
@@ -51,7 +51,7 @@
     {
         CHECKED_MALLOC(qpAqOffset, double, cuCount);
         CHECKED_MALLOC(invQscaleFactor, int, cuCount);
-        CHECKED_MALLOC(qpOffset, double, cuCount);
+        CHECKED_MALLOC(qpCuTreeOffset, double, cuCount);
     }
     CHECKED_MALLOC(propagateCost, uint16_t, cuCount);
 
@@ -121,7 +121,7 @@
 
     X265_FREE(qpAqOffset);
     X265_FREE(invQscaleFactor);
-    X265_FREE(qpOffset);
+    X265_FREE(qpCuTreeOffset);
     X265_FREE(propagateCost);
 }
 
@@ -135,6 +135,7 @@
     sliceType = type;
     frameNum = poc;
     leadingBframes = 0;
+    indB = 0;
     satdCost = (int64_t)-1;
     memset(costEst, -1, sizeof(costEst));
     memset(weightedCostDelta, 0, sizeof(weightedCostDelta));

x265-hg20140324.tar.bz2/source/common/lowres.h -> x265-hg20140928.tar.bz2/source/common/lowres.h Changed

x265-hg20140324.tar.bz2/source/common/mv.h -> x265-hg20140928.tar.bz2/source/common/mv.h Changed

x265-hg20140324.tar.bz2/source/common/param.cpp -> x265-hg20140928.tar.bz2/source/common/param.cpp Changed

@@ -18,27 +18,30 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComSlice.h"
-#include "threading.h"
 #include "common.h"
+#include "slice.h"
+#include "threading.h"
 #include "param.h"
 #include "cpu.h"
+#include "TLibCommon/TComRom.h"
 #include "x265.h"
 
 #if _MSC_VER
 #pragma warning(disable: 4996) // POSIX functions are just fine, thanks
 #pragma warning(disable: 4706) // assignment within conditional
+#pragma warning(disable: 4127) // conditional expression is constant
 #endif
 
 #if _WIN32
-#define strcasecmp _stricmp 
+#define strcasecmp _stricmp
 #endif
 
-#if !HAVE_STROTOK_R
-/* 
+#if !defined(HAVE_STRTOK_R)
+
+/*
  * adapted from public domain strtok_r() by Charlie Gordon
  *
  *   from comp.lang.c  9/14/2007
@@ -49,10 +52,10 @@
  *      http://groups.google.com/group/comp.lang.c/msg/7c7b39328fefab9c
  */
 
-char* strtok_r(
-    char *str, 
-    const char *delim, 
-    char **nextp)
+#undef strtok_r
+char* strtok_r(char *      str,
+               const char *delim,
+               char **     nextp)
 {
     if (!str)
         str = *nextp;
@@ -73,7 +76,8 @@
 
     return ret;
 }
-#endif
+
+#endif // if !defined(HAVE_STRTOK_R)
 
 using namespace x265;
 
@@ -94,18 +98,34 @@
 {
     memset(param, 0, sizeof(x265_param));
 
-    /* Applying non-zero default values to all elements in the param structure */
+    /* Applying default values to all elements in the param structure */
     param->cpuid = x265::cpu_detect();
-    param->logLevel = X265_LOG_INFO;
     param->bEnableWavefront = 1;
-    param->frameNumThreads = 0;
     param->poolNumThreads = 0;
+    param->frameNumThreads = 0;
+
+    param->logLevel = X265_LOG_INFO;
     param->csvfn = NULL;
+    param->rc.lambdaFileName = NULL;
+    param->bLogCuStats = 0;
+    param->decodedPictureHashSEI = 0;
+
+    /* Quality Measurement Metrics */
+    param->bEnablePsnr = 0;
+    param->bEnableSsim = 0;
 
     /* Source specifications */
     param->internalBitDepth = x265_max_bit_depth;
     param->internalCsp = X265_CSP_I420;
 
+    param->levelIdc = 0;
+    param->bHighTier = 0;
+    param->interlaceMode = 0;
+    param->bRepeatHeaders = 0;
+    param->bEnableAccessUnitDelimiters = 0;
+    param->bEmitHRDSEI = 0;
+    param->bEmitInfoSEI = 1;
+
     /* CU definitions */
     param->maxCUSize = 64;
     param->tuQTMaxInterDepth = 1;
@@ -124,6 +144,7 @@
     /* Intra Coding Tools */
     param->bEnableConstrainedIntra = 0;
     param->bEnableStrongIntraSmoothing = 1;
+    param->bEnableFastIntra = 0;
 
     /* Inter Coding tools */
     param->searchMethod = X265_HEX_SEARCH;
@@ -134,8 +155,8 @@
     param->bEnableWeightedBiPred = 0;
     param->bEnableEarlySkip = 0;
     param->bEnableCbfFastMode = 0;
-    param->bEnableAMP = 1;
-    param->bEnableRectInter = 1;
+    param->bEnableAMP = 0;
+    param->bEnableRectInter = 0;
     param->rdLevel = 3;
     param->bEnableSignHiding = 1;
     param->bEnableTransformSkip = 0;
@@ -154,6 +175,11 @@
     param->cbQpOffset = 0;
     param->crQpOffset = 0;
     param->rdPenalty = 0;
+    param->psyRd = 0.0;
+    param->psyRdoq = 0.0;
+    param->bIntraInBFrames = 1;
+    param->bLossless = 0;
+    param->bCULossless = 0;
 
     /* Rate control options */
     param->rc.vbvMaxBitrate = 0;
@@ -168,17 +194,19 @@
     param->rc.qpStep = 4;
     param->rc.rateControlMode = X265_RC_CRF;
     param->rc.qp = 32;
-    param->rc.aqMode = X265_AQ_VARIANCE;
+    param->rc.aqMode = X265_AQ_AUTO_VARIANCE;
     param->rc.aqStrength = 1.0;
     param->rc.cuTree = 1;
-
-    /* Quality Measurement Metrics */
-    param->bEnablePsnr = 0;
-    param->bEnableSsim = 0;
+    param->rc.rfConstantMax = 0;
+    param->rc.rfConstantMin = 0;
+    param->rc.bStatRead = 0;
+    param->rc.bStatWrite = 0;
+    param->rc.statFileName = NULL;
+    param->rc.complexityBlur = 20;
+    param->rc.qblur = 0.5;
+    param->rc.bEnableSlowFirstPass = 0;
 
     /* Video Usability Information (VUI) */
-    param->vui.bEnableVuiParametersPresentFlag = 0;
-    param->vui.bEnableAspectRatioIdc = 0;
     param->vui.aspectRatioIdc = 0;
     param->vui.sarWidth = 0;
     param->vui.sarHeight = 0;
@@ -198,41 +226,6 @@
     param->vui.defDispWinRightOffset = 0;
     param->vui.defDispWinTopOffset = 0;
     param->vui.defDispWinBottomOffset = 0;
-    param->vui.bEnableVuiTimingInfoPresentFlag = 0;
-    param->vui.bEnableVuiHrdParametersPresentFlag = 0;
-    param->vui.bEnableNalHrdParametersPresentFlag = 0;
-    param->vui.bEnableBitstreamRestrictionFlag = 0;
-    param->vui.bEnableSubPicHrdParamsPresentFlag = 0;
-}
-
-extern "C"
-int x265_param_apply_profile(x265_param *param, const char *profile)
-{
-    if (!profile)
-        return 0;
-    if (!strcmp(profile, "main"))
-    {}
-    else if (!strcmp(profile, "main10"))
-    {
-#if HIGH_BIT_DEPTH
-        param->internalBitDepth = 10;
-#else
-        x265_log(param, X265_LOG_WARNING, "Main10 not supported, not compiled for 16bpp.\n");
-        return -1;
-#endif
-    }
-    else if (!strcmp(profile, "mainstillpicture"))
-    {
-        param->keyframeMax = 1;
-        param->bOpenGOP = 0;
-    }
-    else
-    {
-        x265_log(param, X265_LOG_ERROR, "unknown profile <%s>\n", profile);
-        return -1;
-    }
-
-    return 0;
 }
 
 extern "C"
@@ -256,17 +249,18 @@
             param->bFrameAdaptive = 0;
             param->subpelRefine = 0;
             param->searchMethod = X265_DIA_SEARCH;
-            param->bEnableRectInter = 0;
-            param->bEnableAMP = 0;
             param->bEnableEarlySkip = 1;
             param->bEnableCbfFastMode = 1;
             param->bEnableSAO = 0;
             param->bEnableSignHiding = 0;
             param->bEnableWeightedPred = 0;
+            param->rdLevel = 2;
             param->maxNumReferences = 1;
+            param->bEnableLoopFilter = 0;
             param->rc.aqStrength = 0.0;
             param->rc.aqMode = X265_AQ_NONE;
             param->rc.cuTree = 0;
+            param->bEnableFastIntra = 1;
         }
         else if (!strcmp(preset, "superfast"))
         {
@@ -275,15 +269,16 @@
             param->searchRange = 44;
             param->bFrameAdaptive = 0;
             param->subpelRefine = 1;
-            param->bEnableRectInter = 0;
-            param->bEnableAMP = 0;
             param->bEnableEarlySkip = 1;
             param->bEnableCbfFastMode = 1;
             param->bEnableWeightedPred = 0;
+            param->rdLevel = 2;
             param->maxNumReferences = 1;
             param->rc.aqStrength = 0.0;
             param->rc.aqMode = X265_AQ_NONE;
             param->rc.cuTree = 0;
+            param->bEnableSAO = 0;
+            param->bEnableFastIntra = 1;
         }
         else if (!strcmp(preset, "veryfast"))
         {
@@ -291,29 +286,31 @@
             param->maxCUSize = 32;
             param->bFrameAdaptive = 0;
             param->subpelRefine = 1;
-            param->bEnableRectInter = 0;
-            param->bEnableAMP = 0;
             param->bEnableEarlySkip = 1;
             param->bEnableCbfFastMode = 1;
+            param->rdLevel = 2;
             param->maxNumReferences = 1;
             param->rc.cuTree = 0;
+            param->bEnableFastIntra = 1;
         }
         else if (!strcmp(preset, "faster"))
         {
             param->lookaheadDepth = 15;
             param->bFrameAdaptive = 0;
-            param->bEnableRectInter = 0;
-            param->bEnableAMP = 0;
             param->bEnableEarlySkip = 1;
             param->bEnableCbfFastMode = 1;
+            param->rdLevel = 2;
             param->maxNumReferences = 1;
             param->rc.cuTree = 0;
+            param->bEnableFastIntra = 1;
         }
         else if (!strcmp(preset, "fast"))
         {
             param->lookaheadDepth = 15;
-            param->bEnableRectInter = 0;
-            param->bEnableAMP = 0;
+            param->bFrameAdaptive = 0;
+            param->rdLevel = 2;
+            param->maxNumReferences = 2;
+            param->bEnableFastIntra = 1;
         }
         else if (!strcmp(preset, "medium"))
         {
@@ -321,6 +318,7 @@
         }
         else if (!strcmp(preset, "slow"))
         {
+            param->bEnableRectInter = 1;
             param->lookaheadDepth = 25;
             param->rdLevel = 4;
             param->subpelRefine = 3;
@@ -329,6 +327,9 @@
         }
         else if (!strcmp(preset, "slower"))
         {
+            param->bEnableWeightedBiPred = 1;
+            param->bEnableAMP = 1;
+            param->bEnableRectInter = 1;
             param->lookaheadDepth = 30;
             param->bframes = 8;
             param->tuQTMaxInterDepth = 2;
@@ -340,6 +341,9 @@
         }
         else if (!strcmp(preset, "veryslow"))
         {
+            param->bEnableWeightedBiPred = 1;
+            param->bEnableAMP = 1;
+            param->bEnableRectInter = 1;
             param->lookaheadDepth = 40;
             param->bframes = 8;
             param->tuQTMaxInterDepth = 3;
@@ -352,6 +356,9 @@
         }
         else if (!strcmp(preset, "placebo"))
         {
+            param->bEnableWeightedBiPred = 1;
+            param->bEnableAMP = 1;
+            param->bEnableRectInter = 1;
             param->lookaheadDepth = 60;
             param->searchRange = 92;
             param->bframes = 8;
@@ -363,6 +370,7 @@
             param->searchMethod = X265_STAR_SEARCH;
             param->bEnableTransformSkip = 1;
             param->maxNumReferences = 5;
+            param->rc.bEnableSlowFirstPass = 1;
             // TODO: optimized esa
         }
         else
@@ -373,10 +381,14 @@
         if (!strcmp(tune, "psnr"))
         {
             param->rc.aqStrength = 0.0;
+            param->psyRd = 0.0;
+            param->psyRdoq = 0.0;
         }
         else if (!strcmp(tune, "ssim"))
         {
             param->rc.aqMode = X265_AQ_AUTO_VARIANCE;
+            param->psyRd = 0.0;
+            param->psyRdoq = 0.0;
         }
         else if (!strcmp(tune, "fastdecode") ||
                  !strcmp(tune, "fast-decode"))
@@ -385,6 +397,7 @@
             param->bEnableSAO = 0;
             param->bEnableWeightedPred = 0;
             param->bEnableWeightedBiPred = 0;
+            param->bIntraInBFrames = 0;
         }
         else if (!strcmp(tune, "zerolatency") ||
                  !strcmp(tune, "zero-latency"))
@@ -457,6 +470,10 @@
     if (!name)
         return X265_PARAM_BAD_NAME;
 
+    // skip -- prefix if provided
+    if (name[0] == '-' && name[1] == '-')
+        name += 2;
+
     // s/_/-/g
     if (strlen(name) + 1 < sizeof(nameBuf) && strchr(name, '_'))
     {
@@ -518,8 +535,19 @@
         }
     }
     OPT("csv") p->csvfn = value;
+    OPT("scaling-list") p->scalingLists = value;
+    OPT("lambda-file") p->rc.lambdaFileName = value;
     OPT("threads") p->poolNumThreads = atoi(value);
     OPT("frame-threads") p->frameNumThreads = atoi(value);
+    OPT2("level-idc", "level")
+    {
+        /* allow "5.1" or "51", both converted to integer 51 */
+        if (atof(value) < 7)
+            p->levelIdc = (int)(10 * atof(value) + .5);
+        else
+            p->levelIdc = atoi(value);
+    }
+    OPT("high-tier") p->bHighTier = atobool(value);
     OPT2("log-level", "log")
     {
         p->logLevel = atoi(value);
@@ -529,6 +557,8 @@
             p->logLevel = parseName(value, logLevelNames, bError) - 1;
         }
     }
+    OPT("cu-stats") p->bLogCuStats = atobool(value);
+    OPT("repeat-headers") p->bRepeatHeaders = atobool(value);
     OPT("wpp") p->bEnableWavefront = atobool(value);
     OPT("ctu") p->maxCUSize = (uint32_t)atoi(value);
     OPT("tu-intra-depth") p->tuQTMaxIntraDepth = (uint32_t)atoi(value);
@@ -545,7 +575,10 @@
     OPT("no-tskip-fast") p->bEnableTSkipFast = atobool(value);
     OPT("tskip-fast") p->bEnableTSkipFast = atobool(value);
     OPT("strong-intra-smoothing") p->bEnableStrongIntraSmoothing = atobool(value);
+    OPT("lossless") p->bLossless = atobool(value);
+    OPT("cu-lossless") p->bCULossless = atobool(value);
     OPT("constrained-intra") p->bEnableConstrainedIntra = atobool(value);
+    OPT("fast-intra") p->bEnableFastIntra = atobool(value);
     OPT("open-gop") p->bOpenGOP = atobool(value);
     OPT("scenecut")
     {
@@ -578,14 +611,17 @@
             bError = false;
             p->interlaceMode = parseName(value, x265_interlace_names, bError);
         }
-        p->vui.bEnableVuiParametersPresentFlag |= !!p->interlaceMode;
     }
     OPT("ref") p->maxNumReferences = atoi(value);
     OPT("weightp") p->bEnableWeightedPred = atobool(value);
+    OPT("weightb") p->bEnableWeightedBiPred = atobool(value);
     OPT("cbqpoffs") p->cbQpOffset = atoi(value);
     OPT("crqpoffs") p->crQpOffset = atoi(value);
     OPT("rd") p->rdLevel = atoi(value);
+    OPT("psy-rd") p->psyRd = atof(value);
+    OPT("psy-rdoq") p->psyRdoq = atof(value);
     OPT("signhide") p->bEnableSignHiding = atobool(value);
+    OPT("b-intra") p->bIntraInBFrames = atobool(value);
     OPT("lft") p->bEnableLoopFilter = atobool(value);
     OPT("sao") p->bEnableSAO = atobool(value);
     OPT("sao-lcu-bounds") p->saoLcuBoundary = atoi(value);
@@ -593,13 +629,19 @@
     OPT("ssim") p->bEnableSsim = atobool(value);
     OPT("psnr") p->bEnablePsnr = atobool(value);
     OPT("hash") p->decodedPictureHashSEI = atoi(value);
+    OPT("aud") p->bEnableAccessUnitDelimiters = atobool(value);
+    OPT("info") p->bEmitInfoSEI = atobool(value);
     OPT("b-pyramid") p->bBPyramid = atobool(value);
+    OPT("hrd") p->bEmitHRDSEI = atobool(value);
+    OPT2("ipratio", "ip-factor") p->rc.ipFactor = atof(value);
+    OPT2("pbratio", "pb-factor") p->rc.pbFactor = atof(value);
     OPT("aq-mode") p->rc.aqMode = atoi(value);
     OPT("aq-strength") p->rc.aqStrength = atof(value);
     OPT("vbv-maxrate") p->rc.vbvMaxBitrate = atoi(value);
     OPT("vbv-bufsize") p->rc.vbvBufferSize = atoi(value);
     OPT("vbv-init")    p->rc.vbvBufferInit = atof(value);
     OPT("crf-max")     p->rc.rfConstantMax = atof(value);
+    OPT("crf-min")     p->rc.rfConstantMin = atof(value);
     OPT("crf")
     {
         p->rc.rfConstant = atof(value);
@@ -615,29 +657,14 @@
         p->rc.qp = atoi(value);
         p->rc.rateControlMode = X265_RC_CQP;
     }
+    OPT("input-res") bError |= sscanf(value, "%dx%d", &p->sourceWidth, &p->sourceHeight) != 2;
     OPT("input-csp") p->internalCsp = parseName(value, x265_source_csp_names, bError);
     OPT("me")        p->searchMethod = parseName(value, x265_motion_est_names, bError);
     OPT("cutree")    p->rc.cuTree = atobool(value);
-    OPT("vui")
-    {
-        int bvalue = atobool(value);
-
-        p->vui.bEnableVuiParametersPresentFlag = bvalue;
-        p->vui.bEnableAspectRatioIdc = bvalue;
-        p->vui.bEnableOverscanInfoPresentFlag = bvalue;
-        p->vui.bEnableVideoSignalTypePresentFlag = bvalue;
-        p->vui.bEnableColorDescriptionPresentFlag = bvalue;
-        p->vui.bEnableChromaLocInfoPresentFlag = bvalue;
-        p->vui.bEnableDefaultDisplayWindowFlag = bvalue;
-        p->vui.bEnableVuiTimingInfoPresentFlag = bvalue;
-        p->vui.bEnableVuiHrdParametersPresentFlag = bvalue;
-        p->vui.bEnableBitstreamRestrictionFlag = bvalue;
-        p->vui.bEnableSubPicHrdParamsPresentFlag = bvalue;
-    }
+    OPT("slow-firstpass") p->rc.bEnableSlowFirstPass = atobool(value);
+    OPT("analysis-mode") p->analysisMode = parseName(value, x265_analysis_names, bError);
     OPT("sar")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableAspectRatioIdc = 1;
         p->vui.aspectRatioIdc = parseName(value, x265_sar_names, bError);
         if (bError)
         {
@@ -647,7 +674,6 @@
     }
     OPT("overscan")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         if (!strcmp(value, "show"))
             p->vui.bEnableOverscanInfoPresentFlag = 1;
         else if (!strcmp(value, "crop"))
@@ -662,47 +688,40 @@
     }
     OPT("videoformat")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableVideoSignalTypePresentFlag = 1;
         p->vui.videoFormat = parseName(value, x265_video_format_names, bError);
     }
     OPT("range")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableVideoSignalTypePresentFlag = 1; 
+        p->vui.bEnableVideoSignalTypePresentFlag = 1;
         p->vui.bEnableVideoFullRangeFlag = parseName(value, x265_fullrange_names, bError);
     }
     OPT("colorprim")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableVideoSignalTypePresentFlag = 1;
         p->vui.bEnableColorDescriptionPresentFlag = 1;
         p->vui.colorPrimaries = parseName(value, x265_colorprim_names, bError);
     }
     OPT("transfer")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableVideoSignalTypePresentFlag = 1;
         p->vui.bEnableColorDescriptionPresentFlag = 1;
         p->vui.transferCharacteristics = parseName(value, x265_transfer_names, bError);
     }
     OPT("colormatrix")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableVideoSignalTypePresentFlag = 1;
         p->vui.bEnableColorDescriptionPresentFlag = 1;
         p->vui.matrixCoeffs = parseName(value, x265_colmatrix_names, bError);
     }
     OPT("chromaloc")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableChromaLocInfoPresentFlag = 1;
         p->vui.chromaSampleLocTypeTopField = atoi(value);
         p->vui.chromaSampleLocTypeBottomField = p->vui.chromaSampleLocTypeTopField;
     }
     OPT("crop-rect")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
         p->vui.bEnableDefaultDisplayWindowFlag = 1;
         bError |= sscanf(value, "%d,%d,%d,%d",
                          &p->vui.defDispWinLeftOffset,
@@ -710,36 +729,14 @@
                          &p->vui.defDispWinRightOffset,
                          &p->vui.defDispWinBottomOffset) != 4;
     }
-    OPT("timinginfo")
-    {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableVuiTimingInfoPresentFlag = atobool(value);
-    }
-    OPT("hrd")
-    {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableVuiTimingInfoPresentFlag = 1;
-        p->vui.bEnableVuiHrdParametersPresentFlag = atobool(value);
-    }
-    OPT("nal-hrd")
-    {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableVuiTimingInfoPresentFlag = 1;
-        p->vui.bEnableVuiHrdParametersPresentFlag = 1;
-        p->vui.bEnableNalHrdParametersPresentFlag = parseName(value, x265_nal_hrd_names, bError);
-    }
-    OPT("bitstreamrestriction")
-    {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableBitstreamRestrictionFlag = atobool(value);
-    }
-    OPT("subpichrd")
+    OPT("nr") p->noiseReduction = atoi(value);
+    OPT("pass")
     {
-        p->vui.bEnableVuiParametersPresentFlag = 1;
-        p->vui.bEnableVuiHrdParametersPresentFlag = 1;
-        p->vui.bEnableNalHrdParametersPresentFlag = 1;
-        p->vui.bEnableSubPicHrdParamsPresentFlag = atobool(value);
+        int pass = Clip3(0, 3, atoi(value));
+        p->rc.bStatWrite = pass & 1;
+        p->rc.bStatRead = pass & 2;
     }
+    OPT("stats") p->rc.statFileName = strdup(value);
     else
         return X265_PARAM_BAD_NAME;
 #undef OPT
@@ -751,7 +748,6 @@
     return bError ? X265_PARAM_BAD_VALUE : 0;
 }
 
-
 namespace x265 {
 // internal encoder functions
 
@@ -779,9 +775,9 @@
     }
     int cpu;
     if (isdigit(value[0]))
-       cpu = x265_atoi(value, bError);
+        cpu = x265_atoi(value, bError);
     else
-       cpu = !strcmp(value, "auto") || x265_atobool(value, bError) ? x265::cpu_detect() : 0;
+        cpu = !strcmp(value, "auto") || x265_atobool(value, bError) ? x265::cpu_detect() : 0;
 
     if (bError)
     {
@@ -792,11 +788,15 @@
         for (init = buf; (tok = strtok_r(init, ",", &saveptr)); init = NULL)
         {
             int i;
-            for (i = 0; x265::cpu_names[i].flags && strcasecmp(tok, x265::cpu_names[i].name); i++);
+            for (i = 0; x265::cpu_names[i].flags && strcasecmp(tok, x265::cpu_names[i].name); i++)
+            {
+            }
+
             cpu |= x265::cpu_names[i].flags;
             if (!x265::cpu_names[i].flags)
                 bError = 1;
         }
+
         free(buf);
         if ((cpu & X265_CPU_SSSE3) && !(cpu & X265_CPU_SSE2_IS_SLOW))
             cpu |= X265_CPU_SSE2_IS_FAST;
@@ -827,8 +827,6 @@
 
 void setParamAspectRatio(x265_param *p, int width, int height)
 {
-    p->vui.bEnableVuiParametersPresentFlag = 1;
-    p->vui.bEnableAspectRatioIdc = 1;
     p->vui.aspectRatioIdc = X265_EXTENDED_SAR;
     p->vui.sarWidth = width;
     p->vui.sarHeight = height;
@@ -844,7 +842,7 @@
 
 void getParamAspectRatio(x265_param *p, int& width, int& height)
 {
-    if (!p->vui.bEnableVuiParametersPresentFlag || !p->vui.bEnableAspectRatioIdc || !p->vui.aspectRatioIdc)
+    if (!p->vui.aspectRatioIdc)
     {
         width = height = 0;
     }
@@ -883,15 +881,10 @@
     if (check_failed == 1)
         return check_failed;
 
-    uint32_t maxCUDepth = (uint32_t)g_convertToBit[param->maxCUSize];
-    uint32_t tuQTMaxLog2Size = maxCUDepth + 2 - 1;
+    uint32_t maxLog2CUSize = (uint32_t)g_log2Size[param->maxCUSize];
+    uint32_t tuQTMaxLog2Size = maxLog2CUSize - 1;
     uint32_t tuQTMinLog2Size = 2; //log2(4)
 
-    CHECK((param->maxCUSize >> maxCUDepth) < 4,
-          "Minimum partition width size should be larger than or equal to 8");
-    CHECK(param->internalCsp != X265_CSP_I420 && param->internalCsp != X265_CSP_I444,
-          "Only 4:2:0 and 4:4:4 color spaces is supported at this time");
-
     /* These checks might be temporary */
 #if HIGH_BIT_DEPTH
     CHECK(param->internalBitDepth != 10,
@@ -924,16 +917,16 @@
     CHECK(param->crQpOffset < -12, "Min. Chroma Cr QP Offset is -12");
     CHECK(param->crQpOffset >  12, "Max. Chroma Cr QP Offset is  12");
 
-    CHECK((1u << tuQTMaxLog2Size) > param->maxCUSize,
+    CHECK(tuQTMaxLog2Size > maxLog2CUSize,
           "QuadtreeTULog2MaxSize must be log2(maxCUSize) or smaller.");
 
     CHECK(param->tuQTMaxInterDepth < 1 || param->tuQTMaxInterDepth > 4,
           "QuadtreeTUMaxDepthInter must be greater than 0 and less than 5");
-    CHECK(param->maxCUSize < (1u << (tuQTMinLog2Size + param->tuQTMaxInterDepth - 1)),
+    CHECK(maxLog2CUSize < tuQTMinLog2Size + param->tuQTMaxInterDepth - 1,
           "QuadtreeTUMaxDepthInter must be less than or equal to the difference between log2(maxCUSize) and QuadtreeTULog2MinSize plus 1");
     CHECK(param->tuQTMaxIntraDepth < 1 || param->tuQTMaxIntraDepth > 4,
           "QuadtreeTUMaxDepthIntra must be greater 0 and less than 5");
-    CHECK(param->maxCUSize < (1u << (tuQTMinLog2Size + param->tuQTMaxIntraDepth - 1)),
+    CHECK(maxLog2CUSize < tuQTMinLog2Size + param->tuQTMaxIntraDepth - 1,
           "QuadtreeTUMaxDepthInter must be less than or equal to the difference between log2(maxCUSize) and QuadtreeTULog2MinSize plus 1");
 
     CHECK(param->maxNumMergeCand < 1, "MaxNumMergeCand must be 1 or greater.");
@@ -942,16 +935,20 @@
     CHECK(param->maxNumReferences < 1, "maxNumReferences must be 1 or greater.");
     CHECK(param->maxNumReferences > MAX_NUM_REF, "maxNumReferences must be 16 or smaller.");
 
-    CHECK(param->sourceWidth  % TComSPS::getWinUnitX(param->internalCsp) != 0,
+    CHECK(param->sourceWidth < (int)param->maxCUSize || param->sourceHeight < (int)param->maxCUSize,
+          "Picture size must be at least one CTU");
+    CHECK(param->internalCsp < X265_CSP_I420 || X265_CSP_I444 < param->internalCsp,
+          "Color space must be i420, i422, or i444");
+    CHECK(param->sourceWidth % g_winUnitX[param->internalCsp] != 0,
           "Picture width must be an integer multiple of the specified chroma subsampling");
-    CHECK(param->sourceHeight % TComSPS::getWinUnitY(param->internalCsp) != 0,
+    CHECK(param->sourceHeight % g_winUnitY[param->internalCsp] != 0,
           "Picture height must be an integer multiple of the specified chroma subsampling");
 
-    CHECK(param->rc.rateControlMode<X265_RC_ABR || param->rc.rateControlMode> X265_RC_CRF,
+    CHECK(param->rc.rateControlMode > X265_RC_CRF || param->rc.rateControlMode < X265_RC_ABR,
           "Rate control mode is out of range");
     CHECK(param->rdLevel < 0 || param->rdLevel > 6,
           "RD Level is out of range");
-    CHECK(param->bframes > param->lookaheadDepth,
+    CHECK(param->bframes > param->lookaheadDepth && !param->rc.bStatRead,
           "Lookahead depth must be greater than the max consecutive bframe count");
     CHECK(param->bframes < 0,
           "bframe count should be greater than zero");
@@ -963,8 +960,10 @@
           "Aq-Mode is out of range");
     CHECK(param->rc.aqStrength < 0 || param->rc.aqStrength > 3,
           "Aq-Strength is out of range");
-
+    CHECK(param->psyRd < 0 || 2.0 < param->psyRd, "Psy-rd strength must be between 0 and 2.0");
+    CHECK(param->psyRdoq < 0 || 10.0 < param->psyRdoq, "Psy-rdoq strength must be between 0 and 10.0");
     CHECK(param->bEnableWavefront < 0, "WaveFrontSynchro cannot be negative");
+    CHECK(!param->bEnableWavefront && param->rc.vbvBufferSize, "VBV requires wave-front parallelism (--wpp)");
     CHECK((param->vui.aspectRatioIdc < 0
            || param->vui.aspectRatioIdc > 16)
           && param->vui.aspectRatioIdc != X265_EXTENDED_SAR,
@@ -1006,21 +1005,22 @@
           "Default Display Window Top Offset must be 0 or greater");
     CHECK(param->vui.defDispWinBottomOffset < 0,
           "Default Display Window Bottom Offset must be 0 or greater");
-    CHECK(param->vui.bEnableNalHrdParametersPresentFlag
-          && param->rc.vbvBufferSize <= 0,
-          "If nal-hrd specified then vbv buffersize must also be specified");
-    CHECK(param->rc.rfConstant < 0 || param->rc.rfConstant > 51,
-          "Valid quality based VBR range 0 - 51");
+    CHECK(param->rc.rfConstant < -6 * (param->internalBitDepth - 8) || param->rc.rfConstant > 51,
+          "Valid quality based range: -qpBDOffsetY to 51");
+    CHECK(param->rc.rfConstantMax < -6 * (param->internalBitDepth - 8) || param->rc.rfConstantMax > 51,
+          "Valid quality based range: -qpBDOffsetY to 51");
+    CHECK(param->rc.rfConstantMin < -6 * (param->internalBitDepth - 8) || param->rc.rfConstantMin > 51,
+          "Valid quality based range: -qpBDOffsetY to 51");
     CHECK(param->bFrameAdaptive < 0 || param->bFrameAdaptive > 2,
           "Valid adaptive b scheduling values 0 - none, 1 - fast, 2 - full");
-    CHECK(param->logLevel < -1 || param->logLevel > X265_LOG_FULL,
+    CHECK(param->logLevel<-1 || param->logLevel> X265_LOG_FULL,
           "Valid Logging level -1:none 0:error 1:warning 2:info 3:debug 4:full");
     CHECK(param->scenecutThreshold < 0,
           "scenecutThreshold must be greater than 0");
     CHECK(param->rdPenalty < 0 || param->rdPenalty > 2,
-          "Valid penalty for 32x32 intra TU in non-I slices. 0:disabled 1:RD-penalty 2:maximum"); 
+          "Valid penalty for 32x32 intra TU in non-I slices. 0:disabled 1:RD-penalty 2:maximum");
     CHECK(param->keyframeMax < -1,
-          "Invalid max IDR period in frames. value should be greater than -1"); 
+          "Invalid max IDR period in frames. value should be greater than -1");
     CHECK(param->decodedPictureHashSEI < 0 || param->decodedPictureHashSEI > 3,
           "Invalid hash option. Decoded Picture Hash SEI 0: disabled, 1: MD5, 2: CRC, 3: Checksum");
     CHECK(param->rc.vbvBufferSize < 0,
@@ -1031,15 +1031,34 @@
           "Valid initial VBV buffer occupancy must be a fraction 0 - 1, or size in kbits");
     CHECK(param->rc.bitrate < 0,
           "Target bitrate can not be less than zero");
-    CHECK(param->bFrameBias < 0, "Bias towards B frame decisions must be 0 or greater");
+    if (param->noiseReduction)
+        CHECK(100 > param->noiseReduction || param->noiseReduction > 1000, "Valid noise reduction range 100 - 1000");
+    CHECK(param->rc.rateControlMode == X265_RC_CRF && param->rc.bStatRead,
+          "Constant rate-factor is incompatible with 2pass");
+    CHECK(param->rc.rateControlMode == X265_RC_CQP && param->rc.bStatRead,
+          "Constant QP is incompatible with 2pass");
     return check_failed;
 }
 
-int x265_set_globals(x265_param *param)
+void x265_param_apply_fastfirstpass(x265_param* param)
 {
-    uint32_t maxCUDepth = (uint32_t)g_convertToBit[param->maxCUSize];
-    uint32_t tuQTMinLog2Size = 2; //log2(4)
+    /* Set faster options in case of turbo firstpass */
+    if (param->rc.bStatWrite && !param->rc.bStatRead)
+    {
+        param->maxNumReferences = 1;
+        param->maxNumMergeCand = 1;
+        param->bEnableRectInter = 0;
+        param->bEnableFastIntra = 1;
+        param->bEnableAMP = 0;
+        param->searchMethod = X265_DIA_SEARCH;
+        param->subpelRefine = X265_MIN(2, param->subpelRefine);
+        param->bEnableEarlySkip = 1;
+        param->rdLevel = X265_MIN(2, param->rdLevel);
+    }
+}
 
+int x265_set_globals(x265_param *param)
+{
     static int once /* = 0 */;
 
     if (ATOMIC_CAS32(&once, 0, 1) == 1)
@@ -1052,27 +1071,20 @@
     }
     else
     {
+        uint32_t maxLog2CUSize = (uint32_t)g_log2Size[param->maxCUSize];
+
         // set max CU width & height
-        g_maxCUSize = param->maxCUSize;
+        g_maxCUSize     = param->maxCUSize;
+        g_maxLog2CUSize = maxLog2CUSize;
 
         // compute actual CU depth with respect to config depth and max transform size
-        g_addCUDepth = 0;
-        while ((param->maxCUSize >> maxCUDepth) > (1u << (tuQTMinLog2Size + g_addCUDepth)))
-        {
-            g_addCUDepth++;
-        }
-
-        maxCUDepth += g_addCUDepth;
-        g_addCUDepth++;
-        g_maxCUDepth = maxCUDepth;
+        g_maxCUDepth   = maxLog2CUSize - MIN_LOG2_CU_SIZE;
+        g_maxFullDepth = maxLog2CUSize - LOG2_UNIT_SIZE;
 
         // initialize partition order
         uint32_t* tmp = &g_zscanToRaster[0];
-        initZscanToRaster(g_maxCUDepth + 1, 1, 0, tmp);
-        initRasterToZscan(g_maxCUSize, g_maxCUSize, g_maxCUDepth + 1);
-
-        // initialize conversion matrix from partition index to pel
-        initRasterToPelXY(g_maxCUSize, g_maxCUSize, g_maxCUDepth + 1);
+        initZscanToRaster(g_maxFullDepth, 1, 0, tmp);
+        initRasterToZscan(g_maxFullDepth);
     }
     return 0;
 }
@@ -1110,8 +1122,12 @@
         x265_log(param, X265_LOG_INFO, "RDpenalty                    : %d\n", param->rdPenalty);
     }
     x265_log(param, X265_LOG_INFO, "Lookahead / bframes / badapt        : %d / %d / %d\n", param->lookaheadDepth, param->bframes, param->bFrameAdaptive);
-    x265_log(param, X265_LOG_INFO, "b-pyramid / weightp / refs          : %d / %d / %d\n", param->bBPyramid, param->bEnableWeightedPred, param->maxNumReferences);
-    switch (param->rc.rateControlMode)
+    x265_log(param, X265_LOG_INFO, "b-pyramid / weightp / weightb / refs: %d / %d / %d / %d\n",
+             param->bBPyramid, param->bEnableWeightedPred, param->bEnableWeightedBiPred, param->maxNumReferences);
+
+    if (param->bLossless)
+        x265_log(param, X265_LOG_INFO, "Rate Control                        : Lossless\n");
+    else switch (param->rc.rateControlMode)
     {
     case X265_RC_ABR:
         x265_log(param, X265_LOG_INFO, "Rate Control / AQ-Strength / CUTree : ABR-%d kbps / %0.1f / %d\n", param->rc.bitrate,
@@ -1126,6 +1142,7 @@
                  param->rc.aqStrength, param->rc.cuTree);
         break;
     }
+
     if (param->rc.vbvBufferSize)
     {
         x265_log(param, X265_LOG_INFO, "VBV/HRD buffer / max-rate / init    : %d / %d / %.3f\n",
@@ -1140,6 +1157,12 @@
     TOOLOPT(param->bEnableConstrainedIntra, "cip");
     TOOLOPT(param->bEnableEarlySkip, "esd");
     fprintf(stderr, "rd=%d ", param->rdLevel);
+    if (param->psyRd > 0.)
+        fprintf(stderr, "psy-rd=%.2lf ", param->psyRd);
+    if (param->psyRdoq > 0.)
+        fprintf(stderr, "psy-rdoq=%.2lf ", param->psyRdoq);
+    if (param->noiseReduction)
+        fprintf(stderr, "nr=%d ", param->noiseReduction);
 
     TOOLOPT(param->bEnableLoopFilter, "lft");
     if (param->bEnableSAO)
@@ -1149,7 +1172,9 @@
         else
             fprintf(stderr, "sao-frame ");
     }
-    TOOLOPT(param->bEnableSignHiding, "sign-hide");
+    TOOLOPT(param->bEnableSignHiding, "signhide");
+    TOOLOPT(param->bCULossless, "cu-lossless");
+    TOOLOPT(param->bEnableFastIntra, "fast-intra");
     if (param->bEnableTransformSkip)
     {
         if (param->bEnableTSkipFast)
@@ -1157,7 +1182,8 @@
         else
             fprintf(stderr, "tskip ");
     }
-    TOOLOPT(param->bEnableWeightedBiPred, "weightbp");
+    TOOLOPT(param->rc.bStatWrite, "stats-write");
+    TOOLOPT(param->rc.bStatRead,  "stats-read");
     fprintf(stderr, "\n");
     fflush(stderr);
 }
@@ -1166,15 +1192,17 @@
 {
     char *buf, *s;
 
-    buf = s = X265_MALLOC(char, 2000);
+    buf = s = X265_MALLOC(char, MAXPARAMSIZE);
     if (!buf)
         return NULL;
 
 #define BOOL(param, cliopt) \
     s += sprintf(s, " %s", (param) ? cliopt : "no-"cliopt);
 
+    s += sprintf(s, "%dx%d", p->sourceWidth,p->sourceHeight);
+    s += sprintf(s, " fps=%u/%u", p->fpsNum, p->fpsDenom);
+    s += sprintf(s, " bitdepth=%d", p->internalBitDepth);
     BOOL(p->bEnableWavefront, "wpp");
-    s += sprintf(s, " fps=%d/%d", p->fpsNum, p->fpsDenom);
     s += sprintf(s, " ctu=%d", p->maxCUSize);
     s += sprintf(s, " tu-intra-depth=%d", p->tuQTMaxIntraDepth);
     s += sprintf(s, " tu-inter-depth=%d", p->tuQTMaxInterDepth);
@@ -1190,7 +1218,10 @@
     BOOL(p->bEnableTransformSkip, "tskip");
     BOOL(p->bEnableTSkipFast, "tskip-fast");
     BOOL(p->bEnableStrongIntraSmoothing, "strong-intra-smoothing");
+    BOOL(p->bLossless, "lossless");
+    BOOL(p->bCULossless, "cu-lossless");
     BOOL(p->bEnableConstrainedIntra, "constrained-intra");
+    BOOL(p->bEnableFastIntra, "fast-intra");
     BOOL(p->bOpenGOP, "open-gop");
     s += sprintf(s, " interlace=%d", p->interlaceMode);
     s += sprintf(s, " keyint=%d", p->keyframeMax);
@@ -1202,23 +1233,124 @@
     s += sprintf(s, " b-adapt=%d", p->bFrameAdaptive);
     s += sprintf(s, " ref=%d", p->maxNumReferences);
     BOOL(p->bEnableWeightedPred, "weightp");
-    s += sprintf(s, " bitrate=%d", p->rc.bitrate);
-    s += sprintf(s, " qp=%d", p->rc.qp);
+    BOOL(p->bEnableWeightedBiPred, "weightb");
     s += sprintf(s, " aq-mode=%d", p->rc.aqMode);
     s += sprintf(s, " aq-strength=%.2f", p->rc.aqStrength);
     s += sprintf(s, " cbqpoffs=%d", p->cbQpOffset);
     s += sprintf(s, " crqpoffs=%d", p->crQpOffset);
     s += sprintf(s, " rd=%d", p->rdLevel);
+    s += sprintf(s, " psy-rd=%.2f", p->psyRd);
+    s += sprintf(s, " psy-rdoq=%.2f", p->psyRdoq);
     BOOL(p->bEnableSignHiding, "signhide");
     BOOL(p->bEnableLoopFilter, "lft");
     BOOL(p->bEnableSAO, "sao");
     s += sprintf(s, " sao-lcu-bounds=%d", p->saoLcuBoundary);
     s += sprintf(s, " sao-lcu-opt=%d", p->saoLcuBasedOptimization);
-    s += sprintf(s, " b-pyramid=%d", p->bBPyramid);
+    BOOL(p->bBPyramid, "b-pyramid");
     BOOL(p->rc.cuTree, "cutree");
+    s += sprintf(s, " rc=%s", p->rc.rateControlMode == X265_RC_ABR ? (
+         p->rc.bStatRead ? "2 pass" : p->rc.bitrate == p->rc.vbvMaxBitrate ? "cbr" : "abr")
+         : p->rc.rateControlMode == X265_RC_CRF ? "crf" : "cqp");
+    if (p->rc.rateControlMode == X265_RC_ABR || p->rc.rateControlMode == X265_RC_CRF)
+    {
+        if (p->rc.rateControlMode == X265_RC_CRF)
+            s += sprintf(s, " crf=%.1f", p->rc.rfConstant);
+        else
+            s += sprintf(s, " bitrate=%d ratetol=%.1f",
+                         p->rc.bitrate, p->rc.rateTolerance);
+        s += sprintf(s, " qcomp=%.2f qpmin=%d qpmax=%d qpstep=%d",
+                     p->rc.qCompress, QP_MIN, QP_MAX_SPEC, p->rc.qpStep);
+        if (p->rc.bStatRead)
+            s += sprintf( s, " cplxblur=%.1f qblur=%.1f",
+                          p->rc.complexityBlur, p->rc.qblur);
+        if (p->rc.vbvBufferSize)
+        {
+            s += sprintf(s, " vbv-maxrate=%d vbv-bufsize=%d",
+                          p->rc.vbvMaxBitrate, p->rc.vbvBufferSize);
+            if (p->rc.rateControlMode == X265_RC_CRF)
+                s += sprintf(s, " crf-max=%.1f", p->rc.rfConstantMax);
+        }
+    }
+    else if (p->rc.rateControlMode == X265_RC_CQP)
+        s += sprintf(s, " qp=%d", p->rc.qp);
+    if (!(p->rc.rateControlMode == X265_RC_CQP && p->rc.qp == 0))
+    {
+        s += sprintf(s, " ipratio=%.2f", p->rc.ipFactor);
+        if (p->bframes)
+            s += sprintf(s, " pbratio=%.2f", p->rc.pbFactor);
+    }
 #undef BOOL
-
     return buf;
 }
 
+bool parseLambdaFile(x265_param *param)
+{
+    if (!param->rc.lambdaFileName)
+        return false;
+
+    FILE *lfn = fopen(param->rc.lambdaFileName, "r");
+    if (!lfn)
+    {
+        x265_log(param, X265_LOG_ERROR, "unable to read lambda file <%s>\n", param->rc.lambdaFileName);
+        return true;
+    }
+
+    char line[2048];
+    char *toksave = NULL, *tok = NULL, *buf = NULL;
+
+    for (int t = 0; t < 3; t++)
+    {
+        double *table = t ? x265_lambda2_tab : x265_lambda_tab;
+
+        for (int i = 0; i < QP_MAX_MAX + 1; i++)
+        {
+            double value;
+
+            do
+            {
+                if (!tok)
+                {
+                    /* consume a line of text file */
+                    if (!fgets(line, sizeof(line), lfn))
+                    {
+                        fclose(lfn);
+
+                        if (t < 2)
+                        {
+                            x265_log(param, X265_LOG_ERROR, "lambda file is incomplete\n");
+                            return true;
+                        }
+                        else
+                            return false;
+                    }
+
+                    /* truncate at first hash */
+                    char *hash = strchr(line, '#');
+                    if (hash) *hash = 0;
+                    buf = line;
+                }
+
+                tok = strtok_r(buf, " ,", &toksave);
+                buf = NULL;
+                if (tok && sscanf(tok, "%lf", &value) == 1)
+                    break;
+            }
+            while (1);
+
+            if (t == 2)
+            {
+                x265_log(param, X265_LOG_ERROR, "lambda file contains too many values\n");
+                fclose(lfn);
+                return true;
+            }
+            else
+                x265_log(param, X265_LOG_DEBUG, "lambda%c[%d] = %lf\n", t ? '2' : ' ', i, value);
+            table[i] = value;
+        }
+    }
+
+    fclose(lfn);
+    return false;
+}
+
 }

x265-hg20140324.tar.bz2/source/common/param.h -> x265-hg20140928.tar.bz2/source/common/param.h Changed

@@ -18,26 +18,28 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_PARAM_H
 #define X265_PARAM_H
 
 namespace x265 {
-
 int   x265_check_params(x265_param *param);
 int   x265_set_globals(x265_param *param);
 void  x265_print_params(x265_param *param);
+void  x265_param_apply_fastfirstpass(x265_param *p);
 char* x265_param2string(x265_param *param);
 int   x265_atoi(const char *str, bool& bError);
 int   parseCpuName(const char *value, bool& bError);
 void  setParamAspectRatio(x265_param *p, int width, int height);
 void  getParamAspectRatio(x265_param *p, int& width, int& height);
+bool  parseLambdaFile(x265_param *param);
 
 /* this table is kept internal to avoid confusion, since log level indices start at -1 */
 static const char * const logLevelNames[] = { "none", "error", "warning", "info", "debug", "full", 0 };
 
+#define MAXPARAMSIZE 2000
 }
 
 #endif // ifndef X265_PARAM_H

x265-hg20140324.tar.bz2/source/common/piclist.cpp -> x265-hg20140928.tar.bz2/source/common/piclist.cpp Changed

@@ -18,17 +18,18 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
+#include "common.h"
 #include "piclist.h"
-#include "TLibCommon/TComPic.h"
+#include "frame.h"
 
 using namespace x265;
 
-void PicList::pushFront(TComPic& pic)
+void PicList::pushFront(Frame& pic)
 {
-    assert(!pic.m_next && !pic.m_prev); // ensure pic is not in a list
+    X265_CHECK(!pic.m_next && !pic.m_prev, "piclist: picture already in list\n"); // ensure pic is not in a list
     pic.m_next = m_start;
     pic.m_prev = NULL;
 
@@ -44,9 +45,9 @@
     m_count++;
 }
 
-void PicList::pushBack(TComPic& pic)
+void PicList::pushBack(Frame& pic)
 {
-    assert(!pic.m_next && !pic.m_prev); // ensure pic is not in a list
+    X265_CHECK(!pic.m_next && !pic.m_prev, "piclist: picture already in list\n"); // ensure pic is not in a list
     pic.m_next = NULL;
     pic.m_prev = m_end;
 
@@ -62,11 +63,11 @@
     m_count++;
 }
 
-TComPic *PicList::popFront()
+Frame *PicList::popFront()
 {
     if (m_start)
     {
-        TComPic *temp = m_start;
+        Frame *temp = m_start;
         m_count--;
 
         if (m_count)
@@ -85,11 +86,19 @@
         return NULL;
 }
 
-TComPic *PicList::popBack()
+Frame* PicList::getPOC(int poc)
+{
+    Frame *pic = m_start;
+    while (pic && pic->getPOC() != poc)
+        pic = pic->m_next;
+    return pic;
+}
+
+Frame *PicList::popBack()
 {
     if (m_end)
     {
-        TComPic* temp = m_end;
+        Frame* temp = m_end;
         m_count--;
 
         if (m_count)
@@ -108,16 +117,16 @@
         return NULL;
 }
 
-void PicList::remove(TComPic& pic)
+void PicList::remove(Frame& pic)
 {
 #if _DEBUG
-    TComPic *tmp = m_start;
+    Frame *tmp = m_start;
     while (tmp && tmp != &pic)
     {
         tmp = tmp->m_next;
     }
 
-    assert(tmp == &pic); // verify pic is in this list
+    X265_CHECK(tmp == &pic, "piclist: pic being removed was not in list\n"); // verify pic is in this list
 #endif
 
     m_count--;

x265-hg20140324.tar.bz2/source/common/piclist.h -> x265-hg20140928.tar.bz2/source/common/piclist.h Changed

x265-hg20140324.tar.bz2/source/common/pixel.cpp -> x265-hg20140928.tar.bz2/source/common/pixel.cpp Changed

@@ -21,7 +21,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "TLibCommon/TComRom.h"
@@ -375,59 +375,73 @@
     return cost;
 }
 
-void blockcopy_p_p(int bx, int by, pixel *a, intptr_t stridea, pixel *b, intptr_t strideb)
+template<int size>
+int pixel_ssd_s_c(short *a, intptr_t dstride)
 {
-    for (int y = 0; y < by; y++)
+    int sum = 0;
+    for (int y = 0; y < size; y++)
     {
-        for (int x = 0; x < bx; x++)
+        for (int x = 0; x < size; x++)
         {
-            a[x] = b[x];
+            sum += a[x] * a[x];
         }
-
-        a += stridea;
-        b += strideb;
+        a += dstride;
     }
+    return sum;
 }
 
-void blockcopy_p_s(int bx, int by, pixel *a, intptr_t stridea, int16_t *b, intptr_t strideb)
+template<int size>
+void blockfil_s_c(int16_t *dst, intptr_t dstride, int16_t val)
 {
-    for (int y = 0; y < by; y++)
+    for (int y = 0; y < size; y++)
     {
-        for (int x = 0; x < bx; x++)
+        for (int x = 0; x < size; x++)
         {
-            assert((b[x] >= 0) && (b[x] <= ((1 << X265_DEPTH) - 1)));
-            a[x] = (pixel)b[x];
+            dst[y * dstride + x] = val;
         }
+    }
+}
 
-        a += stridea;
-        b += strideb;
+void convert16to32_shl(int32_t *dst, int16_t *src, intptr_t stride, int shift, int size)
+{
+    for (int i = 0; i < size; i++)
+    {
+        for (int j = 0; j < size; j++)
+        {
+            dst[i * size + j] = ((int)src[i * stride + j]) << shift;
+        }
     }
 }
 
 template<int size>
-void blockfil_s_c(int16_t *dst, intptr_t dstride, int16_t val)
+void convert16to32_shr(int32_t *dst, int16_t *src, intptr_t stride, int shift, int offset)
 {
-    for (int y = 0; y < size; y++)
+    for (int i = 0; i < size; i++)
     {
-        for (int x = 0; x < size; x++)
+        for (int j = 0; j < size; j++)
         {
-            dst[y * dstride + x] = val;
+            dst[i * size + j] = ((int)src[i * stride + j] + offset) >> shift;
         }
     }
 }
 
-void convert16to32_shl(int32_t *dst, int16_t *src, intptr_t stride, int shift, int size)
+void convert32to16_shr(int16_t *dst, int32_t *src, intptr_t stride, int shift, int size)
 {
+    int round = 1 << (shift - 1);
+
     for (int i = 0; i < size; i++)
     {
         for (int j = 0; j < size; j++)
         {
-            dst[i * size + j] = ((int)src[i * stride + j]) << shift;
+            dst[j] = (int16_t)((src[j] + round) >> shift);
         }
+
+        src += size;
+        dst += stride;
     }
 }
 
-void convert32to16_shr(int16_t *dst, int32_t *src, intptr_t stride, int shift, int size)
+void copy_shr(int16_t *dst, int16_t *src, intptr_t stride, int shift, int size)
 {
     int round = 1 << (shift - 1);
 
@@ -443,14 +457,44 @@
     }
 }
 
+template<int size>
+void convert32to16_shl(int16_t *dst, int32_t *src, intptr_t stride, int shift)
+{
+    for (int i = 0; i < size; i++)
+    {
+        for (int j = 0; j < size; j++)
+        {
+            dst[j] = ((int16_t)src[j] << shift);
+        }
+
+        src += size;
+        dst += stride;
+    }
+}
+
+template<int size>
+void copy_shl(int16_t *dst, int16_t *src, intptr_t stride, int shift)
+{
+    for (int i = 0; i < size; i++)
+    {
+        for (int j = 0; j < size; j++)
+        {
+            dst[j] = (src[j] << shift);
+        }
+
+        src += size;
+        dst += stride;
+    }
+}
+
 template<int blockSize>
 void getResidual(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride)
 {
-    for (int uiY = 0; uiY < blockSize; uiY++)
+    for (int y = 0; y < blockSize; y++)
     {
-        for (int uiX = 0; uiX < blockSize; uiX++)
+        for (int x = 0; x < blockSize; x++)
         {
-            residual[uiX] = static_cast<int16_t>(fenc[uiX]) - static_cast<int16_t>(pred[uiX]);
+            residual[x] = static_cast<int16_t>(fenc[x]) - static_cast<int16_t>(pred[x]);
         }
 
         fenc += stride;
@@ -460,33 +504,18 @@
 }
 
 template<int blockSize>
-void calcRecons(pixel* pred, int16_t* residual,
-#if NEW_CALCRECON
-                pixel*,
-#else
-                pixel* recon,
-#endif
-                int16_t* recqt, pixel* recipred, int stride, int qtstride, int ipredstride)
+void calcRecons(pixel* pred, int16_t* residual, int16_t* recqt, pixel* recipred, int stride, int qtstride, int ipredstride)
 {
-    for (int uiY = 0; uiY < blockSize; uiY++)
+    for (int y = 0; y < blockSize; y++)
     {
-        for (int uiX = 0; uiX < blockSize; uiX++)
+        for (int x = 0; x < blockSize; x++)
         {
-#if NEW_CALCRECON
-            recqt[uiX] = (int16_t)ClipY(static_cast<int16_t>(pred[uiX]) + residual[uiX]);
-            recipred[uiX] = (pixel)recqt[uiX];
-#else
-            recon[uiX] = (pixel)ClipY(static_cast<int16_t>(pred[uiX]) + residual[uiX]);
-            recqt[uiX] = (int16_t)recon[uiX];
-            recipred[uiX] = recon[uiX];
-#endif
+            recqt[x] = (int16_t)Clip(static_cast<int16_t>(pred[x]) + residual[x]);
+            recipred[x] = (pixel)recqt[x];
         }
 
         pred += stride;
         residual += stride;
-#if !NEW_CALCRECON
-        recon += stride;
-#endif
         recqt += qtstride;
         recipred += ipredstride;
     }
@@ -526,7 +555,7 @@
 {
     int x, y;
 
-    assert(!(width & 15));
+    X265_CHECK(!(width & 15), "weightp alignment error\n");
 
     for (y = 0; y <= height - 1; y++)
     {
@@ -543,21 +572,6 @@
     }
 }
 
-void pixeladd_ss_c(int bx, int by, int16_t *a, intptr_t dstride, int16_t *b0, int16_t *b1, intptr_t sstride0, intptr_t sstride1)
-{
-    for (int y = 0; y < by; y++)
-    {
-        for (int x = 0; x < bx; x++)
-        {
-            a[x] = (int16_t)ClipY(b0[x] + b1[x]);
-        }
-
-        b0 += sstride0;
-        b1 += sstride1;
-        a += dstride;
-    }
-}
-
 template<int lx, int ly>
 void pixelavg_pp(pixel* dst, intptr_t dstride, pixel* src0, intptr_t sstride0, pixel* src1, intptr_t sstride1, int)
 {
@@ -669,12 +683,12 @@
 
 #define PIXEL_MAX ((1 << X265_DEPTH) - 1)
 #if HIGH_BIT_DEPTH
-    assert(X265_DEPTH == 10);
+    X265_CHECK(X265_DEPTH == 10, "ssim invalid depth\n");
 #define type float
     static const float ssim_c1 = (float)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64);
     static const float ssim_c2 = (float)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63);
 #else
-    assert(X265_DEPTH == 8);
+    X265_CHECK(X265_DEPTH == 8, "ssim invalid depth\n");
 #define type int
     static const int ssim_c1 = (int)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64 + .5);
     static const int ssim_c2 = (int)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63 + .5);
@@ -725,6 +739,43 @@
     return sum + ((uint64_t)sqr << 32);
 }
 
+#if defined(_MSC_VER)
+#pragma warning(disable: 4127) // conditional expression is constant
+#endif
+
+template<int size>
+int psyCost(pixel *source, intptr_t sstride, pixel *recon, intptr_t rstride)
+{
+    static pixel zeroBuf[8] /* = { 0 } */;
+
+    if (size)
+    {
+        int dim = 1 << (size + 2);
+        uint32_t totEnergy = 0;
+        for (int i = 0; i < dim; i += 8)
+        {
+            for (int j = 0; j < dim; j+= 8)
+            {
+                /* AC energy, measured by sa8d (AC + DC) minus SAD (DC) */
+                int sourceEnergy = sa8d_8x8(source + i * sstride + j, sstride, zeroBuf, 0) - 
+                                   (sad<8, 8>(source + i * sstride + j, sstride, zeroBuf, 0) >> 2);
+                int reconEnergy =  sa8d_8x8(recon + i * rstride + j, rstride, zeroBuf, 0) - 
+                                   (sad<8, 8>(recon + i * rstride + j, rstride, zeroBuf, 0) >> 2);
+
+                totEnergy += abs(sourceEnergy - reconEnergy);
+            }
+        }
+        return totEnergy;
+    }
+    else
+    {
+        /* 4x4 is too small for sa8d */
+        int sourceEnergy = satd_4x4(source, sstride, zeroBuf, 0) - (sad<4, 4>(source, sstride, zeroBuf, 0) >> 2);
+        int reconEnergy = satd_4x4(recon, rstride, zeroBuf, 0) - (sad<4, 4>(recon, rstride, zeroBuf, 0) >> 2);
+        return abs(sourceEnergy - reconEnergy);
+    }
+}
+
 void plane_copy_deinterleave_chroma(pixel *dstu, intptr_t dstuStride, pixel *dstv, intptr_t dstvStride,
                                     pixel *src,  intptr_t srcStride, int w, int h)
 {
@@ -775,7 +826,7 @@
     {
         for (int x = 0; x < bx; x++)
         {
-            assert((b[x] >= 0) && (b[x] <= ((1 << X265_DEPTH) - 1)));
+            X265_CHECK((b[x] >= 0) && (b[x] <= ((1 << X265_DEPTH) - 1)), "blockcopy pixel size fail\n");
             a[x] = (pixel)b[x];
         }
 
@@ -822,7 +873,7 @@
     {
         for (int x = 0; x < bx; x++)
         {
-            a[x] = (pixel)ClipY(b0[x] + b1[x]);
+            a[x] = Clip(b0[x] + b1[x]);
         }
 
         b0 += sstride0;
@@ -843,8 +894,8 @@
     {
         for (int x = 0; x < bx; x += 2)
         {
-            dst[x + 0] = (pixel)ClipY((src0[x + 0] + src1[x + 0] + offset) >> shiftNum);
-            dst[x + 1] = (pixel)ClipY((src0[x + 1] + src1[x + 1] + offset) >> shiftNum);
+            dst[x + 0] = Clip((src0[x + 0] + src1[x + 0] + offset) >> shiftNum);
+            dst[x + 1] = Clip((src0[x + 1] + src1[x + 1] + offset) >> shiftNum);
         }
 
         src0 += src0Stride;
@@ -852,6 +903,51 @@
         dst  += dstStride;
     }
 }
+
+void planecopy_cp_c(uint8_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift)
+{
+    for (int r = 0; r < height; r++)
+    {
+        for (int c = 0; c < width; c++)
+        {
+            dst[c] = ((pixel)src[c]) << shift;
+        }
+
+        dst += dstStride;
+        src += srcStride;
+    }
+}
+
+void planecopy_sp_c(uint16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask)
+{
+    for (int r = 0; r < height; r++)
+    {
+        for (int c = 0; c < width; c++)
+        {
+            dst[c] = (pixel)((src[c] >> shift) & mask);
+        }
+
+        dst += dstStride;
+        src += srcStride;
+    }
+}
+
+/* Estimate the total amount of influence on future quality that could be had if we
+ * were to improve the reference samples used to inter predict any given CU. */
+void estimateCUPropagateCost(int *dst, uint16_t *propagateIn, int32_t *intraCosts, uint16_t *interCosts,
+                             int32_t *invQscales, double *fpsFactor, int len)
+{
+    double fps = *fpsFactor / 256;
+
+    for (int i = 0; i < len; i++)
+    {
+        double intraCost       = intraCosts[i] * invQscales[i];
+        double propagateAmount = (double)propagateIn[i] + intraCost * fps;
+        double propagateNum    = (double)intraCosts[i] - (interCosts[i] & ((1 << 14) - 1));
+        double propagateDenom  = (double)intraCosts[i];
+        dst[i] = (int)(propagateAmount * propagateNum / propagateDenom + 0.5);
+    }
+}
 }  // end anonymous namespace
 
 namespace x265 {
@@ -916,32 +1012,51 @@
     p.satd[LUMA_16x64] = satd8<16, 64>;
 
 #define CHROMA_420(W, H) \
-    p.chroma[X265_CSP_I420].addAvg[CHROMA_ ## W ## x ## H]  = addAvg<W, H>; \
+    p.chroma[X265_CSP_I420].addAvg[CHROMA_ ## W ## x ## H]  = addAvg<W, H>;         \
     p.chroma[X265_CSP_I420].copy_pp[CHROMA_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
     p.chroma[X265_CSP_I420].copy_sp[CHROMA_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
     p.chroma[X265_CSP_I420].copy_ps[CHROMA_ ## W ## x ## H] = blockcopy_ps_c<W, H>; \
-    p.chroma[X265_CSP_I420].copy_ss[CHROMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>; \
-    p.chroma[X265_CSP_I420].sub_ps[CHROMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>; \
-    p.chroma[X265_CSP_I420].add_ps[CHROMA_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
+    p.chroma[X265_CSP_I420].copy_ss[CHROMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>;
+
+#define CHROMA_422(W, H) \
+    p.chroma[X265_CSP_I422].addAvg[CHROMA422_ ## W ## x ## H] = addAvg<W, H>;         \
+    p.chroma[X265_CSP_I422].copy_pp[CHROMA422_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
+    p.chroma[X265_CSP_I422].copy_sp[CHROMA422_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
+    p.chroma[X265_CSP_I422].copy_ps[CHROMA422_ ## W ## x ## H] = blockcopy_ps_c<W, H>; \
+    p.chroma[X265_CSP_I422].copy_ss[CHROMA422_ ## W ## x ## H] = blockcopy_ss_c<W, H>;
 
 #define CHROMA_444(W, H) \
     p.chroma[X265_CSP_I444].addAvg[LUMA_ ## W ## x ## H]  = addAvg<W, H>; \
     p.chroma[X265_CSP_I444].copy_pp[LUMA_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
     p.chroma[X265_CSP_I444].copy_sp[LUMA_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
     p.chroma[X265_CSP_I444].copy_ps[LUMA_ ## W ## x ## H] = blockcopy_ps_c<W, H>; \
-    p.chroma[X265_CSP_I444].copy_ss[LUMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>; \
-    p.chroma[X265_CSP_I444].sub_ps[LUMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>; \
-    p.chroma[X265_CSP_I444].add_ps[LUMA_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
+    p.chroma[X265_CSP_I444].copy_ss[LUMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>;
 
 #define LUMA(W, H) \
     p.luma_addAvg[LUMA_ ## W ## x ## H]  = addAvg<W, H>; \
     p.luma_copy_pp[LUMA_ ## W ## x ## H] = blockcopy_pp_c<W, H>; \
     p.luma_copy_sp[LUMA_ ## W ## x ## H] = blockcopy_sp_c<W, H>; \
     p.luma_copy_ps[LUMA_ ## W ## x ## H] = blockcopy_ps_c<W, H>; \
-    p.luma_copy_ss[LUMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>; \
+    p.luma_copy_ss[LUMA_ ## W ## x ## H] = blockcopy_ss_c<W, H>;
+
+#define LUMA_PIXELSUB(W, H) \
     p.luma_sub_ps[LUMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>; \
     p.luma_add_ps[LUMA_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
 
+#define CHROMA_PIXELSUB_420(W, H) \
+    p.chroma[X265_CSP_I420].sub_ps[CHROMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>;  \
+    p.chroma[X265_CSP_I420].add_ps[CHROMA_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
+
+#define CHROMA_PIXELSUB_422(W, H) \
+    p.chroma[X265_CSP_I422].sub_ps[CHROMA422_ ## W ## x ## H] = pixel_sub_ps_c<W, H>; \
+    p.chroma[X265_CSP_I422].add_ps[CHROMA422_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
+
+#define CHROMA_PIXELSUB_444(W, H) \
+    p.chroma[X265_CSP_I444].sub_ps[LUMA_ ## W ## x ## H] = pixel_sub_ps_c<W, H>; \
+    p.chroma[X265_CSP_I444].add_ps[LUMA_ ## W ## x ## H] = pixel_add_ps_c<W, H>;
+
+
+
     LUMA(4, 4);
     LUMA(8, 8);
     CHROMA_420(4, 4);
@@ -992,6 +1107,49 @@
     LUMA(16, 64);
     CHROMA_420(8,  32);
 
+    LUMA_PIXELSUB(4, 4);
+    LUMA_PIXELSUB(8, 8);
+    LUMA_PIXELSUB(16, 16);
+    LUMA_PIXELSUB(32, 32);
+    LUMA_PIXELSUB(64, 64);
+    CHROMA_PIXELSUB_420(4, 4)
+    CHROMA_PIXELSUB_420(8, 8)
+    CHROMA_PIXELSUB_420(16, 16)
+    CHROMA_PIXELSUB_420(32, 32)
+    CHROMA_PIXELSUB_422(4, 8)
+    CHROMA_PIXELSUB_422(8, 16)
+    CHROMA_PIXELSUB_422(16, 32)
+    CHROMA_PIXELSUB_422(32, 64)
+    CHROMA_PIXELSUB_444(8, 8)
+    CHROMA_PIXELSUB_444(16, 16)
+    CHROMA_PIXELSUB_444(32, 32)
+    CHROMA_PIXELSUB_444(64, 64)
+
+    CHROMA_422(4, 8);
+    CHROMA_422(4, 4);
+    CHROMA_422(2, 8);
+    CHROMA_422(8,  16);
+    CHROMA_422(8,  8);
+    CHROMA_422(4,  16);
+    CHROMA_422(8,  12);
+    CHROMA_422(6,  16);
+    CHROMA_422(8,  4);
+    CHROMA_422(2,  16);
+    CHROMA_422(16, 32);
+    CHROMA_422(16, 16);
+    CHROMA_422(8,  32);
+    CHROMA_422(16, 24);
+    CHROMA_422(12, 32);
+    CHROMA_422(16, 8);
+    CHROMA_422(4,  32);
+    CHROMA_422(32, 64);
+    CHROMA_422(32, 32);
+    CHROMA_422(16, 64);
+    CHROMA_422(32, 48);
+    CHROMA_422(24, 64);
+    CHROMA_422(32, 16);
+    CHROMA_422(8,  64);
+
     CHROMA_444(4,  4);
     CHROMA_444(8,  8);
     CHROMA_444(4,  8);
@@ -1022,9 +1180,6 @@
     SET_FUNC_PRIMITIVE_TABLE_C(sse_sp, sse, pixelcmp_sp_t, int16_t, pixel)
     SET_FUNC_PRIMITIVE_TABLE_C(sse_ss, sse, pixelcmp_ss_t, int16_t, int16_t)
 
-    p.blockcpy_pp = blockcopy_p_p;
-    p.blockcpy_ps = blockcopy_p_s;
-
     p.blockfill_s[BLOCK_4x4]   = blockfil_s_c<4>;
     p.blockfill_s[BLOCK_8x8]   = blockfil_s_c<8>;
     p.blockfill_s[BLOCK_16x16] = blockfil_s_c<16>;
@@ -1032,7 +1187,21 @@
     p.blockfill_s[BLOCK_64x64] = blockfil_s_c<64>;
 
     p.cvt16to32_shl = convert16to32_shl;
+    p.cvt16to32_shr[BLOCK_4x4] = convert16to32_shr<4>;
+    p.cvt16to32_shr[BLOCK_8x8] = convert16to32_shr<8>;
+    p.cvt16to32_shr[BLOCK_16x16] = convert16to32_shr<16>;
+    p.cvt16to32_shr[BLOCK_32x32] = convert16to32_shr<32>;
     p.cvt32to16_shr = convert32to16_shr;
+    p.cvt32to16_shl[BLOCK_4x4] = convert32to16_shl<4>;
+    p.cvt32to16_shl[BLOCK_8x8] = convert32to16_shl<8>;
+    p.cvt32to16_shl[BLOCK_16x16] = convert32to16_shl<16>;
+    p.cvt32to16_shl[BLOCK_32x32] = convert32to16_shl<32>;
+
+    p.copy_shr = copy_shr;
+    p.copy_shl[BLOCK_4x4] = copy_shl<4>;
+    p.copy_shl[BLOCK_8x8] = copy_shl<8>;
+    p.copy_shl[BLOCK_16x16] = copy_shl<16>;
+    p.copy_shl[BLOCK_32x32] = copy_shl<32>;
 
     p.sa8d[BLOCK_4x4]   = satd_4x4;
     p.sa8d[BLOCK_8x8]   = sa8d_8x8;
@@ -1040,6 +1209,12 @@
     p.sa8d[BLOCK_32x32] = sa8d16<32, 32>;
     p.sa8d[BLOCK_64x64] = sa8d16<64, 64>;
 
+    p.psy_cost[BLOCK_4x4] = psyCost<BLOCK_4x4>;
+    p.psy_cost[BLOCK_8x8] = psyCost<BLOCK_8x8>;
+    p.psy_cost[BLOCK_16x16] = psyCost<BLOCK_16x16>;
+    p.psy_cost[BLOCK_32x32] = psyCost<BLOCK_32x32>;
+    p.psy_cost[BLOCK_64x64] = psyCost<BLOCK_64x64>;
+
     p.sa8d_inter[LUMA_4x4]   = satd_4x4;
     p.sa8d_inter[LUMA_8x8]   = sa8d_8x8;
     p.sa8d_inter[LUMA_8x4]   = satd_8x4;
@@ -1077,17 +1252,20 @@
     p.calcrecon[BLOCK_32x32] = calcRecons<32>;
     p.calcrecon[BLOCK_64x64] = NULL;
 
-    p.transpose[0] = transpose<4>;
-    p.transpose[1] = transpose<8>;
-    p.transpose[2] = transpose<16>;
-    p.transpose[3] = transpose<32>;
-    p.transpose[4] = transpose<64>;
+    p.transpose[BLOCK_4x4] = transpose<4>;
+    p.transpose[BLOCK_8x8] = transpose<8>;
+    p.transpose[BLOCK_16x16] = transpose<16>;
+    p.transpose[BLOCK_32x32] = transpose<32>;
+    p.transpose[BLOCK_64x64] = transpose<64>;
+
+    p.ssd_s[BLOCK_4x4] = pixel_ssd_s_c<4>;
+    p.ssd_s[BLOCK_8x8] = pixel_ssd_s_c<8>;
+    p.ssd_s[BLOCK_16x16] = pixel_ssd_s_c<16>;
+    p.ssd_s[BLOCK_32x32] = pixel_ssd_s_c<32>;
 
     p.weight_pp = weight_pp_c;
     p.weight_sp = weight_sp_c;
 
-    p.pixeladd_ss = pixeladd_ss_c;
-
     p.scale1D_128to64 = scale1D_128to64;
     p.scale2D_64to32 = scale2D_64to32;
     p.frame_init_lowres_core = frame_init_lowres_core;
@@ -1099,5 +1277,8 @@
     p.var[BLOCK_32x32] = pixel_var<32>;
     p.var[BLOCK_64x64] = pixel_var<64>;
     p.plane_copy_deinterleave_c = plane_copy_deinterleave_chroma;
+    p.planecopy_cp = planecopy_cp_c;
+    p.planecopy_sp = planecopy_sp_c;
+    p.propagateCost = estimateCUPropagateCost;
 }
 }

x265-hg20140324.tar.bz2/source/common/primitives.cpp -> x265-hg20140928.tar.bz2/source/common/primitives.cpp Changed

@@ -18,18 +18,17 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "TLibCommon/TComRom.h"
-#include "TLibEncoder/TEncSbac.h"
 #include "primitives.h"
 #include "common.h"
 
 namespace x265 {
 // x265 private namespace
 
-uint8_t lumaPartitioneMapTable[] =
+extern const uint8_t lumaPartitionMapTable[] =
 {
 //  4          8          12          16          20  24          28  32          36  40  44  48          52  56  60  64
     LUMA_4x4,  LUMA_4x8,  255,        LUMA_4x16,  255, 255,        255, 255,        255, 255, 255, 255,        255, 255, 255, 255,        // 4
@@ -67,6 +66,49 @@
     Setup_C_IPredPrimitives(p);      // intrapred.cpp
     Setup_C_LoopFilterPrimitives(p); // loopfilter.cpp
 }
+
+void Setup_Alias_Primitives(EncoderPrimitives &p)
+{
+    /* copy reusable luma primitives to chroma 4:4:4 */
+    for (int i = 0; i < NUM_LUMA_PARTITIONS; i++)
+    {
+        p.chroma[X265_CSP_I444].copy_pp[i] = p.luma_copy_pp[i];
+        p.chroma[X265_CSP_I444].copy_ps[i] = p.luma_copy_ps[i];
+        p.chroma[X265_CSP_I444].copy_sp[i] = p.luma_copy_sp[i];
+        p.chroma[X265_CSP_I444].copy_ss[i] = p.luma_copy_ss[i];
+        p.chroma[X265_CSP_I444].addAvg[i]  = p.luma_addAvg[i];
+    }
+
+    for (int i = 0; i < NUM_SQUARE_BLOCKS; i++)
+    {
+        p.chroma[X265_CSP_I444].add_ps[i]  = p.luma_add_ps[i];
+        p.chroma[X265_CSP_I444].sub_ps[i]  = p.luma_sub_ps[i];
+    }
+
+    for (int i = 0; i < NUM_SQUARE_BLOCKS; i++)
+    {
+        int partL = partitionFromLog2Size(i + 2);
+        p.square_copy_pp[i] = p.luma_copy_pp[partL];
+        p.square_copy_ps[i] = p.luma_copy_ps[partL];
+        p.square_copy_sp[i] = p.luma_copy_sp[partL];
+        p.square_copy_ss[i] = p.luma_copy_ss[partL];
+    }
+
+    primitives.sa8d[BLOCK_4x4]   = primitives.sa8d_inter[LUMA_4x4];
+    primitives.sa8d[BLOCK_8x8]   = primitives.sa8d_inter[LUMA_8x8];
+    primitives.sa8d[BLOCK_16x16] = primitives.sa8d_inter[LUMA_16x16];
+    primitives.sa8d[BLOCK_32x32] = primitives.sa8d_inter[LUMA_32x32];
+    primitives.sa8d[BLOCK_64x64] = primitives.sa8d_inter[LUMA_64x64];
+
+    // SA8D devolves to SATD for blocks not even multiples of 8x8
+    primitives.sa8d_inter[LUMA_4x4]   = primitives.satd[LUMA_4x4];
+    primitives.sa8d_inter[LUMA_4x8]   = primitives.satd[LUMA_4x8];
+    primitives.sa8d_inter[LUMA_4x16]  = primitives.satd[LUMA_4x16];
+    primitives.sa8d_inter[LUMA_8x4]   = primitives.satd[LUMA_8x4];
+    primitives.sa8d_inter[LUMA_16x4]  = primitives.satd[LUMA_16x4];
+    primitives.sa8d_inter[LUMA_16x12] = primitives.satd[LUMA_16x12];
+    primitives.sa8d_inter[LUMA_12x16] = primitives.satd[LUMA_12x16];
+}
 }
 using namespace x265;
 
@@ -90,6 +132,8 @@
         x265_log(param, X265_LOG_WARNING, "Assembly not supported in this binary\n");
 #endif
 
+        Setup_Alias_Primitives(primitives);
+
         initROM();
     }

x265-hg20140324.tar.bz2/source/common/primitives.h -> x265-hg20140928.tar.bz2/source/common/primitives.h Changed

@@ -7,6 +7,7 @@
  *          Mahesh Pittala <mahesh@multicorewareinc.com>
  *          Rajesh Paulraj <rajesh@multicorewareinc.com>
  *          Praveen Kumar Tiwari <praveen@multicorewareinc.com>
+ *          Min Chen <chenm003@163.com>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -23,7 +24,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_PRIMITIVES_H
@@ -32,18 +33,22 @@
 #include "common.h"
 #include "cpu.h"
 
-#define NEW_CALCRECON 1 // TODO: remove recon[] arg
-
 namespace x265 {
 // x265 private namespace
 
 enum LumaPartitions
-{ // Square     Rectangular             Asymmetrical (0.75, 0.25)
-    LUMA_4x4,
-    LUMA_8x8,   LUMA_8x4,   LUMA_4x8,
-    LUMA_16x16, LUMA_16x8,  LUMA_8x16,  LUMA_16x12, LUMA_12x16, LUMA_16x4,  LUMA_4x16,
-    LUMA_32x32, LUMA_32x16, LUMA_16x32, LUMA_32x24, LUMA_24x32, LUMA_32x8,  LUMA_8x32,
-    LUMA_64x64, LUMA_64x32, LUMA_32x64, LUMA_64x48, LUMA_48x64, LUMA_64x16, LUMA_16x64,
+{
+    // Square
+    LUMA_4x4,   LUMA_8x8,   LUMA_16x16, LUMA_32x32, LUMA_64x64,
+    // Rectangular
+    LUMA_8x4,   LUMA_4x8,
+    LUMA_16x8,  LUMA_8x16,  
+    LUMA_32x16, LUMA_16x32,
+    LUMA_64x32, LUMA_32x64,
+    // Asymmetrical (0.75, 0.25)
+    LUMA_16x12, LUMA_12x16, LUMA_16x4,  LUMA_4x16,
+    LUMA_32x24, LUMA_24x32, LUMA_32x8,  LUMA_8x32,
+    LUMA_64x48, LUMA_48x64, LUMA_64x16, LUMA_16x64,
     NUM_LUMA_PARTITIONS
 };
 
@@ -52,14 +57,30 @@
 // be indexed by the luma partition enum
 enum Chroma420Partitions
 {
-    CHROMA_2x2, // never used by HEVC
-    CHROMA_4x4,   CHROMA_4x2,   CHROMA_2x4,
-    CHROMA_8x8,   CHROMA_8x4,   CHROMA_4x8,   CHROMA_8x6,   CHROMA_6x8,   CHROMA_8x2,  CHROMA_2x8,
-    CHROMA_16x16, CHROMA_16x8,  CHROMA_8x16,  CHROMA_16x12, CHROMA_12x16, CHROMA_16x4, CHROMA_4x16,
-    CHROMA_32x32, CHROMA_32x16, CHROMA_16x32, CHROMA_32x24, CHROMA_24x32, CHROMA_32x8, CHROMA_8x32,
+    CHROMA_2x2,   CHROMA_4x4,   CHROMA_8x8,   CHROMA_16x16, CHROMA_32x32,
+    CHROMA_4x2,   CHROMA_2x4,
+    CHROMA_8x4,   CHROMA_4x8,
+    CHROMA_16x8,  CHROMA_8x16,
+    CHROMA_32x16, CHROMA_16x32,
+    CHROMA_8x6,   CHROMA_6x8,   CHROMA_8x2,  CHROMA_2x8,
+    CHROMA_16x12, CHROMA_12x16, CHROMA_16x4, CHROMA_4x16,
+    CHROMA_32x24, CHROMA_24x32, CHROMA_32x8, CHROMA_8x32,
     NUM_CHROMA_PARTITIONS
 };
 
+enum Chroma422Partitions
+{
+    CHROMA422_2x4,   CHROMA422_4x8,   CHROMA422_8x16,  CHROMA422_16x32, CHROMA422_32x64,
+    CHROMA422_4x4,   CHROMA422_2x8,
+    CHROMA422_8x8,   CHROMA422_4x16,
+    CHROMA422_16x16, CHROMA422_8x32,
+    CHROMA422_32x32, CHROMA422_16x64,
+    CHROMA422_8x12,  CHROMA422_6x16,  CHROMA422_8x4,   CHROMA422_2x16,
+    CHROMA422_16x24, CHROMA422_12x32, CHROMA422_16x8,  CHROMA422_4x32,
+    CHROMA422_32x48, CHROMA422_24x64, CHROMA422_32x16, CHROMA422_8x64,
+    NUM_CHROMA_PARTITIONS422
+};
+
 enum SquareBlocks   // Routines can be indexed using log2n(width)-2
 {
     BLOCK_4x4,
@@ -70,6 +91,8 @@
     NUM_SQUARE_BLOCKS
 };
 
+enum { NUM_TR_SIZE = 4 };
+
 // NOTE: Not all DCT functions support dest stride
 enum Dcts
 {
@@ -94,45 +117,56 @@
 // Returns a LumaPartitions enum for the given size, always expected to return a valid enum
 inline int partitionFromSizes(int width, int height)
 {
-    assert(((width | height) & ~(4 | 8 | 16 | 32 | 64)) == 0);
-    extern uint8_t lumaPartitioneMapTable[];
+    X265_CHECK(((width | height) & ~(4 | 8 | 16 | 32 | 64)) == 0, "Invalid block width/height\n");
+    extern const uint8_t lumaPartitionMapTable[];
     int w = (width >> 2) - 1;
     int h = (height >> 2) - 1;
-    int part = (int)lumaPartitioneMapTable[(w << 4) + h];
-    assert(part != 255);
+    int part = (int)lumaPartitionMapTable[(w << 4) + h];
+    X265_CHECK(part != 255, "Invalid block width %d height %d\n", width, height);
     return part;
 }
 
+inline int partitionFromLog2Size(int log2Size)
+{
+    X265_CHECK(2 <= log2Size && log2Size <= 6, "Invalid block size\n");
+    return log2Size - 2;
+}
+
 typedef int  (*pixelcmp_t)(pixel *fenc, intptr_t fencstride, pixel *fref, intptr_t frefstride); // fenc is aligned
 typedef int  (*pixelcmp_ss_t)(int16_t *fenc, intptr_t fencstride, int16_t *fref, intptr_t frefstride);
 typedef int  (*pixelcmp_sp_t)(int16_t *fenc, intptr_t fencstride, pixel *fref, intptr_t frefstride);
+typedef int  (*pixel_ssd_s_t)(int16_t *fenc, intptr_t fencstride);
 typedef void (*pixelcmp_x4_t)(pixel *fenc, pixel *fref0, pixel *fref1, pixel *fref2, pixel *fref3, intptr_t frefstride, int32_t *res);
 typedef void (*pixelcmp_x3_t)(pixel *fenc, pixel *fref0, pixel *fref1, pixel *fref2, intptr_t frefstride, int32_t *res);
-typedef void (*blockcpy_pp_t)(int bx, int by, pixel *dst, intptr_t dstride, pixel *src, intptr_t sstride); // dst is aligned
 typedef void (*blockcpy_sp_t)(int bx, int by, int16_t *dst, intptr_t dstride, pixel *src, intptr_t sstride); // dst is aligned
-typedef void (*blockcpy_ps_t)(int bx, int by, pixel *dst, intptr_t dstride, int16_t *src, intptr_t sstride); // dst is aligned
 typedef void (*blockcpy_sc_t)(int bx, int by, int16_t *dst, intptr_t dstride, uint8_t *src, intptr_t sstride); // dst is aligned
 typedef void (*pixelsub_ps_t)(int bx, int by, int16_t *dst, intptr_t dstride, pixel *src0, pixel *src1, intptr_t sstride0, intptr_t sstride1);
-typedef void (*pixeladd_ss_t)(int bx, int by, int16_t *dst, intptr_t dstride, int16_t *src0, int16_t *src1, intptr_t sstride0, intptr_t sstride1);
 typedef void (*pixelavg_pp_t)(pixel *dst, intptr_t dstride, pixel *src0, intptr_t sstride0, pixel *src1, intptr_t sstride1, int weight);
 typedef void (*blockfill_s_t)(int16_t *dst, intptr_t dstride, int16_t val);
 
-typedef void (*intra_planar_t)(pixel* above, pixel* left, pixel* dst, intptr_t dstStride);
 typedef void (*intra_pred_t)(pixel* dst, intptr_t dstStride, pixel *refLeft, pixel *refAbove, int dirMode, int bFilter);
-typedef void (*intra_allangs_t)(pixel *dst, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma);
+typedef void (*intra_allangs_t)(pixel *dst, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma);
 
 typedef void (*cvt16to32_shl_t)(int32_t *dst, int16_t *src, intptr_t, int, int);
+typedef void (*cvt16to32_shr_t)(int32_t *dst, int16_t *src, intptr_t, int, int);
 typedef void (*cvt32to16_shr_t)(int16_t *dst, int32_t *src, intptr_t, int, int);
+typedef void (*cvt32to16_shl_t)(int16_t *dst, int32_t *src, intptr_t, int);
+typedef uint32_t (*copy_cnt_t)(int16_t* coeff, int16_t* residual, intptr_t stride);
+typedef void (*copy_shr_t)(int16_t *dst, int16_t *src, intptr_t stride, int shift, int size);
+typedef void (*copy_shl_t)(int16_t *dst, int16_t *src, intptr_t stride, int shift);
 
 typedef void (*dct_t)(int16_t *src, int32_t *dst, intptr_t stride);
 typedef void (*idct_t)(int32_t *src, int16_t *dst, intptr_t stride);
+typedef void (*denoiseDct_t)(int32_t* dctCoef, uint32_t* resSum, uint16_t* offset, int numCoeff);
+
 typedef void (*calcresidual_t)(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
-typedef void (*calcrecon_t)(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+typedef void (*calcrecon_t)(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
 typedef void (*transpose_t)(pixel* dst, pixel* src, intptr_t stride);
-typedef uint32_t (*quant_t)(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int32_t *qCoef, int qBits, int add, int numCoeff, int32_t* lastPos);
-typedef void (*dequant_scaling_t)(const int32_t* src, const int32_t *dequantCoef, int32_t* dst, int num, int mcqp_miper, int shift);
-typedef void (*dequant_normal_t)(const int32_t* quantCoef, int32_t* coef, int num, int scale, int shift);
-typedef int  (*count_nonzero_t)(const int32_t *quantCoeff, int numCoeff);
+typedef uint32_t (*quant_t)(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
+typedef uint32_t (*nquant_t)(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
+typedef void (*dequant_scaling_t)(const int16_t* src, const int32_t *dequantCoef, int32_t* dst, int num, int mcqp_miper, int shift);
+typedef void (*dequant_normal_t)(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift);
+typedef int  (*count_nonzero_t)(const int16_t *quantCoeff, int numCoeff);
 
 typedef void (*weightp_pp_t)(pixel *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
 typedef void (*weightp_sp_t)(int16_t *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
@@ -143,7 +177,7 @@
 typedef void (*ssim_4x4x2_core_t)(const pixel *pix1, intptr_t stride1, const pixel *pix2, intptr_t stride2, int sums[2][4]);
 typedef float (*ssim_end4_t)(int sum0[5][4], int sum1[5][4], int width);
 typedef uint64_t (*var_t)(pixel *pix, intptr_t stride);
-typedef void (*plane_copy_deinterleave_t)(pixel *dstu, intptr_t dstuStride, pixel *dstv, intptr_t dstvStride, pixel *src,  intptr_t srcStride, int w, int h);
+typedef void (*plane_copy_deinterleave_t)(pixel *dstu, intptr_t dstuStride, pixel *dstv, intptr_t dstvStride, pixel *src, intptr_t srcStride, int w, int h);
 
 typedef void (*filter_pp_t) (pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx);
 typedef void (*filter_hps_t) (pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx, int isRowExt);
@@ -163,6 +197,10 @@
 typedef void (*addAvg_t)(int16_t* src0, int16_t* src1, pixel* dst, intptr_t src0Stride, intptr_t src1Stride, intptr_t dstStride);
 
 typedef void (*saoCuOrgE0_t)(pixel * rec, int8_t * offsetEo, int lcuWidth, int8_t signLeft);
+typedef void (*planecopy_cp_t) (uint8_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift);
+typedef void (*planecopy_sp_t) (uint16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask);
+
+typedef void (*cutree_propagate_cost) (int *dst, uint16_t *propagateIn, int32_t *intraCosts, uint16_t *interCosts, int32_t *invQscales, double *fpsFactor, int len);
 
 /* Define a structure containing function pointers to optimized encoder
  * primitives.  Each pointer can reference either an assembly routine,
@@ -175,22 +213,31 @@
     pixelcmp_t      sse_pp[NUM_LUMA_PARTITIONS];     // Sum of Square Error (pixel, pixel) fenc alignment not assumed
     pixelcmp_ss_t   sse_ss[NUM_LUMA_PARTITIONS];     // Sum of Square Error (short, short) fenc alignment not assumed
     pixelcmp_sp_t   sse_sp[NUM_LUMA_PARTITIONS];     // Sum of Square Error (short, pixel) fenc alignment not assumed
+    pixel_ssd_s_t   ssd_s[NUM_SQUARE_BLOCKS - 1];    // Sum of Square Error (short) fenc alignment not assumed
     pixelcmp_t      satd[NUM_LUMA_PARTITIONS];       // Sum of Transformed differences (HADAMARD)
     pixelcmp_t      sa8d_inter[NUM_LUMA_PARTITIONS]; // sa8d primitives for motion search partitions
     pixelcmp_t      sa8d[NUM_SQUARE_BLOCKS];         // sa8d primitives for square intra blocks
+    pixelcmp_t      psy_cost[NUM_SQUARE_BLOCKS];     // difference in AC energy between two blocks
 
     blockfill_s_t   blockfill_s[NUM_SQUARE_BLOCKS];  // block fill with value
-    blockcpy_pp_t   blockcpy_pp;                     // block copy pixel from pixel
-    blockcpy_ps_t   blockcpy_ps;                     // block copy pixel from short
     cvt16to32_shl_t cvt16to32_shl;
+    cvt16to32_shr_t cvt16to32_shr[NUM_SQUARE_BLOCKS - 1];
     cvt32to16_shr_t cvt32to16_shr;
+    cvt32to16_shl_t cvt32to16_shl[NUM_SQUARE_BLOCKS - 1];
+    copy_cnt_t      copy_cnt[NUM_SQUARE_BLOCKS - 1];
+    copy_shr_t      copy_shr;
+    copy_shl_t      copy_shl[NUM_SQUARE_BLOCKS - 1];
 
     copy_pp_t       luma_copy_pp[NUM_LUMA_PARTITIONS];
     copy_sp_t       luma_copy_sp[NUM_LUMA_PARTITIONS];
     copy_ps_t       luma_copy_ps[NUM_LUMA_PARTITIONS];
     copy_ss_t       luma_copy_ss[NUM_LUMA_PARTITIONS];
-    pixel_sub_ps_t  luma_sub_ps[NUM_LUMA_PARTITIONS];
-    pixel_add_ps_t  luma_add_ps[NUM_LUMA_PARTITIONS];
+    pixel_sub_ps_t  luma_sub_ps[NUM_SQUARE_BLOCKS];
+    pixel_add_ps_t  luma_add_ps[NUM_SQUARE_BLOCKS];
+    copy_pp_t       square_copy_pp[NUM_SQUARE_BLOCKS];
+    copy_sp_t       square_copy_sp[NUM_SQUARE_BLOCKS];
+    copy_ps_t       square_copy_ps[NUM_SQUARE_BLOCKS];
+    copy_ss_t       square_copy_ss[NUM_SQUARE_BLOCKS];
 
     filter_pp_t     luma_hpp[NUM_LUMA_PARTITIONS];
     filter_hps_t    luma_hps[NUM_LUMA_PARTITIONS];
@@ -204,21 +251,22 @@
 
     weightp_sp_t    weight_sp;
     weightp_pp_t    weight_pp;
-    pixeladd_ss_t   pixeladd_ss;
     pixelavg_pp_t   pixelavg_pp[NUM_LUMA_PARTITIONS];
     addAvg_t        luma_addAvg[NUM_LUMA_PARTITIONS];
 
-    intra_pred_t    intra_pred[NUM_SQUARE_BLOCKS - 1][NUM_INTRA_MODE];
-    intra_allangs_t intra_pred_allangs[NUM_SQUARE_BLOCKS - 1];
+    intra_pred_t    intra_pred[NUM_INTRA_MODE][NUM_TR_SIZE];
+    intra_allangs_t intra_pred_allangs[NUM_TR_SIZE];
     scale_t         scale1D_128to64;
     scale_t         scale2D_64to32;
 
     dct_t           dct[NUM_DCTS];
     idct_t          idct[NUM_IDCTS];
     quant_t         quant;
+    nquant_t        nquant;
     dequant_scaling_t dequant_scaling;
     dequant_normal_t dequant_normal;
     count_nonzero_t count_nonzero;
+    denoiseDct_t    denoiseDct;
 
     calcresidual_t  calcresidual[NUM_SQUARE_BLOCKS];
     calcrecon_t     calcrecon[NUM_SQUARE_BLOCKS];
@@ -233,6 +281,10 @@
     extendCURowBorder_t extendRowBorder;
     // sao primitives
     saoCuOrgE0_t      saoCuOrgE0;
+    planecopy_cp_t    planecopy_cp;
+    planecopy_sp_t    planecopy_sp;
+
+    cutree_propagate_cost    propagateCost;
 
     struct
     {
@@ -242,13 +294,13 @@
         filter_ss_t     filter_vss[NUM_LUMA_PARTITIONS];
         filter_pp_t     filter_hpp[NUM_LUMA_PARTITIONS];
         filter_hps_t    filter_hps[NUM_LUMA_PARTITIONS];
+        addAvg_t        addAvg[NUM_LUMA_PARTITIONS];
         copy_pp_t       copy_pp[NUM_LUMA_PARTITIONS];
         copy_sp_t       copy_sp[NUM_LUMA_PARTITIONS];
         copy_ps_t       copy_ps[NUM_LUMA_PARTITIONS];
         copy_ss_t       copy_ss[NUM_LUMA_PARTITIONS];
-        pixel_sub_ps_t  sub_ps[NUM_LUMA_PARTITIONS];
-        pixel_add_ps_t  add_ps[NUM_LUMA_PARTITIONS];
-        addAvg_t        addAvg[NUM_LUMA_PARTITIONS];
+        pixel_sub_ps_t  sub_ps[NUM_SQUARE_BLOCKS];
+        pixel_add_ps_t  add_ps[NUM_SQUARE_BLOCKS];
     } chroma[4]; // X265_CSP_COUNT - do not want to include x265.h here
 };
 
@@ -261,6 +313,7 @@
 void Setup_C_Primitives(EncoderPrimitives &p);
 void Setup_Instrinsic_Primitives(EncoderPrimitives &p, int cpuMask);
 void Setup_Assembly_Primitives(EncoderPrimitives &p, int cpuMask);
+void Setup_Alias_Primitives(EncoderPrimitives &p);
 }
 
 #endif // ifndef X265_PRIMITIVES_H

x265-hg20140928.tar.bz2/source/common/quant.cpp Added

@@ -0,0 +1,1111 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "common.h"
+#include "primitives.h"
+#include "quant.h"
+#include "frame.h"
+#include "entropy.h"
+#include "TLibCommon/TComDataCU.h"
+#include "TLibCommon/TComYuv.h"
+#include "TLibCommon/ContextTables.h"
+
+using namespace x265;
+
+#define SIGN(x,y) ((x^(y >> 31))-(y >> 31))
+
+namespace {
+
+struct coeffGroupRDStats
+{
+    int     nnzBeforePos0;     /* indicates coeff other than pos 0 are coded */
+    int64_t codedLevelAndDist; /* distortion and level cost of coded coefficients */
+    int64_t uncodedDist;       /* uncoded distortion cost of coded coefficients */ 
+    int64_t sigCost;           /* cost of signaling significant coeff bitmap */
+    int64_t sigCost0;          /* cost of signaling sig coeff bit of coeff 0 */
+};
+
+inline int fastMin(int x, int y)
+{
+    return y + ((x - y) & ((x - y) >> (sizeof(int) * CHAR_BIT - 1))); // min(x, y)
+}
+
+inline int getICRate(uint32_t absLevel, int32_t diffLevel, const int *greaterOneBits, const int *levelAbsBits, uint32_t absGoRice, uint32_t c1c2Idx)
+{
+    X265_CHECK(c1c2Idx <= 3, "c1c2Idx check failure\n");
+    X265_CHECK(absGoRice <= 4, "absGoRice check failure\n");
+    if (!absLevel)
+    {
+        X265_CHECK(diffLevel < 0, "diffLevel check failure\n");
+        return 0;
+    }
+    int rate = 0;
+
+    if (diffLevel < 0)
+    {
+        X265_CHECK(absLevel <= 2, "absLevel check failure\n");
+        rate += greaterOneBits[(absLevel == 2)];
+
+        if (absLevel == 2)
+            rate += levelAbsBits[0];
+    }
+    else
+    {
+        uint32_t symbol = diffLevel;
+        const uint32_t maxVlc = g_goRiceRange[absGoRice];
+        bool expGolomb = (symbol > maxVlc);
+
+        if (expGolomb)
+        {
+            absLevel = symbol - maxVlc;
+
+            // NOTE: mapping to x86 hardware instruction BSR
+            unsigned long size;
+            CLZ32(size, absLevel);
+            int egs = size * 2 + 1;
+
+            rate += egs << 15;
+
+            // NOTE: in here, expGolomb=true means (symbol >= maxVlc + 1)
+            X265_CHECK(fastMin(symbol, (maxVlc + 1)) == (int)maxVlc + 1, "min check failure\n");
+            symbol = maxVlc + 1;
+        }
+
+        uint32_t prefLen = (symbol >> absGoRice) + 1;
+        uint32_t numBins = fastMin(prefLen + absGoRice, 8 /* g_goRicePrefixLen[absGoRice] + absGoRice */);
+
+        rate += numBins << 15;
+
+        if (c1c2Idx & 1)
+            rate += greaterOneBits[1];
+
+        if (c1c2Idx == 3)
+            rate += levelAbsBits[1];
+    }
+    return rate;
+}
+
+/* Calculates the cost for specific absolute transform level */
+inline uint32_t getICRateCost(uint32_t absLevel, int32_t diffLevel, const int *greaterOneBits, const int *levelAbsBits, uint32_t absGoRice, uint32_t c1c2Idx)
+{
+    X265_CHECK(absLevel, "absLevel should not be zero\n");
+
+    if (diffLevel < 0)
+    {
+        X265_CHECK((absLevel == 1) || (absLevel == 2), "absLevel range check failure\n");
+
+        uint32_t rate = greaterOneBits[(absLevel == 2)];
+        if (absLevel == 2)
+            rate += levelAbsBits[0];
+        return rate;
+    }
+    else
+    {
+        uint32_t rate;
+        uint32_t symbol = diffLevel;
+        if ((symbol >> absGoRice) < COEF_REMAIN_BIN_REDUCTION)
+        {
+            uint32_t length = symbol >> absGoRice;
+            rate = (length + 1 + absGoRice) << 15;
+        }
+        else
+        {
+            uint32_t length = 0;
+            symbol = (symbol >> absGoRice) - COEF_REMAIN_BIN_REDUCTION;
+            if (symbol)
+            {
+                unsigned long idx;
+                CLZ32(idx, symbol + 1);
+                length = idx;
+            }
+
+            rate = (COEF_REMAIN_BIN_REDUCTION + length + absGoRice + 1 + length) << 15;
+        }
+        if (c1c2Idx & 1)
+            rate += greaterOneBits[1];
+        if (c1c2Idx == 3)
+            rate += levelAbsBits[1];
+        return rate;
+    }
+}
+
+}
+
+Quant::Quant()
+{
+    m_resiDctCoeff = NULL;
+    m_fencDctCoeff = NULL;
+    m_fencShortBuf = NULL;
+}
+
+bool Quant::init(bool useRDOQ, double psyScale, const ScalingList& scalingList)
+{
+    m_useRDOQ = useRDOQ;
+    m_psyRdoqScale = (int64_t)(psyScale * 256.0);
+    m_scalingList = &scalingList;
+    m_resiDctCoeff = X265_MALLOC(int32_t, MAX_TR_SIZE * MAX_TR_SIZE * 2);
+    m_fencDctCoeff = m_resiDctCoeff + (MAX_TR_SIZE * MAX_TR_SIZE);
+    m_fencShortBuf = X265_MALLOC(int16_t, MAX_TR_SIZE * MAX_TR_SIZE);
+    
+    return m_resiDctCoeff && m_fencShortBuf;
+}
+
+Quant::~Quant()
+{
+    X265_FREE(m_resiDctCoeff);
+    X265_FREE(m_fencShortBuf);
+}
+
+void Quant::setQPforQuant(TComDataCU* cu)
+{
+    int qpy = cu->getQP(0);
+    int chFmt = cu->getChromaFormat();
+
+    m_qpParam[TEXT_LUMA].setQpParam(qpy + QP_BD_OFFSET);
+    setChromaQP(qpy + cu->m_slice->m_pps->chromaCbQpOffset, TEXT_CHROMA_U, chFmt);
+    setChromaQP(qpy + cu->m_slice->m_pps->chromaCrQpOffset, TEXT_CHROMA_V, chFmt);
+}
+
+void Quant::setChromaQP(int qpin, TextType ttype, int chFmt)
+{
+    int qp = Clip3(-QP_BD_OFFSET, 57, qpin);
+    if (qp >= 30)
+    {
+        if (chFmt == X265_CSP_I420)
+            qp = g_chromaScale[qp];
+        else
+            qp = X265_MIN(qp, 51);
+    }
+    m_qpParam[ttype].setQpParam(qp + QP_BD_OFFSET);
+}
+
+/* To minimize the distortion only. No rate is considered */
+uint32_t Quant::signBitHidingHDQ(int16_t* coeff, int32_t* deltaU, uint32_t numSig, const TUEntropyCodingParameters &codeParams)
+{
+    const uint32_t log2TrSizeCG = codeParams.log2TrSizeCG;
+    const uint16_t *scan = codeParams.scan;
+    bool lastCG = true;
+
+    for (int cg = (1 << (log2TrSizeCG * 2)) - 1; cg >= 0; cg--)
+    {
+        int cgStartPos = cg << LOG2_SCAN_SET_SIZE;
+        int n;
+
+        for (n = SCAN_SET_SIZE - 1; n >= 0; --n)
+            if (coeff[scan[n + cgStartPos]])
+                break;
+        if (n < 0)
+            continue;
+
+        int lastNZPosInCG = n;
+
+        for (n = 0;; n++)
+            if (coeff[scan[n + cgStartPos]])
+                break;
+
+        int firstNZPosInCG = n;
+
+        if (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD)
+        {
+            uint32_t signbit = coeff[scan[cgStartPos + firstNZPosInCG]] > 0 ? 0 : 1;
+            uint32_t absSum = 0;
+
+            for (n = firstNZPosInCG; n <= lastNZPosInCG; n++)
+                absSum += coeff[scan[n + cgStartPos]];
+
+            if (signbit != (absSum & 0x1)) // compare signbit with sum_parity
+            {
+                int minCostInc = MAX_INT,  minPos = -1, curCost = MAX_INT;
+                int16_t finalChange = 0, curChange = 0;
+
+                for (n = (lastCG ? lastNZPosInCG : SCAN_SET_SIZE - 1); n >= 0; --n)
+                {
+                    uint32_t blkPos = scan[n + cgStartPos];
+                    if (coeff[blkPos])
+                    {
+                        if (deltaU[blkPos] > 0)
+                        {
+                            curCost = -deltaU[blkPos];
+                            curChange = 1;
+                        }
+                        else
+                        {
+                            if (n == firstNZPosInCG && abs(coeff[blkPos]) == 1)
+                                curCost = MAX_INT;
+                            else
+                            {
+                                curCost = deltaU[blkPos];
+                                curChange = -1;
+                            }
+                        }
+                    }
+                    else
+                    {
+                        if (n < firstNZPosInCG)
+                        {
+                            uint32_t thisSignBit = m_resiDctCoeff[blkPos] >= 0 ? 0 : 1;
+                            if (thisSignBit != signbit)
+                                curCost = MAX_INT;
+                            else
+                            {
+                                curCost = -deltaU[blkPos];
+                                curChange = 1;
+                            }
+                        }
+                        else
+                        {
+                            curCost = -deltaU[blkPos];
+                            curChange = 1;
+                        }
+                    }
+
+                    if (curCost < minCostInc)
+                    {
+                        minCostInc = curCost;
+                        finalChange = curChange;
+                        minPos = blkPos;
+                    }
+                }
+
+                /* do not allow change to violate coeff clamp */
+                if (coeff[minPos] == 32767 || coeff[minPos] == -32768)
+                    finalChange = -1;
+
+                if (!coeff[minPos])
+                    numSig++;
+                else if (finalChange == -1 && abs(coeff[minPos]) == 1)
+                    numSig--;
+
+                if (m_resiDctCoeff[minPos] >= 0)
+                    coeff[minPos] += finalChange;
+                else
+                    coeff[minPos] -= finalChange;
+            }
+        }
+
+        lastCG = false;
+    }
+
+    return numSig;
+}
+
+uint32_t Quant::transformNxN(TComDataCU* cu, pixel* fenc, uint32_t fencStride, int16_t* residual, uint32_t stride,
+                             coeff_t* coeff, uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx, bool useTransformSkip)
+{
+    if (cu->getCUTransquantBypass(absPartIdx))
+    {
+        X265_CHECK(log2TrSize >= 2 && log2TrSize <= 5, "Block size mistake!\n");
+        return primitives.copy_cnt[log2TrSize - 2](coeff, residual, stride);
+    }
+
+    bool isLuma  = ttype == TEXT_LUMA;
+    bool usePsy  = m_psyRdoqScale && isLuma && !useTransformSkip;
+    bool isIntra = cu->getPredictionMode(absPartIdx) == MODE_INTRA;
+    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize; // Represents scaling through forward transform
+    int trSize = 1 << log2TrSize;
+
+    X265_CHECK((cu->m_slice->m_sps->quadtreeTULog2MaxSize >= log2TrSize), "transform size too large\n");
+    if (useTransformSkip)
+    {
+#if X265_DEPTH <= 10
+        primitives.cvt16to32_shl(m_resiDctCoeff, residual, stride, transformShift, trSize);
+#else
+        if (transformShift >= 0)
+            primitives.cvt16to32_shl(m_resiDctCoeff, residual, stride, transformShift, trSize);
+        else
+        {
+            int shift = -transformShift;
+            int offset = (1 << (shift - 1));
+            primitives.cvt16to32_shr[log2TrSize - 2](m_resiDctCoeff, residual, stride, shift, offset);
+        }
+#endif
+    }
+    else
+    {
+        const uint32_t sizeIdx = log2TrSize - 2;
+        int useDST = !sizeIdx && isLuma && isIntra;
+        int index = DCT_4x4 + sizeIdx - useDST;
+
+        primitives.dct[index](residual, m_resiDctCoeff, stride);
+
+        /* NOTE: if RDOQ is disabled globally, psy-rdoq is also disabled, so
+         * there is no risk of performing this DCT unnecessarily */
+        if (usePsy)
+        {
+            /* perform DCT on source pixels for psy-rdoq */
+            primitives.square_copy_ps[sizeIdx](m_fencShortBuf, trSize, fenc, fencStride);
+            primitives.dct[index](m_fencShortBuf, m_fencDctCoeff, trSize);
+        }
+
+        if (m_nr && !isIntra)
+        {
+            /* denoise is not applied to intra residual, so DST can be ignored */
+            int cat = sizeIdx + 4 * !isLuma;
+            int numCoeff = 1 << (log2TrSize * 2);
+            primitives.denoiseDct(m_resiDctCoeff, m_nr->residualSum[cat], m_nr->offsetDenoise[cat], numCoeff);
+            m_nr->count[cat]++;
+        }
+    }
+
+    if (m_useRDOQ)
+        return rdoQuant(cu, coeff, log2TrSize, ttype, absPartIdx, usePsy);
+    else
+    {
+        int deltaU[32 * 32];
+
+        int scalingListType = ttype + (isLuma ? 3 : 0);
+        int rem = m_qpParam[ttype].rem;
+        int per = m_qpParam[ttype].per;
+        int32_t *quantCoeff = m_scalingList->m_quantCoef[log2TrSize - 2][scalingListType][rem];
+
+        int qbits = QUANT_SHIFT + per + transformShift;
+        int add = (cu->m_slice->m_sliceType == I_SLICE ? 171 : 85) << (qbits - 9);
+        int numCoeff = 1 << (log2TrSize * 2);
+
+        uint32_t numSig = primitives.quant(m_resiDctCoeff, quantCoeff, deltaU, coeff, qbits, add, numCoeff);
+
+        if (numSig >= 2 && cu->m_slice->m_pps->bSignHideEnabled)
+        {
+            TUEntropyCodingParameters codeParams;
+            cu->getTUEntropyCodingParameters(codeParams, absPartIdx, log2TrSize, isLuma);
+            return signBitHidingHDQ(coeff, deltaU, numSig, codeParams);
+        }
+        else
+            return numSig;
+    }
+}
+
+void Quant::invtransformNxN(bool transQuantBypass, int16_t* residual, uint32_t stride, coeff_t* coeff,
+                            uint32_t log2TrSize, TextType ttype, bool bIntra, bool useTransformSkip, uint32_t numSig)
+{
+    if (transQuantBypass)
+    {
+        primitives.copy_shl[log2TrSize - 2](residual, coeff, stride, 0);
+        return;
+    }
+
+    // Values need to pass as input parameter in dequant
+    int rem = m_qpParam[ttype].rem;
+    int per = m_qpParam[ttype].per;
+    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
+    int shift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - transformShift;
+    int numCoeff = 1 << (log2TrSize * 2);
+
+    if (m_scalingList->m_bEnabled)
+    {
+        int scalingListType = (bIntra ? 0 : 3) + ttype;
+        int32_t *dequantCoef = m_scalingList->m_dequantCoef[log2TrSize - 2][scalingListType][rem];
+        primitives.dequant_scaling(coeff, dequantCoef, m_resiDctCoeff, numCoeff, per, shift);
+    }
+    else
+    {
+        int scale = m_scalingList->s_invQuantScales[rem] << per;
+        primitives.dequant_normal(coeff, m_resiDctCoeff, numCoeff, scale, shift);
+    }
+
+    if (useTransformSkip)
+    {
+        int trSize = 1 << log2TrSize;
+
+#if X265_DEPTH <= 10
+        primitives.cvt32to16_shr(residual, m_resiDctCoeff, stride, transformShift, trSize);
+#else
+        if (transformShift > 0)
+            primitives.cvt32to16_shr(residual, m_resiDctCoeff, stride, transformShift, trSize);
+        else
+            primitives.cvt32to16_shl[log2TrSize - 2](residual, m_resiDctCoeff, stride, -transformShift);
+#endif
+    }
+    else
+    {
+        const uint32_t sizeIdx = log2TrSize - 2;
+        int useDST = !sizeIdx && ttype == TEXT_LUMA && bIntra;
+
+        X265_CHECK((int)numSig == primitives.count_nonzero(coeff, 1 << (log2TrSize * 2)), "numSig differ\n");
+
+        // DC only
+        if (numSig == 1 && coeff[0] != 0 && !useDST)
+        {
+            const int shift_1st = 7;
+            const int add_1st = 1 << (shift_1st - 1);
+            const int shift_2nd = 12 - (X265_DEPTH - 8);
+            const int add_2nd = 1 << (shift_2nd - 1);
+
+            int dc_val = (((m_resiDctCoeff[0] * 64 + add_1st) >> shift_1st) * 64 + add_2nd) >> shift_2nd;
+            primitives.blockfill_s[sizeIdx](residual, stride, (int16_t)dc_val);
+            return;
+        }
+
+        primitives.idct[IDCT_4x4 + sizeIdx - useDST](m_resiDctCoeff, residual, stride);
+    }
+}
+
+/* Rate distortion optimized quantization for entropy coding engines using
+ * probability models like CABAC */
+uint32_t Quant::rdoQuant(TComDataCU* cu, int16_t* dstCoeff, uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx, bool usePsy)
+{
+    int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize; /* Represents scaling through forward transform */
+    int scalingListType = (cu->isIntra(absPartIdx) ? 0 : 3) + ttype;
+
+    X265_CHECK(scalingListType < 6, "scaling list type out of range\n");
+
+    int rem = m_qpParam[ttype].rem;
+    int per = m_qpParam[ttype].per;
+    int qbits = QUANT_SHIFT + per + transformShift; /* Right shift of non-RDOQ quantizer level = (coeff*Q + offset)>>q_bits */
+    int add = (1 << (qbits - 1));
+    int32_t *qCoef = m_scalingList->m_quantCoef[log2TrSize - 2][scalingListType][rem];
+
+    int numCoeff = 1 << (log2TrSize * 2);
+
+    uint32_t numSig = primitives.nquant(m_resiDctCoeff, qCoef, dstCoeff, qbits, add, numCoeff);
+
+    X265_CHECK((int)numSig == primitives.count_nonzero(dstCoeff, 1 << (log2TrSize * 2)), "numSig differ\n");
+    if (!numSig)
+        return 0;
+
+    uint32_t trSize = 1 << log2TrSize;
+    int64_t lambda2 = m_qpParam[ttype].lambda2;
+    int64_t psyScale = (m_psyRdoqScale * m_qpParam[ttype].lambda);
+
+    /* unquant constants for measuring distortion. Scaling list quant coefficients have a (1 << 4)
+     * scale applied that must be removed during unquant. Note that in real dequant there is clipping
+     * at several stages. We skip the clipping for simplicity when measuring RD cost */
+    int32_t *unquantScale = m_scalingList->m_dequantCoef[log2TrSize - 2][scalingListType][rem];
+    int unquantShift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - transformShift + (m_scalingList->m_bEnabled ? 4 : 0);
+    int unquantRound = (unquantShift > per) ? 1 << (unquantShift - per - 1) : 0;
+    int scaleBits = SCALE_BITS - 2 * transformShift;
+
+#define UNQUANT(lvl)    (((lvl) * (unquantScale[blkPos] << per) + unquantRound) >> unquantShift)
+#define SIGCOST(bits)   ((lambda2 * (bits)) >> 8)
+#define RDCOST(d, bits) ((((int64_t)d * d) << scaleBits) + SIGCOST(bits))
+#define PSYVALUE(rec)   ((psyScale * (rec)) >> (16 - scaleBits))
+
+    int64_t costCoeff[32 * 32];   /* d*d + lambda * bits */
+    int64_t costUncoded[32 * 32]; /* d*d + lambda * 0    */
+    int64_t costSig[32 * 32];     /* lambda * bits       */
+
+    int rateIncUp[32 * 32];      /* signal overhead of increasing level */
+    int rateIncDown[32 * 32];    /* signal overhead of decreasing level */
+    int sigRateDelta[32 * 32];   /* signal difference between zero and non-zero */
+
+    int64_t costCoeffGroupSig[MLS_GRP_NUM]; /* lambda * bits of group coding cost */
+    uint64_t sigCoeffGroupFlag64 = 0;
+
+    uint32_t ctxSet      = 0;
+    int    c1            = 1;
+    int    c2            = 0;
+    uint32_t goRiceParam = 0;
+    uint32_t c1Idx       = 0;
+    uint32_t c2Idx       = 0;
+    int cgLastScanPos    = -1;
+    int lastScanPos      = -1;
+    const uint32_t cgSize = (1 << MLS_CG_SIZE); /* 4x4 num coef = 16 */
+    bool bIsLuma = ttype == TEXT_LUMA;
+
+    /* total rate distortion cost of transform block, as CBF=0 */
+    int64_t totalUncodedCost = 0;
+
+    /* Total rate distortion cost of this transform block, counting te distortion of uncoded blocks,
+     * the distortion and signal cost of coded blocks, and the coding cost of significant
+     * coefficient and coefficient group bitmaps */
+    int64_t totalRdCost = 0;
+
+    TUEntropyCodingParameters codeParams;
+    cu->getTUEntropyCodingParameters(codeParams, absPartIdx, log2TrSize, bIsLuma);
+    const uint32_t cgNum = 1 << (codeParams.log2TrSizeCG * 2);
+
+    /* TODO: update bit estimates if dirty */
+    EstBitsSbac& estBitsSbac = m_entropyCoder->m_estBitsSbac;
+
+    uint32_t scanPos;
+    coeffGroupRDStats cgRdStats;
+
+    /* iterate over coding groups in reverse scan order */
+    for (int cgScanPos = cgNum - 1; cgScanPos >= 0; cgScanPos--)
+    {
+        const uint32_t cgBlkPos = codeParams.scanCG[cgScanPos];
+        const uint32_t cgPosY   = cgBlkPos >> codeParams.log2TrSizeCG;
+        const uint32_t cgPosX   = cgBlkPos - (cgPosY << codeParams.log2TrSizeCG);
+        const uint64_t cgBlkPosMask = ((uint64_t)1 << cgBlkPos);
+        memset(&cgRdStats, 0, sizeof(coeffGroupRDStats));
+
+        const int patternSigCtx = calcPatternSigCtx(sigCoeffGroupFlag64, cgPosX, cgPosY, codeParams.log2TrSizeCG);
+
+        /* iterate over coefficients in each group in reverse scan order */
+        for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
+        {
+            scanPos              = (cgScanPos << MLS_CG_SIZE) + scanPosinCG;
+            uint32_t blkPos      = codeParams.scan[scanPos];
+            uint16_t maxAbsLevel = (int16_t)abs(dstCoeff[blkPos]);             /* abs(quantized coeff) */
+            int signCoef         = m_resiDctCoeff[blkPos];            /* pre-quantization DCT coeff */
+            int predictedCoef    = m_fencDctCoeff[blkPos] - signCoef; /* predicted DCT = source DCT - residual DCT*/
+
+            /* RDOQ measures distortion as the squared difference between the unquantized coded level
+             * and the original DCT coefficient. The result is shifted scaleBits to account for the
+             * FIX15 nature of the CABAC cost tables minus the forward transform scale */
+
+            /* cost of not coding this coefficient (all distortion, no signal bits) */
+            costUncoded[scanPos] = (int64_t)(signCoef * signCoef) << scaleBits;
+            if (usePsy && blkPos)
+                /* when no residual coefficient is coded, predicted coef == recon coef */
+                costUncoded[scanPos] -= PSYVALUE(predictedCoef);
+
+            totalUncodedCost += costUncoded[scanPos];
+
+            if (maxAbsLevel && lastScanPos < 0)
+            {
+                /* remember the first non-zero coef found in this reverse scan as the last pos */
+                lastScanPos   = scanPos;
+                ctxSet        = (scanPos < SCAN_SET_SIZE || !bIsLuma) ? 0 : 2;
+                cgLastScanPos = cgScanPos;
+            }
+
+            if (lastScanPos < 0)
+            {
+                /* coefficients after lastNZ have no distortion signal cost */
+                costCoeff[scanPos] = 0;
+                costSig[scanPos] = 0;
+
+                /* No non-zero coefficient yet found, but this does not mean
+                 * there is no uncoded-cost for this coefficient. Pre-
+                 * quantization the coefficient may have been non-zero */
+                totalRdCost += costUncoded[scanPos];
+            }
+            else
+            {
+                const uint32_t c1c2Idx = ((c1Idx - 8) >> (sizeof(int) * CHAR_BIT - 1)) + (((-(int)c2Idx) >> (sizeof(int) * CHAR_BIT - 1)) + 1) * 2;
+                const uint32_t baseLevel = ((uint32_t)0xD9 >> (c1c2Idx * 2)) & 3;  // {1, 2, 1, 3}
+
+                X265_CHECK(!!((int)c1Idx < C1FLAG_NUMBER) == (int)((c1Idx - 8) >> (sizeof(int) * CHAR_BIT - 1)), "scan validation 1\n");
+                X265_CHECK(!!(c2Idx == 0) == ((-(int)c2Idx) >> (sizeof(int) * CHAR_BIT - 1)) + 1, "scan validation 2\n");
+                X265_CHECK((int)baseLevel == ((c1Idx < C1FLAG_NUMBER) ? (2 + (c2Idx == 0)) : 1), "scan validation 3\n");
+
+                // coefficient level estimation
+                const uint32_t oneCtx = 4 * ctxSet + c1;
+                const uint32_t absCtx = ctxSet + c2;
+                const int *greaterOneBits = estBitsSbac.greaterOneBits[oneCtx];
+                const int *levelAbsBits = estBitsSbac.levelAbsBits[absCtx];
+
+                uint16_t level = 0;
+                uint32_t sigCoefBits = 0;
+                costCoeff[scanPos] = MAX_INT64;
+
+                if ((int)scanPos == lastScanPos)
+                    sigRateDelta[blkPos] = 0;
+                else
+                {
+                    const uint32_t ctxSig = getSigCtxInc(patternSigCtx, log2TrSize, trSize, blkPos, bIsLuma, codeParams.firstSignificanceMapContext);
+                    if (maxAbsLevel < 3)
+                    {
+                        /* set default costs to uncoded costs */
+                        costSig[scanPos] = SIGCOST(estBitsSbac.significantBits[ctxSig][0]);
+                        costCoeff[scanPos] = costUncoded[scanPos] + costSig[scanPos];
+                    }
+                    sigRateDelta[blkPos] = estBitsSbac.significantBits[ctxSig][1] - estBitsSbac.significantBits[ctxSig][0];
+                    sigCoefBits = estBitsSbac.significantBits[ctxSig][1];
+                }
+                if (maxAbsLevel)
+                {
+                    uint16_t minAbsLevel = X265_MAX(maxAbsLevel - 1, 1);
+                    for (uint16_t lvl = maxAbsLevel; lvl >= minAbsLevel; lvl--)
+                    {
+                        uint32_t levelBits = getICRateCost(lvl, lvl - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx) + IEP_RATE;
+
+                        int unquantAbsLevel = UNQUANT(lvl);
+                        int d = abs(signCoef) - unquantAbsLevel;
+                        int64_t curCost = RDCOST(d, sigCoefBits + levelBits);
+
+                        /* Psy RDOQ: bias in favor of higher AC coefficients in the reconstructed frame */
+                        if (usePsy && blkPos)
+                        {
+                            int reconCoef = abs(unquantAbsLevel + SIGN(predictedCoef, signCoef));
+                            curCost -= PSYVALUE(reconCoef);
+                        }
+
+                        if (curCost < costCoeff[scanPos])
+                        {
+                            level = lvl;
+                            costCoeff[scanPos] = curCost;
+                            costSig[scanPos] = SIGCOST(sigCoefBits);
+                        }
+                    }
+                }
+
+                dstCoeff[blkPos] = level;
+                totalRdCost += costCoeff[scanPos];
+
+                /* record costs for sign-hiding performed at the end */
+                if (level)
+                {
+                    int rateNow = getICRate(level, level - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx);
+                    rateIncUp[blkPos] = getICRate(level + 1, level + 1 - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx) - rateNow;
+                    rateIncDown[blkPos] = getICRate(level - 1, level - 1 - baseLevel, greaterOneBits, levelAbsBits, goRiceParam, c1c2Idx) - rateNow;
+                }
+                else
+                {
+                    rateIncUp[blkPos] = greaterOneBits[0];
+                    rateIncDown[blkPos] = 0;
+                }
+
+                /* Update CABAC estimation state */
+                if (level >= baseLevel && goRiceParam < 4 && level > (3U << goRiceParam))
+                    goRiceParam++;
+
+                c1Idx -= (-(int32_t)level) >> 31;
+
+                /* update bin model */
+                if (level > 1)
+                {
+                    c1 = 0;
+                    c2 += (uint32_t)(c2 - 2) >> 31;
+                    c2Idx++;
+                }
+                else if ((c1 < 3) && (c1 > 0) && level)
+                    c1++;
+
+                /* context set update */
+                if (!(scanPos % SCAN_SET_SIZE) && scanPos)
+                {
+                    c2 = 0;
+                    goRiceParam = 0;
+
+                    c1Idx = 0;
+                    c2Idx = 0;
+                    ctxSet = (scanPos == SCAN_SET_SIZE || !bIsLuma) ? 0 : 2;
+                    X265_CHECK(c1 >= 0, "c1 is negative\n");
+                    ctxSet -= ((int32_t)(c1 - 1) >> 31);
+                    c1 = 1;
+                }
+            }
+
+            cgRdStats.sigCost += costSig[scanPos];
+            if (!scanPosinCG)
+                cgRdStats.sigCost0 = costSig[scanPos];
+
+            if (dstCoeff[blkPos])
+            {
+                sigCoeffGroupFlag64 |= cgBlkPosMask;
+                cgRdStats.codedLevelAndDist += costCoeff[scanPos] - costSig[scanPos];
+                cgRdStats.uncodedDist += costUncoded[scanPos];
+                cgRdStats.nnzBeforePos0 += scanPosinCG;
+            }
+        } /* end for (scanPosinCG) */
+
+        costCoeffGroupSig[cgScanPos] = 0;
+
+        if (cgLastScanPos < 0)
+        {
+            /* nothing to do at this point */
+        }
+        else if (!cgScanPos || cgScanPos == cgLastScanPos)
+        {
+            /* coeff group 0 is implied to be present, no signal cost */
+            /* coeff group with last NZ is implied to be present, handled below */
+        }
+        else if (sigCoeffGroupFlag64 & cgBlkPosMask)
+        {
+            if (!cgRdStats.nnzBeforePos0)
+            {
+                /* if only coeff 0 in this CG is coded, its significant coeff bit is implied */
+                totalRdCost -= cgRdStats.sigCost0;
+                cgRdStats.sigCost -= cgRdStats.sigCost0;
+            }
+
+            /* there are coded coefficients in this group, but now we include the signaling cost
+             * of the significant coefficient group flag and evaluate whether the RD cost of the
+             * coded group is more than the RD cost of the uncoded group */
+
+            uint32_t sigCtx = getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codeParams.log2TrSizeCG);
+
+            int64_t costZeroCG = totalRdCost + SIGCOST(estBitsSbac.significantCoeffGroupBits[sigCtx][0]);
+            costZeroCG += cgRdStats.uncodedDist;       /* add distortion for resetting non-zero levels to zero levels */
+            costZeroCG -= cgRdStats.codedLevelAndDist; /* remove distortion and level cost of coded coefficients */
+            costZeroCG -= cgRdStats.sigCost;           /* remove signaling cost of significant coeff bitmap */
+
+            costCoeffGroupSig[cgScanPos] = SIGCOST(estBitsSbac.significantCoeffGroupBits[sigCtx][1]);
+            totalRdCost += costCoeffGroupSig[cgScanPos];  /* add the cost of 1 bit in significant CG bitmap */
+
+            if (costZeroCG < totalRdCost)
+            {
+                sigCoeffGroupFlag64 &= ~cgBlkPosMask;
+                totalRdCost = costZeroCG;
+                costCoeffGroupSig[cgScanPos] = SIGCOST(estBitsSbac.significantCoeffGroupBits[sigCtx][0]);
+
+                /* reset all coeffs to 0. UNCODE THIS COEFF GROUP! */
+                for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
+                {
+                    scanPos = cgScanPos * cgSize + scanPosinCG;
+                    uint32_t blkPos = codeParams.scan[scanPos];
+                    if (dstCoeff[blkPos])
+                    {
+                        costCoeff[scanPos] = costUncoded[scanPos];
+                        costSig[scanPos] = 0;
+                    }
+                    dstCoeff[blkPos] = 0;
+                }
+            }
+        }
+        else
+        {
+            /* there were no coded coefficients in this coefficient group */
+            uint32_t ctxSig = getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codeParams.log2TrSizeCG);
+            costCoeffGroupSig[cgScanPos] = SIGCOST(estBitsSbac.significantCoeffGroupBits[ctxSig][0]);
+            totalRdCost += costCoeffGroupSig[cgScanPos];  /* add cost of 0 bit in significant CG bitmap */
+            totalRdCost -= cgRdStats.sigCost;             /* remove cost of significant coefficient bitmap */
+        }
+    } /* end for (cgScanPos) */
+
+    X265_CHECK(lastScanPos >= 0, "numSig non zero, but no coded CG\n");
+
+    /* calculate RD cost of uncoded block CBF=0, and add cost of CBF=1 to total */
+    int64_t bestCost;
+    if (!cu->isIntra(absPartIdx) && bIsLuma && !cu->getTransformIdx(absPartIdx))
+    {
+        bestCost = totalUncodedCost + SIGCOST(estBitsSbac.blockRootCbpBits[0]);
+        totalRdCost += SIGCOST(estBitsSbac.blockRootCbpBits[1]);
+    }
+    else
+    {
+        int ctx = ctxCbf[ttype][cu->getTransformIdx(absPartIdx)];
+        bestCost = totalUncodedCost + SIGCOST(estBitsSbac.blockCbpBits[ctx][0]);
+        totalRdCost += SIGCOST(estBitsSbac.blockCbpBits[ctx][1]);
+    }
+
+    /* This loop starts with the last non-zero found in the first loop and then refines this last
+     * non-zero by measuring the true RD cost of the last NZ at this position, and then the RD costs
+     * at all previous coefficients until a coefficient greater than 1 is encountered or we run out
+     * of coefficients to evaluate.  This will factor in the cost of coding empty groups and empty
+     * coeff prior to the last NZ. The base best cost is the RD cost of CBF=0 */
+    int  bestLastIdx = 0;
+    bool foundLast = false;
+    for (int cgScanPos = cgLastScanPos; cgScanPos >= 0 && !foundLast; cgScanPos--)
+    {
+        if (!cgScanPos || cgScanPos == cgLastScanPos)
+        {
+            /* the presence of these coefficient groups are inferred, they have no bit in
+             * sigCoeffGroupFlag64 and no saved costCoeffGroupSig[] cost */
+        }
+        else if (sigCoeffGroupFlag64 & (1ULL << codeParams.scanCG[cgScanPos]))
+        {
+            /* remove cost of significant coeff group flag, the group's presence would be inferred
+             * from lastNZ if it were present in this group */
+            totalRdCost -= costCoeffGroupSig[cgScanPos];
+        }
+        else
+        {
+            /* remove cost of signaling this empty group as not present */
+            totalRdCost -= costCoeffGroupSig[cgScanPos];
+            continue;
+        }
+
+        for (int scanPosinCG = cgSize - 1; scanPosinCG >= 0; scanPosinCG--)
+        {
+            scanPos = cgScanPos * cgSize + scanPosinCG;
+            if ((int)scanPos > lastScanPos)
+                continue;
+
+            /* if the coefficient was coded, measure the RD cost of it as the last non-zero and then
+             * continue as if it were uncoded. If the coefficient was already uncoded, remove the
+             * cost of signaling it as not-significant */
+            uint32_t blkPos = codeParams.scan[scanPos];
+            if (dstCoeff[blkPos])
+            {
+                /* Swap the cost of signaling its significant coeff bit with the cost of
+                 * signaling its lastNZ pos */
+                uint32_t posY = blkPos >> log2TrSize;
+                uint32_t posX = blkPos - (posY << log2TrSize);
+                uint32_t bitsLastNZ = codeParams.scanType == SCAN_VER ? getRateLast(posY, posX) : getRateLast(posX, posY);
+                int64_t costAsLast = totalRdCost - costSig[scanPos] + SIGCOST(bitsLastNZ);
+
+                if (costAsLast < bestCost)
+                {
+                    bestLastIdx = scanPos + 1;
+                    bestCost = costAsLast;
+                }
+                if (dstCoeff[blkPos] > 1)
+                {
+                    foundLast = true;
+                    break;
+                }
+
+                totalRdCost -= costCoeff[scanPos];
+                totalRdCost += costUncoded[scanPos];
+            }
+            else
+                totalRdCost -= costSig[scanPos];
+        }
+    }
+
+    /* recount non-zero coefficients and re-apply sign of DCT coef */
+    numSig = 0;
+    for (int pos = 0; pos < bestLastIdx; pos++)
+    {
+        int blkPos = codeParams.scan[pos];
+        int level  = dstCoeff[blkPos];
+        numSig += (level != 0);
+
+        uint32_t mask = (int32_t)m_resiDctCoeff[blkPos] >> 31;
+        dstCoeff[blkPos] = (int16_t)((level ^ mask) - mask);
+    }
+
+    /* clean uncoded coefficients */
+    for (int pos = bestLastIdx; pos <= lastScanPos; pos++)
+        dstCoeff[codeParams.scan[pos]] = 0;
+
+    /* rate-distortion based sign-hiding */
+    if (cu->m_slice->m_pps->bSignHideEnabled && numSig >= 2)
+    {
+        int lastCG = true;
+        for (int subSet = cgLastScanPos; subSet >= 0; subSet--)
+        {
+            int subPos = subSet << LOG2_SCAN_SET_SIZE;
+            int n;
+
+            /* measure distance between first and last non-zero coef in this
+             * coding group */
+            for (n = SCAN_SET_SIZE - 1; n >= 0; --n)
+                if (dstCoeff[codeParams.scan[n + subPos]])
+                    break;
+            if (n < 0)
+                continue;
+
+            int lastNZPosInCG = n;
+
+            for (n = 0;; n++)
+                if (dstCoeff[codeParams.scan[n + subPos]])
+                    break;
+
+            int firstNZPosInCG = n;
+
+            if (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD)
+            {
+                uint32_t signbit = (dstCoeff[codeParams.scan[subPos + firstNZPosInCG]] > 0 ? 0 : 1);
+                int absSum = 0;
+
+                for (n = firstNZPosInCG; n <= lastNZPosInCG; n++)
+                    absSum += dstCoeff[codeParams.scan[n + subPos]];
+
+                if (signbit != (absSum & 1U))
+                {
+                    /* We must find a coeff to toggle up or down so the sign bit of the first non-zero coeff
+                     * is properly implied. Note dstCoeff[] are signed by this point but curChange and
+                     * finalChange imply absolute levels (+1 is away from zero, -1 is towards zero) */
+
+                    int64_t minCostInc = MAX_INT64, curCost = MAX_INT64;
+                    int minPos = -1;
+                    int16_t finalChange = 0, curChange = 0;
+
+                    for (n = (lastCG ? lastNZPosInCG : SCAN_SET_SIZE - 1); n >= 0; --n)
+                    {
+                        uint32_t blkPos = codeParams.scan[n + subPos];
+                        int signCoef    = m_resiDctCoeff[blkPos]; /* pre-quantization DCT coeff */
+                        int absLevel    = abs(dstCoeff[blkPos]);
+
+                        int d = abs(signCoef) - UNQUANT(absLevel);
+                        int64_t origDist = (((int64_t)d * d)) << scaleBits;
+
+#define DELTARDCOST(d, deltabits) ((((int64_t)d * d) << scaleBits) - origDist + ((lambda2 * (int64_t)(deltabits)) >> 8))
+
+                        if (dstCoeff[blkPos])
+                        {
+                            d = abs(signCoef) - UNQUANT(absLevel + 1);
+                            int64_t costUp = DELTARDCOST(d, rateIncUp[blkPos]);
+
+                            /* if decrementing would make the coeff 0, we can include the
+                             * significant coeff flag cost savings */
+                            d = abs(signCoef) - UNQUANT(absLevel - 1);
+                            bool isOne = abs(dstCoeff[blkPos]) == 1;
+                            int downBits = rateIncDown[blkPos] - (isOne ? (IEP_RATE + sigRateDelta[blkPos]) : 0);
+                            int64_t costDown = DELTARDCOST(d, downBits);
+
+                            if (lastCG && lastNZPosInCG == n && isOne)
+                                costDown -= 4 * IEP_RATE;
+
+                            if (costUp < costDown)
+                            {
+                                curCost = costUp;
+                                curChange =  1;
+                            }
+                            else
+                            {
+                                curChange = -1;
+                                if (n == firstNZPosInCG && isOne)
+                                    curCost = MAX_INT64;
+                                else
+                                    curCost = costDown;
+                            }
+                        }
+                        else if (n < firstNZPosInCG && signbit != (signCoef >= 0 ? 0 : 1U))
+                        {
+                            /* don't try to make a new coded coeff before the first coeff if its
+                             * sign would be different than the first coeff, the inferred sign would
+                             * still be wrong and we'd have to do this again. */
+                            curCost = MAX_INT64;
+                        }
+                        else
+                        {
+                            /* evaluate changing an uncoded coeff 0 to a coded coeff +/-1 */
+                            d = abs(signCoef) - UNQUANT(1);
+                            curCost = DELTARDCOST(d, rateIncUp[blkPos] + IEP_RATE + sigRateDelta[blkPos]);
+                            curChange = 1;
+                        }
+
+                        if (curCost < minCostInc)
+                        {
+                            minCostInc = curCost;
+                            finalChange = curChange;
+                            minPos = blkPos;
+                        }
+                    }
+
+                    if (dstCoeff[minPos] == 32767 || dstCoeff[minPos] == -32768)
+                        /* don't allow sign hiding to violate the SPEC range */
+                        finalChange = -1;
+
+                    if (dstCoeff[minPos] == 0)
+                        numSig++;
+                    else if (finalChange == -1 && abs(dstCoeff[minPos]) == 1)
+                        numSig--;
+
+                    if (m_resiDctCoeff[minPos] >= 0)
+                        dstCoeff[minPos] += finalChange;
+                    else
+                        dstCoeff[minPos] -= finalChange;
+                }
+            }
+
+            lastCG = false;
+        }
+    }
+
+    return numSig;
+}
+
+/* Pattern decision for context derivation process of significant_coeff_flag */
+uint32_t Quant::calcPatternSigCtx(uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, uint32_t log2TrSizeCG)
+{
+    if (!log2TrSizeCG)
+        return 0;
+
+    const uint32_t trSizeCG = 1 << log2TrSizeCG;
+    X265_CHECK(trSizeCG <= 8, "transform CG is too large\n");
+    const uint32_t sigPos = (uint32_t)(sigCoeffGroupFlag64 >> (1 + (cgPosY << log2TrSizeCG) + cgPosX));
+    const uint32_t sigRight = ((int32_t)(cgPosX - (trSizeCG - 1)) >> 31) & (sigPos & 1);
+    const uint32_t sigLower = ((int32_t)(cgPosY - (trSizeCG - 1)) >> 31) & (sigPos >> (trSizeCG - 2)) & 2;
+
+    return sigRight + sigLower;
+}
+
+/* Context derivation process of coeff_abs_significant_flag */
+uint32_t Quant::getSigCtxInc(uint32_t patternSigCtx, uint32_t log2TrSize, uint32_t trSize, uint32_t blkPos, bool bIsLuma,
+                             uint32_t firstSignificanceMapContext)
+{
+    static const uint8_t ctxIndMap[16] =
+    {
+        0, 1, 4, 5,
+        2, 3, 4, 5,
+        6, 6, 8, 8,
+        7, 7, 8, 8
+    };
+
+    if (!blkPos) // special case for the DC context variable
+        return 0;
+
+    if (log2TrSize == 2) // 4x4
+        return ctxIndMap[blkPos];
+
+    const uint32_t posY = blkPos >> log2TrSize;
+    const uint32_t posX = blkPos & (trSize - 1);
+    X265_CHECK((blkPos - (posY << log2TrSize)) == posX, "block pos check failed\n");
+
+    int posXinSubset = blkPos & 3;
+    X265_CHECK((posX & 3) == (blkPos & 3), "pos alignment fail\n");
+    int posYinSubset = posY & 3;
+
+    // NOTE: [patternSigCtx][posXinSubset][posYinSubset]
+    static const uint8_t table_cnt[4][4][4] =
+    {
+        // patternSigCtx = 0
+        {
+            { 2, 1, 1, 0 },
+            { 1, 1, 0, 0 },
+            { 1, 0, 0, 0 },
+            { 0, 0, 0, 0 },
+        },
+        // patternSigCtx = 1
+        {
+            { 2, 1, 0, 0 },
+            { 2, 1, 0, 0 },
+            { 2, 1, 0, 0 },
+            { 2, 1, 0, 0 },
+        },
+        // patternSigCtx = 2
+        {
+            { 2, 2, 2, 2 },
+            { 1, 1, 1, 1 },
+            { 0, 0, 0, 0 },
+            { 0, 0, 0, 0 },
+        },
+        // patternSigCtx = 3
+        {
+            { 2, 2, 2, 2 },
+            { 2, 2, 2, 2 },
+            { 2, 2, 2, 2 },
+            { 2, 2, 2, 2 },
+        }
+    };
+
+    int cnt = table_cnt[patternSigCtx][posXinSubset][posYinSubset];
+    int offset = firstSignificanceMapContext;
+
+    offset += cnt;
+
+    return (bIsLuma && (posX | posY) >= 4) ? 3 + offset : offset;
+}
+
+/* Calculates the cost of signaling the last significant coefficient in the block */
+inline uint32_t Quant::getRateLast(uint32_t posx, uint32_t posy) const
+{
+    uint32_t ctxX = getGroupIdx(posx);
+    uint32_t ctxY = getGroupIdx(posy);
+    uint32_t cost = m_entropyCoder->m_estBitsSbac.lastXBits[ctxX] + m_entropyCoder->m_estBitsSbac.lastYBits[ctxY];
+
+    int32_t maskX = (int32_t)(2 - posx) >> 31;
+    int32_t maskY = (int32_t)(2 - posy) >> 31;
+
+    cost += maskX & (IEP_RATE * ((ctxX - 2) >> 1));
+    cost += maskY & (IEP_RATE * ((ctxY - 2) >> 1));
+    return cost;
+}
+
+/* Context derivation process of coeff_abs_significant_flag */
+uint32_t Quant::getSigCoeffGroupCtxInc(uint64_t cgGroupMask, uint32_t cgPosX, uint32_t cgPosY, uint32_t log2TrSizeCG)
+{
+    const uint32_t trSizeCG = 1 << log2TrSizeCG;
+
+    const uint32_t sigPos = (uint32_t)(cgGroupMask >> (1 + (cgPosY << log2TrSizeCG) + cgPosX));
+    const uint32_t sigRight = ((int32_t)(cgPosX - (trSizeCG - 1)) >> 31) & sigPos;
+    const uint32_t sigLower = ((int32_t)(cgPosY - (trSizeCG - 1)) >> 31) & (sigPos >> (trSizeCG - 1));
+
+    return (sigRight | sigLower) & 1;
+}

x265-hg20140928.tar.bz2/source/common/quant.h Added

@@ -0,0 +1,137 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_QUANT_H
+#define X265_QUANT_H
+
+#include "common.h"
+#include "scalinglist.h"
+#include "TLibCommon/TComRom.h"
+#include "TLibCommon/ContextTables.h"
+
+namespace x265 {
+// private namespace
+
+class TComDataCU;
+class Entropy;
+struct TUEntropyCodingParameters;
+
+struct QpParam
+{
+    int rem;
+    int per;
+    int qp;
+    int64_t lambda2; /* FIX8 */
+    int64_t lambda;  /* FIX8 */
+
+    QpParam() : qp(MAX_INT) {}
+
+    void setQpParam(int qpScaled)
+    {
+        if (qp != qpScaled)
+        {
+            rem = qpScaled % 6;
+            per = qpScaled / 6;
+            qp  = qpScaled;
+            lambda2 = (int64_t)(x265_lambda2_tab[qp - QP_BD_OFFSET] * 256. + 0.5);
+            lambda  = (int64_t)(x265_lambda_tab[qp - QP_BD_OFFSET] * 256. + 0.5);
+        }
+    }
+};
+
+class Quant
+{
+public:
+
+    NoiseReduction*    m_nr;
+    Entropy*           m_entropyCoder;
+
+protected:
+
+    const ScalingList* m_scalingList;
+
+    QpParam            m_qpParam[3];
+
+    bool               m_useRDOQ;
+    int64_t            m_psyRdoqScale;
+    int32_t*           m_resiDctCoeff;
+    int32_t*           m_fencDctCoeff;
+    int16_t*           m_fencShortBuf;
+
+    enum { IEP_RATE = 32768 }; /* FIX15 cost of an equal probable bit */
+
+public:
+
+    Quant();
+    ~Quant();
+
+    /* one-time setup */
+    bool init(bool useRDOQ, double psyScale, const ScalingList& scalingList);
+
+    /* CU setup */
+    void setQPforQuant(TComDataCU* cu);
+
+    uint32_t transformNxN(TComDataCU* cu, pixel *fenc, uint32_t fencstride, int16_t* residual, uint32_t stride, coeff_t* coeff,
+                          uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx, bool useTransformSkip);
+
+    void invtransformNxN(bool transQuantBypass, int16_t* residual, uint32_t stride, coeff_t* coeff,
+                         uint32_t log2TrSize, TextType ttype, bool bIntra, bool useTransformSkip, uint32_t numSig);
+
+    /* static methods shared with entropy.cpp */
+    static uint32_t calcPatternSigCtx(uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, uint32_t log2TrSizeCG);
+    static uint32_t getSigCtxInc(uint32_t patternSigCtx, uint32_t log2TrSize, uint32_t trSize, uint32_t blkPos, bool bIsLuma, uint32_t firstSignificanceMapContext);
+    static uint32_t getSigCoeffGroupCtxInc(uint64_t sigCoeffGroupFlag64, uint32_t cgPosX, uint32_t cgPosY, uint32_t log2TrSizeCG);
+
+protected:
+
+    void setChromaQP(int qpin, TextType ttype, int chFmt);
+
+    uint32_t signBitHidingHDQ(int16_t* qcoeff, int32_t* deltaU, uint32_t numSig, const TUEntropyCodingParameters &codingParameters);
+
+    uint32_t rdoQuant(TComDataCU* cu, int16_t* dstCoeff, uint32_t log2TrSize, TextType ttype, uint32_t absPartIdx, bool usePsy);
+    inline uint32_t getRateLast(uint32_t posx, uint32_t posy) const;
+};
+
+static inline uint32_t getGroupIdx(const uint32_t idx)
+{
+    // TODO: Why is this not a table lookup?
+
+    uint32_t group = (idx >> 3);
+
+    if (idx >= 24)
+        group = 2;
+    uint32_t groupIdx = ((idx >> (group + 1)) - 2) + 4 + (group << 1);
+    if (idx <= 3)
+        groupIdx = idx;
+
+#ifdef _DEBUG
+    static const uint8_t g_groupIdx[32] = { 0, 1, 2, 3, 4, 4, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9 };
+    assert(groupIdx == g_groupIdx[idx]);
+#endif
+
+    return groupIdx;
+}
+
+}
+
+#endif // ifndef X265_QUANT_H

x265-hg20140928.tar.bz2/source/common/scalinglist.cpp Added

@@ -0,0 +1,380 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "common.h"
+#include "primitives.h"
+#include "TLibCommon/TComRom.h"
+#include "scalinglist.h"
+
+namespace {
+// file-anonymous namespace
+
+/* Strings for scaling list file parsing */
+const char MatrixType[4][6][20] =
+{
+    {
+        "INTRA4X4_LUMA",
+        "INTRA4X4_CHROMAU",
+        "INTRA4X4_CHROMAV",
+        "INTER4X4_LUMA",
+        "INTER4X4_CHROMAU",
+        "INTER4X4_CHROMAV"
+    },
+    {
+        "INTRA8X8_LUMA",
+        "INTRA8X8_CHROMAU",
+        "INTRA8X8_CHROMAV",
+        "INTER8X8_LUMA",
+        "INTER8X8_CHROMAU",
+        "INTER8X8_CHROMAV"
+    },
+    {
+        "INTRA16X16_LUMA",
+        "INTRA16X16_CHROMAU",
+        "INTRA16X16_CHROMAV",
+        "INTER16X16_LUMA",
+        "INTER16X16_CHROMAU",
+        "INTER16X16_CHROMAV"
+    },
+    {
+        "INTRA32X32_LUMA",
+        "INTER32X32_LUMA",
+    },
+};
+const char MatrixType_DC[4][12][22] =
+{
+    {
+    },
+    {
+    },
+    {
+        "INTRA16X16_LUMA_DC",
+        "INTRA16X16_CHROMAU_DC",
+        "INTRA16X16_CHROMAV_DC",
+        "INTER16X16_LUMA_DC",
+        "INTER16X16_CHROMAU_DC",
+        "INTER16X16_CHROMAV_DC"
+    },
+    {
+        "INTRA32X32_LUMA_DC",
+        "INTER32X32_LUMA_DC",
+    },
+};
+
+int quantTSDefault4x4[16] =
+{
+    16, 16, 16, 16,
+    16, 16, 16, 16,
+    16, 16, 16, 16,
+    16, 16, 16, 16
+};
+
+int quantIntraDefault8x8[64] =
+{
+    16, 16, 16, 16, 17, 18, 21, 24,
+    16, 16, 16, 16, 17, 19, 22, 25,
+    16, 16, 17, 18, 20, 22, 25, 29,
+    16, 16, 18, 21, 24, 27, 31, 36,
+    17, 17, 20, 24, 30, 35, 41, 47,
+    18, 19, 22, 27, 35, 44, 54, 65,
+    21, 22, 25, 31, 41, 54, 70, 88,
+    24, 25, 29, 36, 47, 65, 88, 115
+};
+
+int quantInterDefault8x8[64] =
+{
+    16, 16, 16, 16, 17, 18, 20, 24,
+    16, 16, 16, 17, 18, 20, 24, 25,
+    16, 16, 17, 18, 20, 24, 25, 28,
+    16, 17, 18, 20, 24, 25, 28, 33,
+    17, 18, 20, 24, 25, 28, 33, 41,
+    18, 20, 24, 25, 28, 33, 41, 54,
+    20, 24, 25, 28, 33, 41, 54, 71,
+    24, 25, 28, 33, 41, 54, 71, 91
+};
+
+}
+
+namespace x265 {
+// private namespace
+
+const int     ScalingList::s_numCoefPerSize[NUM_SIZES] = { 16, 64, 256, 1024 };
+const int32_t ScalingList::s_quantScales[NUM_REM] = { 26214, 23302, 20560, 18396, 16384, 14564 };
+const int32_t ScalingList::s_invQuantScales[NUM_REM] = { 40, 45, 51, 57, 64, 72 };
+
+ScalingList::ScalingList()
+{
+    memset(m_quantCoef, 0, sizeof(m_quantCoef));
+    memset(m_dequantCoef, 0, sizeof(m_dequantCoef));
+    memset(m_scalingListCoef, 0, sizeof(m_scalingListCoef));
+}
+
+bool ScalingList::init()
+{
+    bool ok = true;
+    for (int sizeId = 0; sizeId < NUM_SIZES; sizeId++)
+    {
+        for (int listId = 0; listId < NUM_LISTS; listId++)
+        {
+            m_scalingListCoef[sizeId][listId] = X265_MALLOC(int32_t, X265_MIN(MAX_MATRIX_COEF_NUM, s_numCoefPerSize[sizeId]));
+            ok &= !!m_scalingListCoef[sizeId][listId];
+            for (int rem = 0; rem < NUM_REM; rem++)
+            {
+                m_quantCoef[sizeId][listId][rem] = X265_MALLOC(int32_t, s_numCoefPerSize[sizeId]);
+                m_dequantCoef[sizeId][listId][rem] = X265_MALLOC(int32_t, s_numCoefPerSize[sizeId]);
+                ok &= m_quantCoef[sizeId][listId][rem] && m_dequantCoef[sizeId][listId][rem];
+            }
+        }
+    }
+    return ok;
+}
+
+ScalingList::~ScalingList()
+{
+    for (int sizeId = 0; sizeId < NUM_SIZES; sizeId++)
+    {
+        for (int listId = 0; listId < NUM_LISTS; listId++)
+        {
+            X265_FREE(m_scalingListCoef[sizeId][listId]);
+            for (int rem = 0; rem < NUM_REM; rem++)
+            {
+                X265_FREE(m_quantCoef[sizeId][listId][rem]);
+                X265_FREE(m_dequantCoef[sizeId][listId][rem]);
+            }
+        }
+    }
+}
+
+/* returns predicted list index if a match is found, else -1 */ 
+int ScalingList::checkPredMode(int size, int list)
+{
+    for (int predList = list; predList >= 0; predList--)
+    {
+        // check DC value
+        if (size < BLOCK_16x16 && m_scalingListDC[size][list] != m_scalingListDC[size][predList])
+            continue;
+
+        // check value of matrix
+        if (!memcmp(m_scalingListCoef[size][list],
+                    list == predList ? getScalingListDefaultAddress(size, predList) : m_scalingListCoef[size][predList],
+                    sizeof(int32_t) * X265_MIN(MAX_MATRIX_COEF_NUM, s_numCoefPerSize[size])))
+            return predList;
+    }
+
+    return -1;
+}
+
+/* check if use default quantization matrix
+ * returns true if default quantization matrix is used in all sizes */
+bool ScalingList::checkDefaultScalingList()
+{
+    int defaultCounter = 0;
+
+    for (int s = 0; s < NUM_SIZES; s++)
+        for (int l = 0; l < NUM_LISTS; l++)
+            if (!memcmp(m_scalingListCoef[s][l], getScalingListDefaultAddress(s, l),
+                        sizeof(int32_t) * X265_MIN(MAX_MATRIX_COEF_NUM, s_numCoefPerSize[s])) &&
+                ((s < BLOCK_16x16) || (m_scalingListDC[s][l] == 16)))
+                defaultCounter++;
+
+    return defaultCounter != (NUM_LISTS * NUM_SIZES - 4); // -4 for 32x32
+}
+
+/* get address of default quantization matrix */
+int32_t* ScalingList::getScalingListDefaultAddress(int sizeId, int listId)
+{
+    switch (sizeId)
+    {
+    case BLOCK_4x4:
+        return quantTSDefault4x4;
+    case BLOCK_8x8:
+        return (listId < 3) ? quantIntraDefault8x8 : quantInterDefault8x8;
+    case BLOCK_16x16:
+        return (listId < 3) ? quantIntraDefault8x8 : quantInterDefault8x8;
+    case BLOCK_32x32:
+        return (listId < 1) ? quantIntraDefault8x8 : quantInterDefault8x8;
+    default:
+        break;
+    }
+
+    X265_CHECK(0, "invalid scaling list size\n");
+    return NULL;
+}
+
+void ScalingList::processDefaultMarix(int sizeId, int listId)
+{
+    ::memcpy(m_scalingListCoef[sizeId][listId], getScalingListDefaultAddress(sizeId, listId), sizeof(int) * X265_MIN(MAX_MATRIX_COEF_NUM, s_numCoefPerSize[sizeId]));
+    m_scalingListDC[sizeId][listId] = SCALING_LIST_DC;
+}
+
+void ScalingList::setDefaultScalingList()
+{
+    for (int sizeId = 0; sizeId < NUM_SIZES; sizeId++)
+        for (int listId = 0; listId < NUM_LISTS; listId++)
+            processDefaultMarix(sizeId, listId);
+    m_bEnabled = true;
+    m_bDataPresent = false;
+}
+
+bool ScalingList::parseScalingList(const char* filename)
+{
+    FILE *fp = fopen(filename, "r");
+    if (!fp)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "can't open scaling list file %s\n", filename);
+        return true;
+    }
+
+    char line[1024];
+    int32_t *src = NULL;
+
+    for (int sizeIdc = 0; sizeIdc < NUM_SIZES; sizeIdc++)
+    {
+        int size = X265_MIN(MAX_MATRIX_COEF_NUM, s_numCoefPerSize[sizeIdc]);
+        for (int listIdc = 0; listIdc < NUM_LISTS; listIdc++)
+        {
+            src = m_scalingListCoef[sizeIdc][listIdc];
+
+            fseek(fp, 0, 0);
+            do
+            {
+                char *ret = fgets(line, 1024, fp);
+                if (!ret || (!strstr(line, MatrixType[sizeIdc][listIdc]) && feof(fp)))
+                {
+                    x265_log(NULL, X265_LOG_ERROR, "can't read matrix from %s\n", filename);
+                    return true;
+                }
+            }
+            while (!strstr(line, MatrixType[sizeIdc][listIdc]));
+
+            for (int i = 0; i < size; i++)
+            {
+                int data;
+                if (fscanf(fp, "%d,", &data) != 1)
+                {
+                    x265_log(NULL, X265_LOG_ERROR, "can't read matrix from %s\n", filename);
+                    return true;
+                }
+                src[i] = data;
+            }
+
+            // set DC value for default matrix check
+            m_scalingListDC[sizeIdc][listIdc] = src[0];
+
+            if (sizeIdc > BLOCK_8x8)
+            {
+                fseek(fp, 0, 0);
+                do
+                {
+                    char *ret = fgets(line, 1024, fp);
+                    if (!ret || (!strstr(line, MatrixType_DC[sizeIdc][listIdc]) && feof(fp)))
+                    {
+                        x265_log(NULL, X265_LOG_ERROR, "can't read DC from %s\n", filename);
+                        return true;
+                    }
+                }
+                while (!strstr(line, MatrixType_DC[sizeIdc][listIdc]));
+
+                int data;
+                if (fscanf(fp, "%d,", &data) != 1)
+                {
+                    x265_log(NULL, X265_LOG_ERROR, "can't read matrix from %s\n", filename);
+                    return true;
+                }
+
+                // overwrite DC value when size of matrix is larger than 16x16
+                m_scalingListDC[sizeIdc][listIdc] = data;
+            }
+        }
+    }
+
+    fclose(fp);
+
+    m_bEnabled = true;
+    m_bDataPresent = !checkDefaultScalingList();
+
+    return false;
+}
+
+/** set quantized matrix coefficient for encode */
+void ScalingList::setupQuantMatrices()
+{
+    for (int size = 0; size < NUM_SIZES; size++)
+    {
+        int width = 1 << (size + 2);
+        int ratio = width / X265_MIN(MAX_MATRIX_SIZE_NUM, width);
+        int stride = X265_MIN(MAX_MATRIX_SIZE_NUM, width);
+        int count = s_numCoefPerSize[size];
+
+        for (int list = 0; list < NUM_LISTS; list++)
+        {
+            int32_t *coeff = m_scalingListCoef[size][list];
+            int32_t dc = m_scalingListDC[size][list];
+
+            for (int rem = 0; rem < NUM_REM; rem++)
+            {
+                int32_t *quantCoeff   = m_quantCoef[size][list][rem];
+                int32_t *dequantCoeff = m_dequantCoef[size][list][rem];
+
+                if (m_bEnabled)
+                {
+                    processScalingListEnc(coeff, quantCoeff, s_quantScales[rem] << 4, width, width, ratio, stride, dc);
+                    processScalingListDec(coeff, dequantCoeff, s_invQuantScales[rem], width, width, ratio, stride, dc);
+                }
+                else
+                {
+                    /* flat quant and dequant coefficients */
+                    for (int i = 0; i < count; i++)
+                    {
+                        quantCoeff[i] = s_quantScales[rem];
+                        dequantCoeff[i] = s_invQuantScales[rem];
+                    }
+                }
+            }
+        }
+    }
+}
+
+void ScalingList::processScalingListEnc(int32_t *coeff, int32_t *quantcoeff, int32_t quantScales, int height, int width,
+                                        int ratio, int stride, int32_t dc)
+{
+    for (int j = 0; j < height; j++)
+        for (int i = 0; i < width; i++)
+            quantcoeff[j * width + i] = quantScales / coeff[stride * (j / ratio) + i / ratio];
+
+    if (ratio > 1)
+        quantcoeff[0] = quantScales / dc;
+}
+
+void ScalingList::processScalingListDec(int32_t *coeff, int32_t *dequantcoeff, int32_t invQuantScales, int height, int width,
+                                        int ratio, int stride, int32_t dc)
+{
+    for (int j = 0; j < height; j++)
+        for (int i = 0; i < width; i++)
+            dequantcoeff[j * width + i] = invQuantScales * coeff[stride * (j / ratio) + i / ratio];
+
+    if (ratio > 1)
+        dequantcoeff[0] = invQuantScales * dc;
+}
+
+}

x265-hg20140928.tar.bz2/source/common/scalinglist.h Added

@@ -0,0 +1,80 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_SCALINGLIST_H
+#define X265_SCALINGLIST_H
+
+#include "common.h"
+
+namespace x265 {
+// private namespace
+
+class ScalingList
+{
+public:
+
+    enum { NUM_SIZES = 4 };            // 4x4, 8x8, 16x16, 32x32
+    enum { NUM_LISTS = 6 };            // number of quantization matrix lists (YUV * inter/intra)
+    enum { NUM_REM = 6 };              // number of remainders of QP/6
+    enum { MAX_MATRIX_COEF_NUM = 64 }; // max coefficient number per quantization matrix
+    enum { MAX_MATRIX_SIZE_NUM = 8 };  // max size number for quantization matrix
+
+    static const int     s_numCoefPerSize[NUM_SIZES];
+    static const int32_t s_invQuantScales[NUM_REM];
+    static const int32_t s_quantScales[NUM_REM];
+
+    int32_t  m_scalingListDC[NUM_SIZES][NUM_LISTS];   // the DC value of the matrix coefficient for 16x16
+    int32_t* m_scalingListCoef[NUM_SIZES][NUM_LISTS]; // quantization matrix
+
+    int32_t* m_quantCoef[NUM_SIZES][NUM_LISTS][NUM_REM];   // array of quantization matrix coefficient 4x4
+    int32_t* m_dequantCoef[NUM_SIZES][NUM_LISTS][NUM_REM]; // array of dequantization matrix coefficient 4x4
+
+    bool     m_bEnabled;
+    bool     m_bDataPresent; // non-default scaling lists must be signaled
+
+    ScalingList();
+    ~ScalingList();
+
+    bool     init();
+    void     setDefaultScalingList();
+    bool     parseScalingList(const char* filename);
+    void     setupQuantMatrices();
+
+    /* used during SPS coding */
+    int      checkPredMode(int sizeId, int listId);
+
+protected:
+
+    static const int SCALING_LIST_DC = 16;    // default DC value
+
+    int32_t* getScalingListDefaultAddress(int sizeId, int listId);
+    void     processDefaultMarix(int sizeId, int listId);
+    bool     checkDefaultScalingList();
+
+    void     processScalingListEnc(int32_t *coeff, int32_t *quantcoeff, int32_t quantScales, int height, int width, int ratio, int stride, int32_t dc);
+    void     processScalingListDec(int32_t *coeff, int32_t *dequantcoeff, int32_t invQuantScales, int height, int width, int ratio, int stride, int32_t dc);
+};
+
+}
+
+#endif // ifndef X265_SCALINGLIST_H

x265-hg20140324.tar.bz2/source/common/shortyuv.cpp -> x265-hg20140928.tar.bz2/source/common/shortyuv.cpp Changed

x265-hg20140324.tar.bz2/source/common/shortyuv.h -> x265-hg20140928.tar.bz2/source/common/shortyuv.h Changed

@@ -1,5 +1,5 @@
 /*****************************************************************************
- * x265: TShortYUV classes for short sized YUV-style frames
+ * x265: ShortYUV class for short sized YUV-style frames
  *****************************************************************************
  * Copyright (C) 2013 x265 project
  *
@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com
+ * For more information, contact us at license @ x265.com
  *****************************************************************************/
 
 #ifndef X265_SHORTYUV_H
@@ -37,9 +37,7 @@
 {
 public:
 
-    int16_t* m_bufY;
-    int16_t* m_bufCb;
-    int16_t* m_bufCr;
+    int16_t* m_buf[3];
 
     uint32_t m_width;
     uint32_t m_height;
@@ -51,22 +49,22 @@
     int m_vChromaShift;
 
     ShortYuv();
-    virtual ~ShortYuv();
+    ~ShortYuv();
 
-    static int getAddrOffset(uint32_t idx, uint32_t width)
+    int getChromaAddrOffset(uint32_t idx, uint32_t width)
     {
-        int blkX = g_rasterToPelX[g_zscanToRaster[idx]];
-        int blkY = g_rasterToPelY[g_zscanToRaster[idx]];
+        int blkX = g_zscanToPelX[idx] >> m_hChromaShift;
+        int blkY = g_zscanToPelY[idx] >> m_vChromaShift;
 
         return blkX + blkY * width;
     }
 
-    static int getAddrOffset(uint32_t idx, uint32_t size, uint32_t width)
+    static int getAddrOffset(uint32_t idx, uint32_t width)
     {
-        int blkX = (idx * size) &  (width - 1);
-        int blkY = (idx * size) & ~(width - 1);
+        int blkX = g_zscanToPelX[idx];
+        int blkY = g_zscanToPelY[idx];
 
-        return blkX + blkY * size;
+        return blkX + blkY * width;
     }
 
     bool create(uint32_t width, uint32_t height, int csp);
@@ -74,36 +72,37 @@
     void destroy();
     void clear();
 
-    int16_t* getLumaAddr()  { return m_bufY; }
+    int16_t* getLumaAddr()  { return m_buf[0]; }
+
+    int16_t* getCbAddr()    { return m_buf[1]; }
 
-    int16_t* getCbAddr()    { return m_bufCb; }
+    int16_t* getCrAddr()    { return m_buf[2]; }
 
-    int16_t* getCrAddr()    { return m_bufCr; }
+    int16_t* getChromaAddr(uint32_t chromaId)    { return m_buf[chromaId]; }
 
     // Access starting position of YUV partition unit buffer
-    int16_t* getLumaAddr(uint32_t partUnitIdx) { return m_bufY + getAddrOffset(partUnitIdx, m_width); }
+    int16_t* getLumaAddr(uint32_t partUnitIdx) { return m_buf[0] + getAddrOffset(partUnitIdx, m_width); }
 
-    int16_t* getCbAddr(uint32_t partUnitIdx) { return m_bufCb + (getAddrOffset(partUnitIdx, m_cwidth) >> m_hChromaShift); }
+    int16_t* getCbAddr(uint32_t partUnitIdx) { return m_buf[1] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
-    int16_t* getCrAddr(uint32_t partUnitIdx) { return m_bufCr + (getAddrOffset(partUnitIdx, m_cwidth) >> m_hChromaShift); }
+    int16_t* getCrAddr(uint32_t partUnitIdx) { return m_buf[2] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
-    // Access starting position of YUV transform unit buffer
-    int16_t* getLumaAddr(uint32_t partIdx, uint32_t size) { return m_bufY + getAddrOffset(partIdx, size, m_width); }
+    int16_t* getChromaAddr(uint32_t chromaId, uint32_t partUnitIdx) { return m_buf[chromaId] + getChromaAddrOffset(partUnitIdx, m_cwidth); }
 
-    int16_t* getCbAddr(uint32_t partIdx, uint32_t size) { return m_bufCb + getAddrOffset(partIdx, size, m_cwidth); }
+    void subtract(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t log2Size);
 
-    int16_t* getCrAddr(uint32_t partIdx, uint32_t size) { return m_bufCr + getAddrOffset(partIdx, size, m_cwidth); }
+    void copyPartToPartLuma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t log2Size);
+    void copyPartToPartChroma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t log2SizeL);
+    void copyPartToPartLuma(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t log2Size);
+    void copyPartToPartChroma(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t log2SizeL);
 
-    void subtract(TComYuv* srcYuv0, TComYuv* srcYuv1, uint32_t partSize);
-    void addClip(ShortYuv* srcYuv0, ShortYuv* srcYuv1, uint32_t partSize);
+    // -------------------------------------------------------------------------------------------------------------------
+    // member functions to support multiple color space formats
+    // -------------------------------------------------------------------------------------------------------------------
 
-    void copyPartToPartLuma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t width, uint32_t height);
-    void copyPartToPartChroma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, bool bChromaSame);
-    void copyPartToPartShortChroma(ShortYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, uint32_t chromaId);
+    int  getHorzChromaShift()  { return m_hChromaShift; }
 
-    void copyPartToPartLuma(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t width, uint32_t height);
-    void copyPartToPartChroma(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, bool bChromaSame);
-    void copyPartToPartYuvChroma(TComYuv* dstPicYuv, uint32_t partIdx, uint32_t lumaSize, uint32_t chromaId);
+    int  getVertChromaShift()  { return m_vChromaShift; }
 };
 }

x265-hg20140928.tar.bz2/source/common/slice.cpp Added

@@ -0,0 +1,203 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "common.h"
+#include "frame.h"
+#include "piclist.h"
+#include "slice.h"
+
+using namespace x265;
+
+void Slice::setRefPicList(PicList& picList)
+{
+    if (m_sliceType == I_SLICE)
+    {
+        ::memset(m_refPicList, 0, sizeof(m_refPicList));
+        m_numRefIdx[1] = m_numRefIdx[0] = 0;
+        return;
+    }
+
+    Frame* refPic = NULL;
+    Frame* refPicSetStCurr0[MAX_NUM_REF];
+    Frame* refPicSetStCurr1[MAX_NUM_REF];
+    Frame* refPicSetLtCurr[MAX_NUM_REF];
+    int numPocStCurr0 = 0;
+    int numPocStCurr1 = 0;
+    int numPocLtCurr = 0;
+    int i;
+
+    for (i = 0; i < m_rps.numberOfNegativePictures; i++)
+    {
+        if (m_rps.bUsed[i])
+        {
+            refPic = picList.getPOC(m_poc + m_rps.deltaPOC[i]);
+            refPicSetStCurr0[numPocStCurr0] = refPic;
+            numPocStCurr0++;
+        }
+    }
+
+    for (; i < m_rps.numberOfNegativePictures + m_rps.numberOfPositivePictures; i++)
+    {
+        if (m_rps.bUsed[i])
+        {
+            refPic = picList.getPOC(m_poc + m_rps.deltaPOC[i]);
+            refPicSetStCurr1[numPocStCurr1] = refPic;
+            numPocStCurr1++;
+        }
+    }
+
+    X265_CHECK(m_rps.numberOfPictures == m_rps.numberOfNegativePictures + m_rps.numberOfPositivePictures,
+               "unexpected picture in RPS\n");
+
+    // ref_pic_list_init
+    Frame* rpsCurrList0[MAX_NUM_REF + 1];
+    Frame* rpsCurrList1[MAX_NUM_REF + 1];
+    int numPocTotalCurr = numPocStCurr0 + numPocStCurr1 + numPocLtCurr;
+
+    int cIdx = 0;
+    for (i = 0; i < numPocStCurr0; i++, cIdx++)
+        rpsCurrList0[cIdx] = refPicSetStCurr0[i];
+
+    for (i = 0; i < numPocStCurr1; i++, cIdx++)
+        rpsCurrList0[cIdx] = refPicSetStCurr1[i];
+
+    for (i = 0; i < numPocLtCurr; i++, cIdx++)
+        rpsCurrList0[cIdx] = refPicSetLtCurr[i];
+
+    X265_CHECK(cIdx == numPocTotalCurr, "RPS index check fail\n");
+
+    if (m_sliceType == B_SLICE)
+    {
+        cIdx = 0;
+        for (i = 0; i < numPocStCurr1; i++, cIdx++)
+            rpsCurrList1[cIdx] = refPicSetStCurr1[i];
+
+        for (i = 0; i < numPocStCurr0; i++, cIdx++)
+            rpsCurrList1[cIdx] = refPicSetStCurr0[i];
+
+        for (i = 0; i < numPocLtCurr; i++, cIdx++)
+            rpsCurrList1[cIdx] = refPicSetLtCurr[i];
+
+        X265_CHECK(cIdx == numPocTotalCurr, "RPS index check fail\n");
+    }
+
+    for (int rIdx = 0; rIdx < m_numRefIdx[0]; rIdx++)
+    {
+        cIdx = rIdx % numPocTotalCurr;
+        X265_CHECK(cIdx >= 0 && cIdx < numPocTotalCurr, "RPS index check fail\n");
+        m_refPicList[0][rIdx] = rpsCurrList0[cIdx];
+    }
+
+    if (m_sliceType != B_SLICE)
+    {
+        m_numRefIdx[1] = 0;
+        ::memset(m_refPicList[1], 0, sizeof(m_refPicList[1]));
+    }
+    else
+    {
+        for (int rIdx = 0; rIdx < m_numRefIdx[1]; rIdx++)
+        {
+            cIdx = rIdx % numPocTotalCurr;
+            X265_CHECK(cIdx >= 0 && cIdx < numPocTotalCurr, "RPS index check fail\n");
+            m_refPicList[1][rIdx] = rpsCurrList1[cIdx];
+        }
+    }
+
+    for (int dir = 0; dir < 2; dir++)
+        for (int numRefIdx = 0; numRefIdx < m_numRefIdx[dir]; numRefIdx++)
+            m_refPOCList[dir][numRefIdx] = m_refPicList[dir][numRefIdx]->getPOC();
+}
+
+void Slice::disableWeights()
+{
+    for (int l = 0; l < 2; l++)
+        for (int i = 0; i < MAX_NUM_REF; i++)
+            for (int yuv = 0; yuv < 3; yuv++)
+            {
+                WeightParam& wp = m_weightPredTable[l][i][yuv];
+                wp.bPresentFlag = false;
+                wp.log2WeightDenom = 0;
+                wp.inputWeight = 1;
+                wp.inputOffset = 0;
+            }
+}
+
+/* Sorts the deltaPOC and Used by current values in the RPS based on the
+ * deltaPOC values.  deltaPOC values are sorted with -ve values before the +ve
+ * values.  -ve values are in decreasing order.  +ve values are in increasing
+ * order */
+void RPS::sortDeltaPOC()
+{
+    // sort in increasing order (smallest first)
+    for (int j = 1; j < numberOfPictures; j++)
+    {
+        int dPOC = deltaPOC[j];
+        bool used = bUsed[j];
+        for (int k = j - 1; k >= 0; k--)
+        {
+            int temp = deltaPOC[k];
+            if (dPOC < temp)
+            {
+                deltaPOC[k + 1] = temp;
+                bUsed[k + 1] = bUsed[k];
+                deltaPOC[k] = dPOC;
+                bUsed[k] = used;
+            }
+        }
+    }
+
+    // flip the negative values to largest first
+    int numNegPics = numberOfNegativePictures;
+    for (int j = 0, k = numNegPics - 1; j < numNegPics >> 1; j++, k--)
+    {
+        int dPOC = deltaPOC[j];
+        bool used = bUsed[j];
+        deltaPOC[j] = deltaPOC[k];
+        bUsed[j] = bUsed[k];
+        deltaPOC[k] = dPOC;
+        bUsed[k] = used;
+    }
+}
+
+uint32_t Slice::realEndAddress(uint32_t endCUAddr)
+{
+    // Calculate end address
+    uint32_t internalAddress = (endCUAddr - 1) % NUM_CU_PARTITIONS;
+    uint32_t externalAddress = (endCUAddr - 1) / NUM_CU_PARTITIONS;
+    uint32_t xmax = m_sps->picWidthInLumaSamples  - (externalAddress % m_pic->getFrameWidthInCU()) * g_maxCUSize;
+    uint32_t ymax = m_sps->picHeightInLumaSamples - (externalAddress / m_pic->getFrameWidthInCU()) * g_maxCUSize;
+
+    while (g_zscanToPelX[internalAddress] >= xmax || g_zscanToPelY[internalAddress] >= ymax)
+        internalAddress--;
+
+    internalAddress++;
+    if (internalAddress == NUM_CU_PARTITIONS)
+    {
+        internalAddress = 0;
+        externalAddress++;
+    }
+
+    return externalAddress * NUM_CU_PARTITIONS + internalAddress;
+}
+
+

x265-hg20140928.tar.bz2/source/common/slice.h Added

@@ -0,0 +1,342 @@
+/*****************************************************************************
+ * Copyright (C) 2014 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_SLICE_H
+#define X265_SLICE_H
+
+#include "common.h"
+
+namespace x265 {
+// private namespace
+
+class Frame;
+class PicList;
+class MotionReference;
+
+struct RPS
+{
+    int  numberOfPictures;
+    int  numberOfNegativePictures;
+    int  numberOfPositivePictures;
+
+    int  poc[MAX_NUM_REF_PICS];
+    int  deltaPOC[MAX_NUM_REF_PICS];
+    bool bUsed[MAX_NUM_REF_PICS];
+
+    RPS()
+        : numberOfPictures(0)
+        , numberOfNegativePictures(0)
+        , numberOfPositivePictures(0)
+    {
+        ::memset(deltaPOC, 0, sizeof(deltaPOC));
+        ::memset(poc, 0, sizeof(poc));
+        ::memset(bUsed, 0, sizeof(bUsed));
+    }
+
+    void sortDeltaPOC();
+};
+
+namespace Profile {
+    enum Name
+    {
+        NONE = 0,
+        MAIN = 1,
+        MAIN10 = 2,
+        MAINSTILLPICTURE = 3,
+    };
+}
+
+namespace Level {
+    enum Tier
+    {
+        MAIN = 0,
+        HIGH = 1,
+    };
+
+    enum Name
+    {
+        NONE = 0,
+        LEVEL1 = 30,
+        LEVEL2 = 60,
+        LEVEL2_1 = 63,
+        LEVEL3 = 90,
+        LEVEL3_1 = 93,
+        LEVEL4 = 120,
+        LEVEL4_1 = 123,
+        LEVEL5 = 150,
+        LEVEL5_1 = 153,
+        LEVEL5_2 = 156,
+        LEVEL6 = 180,
+        LEVEL6_1 = 183,
+        LEVEL6_2 = 186,
+    };
+}
+
+struct ProfileTierLevel
+{
+    bool    tierFlag;
+    int     profileIdc;
+    bool    profileCompatibilityFlag[32];
+    int     levelIdc;
+    bool    progressiveSourceFlag;
+    bool    interlacedSourceFlag;
+    bool    nonPackedConstraintFlag;
+    bool    frameOnlyConstraintFlag;
+};
+
+struct HRDInfo
+{
+    uint32_t bitRateScale;
+    uint32_t cpbSizeScale;
+    uint32_t initialCpbRemovalDelayLength;
+    uint32_t cpbRemovalDelayLength;
+    uint32_t dpbOutputDelayLength;
+    uint32_t bitRateValue;
+    uint32_t cpbSizeValue;
+    bool     cbrFlag;
+
+    HRDInfo()
+        : bitRateScale(0)
+        , cpbSizeScale(0)
+        , initialCpbRemovalDelayLength(1)
+        , cpbRemovalDelayLength(1)
+        , dpbOutputDelayLength(1)
+        , cbrFlag(false)
+    {
+    }
+};
+
+struct TimingInfo
+{
+    uint32_t numUnitsInTick;
+    uint32_t timeScale;
+};
+
+struct VPS
+{
+    uint32_t         numReorderPics;
+    uint32_t         maxDecPicBuffering;
+    HRDInfo          hrdParameters;
+    ProfileTierLevel ptl;
+};
+
+struct Window
+{
+    bool bEnabled;
+    int  leftOffset;
+    int  rightOffset;
+    int  topOffset;
+    int  bottomOffset;
+
+    Window()
+    {
+        bEnabled = false;
+    }
+};
+
+struct VUI
+{
+    bool       aspectRatioInfoPresentFlag;
+    int        aspectRatioIdc;
+    int        sarWidth;
+    int        sarHeight;
+
+    bool       overscanInfoPresentFlag;
+    bool       overscanAppropriateFlag;
+
+    bool       videoSignalTypePresentFlag;
+    int        videoFormat;
+    bool       videoFullRangeFlag;
+
+    bool       colourDescriptionPresentFlag;
+    int        colourPrimaries;
+    int        transferCharacteristics;
+    int        matrixCoefficients;
+
+    bool       chromaLocInfoPresentFlag;
+    int        chromaSampleLocTypeTopField;
+    int        chromaSampleLocTypeBottomField;
+
+    Window     defaultDisplayWindow;
+
+    bool       frameFieldInfoPresentFlag;
+    bool       fieldSeqFlag;
+
+    bool       hrdParametersPresentFlag;
+    HRDInfo    hrdParameters;
+
+    TimingInfo timingInfo;
+};
+
+struct SPS
+{
+    int      chromaFormatIdc;        // use param
+    uint32_t picWidthInLumaSamples;  // use param
+    uint32_t picHeightInLumaSamples; // use param
+
+    int      log2MinCodingBlockSize;
+    int      log2DiffMaxMinCodingBlockSize;
+
+    uint32_t quadtreeTULog2MaxSize;
+    uint32_t quadtreeTULog2MinSize;
+
+    uint32_t quadtreeTUMaxDepthInter; // use param
+    uint32_t quadtreeTUMaxDepthIntra; // use param
+
+    bool     bUseSAO; // use param
+    bool     bUseAMP; // use param
+    uint32_t maxAMPDepth;
+
+    uint32_t maxDecPicBuffering; // these are dups of VPS values
+    int      numReorderPics;
+
+    bool     bUseStrongIntraSmoothing; // use param
+
+    Window   conformanceWindow;
+    VUI      vuiParameters;
+};
+
+struct PPS
+{
+    uint32_t maxCuDQPDepth;
+
+    int      chromaCbQpOffset;       // use param
+    int      chromaCrQpOffset;       // use param
+
+    bool     bUseWeightPred;         // use param
+    bool     bUseWeightedBiPred;     // use param
+    bool     bUseDQP;
+    bool     bConstrainedIntraPred;  // use param
+
+    bool     bTransquantBypassEnabled;  // Indicates presence of cu_transquant_bypass_flag in CUs.
+    bool     bTransformSkipEnabled;     // use param
+    bool     bEntropyCodingSyncEnabled; // use param
+    bool     bSignHideEnabled;          // use param
+
+    bool     bDeblockingFilterControlPresent;
+    bool     bPicDisableDeblockingFilter;
+    int      deblockingFilterBetaOffsetDiv2;
+    int      deblockingFilterTcOffsetDiv2;
+};
+
+struct WeightParam
+{
+    // Explicit weighted prediction parameters parsed in slice header,
+    bool     bPresentFlag;
+    uint32_t log2WeightDenom;
+    int      inputWeight;
+    int      inputOffset;
+
+    /* makes a non-h265 weight (i.e. fix7), into an h265 weight */
+    void setFromWeightAndOffset(int w, int o, int denom, bool bNormalize)
+    {
+        inputOffset = o;
+        log2WeightDenom = denom;
+        inputWeight = w;
+        while (bNormalize && log2WeightDenom > 0 && (inputWeight > 127))
+        {
+            log2WeightDenom--;
+            inputWeight >>= 1;
+        }
+
+        inputWeight = X265_MIN(inputWeight, 127);
+    }
+};
+
+class Slice
+{
+public:
+
+    const SPS*  m_sps;
+    const PPS*  m_pps;
+    Frame*      m_pic;
+    WeightParam m_weightPredTable[2][MAX_NUM_REF][3]; // [list][refIdx][0:Y, 1:U, 2:V]
+    MotionReference (*m_mref)[MAX_NUM_REF + 1];
+    RPS         m_rps;
+
+    NalUnitType m_nalUnitType;
+    SliceType   m_sliceType;
+    int         m_sliceQp;
+    int         m_poc;
+    
+    int         m_lastIDR;
+
+    bool        m_bCheckLDC;       // TODO: is this necessary?
+    bool        m_sLFaseFlag;      // loop filter boundary flag
+    bool        m_colFromL0Flag;   // collocated picture from List0 or List1 flag
+    uint32_t    m_colRefIdx;       // never modified
+    
+    int         m_numRefIdx[2];
+    Frame*      m_refPicList[2][MAX_NUM_REF + 1];
+    int         m_refPOCList[2][MAX_NUM_REF + 1];
+
+    uint32_t    m_maxNumMergeCand; // use param
+    uint32_t    m_endCUAddr;
+
+    Slice()
+    {
+        m_lastIDR = 0;
+        m_sLFaseFlag = true;
+        m_numRefIdx[0] = m_numRefIdx[1] = 0;
+        for (int i = 0; i < MAX_NUM_REF; i++)
+        {
+            m_refPicList[0][i] = NULL;
+            m_refPicList[1][i] = NULL;
+            m_refPOCList[0][i] = 0;
+            m_refPOCList[1][i] = 0;
+        }
+
+        disableWeights();
+    }
+
+    void disableWeights();
+
+    void setRefPicList(PicList& picList);
+
+    bool getRapPicFlag() const
+    {
+        return m_nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL
+            || m_nalUnitType == NAL_UNIT_CODED_SLICE_CRA;
+    }
+
+    bool getIdrPicFlag() const
+    {
+        return m_nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL;
+    }
+
+    bool isIRAP() const   { return m_nalUnitType >= 16 && m_nalUnitType <= 23; }
+
+    bool isIntra()  const { return m_sliceType == I_SLICE; }
+
+    bool isInterB() const { return m_sliceType == B_SLICE; }
+
+    bool isInterP() const { return m_sliceType == P_SLICE; }
+
+    uint32_t realEndAddress(uint32_t endCUAddr);
+};
+
+#define IS_REFERENCED(slice) (slice->m_pic->m_lowres.sliceType != X265_TYPE_B) 
+
+}
+
+#endif // ifndef X265_SLICE_H

x265-hg20140324.tar.bz2/source/common/threading.cpp -> x265-hg20140928.tar.bz2/source/common/threading.cpp Changed

x265-hg20140324.tar.bz2/source/common/threading.h -> x265-hg20140928.tar.bz2/source/common/threading.h Changed

@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com
+ * For more information, contact us at license @ x265.com
  *****************************************************************************/
 
 #ifndef X265_THREADING_H
@@ -31,6 +31,7 @@
 
 #ifdef _WIN32
 #include <windows.h>
+#include "winxp.h"  // XP workarounds for CONDITION_VARIABLE and ATOMIC_OR
 #else
 #include <pthread.h>
 #include <semaphore.h>
@@ -98,29 +99,9 @@
 
 #endif // if !_WIN64
 
-#if _WIN32_WINNT <= _WIN32_WINNT_WINXP
-/* Windows XP did not define this intrinsic */
-FORCEINLINE LONGLONG x265_interlocked_OR64(__inout LONGLONG volatile *Destination,
-                                           __in    LONGLONG           Value)
-{
-    LONGLONG Old;
-
-    do
-    {
-        Old = *Destination;
-    }
-    while (_InterlockedCompareExchange64(Destination, Old | Value, Old) != Old);
-
-    return Old;
-}
-
-#define ATOMIC_OR(ptr, mask)            x265_interlocked_OR64((volatile LONG64*)ptr, mask)
-#if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
-#pragma intrinsic(_InterlockedCompareExchange64)
+#ifndef ATOMIC_OR
+#define ATOMIC_OR(ptr, mask)                InterlockedOr64((volatile LONG64*)ptr, mask)
 #endif
-#else // if _WIN32_WINNT <= _WIN32_WINNT_WINXP
-#define ATOMIC_OR(ptr, mask)            InterlockedOr64((volatile LONG64*)ptr, mask)
-#endif // if _WIN32_WINNT <= _WIN32_WINNT_WINXP
 
 #define CLZ32(id, x)                        _BitScanReverse(&id, x)
 #define CTZ64(id, x)                        _BitScanForward64(&id, x)
@@ -132,7 +113,6 @@
 
 #endif // ifdef __GNUC__
 
-
 namespace x265 {
 // x265 private namespace
 
@@ -223,6 +203,7 @@
     ~ThreadSafeInteger()
     {
         DeleteCriticalSection(&m_cs);
+        XP_CONDITION_VAR_FREE(&m_cv);
     }
 
     int waitForChange(int prev)
@@ -321,10 +302,14 @@
     void wait()
     {
         pthread_mutex_lock(&m_mutex);
+
         /* blocking wait on conditional variable, mutex is atomically released
          * while blocked. When condition is signaled, mutex is re-acquired */
         while (m_counter == 0)
+        {
             pthread_cond_wait(&m_cond, &m_mutex);
+        }
+
         m_counter--;
         pthread_mutex_unlock(&m_mutex);
     }
@@ -332,6 +317,7 @@
     bool timedWait(uint32_t waitms)
     {
         bool bTimedOut = false;
+
         pthread_mutex_lock(&m_mutex);
         if (m_counter == 0)
         {
@@ -375,7 +361,6 @@
     uint32_t        m_counter;
 };
 
-
 /* This class is intended for use in signaling state changes safely between CPU
  * cores. One thread should be a writer and multiple threads may be readers. The
  * mutex's main purpose is to serve as a memory fence to ensure writes made by

x265-hg20140324.tar.bz2/source/common/threadpool.cpp -> x265-hg20140928.tar.bz2/source/common/threadpool.cpp Changed

@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com
+ * For more information, contact us at license @ x265.com
  *****************************************************************************/
 
 #include "common.h"
@@ -97,7 +97,8 @@
 
 public:
 
-    static ThreadPoolImpl *instance;
+    static ThreadPoolImpl *s_instance;
+    static Lock s_createLock;
 
     JobProvider *m_firstProvider;
     JobProvider *m_lastProvider;
@@ -155,7 +156,7 @@
         {
             // FindJob() may perform actual work and return true.  If
             // it does we restart the job search
-            if (cur->findJob() == true)
+            if (cur->findJob(m_id) == true)
                 break;
 
             cur = cur->m_nextProvider;
@@ -206,30 +207,40 @@
     }
 }
 
-ThreadPoolImpl *ThreadPoolImpl::instance;
+ThreadPoolImpl *ThreadPoolImpl::s_instance;
+Lock ThreadPoolImpl::s_createLock;
 
 /* static */
 ThreadPool *ThreadPool::allocThreadPool(int numthreads)
 {
-    if (ThreadPoolImpl::instance)
-        return ThreadPoolImpl::instance->AddReference();
+    if (ThreadPoolImpl::s_instance)
+        return ThreadPoolImpl::s_instance->AddReference();
 
-    ThreadPoolImpl::instance = new ThreadPoolImpl(numthreads);
-    return ThreadPoolImpl::instance;
+    /* acquire the lock to create the instance */
+    ThreadPoolImpl::s_createLock.acquire();
+
+    if (ThreadPoolImpl::s_instance)
+        /* pool was allocated while we waited for the lock */
+        ThreadPoolImpl::s_instance->AddReference();
+    else
+        ThreadPoolImpl::s_instance = new ThreadPoolImpl(numthreads);
+    ThreadPoolImpl::s_createLock.release();
+
+    return ThreadPoolImpl::s_instance;
 }
 
 ThreadPool *ThreadPool::getThreadPool()
 {
-    assert(ThreadPoolImpl::instance);
-    return ThreadPoolImpl::instance;
+    X265_CHECK(ThreadPoolImpl::s_instance, "getThreadPool() called prior to allocThreadPool()\n");
+    return ThreadPoolImpl::s_instance;
 }
 
 void ThreadPoolImpl::release()
 {
     if (--m_referenceCount == 0)
     {
-        assert(this == ThreadPoolImpl::instance);
-        ThreadPoolImpl::instance = NULL;
+        X265_CHECK(this == ThreadPoolImpl::s_instance, "multiple thread pool instances detected\n");
+        ThreadPoolImpl::s_instance = NULL;
         this->Stop();
         delete this;
     }
@@ -413,7 +424,7 @@
 void JobProvider::enqueue()
 {
     // Add this provider to the end of the thread pool's job provider list
-    assert(!m_nextProvider && !m_prevProvider && m_pool);
+    X265_CHECK(!m_nextProvider && !m_prevProvider && m_pool, "job provider was already queued\n");
     m_pool->enqueueJobProvider(*this);
     m_pool->pokeIdleThread();
 }

x265-hg20140324.tar.bz2/source/common/threadpool.h -> x265-hg20140928.tar.bz2/source/common/threadpool.h Changed

x265-hg20140324.tar.bz2/source/common/vec/dct-sse3.cpp -> x265-hg20140928.tar.bz2/source/common/vec/dct-sse3.cpp Changed

x265-hg20140324.tar.bz2/source/common/vec/dct-sse41.cpp -> x265-hg20140928.tar.bz2/source/common/vec/dct-sse41.cpp Changed

@@ -25,11 +25,10 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "primitives.h"
-#include "TLibCommon/TypeDef.h"    // TCoeff, int, uint32_t
 #include "TLibCommon/TComRom.h"
 #include <xmmintrin.h> // SSE
 #include <smmintrin.h> // SSE4.1
@@ -37,9 +36,9 @@
 using namespace x265;
 
 namespace {
-void dequant_scaling(const int32_t* quantCoef, const int32_t *deQuantCoef, int32_t* coef, int num, int per, int shift)
+void dequant_scaling(const int16_t* quantCoef, const int32_t *deQuantCoef, int32_t* coef, int num, int per, int shift)
 {
-    assert(num <= 32 * 32);
+    X265_CHECK(num <= 32 * 32, "dequant num too large\n");
 
     int valueToAdd;
 
@@ -54,13 +53,11 @@
         {
             __m128i quantCoef1, quantCoef2, deQuantCoef1, deQuantCoef2, quantCoef12, sign;
 
-            quantCoef1 = _mm_loadu_si128((__m128i*)(quantCoef + n));
-            quantCoef2 = _mm_loadu_si128((__m128i*)(quantCoef + n + 4));
+            quantCoef12 = _mm_loadu_si128((__m128i*)(quantCoef + n));
 
             deQuantCoef1 = _mm_loadu_si128((__m128i*)(deQuantCoef + n));
             deQuantCoef2 = _mm_loadu_si128((__m128i*)(deQuantCoef + n + 4));
 
-            quantCoef12 = _mm_packs_epi32(quantCoef1, quantCoef2);
             sign = _mm_srai_epi16(quantCoef12, 15);
             quantCoef1 = _mm_unpacklo_epi16(quantCoef12, sign);
             quantCoef2 = _mm_unpackhi_epi16(quantCoef12, sign);
@@ -82,13 +79,11 @@
         {
             __m128i quantCoef1, quantCoef2, deQuantCoef1, deQuantCoef2, quantCoef12, sign;
 
-            quantCoef1 = _mm_loadu_si128((__m128i*)(quantCoef + n));
-            quantCoef2 = _mm_loadu_si128((__m128i*)(quantCoef + n + 4));
+            quantCoef12 = _mm_loadu_si128((__m128i*)(quantCoef + n));
 
             deQuantCoef1 = _mm_loadu_si128((__m128i*)(deQuantCoef + n));
             deQuantCoef2 = _mm_loadu_si128((__m128i*)(deQuantCoef + n + 4));
 
-            quantCoef12 = _mm_packs_epi32(quantCoef1, quantCoef2);
             sign = _mm_srai_epi16(quantCoef12, 15);
             quantCoef1 = _mm_unpacklo_epi16(quantCoef12, sign);
             quantCoef2 = _mm_unpackhi_epi16(quantCoef12, sign);

x265-hg20140324.tar.bz2/source/common/vec/dct-ssse3.cpp -> x265-hg20140928.tar.bz2/source/common/vec/dct-ssse3.cpp Changed

x265-hg20140324.tar.bz2/source/common/vec/vec-primitives.cpp -> x265-hg20140928.tar.bz2/source/common/vec/vec-primitives.cpp Changed

x265-hg20140324.tar.bz2/source/common/version.cpp -> x265-hg20140928.tar.bz2/source/common/version.cpp Changed

x265-hg20140324.tar.bz2/source/common/wavefront.cpp -> x265-hg20140928.tar.bz2/source/common/wavefront.cpp Changed

x265-hg20140324.tar.bz2/source/common/wavefront.h -> x265-hg20140928.tar.bz2/source/common/wavefront.h Changed

x265-hg20140928.tar.bz2/source/common/winxp.cpp Added

@@ -0,0 +1,130 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com
+ *****************************************************************************/
+
+#include "threading.h"
+
+#if defined(_WIN32) && (_WIN32_WINNT < 0x0600) // _WIN32_WINNT_VISTA
+
+namespace x265 {
+/* Mimic CONDITION_VARIABLE functions only supported on Vista+ */
+
+int WINAPI cond_init(ConditionVariable *cond)
+{ // InitializeConditionVariable
+    cond->semaphore = CreateSemaphore(NULL, 0, 0x7fffffff, NULL);
+    if (!cond->semaphore)
+        return -1;
+    cond->waitersDone = CreateEvent(NULL, FALSE, FALSE, NULL);
+    if (!cond->waitersDone)
+        return -1;
+
+    InitializeCriticalSection(&cond->waiterCountMutex);
+    InitializeCriticalSection(&cond->broadcastMutex);
+    cond->waiterCount = 0;
+    cond->bIsBroadcast = false;
+
+    return 0;
+}
+
+void WINAPI cond_broadcast(ConditionVariable *cond)
+{ // WakeAllConditionVariable
+    EnterCriticalSection(&cond->broadcastMutex);
+    EnterCriticalSection(&cond->waiterCountMutex);
+    int haveWaiter = 0;
+
+    if (cond->waiterCount)
+    {
+        cond->bIsBroadcast = 1;
+        haveWaiter = 1;
+    }
+
+    if (haveWaiter)
+    {
+        ReleaseSemaphore(cond->semaphore, cond->waiterCount, NULL);
+        LeaveCriticalSection(&cond->waiterCountMutex);
+        WaitForSingleObject(cond->waitersDone, INFINITE);
+        cond->bIsBroadcast = 0;
+    }
+    else
+        LeaveCriticalSection(&cond->waiterCountMutex);
+
+    LeaveCriticalSection(&cond->broadcastMutex);
+}
+
+void WINAPI cond_signal(ConditionVariable *cond)
+{ // WakeConditionVariable
+    EnterCriticalSection(&cond->broadcastMutex);
+    EnterCriticalSection(&cond->waiterCountMutex);
+    int haveWaiter = cond->waiterCount;
+    LeaveCriticalSection(&cond->waiterCountMutex);
+
+    if (haveWaiter)
+    {
+        ReleaseSemaphore(cond->semaphore, 1, NULL);
+        WaitForSingleObject(cond->waitersDone, INFINITE);
+    }
+
+    LeaveCriticalSection(&cond->broadcastMutex);
+}
+
+BOOL WINAPI cond_wait(ConditionVariable *cond, CRITICAL_SECTION *mutex, DWORD wait)
+{ // SleepConditionVariableCS
+    EnterCriticalSection(&cond->broadcastMutex);
+    EnterCriticalSection(&cond->waiterCountMutex);
+    cond->waiterCount++;
+    LeaveCriticalSection(&cond->waiterCountMutex);
+    LeaveCriticalSection(&cond->broadcastMutex);
+
+    // unlock the external mutex
+    LeaveCriticalSection(mutex);
+    BOOL ret = WaitForSingleObject(cond->semaphore, wait);
+
+    EnterCriticalSection(&cond->waiterCountMutex);
+    cond->waiterCount--;
+    int last_waiter = !cond->waiterCount || !cond->bIsBroadcast;
+    LeaveCriticalSection(&cond->waiterCountMutex);
+
+    if (last_waiter)
+        SetEvent(cond->waitersDone);
+
+    // lock the external mutex
+    EnterCriticalSection(mutex);
+
+    // returns false on timeout or error
+    return ret;
+}
+
+/* Native CONDITION_VARIABLE instances are not freed, so this is a special case */
+void cond_destroy(ConditionVariable *cond)
+{
+    CloseHandle(cond->semaphore);
+    CloseHandle(cond->waitersDone);
+    DeleteCriticalSection(&cond->broadcastMutex);
+    DeleteCriticalSection(&cond->waiterCountMutex);
+}
+} // namespace x265
+
+#elif defined(_MSC_VER)
+
+namespace { int _avoid_linker_warnings = 0; }
+
+#endif // _WIN32_WINNT <= _WIN32_WINNT_WINXP

x265-hg20140928.tar.bz2/source/common/winxp.h Added

@@ -0,0 +1,91 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com
+ *****************************************************************************/
+
+#ifndef X265_WINXP_H
+#define X265_WINXP_H
+
+#if defined(_WIN32) && (_WIN32_WINNT < 0x0600) // _WIN32_WINNT_VISTA
+
+#ifdef _MSC_VER
+#include <intrin.h> // _InterlockedCompareExchange64
+#endif
+
+namespace x265 {
+/* non-native condition variable */
+typedef struct
+{
+    CRITICAL_SECTION broadcastMutex;
+    CRITICAL_SECTION waiterCountMutex;
+    HANDLE semaphore;
+    HANDLE waitersDone;
+    volatile int waiterCount;
+    volatile int bIsBroadcast;
+} ConditionVariable;
+
+int WINAPI cond_init(ConditionVariable *cond);
+void WINAPI cond_broadcast(ConditionVariable *cond);
+void WINAPI cond_signal(ConditionVariable *cond);
+BOOL WINAPI cond_wait(ConditionVariable *cond, CRITICAL_SECTION *mutex, DWORD wait);
+void cond_destroy(ConditionVariable *cond);
+
+/* map missing API symbols to our structure and functions */
+#define CONDITION_VARIABLE          x265::ConditionVariable
+#define InitializeConditionVariable x265::cond_init
+#define SleepConditionVariableCS    x265::cond_wait
+#define WakeConditionVariable       x265::cond_signal
+#define WakeAllConditionVariable    x265::cond_broadcast
+#define XP_CONDITION_VAR_FREE       x265::cond_destroy
+
+#if defined(_MSC_VER)
+
+/* Windows XP did not define atomic OR 64, but gcc has a good version, so
+ * only use this workaround when targeting XP with MSVC */
+FORCEINLINE LONGLONG interlocked_OR64(__inout LONGLONG volatile *Destination,
+                                      __in    LONGLONG           Value)
+{
+    LONGLONG Old;
+
+    do
+    {
+        Old = *Destination;
+    }
+    while (_InterlockedCompareExchange64(Destination, Old | Value, Old) != Old);
+
+    return Old;
+}
+
+#define ATOMIC_OR(ptr, mask) x265::interlocked_OR64((volatile LONG64*)ptr, mask)
+
+#if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
+#pragma intrinsic(_InterlockedCompareExchange64)
+#endif
+#endif // defined(_MSC_VER)
+} // namespace x265
+
+#else // if defined(_WIN32) && (_WIN32_WINNT < 0x0600)
+
+#define XP_CONDITION_VAR_FREE(x)
+
+#endif // _WIN32_WINNT <= _WIN32_WINNT_WINXP
+
+#endif // ifndef X265_WINXP_H

x265-hg20140324.tar.bz2/source/common/x86/asm-primitives.cpp -> x265-hg20140928.tar.bz2/source/common/x86/asm-primitives.cpp Changed

@@ -20,12 +20,13 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "primitives.h"
 #include "x265.h"
 #include "cpu.h"
+#include "TLibCommon/TComRom.h"
 
 extern "C" {
 #include "pixel.h"
@@ -190,12 +191,14 @@
     p.sse_ss[LUMA_64x64]   = x265_pixel_ssd_ss_64x64_ ## cpu;
 
 #define SA8D_INTER_FROM_BLOCK(cpu) \
-    p.sa8d_inter[LUMA_4x8]  = x265_pixel_satd_4x8_ ## cpu; \
-    p.sa8d_inter[LUMA_8x4]  = x265_pixel_satd_8x4_ ## cpu; \
+    p.sa8d_inter[LUMA_4x8]   = x265_pixel_satd_4x8_ ## cpu; \
+    p.sa8d_inter[LUMA_8x4]   = x265_pixel_satd_8x4_ ## cpu; \
     p.sa8d_inter[LUMA_4x16]  = x265_pixel_satd_4x16_ ## cpu; \
     p.sa8d_inter[LUMA_16x4]  = x265_pixel_satd_16x4_ ## cpu; \
-    p.sa8d_inter[LUMA_12x16]  = x265_pixel_satd_12x16_ ## cpu; \
-    p.sa8d_inter[LUMA_16x12]  = x265_pixel_satd_16x12_ ## cpu; \
+    p.sa8d_inter[LUMA_12x16] = x265_pixel_satd_12x16_ ## cpu; \
+    p.sa8d_inter[LUMA_8x8]   = x265_pixel_sa8d_8x8_ ## cpu; \
+    p.sa8d_inter[LUMA_16x16] = x265_pixel_sa8d_16x16_ ## cpu; \
+    p.sa8d_inter[LUMA_16x12] = x265_pixel_satd_16x12_ ## cpu; \
     p.sa8d_inter[LUMA_16x8]  = x265_pixel_sa8d_16x8_ ## cpu; \
     p.sa8d_inter[LUMA_8x16]  = x265_pixel_sa8d_8x16_ ## cpu; \
     p.sa8d_inter[LUMA_32x24] = x265_pixel_sa8d_32x24_ ## cpu; \
@@ -248,6 +251,12 @@
     p.chroma[X265_CSP_I420].filter_vpp[CHROMA_ ## W ## x ## H] = x265_interp_4tap_vert_pp_ ## W ## x ## H ## cpu; \
     p.chroma[X265_CSP_I420].filter_vps[CHROMA_ ## W ## x ## H] = x265_interp_4tap_vert_ps_ ## W ## x ## H ## cpu;
 
+#define SETUP_CHROMA_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].filter_hpp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_horiz_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_hps[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_horiz_ps_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_vpp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_vps[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_ps_ ## W ## x ## H ## cpu;
+
 #define SETUP_CHROMA_FUNC_DEF_444(W, H, cpu) \
     p.chroma[X265_CSP_I444].filter_hpp[LUMA_ ## W ## x ## H] = x265_interp_4tap_horiz_pp_ ## W ## x ## H ## cpu; \
     p.chroma[X265_CSP_I444].filter_hps[LUMA_ ## W ## x ## H] = x265_interp_4tap_horiz_ps_ ## W ## x ## H ## cpu; \
@@ -257,12 +266,18 @@
 #define SETUP_CHROMA_SP_FUNC_DEF_420(W, H, cpu) \
     p.chroma[X265_CSP_I420].filter_vsp[CHROMA_ ## W ## x ## H] = x265_interp_4tap_vert_sp_ ## W ## x ## H ## cpu;
 
+#define SETUP_CHROMA_SP_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].filter_vsp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_sp_ ## W ## x ## H ## cpu;
+
 #define SETUP_CHROMA_SP_FUNC_DEF_444(W, H, cpu) \
     p.chroma[X265_CSP_I444].filter_vsp[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_sp_ ## W ## x ## H ## cpu;
 
 #define SETUP_CHROMA_SS_FUNC_DEF_420(W, H, cpu) \
     p.chroma[X265_CSP_I420].filter_vss[CHROMA_ ## W ## x ## H] = x265_interp_4tap_vert_ss_ ## W ## x ## H ## cpu;
 
+#define SETUP_CHROMA_SS_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].filter_vss[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_ss_ ## W ## x ## H ## cpu;
+
 #define SETUP_CHROMA_SS_FUNC_DEF_444(W, H, cpu) \
     p.chroma[X265_CSP_I444].filter_vss[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_ss_ ## W ## x ## H ## cpu;
 
@@ -292,8 +307,33 @@
     SETUP_CHROMA_FUNC_DEF_420(32, 8, cpu); \
     SETUP_CHROMA_FUNC_DEF_420(8, 32, cpu);
 
+#define CHROMA_FILTERS_422(cpu) \
+    SETUP_CHROMA_FUNC_DEF_422(4, 8, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(4, 4, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(2, 8, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 8, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(4, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 12, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(6, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 4, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(2, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(16, 8, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(4, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(32, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(32, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_422(8, 64, cpu);
+
 #define CHROMA_FILTERS_444(cpu) \
-    SETUP_CHROMA_FUNC_DEF_444(4, 4, cpu); \
     SETUP_CHROMA_FUNC_DEF_444(8, 8, cpu); \
     SETUP_CHROMA_FUNC_DEF_444(8, 4, cpu); \
     SETUP_CHROMA_FUNC_DEF_444(4, 8, cpu); \
@@ -310,12 +350,22 @@
     SETUP_CHROMA_FUNC_DEF_444(32, 24, cpu); \
     SETUP_CHROMA_FUNC_DEF_444(24, 32, cpu); \
     SETUP_CHROMA_FUNC_DEF_444(32, 8, cpu); \
-    SETUP_CHROMA_FUNC_DEF_444(8, 32, cpu);
+    SETUP_CHROMA_FUNC_DEF_444(8, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(64, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(64, 32, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(32, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(64, 48, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(48, 64, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(64, 16, cpu); \
+    SETUP_CHROMA_FUNC_DEF_444(16, 64, cpu);
 
 #define CHROMA_SP_FILTERS_SSE4_420(cpu) \
     SETUP_CHROMA_SP_FUNC_DEF_420(4, 4, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(4, 2, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_420(2, 4, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(4, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_420(6, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_420(2, 8, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(16, 16, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(16, 8, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(16, 12, cpu); \
@@ -337,8 +387,35 @@
     SETUP_CHROMA_SP_FUNC_DEF_420(8, 16, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_420(8, 32, cpu);
 
+#define CHROMA_SP_FILTERS_SSE4_422(cpu) \
+    SETUP_CHROMA_SP_FUNC_DEF_422(4, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(4, 4, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(2, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(4, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(6, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(2, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(16, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(4, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(32, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(32, 16, cpu);
+
+#define CHROMA_SP_FILTERS_422(cpu) \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 4, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 12, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_422(8, 64, cpu);
+
 #define CHROMA_SP_FILTERS_SSE4_444(cpu) \
-    SETUP_CHROMA_SP_FUNC_DEF_444(4, 4, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_444(4, 8, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_444(16, 16, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_444(16, 8, cpu); \
@@ -351,7 +428,14 @@
     SETUP_CHROMA_SP_FUNC_DEF_444(16, 32, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_444(32, 24, cpu); \
     SETUP_CHROMA_SP_FUNC_DEF_444(24, 32, cpu); \
-    SETUP_CHROMA_SP_FUNC_DEF_444(32, 8, cpu);
+    SETUP_CHROMA_SP_FUNC_DEF_444(32, 8, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(64, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(64, 32, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(32, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(64, 48, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(48, 64, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(64, 16, cpu); \
+    SETUP_CHROMA_SP_FUNC_DEF_444(16, 64, cpu);
 
 #define CHROMA_SP_FILTERS_444(cpu) \
     SETUP_CHROMA_SP_FUNC_DEF_444(8, 8, cpu); \
@@ -387,8 +471,35 @@
     SETUP_CHROMA_SS_FUNC_DEF_420(2, 8, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_420(6, 8, cpu);
 
+#define CHROMA_SS_FILTERS_422(cpu) \
+    SETUP_CHROMA_SS_FUNC_DEF_422(4, 8, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(4, 4, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 8, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(4, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 12, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 4, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(16, 8, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(4, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(32, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(32, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(8, 64, cpu);
+
+#define CHROMA_SS_FILTERS_SSE4_422(cpu) \
+    SETUP_CHROMA_SS_FUNC_DEF_422(2, 8, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(2, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_422(6, 16, cpu);
+
 #define CHROMA_SS_FILTERS_444(cpu) \
-    SETUP_CHROMA_SS_FUNC_DEF_444(4, 4, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_444(8, 8, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_444(8, 4, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_444(4, 8, cpu); \
@@ -405,7 +516,14 @@
     SETUP_CHROMA_SS_FUNC_DEF_444(32, 24, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_444(24, 32, cpu); \
     SETUP_CHROMA_SS_FUNC_DEF_444(32, 8, cpu); \
-    SETUP_CHROMA_SS_FUNC_DEF_444(8, 32, cpu);
+    SETUP_CHROMA_SS_FUNC_DEF_444(8, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(64, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(64, 32, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(32, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(64, 48, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(48, 64, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(64, 16, cpu); \
+    SETUP_CHROMA_SS_FUNC_DEF_444(16, 64, cpu);
 
 #if HIGH_BIT_DEPTH    // temporary, until all 10bit functions are completed
 #define SETUP_LUMA_FUNC_DEF(W, H, cpu) \
@@ -420,7 +538,7 @@
     p.luma_hps[LUMA_ ## W ## x ## H] = x265_interp_8tap_horiz_ps_ ## W ## x ## H ## cpu; \
     p.luma_vpp[LUMA_ ## W ## x ## H] = x265_interp_8tap_vert_pp_ ## W ## x ## H ## cpu; \
     p.luma_vps[LUMA_ ## W ## x ## H] = x265_interp_8tap_vert_ps_ ## W ## x ## H ## cpu;
-#endif
+#endif // if HIGH_BIT_DEPTH
 
 #define SETUP_LUMA_SUB_FUNC_DEF(W, H, cpu) \
     p.luma_sub_ps[LUMA_ ## W ## x ## H] = x265_pixel_sub_ps_ ## W ## x ## H ## cpu; \
@@ -436,9 +554,9 @@
     p.luma_copy_ ## type[LUMA_ ## W ## x ## H] = x265_blockcopy_ ## type ## _ ## W ## x ## H ## cpu;
 
 #define SETUP_CHROMA_BLOCKCOPY(type, W, H, cpu) \
-    p.chroma[X265_CSP_I420].copy_ ## type[CHROMA_ ## W ## x ## H] = x265_blockcopy_ ## type ##_ ## W ## x ## H ## cpu;
+    p.chroma[X265_CSP_I420].copy_ ## type[CHROMA_ ## W ## x ## H] = x265_blockcopy_ ## type ## _ ## W ## x ## H ## cpu;
 
-#define CHROMA_BLOCKCOPY(type ,cpu) \
+#define CHROMA_BLOCKCOPY(type, cpu) \
     SETUP_CHROMA_BLOCKCOPY(type, 2,  4,  cpu); \
     SETUP_CHROMA_BLOCKCOPY(type, 2,  8,  cpu); \
     SETUP_CHROMA_BLOCKCOPY(type, 4,  2,  cpu); \
@@ -464,6 +582,35 @@
     SETUP_CHROMA_BLOCKCOPY(type, 32, 24, cpu); \
     SETUP_CHROMA_BLOCKCOPY(type, 32, 32, cpu);
 
+#define SETUP_CHROMA_BLOCKCOPY_422(type, W, H, cpu) \
+    p.chroma[X265_CSP_I422].copy_ ## type[CHROMA422_ ## W ## x ## H] = x265_blockcopy_ ## type ## _ ## W ## x ## H ## cpu;
+
+#define CHROMA_BLOCKCOPY_422(type, cpu) \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 2,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 2, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 4,  4,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 4,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 4, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 4, 32,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 6, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8,  4,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8, 12,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8, 32,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 8, 64,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 12, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 16,  8, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 16, 16, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 16, 24, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 16, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 16, 64, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 24, 64, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 32, 16, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 32, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 32, 48, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_422(type, 32, 64, cpu);
+
 #define LUMA_BLOCKCOPY(type, cpu) \
     SETUP_LUMA_BLOCKCOPY(type, 4,   4, cpu); \
     SETUP_LUMA_BLOCKCOPY(type, 8,   8, cpu); \
@@ -495,10 +642,13 @@
     p.chroma[X265_CSP_I420].copy_sp[CHROMA_ ## W ## x ## H] = x265_blockcopy_sp_ ## W ## x ## H ## cpu;
 
 #define CHROMA_BLOCKCOPY_SP(cpu) \
+    SETUP_CHROMA_BLOCKCOPY_SP(2,  4,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP(2,  8,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(4,  2,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(4,  4,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(4,  8,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(4,  16, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP(6,  8,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(8,  2,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(8,  4,  cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(8,  6,  cpu); \
@@ -517,36 +667,54 @@
     SETUP_CHROMA_BLOCKCOPY_SP(32, 24, cpu); \
     SETUP_CHROMA_BLOCKCOPY_SP(32, 32, cpu);
 
-
-#define SETUP_CHROMA_LUMA(W1, H1, W2, H2, cpu) \
-    p.chroma[X265_CSP_I420].sub_ps[LUMA_ ## W1 ## x ## H1] = x265_pixel_sub_ps_ ## W2 ## x ## H2 ## cpu; \
-    p.chroma[X265_CSP_I420].add_ps[LUMA_ ## W1 ## x ## H1] = x265_pixel_add_ps_ ## W2 ## x ## H2 ## cpu;
+#define SETUP_CHROMA_BLOCKCOPY_SP_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].copy_sp[CHROMA422_ ## W ## x ## H] = x265_blockcopy_sp_ ## W ## x ## H ## cpu;
+
+#define CHROMA_BLOCKCOPY_SP_422(cpu) \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(2,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(2, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(4,  4,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(4,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(4, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(4,  32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(6, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8,  4,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8,  8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8, 12,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8, 16,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8,  32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(8,  64, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(12, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(16, 8,  cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(16, 16, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(16, 24, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(16, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(16, 64, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(24, 64, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(32, 16, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(32, 32, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(32, 48, cpu); \
+    SETUP_CHROMA_BLOCKCOPY_SP_422(32, 64, cpu);
+
+#define SETUP_CHROMA_PIXELSUB(W, H, cpu) \
+    p.chroma[X265_CSP_I420].sub_ps[CHROMA_ ## W ## x ## H] = x265_pixel_sub_ps_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I420].add_ps[CHROMA_ ## W ## x ## H] = x265_pixel_add_ps_ ## W ## x ## H ## cpu;
 
 #define CHROMA_PIXELSUB_PS(cpu) \
-    SETUP_CHROMA_LUMA(8,   8, 4,  4,  cpu); \
-    SETUP_CHROMA_LUMA(8,   4, 4,  2,  cpu); \
-    SETUP_CHROMA_LUMA(4,   8, 2,  4,  cpu); \
-    SETUP_CHROMA_LUMA(16, 16, 8,  8,  cpu); \
-    SETUP_CHROMA_LUMA(16,  8, 8,  4,  cpu); \
-    SETUP_CHROMA_LUMA(8,  16, 4,  8,  cpu); \
-    SETUP_CHROMA_LUMA(16, 12, 8,  6,  cpu); \
-    SETUP_CHROMA_LUMA(12, 16, 6,  8,  cpu); \
-    SETUP_CHROMA_LUMA(16,  4, 8,  2,  cpu); \
-    SETUP_CHROMA_LUMA(4,  16, 2,  8,  cpu); \
-    SETUP_CHROMA_LUMA(32, 32, 16, 16, cpu); \
-    SETUP_CHROMA_LUMA(32, 16, 16, 8,  cpu); \
-    SETUP_CHROMA_LUMA(16, 32, 8,  16, cpu); \
-    SETUP_CHROMA_LUMA(32, 24, 16, 12, cpu); \
-    SETUP_CHROMA_LUMA(24, 32, 12, 16, cpu); \
-    SETUP_CHROMA_LUMA(32,  8, 16, 4,  cpu); \
-    SETUP_CHROMA_LUMA(8,  32, 4,  16, cpu); \
-    SETUP_CHROMA_LUMA(64, 64, 32, 32, cpu); \
-    SETUP_CHROMA_LUMA(64, 32, 32, 16, cpu); \
-    SETUP_CHROMA_LUMA(32, 64, 16, 32, cpu); \
-    SETUP_CHROMA_LUMA(64, 48, 32, 24, cpu); \
-    SETUP_CHROMA_LUMA(48, 64, 24, 32, cpu); \
-    SETUP_CHROMA_LUMA(64, 16, 32, 8,  cpu); \
-    SETUP_CHROMA_LUMA(16, 64, 8,  32, cpu);
+    SETUP_CHROMA_PIXELSUB(4,  4,  cpu); \
+    SETUP_CHROMA_PIXELSUB(8,  8,  cpu); \
+    SETUP_CHROMA_PIXELSUB(16, 16, cpu); \
+    SETUP_CHROMA_PIXELSUB(32, 32, cpu);
+
+#define SETUP_CHROMA_PIXELSUB_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].sub_ps[CHROMA422_ ## W ## x ## H] = x265_pixel_sub_ps_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].add_ps[CHROMA422_ ## W ## x ## H] = x265_pixel_add_ps_ ## W ## x ## H ## cpu;
+
+#define CHROMA_PIXELSUB_PS_422(cpu) \
+    SETUP_CHROMA_PIXELSUB_422(4,  8,  cpu); \
+    SETUP_CHROMA_PIXELSUB_422(8, 16,  cpu); \
+    SETUP_CHROMA_PIXELSUB_422(16, 32, cpu); \
+    SETUP_CHROMA_PIXELSUB_422(32, 64, cpu);
 
 #define LUMA_FILTERS(cpu) \
     SETUP_LUMA_FUNC_DEF(4,   4, cpu); \
@@ -578,29 +746,9 @@
 #define LUMA_PIXELSUB(cpu) \
     SETUP_LUMA_SUB_FUNC_DEF(4,   4, cpu); \
     SETUP_LUMA_SUB_FUNC_DEF(8,   8, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(8,   4, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(4,   8, cpu); \
     SETUP_LUMA_SUB_FUNC_DEF(16, 16, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(16,  8, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(8,  16, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(16, 12, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(12, 16, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(16,  4, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(4,  16, cpu); \
     SETUP_LUMA_SUB_FUNC_DEF(32, 32, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(32, 16, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(16, 32, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(32, 24, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(24, 32, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(32,  8, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(8,  32, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(64, 64, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(64, 32, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(32, 64, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(64, 48, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(48, 64, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(64, 16, cpu); \
-    SETUP_LUMA_SUB_FUNC_DEF(16, 64, cpu);
+    SETUP_LUMA_SUB_FUNC_DEF(64, 64, cpu);
 
 #define LUMA_SP_FILTERS(cpu) \
     SETUP_LUMA_SP_FUNC_DEF(4,   4, cpu); \
@@ -752,29 +900,58 @@
     SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 8,  cpu); \
     SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 16, cpu); \
     SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 24, cpu); \
-    SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 32, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF(32, 32, cpu);
+
+#define SETUP_CHROMA_ADDAVG_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].addAvg[CHROMA422_ ## W ## x ## H] = x265_addAvg_ ## W ## x ## H ## cpu;
+
+#define CHROMA_ADDAVG_422(cpu) \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(2,  8,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(2, 16,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(4,  4,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(4,  8,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(4, 16,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(4, 32,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(6, 16,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8,  4,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8,  8,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8, 12,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8, 16,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8, 32,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(8, 64,  cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(16,  8, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(32, 16, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_ADDAVG_FUNC_DEF_422(32, 64, cpu);
 
 #define SETUP_INTRA_ANG_COMMON(mode, fno, cpu) \
-    p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
+    p.intra_pred[mode][BLOCK_4x4] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_8x8] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_16x16] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_32x32] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
 
 #define SETUP_INTRA_ANG(mode, fno, cpu) \
-    p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
+    p.intra_pred[mode][BLOCK_8x8] = x265_intra_pred_ang8_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_16x16] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_32x32] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
 
 #define SETUP_INTRA_ANG4(mode, fno, cpu) \
-    p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu;
+    p.intra_pred[mode][BLOCK_4x4] = x265_intra_pred_ang4_ ## fno ## _ ## cpu;
 
 #define SETUP_INTRA_ANG16_32(mode, fno, cpu) \
-    p.intra_pred[BLOCK_16x16][mode] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_32x32][mode] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
+    p.intra_pred[mode][BLOCK_16x16] = x265_intra_pred_ang16_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_32x32] = x265_intra_pred_ang32_ ## fno ## _ ## cpu;
 
 #define SETUP_INTRA_ANG4_8(mode, fno, cpu) \
-    p.intra_pred[BLOCK_4x4][mode] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \
-    p.intra_pred[BLOCK_8x8][mode] = x265_intra_pred_ang8_ ## fno ## _ ## cpu;
+    p.intra_pred[mode][BLOCK_4x4] = x265_intra_pred_ang4_ ## fno ## _ ## cpu; \
+    p.intra_pred[mode][BLOCK_8x8] = x265_intra_pred_ang8_ ## fno ## _ ## cpu;
 
 #define INTRA_ANG_SSSE3(cpu) \
     SETUP_INTRA_ANG_COMMON(2, 2, cpu); \
@@ -817,7 +994,7 @@
     SETUP_INTRA_ANG4(19, 17, cpu); \
     SETUP_INTRA_ANG4(20, 16, cpu); \
     SETUP_INTRA_ANG4(21, 15, cpu); \
-    SETUP_INTRA_ANG4(22, 14, cpu);\
+    SETUP_INTRA_ANG4(22, 14, cpu); \
     SETUP_INTRA_ANG4(23, 13, cpu); \
     SETUP_INTRA_ANG4(24, 12, cpu); \
     SETUP_INTRA_ANG4(25, 11, cpu); \
@@ -834,7 +1011,7 @@
     SETUP_INTRA_ANG4_8(19, 17, cpu); \
     SETUP_INTRA_ANG4_8(20, 16, cpu); \
     SETUP_INTRA_ANG4_8(21, 15, cpu); \
-    SETUP_INTRA_ANG4_8(22, 14, cpu);\
+    SETUP_INTRA_ANG4_8(22, 14, cpu); \
     SETUP_INTRA_ANG4_8(23, 13, cpu); \
     SETUP_INTRA_ANG4_8(24, 12, cpu); \
     SETUP_INTRA_ANG4_8(25, 11, cpu); \
@@ -870,7 +1047,6 @@
 
 #define CHROMA_VERT_FILTERS(cpu) \
     SETUP_CHROMA_VERT_FUNC_DEF(4, 4, cpu); \
-    SETUP_CHROMA_VERT_FUNC_DEF(4, 2, cpu); \
     SETUP_CHROMA_VERT_FUNC_DEF(8, 8, cpu); \
     SETUP_CHROMA_VERT_FUNC_DEF(8, 4, cpu); \
     SETUP_CHROMA_VERT_FUNC_DEF(4, 8, cpu); \
@@ -894,8 +1070,75 @@
 #define CHROMA_VERT_FILTERS_SSE4(cpu) \
     SETUP_CHROMA_VERT_FUNC_DEF(2, 4, cpu); \
     SETUP_CHROMA_VERT_FUNC_DEF(2, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF(4, 2, cpu); \
     SETUP_CHROMA_VERT_FUNC_DEF(6, 8, cpu);
 
+#define SETUP_CHROMA_VERT_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].filter_vss[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_ss_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_vpp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_vps[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_ps_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_vsp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_vert_sp_ ## W ## x ## H ## cpu;
+
+#define CHROMA_VERT_FILTERS_422(cpu) \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(4, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(4, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 12, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 4, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(16, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(4, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(32, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(32, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(8, 64, cpu);
+
+#define CHROMA_VERT_FILTERS_SSE4_422(cpu) \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(2, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(2, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(4, 4, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_422(6, 16, cpu);
+
+#define SETUP_CHROMA_VERT_FUNC_DEF_444(W, H, cpu) \
+    p.chroma[X265_CSP_I444].filter_vss[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_ss_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I444].filter_vpp[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I444].filter_vps[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_ps_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I444].filter_vsp[LUMA_ ## W ## x ## H] = x265_interp_4tap_vert_sp_ ## W ## x ## H ## cpu;
+
+#define CHROMA_VERT_FILTERS_444(cpu) \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(8, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(8, 4, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(4, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(8, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 12, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(12, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 4, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(4, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(32, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(32, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(32, 24, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(24, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(32, 8, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(8, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(64, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(64, 32, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(32, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(64, 48, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(48, 64, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(64, 16, cpu); \
+    SETUP_CHROMA_VERT_FUNC_DEF_444(16, 64, cpu);
+
 #define SETUP_CHROMA_HORIZ_FUNC_DEF(W, H, cpu) \
     p.chroma[X265_CSP_I420].filter_hpp[CHROMA_ ## W ## x ## H] = x265_interp_4tap_horiz_pp_ ## W ## x ## H ## cpu; \
     p.chroma[X265_CSP_I420].filter_hps[CHROMA_ ## W ## x ## H] = x265_interp_4tap_horiz_ps_ ## W ## x ## H ## cpu;
@@ -924,10 +1167,97 @@
     SETUP_CHROMA_HORIZ_FUNC_DEF(32, 24, cpu); \
     SETUP_CHROMA_HORIZ_FUNC_DEF(24, 32, cpu); \
     SETUP_CHROMA_HORIZ_FUNC_DEF(32, 8, cpu); \
-    SETUP_CHROMA_HORIZ_FUNC_DEF(8, 32, cpu)
+    SETUP_CHROMA_HORIZ_FUNC_DEF(8, 32, cpu);
+
+#define SETUP_CHROMA_HORIZ_FUNC_DEF_422(W, H, cpu) \
+    p.chroma[X265_CSP_I422].filter_hpp[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_horiz_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I422].filter_hps[CHROMA422_ ## W ## x ## H] = x265_interp_4tap_horiz_ps_ ## W ## x ## H ## cpu;
+
+#define CHROMA_HORIZ_FILTERS_422(cpu) \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(4, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(4, 4, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(2, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(4, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 12, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(6, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 4, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(2, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(16, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(16, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(16, 24, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(12, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(16, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(4, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(32, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(32, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(16, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(32, 48, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(24, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(32, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_422(8, 64, cpu);
+
+#define SETUP_CHROMA_HORIZ_FUNC_DEF_444(W, H, cpu) \
+    p.chroma[X265_CSP_I444].filter_hpp[LUMA_ ## W ## x ## H] = x265_interp_4tap_horiz_pp_ ## W ## x ## H ## cpu; \
+    p.chroma[X265_CSP_I444].filter_hps[LUMA_ ## W ## x ## H] = x265_interp_4tap_horiz_ps_ ## W ## x ## H ## cpu;
+
+#define CHROMA_HORIZ_FILTERS_444(cpu) \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(8, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(8, 4, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(4, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(8, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 12, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(12, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 4, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(4, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(32, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(32, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(32, 24, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(24, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(32, 8, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(8, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(64, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(64, 32, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(32, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(64, 48, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(48, 64, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(64, 16, cpu); \
+    SETUP_CHROMA_HORIZ_FUNC_DEF_444(16, 64, cpu);
 
 namespace x265 {
 // private x265 namespace
+
+#if HIGH_BIT_DEPTH
+/* Very similar to CRef in intrapred.cpp, except it uses optimized primitives */
+template<int log2Size>
+void intra_allangs(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma)
+{
+    const int size = 1 << log2Size;
+    const int sizeIdx = log2Size - 2;
+    ALIGN_VAR_32(pixel, buffer[32 * 32]);
+
+    for (int mode = 2; mode <= 34; mode++)
+    {
+        pixel *left  = (g_intraFilterFlags[mode] & size ? left1  : left0);
+        pixel *above = (g_intraFilterFlags[mode] & size ? above1 : above0);
+        pixel *out = dest + ((mode - 2) << (log2Size * 2));
+
+        if (mode < 18)
+        {
+            primitives.intra_pred[mode][sizeIdx](buffer, size, left, above, mode, bLuma);
+            primitives.transpose[sizeIdx](out, buffer, size);
+        }
+        else
+            primitives.intra_pred[mode][sizeIdx](out, size, left, above, mode, bLuma);
+    }
+}
+#endif
+
 void Setup_Assembly_Primitives(EncoderPrimitives &p, int cpuMask)
 {
 #if HIGH_BIT_DEPTH
@@ -1007,22 +1337,38 @@
         p.sad_x4[LUMA_12x16] = x265_pixel_sad_x4_12x16_mmx2;
 
         p.cvt32to16_shr = x265_cvt32to16_shr_sse2;
+        p.cvt32to16_shl[BLOCK_4x4] = x265_cvt32to16_shl_4_sse2;
+        p.cvt32to16_shl[BLOCK_8x8] = x265_cvt32to16_shl_8_sse2;
+        p.cvt32to16_shl[BLOCK_16x16] = x265_cvt32to16_shl_16_sse2;
+        p.cvt32to16_shl[BLOCK_32x32] = x265_cvt32to16_shl_32_sse2;
 
         CHROMA_PIXELSUB_PS(_sse2);
+        CHROMA_PIXELSUB_PS_422(_sse2);
         LUMA_PIXELSUB(_sse2);
 
         CHROMA_BLOCKCOPY(ss, _sse2);
+        CHROMA_BLOCKCOPY_422(ss, _sse2);
         LUMA_BLOCKCOPY(ss, _sse2);
 
         CHROMA_VERT_FILTERS(_sse2);
-        p.chroma_p2s[X265_CSP_I420] = x265_chroma_p2s_sse2;
+        CHROMA_VERT_FILTERS_422(_sse2);
+        CHROMA_VERT_FILTERS_444(_sse2);
         p.luma_p2s = x265_luma_p2s_sse2;
+        p.chroma_p2s[X265_CSP_I420] = x265_chroma_p2s_sse2;
+        p.chroma_p2s[X265_CSP_I422] = x265_chroma_p2s_sse2;
+        p.chroma_p2s[X265_CSP_I444] = x265_luma_p2s_sse2; // for i444 , chroma_p2s can be replaced by luma_p2s
 
         p.blockfill_s[BLOCK_4x4] = x265_blockfill_s_4x4_sse2;
         p.blockfill_s[BLOCK_8x8] = x265_blockfill_s_8x8_sse2;
         p.blockfill_s[BLOCK_16x16] = x265_blockfill_s_16x16_sse2;
         p.blockfill_s[BLOCK_32x32] = x265_blockfill_s_32x32_sse2;
 
+        // TODO: overflow on 12-bits mode!
+        p.ssd_s[BLOCK_4x4] = x265_pixel_ssd_s_4_sse2;
+        p.ssd_s[BLOCK_8x8] = x265_pixel_ssd_s_8_sse2;
+        p.ssd_s[BLOCK_16x16] = x265_pixel_ssd_s_16_sse2;
+        p.ssd_s[BLOCK_32x32] = x265_pixel_ssd_s_32_sse2;
+
         p.calcresidual[BLOCK_4x4] = x265_getResidual4_sse2;
         p.calcresidual[BLOCK_8x8] = x265_getResidual8_sse2;
         p.calcresidual[BLOCK_16x16] = x265_getResidual16_sse2;
@@ -1036,7 +1382,6 @@
         p.dct[DCT_4x4] = x265_dct4_sse2;
         p.idct[IDCT_4x4] = x265_idct4_sse2;
         p.idct[IDST_4x4] = x265_idst4_sse2;
-        p.count_nonzero = x265_count_nonzero_sse2;
 
         LUMA_SS_FILTERS(_sse2);
     }
@@ -1049,40 +1394,65 @@
 
         p.dct[DST_4x4] = x265_dst4_ssse3;
         p.idct[IDCT_8x8] = x265_idct8_ssse3;
+        p.count_nonzero = x265_count_nonzero_ssse3;
     }
     if (cpuMask & X265_CPU_SSE4)
     {
         LUMA_ADDAVG(_sse4);
         CHROMA_ADDAVG(_sse4);
+        CHROMA_ADDAVG_422(_sse4);
         LUMA_FILTERS(_sse4);
         CHROMA_HORIZ_FILTERS(_sse4);
         CHROMA_VERT_FILTERS_SSE4(_sse4);
+        CHROMA_HORIZ_FILTERS_422(_sse4);
+        CHROMA_VERT_FILTERS_SSE4_422(_sse4);
+        CHROMA_HORIZ_FILTERS_444(_sse4);
 
         p.dct[DCT_8x8] = x265_dct8_sse4;
         p.quant = x265_quant_sse4;
+        p.nquant = x265_nquant_sse4;
         p.dequant_normal = x265_dequant_normal_sse4;
         p.cvt16to32_shl = x265_cvt16to32_shl_sse4;
-        p.intra_pred[BLOCK_4x4][0] = x265_intra_pred_planar4_sse4;
-        p.intra_pred[BLOCK_8x8][0] = x265_intra_pred_planar8_sse4;
-        p.intra_pred[BLOCK_16x16][0] = x265_intra_pred_planar16_sse4;
-        p.intra_pred[BLOCK_32x32][0] = x265_intra_pred_planar32_sse4;
-
-        p.intra_pred[BLOCK_4x4][1] = x265_intra_pred_dc4_sse4;
-        p.intra_pred[BLOCK_8x8][1] = x265_intra_pred_dc8_sse4;
-        p.intra_pred[BLOCK_16x16][1] = x265_intra_pred_dc16_sse4;
-        p.intra_pred[BLOCK_32x32][1] = x265_intra_pred_dc32_sse4;
+        p.cvt16to32_shr[BLOCK_4x4] = x265_cvt16to32_shr_4_sse4;
+        p.cvt16to32_shr[BLOCK_8x8] = x265_cvt16to32_shr_8_sse4;
+        p.cvt16to32_shr[BLOCK_16x16] = x265_cvt16to32_shr_16_sse4;
+        p.cvt16to32_shr[BLOCK_32x32] = x265_cvt16to32_shr_32_sse4;
+        p.intra_pred[0][BLOCK_4x4] = x265_intra_pred_planar4_sse4;
+        p.intra_pred[0][BLOCK_8x8] = x265_intra_pred_planar8_sse4;
+        p.intra_pred[0][BLOCK_16x16] = x265_intra_pred_planar16_sse4;
+        p.intra_pred[0][BLOCK_32x32] = x265_intra_pred_planar32_sse4;
+
+        p.intra_pred[1][BLOCK_4x4] = x265_intra_pred_dc4_sse4;
+        p.intra_pred[1][BLOCK_8x8] = x265_intra_pred_dc8_sse4;
+        p.intra_pred[1][BLOCK_16x16] = x265_intra_pred_dc16_sse4;
+        p.intra_pred[1][BLOCK_32x32] = x265_intra_pred_dc32_sse4;
+        p.planecopy_cp = x265_upShift_8_sse4;
 
         INTRA_ANG_SSE4_COMMON(sse4);
         INTRA_ANG_SSE4_HIGH(sse4);
-
     }
     if (cpuMask & X265_CPU_XOP)
     {
+        p.frame_init_lowres_core = x265_frame_init_lowres_core_xop;
+        SA8D_INTER_FROM_BLOCK(xop);
+        INIT7(satd, _xop);
+        HEVC_SATD(xop);
     }
     if (cpuMask & X265_CPU_AVX2)
     {
+        p.dct[DCT_4x4] = x265_dct4_avx2;
+        p.quant = x265_quant_avx2;
+        p.nquant = x265_nquant_avx2;
+        p.dequant_normal = x265_dequant_normal_avx2;
+#if X86_64
+        p.dct[DCT_8x8] = x265_dct8_avx2;
+        p.dct[DCT_16x16] = x265_dct16_avx2;
+        p.dct[DCT_32x32] = x265_dct32_avx2;
+        p.idct[IDCT_8x8] = x265_idct8_avx2;
+        p.idct[IDCT_16x16] = x265_idct16_avx2;
+        p.idct[IDCT_32x32] = x265_idct32_avx2;
+#endif
     }
-
     /* at HIGH_BIT_DEPTH, pixel == short so we can reuse a number of primitives */
     for (int i = 0; i < NUM_LUMA_PARTITIONS; i++)
     {
@@ -1104,6 +1474,21 @@
         p.chroma[X265_CSP_I420].copy_pp[i] = (copy_pp_t)p.chroma[X265_CSP_I420].copy_ss[i];
     }
 
+    for (int i = 0; i < NUM_CHROMA_PARTITIONS; i++)
+    {
+        p.chroma[X265_CSP_I422].copy_ps[i] = (copy_ps_t)p.chroma[X265_CSP_I422].copy_ss[i];
+        p.chroma[X265_CSP_I422].copy_sp[i] = (copy_sp_t)p.chroma[X265_CSP_I422].copy_ss[i];
+        p.chroma[X265_CSP_I422].copy_pp[i] = (copy_pp_t)p.chroma[X265_CSP_I422].copy_ss[i];
+    }
+
+    if (p.intra_pred[0][0] && p.transpose[0])
+    {
+        p.intra_pred_allangs[BLOCK_4x4] = intra_allangs<2>;
+        p.intra_pred_allangs[BLOCK_8x8] = intra_allangs<3>;
+        p.intra_pred_allangs[BLOCK_16x16] = intra_allangs<4>;
+        p.intra_pred_allangs[BLOCK_32x32] = intra_allangs<5>;
+    }
+
 #else // if HIGH_BIT_DEPTH
     if (cpuMask & X265_CPU_SSE2)
     {
@@ -1121,6 +1506,7 @@
         LUMA_VAR(_sse2);
 
         ASSGN_SSE(sse2);
+        ASSGN_SSE_SS(sse2);
         INIT2(sad, _sse2);
         SAD(sse2);
         INIT2(sad_x3, _sse2);
@@ -1129,14 +1515,19 @@
 
         CHROMA_BLOCKCOPY(ss, _sse2);
         CHROMA_BLOCKCOPY(pp, _sse2);
+        CHROMA_BLOCKCOPY_422(ss, _sse2);
+        CHROMA_BLOCKCOPY_422(pp, _sse2);
         LUMA_BLOCKCOPY(ss, _sse2);
         LUMA_BLOCKCOPY(pp, _sse2);
         LUMA_BLOCKCOPY(sp, _sse2);
         CHROMA_BLOCKCOPY_SP(_sse2);
+        CHROMA_BLOCKCOPY_SP_422(_sse2);
 
         CHROMA_SS_FILTERS_420(_sse2);
+        CHROMA_SS_FILTERS_422(_sse2);
         CHROMA_SS_FILTERS_444(_sse2);
         CHROMA_SP_FILTERS_420(_sse2);
+        CHROMA_SP_FILTERS_422(_sse2);
         CHROMA_SP_FILTERS_444(_sse2);
         LUMA_SS_FILTERS(_sse2);
 
@@ -1150,10 +1541,19 @@
         p.blockfill_s[BLOCK_16x16] = x265_blockfill_s_16x16_sse2;
         p.blockfill_s[BLOCK_32x32] = x265_blockfill_s_32x32_sse2;
 
+        p.ssd_s[BLOCK_4x4] = x265_pixel_ssd_s_4_sse2;
+        p.ssd_s[BLOCK_8x8] = x265_pixel_ssd_s_8_sse2;
+        p.ssd_s[BLOCK_16x16] = x265_pixel_ssd_s_16_sse2;
+        p.ssd_s[BLOCK_32x32] = x265_pixel_ssd_s_32_sse2;
+
         p.frame_init_lowres_core = x265_frame_init_lowres_core_sse2;
         SA8D_INTER_FROM_BLOCK(sse2);
 
         p.cvt32to16_shr = x265_cvt32to16_shr_sse2;
+        p.cvt32to16_shl[BLOCK_4x4] = x265_cvt32to16_shl_4_sse2;
+        p.cvt32to16_shl[BLOCK_8x8] = x265_cvt32to16_shl_8_sse2;
+        p.cvt32to16_shl[BLOCK_16x16] = x265_cvt32to16_shl_16_sse2;
+        p.cvt32to16_shl[BLOCK_32x32] = x265_cvt32to16_shl_32_sse2;
         p.calcrecon[BLOCK_4x4] = x265_calcRecons4_sse2;
         p.calcrecon[BLOCK_8x8] = x265_calcRecons8_sse2;
         p.calcresidual[BLOCK_4x4] = x265_getResidual4_sse2;
@@ -1168,7 +1568,11 @@
         p.dct[DCT_4x4] = x265_dct4_sse2;
         p.idct[IDCT_4x4] = x265_idct4_sse2;
         p.idct[IDST_4x4] = x265_idst4_sse2;
-        p.count_nonzero = x265_count_nonzero_sse2;
+        p.planecopy_sp = x265_downShift_16_sse2;
+        p.copy_shl[BLOCK_4x4] = x265_copy_shl_4_sse2;
+        p.copy_shl[BLOCK_8x8] = x265_copy_shl_8_sse2;
+        p.copy_shl[BLOCK_16x16] = x265_copy_shl_16_sse2;
+        p.copy_shl[BLOCK_32x32] = x265_copy_shl_32_sse2;
     }
     if (cpuMask & X265_CPU_SSSE3)
     {
@@ -1198,10 +1602,12 @@
         p.luma_hvpp[LUMA_8x8] = x265_interp_8tap_hv_pp_8x8_ssse3;
         p.luma_p2s = x265_luma_p2s_ssse3;
         p.chroma_p2s[X265_CSP_I420] = x265_chroma_p2s_ssse3;
+        p.chroma_p2s[X265_CSP_I422] = x265_chroma_p2s_ssse3;
         p.chroma_p2s[X265_CSP_I444] = x265_luma_p2s_ssse3; // for i444 , chroma_p2s can be replaced by luma_p2s
 
         p.dct[DST_4x4] = x265_dst4_ssse3;
         p.idct[IDCT_8x8] = x265_idct8_ssse3;
+        p.count_nonzero = x265_count_nonzero_ssse3;
     }
     if (cpuMask & X265_CPU_SSE4)
     {
@@ -1209,7 +1615,18 @@
 
         LUMA_ADDAVG(_sse4);
         CHROMA_ADDAVG(_sse4);
+        CHROMA_ADDAVG_422(_sse4);
         p.cvt16to32_shl = x265_cvt16to32_shl_sse4;
+        p.cvt16to32_shr[BLOCK_4x4] = x265_cvt16to32_shr_4_sse4;
+        p.cvt16to32_shr[BLOCK_8x8] = x265_cvt16to32_shr_8_sse4;
+        p.cvt16to32_shr[BLOCK_16x16] = x265_cvt16to32_shr_16_sse4;
+        p.cvt16to32_shr[BLOCK_32x32] = x265_cvt16to32_shr_32_sse4;
+
+        // TODO: check POPCNT flag!
+        p.copy_cnt[BLOCK_4x4] = x265_copy_cnt_4_sse4;
+        p.copy_cnt[BLOCK_8x8] = x265_copy_cnt_8_sse4;
+        p.copy_cnt[BLOCK_16x16] = x265_copy_cnt_16_sse4;
+        p.copy_cnt[BLOCK_32x32] = x265_copy_cnt_32_sse4;
 
         HEVC_SATD(sse4);
         SA8D_INTER_FROM_BLOCK(sse4);
@@ -1225,13 +1642,17 @@
         LUMA_SSE_SP(_sse4);
 
         CHROMA_PIXELSUB_PS(_sse4);
+        CHROMA_PIXELSUB_PS_422(_sse4);
         LUMA_PIXELSUB(_sse4);
 
         CHROMA_FILTERS_420(_sse4);
+        CHROMA_FILTERS_422(_sse4);
         CHROMA_FILTERS_444(_sse4);
         CHROMA_SS_FILTERS_SSE4_420(_sse4);
-        CHROMA_SP_FILTERS_SSE4_444(_sse4);
+        CHROMA_SS_FILTERS_SSE4_422(_sse4);
         CHROMA_SP_FILTERS_SSE4_420(_sse4);
+        CHROMA_SP_FILTERS_SSE4_422(_sse4);
+        CHROMA_SP_FILTERS_SSE4_444(_sse4);
         LUMA_SP_FILTERS(_sse4);
         LUMA_FILTERS(_sse4);
         ASSGN_SSE_SS(sse4);
@@ -1240,39 +1661,39 @@
         p.chroma[X265_CSP_I420].copy_sp[CHROMA_2x8] = x265_blockcopy_sp_2x8_sse4;
         p.chroma[X265_CSP_I420].copy_sp[CHROMA_6x8] = x265_blockcopy_sp_6x8_sse4;
         CHROMA_BLOCKCOPY(ps, _sse4);
+        CHROMA_BLOCKCOPY_422(ps, _sse4);
         LUMA_BLOCKCOPY(ps, _sse4);
 
-        p.chroma[X265_CSP_I420].filter_vsp[CHROMA_2x4] = x265_interp_4tap_vert_sp_2x4_sse4;
-        p.chroma[X265_CSP_I420].filter_vsp[CHROMA_2x8] = x265_interp_4tap_vert_sp_2x8_sse4;
-        p.chroma[X265_CSP_I420].filter_vsp[CHROMA_6x8] = x265_interp_4tap_vert_sp_6x8_sse4;
-
         p.calcrecon[BLOCK_16x16] = x265_calcRecons16_sse4;
         p.calcrecon[BLOCK_32x32] = x265_calcRecons32_sse4;
         p.calcresidual[BLOCK_16x16] = x265_getResidual16_sse4;
         p.calcresidual[BLOCK_32x32] = x265_getResidual32_sse4;
         p.quant = x265_quant_sse4;
+        p.nquant = x265_nquant_sse4;
         p.dequant_normal = x265_dequant_normal_sse4;
         p.weight_pp = x265_weight_pp_sse4;
         p.weight_sp = x265_weight_sp_sse4;
-        p.intra_pred[BLOCK_4x4][0] = x265_intra_pred_planar4_sse4;
-        p.intra_pred[BLOCK_8x8][0] = x265_intra_pred_planar8_sse4;
-        p.intra_pred[BLOCK_16x16][0] = x265_intra_pred_planar16_sse4;
-        p.intra_pred[BLOCK_32x32][0] = x265_intra_pred_planar32_sse4;
+        p.intra_pred[0][BLOCK_4x4] = x265_intra_pred_planar4_sse4;
+        p.intra_pred[0][BLOCK_8x8] = x265_intra_pred_planar8_sse4;
+        p.intra_pred[0][BLOCK_16x16] = x265_intra_pred_planar16_sse4;
+        p.intra_pred[0][BLOCK_32x32] = x265_intra_pred_planar32_sse4;
 
         p.intra_pred_allangs[BLOCK_4x4] = x265_all_angs_pred_4x4_sse4;
         p.intra_pred_allangs[BLOCK_8x8] = x265_all_angs_pred_8x8_sse4;
         p.intra_pred_allangs[BLOCK_16x16] = x265_all_angs_pred_16x16_sse4;
         p.intra_pred_allangs[BLOCK_32x32] = x265_all_angs_pred_32x32_sse4;
 
-        p.intra_pred[BLOCK_4x4][1] = x265_intra_pred_dc4_sse4;
-        p.intra_pred[BLOCK_8x8][1] = x265_intra_pred_dc8_sse4;
-        p.intra_pred[BLOCK_16x16][1] = x265_intra_pred_dc16_sse4;
-        p.intra_pred[BLOCK_32x32][1] = x265_intra_pred_dc32_sse4;
+        p.intra_pred[1][BLOCK_4x4] = x265_intra_pred_dc4_sse4;
+        p.intra_pred[1][BLOCK_8x8] = x265_intra_pred_dc8_sse4;
+        p.intra_pred[1][BLOCK_16x16] = x265_intra_pred_dc16_sse4;
+        p.intra_pred[1][BLOCK_32x32] = x265_intra_pred_dc32_sse4;
 
         INTRA_ANG_SSE4_COMMON(sse4);
         INTRA_ANG_SSE4(sse4);
 
         p.dct[DCT_8x8] = x265_dct8_sse4;
+        p.copy_shr = x265_copy_shr_sse4;
+        p.denoiseDct = x265_denoise_dct_sse4;
     }
     if (cpuMask & X265_CPU_AVX)
     {
@@ -1291,6 +1712,29 @@
 
         p.ssim_4x4x2_core = x265_pixel_ssim_4x4x2_core_avx;
         p.ssim_end_4 = x265_pixel_ssim_end4_avx;
+        p.luma_copy_ss[LUMA_64x16] = x265_blockcopy_ss_64x16_avx;
+        p.luma_copy_ss[LUMA_64x32] = x265_blockcopy_ss_64x32_avx;
+        p.luma_copy_ss[LUMA_64x48] = x265_blockcopy_ss_64x48_avx;
+        p.luma_copy_ss[LUMA_64x64] = x265_blockcopy_ss_64x64_avx;
+
+        p.chroma[X265_CSP_I420].copy_pp[CHROMA_32x8] = x265_blockcopy_pp_32x8_avx;
+        p.luma_copy_pp[LUMA_32x8] = x265_blockcopy_pp_32x8_avx;
+
+        p.chroma[X265_CSP_I420].copy_pp[CHROMA_32x16] = x265_blockcopy_pp_32x16_avx;
+        p.chroma[X265_CSP_I422].copy_pp[CHROMA422_32x16] = x265_blockcopy_pp_32x16_avx;
+        p.luma_copy_pp[LUMA_32x16] = x265_blockcopy_pp_32x16_avx;
+
+        p.chroma[X265_CSP_I420].copy_pp[CHROMA_32x24] = x265_blockcopy_pp_32x24_avx;
+        p.luma_copy_pp[LUMA_32x24] = x265_blockcopy_pp_32x24_avx;
+
+        p.chroma[X265_CSP_I420].copy_pp[CHROMA_32x32] = x265_blockcopy_pp_32x32_avx;
+        p.chroma[X265_CSP_I422].copy_pp[CHROMA422_32x32] = x265_blockcopy_pp_32x32_avx;
+        p.luma_copy_pp[LUMA_32x32]  = x265_blockcopy_pp_32x32_avx;
+
+        p.chroma[X265_CSP_I422].copy_pp[CHROMA422_32x48] = x265_blockcopy_pp_32x48_avx;
+
+        p.chroma[X265_CSP_I422].copy_pp[CHROMA422_32x64] = x265_blockcopy_pp_32x64_avx;
+        p.luma_copy_pp[LUMA_32x64]  = x265_blockcopy_pp_32x64_avx;
     }
     if (cpuMask & X265_CPU_XOP)
     {
@@ -1307,35 +1751,46 @@
         INIT2_NAME(sse_pp, ssd, _avx2);
         p.sad_x4[LUMA_16x12] = x265_pixel_sad_x4_16x12_avx2;
         p.sad_x4[LUMA_16x32] = x265_pixel_sad_x4_16x32_avx2;
+        p.ssd_s[BLOCK_32x32] = x265_pixel_ssd_s_32_avx2;
+
+        /* Need to update assembly code as per changed interface of the copy_cnt primitive, once
+         * code is updated, avx2 version will be enabled */
+
+        p.copy_cnt[BLOCK_8x8] = x265_copy_cnt_8_avx2;
+        p.copy_cnt[BLOCK_16x16] = x265_copy_cnt_16_avx2;
+        p.copy_cnt[BLOCK_32x32] = x265_copy_cnt_32_avx2;
+
+
+        p.cvt32to16_shl[BLOCK_4x4] = x265_cvt32to16_shl_4_avx2;
+        p.cvt32to16_shl[BLOCK_8x8] = x265_cvt32to16_shl_8_avx2;
+        p.cvt32to16_shl[BLOCK_16x16] = x265_cvt32to16_shl_16_avx2;
+        p.cvt32to16_shl[BLOCK_32x32] = x265_cvt32to16_shl_32_avx2;
+        p.denoiseDct = x265_denoise_dct_avx2;
+        p.dct[DCT_4x4] = x265_dct4_avx2;
+        p.quant = x265_quant_avx2;
+        p.nquant = x265_nquant_avx2;
+        p.dequant_normal = x265_dequant_normal_avx2;
+        p.chroma[X265_CSP_I420].copy_ss[CHROMA_16x4] = x265_blockcopy_ss_16x4_avx;
+        p.chroma[X265_CSP_I420].copy_ss[CHROMA_16x12] = x265_blockcopy_ss_16x12_avx;
+        p.chroma[X265_CSP_I420].copy_ss[CHROMA_16x8] = x265_blockcopy_ss_16x8_avx;
+        p.chroma[X265_CSP_I420].copy_ss[CHROMA_16x16] = x265_blockcopy_ss_16x16_avx;
+        p.chroma[X265_CSP_I420].copy_ss[CHROMA_16x32] = x265_blockcopy_ss_16x32_avx;
+        p.chroma[X265_CSP_I422].copy_ss[CHROMA422_16x8] = x265_blockcopy_ss_16x8_avx;
+        p.chroma[X265_CSP_I422].copy_ss[CHROMA422_16x16] = x265_blockcopy_ss_16x16_avx;
+        p.chroma[X265_CSP_I422].copy_ss[CHROMA422_16x24] = x265_blockcopy_ss_16x24_avx;
+        p.chroma[X265_CSP_I422].copy_ss[CHROMA422_16x32] = x265_blockcopy_ss_16x32_avx;
+        p.chroma[X265_CSP_I422].copy_ss[CHROMA422_16x64] = x265_blockcopy_ss_16x64_avx;
+
+#if X86_64
+        p.dct[DCT_8x8] = x265_dct8_avx2;
+        p.dct[DCT_16x16] = x265_dct16_avx2;
+        p.dct[DCT_32x32] = x265_dct32_avx2;
+        p.idct[IDCT_8x8] = x265_idct8_avx2;
+        p.idct[IDCT_16x16] = x265_idct16_avx2;
+        p.idct[IDCT_32x32] = x265_idct32_avx2;
+#endif
     }
 #endif // if HIGH_BIT_DEPTH
-
-    /* copy reusable luma primitives to chroma 4:4:4 */
-    for (int i = 0; i < NUM_LUMA_PARTITIONS; i++)
-    {
-        p.chroma[X265_CSP_I444].copy_pp[i] = p.luma_copy_pp[i];
-        p.chroma[X265_CSP_I444].copy_ps[i] = p.luma_copy_ps[i];
-        p.chroma[X265_CSP_I444].copy_sp[i] = p.luma_copy_sp[i];
-        p.chroma[X265_CSP_I444].copy_ss[i] = p.luma_copy_ss[i];
-        p.chroma[X265_CSP_I444].add_ps[i]  = p.luma_add_ps[i];
-        p.chroma[X265_CSP_I444].sub_ps[i]  = p.luma_sub_ps[i];
-        p.chroma[X265_CSP_I444].addAvg[i]  = p.luma_addAvg[i];
-    }
-
-    primitives.sa8d[BLOCK_4x4]   = primitives.sa8d_inter[LUMA_4x4];
-    primitives.sa8d[BLOCK_8x8]   = primitives.sa8d_inter[LUMA_8x8];
-    primitives.sa8d[BLOCK_16x16] = primitives.sa8d_inter[LUMA_16x16];
-    primitives.sa8d[BLOCK_32x32] = primitives.sa8d_inter[LUMA_32x32];
-    primitives.sa8d[BLOCK_64x64] = primitives.sa8d_inter[LUMA_64x64];
-
-    // SA8D devolves to SATD for blocks not even multiples of 8x8
-    primitives.sa8d_inter[LUMA_4x4]   = primitives.satd[LUMA_4x4];
-    primitives.sa8d_inter[LUMA_4x8]   = primitives.satd[LUMA_4x8];
-    primitives.sa8d_inter[LUMA_4x16]  = primitives.satd[LUMA_4x16];
-    primitives.sa8d_inter[LUMA_8x4]   = primitives.satd[LUMA_8x4];
-    primitives.sa8d_inter[LUMA_16x4]  = primitives.satd[LUMA_16x4];
-    primitives.sa8d_inter[LUMA_16x12] = primitives.satd[LUMA_16x12];
-    primitives.sa8d_inter[LUMA_12x16] = primitives.satd[LUMA_12x16];
 }
 }

x265-hg20140324.tar.bz2/source/common/x86/blockcopy8.asm -> x265-hg20140928.tar.bz2/source/common/x86/blockcopy8.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/blockcopy8.h -> x265-hg20140928.tar.bz2/source/common/x86/blockcopy8.h Changed

@@ -18,18 +18,43 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_BLOCKCOPY8_H
 #define X265_BLOCKCOPY8_H
 
 void x265_cvt32to16_shr_sse2(int16_t * dst, int *src, intptr_t, int, int);
+void x265_cvt32to16_shl_4_sse2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_8_sse2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_16_sse2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_32_sse2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_4_avx2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_8_avx2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_16_avx2(int16_t * dst, int *src, intptr_t, int);
+void x265_cvt32to16_shl_32_avx2(int16_t * dst, int *src, intptr_t, int);
 void x265_cvt16to32_shl_sse4(int32_t * dst, int16_t * src, intptr_t, int32_t, int32_t);
+void x265_cvt16to32_shr_4_sse4(int32_t * dst, int16_t * src, intptr_t, int32_t, int32_t);
+void x265_cvt16to32_shr_8_sse4(int32_t * dst, int16_t * src, intptr_t, int32_t, int32_t);
+void x265_cvt16to32_shr_16_sse4(int32_t * dst, int16_t * src, intptr_t, int32_t, int32_t);
+void x265_cvt16to32_shr_32_sse4(int32_t * dst, int16_t * src, intptr_t, int32_t, int32_t);
+void x265_copy_shr_sse4(int16_t * dst, int16_t *src, intptr_t, int, int);
+void x265_copy_shl_4_sse2(int16_t * dst, int16_t *src, intptr_t, int);
+void x265_copy_shl_8_sse2(int16_t * dst, int16_t *src, intptr_t, int);
+void x265_copy_shl_16_sse2(int16_t * dst, int16_t *src, intptr_t, int);
+void x265_copy_shl_32_sse2(int16_t * dst, int16_t *src, intptr_t, int);
+uint32_t x265_copy_cnt_4_sse4(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_8_sse4(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_16_sse4(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_32_sse4(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_4_avx2(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_8_avx2(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_16_avx2(int16_t * dst, int16_t * src, intptr_t);
+uint32_t x265_copy_cnt_32_avx2(int16_t * dst, int16_t * src, intptr_t);
 
 #define SETUP_BLOCKCOPY_FUNC(W, H, cpu) \
     void x265_blockcopy_pp_ ## W ## x ## H ## cpu(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb); \
-    void x265_blockcopy_sp_ ## W ## x ## H ## cpu(pixel * a, intptr_t stridea, int16_t * b, intptr_t strideb);\
+    void x265_blockcopy_sp_ ## W ## x ## H ## cpu(pixel * a, intptr_t stridea, int16_t * b, intptr_t strideb); \
     void x265_blockcopy_ss_ ## W ## x ## H ## cpu(int16_t * a, intptr_t stridea, int16_t * b, intptr_t strideb);
 
 #define SETUP_BLOCKCOPY_PS(W, H, cpu) \
@@ -75,12 +100,33 @@
 #define BLOCKCOPY_SP(cpu) \
     SETUP_BLOCKCOPY_SP(2, 4, cpu); \
     SETUP_BLOCKCOPY_SP(2, 8, cpu); \
-    SETUP_BLOCKCOPY_SP(6, 8, cpu);
+    SETUP_BLOCKCOPY_SP(6, 8, cpu); \
+    \
+    SETUP_BLOCKCOPY_SP(2, 16, cpu); \
+    SETUP_BLOCKCOPY_SP(4, 32, cpu); \
+    SETUP_BLOCKCOPY_SP(6, 16, cpu); \
+    SETUP_BLOCKCOPY_SP(8, 12, cpu); \
+    SETUP_BLOCKCOPY_SP(8, 64, cpu); \
+    SETUP_BLOCKCOPY_SP(12, 32, cpu); \
+    SETUP_BLOCKCOPY_SP(16, 24, cpu); \
+    SETUP_BLOCKCOPY_SP(24, 64, cpu); \
+    SETUP_BLOCKCOPY_SP(32, 48, cpu);
 
 #define BLOCKCOPY_SS_PP(cpu) \
     SETUP_BLOCKCOPY_SS_PP(2, 4, cpu); \
     SETUP_BLOCKCOPY_SS_PP(2, 8, cpu); \
-    SETUP_BLOCKCOPY_SS_PP(6, 8, cpu);
+    SETUP_BLOCKCOPY_SS_PP(6, 8, cpu); \
+    \
+    SETUP_BLOCKCOPY_SS_PP(2, 16, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(4, 32, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(6, 16, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(8, 12, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(8, 64, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(12, 32, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(16, 24, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(24, 64, cpu); \
+    SETUP_BLOCKCOPY_SS_PP(32, 48, cpu);
+    
 
 #define BLOCKCOPY_PS(cpu) \
     SETUP_BLOCKCOPY_PS(2, 4, cpu); \
@@ -113,18 +159,47 @@
     SETUP_BLOCKCOPY_PS(64, 16, cpu); \
     SETUP_BLOCKCOPY_PS(64, 32, cpu); \
     SETUP_BLOCKCOPY_PS(64, 48, cpu); \
-    SETUP_BLOCKCOPY_PS(64, 64, cpu);
-
+    SETUP_BLOCKCOPY_PS(64, 64, cpu); \
+    \
+    SETUP_BLOCKCOPY_PS(2, 16, cpu); \
+    SETUP_BLOCKCOPY_PS(4, 32, cpu); \
+    SETUP_BLOCKCOPY_PS(6, 16, cpu); \
+    SETUP_BLOCKCOPY_PS(8, 12, cpu); \
+    SETUP_BLOCKCOPY_PS(8, 64, cpu); \
+    SETUP_BLOCKCOPY_PS(12, 32, cpu); \
+    SETUP_BLOCKCOPY_PS(16, 24, cpu); \
+    SETUP_BLOCKCOPY_PS(24, 64, cpu); \
+    SETUP_BLOCKCOPY_PS(32, 48, cpu);
 
 BLOCKCOPY_COMMON(_sse2);
 BLOCKCOPY_SS_PP(_sse2);
 BLOCKCOPY_SP(_sse4);
 BLOCKCOPY_PS(_sse4);
 
+BLOCKCOPY_SP(_sse2);
+
 void x265_blockfill_s_4x4_sse2(int16_t *dst, intptr_t dstride, int16_t val);
 void x265_blockfill_s_8x8_sse2(int16_t *dst, intptr_t dstride, int16_t val);
 void x265_blockfill_s_16x16_sse2(int16_t *dst, intptr_t dstride, int16_t val);
 void x265_blockfill_s_32x32_sse2(int16_t *dst, intptr_t dstride, int16_t val);
+void x265_blockcopy_ss_16x4_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x8_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x12_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x16_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x24_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x32_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_16x64_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_64x16_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_64x32_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_64x48_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+void x265_blockcopy_ss_64x64_avx(int16_t *dest, intptr_t deststride, int16_t *src, intptr_t srcstride);
+
+void x265_blockcopy_pp_32x8_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
+void x265_blockcopy_pp_32x16_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
+void x265_blockcopy_pp_32x24_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
+void x265_blockcopy_pp_32x32_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
+void x265_blockcopy_pp_32x48_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
+void x265_blockcopy_pp_32x64_avx(pixel * a, intptr_t stridea, pixel * b, intptr_t strideb);
 
 #undef BLOCKCOPY_COMMON
 #undef BLOCKCOPY_SS_PP

x265-hg20140324.tar.bz2/source/common/x86/const-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/const-a.asm Changed

@@ -4,7 +4,7 @@
 ;* Copyright (C) 2010-2013 x264 project
 ;*
 ;* Authors: Loren Merritt <lorenm@u.washington.edu>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
+;*          Fiona Glaser <fiona@x264.com>
 ;*          Min Chen <chenm003@163.com> <min.chen@multicorewareinc.com>
 ;*
 ;* This program is free software; you can redistribute it and/or modify
@@ -22,13 +22,15 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@x264.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************
 
 %include "x86inc.asm"
 
 SECTION_RODATA 32
 
+const pb_1,        times 32 db 1
+
 const hsub_mul,    times 16 db 1, -1
 const pw_1,        times 16 dw 1
 const pw_16,       times 16 dw 16
@@ -48,11 +50,16 @@
 const pb_unpackwq2, db 4,5,4,5,4,5,4,5,6,7,6,7,6,7,6,7
 const pw_swap,      times 2 db 6,7,4,5,2,3,0,1
 
+const pb_4,        times 16 db 4
+const pb_16,       times 16 db 16
+const pb_64,       times 16 db 64
 const pb_01,       times  8 db 0,1
 const pb_0,        times 16 db 0
-const pb_1,        times 32 db 1
 const pb_a1,       times 16 db 0xa1
 const pb_3,        times 16 db 3
+const pb_8,        times 16 db 8
+const pb_32,       times 16 db 32
+const pb_128,      times 16 db 128
 const pb_shuf8x8c, db 0,0,0,0,2,2,2,2,4,4,4,4,6,6,6,6
 
 const pw_2,        times 8 dw 2
@@ -70,7 +77,7 @@
 const pw_ppmmppmm, dw 1,1,-1,-1,1,1,-1,-1
 const pw_pmpmpmpm, dw 1,-1,1,-1,1,-1,1,-1
 const pw_pmmpzzzz, dw 1,-1,-1,1,0,0,0,0
-const pd_1,        times 4 dd 1
+const pd_1,        times 8 dd 1
 const pd_2,        times 4 dd 2
 const pd_4,        times 4 dd 4
 const pd_8,        times 4 dd 8
@@ -83,6 +90,7 @@
 const pd_1024,     times 4 dd 1024
 const pd_2048,     times 4 dd 2048
 const pd_ffff,     times 4 dd 0xffff
+const pd_32767,    times 4 dd 32767
 const pd_n32768,   times 4 dd 0xffff8000
 const pw_ff00,     times 8 dw 0xff00

x265-hg20140324.tar.bz2/source/common/x86/cpu-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/cpu-a.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/dct8.asm -> x265-hg20140928.tar.bz2/source/common/x86/dct8.asm Changed

@@ -4,6 +4,7 @@
 ;* Authors: Nabajit Deka <nabajit@multicorewareinc.com>
 ;*          Min Chen <chenm003@163.com> <min.chen@multicorewareinc.com>
 ;*          Li Cao <li@multicorewareinc.com>
+;*          Praveen Kumar Tiwari <Praveen@multicorewareinc.com>
 ;*
 ;* This program is free software; you can redistribute it and/or modify
 ;* it under the terms of the GNU General Public License as published by
@@ -20,21 +21,232 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@multicorewareinc.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************/
 
 ;TO-DO : Further optimize the routines.
 
 %include "x86inc.asm"
 %include "x86util.asm"
-
 SECTION_RODATA 32
+tab_dct8:       dw 64, 64, 64, 64, 64, 64, 64, 64
+                dw 89, 75, 50, 18, -18, -50, -75, -89
+                dw 83, 36, -36, -83, -83, -36, 36, 83
+                dw 75, -18, -89, -50, 50, 89, 18, -75
+                dw 64, -64, -64, 64, 64, -64, -64, 64
+                dw 50, -89, 18, 75, -75, -18, 89, -50
+                dw 36, -83, 83, -36, -36, 83, -83, 36
+                dw 18, -50, 75, -89, 89, -75, 50, -18
+
+dct8_shuf:      times 2 db 6, 7, 4, 5, 2, 3, 0, 1, 14, 15, 12, 13, 10, 11, 8, 9
+
+tab_dct16_1:    dw 64, 64, 64, 64, 64, 64, 64, 64
+                dw 90, 87, 80, 70, 57, 43, 25,  9
+                dw 89, 75, 50, 18, -18, -50, -75, -89
+                dw 87, 57,  9, -43, -80, -90, -70, -25
+                dw 83, 36, -36, -83, -83, -36, 36, 83
+                dw 80,  9, -70, -87, -25, 57, 90, 43
+                dw 75, -18, -89, -50, 50, 89, 18, -75
+                dw 70, -43, -87,  9, 90, 25, -80, -57
+                dw 64, -64, -64, 64, 64, -64, -64, 64
+                dw 57, -80, -25, 90, -9, -87, 43, 70
+                dw 50, -89, 18, 75, -75, -18, 89, -50
+                dw 43, -90, 57, 25, -87, 70,  9, -80
+                dw 36, -83, 83, -36, -36, 83, -83, 36
+                dw 25, -70, 90, -80, 43,  9, -57, 87
+                dw 18, -50, 75, -89, 89, -75, 50, -18
+                dw  9, -25, 43, -57, 70, -80, 87, -90
+
+
+tab_dct16_2:    dw 64, 64, 64, 64, 64, 64, 64, 64
+                dw -9, -25, -43, -57, -70, -80, -87, -90
+                dw -89, -75, -50, -18, 18, 50, 75, 89
+                dw 25, 70, 90, 80, 43, -9, -57, -87
+                dw 83, 36, -36, -83, -83, -36, 36, 83
+                dw -43, -90, -57, 25, 87, 70, -9, -80
+                dw -75, 18, 89, 50, -50, -89, -18, 75
+                dw 57, 80, -25, -90, -9, 87, 43, -70
+                dw 64, -64, -64, 64, 64, -64, -64, 64
+                dw -70, -43, 87,  9, -90, 25, 80, -57
+                dw -50, 89, -18, -75, 75, 18, -89, 50
+                dw 80, -9, -70, 87, -25, -57, 90, -43
+                dw 36, -83, 83, -36, -36, 83, -83, 36
+                dw -87, 57, -9, -43, 80, -90, 70, -25
+                dw -18, 50, -75, 89, -89, 75, -50, 18
+                dw 90, -87, 80, -70, 57, -43, 25, -9
+
+dct16_shuf1:     times 2 db 14, 15, 12, 13, 10, 11, 8, 9, 6, 7, 4, 5, 2, 3, 0, 1
+
+dct16_shuf2:    times 2 db 0, 1, 14, 15, 2, 3, 12, 13, 4, 5, 10, 11, 6, 7, 8, 9
+
+tab_dct32_1:    dw 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
+                dw 90, 90, 88, 85, 82, 78, 73, 67, 61, 54, 46, 38, 31, 22, 13,  4
+                dw 90, 87, 80, 70, 57, 43, 25,  9, -9, -25, -43, -57, -70, -80, -87, -90
+                dw 90, 82, 67, 46, 22, -4, -31, -54, -73, -85, -90, -88, -78, -61, -38, -13
+                dw 89, 75, 50, 18, -18, -50, -75, -89, -89, -75, -50, -18, 18, 50, 75, 89
+                dw 88, 67, 31, -13, -54, -82, -90, -78, -46, -4, 38, 73, 90, 85, 61, 22
+                dw 87, 57,  9, -43, -80, -90, -70, -25, 25, 70, 90, 80, 43, -9, -57, -87
+                dw 85, 46, -13, -67, -90, -73, -22, 38, 82, 88, 54, -4, -61, -90, -78, -31
+                dw 83, 36, -36, -83, -83, -36, 36, 83, 83, 36, -36, -83, -83, -36, 36, 83
+                dw 82, 22, -54, -90, -61, 13, 78, 85, 31, -46, -90, -67,  4, 73, 88, 38
+                dw 80,  9, -70, -87, -25, 57, 90, 43, -43, -90, -57, 25, 87, 70, -9, -80
+                dw 78, -4, -82, -73, 13, 85, 67, -22, -88, -61, 31, 90, 54, -38, -90, -46
+                dw 75, -18, -89, -50, 50, 89, 18, -75, -75, 18, 89, 50, -50, -89, -18, 75
+                dw 73, -31, -90, -22, 78, 67, -38, -90, -13, 82, 61, -46, -88, -4, 85, 54
+                dw 70, -43, -87,  9, 90, 25, -80, -57, 57, 80, -25, -90, -9, 87, 43, -70
+                dw 67, -54, -78, 38, 85, -22, -90,  4, 90, 13, -88, -31, 82, 46, -73, -61
+                dw 64, -64, -64, 64, 64, -64, -64, 64, 64, -64, -64, 64, 64, -64, -64, 64
+                dw 61, -73, -46, 82, 31, -88, -13, 90, -4, -90, 22, 85, -38, -78, 54, 67
+                dw 57, -80, -25, 90, -9, -87, 43, 70, -70, -43, 87,  9, -90, 25, 80, -57
+                dw 54, -85, -4, 88, -46, -61, 82, 13, -90, 38, 67, -78, -22, 90, -31, -73
+                dw 50, -89, 18, 75, -75, -18, 89, -50, -50, 89, -18, -75, 75, 18, -89, 50
+                dw 46, -90, 38, 54, -90, 31, 61, -88, 22, 67, -85, 13, 73, -82,  4, 78
+                dw 43, -90, 57, 25, -87, 70,  9, -80, 80, -9, -70, 87, -25, -57, 90, -43
+                dw 38, -88, 73, -4, -67, 90, -46, -31, 85, -78, 13, 61, -90, 54, 22, -82
+                dw 36, -83, 83, -36, -36, 83, -83, 36, 36, -83, 83, -36, -36, 83, -83, 36
+                dw 31, -78, 90, -61,  4, 54, -88, 82, -38, -22, 73, -90, 67, -13, -46, 85
+                dw 25, -70, 90, -80, 43,  9, -57, 87, -87, 57, -9, -43, 80, -90, 70, -25
+                dw 22, -61, 85, -90, 73, -38, -4, 46, -78, 90, -82, 54, -13, -31, 67, -88
+                dw 18, -50, 75, -89, 89, -75, 50, -18, -18, 50, -75, 89, -89, 75, -50, 18
+                dw 13, -38, 61, -78, 88, -90, 85, -73, 54, -31,  4, 22, -46, 67, -82, 90
+                dw 9, -25, 43, -57, 70, -80, 87, -90, 90, -87, 80, -70, 57, -43, 25, -9
+                dw 4, -13, 22, -31, 38, -46, 54, -61, 67, -73, 78, -82, 85, -88, 90, -90
+
+tab_dct32_2:    dw 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
+                dw -4, -13, -22, -31, -38, -46, -54, -61, -67, -73, -78, -82, -85, -88, -90, -90
+                dw -90, -87, -80, -70, -57, -43, -25, -9,  9, 25, 43, 57, 70, 80, 87, 90
+                dw 13, 38, 61, 78, 88, 90, 85, 73, 54, 31,  4, -22, -46, -67, -82, -90
+                dw 89, 75, 50, 18, -18, -50, -75, -89, -89, -75, -50, -18, 18, 50, 75, 89
+                dw -22, -61, -85, -90, -73, -38,  4, 46, 78, 90, 82, 54, 13, -31, -67, -88
+                dw -87, -57, -9, 43, 80, 90, 70, 25, -25, -70, -90, -80, -43,  9, 57, 87
+                dw 31, 78, 90, 61,  4, -54, -88, -82, -38, 22, 73, 90, 67, 13, -46, -85
+                dw 83, 36, -36, -83, -83, -36, 36, 83, 83, 36, -36, -83, -83, -36, 36, 83
+                dw -38, -88, -73, -4, 67, 90, 46, -31, -85, -78, -13, 61, 90, 54, -22, -82
+                dw -80, -9, 70, 87, 25, -57, -90, -43, 43, 90, 57, -25, -87, -70,  9, 80
+                dw 46, 90, 38, -54, -90, -31, 61, 88, 22, -67, -85, -13, 73, 82,  4, -78
+                dw 75, -18, -89, -50, 50, 89, 18, -75, -75, 18, 89, 50, -50, -89, -18, 75
+                dw -54, -85,  4, 88, 46, -61, -82, 13, 90, 38, -67, -78, 22, 90, 31, -73
+                dw -70, 43, 87, -9, -90, -25, 80, 57, -57, -80, 25, 90,  9, -87, -43, 70
+                dw 61, 73, -46, -82, 31, 88, -13, -90, -4, 90, 22, -85, -38, 78, 54, -67
+                dw 64, -64, -64, 64, 64, -64, -64, 64, 64, -64, -64, 64, 64, -64, -64, 64
+                dw -67, -54, 78, 38, -85, -22, 90,  4, -90, 13, 88, -31, -82, 46, 73, -61
+                dw -57, 80, 25, -90,  9, 87, -43, -70, 70, 43, -87, -9, 90, -25, -80, 57
+                dw 73, 31, -90, 22, 78, -67, -38, 90, -13, -82, 61, 46, -88,  4, 85, -54
+                dw 50, -89, 18, 75, -75, -18, 89, -50, -50, 89, -18, -75, 75, 18, -89, 50
+                dw -78, -4, 82, -73, -13, 85, -67, -22, 88, -61, -31, 90, -54, -38, 90, -46
+                dw -43, 90, -57, -25, 87, -70, -9, 80, -80,  9, 70, -87, 25, 57, -90, 43
+                dw 82, -22, -54, 90, -61, -13, 78, -85, 31, 46, -90, 67,  4, -73, 88, -38
+                dw 36, -83, 83, -36, -36, 83, -83, 36, 36, -83, 83, -36, -36, 83, -83, 36
+                dw -85, 46, 13, -67, 90, -73, 22, 38, -82, 88, -54, -4, 61, -90, 78, -31
+                dw -25, 70, -90, 80, -43, -9, 57, -87, 87, -57,  9, 43, -80, 90, -70, 25
+                dw 88, -67, 31, 13, -54, 82, -90, 78, -46,  4, 38, -73, 90, -85, 61, -22
+                dw 18, -50, 75, -89, 89, -75, 50, -18, -18, 50, -75, 89, -89, 75, -50, 18
+                dw -90, 82, -67, 46, -22, -4, 31, -54, 73, -85, 90, -88, 78, -61, 38, -13
+                dw -9, 25, -43, 57, -70, 80, -87, 90, -90, 87, -80, 70, -57, 43, -25,  9
+                dw 90, -90, 88, -85, 82, -78, 73, -67, 61, -54, 46, -38, 31, -22, 13, -4
+
+avx2_idct8_1:   times 4 dw 64, 83, 64, 36
+                times 4 dw 64, 36, -64, -83
+                times 4 dw 64, -36, -64, 83
+                times 4 dw 64, -83, 64, -36
+
+avx2_idct8_2:   times 4 dw 89, 75, 50, 18
+                times 4 dw 75, -18, -89, -50
+                times 4 dw 50, -89, 18, 75
+                times 4 dw 18, -50, 75, -89
+
+idct8_shuf1:    dd 0, 2, 4, 6, 1, 3, 5, 7
+
+idct8_shuf2:    times 2 db 0, 1, 2, 3, 8, 9, 10, 11, 4, 5, 6, 7, 12, 13, 14, 15
+
+idct8_shuf3:    times 2 db 12, 13, 14, 15, 8, 9, 10, 11, 4, 5, 6, 7, 0, 1, 2, 3
+
+tab_idct16_1:   dw 90, 87, 80, 70, 57, 43, 25, 9
+                dw 87, 57, 9, -43, -80, -90, -70, -25
+                dw 80, 9, -70, -87, -25, 57, 90, 43
+                dw 70, -43, -87, 9, 90, 25, -80, -57
+                dw 57, -80, -25, 90, -9, -87, 43, 70
+                dw 43, -90, 57, 25, -87, 70, 9, -80
+                dw 25, -70, 90, -80, 43, 9, -57, 87
+                dw 9, -25, 43, -57, 70, -80, 87, -90
+
+tab_idct16_2:   dw 64, 89, 83, 75, 64, 50, 36, 18
+                dw 64, 75, 36, -18, -64, -89, -83, -50
+                dw 64, 50, -36, -89, -64, 18, 83, 75
+                dw 64, 18, -83, -50, 64, 75, -36, -89
+                dw 64, -18, -83, 50, 64, -75, -36, 89
+                dw 64, -50, -36, 89, -64, -18, 83, -75
+                dw 64, -75, 36, 18, -64, 89, -83, 50
+                dw 64, -89, 83, -75, 64, -50, 36, -18
+
+idct16_shuff:   dd 0, 4, 2, 6, 1, 5, 3, 7
+
+idct16_shuff1:  dd 2, 6, 0, 4, 3, 7, 1, 5
+
+tab_idct32_1:   dw 90 ,90 ,88 ,85, 82, 78, 73, 67, 61, 54, 46, 38, 31, 22, 13, 4
+                dw 90, 82, 67, 46, 22, -4, -31, -54, -73, -85, -90, -88, -78, -61, -38, -13
+                dw 88, 67, 31, -13, -54, -82, -90, -78, -46, -4, 38, 73, 90, 85, 61, 22
+                dw 85, 46, -13, -67, -90, -73, -22, 38, 82, 88, 54, -4, -61, -90, -78, -31
+                dw 82, 22, -54, -90, -61, 13, 78, 85, 31, -46, -90, -67, 4, 73, 88, 38
+                dw 78, -4, -82, -73, 13, 85, 67, -22, -88, -61, 31, 90, 54, -38, -90, -46
+                dw 73, -31, -90, -22, 78, 67, -38, -90, -13, 82, 61, -46, -88, -4, 85, 54
+                dw 67, -54, -78, 38, 85, -22, -90, 4, 90, 13, -88, -31, 82, 46, -73, -61
+                dw 61, -73, -46, 82, 31, -88, -13, 90, -4, -90, 22, 85, -38, -78, 54, 67
+                dw 54, -85, -4, 88, -46, -61, 82, 13, -90, 38, 67, -78, -22, 90, -31, -73
+                dw 46, -90, 38, 54, -90, 31, 61, -88, 22, 67, -85, 13, 73, -82, 4, 78
+                dw 38, -88, 73, -4, -67, 90, -46, -31, 85, -78, 13, 61, -90, 54, 22, -82
+                dw 31, -78, 90, -61, 4, 54, -88, 82, -38, -22, 73, -90, 67, -13, -46, 85
+                dw 22, -61, 85, -90, 73, -38, -4, 46, -78, 90, -82, 54, -13, -31, 67, -88
+                dw 13, -38, 61, -78, 88, -90, 85, -73, 54, -31, 4, 22, -46, 67, -82, 90
+                dw 4, -13, 22, -31, 38, -46, 54, -61, 67, -73, 78, -82, 85, -88, 90, -90
+
+
+tab_idct32_2:   dw 64, 89, 83, 75, 64, 50, 36, 18
+                dw 64, 75, 36, -18, -64, -89, -83, -50
+                dw 64, 50, -36, -89, -64, 18, 83, 75
+                dw 64, 18, -83, -50, 64, 75, -36, -89
+                dw 64, -18, -83, 50, 64, -75, -36, 89
+                dw 64, -50, -36, 89, -64, -18, 83, -75
+                dw 64, -75, 36, 18, -64, 89, -83, 50
+                dw 64, -89, 83, -75, 64, -50, 36, -18
+
+
+tab_idct32_3:   dw 90, 87, 80, 70, 57, 43, 25, 9
+                dw 87, 57, 9, -43, -80, -90, -70, -25
+                dw 80, 9, -70, -87, -25, 57, 90, 43
+                dw 70, -43, -87, 9, 90, 25, -80, -57
+                dw 57, -80, -25, 90, -9, -87, 43, 70
+                dw 43, -90, 57, 25, -87, 70, 9, -80
+                dw 25, -70, 90, -80, 43, 9, -57, 87
+                dw 9, -25, 43, -57, 70, -80, 87, -90
+
+tab_idct32_4:   dw 64, 90, 89, 87, 83, 80, 75, 70, 64, 57, 50, 43, 36, 25, 18, 9
+                dw 64, 87, 75, 57, 36, 9, -18, -43, -64, -80, -89, -90, -83, -70, -50, -25
+                dw 64, 80, 50, 9, -36, -70, -89, -87, -64, -25, 18, 57, 83, 90, 75, 43
+                dw 64, 70, 18, -43, -83, -87, -50, 9, 64, 90, 75, 25, -36, -80, -89, -57
+                dw 64, 57, -18, -80, -83, -25, 50, 90, 64, -9, -75, -87, -36, 43, 89, 70
+                dw 64, 43, -50, -90, -36, 57, 89, 25, -64, -87, -18, 70, 83, 9, -75, -80
+                dw 64, 25, -75, -70, 36, 90, 18, -80, -64, 43, 89, 9, -83, -57, 50, 87
+                dw 64, 9, -89, -25, 83, 43, -75, -57, 64, 70, -50, -80, 36, 87, -18, -90
+                dw 64, -9, -89, 25, 83, -43, -75, 57, 64, -70, -50, 80, 36, -87, -18, 90
+                dw 64, -25, -75, 70, 36, -90, 18, 80, -64, -43, 89, -9, -83, 57, 50, -87
+                dw 64, -43, -50, 90, -36, -57, 89, -25, -64, 87, -18, -70, 83, -9, -75, 80
+                dw 64, -57, -18, 80, -83, 25, 50, -90, 64, 9, -75, 87, -36, -43, 89, -70
+                dw 64, -70, 18, 43, -83, 87, -50, -9, 64, -90, 75, -25, -36, 80, -89, 57
+                dw 64, -80, 50, -9, -36, 70, -89, 87, -64, 25, 18, -57, 83, -90, 75, -43
+                dw 64, -87, 75, -57, 36, -9, -18, 43, -64, 80, -89, 90, -83, 70, -50, 25
+                dw 64, -90, 89, -87, 83, -80, 75, -70, 64, -57, 50, -43, 36, -25, 18, -9
+
+avx2_dct4:      dw 64, 64, 64, 64, 64, 64, 64, 64, 64, -64, 64, -64, 64, -64, 64, -64
+                dw 83, 36, 83, 36, 83, 36, 83, 36, 36, -83, 36, -83, 36, -83, 36, -83
 
 tab_dct4:       times 4 dw 64, 64
                 times 4 dw 83, 36
                 times 4 dw 64, -64
                 times 4 dw 36, -83
 
+dct4_shuf:      db 0, 1, 2, 3, 8, 9, 10, 11, 6, 7, 4, 5, 14, 15, 12, 13
+
 tab_dst4:       times 2 dw 29, 55, 74, 84
                 times 2 dw 74, 74,  0, -74
                 times 2 dw 84, -29, -74, 55
@@ -92,6 +304,7 @@
 cextern pd_128
 cextern pd_256
 cextern pd_512
+cextern pd_1024
 cextern pd_2048
 cextern pw_ppppmmmm
 
@@ -184,7 +397,74 @@
     paddd       m2, m7
     psrad       m2, 8
     movu        [r1 + 3 * 16], m2
+    RET
 
+; DCT 4x4
+;
+; Input parameters:
+; - r0:     source
+; - r1:     destination
+; - r2:     source stride
+INIT_YMM avx2
+cglobal dct4, 3, 4, 8, src, dst, srcStride
+%if BIT_DEPTH == 10
+    %define DCT_SHIFT 3
+    vbroadcasti128 m7, [pd_4]
+%elif BIT_DEPTH == 8
+    %define DCT_SHIFT 1
+    vbroadcasti128 m7, [pd_1]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+    add             r2d, r2d
+    lea             r3, [avx2_dct4]
+
+    vbroadcasti128  m4, [dct4_shuf]
+    mova            m5, [r3]
+    mova            m6, [r3 + 32]
+    movq            xm0, [r0]
+    movhps          xm0, [r0 + r2]
+    lea             r0, [r0 + 2 * r2]
+    movq            xm1, [r0]
+    movhps          xm1, [r0 + r2]
+
+    vinserti128     m0, m0, xm1, 1
+    pshufb          m0, m4
+    vpermq          m1, m0, 11011101b
+    vpermq          m0, m0, 10001000b
+    paddw           m2, m0, m1
+    psubw           m0, m1
+
+    pmaddwd         m2, m5
+    paddd           m2, m7
+    psrad           m2, DCT_SHIFT
+
+    pmaddwd         m0, m6
+    paddd           m0, m7
+    psrad           m0, DCT_SHIFT
+
+    packssdw        m2, m0
+    pshufb          m2, m4
+    vpermq          m1, m2, 11011101b
+    vpermq          m2, m2, 10001000b
+    vbroadcasti128  m7, [pd_128]
+
+    pmaddwd         m0, m2, m5
+    pmaddwd         m3, m1, m5
+    paddd           m3, m0
+    paddd           m3, m7
+    psrad           m3, 8
+
+    pmaddwd         m2, m6
+    pmaddwd         m1, m6
+    psubd           m2, m1
+    paddd           m2, m7
+    psrad           m2, 8
+
+    movu            [r1], xm3
+    movu            [r1 + mmsize/2], m2
+    vextracti128    [r1 + mmsize], m3, 1
+    vextracti128    [r1 + mmsize + mmsize/2], m2, 1
     RET
 
 ;-------------------------------------------------------
@@ -834,8 +1114,14 @@
 
     ret
 
-cglobal idct8, 3,7,8,0-16*mmsize
+cglobal idct8, 3,7,8 ;,0-16*mmsize
+    ; alignment stack to 64-bytes
     mov         r5, rsp
+    sub         rsp, 16*mmsize + gprsize
+    and         rsp, ~(64-1)
+    mov         [rsp + 16*mmsize], r5
+    mov         r5, rsp
+
     lea         r4, [tab_idct8_3]
     lea         r6, [tab_dct4]
 
@@ -866,4 +1152,1453 @@
 
     call        patial_butterfly_inverse_internal_pass2
 
+    ; restore origin stack pointer
+    mov         rsp, [rsp + 16*mmsize]
+    RET
+
+
+;-----------------------------------------------------------------------------
+; void denoise_dct(int32_t *dct, uint32_t *sum, uint16_t *offset, int size)
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal denoise_dct, 4, 4, 6
+    pxor     m5,  m5
+    shr      r3d, 2
+.loop:
+    mova     m0, [r0]
+    pabsd    m1, m0
+    mova     m2, [r1]
+    paddd    m2, m1
+    mova     [r1], m2
+    pmovzxwd m3, [r2]
+    psubd    m1, m3
+    pcmpgtd  m4, m1, m5
+    pand     m1, m4
+    psignd   m1, m0
+    mova     [r0], m1
+    add      r0, 16
+    add      r1, 16
+    add      r2, 8
+    dec      r3d
+    jnz .loop
+    RET
+
+INIT_YMM avx2
+cglobal denoise_dct, 4, 4, 6
+    pxor     m5,  m5
+    shr      r3d, 3
+.loop:
+    movu     m0, [r0]
+    pabsd    m1, m0
+    movu     m2, [r1]
+    paddd    m2, m1
+    movu     [r1], m2
+    pmovzxwd m3, [r2]
+    psubd    m1, m3
+    pcmpgtd  m4, m1, m5
+    pand     m1, m4
+    psignd   m1, m0
+    movu     [r0], m1
+    add      r0, 32
+    add      r1, 32
+    add      r2, 16
+    dec      r3d
+    jnz .loop
+    RET
+%if ARCH_X86_64 == 1
+%macro DCT8_PASS_1 4
+    vpbroadcastq    m0,                 [r6 + %1]
+    pmaddwd         m2,                 m%3, m0
+    pmaddwd         m0,                 m%4
+    phaddd          m2,                 m0
+    paddd           m2,                 m5
+    psrad           m2,                 DCT_SHIFT
+    packssdw        m2,                 m2
+    vpermq          m2,                 m2, 0x08
+    mova            [r5 + %2],          xm2
+%endmacro
+
+%macro DCT8_PASS_2 1
+    vbroadcasti128  m4,                 [r6 + %1]
+    pmaddwd         m6,                 m0, m4
+    pmaddwd         m7,                 m1, m4
+    pmaddwd         m8,                 m2, m4
+    pmaddwd         m9,                 m3, m4
+    phaddd          m6,                 m7
+    phaddd          m8,                 m9
+    phaddd          m6,                 m8
+    paddd           m6,                 m5
+    psrad           m6,                 DCT_SHIFT2
+%endmacro
+
+INIT_YMM avx2
+cglobal dct8, 3, 7, 10, 0-8*16
+%if BIT_DEPTH == 10
+    %define         DCT_SHIFT          4
+    vbroadcasti128  m5,                [pd_8]
+%elif BIT_DEPTH == 8
+    %define         DCT_SHIFT          2
+    vbroadcasti128  m5,                [pd_2]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+%define             DCT_SHIFT2         9
+
+    add             r2d,               r2d
+    lea             r3,                [r2 * 3]
+    lea             r4,                [r0 + r2 * 4]
+    mov             r5,                rsp
+    lea             r6,                [tab_dct8]
+    mova            m6,                [dct8_shuf]
+
+    ;pass1
+    mova            xm0,               [r0]
+    vinserti128     m0,                m0, [r4], 1
+    mova            xm1,               [r0 + r2]
+    vinserti128     m1,                m1, [r4 + r2], 1
+    mova            xm2,               [r0 + r2 * 2]
+    vinserti128     m2,                m2, [r4 + r2 * 2], 1
+    mova            xm3,               [r0 + r3]
+    vinserti128     m3,                m3,  [r4 + r3], 1
+
+    punpcklqdq      m4,                m0, m1
+    punpckhqdq      m0,                m1
+    punpcklqdq      m1,                m2, m3
+    punpckhqdq      m2,                m3
+
+    pshufb          m0,                m6
+    pshufb          m2,                m6
+
+    paddw           m3,                m4, m0
+    paddw           m7,                m1, m2
+
+    psubw           m4,                m0
+    psubw           m1,                m2
+
+    DCT8_PASS_1     0 * 16,             0 * 16, 3, 7
+    DCT8_PASS_1     1 * 16,             2 * 16, 4, 1
+    DCT8_PASS_1     2 * 16,             4 * 16, 3, 7
+    DCT8_PASS_1     3 * 16,             6 * 16, 4, 1
+    DCT8_PASS_1     4 * 16,             1 * 16, 3, 7
+    DCT8_PASS_1     5 * 16,             3 * 16, 4, 1
+    DCT8_PASS_1     6 * 16,             5 * 16, 3, 7
+    DCT8_PASS_1     7 * 16,             7 * 16, 4, 1
+
+    ;pass2
+    mov             r2d,               32
+    lea             r3,                [r2 * 3]
+    lea             r4,                [r1 + r2 * 4]
+    vbroadcasti128  m5,                [pd_256]
+
+    mova            m0,                [r5]
+    mova            m1,                [r5 + 32]
+    mova            m2,                [r5 + 64]
+    mova            m3,                [r5 + 96]
+
+    DCT8_PASS_2     0 * 16
+    movu            [r1],              m6
+    DCT8_PASS_2     1 * 16
+    movu            [r1 + r2],         m6
+    DCT8_PASS_2     2 * 16
+    movu            [r1 + r2 * 2],     m6
+    DCT8_PASS_2     3 * 16
+    movu            [r1 + r3],         m6
+    DCT8_PASS_2     4 * 16
+    movu            [r4],              m6
+    DCT8_PASS_2     5 * 16
+    movu            [r4 + r2],         m6
+    DCT8_PASS_2     6 * 16
+    movu            [r4 + r2 * 2],     m6
+    DCT8_PASS_2     7 * 16
+    movu            [r4 + r3],         m6
+    RET
+
+%macro DCT16_PASS_1_E 2
+    vpbroadcastq    m7,                [r7 + %1]
+
+    pmaddwd         m4,                m0, m7
+    pmaddwd         m6,                m2, m7
+    phaddd          m4,                m6
+
+    paddd           m4,                m9
+    psrad           m4,                DCT_SHIFT
+
+    packssdw        m4,                m4
+    vpermq          m4,                m4, 0x08
+
+    mova            [r5 + %2],         xm4
+%endmacro
+
+%macro DCT16_PASS_1_O 2
+    vbroadcasti128  m7,                [r7 + %1]
+
+    pmaddwd         m10,               m0, m7
+    pmaddwd         m11,               m2, m7
+    phaddd          m10,               m11                 ; [d0 d0 d1 d1 d4 d4 d5 d5]
+
+    pmaddwd         m11,               m4, m7
+    pmaddwd         m12,               m6, m7
+    phaddd          m11,               m12                 ; [d2 d2 d3 d3 d6 d6 d7 d7]
+
+    phaddd          m10,               m11                 ; [d0 d1 d2 d3 d4 d5 d6 d7]
+
+    paddd           m10,               m9
+    psrad           m10,               DCT_SHIFT
+
+    packssdw        m10,               m10                 ; [w0 w1 w2 w3 - - - - w4 w5 w6 w7 - - - -]
+    vpermq          m10,               m10, 0x08
+
+    mova            [r5 + %2],         xm10
+%endmacro
+
+%macro DCT16_PASS_2 1
+    vbroadcasti128  m8,                [r7 + %1]
+    vbroadcasti128  m13,               [r8 + %1]
+
+    pmaddwd         m10,               m0, m8
+    pmaddwd         m11,               m1, m13
+    paddd           m10,               m11
+
+    pmaddwd         m11,               m2, m8
+    pmaddwd         m12,               m3, m13
+    paddd           m11,               m12
+    phaddd          m10,               m11
+
+    pmaddwd         m11,               m4, m8
+    pmaddwd         m12,               m5, m13
+    paddd           m11,               m12
+
+    pmaddwd         m12,               m6, m8
+    pmaddwd         m13,               m7, m13
+    paddd           m12,               m13
+    phaddd          m11,               m12
+
+    phaddd          m10,               m11
+    paddd           m10,               m9
+    psrad           m10,               DCT_SHIFT2
+%endmacro
+INIT_YMM avx2
+cglobal dct16, 3, 9, 15, 0-16*mmsize
+%if BIT_DEPTH == 10
+    %define         DCT_SHIFT          5
+    vbroadcasti128  m9,                [pd_16]
+%elif BIT_DEPTH == 8
+    %define         DCT_SHIFT          3
+    vbroadcasti128  m9,                [pd_4]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+%define             DCT_SHIFT2         10
+
+    add             r2d,               r2d
+
+    mova            m13,               [dct16_shuf1]
+    mova            m14,               [dct16_shuf2]
+    lea             r7,                [tab_dct16_1 + 8 * 16]
+    lea             r8,                [tab_dct16_2 + 8 * 16]
+    lea             r3,                [r2 * 3]
+    mov             r5,                rsp
+    mov             r4d,               2                   ; Each iteration process 8 rows, so 16/8 iterations
+
+.pass1:
+    lea             r6,                [r0 + r2 * 4]
+
+    movu            m2,                [r0]
+    movu            m1,                [r6]
+    vperm2i128      m0,                m2, m1, 0x20        ; [row0lo  row4lo]
+    vperm2i128      m1,                m2, m1, 0x31        ; [row0hi  row4hi]
+
+    movu            m4,                [r0 + r2]
+    movu            m3,                [r6 + r2]
+    vperm2i128      m2,                m4, m3, 0x20        ; [row1lo  row5lo]
+    vperm2i128      m3,                m4, m3, 0x31        ; [row1hi  row5hi]
+
+    movu            m6,                [r0 + r2 * 2]
+    movu            m5,                [r6 + r2 * 2]
+    vperm2i128      m4,                m6, m5, 0x20        ; [row2lo  row6lo]
+    vperm2i128      m5,                m6, m5, 0x31        ; [row2hi  row6hi]
+
+    movu            m8,                [r0 + r3]
+    movu            m7,                [r6 + r3]
+    vperm2i128      m6,                m8, m7, 0x20        ; [row3lo  row7lo]
+    vperm2i128      m7,                m8, m7, 0x31        ; [row3hi  row7hi]
+
+    pshufb          m1,                m13
+    pshufb          m3,                m13
+    pshufb          m5,                m13
+    pshufb          m7,                m13
+
+    paddw           m8,                m0, m1              ;E
+    psubw           m0,                m1                  ;O
+
+    paddw           m1,                m2, m3              ;E
+    psubw           m2,                m3                  ;O
+
+    paddw           m3,                m4, m5              ;E
+    psubw           m4,                m5                  ;O
+
+    paddw           m5,                m6, m7              ;E
+    psubw           m6,                m7                  ;O
+
+    DCT16_PASS_1_O  -7 * 16,           1 * 32
+    DCT16_PASS_1_O  -5 * 16,           3 * 32
+    DCT16_PASS_1_O  -3 * 16,           1 * 32 + 16
+    DCT16_PASS_1_O  -1 * 16,           3 * 32 + 16
+    DCT16_PASS_1_O  1 * 16,            5 * 32
+    DCT16_PASS_1_O  3 * 16,            7 * 32
+    DCT16_PASS_1_O  5 * 16,            5 * 32 + 16
+    DCT16_PASS_1_O  7 * 16,            7 * 32 + 16
+
+    pshufb          m8,                m14
+    pshufb          m1,                m14
+    phaddw          m0,                m8, m1
+
+    pshufb          m3,                m14
+    pshufb          m5,                m14
+    phaddw          m2,                m3, m5
+
+    DCT16_PASS_1_E  -8 * 16,           0 * 32
+    DCT16_PASS_1_E  -4 * 16,           0 * 32 + 16
+    DCT16_PASS_1_E  0 * 16,            4 * 32
+    DCT16_PASS_1_E  4 * 16,            4 * 32 + 16
+
+    phsubw          m0,                m8, m1
+    phsubw          m2,                m3, m5
+
+    DCT16_PASS_1_E  -6 * 16,           2 * 32
+    DCT16_PASS_1_E  -2 * 16,           2 * 32 + 16
+    DCT16_PASS_1_E  2 * 16,            6 * 32
+    DCT16_PASS_1_E  6 * 16,            6 * 32 + 16
+
+    lea             r0,                [r0 + 8 * r2]
+    add             r5,                256
+
+    dec             r4d
+    jnz             .pass1
+
+    mov             r5,                rsp
+    mov             r4d,               2
+    mov             r2d,               64
+    lea             r3,                [r2 * 3]
+    vbroadcasti128  m9,                [pd_512]
+
+.pass2:
+    mova            m0,                [r5 + 0 * 32]        ; [row0lo  row4lo]
+    mova            m1,                [r5 + 8 * 32]        ; [row0hi  row4hi]
+
+    mova            m2,                [r5 + 1 * 32]        ; [row1lo  row5lo]
+    mova            m3,                [r5 + 9 * 32]        ; [row1hi  row5hi]
+
+    mova            m4,                [r5 + 2 * 32]        ; [row2lo  row6lo]
+    mova            m5,                [r5 + 10 * 32]       ; [row2hi  row6hi]
+
+    mova            m6,                [r5 + 3 * 32]        ; [row3lo  row7lo]
+    mova            m7,                [r5 + 11 * 32]       ; [row3hi  row7hi]
+
+    DCT16_PASS_2    -8 * 16
+    movu            [r1],              m10
+    DCT16_PASS_2    -7 * 16
+    movu            [r1 + r2],         m10
+    DCT16_PASS_2    -6 * 16
+    movu            [r1 + r2 * 2],     m10
+    DCT16_PASS_2    -5 * 16
+    movu            [r1 + r3],         m10
+
+    lea             r6,                [r1 + r2 * 4]
+    DCT16_PASS_2    -4 * 16
+    movu            [r6],              m10
+    DCT16_PASS_2    -3 * 16
+    movu            [r6 + r2],         m10
+    DCT16_PASS_2    -2 * 16
+    movu            [r6 + r2 * 2],     m10
+    DCT16_PASS_2    -1 * 16
+    movu            [r6 + r3],         m10
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT16_PASS_2    0 * 16
+    movu            [r6],              m10
+    DCT16_PASS_2    1 * 16
+    movu            [r6 + r2],         m10
+    DCT16_PASS_2    2 * 16
+    movu            [r6 + r2 * 2],     m10
+    DCT16_PASS_2    3 * 16
+    movu            [r6 + r3],         m10
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT16_PASS_2    4 * 16
+    movu            [r6],              m10
+    DCT16_PASS_2    5 * 16
+    movu            [r6 + r2],         m10
+    DCT16_PASS_2    6 * 16
+    movu            [r6 + r2 * 2],     m10
+    DCT16_PASS_2    7 * 16
+    movu            [r6 + r3],         m10
+
+    add             r1,                32
+    add             r5,                128
+
+    dec             r4d
+    jnz             .pass2
+    RET
+
+%macro DCT32_PASS_1 4
+    vbroadcasti128  m8,                [r7 + %1]
+
+    pmaddwd         m11,               m%3, m8
+    pmaddwd         m12,               m%4, m8
+    phaddd          m11,               m12
+
+    vbroadcasti128  m8,                [r7 + %1 + 32]
+    vbroadcasti128  m10,               [r7 + %1 + 48]
+    pmaddwd         m12,               m5, m8
+    pmaddwd         m13,               m6, m10
+    phaddd          m12,               m13
+
+    pmaddwd         m13,               m4, m8
+    pmaddwd         m14,               m7, m10
+    phaddd          m13,               m14
+
+    phaddd          m12,               m13
+
+    phaddd          m11,               m12
+    paddd           m11,               m9
+    psrad           m11,               DCT_SHIFT
+
+    vpermq          m11,               m11, 0xD8
+    packssdw        m11,               m11
+    movq            [r5 + %2],         xm11
+    vextracti128    xm10,              m11, 1
+    movq            [r5 + %2 + 64],    xm10
+%endmacro
+
+%macro DCT32_PASS_2 1
+    mova            m8,                [r7 + %1]
+    mova            m10,               [r8 + %1]
+    pmaddwd         m11,               m0, m8
+    pmaddwd         m12,               m1, m10
+    paddd           m11,               m12
+
+    pmaddwd         m12,               m2, m8
+    pmaddwd         m13,               m3, m10
+    paddd           m12,               m13
+
+    phaddd          m11,               m12
+
+    pmaddwd         m12,               m4, m8
+    pmaddwd         m13,               m5, m10
+    paddd           m12,               m13
+
+    pmaddwd         m13,               m6, m8
+    pmaddwd         m14,               m7, m10
+    paddd           m13,               m14
+
+    phaddd          m12,               m13
+
+    phaddd          m11,               m12
+    vextracti128    xm10,              m11, 1
+    paddd           xm11,              xm10
+
+    paddd           xm11,               xm9
+    psrad           xm11,               DCT_SHIFT2
+
+%endmacro
+
+INIT_YMM avx2
+cglobal dct32, 3, 9, 16, 0-64*mmsize
+%if BIT_DEPTH == 10
+    %define         DCT_SHIFT          6
+    vpbroadcastq    m9,                [pd_32]
+%elif BIT_DEPTH == 8
+    %define         DCT_SHIFT          4
+    vpbroadcastq    m9,                [pd_8]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+%define             DCT_SHIFT2         11
+
+    add             r2d,               r2d
+
+    lea             r7,                [tab_dct32_1]
+    lea             r8,                [tab_dct32_2]
+    lea             r3,                [r2 * 3]
+    mov             r5,                rsp
+    mov             r4d,               8
+    mova            m15,               [dct16_shuf1]
+
+.pass1:
+    movu            m2,                [r0]
+    movu            m1,                [r0 + 32]
+    pshufb          m1,                m15
+    vpermq          m1,                m1, 0x4E
+    psubw           m7,                m2, m1
+    paddw           m2,                m1
+
+    movu            m1,                [r0 + r2 * 2]
+    movu            m0,                [r0 + r2 * 2 + 32]
+    pshufb          m0,                m15
+    vpermq          m0,                m0, 0x4E
+    psubw           m8,                m1, m0
+    paddw           m1,                m0
+    vperm2i128      m0,                m2, m1, 0x20        ; [row0lo  row2lo] for E
+    vperm2i128      m3,                m2, m1, 0x31        ; [row0hi  row2hi] for E
+    pshufb          m3,                m15
+    psubw           m1,                m0, m3
+    paddw           m0,                m3
+
+    vperm2i128      m5,                m7, m8, 0x20        ; [row0lo  row2lo] for O
+    vperm2i128      m6,                m7, m8, 0x31        ; [row0hi  row2hi] for O
+
+
+    movu            m4,                [r0 + r2]
+    movu            m2,                [r0 + r2 + 32]
+    pshufb          m2,                m15
+    vpermq          m2,                m2, 0x4E
+    psubw           m10,               m4, m2
+    paddw           m4,                m2
+
+    movu            m3,                [r0 + r3]
+    movu            m2,                [r0 + r3 + 32]
+    pshufb          m2,                m15
+    vpermq          m2,                m2, 0x4E
+    psubw           m11,               m3, m2
+    paddw           m3,                m2
+    vperm2i128      m2,                m4, m3, 0x20        ; [row1lo  row3lo] for E
+    vperm2i128      m8,                m4, m3, 0x31        ; [row1hi  row3hi] for E
+    pshufb          m8,                m15
+    psubw           m3,                m2, m8
+    paddw           m2,                m8
+
+    vperm2i128      m4,                m10, m11, 0x20      ; [row1lo  row3lo] for O
+    vperm2i128      m7,                m10, m11, 0x31      ; [row1hi  row3hi] for O
+
+
+    DCT32_PASS_1    0 * 32,            0 * 64, 0, 2
+    DCT32_PASS_1    2 * 32,            2 * 64, 1, 3
+    DCT32_PASS_1    4 * 32,            4 * 64, 0, 2
+    DCT32_PASS_1    6 * 32,            6 * 64, 1, 3
+    DCT32_PASS_1    8 * 32,            8 * 64, 0, 2
+    DCT32_PASS_1    10 * 32,           10 * 64, 1, 3
+    DCT32_PASS_1    12 * 32,           12 * 64, 0, 2
+    DCT32_PASS_1    14 * 32,           14 * 64, 1, 3
+    DCT32_PASS_1    16 * 32,           16 * 64, 0, 2
+    DCT32_PASS_1    18 * 32,           18 * 64, 1, 3
+    DCT32_PASS_1    20 * 32,           20 * 64, 0, 2
+    DCT32_PASS_1    22 * 32,           22 * 64, 1, 3
+    DCT32_PASS_1    24 * 32,           24 * 64, 0, 2
+    DCT32_PASS_1    26 * 32,           26 * 64, 1, 3
+    DCT32_PASS_1    28 * 32,           28 * 64, 0, 2
+    DCT32_PASS_1    30 * 32,           30 * 64, 1, 3
+
+    add             r5,                8
+    lea             r0,                [r0 + r2 * 4]
+
+    dec             r4d
+    jnz             .pass1
+
+    mov             r2d,               128
+    lea             r3,                [r2 * 3]
+    mov             r5,                rsp
+    mov             r4d,               8
+    vpbroadcastq    m9,                [pd_1024]
+
+.pass2:
+    mova            m0,                [r5 + 0 * 64]
+    mova            m1,                [r5 + 0 * 64 + 32]
+
+    mova            m2,                [r5 + 1 * 64]
+    mova            m3,                [r5 + 1 * 64 + 32]
+
+    mova            m4,                [r5 + 2 * 64]
+    mova            m5,                [r5 + 2 * 64 + 32]
+
+    mova            m6,                [r5 + 3 * 64]
+    mova            m7,                [r5 + 3 * 64 + 32]
+
+    DCT32_PASS_2    0 * 32
+    movu            [r1],              xm11
+    DCT32_PASS_2    1 * 32
+    movu            [r1 + r2],         xm11
+    DCT32_PASS_2    2 * 32
+    movu            [r1 + r2 * 2],     xm11
+    DCT32_PASS_2    3 * 32
+    movu            [r1 + r3],         xm11
+
+    lea             r6,                [r1 + r2 * 4]
+    DCT32_PASS_2    4 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    5 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    6 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    7 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    8 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    9 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    10 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    11 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    12 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    13 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    14 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    15 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    16 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    17 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    18 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    19 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    20 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    21 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    22 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    23 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    24 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    25 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    26 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    27 * 32
+    movu            [r6 + r3],         xm11
+
+    lea             r6,                [r6 + r2 * 4]
+    DCT32_PASS_2    28 * 32
+    movu            [r6],              xm11
+    DCT32_PASS_2    29 * 32
+    movu            [r6 + r2],         xm11
+    DCT32_PASS_2    30 * 32
+    movu            [r6 + r2 * 2],     xm11
+    DCT32_PASS_2    31 * 32
+    movu            [r6 + r3],         xm11
+
+    add             r5,                256
+    add             r1,                16
+
+    dec             r4d
+    jnz             .pass2
+    RET
+
+%macro IDCT8_PASS_1 1
+    vpbroadcastd    m7,                [r5 + %1]
+    vpbroadcastd    m10,               [r5 + %1 + 4]
+    pmaddwd         m5,                m4, m7
+    pmaddwd         m6,                m0, m10
+    paddd           m5,                m6
+
+    vpbroadcastd    m7,                [r6 + %1]
+    vpbroadcastd    m10,               [r6 + %1 + 4]
+    pmaddwd         m6,                m1, m7
+    pmaddwd         m3,                m2, m10
+    paddd           m6,                m3
+
+    paddd           m3,                m5, m6
+    paddd           m3,                m11
+    psrad           m3,                IDCT_SHIFT1
+
+    psubd           m5,                m6
+    paddd           m5,                m11
+    psrad           m5,                IDCT_SHIFT1
+
+    vpbroadcastd    m7,                [r5 + %1 + 32]
+    vpbroadcastd    m10,               [r5 + %1 + 36]
+    pmaddwd         m6,                m4, m7
+    pmaddwd         m8,                m0, m10
+    paddd           m6,                m8
+
+    vpbroadcastd    m7,                [r6 + %1 + 32]
+    vpbroadcastd    m10,               [r6 + %1 + 36]
+    pmaddwd         m8,                m1, m7
+    pmaddwd         m9,                m2, m10
+    paddd           m8,                m9
+
+    paddd           m9,                m6, m8
+    paddd           m9,                m11
+    psrad           m9,                IDCT_SHIFT1
+
+    psubd           m6,                m8
+    paddd           m6,                m11
+    psrad           m6,                IDCT_SHIFT1
+
+    packssdw        m3,                m9
+    vpermq          m3,                m3, 0xD8
+
+    packssdw        m6,                m5
+    vpermq          m6,                m6, 0xD8
+%endmacro
+
+%macro IDCT8_PASS_2 0
+    punpcklqdq      m2,                m0, m1
+    punpckhqdq      m0,                m1
+
+    pmaddwd         m3,                m2, [r5]
+    pmaddwd         m5,                m2, [r5 + 32]
+    pmaddwd         m6,                m2, [r5 + 64]
+    pmaddwd         m7,                m2, [r5 + 96]
+    phaddd          m3,                m5
+    phaddd          m6,                m7
+    pshufb          m3,                [idct8_shuf2]
+    pshufb          m6,                [idct8_shuf2]
+    punpcklqdq      m7,                m3, m6
+    punpckhqdq      m3,                m6
+
+    pmaddwd         m5,                m0, [r6]
+    pmaddwd         m6,                m0, [r6 + 32]
+    pmaddwd         m8,                m0, [r6 + 64]
+    pmaddwd         m9,                m0, [r6 + 96]
+    phaddd          m5,                m6
+    phaddd          m8,                m9
+    pshufb          m5,                [idct8_shuf2]
+    pshufb          m8,                [idct8_shuf2]
+    punpcklqdq      m6,                m5, m8
+    punpckhqdq      m5,                m8
+
+    paddd           m8,                m7, m6
+    paddd           m8,                m12
+    psrad           m8,                IDCT_SHIFT2
+
+    psubd           m7,                m6
+    paddd           m7,                m12
+    psrad           m7,                IDCT_SHIFT2
+
+    pshufb          m7,                [idct8_shuf3]
+    packssdw        m8,                 m7
+
+    paddd           m9,                m3, m5
+    paddd           m9,                m12
+    psrad           m9,                IDCT_SHIFT2
+
+    psubd           m3,                m5
+    paddd           m3,                m12
+    psrad           m3,                IDCT_SHIFT2
+
+    pshufb          m3,                [idct8_shuf3]
+    packssdw        m9,                m3
+%endmacro
+
+INIT_YMM avx2
+cglobal idct8, 3, 7, 13, 0-8*16
+%if BIT_DEPTH == 10
+    %define         IDCT_SHIFT2        10
+    vpbroadcastd    m12,                [pd_512]
+%elif BIT_DEPTH == 8
+    %define         IDCT_SHIFT2        12
+    vpbroadcastd    m12,                [pd_2048]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+%define             IDCT_SHIFT1         7
+
+    vbroadcasti128  m11,               [pd_64]
+
+    mov             r4,                rsp
+    lea             r5,                [avx2_idct8_1]
+    lea             r6,                [avx2_idct8_2]
+
+    ;pass1
+    mova            m0,                [r0 + 0 * 32]
+    mova            m1,                [r0 + 4 * 32]
+    packssdw        m0,                m1               ; [0 0 0 0 4 4 4 4 0 0 0 0 4 4 4 4]
+    mova            m1,                [r0 + 2 * 32]
+    mova            m2,                [r0 + 6 * 32]
+    packssdw        m1,                m2               ; [2 2 2 2 6 6 6 6 2 2 2 2 6 6 6 6]
+    mova            m2,                [r0 + 1 * 32]
+    mova            m3,                [r0 + 5 * 32]
+    packssdw        m2,                m3               ; [1 1 1 1 5 5 5 5 1 1 1 1 5 5 5 5]
+    mova            m3,                [r0 + 3 * 32]
+    mova            m4,                [r0 + 7 * 32]
+    packssdw        m3,                m4               ; [3 3 3 3 7 7 7 7 3 3 3 3 7 7 7 7]
+
+    mova            m5,                [idct8_shuf1]
+
+    punpcklwd       m4,                m0, m1           ; [0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2]
+    punpckhwd       m0,                m1               ; [4 6 4 6 4 6 4 6 4 6 4 6 4 6 4 6]
+    vpermd          m4,                m5, m4
+    vpermd          m0,                m5, m0
+
+    punpcklwd       m1,                m2, m3           ; [1 3 1 3 1 3 1 3 1 3 1 3 1 3 1 3]
+    punpckhwd       m2,                m3               ; [5 7 5 7 5 7 5 7 5 7 5 7 5 7 5 7]
+    vpermd          m1,                m5, m1
+    vpermd          m2,                m5, m2
+
+    IDCT8_PASS_1    0
+    mova            [r4],              m3
+    mova            [r4 + 96],         m6
+
+    IDCT8_PASS_1    64
+    mova            [r4 + 32],         m3
+    mova            [r4 + 64],         m6
+
+    ;pass2
+    add             r2d,               r2d
+    lea             r3,                [r2 * 3]
+
+    mova            m0,                [r4]
+    mova            m1,                [r4 + 32]
+    IDCT8_PASS_2
+
+    vextracti128    xm3,               m8, 1
+    mova            [r1],              xm8
+    mova            [r1 + r2],         xm3
+    vextracti128    xm3,               m9, 1
+    mova            [r1 + r2 * 2],     xm9
+    mova            [r1 + r3],         xm3
+
+    lea             r1,                [r1 + r2 * 4]
+    mova            m0,                [r4 + 64]
+    mova            m1,                [r4 + 96]
+    IDCT8_PASS_2
+
+    vextracti128    xm3,               m8, 1
+    mova            [r1],              xm8
+    mova            [r1 + r2],         xm3
+    vextracti128    xm3,               m9, 1
+    mova            [r1 + r2 * 2],     xm9
+    mova            [r1 + r3],         xm3
+    RET
+
+%macro IDCT_PASS1 2
+    vbroadcasti128  m5, [tab_idct16_2 + %1 * 16]
+
+    pmaddwd         m9, m0, m5
+    pmaddwd         m10, m7, m5
+    phaddd          m9, m10
+
+    pmaddwd         m10, m6, m5
+    pmaddwd         m11, m8, m5
+    phaddd          m10, m11
+
+    phaddd          m9, m10
+    vbroadcasti128  m5, [tab_idct16_1 + %1 * 16]
+
+    pmaddwd         m10, m1, m5
+    pmaddwd         m11, m3, m5
+    phaddd          m10, m11
+
+    pmaddwd         m11, m4, m5
+    pmaddwd         m12, m2, m5
+    phaddd          m11, m12
+
+    phaddd          m10, m11
+
+    paddd           m11, m9, m10
+    paddd           m11, m14
+    psrad           m11, IDCT_SHIFT1
+
+    psubd           m9, m10
+    paddd           m9, m14
+    psrad           m9, IDCT_SHIFT1
+
+    vbroadcasti128  m5, [tab_idct16_2 + %1 * 16 + 16]
+
+    pmaddwd         m10, m0, m5
+    pmaddwd         m12, m7, m5
+    phaddd          m10, m12
+
+    pmaddwd         m12, m6, m5
+    pmaddwd         m13, m8, m5
+    phaddd          m12, m13
+
+    phaddd          m10, m12
+    vbroadcasti128  m5, [tab_idct16_1 + %1 * 16  + 16]
+
+    pmaddwd         m12, m1, m5
+    pmaddwd         m13, m3, m5
+    phaddd          m12, m13
+
+    pmaddwd         m13, m4, m5
+    pmaddwd         m5, m2
+    phaddd          m13, m5
+
+    phaddd          m12, m13
+
+    paddd           m5, m10, m12
+    paddd           m5, m14
+    psrad           m5, IDCT_SHIFT1
+
+    psubd           m10, m12
+    paddd           m10, m14
+    psrad           m10, IDCT_SHIFT1
+
+    packssdw        m11, m5
+    packssdw        m9, m10
+
+    mova            m10, [idct16_shuff]
+    mova            m5,  [idct16_shuff1]
+
+    vpermd          m12, m10, m11
+    vpermd          m13, m5, m9
+    mova            [r3 + %1 * 16 * 2], xm12
+    mova            [r3 + %2 * 16 * 2], xm13
+    vextracti128    [r3 + %2 * 16 * 2 + 32], m13, 1
+    vextracti128    [r3 + %1 * 16 * 2 + 32], m12, 1
+%endmacro
+
+;-------------------------------------------------------
+; void idct16(int32_t *src, int16_t *dst, intptr_t stride)
+;-------------------------------------------------------
+INIT_YMM avx2
+cglobal idct16, 3, 7, 16, 0-16*mmsize
+%if BIT_DEPTH == 10
+    %define         IDCT_SHIFT2        10
+    vpbroadcastd    m15,                [pd_512]
+%elif BIT_DEPTH == 8
+    %define         IDCT_SHIFT2        12
+    vpbroadcastd    m15,                [pd_2048]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+%define             IDCT_SHIFT1         7
+
+    vbroadcasti128  m14,               [pd_64]
+
+    add             r2d,               r2d
+    mov             r3, rsp
+    mov             r4d, 2
+
+.pass1:
+    movu            m0, [r0 +  0 * 64]
+    movu            m1, [r0 +  8 * 64]
+    packssdw        m0, m1                    ;[0L 8L 0H 8H]
+
+    movu            m1, [r0 +  1 * 64]
+    movu            m2, [r0 +  9 * 64]
+    packssdw        m1, m2                    ;[1L 9L 1H 9H]
+
+    movu            m2, [r0 +  2 * 64]
+    movu            m3, [r0 + 10 * 64]
+    packssdw        m2, m3                    ;[2L 10L 2H 10H]
+
+    movu            m3, [r0 +  3 * 64]
+    movu            m4, [r0 + 11 * 64]
+    packssdw        m3, m4                    ;[3L 11L 3H 11H]
+
+    movu            m4, [r0 +  4 * 64]
+    movu            m5, [r0 + 12 * 64]
+    packssdw        m4, m5                    ;[4L 12L 4H 12H]
+
+    movu            m5, [r0 +  5 * 64]
+    movu            m6, [r0 + 13 * 64]
+    packssdw        m5, m6                    ;[5L 13L 5H 13H]
+
+    movu            m6, [r0 +  6 * 64]
+    movu            m7, [r0 + 14 * 64]
+    packssdw        m6, m7                    ;[6L 14L 6H 14H]
+
+    movu            m7, [r0 +  7 * 64]
+    movu            m8, [r0 + 15 * 64]
+    packssdw        m7, m8                    ;[7L 15L 7H 15H]
+
+    punpckhwd       m8, m0, m2                ;[8 10]
+    punpcklwd       m0, m2                    ;[0 2]
+
+    punpckhwd       m2, m1, m3                ;[9 11]
+    punpcklwd       m1, m3                    ;[1 3]
+
+    punpckhwd       m3, m4, m6                ;[12 14]
+    punpcklwd       m4, m6                    ;[4 6]
+
+    punpckhwd       m6, m5, m7                ;[13 15]
+    punpcklwd       m5, m7                    ;[5 7]
+
+    punpckhdq       m7, m0, m4                ;[02 22 42 62 03 23 43 63 06 26 46 66 07 27 47 67]
+    punpckldq       m0, m4                    ;[00 20 40 60 01 21 41 61 04 24 44 64 05 25 45 65]
+
+    punpckhdq       m4, m8, m3                ;[82 102 122 142 83 103 123 143 86 106 126 146 87 107 127 147]
+    punpckldq       m8, m3                    ;[80 100 120 140 81 101 121 141 84 104 124 144 85 105 125 145]
+
+    punpckhdq       m3, m1, m5                ;[12 32 52 72 13 33 53 73 16 36 56 76 17 37 57 77]
+    punpckldq       m1, m5                    ;[10 30 50 70 11 31 51 71 14 34 54 74 15 35 55 75]
+
+    punpckhdq       m5, m2, m6                ;[92 112 132 152 93 113 133 153 96 116 136 156 97 117 137 157]
+    punpckldq       m2, m6                    ;[90 110 130 150 91 111 131 151 94 114 134 154 95 115 135 155]
+
+    punpckhqdq      m6, m0, m8                ;[01 21 41 61 81 101 121 141 05 25 45 65 85 105 125 145]
+    punpcklqdq      m0, m8                    ;[00 20 40 60 80 100 120 140 04 24 44 64 84 104 124 144]
+
+    punpckhqdq      m8, m7, m4                ;[03 23 43 63 43 103 123 143 07 27 47 67 87 107 127 147]
+    punpcklqdq      m7, m4                    ;[02 22 42 62 82 102 122 142 06 26 46 66 86 106 126 146]
+
+    punpckhqdq      m4, m1, m2                ;[11 31 51 71 91 111 131 151 15 35 55 75 95 115 135 155]
+    punpcklqdq      m1, m2                    ;[10 30 50 70 90 110 130 150 14 34 54 74 94 114 134 154]
+
+    punpckhqdq      m2, m3, m5                ;[13 33 53 73 93 113 133 153 17 37 57 77 97 117 137 157]
+    punpcklqdq      m3, m5                    ;[12 32 52 72 92 112 132 152 16 36 56 76 96 116 136 156]
+
+    IDCT_PASS1      0, 14
+    IDCT_PASS1      2, 12
+    IDCT_PASS1      4, 10
+    IDCT_PASS1      6, 8
+
+    add             r0, 32
+    add             r3, 16
+    dec             r4d
+    jnz             .pass1
+
+    mov             r3, rsp
+    mov             r4d, 8
+    lea             r5, [tab_idct16_2]
+    lea             r6, [tab_idct16_1]
+
+    vbroadcasti128  m7,  [r5]
+    vbroadcasti128  m8,  [r5 + 16]
+    vbroadcasti128  m9,  [r5 + 32]
+    vbroadcasti128  m10, [r5 + 48]
+    vbroadcasti128  m11, [r5 + 64]
+    vbroadcasti128  m12, [r5 + 80]
+    vbroadcasti128  m13, [r5 + 96]
+
+.pass2:
+    movu            m1, [r3]
+    vpermq          m0, m1, 0xD8
+
+    pmaddwd         m1, m0, m7
+    pmaddwd         m2, m0, m8
+    phaddd          m1, m2
+
+    pmaddwd         m2, m0, m9
+    pmaddwd         m3, m0, m10
+    phaddd          m2, m3
+
+    phaddd          m1, m2
+
+    pmaddwd         m2, m0, m11
+    pmaddwd         m3, m0, m12
+    phaddd          m2, m3
+
+    vbroadcasti128  m14, [r5 + 112]
+    pmaddwd         m3, m0, m13
+    pmaddwd         m4, m0, m14
+    phaddd          m3, m4
+
+    phaddd          m2, m3
+
+    movu            m3, [r3 + 32]
+    vpermq          m0, m3, 0xD8
+
+    vbroadcasti128  m14, [r6]
+    pmaddwd         m3, m0, m14
+    vbroadcasti128  m14, [r6 + 16]
+    pmaddwd         m4, m0, m14
+    phaddd          m3, m4
+
+    vbroadcasti128  m14, [r6 + 32]
+    pmaddwd         m4, m0, m14
+    vbroadcasti128  m14, [r6 + 48]
+    pmaddwd         m5, m0, m14
+    phaddd          m4, m5
+
+    phaddd          m3, m4
+
+    vbroadcasti128  m14, [r6 + 64]
+    pmaddwd         m4, m0, m14
+    vbroadcasti128  m14, [r6 + 80]
+    pmaddwd         m5, m0, m14
+    phaddd          m4, m5
+
+    vbroadcasti128  m14, [r6 + 96]
+    pmaddwd         m6, m0, m14
+    vbroadcasti128  m14, [r6 + 112]
+    pmaddwd         m0, m14
+    phaddd          m6, m0
+
+    phaddd          m4, m6
+
+    paddd           m5, m1, m3
+    paddd           m5, m15
+    psrad           m5, IDCT_SHIFT2
+
+    psubd           m1, m3
+    paddd           m1, m15
+    psrad           m1, IDCT_SHIFT2
+
+    paddd           m6, m2, m4
+    paddd           m6, m15
+    psrad           m6, IDCT_SHIFT2
+
+    psubd           m2, m4
+    paddd           m2, m15
+    psrad           m2, IDCT_SHIFT2
+
+    packssdw        m5, m6
+    packssdw        m1, m2
+    pshufb          m2, m1, [dct16_shuf1]
+
+    mova            [r1], xm5
+    mova            [r1 + 16], xm2
+    vextracti128    [r1 + r2], m5, 1
+    vextracti128    [r1 + r2 + 16], m2, 1
+
+    lea             r1, [r1 + 2 * r2]
+    add             r3, 64
+    dec             r4d
+    jnz             .pass2
     RET
+
+%macro IDCT32_PASS1 1
+    vbroadcasti128  m3, [tab_idct32_1 + %1 * 32]
+    vbroadcasti128  m13, [tab_idct32_1 + %1 * 32 + 16]
+    pmaddwd         m9, m4, m3
+    pmaddwd         m10, m8, m13
+    phaddd          m9, m10
+
+    pmaddwd         m10, m2, m3
+    pmaddwd         m11, m1, m13
+    phaddd          m10, m11
+
+    phaddd          m9, m10
+
+    vbroadcasti128  m3, [tab_idct32_1 + (15 - %1) * 32]
+    vbroadcasti128  m13, [tab_idct32_1 + (15- %1) * 32 + 16]
+    pmaddwd         m10, m4, m3
+    pmaddwd         m11, m8, m13
+    phaddd          m10, m11
+
+    pmaddwd         m11, m2, m3
+    pmaddwd         m12, m1, m13
+    phaddd          m11, m12
+
+    phaddd          m10, m11
+    phaddd          m9, m10                       ;[row0s0 row2s0 row0s15 row2s15 row1s0 row3s0 row1s15 row3s15]
+
+    vbroadcasti128  m3, [tab_idct32_2 + %1 * 16]
+    pmaddwd         m10, m0, m3
+    pmaddwd         m11, m7, m3
+    phaddd          m10, m11
+    phaddd          m10, m10
+
+    vbroadcasti128  m3, [tab_idct32_3 + %1 * 16]
+    pmaddwd         m11, m5, m3
+    pmaddwd         m12, m6, m3
+    phaddd          m11, m12
+    phaddd          m11, m11
+
+    paddd           m12, m10, m11                 ;[row0a0 row2a0 NIL NIL row1sa0 row3a0 NIL NIL]
+    psubd           m10, m11                      ;[row0a15 row2a15 NIL NIL row1a15 row3a15 NIL NIL]
+
+    punpcklqdq      m12, m10                      ;[row0a0 row2a0 row0a15 row2a15 row1a0 row3a0 row1a15 row3a15]
+    paddd           m10, m9, m12
+    paddd           m10, m15
+    psrad           m10, IDCT_SHIFT1
+
+    psubd           m12, m9
+    paddd           m12, m15
+    psrad           m12, IDCT_SHIFT1
+
+    packssdw        m10, m12
+    vextracti128    xm12, m10, 1
+    movd            [r3 + %1 * 64], xm10
+    movd            [r3 + 32 + %1 * 64], xm12
+    pextrd          [r4 - %1 * 64], xm10, 1
+    pextrd          [r4+ 32 - %1 * 64], xm12, 1
+    pextrd          [r3 + 16 * 64 + %1 *64], xm10, 3
+    pextrd          [r3 + 16 * 64 + 32 + %1 * 64], xm12, 3
+    pextrd          [r4 + 16 * 64 - %1 * 64], xm10, 2
+    pextrd          [r4 + 16 * 64 + 32 - %1 * 64], xm12, 2
+%endmacro
+
+;-------------------------------------------------------
+; void idct32(int32_t *src, int16_t *dst, intptr_t stride)
+;-------------------------------------------------------
+
+; TODO: Reduce PHADDD instruction by PADDD
+
+INIT_YMM avx2
+cglobal idct32, 3, 6, 16, 0-32*64
+
+%define             IDCT_SHIFT1         7
+
+    vbroadcasti128  m15, [pd_64]
+
+    mov             r3, rsp
+    lea             r4, [r3 + 15 * 64]
+    mov             r5d, 8
+
+.pass1:
+    movu            xm0,    [r0 +  2 * 128]
+    movu            xm1,    [r0 + 18 * 128]
+    vinserti128     m0, m0, [r0 +  0 * 128], 1
+    vinserti128     m1, m1, [r0 + 16 * 128], 1
+
+    packssdw        m0, m1                      ;[2 18 0 16]
+
+    movu            xm1,    [r0 +  1 * 128]
+    movu            xm2,    [r0 +  9 * 128]
+    vinserti128     m1, m1, [r0 + 17 * 128], 1
+    vinserti128     m2, m2, [r0 + 25 * 128], 1
+    packssdw        m1, m2                      ;[1 9 17 25]
+
+    movu            xm2,    [r0 +  6 * 128]
+    movu            xm3,    [r0 + 22 * 128]
+    vinserti128     m2, m2, [r0 +  4 * 128], 1
+    vinserti128     m3, m3, [r0 + 20 * 128], 1
+    packssdw        m2, m3                      ;[6 22 4 20]
+
+    movu            xm3,    [r0 +  3 * 128]
+    movu            xm4,    [r0 + 11 * 128]
+    vinserti128     m3, m3, [r0 + 19 * 128], 1
+    vinserti128     m4, m4, [r0 + 27 * 128], 1
+    packssdw        m3, m4                      ;[3 11 19 27]
+
+    movu            xm4,    [r0 + 10 * 128]
+    movu            xm5,    [r0 + 26 * 128]
+    vinserti128     m4, m4, [r0 +  8 * 128], 1
+    vinserti128     m5, m5, [r0 + 24 * 128], 1
+    packssdw        m4, m5                      ;[10 26 8 24]
+
+    movu            xm5,    [r0 +  5 * 128]
+    movu            xm6,    [r0 + 13 * 128]
+    vinserti128     m5, m5, [r0 + 21 * 128], 1
+    vinserti128     m6, m6, [r0 + 29 * 128], 1
+    packssdw        m5, m6                      ;[5 13 21 29]
+
+    movu            xm6,    [r0 + 14 * 128]
+    movu            xm7,    [r0 + 30 * 128]
+    vinserti128     m6, m6, [r0 + 12 * 128], 1
+    vinserti128     m7, m7, [r0 + 28 * 128], 1
+    packssdw        m6, m7                      ;[14 30 12 28]
+
+    movu            xm7,    [r0 +  7 * 128]
+    movu            xm8,    [r0 + 15 * 128]
+    vinserti128     m7, m7, [r0 + 23 * 128], 1
+    vinserti128     m8, m8, [r0 + 31 * 128], 1
+    packssdw        m7, m8                      ;[7 15 23 31]
+
+    punpckhwd       m8, m0, m2                  ;[18 22 16 20]
+    punpcklwd       m0, m2                      ;[2 6 0 4]
+
+    punpckhwd       m2, m1, m3                  ;[9 11 25 27]
+    punpcklwd       m1, m3                      ;[1 3 17 19]
+
+    punpckhwd       m3, m4, m6                  ;[26 30 24 28]
+    punpcklwd       m4, m6                      ;[10 14 8 12]
+
+    punpckhwd       m6, m5, m7                  ;[13 15 29 31]
+    punpcklwd       m5, m7                      ;[5 7 21 23]
+
+    punpckhdq       m7, m0, m4                  ;[22 62 102 142 23 63 103 143 02 42 82 122 03 43 83 123]
+    punpckldq       m0, m4                      ;[20 60 100 140 21 61 101 141 00 40 80 120 01 41 81 121]
+
+    punpckhdq       m4, m8, m3                  ;[182 222 262 302 183 223 263 303 162 202 242 282 163 203 243 283]
+    punpckldq       m8, m3                      ;[180 220 260 300 181 221 261 301 160 200 240 280 161 201 241 281]
+
+    punpckhdq       m3, m1, m5                  ;[12 32 52 72 13 33 53 73 172 192 212 232 173 193 213 233]
+    punpckldq       m1, m5                      ;[10 30 50 70 11 31 51 71 170 190 210 230 171 191 211 231]
+
+    punpckhdq       m5, m2, m6                  ;[92 112 132 152 93 113 133 153 252 272 292 312 253 273 293 313]
+    punpckldq       m2, m6                      ;[90 110 130 150 91 111 131 151 250 270 290 310 251 271 291 311]
+
+    punpckhqdq      m6, m0, m8                  ;[21 61 101 141 181 221 261 301 01 41 81 121 161 201 241 281]
+    punpcklqdq      m0, m8                      ;[20 60 100 140 180 220 260 300 00 40 80 120 160 200 240 280]
+
+    punpckhqdq      m8, m7, m4                  ;[23 63 103 143 183 223 263 303 03 43 83 123 163 203 243 283]
+    punpcklqdq      m7, m4                      ;[22 62 102 142 182 222 262 302 02 42 82 122 162 202 242 282]
+
+    punpckhqdq      m4, m1, m2                  ;[11 31 51 71 91 111 131 151 171 191 211 231 251 271 291 311]
+    punpcklqdq      m1, m2                      ;[10 30 50 70 90 110 130 150 170 190 210 230 250 270 290 310]
+
+    punpckhqdq      m2, m3, m5                  ;[13 33 53 73 93 113 133 153 173 193 213 233 253 273 293 313]
+    punpcklqdq      m3, m5                      ;[12 32 52 72 92 112 132 152 172 192 212 232 252 272 292 312]
+
+    vperm2i128      m5, m0, m6, 0x20            ;[20 60 100 140 180 220 260 300 21 61 101 141 181 221 261 301]
+    vperm2i128      m0, m0, m6, 0x31            ;[00 40 80 120 160 200 240 280 01 41 81 121 161 201 241 281]
+
+    vperm2i128      m6, m7, m8, 0x20            ;[22 62 102 142 182 222 262 302 23 63 103 143 183 223 263 303]
+    vperm2i128      m7, m7, m8, 0x31            ;[02 42 82 122 162 202 242 282 03 43 83 123 163 203 243 283]
+
+    vperm2i128      m8, m1, m4, 0x31            ;[170 190 210 230 250 270 290 310 171 191 211 231 251 271 291 311]
+    vperm2i128      m4, m1, m4, 0x20            ;[10 30 50 70 90 110 130 150 11 31 51 71 91 111 131 151]
+
+    vperm2i128      m1, m3, m2, 0x31            ;[172 192 212 232 252 272 292 312 173 193 213 233 253 273 293 313]
+    vperm2i128      m2, m3, m2, 0x20            ;[12 32 52 72 92 112 132 152 13 33 53 73 93 113 133 153]
+
+    IDCT32_PASS1 0
+    IDCT32_PASS1 1
+    IDCT32_PASS1 2
+    IDCT32_PASS1 3
+    IDCT32_PASS1 4
+    IDCT32_PASS1 5
+    IDCT32_PASS1 6
+    IDCT32_PASS1 7
+
+    add             r0, 16
+    add             r3, 4
+    add             r4, 4
+    dec             r5d
+    jnz             .pass1
+
+%if BIT_DEPTH == 10
+    %define         IDCT_SHIFT2        10
+    vpbroadcastd    m15,                [pd_512]
+%elif BIT_DEPTH == 8
+    %define         IDCT_SHIFT2        12
+    vpbroadcastd    m15,                [pd_2048]
+%else
+    %error Unsupported BIT_DEPTH!
+%endif
+
+    mov             r3, rsp
+    add             r2d, r2d
+    mov             r4d, 32
+
+    mova            m7,  [tab_idct32_4]
+    mova            m8,  [tab_idct32_4 + 32]
+    mova            m9,  [tab_idct32_4 + 64]
+    mova            m10, [tab_idct32_4 + 96]
+    mova            m11, [tab_idct32_4 + 128]
+    mova            m12, [tab_idct32_4 + 160]
+    mova            m13, [tab_idct32_4 + 192]
+    mova            m14, [tab_idct32_4 + 224]
+.pass2:
+    movu            m0, [r3]
+    movu            m1, [r3 + 32]
+
+    pmaddwd         m2, m0, m7
+    pmaddwd         m3, m0, m8
+    phaddd          m2, m3
+
+    pmaddwd         m3, m0, m9
+    pmaddwd         m4, m0, m10
+    phaddd          m3, m4
+
+    phaddd          m2, m3
+
+    pmaddwd         m3, m0, m11
+    pmaddwd         m4, m0, m12
+    phaddd          m3, m4
+
+    pmaddwd         m4, m0, m13
+    pmaddwd         m5, m0, m14
+    phaddd          m4, m5
+
+    phaddd          m3, m4
+
+    vperm2i128      m4, m2, m3, 0x31
+    vperm2i128      m2, m2, m3, 0x20
+    paddd           m2, m4
+
+    pmaddwd         m3, m0, [tab_idct32_4 + 256]
+    pmaddwd         m4, m0, [tab_idct32_4 + 288]
+    phaddd          m3, m4
+
+    pmaddwd         m4, m0, [tab_idct32_4 + 320]
+    pmaddwd         m5, m0, [tab_idct32_4 + 352]
+    phaddd          m4, m5
+
+    phaddd          m3, m4
+
+    pmaddwd         m4, m0, [tab_idct32_4 + 384]
+    pmaddwd         m5, m0, [tab_idct32_4 + 416]
+    phaddd          m4, m5
+
+    pmaddwd         m5, m0, [tab_idct32_4 + 448]
+    pmaddwd         m0,     [tab_idct32_4 + 480]
+    phaddd          m5, m0
+
+    phaddd          m4, m5
+
+    vperm2i128      m0, m3, m4, 0x31
+    vperm2i128      m3, m3, m4, 0x20
+    paddd           m3, m0
+
+    pmaddwd         m4, m1, [tab_idct32_1]
+    pmaddwd         m0, m1, [tab_idct32_1 + 32]
+    phaddd          m4, m0
+
+    pmaddwd         m5, m1, [tab_idct32_1 + 64]
+    pmaddwd         m0, m1, [tab_idct32_1 + 96]
+    phaddd          m5, m0
+
+    phaddd          m4, m5
+
+    pmaddwd         m5, m1, [tab_idct32_1 + 128]
+    pmaddwd         m0, m1, [tab_idct32_1 + 160]
+    phaddd          m5, m0
+
+    pmaddwd         m6, m1, [tab_idct32_1 + 192]
+    pmaddwd         m0, m1, [tab_idct32_1 + 224]
+    phaddd          m6, m0
+
+    phaddd          m5, m6
+
+    vperm2i128      m0, m4, m5, 0x31
+    vperm2i128      m4, m4, m5, 0x20
+    paddd           m4, m0
+
+    pmaddwd         m5, m1, [tab_idct32_1 + 256]
+    pmaddwd         m0, m1, [tab_idct32_1 + 288]
+    phaddd          m5, m0
+
+    pmaddwd         m6, m1, [tab_idct32_1 + 320]
+    pmaddwd         m0, m1, [tab_idct32_1 + 352]
+    phaddd          m6, m0
+
+    phaddd          m5, m6
+
+    pmaddwd         m6, m1, [tab_idct32_1 + 384]
+    pmaddwd         m0, m1, [tab_idct32_1 + 416]
+    phaddd          m6, m0
+
+    pmaddwd         m0, m1, [tab_idct32_1 + 448]
+    pmaddwd         m1,     [tab_idct32_1 + 480]
+    phaddd          m0, m1
+
+    phaddd          m6, m0
+
+    vperm2i128      m0, m5, m6, 0x31
+    vperm2i128      m5, m5, m6, 0x20
+    paddd           m5, m0
+
+    paddd           m6, m2, m4
+    paddd           m6, m15
+    psrad           m6, IDCT_SHIFT2
+
+    psubd           m2, m4
+    paddd           m2, m15
+    psrad           m2, IDCT_SHIFT2
+
+    paddd           m4, m3, m5
+    paddd           m4, m15
+    psrad           m4, IDCT_SHIFT2
+
+    psubd           m3, m5
+    paddd           m3, m15
+    psrad           m3, IDCT_SHIFT2
+
+    packssdw        m6, m4
+    packssdw        m2, m3
+
+    vpermq          m6, m6, 0xD8
+    vpermq          m2, m2, 0x8D
+    pshufb          m2, [dct16_shuf1]
+
+    mova            [r1], m6
+    mova            [r1 + 32], m2
+
+    add             r1, r2
+    add             r3, 64
+    dec             r4d
+    jnz             .pass2
+    RET
+%endif

x265-hg20140324.tar.bz2/source/common/x86/dct8.h -> x265-hg20140928.tar.bz2/source/common/x86/dct8.h Changed

@@ -18,17 +18,27 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_DCT8_H
 #define X265_DCT8_H
-
 void x265_dct4_sse2(int16_t *src, int32_t *dst, intptr_t stride);
-void x265_idct4_sse2(int32_t *src, int16_t *dst, intptr_t stride);
-void x265_idct8_ssse3(int32_t *src, int16_t *dst, intptr_t stride);
 void x265_dst4_ssse3(int16_t *src, int32_t *dst, intptr_t stride);
-void x265_idst4_sse2(int32_t *src, int16_t *dst, intptr_t stride);
 void x265_dct8_sse4(int16_t *src, int32_t *dst, intptr_t stride);
+void x265_dct4_avx2(int16_t *src, int32_t *dst, intptr_t stride);
+void x265_dct8_avx2(int16_t *src, int32_t *dst, intptr_t stride);
+void x265_dct16_avx2(int16_t *src, int32_t *dst, intptr_t stride);
+void x265_dct32_avx2(int16_t *src, int32_t *dst, intptr_t stride);
+void x265_idct32_avx2(int32_t *src, int16_t *dst, intptr_t stride);
+
+void x265_idst4_sse2(int32_t *src, int16_t *dst, intptr_t stride);
+void x265_idct4_sse2(int32_t *src, int16_t *dst, intptr_t stride);
+void x265_idct8_ssse3(int32_t *src, int16_t *dst, intptr_t stride);
+void x265_idct8_avx2(int32_t *src, int16_t *dst, intptr_t stride);
+void x265_idct16_avx2(int32_t *src, int16_t *dst, intptr_t stride);
+
+void x265_denoise_dct_sse4(int32_t *dct, uint32_t *sum, uint16_t *offset, int size);
+void x265_denoise_dct_avx2(int32_t *dct, uint32_t *sum, uint16_t *offset, int size);
 
 #endif // ifndef X265_DCT8_H

x265-hg20140324.tar.bz2/source/common/x86/intrapred.h -> x265-hg20140928.tar.bz2/source/common/x86/intrapred.h Changed

@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@x264.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_INTRAPRED_H
@@ -31,10 +31,10 @@
 void x265_intra_pred_dc16_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
 void x265_intra_pred_dc32_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
 
-void x265_intra_pred_planar4_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
-void x265_intra_pred_planar8_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
-void x265_intra_pred_planar16_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
-void x265_intra_pred_planar32_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int filter);
+void x265_intra_pred_planar4_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int);
+void x265_intra_pred_planar8_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int);
+void x265_intra_pred_planar16_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int);
+void x265_intra_pred_planar32_sse4(pixel* dst, intptr_t dstStride, pixel* above, pixel* left, int, int);
 
 #define DECL_ANG(bsize, mode, cpu) \
     void x265_intra_pred_ang ## bsize ## _ ## mode ## _ ## cpu(pixel * dst, intptr_t dstStride, pixel * refLeft, pixel * refAbove, int dirMode, int bFilter);
@@ -157,8 +157,8 @@
 DECL_ANG(32, 33, sse4);
 
 #undef DECL_ANG
-void x265_all_angs_pred_4x4_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma);
-void x265_all_angs_pred_8x8_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma);
-void x265_all_angs_pred_16x16_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma);
-void x265_all_angs_pred_32x32_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, bool bLuma);
+void x265_all_angs_pred_4x4_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma);
+void x265_all_angs_pred_8x8_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma);
+void x265_all_angs_pred_16x16_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma);
+void x265_all_angs_pred_32x32_sse4(pixel *dest, pixel *above0, pixel *left0, pixel *above1, pixel *left1, int bLuma);
 #endif // ifndef X265_INTRAPRED_H

x265-hg20140324.tar.bz2/source/common/x86/intrapred16.asm -> x265-hg20140928.tar.bz2/source/common/x86/intrapred16.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/intrapred8.asm -> x265-hg20140928.tar.bz2/source/common/x86/intrapred8.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/ipfilter16.asm -> x265-hg20140928.tar.bz2/source/common/x86/ipfilter16.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/ipfilter8.asm -> x265-hg20140928.tar.bz2/source/common/x86/ipfilter8.asm Changed

@@ -20,7 +20,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@multicorewareinc.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************/
 
 %include "x86inc.asm"
@@ -211,6 +211,41 @@
 
 RET
 
+;-----------------------------------------------------------------------------
+; void interp_4tap_horiz_pp_2x16(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal interp_4tap_horiz_pp_2x16, 4, 6, 5, src, srcstride, dst, dststride
+%define coef2       m4
+%define Tm0         m3
+%define t2          m2
+%define t1          m1
+%define t0          m0
+
+mov         r4d,        r4m
+
+%ifdef PIC
+lea         r5,          [tab_ChromaCoeff]
+movd        coef2,       [r5 + r4 * 4]
+%else
+movd        coef2,       [tab_ChromaCoeff + r4 * 4]
+%endif
+
+pshufd      coef2,       coef2,      0
+mova        t2,          [tab_c_512]
+mova        Tm0,         [tab_Tm]
+
+mov         r5d,        16/2
+
+.loop:
+FILTER_H4_w2_2   t0, t1, t2
+lea         srcq,       [srcq + srcstrideq * 2]
+lea         dstq,       [dstq + dststrideq * 2]
+dec         r5d
+jnz         .loop
+
+RET
+
 %macro FILTER_H4_w4_2 3
     movh        %2, [srcq - 1]
     pshufb      %2, %2, Tm0
@@ -350,6 +385,42 @@
 
 RET
 
+;-----------------------------------------------------------------------------
+; void interp_4tap_horiz_pp_4x32(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal interp_4tap_horiz_pp_4x32, 4, 6, 5, src, srcstride, dst, dststride
+%define coef2       m4
+%define Tm0         m3
+%define t2          m2
+%define t1          m1
+%define t0          m0
+
+mov         r4d,        r4m
+
+%ifdef PIC
+lea         r5,          [tab_ChromaCoeff]
+movd        coef2,       [r5 + r4 * 4]
+%else
+movd        coef2,       [tab_ChromaCoeff + r4 * 4]
+%endif
+
+pshufd      coef2,       coef2,      0
+mova        t2,          [tab_c_512]
+mova        Tm0,         [tab_Tm]
+
+mov         r5d,        32/2
+
+.loop:
+FILTER_H4_w4_2   t0, t1, t2
+lea         srcq,       [srcq + srcstrideq * 2]
+lea         dstq,       [dstq + dststrideq * 2]
+dec         r5d
+jnz         .loop
+
+RET
+
+
 %macro FILTER_H4_w6 3
     movu        %1, [srcq - 1]
     pshufb      %2, %1, Tm0
@@ -475,6 +546,38 @@
     movu        [dstq + 16],      %2
 %endmacro
 
+%macro FILTER_H4_w16o 5
+    movu        %1, [srcq + %5 - 1]
+    pshufb      %2, %1, Tm0
+    pmaddubsw   %2, coef2
+    pshufb      %1, %1, Tm1
+    pmaddubsw   %1, coef2
+    phaddw      %2, %1
+    movu        %1, [srcq + %5 - 1 + 8]
+    pshufb      %4, %1, Tm0
+    pmaddubsw   %4, coef2
+    pshufb      %1, %1, Tm1
+    pmaddubsw   %1, coef2
+    phaddw      %4, %1
+    pmulhrsw    %2, %3
+    pmulhrsw    %4, %3
+    packuswb    %2, %4
+    movu        [dstq + %5],      %2
+%endmacro
+
+%macro FILTER_H4_w48 4
+    FILTER_H4_w16o %1, %2, %3, %4, 0
+    FILTER_H4_w16o %1, %2, %3, %4, 16
+    FILTER_H4_w16o %1, %2, %3, %4, 32
+%endmacro
+
+%macro FILTER_H4_w64 4
+    FILTER_H4_w16o %1, %2, %3, %4, 0
+    FILTER_H4_w16o %1, %2, %3, %4, 16
+    FILTER_H4_w16o %1, %2, %3, %4, 32
+    FILTER_H4_w16o %1, %2, %3, %4, 48
+%endmacro
+
 ;-----------------------------------------------------------------------------
 ; void interp_4tap_horiz_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
@@ -525,6 +628,11 @@
 IPFILTER_CHROMA 8,  32
 IPFILTER_CHROMA 12, 16
 
+IPFILTER_CHROMA 6,  16
+IPFILTER_CHROMA 8,  12
+IPFILTER_CHROMA 8,  64
+IPFILTER_CHROMA 12, 32
+
 ;-----------------------------------------------------------------------------
 ; void interp_4tap_horiz_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
@@ -577,6 +685,18 @@
 IPFILTER_CHROMA_W 24, 32
 IPFILTER_CHROMA_W 32, 32
 
+IPFILTER_CHROMA_W 16, 24
+IPFILTER_CHROMA_W 16, 64
+IPFILTER_CHROMA_W 32, 48
+IPFILTER_CHROMA_W 24, 64
+IPFILTER_CHROMA_W 32, 64
+
+IPFILTER_CHROMA_W 64, 64
+IPFILTER_CHROMA_W 64, 32
+IPFILTER_CHROMA_W 64, 48
+IPFILTER_CHROMA_W 48, 64
+IPFILTER_CHROMA_W 64, 16
+
 
 %macro FILTER_H8_W8 7-8   ; t0, t1, t2, t3, coef, c512, src, dst
     movu        %1, %7
@@ -613,7 +733,7 @@
 ;----------------------------------------------------------------------------------------------------------------------------
 %macro IPFILTER_LUMA 3
 INIT_XMM sse4
-cglobal interp_8tap_horiz_%3_%1x%2, 4,7,6
+cglobal interp_8tap_horiz_%3_%1x%2, 4,7,8
 
     mov       r4d, r4m
 
@@ -987,7 +1107,7 @@
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W2_H4 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_pp_2x8, 4, 6, 8
+cglobal interp_4tap_vert_pp_2x%2, 4, 6, 8
 
 mov         r4d,       r4m
 sub         r0,        r1
@@ -1067,6 +1187,8 @@
 
 FILTER_V4_W2_H4 2, 8
 
+FILTER_V4_W2_H4 2, 16
+
 ;-----------------------------------------------------------------------------
 ; void interp_4tap_vert_pp_4x2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
@@ -1273,6 +1395,8 @@
 FILTER_V4_W4_H4 4,  8
 FILTER_V4_W4_H4 4, 16
 
+FILTER_V4_W4_H4 4, 32
+
 %macro FILTER_V4_W8_H2 0
 punpcklbw   m1,        m2
 punpcklbw   m7,        m3,        m0
@@ -1640,6 +1764,8 @@
 FILTER_V_PS_W4_H4 4, 8
 FILTER_V_PS_W4_H4 4, 16
 
+FILTER_V_PS_W4_H4 4, 32
+
 ;--------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_8x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;--------------------------------------------------------------------------------------------------------------
@@ -1708,6 +1834,9 @@
 FILTER_V_PS_W8_H8_H16_H2 8, 4
 FILTER_V_PS_W8_H8_H16_H2 8, 6
 
+FILTER_V_PS_W8_H8_H16_H2 8, 12
+FILTER_V_PS_W8_H8_H16_H2 8, 64
+
 ;--------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_8x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;--------------------------------------------------------------------------------------------------------------
@@ -1803,8 +1932,9 @@
 ;------------------------------------------------------------------------------------------------------------
 ;void interp_4tap_vert_ps_6x8(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;------------------------------------------------------------------------------------------------------------
+%macro FILTER_V_PS_W6 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_ps_6x8, 4, 6, 8
+cglobal interp_4tap_vert_ps_6x%2, 4, 6, 8
 
     mov        r4d, r4m
     sub        r0, r1
@@ -1821,7 +1951,7 @@
     pshufb     m5, [tab_Vm + 16]
     mova       m4, [pw_2000]
     lea        r5, [3 * r1]
-    mov        r4d, 2
+    mov        r4d, %2/4
 
 .loop:
     movq       m0, [r0]
@@ -1889,12 +2019,17 @@
     dec        r4d
     jnz        .loop
     RET
+%endmacro
+
+FILTER_V_PS_W6 6, 8
+FILTER_V_PS_W6 6, 16
 
 ;---------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_12x16(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;---------------------------------------------------------------------------------------------------------------
+%macro FILTER_V_PS_W12 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_ps_12x16, 4, 6, 8
+cglobal interp_4tap_vert_ps_12x%2, 4, 6, 8
 
     mov        r4d, r4m
     sub        r0, r1
@@ -1910,7 +2045,7 @@
     pshufb     m1, m0, [tab_Vm]
     pshufb     m0, [tab_Vm + 16]
 
-    mov        r4d, 16/2
+    mov        r4d, %2/2
 
 .loop:
     movu       m2, [r0]
@@ -1970,6 +2105,10 @@
     dec        r4d
     jnz        .loop
     RET
+%endmacro
+
+FILTER_V_PS_W12 12, 16
+FILTER_V_PS_W12 12, 32
 
 ;---------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_16x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
@@ -2059,11 +2198,15 @@
 FILTER_V_PS_W16 16, 16
 FILTER_V_PS_W16 16, 32
 
+FILTER_V_PS_W16 16, 24
+FILTER_V_PS_W16 16, 64
+
 ;--------------------------------------------------------------------------------------------------------------
 ;void interp_4tap_vert_ps_24x32(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;--------------------------------------------------------------------------------------------------------------
+%macro FILTER_V4_PS_W24 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_ps_24x32, 4, 6, 8
+cglobal interp_4tap_vert_ps_24x%2, 4, 6, 8
 
     mov        r4d, r4m
     sub        r0, r1
@@ -2079,7 +2222,7 @@
     pshufb     m1, m0, [tab_Vm]
     pshufb     m0, [tab_Vm + 16]
 
-    mov        r4d, 32/2
+    mov        r4d, %2/2
 
 .loop:
     movu       m2, [r0]
@@ -2170,6 +2313,11 @@
     dec        r4d
     jnz        .loop
     RET
+%endmacro
+
+FILTER_V4_PS_W24 24, 32
+
+FILTER_V4_PS_W24 24, 64
 
 ;---------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_32x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
@@ -2265,6 +2413,9 @@
 FILTER_V_PS_W32 32, 24
 FILTER_V_PS_W32 32, 32
 
+FILTER_V_PS_W32 32, 48
+FILTER_V_PS_W32 32, 64
+
 ;-----------------------------------------------------------------------------
 ; void interp_4tap_vert_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
@@ -2359,12 +2510,16 @@
 FILTER_V4_W8_H8_H16_H32 8, 16
 FILTER_V4_W8_H8_H16_H32 8, 32
 
+FILTER_V4_W8_H8_H16_H32 8, 12
+FILTER_V4_W8_H8_H16_H32 8, 64
+
+
 ;-----------------------------------------------------------------------------
 ;void interp_4tap_vert_pp_6x8(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W6_H4 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_pp_6x8, 4, 6, 8
+cglobal interp_4tap_vert_pp_6x%2, 4, 6, 8
 
 mov         r4d,       r4m
 sub         r0,        r1
@@ -2455,12 +2610,14 @@
 
 FILTER_V4_W6_H4 6, 8
 
+FILTER_V4_W6_H4 6, 16
+
 ;-----------------------------------------------------------------------------
 ; void interp_4tap_vert_pp_12x16(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W12_H2 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_pp_12x16, 4, 6, 8
+cglobal interp_4tap_vert_pp_12x%2, 4, 6, 8
 
 mov         r4d,       r4m
 sub         r0,        r1
@@ -2543,12 +2700,14 @@
 
 FILTER_V4_W12_H2 12, 16
 
+FILTER_V4_W12_H2 12, 32
+
 ;-----------------------------------------------------------------------------
-; void interp_4tap_vert_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+; void interp_4tap_vert_pp_16x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W16_H2 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_pp_%1x%2, 4, 6, 8
+cglobal interp_4tap_vert_pp_16x%2, 4, 6, 8
 
 mov         r4d,       r4m
 sub         r0,        r1
@@ -2563,7 +2722,7 @@
 pshufb      m1,        m0,       [tab_Vm]
 pshufb      m0,        [tab_Vm + 16]
 
-mov         r4d,       %2
+mov         r4d,       %2/2
 
 .loop:
 movu        m2,        [r0]
@@ -2622,7 +2781,7 @@
 
 lea         r2,        [r2 + 2 * r3]
 
-sub         r4,        2
+dec         r4d
 jnz        .loop
 RET
 %endmacro
@@ -2633,12 +2792,15 @@
 FILTER_V4_W16_H2 16, 16
 FILTER_V4_W16_H2 16, 32
 
+FILTER_V4_W16_H2 16, 24
+FILTER_V4_W16_H2 16, 64
+
 ;-----------------------------------------------------------------------------
 ;void interp_4tap_vert_pp_24x32(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W24 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_pp_24x32, 4, 6, 8
+cglobal interp_4tap_vert_pp_24x%2, 4, 6, 8
 
 mov         r4d,       r4m
 sub         r0,        r1
@@ -2754,8 +2916,10 @@
 
 FILTER_V4_W24 24, 32
 
+FILTER_V4_W24 24, 64
+
 ;-----------------------------------------------------------------------------
-; void interp_4tap_vert_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+; void interp_4tap_vert_pp_32x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------
 %macro FILTER_V4_W32 2
 INIT_XMM sse4
@@ -2849,6 +3013,111 @@
 FILTER_V4_W32 32, 24
 FILTER_V4_W32 32, 32
 
+FILTER_V4_W32 32, 48
+FILTER_V4_W32 32, 64
+
+
+;-----------------------------------------------------------------------------
+; void interp_4tap_vert_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+;-----------------------------------------------------------------------------
+%macro FILTER_V4_W16n_H2 2
+INIT_XMM sse4
+cglobal interp_4tap_vert_pp_%1x%2, 4, 7, 8
+
+mov         r4d,       r4m
+sub         r0,        r1
+
+%ifdef PIC
+lea         r5,        [tab_ChromaCoeff]
+movd        m0,        [r5 + r4 * 4]
+%else
+movd        m0,        [tab_ChromaCoeff + r4 * 4]
+%endif
+
+pshufb      m1,        m0,       [tab_Vm]
+pshufb      m0,        [tab_Vm + 16]
+
+mov         r4d,       %2/2
+
+.loop:
+
+mov         r6d,       %1/16
+
+.loopW:
+
+movu        m2,        [r0]
+movu        m3,        [r0 + r1]
+
+punpcklbw   m4,        m2,        m3
+punpckhbw   m2,        m3
+
+pmaddubsw   m4,        m1
+pmaddubsw   m2,        m1
+
+lea         r5,        [r0 + 2 * r1]
+movu        m5,        [r5]
+movu        m6,        [r5 + r1]
+
+punpckhbw   m7,        m5,        m6
+pmaddubsw   m7,        m0
+paddw       m2,        m7
+
+punpcklbw   m7,        m5,        m6
+pmaddubsw   m7,        m0
+paddw       m4,        m7
+
+mova        m7,        [tab_c_512]
+
+pmulhrsw    m4,        m7
+pmulhrsw    m2,        m7
+
+packuswb    m4,        m2
+
+movu        [r2],      m4
+
+punpcklbw   m4,        m3,        m5
+punpckhbw   m3,        m5
+
+pmaddubsw   m4,        m1
+pmaddubsw   m3,        m1
+
+movu        m5,        [r5 + 2 * r1]
+
+punpcklbw   m2,        m6,        m5
+punpckhbw   m6,        m5
+
+pmaddubsw   m2,        m0
+pmaddubsw   m6,        m0
+
+paddw       m4,        m2
+paddw       m3,        m6
+
+pmulhrsw    m4,        m7
+pmulhrsw    m3,        m7
+
+packuswb    m4,        m3
+
+movu        [r2 + r3],      m4
+
+add         r0,        16
+add         r2,        16
+dec         r6d
+jnz         .loopW
+
+lea         r0,        [r0 + r1 * 2 - %1]
+lea         r2,        [r2 + r3 * 2 - %1]
+
+dec         r4d
+jnz        .loop
+RET
+%endmacro
+
+FILTER_V4_W16n_H2 64, 64
+FILTER_V4_W16n_H2 64, 32
+FILTER_V4_W16n_H2 64, 48
+FILTER_V4_W16n_H2 48, 64
+FILTER_V4_W16n_H2 64, 16
+
 
 ;-----------------------------------------------------------------------------
 ; void filterConvertPelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)
@@ -3350,7 +3619,7 @@
 ;-------------------------------------------------------------------------------------------------------------
 %macro FILTER_VER_LUMA 3
 INIT_XMM sse4
-cglobal interp_8tap_vert_%3_%1x%2, 5, 7, 8 ,0-1
+cglobal interp_8tap_vert_%3_%1x%2, 5, 7, 8 ,0-gprsize
     lea       r5, [3 * r1]
     sub       r0, r5
     shl       r4d, 6
@@ -3370,7 +3639,7 @@
 %else
     mova      m3, [pw_2000]
 %endif
-    mov       byte [rsp], %2/4
+    mov       dword [rsp], %2/4
 
 .loopH:
     mov       r4d, (%1/8)
@@ -3420,7 +3689,7 @@
     lea       r2, [r2 + 4 * r3 - 2 * %1]
 %endif
 
-    dec       byte [rsp]
+    dec       dword [rsp]
     jnz       .loopH
 
     RET
@@ -3532,7 +3801,7 @@
 ;--------------------------------------------------------------------------------------------------------------
 %macro FILTER_VER_LUMA_SP 2
 INIT_XMM sse4
-cglobal interp_8tap_vert_sp_%1x%2, 5, 7, 8 ,0-1
+cglobal interp_8tap_vert_sp_%1x%2, 5, 7, 8 ,0-gprsize
 
     add       r1d, r1d
     lea       r5, [r1 + 2 * r1]
@@ -3548,7 +3817,7 @@
 
     mova      m7, [tab_c_526336]
 
-    mov       byte [rsp], %2/4
+    mov       dword [rsp], %2/4
 .loopH:
     mov       r4d, (%1/4)
 .loopW:
@@ -3585,7 +3854,7 @@
     lea       r0, [r0 + 4 * r1 - 2 * %1]
     lea       r2, [r2 + 4 * r3 - %1]
 
-    dec       byte [rsp]
+    dec       dword [rsp]
     jnz       .loopH
 
     RET
@@ -3720,7 +3989,7 @@
 ;--------------------------------------------------------------------------------------------------------------
 %macro FILTER_VER_CHROMA_SP 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_sp_%1x%2, 5, 7, 7 ,0-1
+cglobal interp_4tap_vert_sp_%1x%2, 5, 7, 7 ,0-gprsize
 
     add       r1d, r1d
     sub       r0, r1
@@ -3735,7 +4004,7 @@
 
     mova      m6, [tab_c_526336]
 
-    mov       byte [rsp], %2/4
+    mov       dword [rsp], %2/4
 
 .loopH:
     mov       r4d, (%1/4)
@@ -3773,7 +4042,7 @@
     lea       r0, [r0 + 4 * r1 - 2 * %1]
     lea       r2, [r2 + 4 * r3 - %1]
 
-    dec       byte [rsp]
+    dec       dword [rsp]
     jnz       .loopH
 
     RET
@@ -3794,6 +4063,20 @@
     FILTER_VER_CHROMA_SP 24, 32
     FILTER_VER_CHROMA_SP 32, 8
 
+    FILTER_VER_CHROMA_SP 16, 24
+    FILTER_VER_CHROMA_SP 16, 64
+    FILTER_VER_CHROMA_SP 12, 32
+    FILTER_VER_CHROMA_SP 4, 32
+    FILTER_VER_CHROMA_SP 32, 64
+    FILTER_VER_CHROMA_SP 32, 48
+    FILTER_VER_CHROMA_SP 24, 64
+
+    FILTER_VER_CHROMA_SP 64, 64
+    FILTER_VER_CHROMA_SP 64, 32
+    FILTER_VER_CHROMA_SP 64, 48
+    FILTER_VER_CHROMA_SP 48, 64
+    FILTER_VER_CHROMA_SP 64, 16
+
 
 %macro PROCESS_CHROMA_SP_W2_4R 1
     movd       m0, [r0]
@@ -3879,6 +4162,8 @@
 FILTER_VER_CHROMA_SP_W2_4R 2, 4
 FILTER_VER_CHROMA_SP_W2_4R 2, 8
 
+FILTER_VER_CHROMA_SP_W2_4R 2, 16
+
 ;--------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_sp_4x2(int16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;--------------------------------------------------------------------------------------------------------------
@@ -3931,10 +4216,11 @@
     RET
 
 ;-------------------------------------------------------------------------------------------------------------------
-; void interp_4tap_vertical_sp_6x8(int16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+; void interp_4tap_vertical_sp_6x%2(int16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-------------------------------------------------------------------------------------------------------------------
+%macro FILTER_VER_CHROMA_SP_W6_H4 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_sp_6x8, 5, 7, 7
+cglobal interp_4tap_vert_sp_6x%2, 5, 7, 7
 
     add       r1d, r1d
     sub       r0, r1
@@ -3949,7 +4235,7 @@
 
     mova      m6, [tab_c_526336]
 
-    mov       r4d, 8/4
+    mov       r4d, %2/4
 
 .loopH:
     PROCESS_CHROMA_SP_W4_4R
@@ -4003,6 +4289,11 @@
     jnz       .loopH
 
     RET
+%endmacro
+
+FILTER_VER_CHROMA_SP_W6_H4 6, 8
+
+FILTER_VER_CHROMA_SP_W6_H4 6, 16
 
 %macro PROCESS_CHROMA_SP_W8_2R 0
     movu       m1, [r0]
@@ -4093,6 +4384,10 @@
 FILTER_VER_CHROMA_SP_W8_H2 8, 16
 FILTER_VER_CHROMA_SP_W8_H2 8, 32
 
+FILTER_VER_CHROMA_SP_W8_H2 8, 12
+FILTER_VER_CHROMA_SP_W8_H2 8, 64
+
+
 ;-----------------------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_horiz_ps_2x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx, int isRowExt)
 ;-----------------------------------------------------------------------------------------------------------------------------
@@ -4145,6 +4440,8 @@
 FILTER_HORIZ_CHROMA_2xN 2, 4
 FILTER_HORIZ_CHROMA_2xN 2, 8
 
+FILTER_HORIZ_CHROMA_2xN 2, 16
+
 ;-----------------------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_horiz_ps_4x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx, int isRowExt)
 ;-----------------------------------------------------------------------------------------------------------------------------
@@ -4198,6 +4495,8 @@
 FILTER_HORIZ_CHROMA_4xN 4, 8
 FILTER_HORIZ_CHROMA_4xN 4, 16
 
+FILTER_HORIZ_CHROMA_4xN 4, 32
+
 %macro PROCESS_CHROMA_W6 3
     movu       %1, [srcq]
     pshufb     %2, %1, Tm0
@@ -4277,6 +4576,9 @@
 FILTER_HORIZ_CHROMA 6, 8
 FILTER_HORIZ_CHROMA 12, 16
 
+FILTER_HORIZ_CHROMA 6, 16
+FILTER_HORIZ_CHROMA 12, 32
+
 %macro PROCESS_CHROMA_W8 3
     movu        %1, [srcq]
     pshufb      %2, %1, Tm0
@@ -4341,6 +4643,9 @@
 FILTER_HORIZ_CHROMA_8xN 8, 16
 FILTER_HORIZ_CHROMA_8xN 8, 32
 
+FILTER_HORIZ_CHROMA_8xN 8, 12
+FILTER_HORIZ_CHROMA_8xN 8, 64
+
 %macro PROCESS_CHROMA_W16 4
     movu        %1, [srcq]
     pshufb      %2, %1, Tm0
@@ -4422,6 +4727,38 @@
     movu        [dstq + 48], %4
 %endmacro
 
+%macro PROCESS_CHROMA_W16o 5
+    movu        %1, [srcq + %5]
+    pshufb      %2, %1, Tm0
+    pmaddubsw   %2, coef2
+    pshufb      %1, %1, Tm1
+    pmaddubsw   %1, coef2
+    phaddw      %2, %1
+    movu        %1, [srcq + %5 + 8]
+    pshufb      %4, %1, Tm0
+    pmaddubsw   %4, coef2
+    pshufb      %1, %1, Tm1
+    pmaddubsw   %1, coef2
+    phaddw      %4, %1
+    psubw       %2, %3
+    psubw       %4, %3
+    movu        [dstq + %5 * 2], %2
+    movu        [dstq + %5 * 2 + 16], %4
+%endmacro
+
+%macro PROCESS_CHROMA_W48 4
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 0
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 16
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 32
+%endmacro
+
+%macro PROCESS_CHROMA_W64 4
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 0
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 16
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 32
+    PROCESS_CHROMA_W16o %1, %2, %3, %4, 48
+%endmacro
+
 ;------------------------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_horiz_ps_%1x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx, int isRowExt)
 ;------------------------------------------------------------------------------------------------------------------------------
@@ -4480,6 +4817,119 @@
 FILTER_HORIZ_CHROMA_WxN 32, 24
 FILTER_HORIZ_CHROMA_WxN 32, 32
 
+FILTER_HORIZ_CHROMA_WxN 16, 24
+FILTER_HORIZ_CHROMA_WxN 16, 64
+FILTER_HORIZ_CHROMA_WxN 24, 64
+FILTER_HORIZ_CHROMA_WxN 32, 48
+FILTER_HORIZ_CHROMA_WxN 32, 64
+
+FILTER_HORIZ_CHROMA_WxN 64, 64
+FILTER_HORIZ_CHROMA_WxN 64, 32
+FILTER_HORIZ_CHROMA_WxN 64, 48
+FILTER_HORIZ_CHROMA_WxN 48, 64
+FILTER_HORIZ_CHROMA_WxN 64, 16
+
+
+;---------------------------------------------------------------------------------------------------------------
+; void interp_4tap_vert_ps_%1x%2(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
+;---------------------------------------------------------------------------------------------------------------
+%macro FILTER_V_PS_W16n 2
+INIT_XMM sse4
+cglobal interp_4tap_vert_ps_%1x%2, 4, 7, 8
+
+    mov        r4d, r4m
+    sub        r0, r1
+    add        r3d, r3d
+
+%ifdef PIC
+    lea        r5, [tab_ChromaCoeff]
+    movd       m0, [r5 + r4 * 4]
+%else
+    movd       m0, [tab_ChromaCoeff + r4 * 4]
+%endif
+
+    pshufb     m1, m0, [tab_Vm]
+    pshufb     m0, [tab_Vm + 16]
+    mov        r4d, %2/2
+
+.loop:
+
+    mov         r6d,       %1/16
+
+.loopW:
+
+    movu       m2, [r0]
+    movu       m3, [r0 + r1]
+
+    punpcklbw  m4, m2, m3
+    punpckhbw  m2, m3
+
+    pmaddubsw  m4, m1
+    pmaddubsw  m2, m1
+
+    lea        r5, [r0 + 2 * r1]
+    movu       m5, [r5]
+    movu       m7, [r5 + r1]
+
+    punpcklbw  m6, m5, m7
+    pmaddubsw  m6, m0
+    paddw      m4, m6
+
+    punpckhbw  m6, m5, m7
+    pmaddubsw  m6, m0
+    paddw      m2, m6
+
+    mova       m6, [pw_2000]
+
+    psubw      m4, m6
+    psubw      m2, m6
+
+    movu       [r2], m4
+    movu       [r2 + 16], m2
+
+    punpcklbw  m4, m3, m5
+    punpckhbw  m3, m5
+
+    pmaddubsw  m4, m1
+    pmaddubsw  m3, m1
+
+    movu       m5, [r5 + 2 * r1]
+
+    punpcklbw  m2, m7, m5
+    punpckhbw  m7, m5
+
+    pmaddubsw  m2, m0
+    pmaddubsw  m7, m0
+
+    paddw      m4, m2
+    paddw      m3, m7
+
+    psubw      m4, m6
+    psubw      m3, m6
+
+    movu       [r2 + r3], m4
+    movu       [r2 + r3 + 16], m3
+
+    add         r0,        16
+    add         r2,        32
+    dec         r6d
+    jnz         .loopW
+
+    lea         r0,        [r0 + r1 * 2 - %1]
+    lea         r2,        [r2 + r3 * 2 - %1 * 2]
+
+    dec        r4d
+    jnz        .loop
+    RET
+%endmacro
+
+FILTER_V_PS_W16n 64, 64
+FILTER_V_PS_W16n 64, 32
+FILTER_V_PS_W16n 64, 48
+FILTER_V_PS_W16n 48, 64
+FILTER_V_PS_W16n 64, 16
+
+
 ;------------------------------------------------------------------------------------------------------------
 ;void interp_4tap_vert_ps_2x4(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;------------------------------------------------------------------------------------------------------------
@@ -4556,8 +5006,9 @@
 ;-------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ps_2x8(pixel *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;-------------------------------------------------------------------------------------------------------------
+%macro FILTER_V_PS_W2 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_ps_2x8, 4, 6, 8
+cglobal interp_4tap_vert_ps_2x%2, 4, 6, 8
 
     mov        r4d, r4m
     sub        r0, r1
@@ -4574,7 +5025,7 @@
 
     mova       m1, [pw_2000]
     lea        r5, [3 * r1]
-    mov        r4d, 2
+    mov        r4d, %2/4
 .loop:
     movd       m2, [r0]
     movd       m3, [r0 + r1]
@@ -4635,13 +5086,18 @@
     jnz        .loop
 
 RET
+%endmacro
+
+FILTER_V_PS_W2 2, 8
+
+FILTER_V_PS_W2 2, 16
 
 ;-----------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ss_%1x%2(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------------------------------------------
 %macro FILTER_VER_CHROMA_SS 2
 INIT_XMM sse2
-cglobal interp_4tap_vert_ss_%1x%2, 5, 7, 6 ,0-1
+cglobal interp_4tap_vert_ss_%1x%2, 5, 7, 6 ,0-gprsize
 
     add       r1d, r1d
     add       r3d, r3d
@@ -4655,7 +5111,7 @@
     lea       r6, [tab_ChromaCoeffV + r4]
 %endif
 
-    mov       byte [rsp], %2/4
+    mov       dword [rsp], %2/4
 
 .loopH:
     mov       r4d, (%1/4)
@@ -4686,7 +5142,7 @@
     lea       r0, [r0 + 4 * r1 - 2 * %1]
     lea       r2, [r2 + 4 * r3 - 2 * %1]
 
-    dec       byte [rsp]
+    dec       dword [rsp]
     jnz       .loopH
 
     RET
@@ -4707,6 +5163,21 @@
     FILTER_VER_CHROMA_SS 24, 32
     FILTER_VER_CHROMA_SS 32, 8
 
+    FILTER_VER_CHROMA_SS 16, 24
+    FILTER_VER_CHROMA_SS 12, 32
+    FILTER_VER_CHROMA_SS 4, 32
+    FILTER_VER_CHROMA_SS 32, 64
+    FILTER_VER_CHROMA_SS 16, 64
+    FILTER_VER_CHROMA_SS 32, 48
+    FILTER_VER_CHROMA_SS 24, 64
+
+    FILTER_VER_CHROMA_SS 64, 64
+    FILTER_VER_CHROMA_SS 64, 32
+    FILTER_VER_CHROMA_SS 64, 48
+    FILTER_VER_CHROMA_SS 48, 64
+    FILTER_VER_CHROMA_SS 64, 16
+
+
 ;---------------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vertical_ss_%1x%2(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;---------------------------------------------------------------------------------------------------------------------
@@ -4753,6 +5224,8 @@
 FILTER_VER_CHROMA_SS_W2_4R 2, 4
 FILTER_VER_CHROMA_SS_W2_4R 2, 8
 
+FILTER_VER_CHROMA_SS_W2_4R 2, 16
+
 ;---------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ss_4x2(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;---------------------------------------------------------------------------------------------------------------
@@ -4803,8 +5276,9 @@
 ;-------------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vertical_ss_6x8(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;-------------------------------------------------------------------------------------------------------------------
+%macro FILTER_VER_CHROMA_SS_W6_H4 2
 INIT_XMM sse4
-cglobal interp_4tap_vert_ss_6x8, 5, 7, 6
+cglobal interp_4tap_vert_ss_6x%2, 5, 7, 6
 
     add       r1d, r1d
     add       r3d, r3d
@@ -4818,7 +5292,7 @@
     lea       r6, [tab_ChromaCoeffV + r4]
 %endif
 
-    mov       r4d, 8/4
+    mov       r4d, %2/4
 
 .loopH:
     PROCESS_CHROMA_SP_W4_4R
@@ -4861,6 +5335,12 @@
     jnz       .loopH
 
     RET
+%endmacro
+
+FILTER_VER_CHROMA_SS_W6_H4 6, 8
+
+FILTER_VER_CHROMA_SS_W6_H4 6, 16
+
 
 ;----------------------------------------------------------------------------------------------------------------
 ; void interp_4tap_vert_ss_8x%2(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
@@ -4911,12 +5391,15 @@
 FILTER_VER_CHROMA_SS_W8_H2 8, 16
 FILTER_VER_CHROMA_SS_W8_H2 8, 32
 
+FILTER_VER_CHROMA_SS_W8_H2 8, 12
+FILTER_VER_CHROMA_SS_W8_H2 8, 64
+
 ;-----------------------------------------------------------------------------------------------------------------
 ; void interp_8tap_vert_ss_%1x%2(int16_t *src, intptr_t srcStride, int16_t *dst, intptr_t dstStride, int coeffIdx)
 ;-----------------------------------------------------------------------------------------------------------------
 %macro FILTER_VER_LUMA_SS 2
 INIT_XMM sse2
-cglobal interp_8tap_vert_ss_%1x%2, 5, 7, 7 ,0-1
+cglobal interp_8tap_vert_ss_%1x%2, 5, 7, 7 ,0-gprsize
 
     add        r1d, r1d
     add        r3d, r3d
@@ -4931,7 +5414,7 @@
     lea        r6, [tab_LumaCoeffV + r4]
 %endif
 
-    mov        byte [rsp], %2/4
+    mov        dword [rsp], %2/4
 .loopH:
     mov        r4d, (%1/4)
 .loopW:
@@ -5023,7 +5506,7 @@
     lea        r0, [r0 + 4 * r1 - 2 * %1]
     lea        r2, [r2 + 4 * r3 - 2 * %1]
 
-    dec        byte [rsp]
+    dec        dword [rsp]
     jnz        .loopH
 
     RET

x265-hg20140324.tar.bz2/source/common/x86/ipfilter8.h -> x265-hg20140928.tar.bz2/source/common/x86/ipfilter8.h Changed

x265-hg20140324.tar.bz2/source/common/x86/loopfilter.asm -> x265-hg20140928.tar.bz2/source/common/x86/loopfilter.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/loopfilter.h -> x265-hg20140928.tar.bz2/source/common/x86/loopfilter.h Changed

x265-hg20140324.tar.bz2/source/common/x86/mc-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/mc-a.asm Changed

@@ -4,7 +4,7 @@
 ;* Copyright (C) 2003-2013 x264 project
 ;*
 ;* Authors: Loren Merritt <lorenm@u.washington.edu>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
+;*          Fiona Glaser <fiona@x264.com>
 ;*          Laurent Aimar <fenrir@via.ecp.fr>
 ;*          Dylan Yudaken <dyudaken@gmail.com>
 ;*          Holger Lubitz <holger@lubitz.org>
@@ -26,7 +26,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@x264.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************
 
 %include "x86inc.asm"
@@ -154,6 +154,50 @@
 %endrep
     RET
 
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal addAvg_2x16, 6,7,8, pSrc0, pSrc1, pDst, iStride0, iStride1, iDstStride
+    mova        m6,         [pw_1023]
+    mova        m7,         [pw_1024]
+    mov         r6d,        16/4
+    add         r3,         r3
+    add         r4,         r4
+    add         r5,         r5
+.loop:
+    movd        m1,         [r0]
+    movd        m2,         [r0 + r3]
+    movd        m3,         [r1]
+    movd        m4,         [r1 + r4]
+    lea         r0,         [r0 + r3 * 2]
+    lea         r1,         [r1 + r4 * 2]
+    punpckldq   m1,         m2
+    punpckldq   m3,         m4
+    movd        m2,         [r0]
+    movd        m4,         [r0 + r3]
+    movd        m5,         [r1]
+    movd        m0,         [r1 + r4]
+    lea         r0,         [r0 + r3 * 2]
+    lea         r1,         [r1 + r4 * 2]
+    punpckldq   m2,         m4
+    punpckldq   m5,         m0
+    punpcklqdq  m1,         m2
+    punpcklqdq  m3,         m5
+    paddw       m1,         m3
+    pmulhrsw    m1,         m7
+    paddw       m1,         [pw_512]
+    pxor        m0,         m0
+    pmaxsw      m1,         m0
+    pminsw      m1,         m6
+    movd        [r2],       m1
+    pextrd      [r2 + r5],  m1, 1
+    lea         r2,         [r2 + r5 * 2]
+    pextrd      [r2],       m1, 2
+    pextrd      [r2 + r5],  m1, 3
+    lea         r2,         [r2 + r5 * 2]
+    dec         r6d
+    jnz         .loop
+    RET
+;-----------------------------------------------------------------------------
 
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
@@ -181,7 +225,7 @@
     RET
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
-cglobal addAvg_6x8, 6,7,8, pSrc0, pSrc1, pDst, iStride0, iStride1, iDstStride
+cglobal addAvg_6x8, 6,6,8, pSrc0, pSrc1, pDst, iStride0, iStride1, iDstStride
     mova        m4,             [pw_512]
     mova        m5,             [pw_1023]
     mova        m7,             [pw_1024]
@@ -220,6 +264,42 @@
     RET
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
+cglobal addAvg_6x16, 6,7,8, pSrc0, pSrc1, pDst, iStride0, iStride1, iDstStride
+    mova        m4,             [pw_512]
+    mova        m5,             [pw_1023]
+    mova        m7,             [pw_1024]
+    pxor        m6,             m6
+    mov         r6d,            16/2
+    add         r3,             r3
+    add         r4,             r4
+    add         r5,             r5
+.loop:
+    movu        m0,             [r0]
+    movu        m2,             [r1]
+    movu        m1,             [r0 + r3]
+    movu        m3,             [r1 + r4]
+    dec         r6d
+    lea         r0,             [r0 + r3 * 2]
+    lea         r1,             [r1 + r4 * 2]
+    paddw       m0,             m2
+    paddw       m1,             m3
+    pmulhrsw    m0,             m7
+    pmulhrsw    m1,             m7
+    paddw       m0,             m4
+    paddw       m1,             m4
+    pmaxsw      m0,             m6
+    pmaxsw      m1,             m6
+    pminsw      m0,             m5
+    pminsw      m1,             m5
+    movh        [r2],           m0
+    pextrd      [r2 + 8],       m0, 2
+    movh        [r2 + r5],      m1
+    pextrd      [r2 + r5 + 8],  m1, 2
+    lea         r2,             [r2 + r5 * 2]
+    jnz         .loop
+    RET
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
 cglobal addAvg_8x2, 6,6,8, pSrc0, pSrc1, pDst, iStride0, iStride1, iDstStride
     mova        m4,          [pw_512]
     mova        m5,          [pw_1023]
@@ -335,6 +415,8 @@
 ADDAVG_W4_H4 8
 ADDAVG_W4_H4 16
 
+ADDAVG_W4_H4 32
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W8_H4 1
 INIT_XMM sse4
@@ -382,6 +464,9 @@
 ADDAVG_W8_H4 16
 ADDAVG_W8_H4 32
 
+ADDAVG_W8_H4 12
+ADDAVG_W8_H4 64
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W12_H4 1
 INIT_XMM sse4
@@ -442,6 +527,8 @@
 
 ADDAVG_W12_H4 16
 
+ADDAVG_W12_H4 32
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W16_H4 1
 INIT_XMM sse4
@@ -509,6 +596,8 @@
 ADDAVG_W16_H4 32
 ADDAVG_W16_H4 64
 
+ADDAVG_W16_H4 24
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W24_H2 2
 INIT_XMM sse4
@@ -589,6 +678,8 @@
 
 ADDAVG_W24_H2 24, 32
 
+ADDAVG_W24_H2 24, 64
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W32_H2 1
 INIT_XMM sse4
@@ -691,6 +782,8 @@
 ADDAVG_W32_H2 32
 ADDAVG_W32_H2 64
 
+ADDAVG_W32_H2 48
+
 ;-----------------------------------------------------------------------------
 %macro ADDAVG_W48_H2 1
 INIT_XMM sse4
@@ -1052,6 +1145,48 @@
 
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
+cglobal addAvg_2x16, 6,7,8, src0, src1, dst, src0Stride, src1tride, dstStride
+    mova        m0,         [pw_256]
+    mova        m7,         [pw_128]
+    mov         r6d,        16/4
+    add         r3,         r3
+    add         r4,         r4
+.loop:
+    movd        m1,         [r0]
+    movd        m2,         [r0 + r3]
+    movd        m3,         [r1]
+    movd        m4,         [r1 + r4]
+    lea         r0,         [r0 + r3 * 2]
+    lea         r1,         [r1 + r4 * 2]
+    punpckldq   m1,         m2
+    punpckldq   m3,         m4
+    movd        m2,         [r0]
+    movd        m4,         [r0 + r3]
+    movd        m5,         [r1]
+    movd        m6,         [r1 + r4]
+    lea         r0,         [r0 + r3 * 2]
+    lea         r1,         [r1 + r4 * 2]
+    punpckldq   m2,         m4
+    punpckldq   m5,         m6
+    punpcklqdq  m1,         m2
+    punpcklqdq  m3,         m5
+    paddw       m1,         m3
+    pmulhrsw    m1,         m0
+    paddw       m1,         m7
+    packuswb    m1,         m1
+    pextrw      [r2],       m1, 0
+    pextrw      [r2 + r5],  m1, 1
+    lea         r2,         [r2 + r5 * 2]
+    pextrw      [r2],       m1, 2
+    pextrw      [r2 + r5],  m1, 3
+    lea         r2,         [r2 + r5 * 2]
+    dec         r6d
+    jnz         .loop
+    RET
+;-----------------------------------------------------------------------------
+
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
 cglobal addAvg_4x2, 6,6,4, pSrc0, src0, src1, dst, src0Stride, src1tride, dstStride
 
     mova           m1,          [pw_256]
@@ -1132,6 +1267,9 @@
 ADDAVG_W4_H4 4
 ADDAVG_W4_H4 8
 ADDAVG_W4_H4 16
+
+ADDAVG_W4_H4 32
+
 ;-----------------------------------------------------------------------------
 
 ;-----------------------------------------------------------------------------
@@ -1232,6 +1370,39 @@
 
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
+cglobal addAvg_6x16, 6,7,6, pSrc0, src0, src1, dst, src0Stride, src1tride, dstStride
+    mova        m4,             [pw_256]
+    mova        m5,             [pw_128]
+    mov         r6d,            16/2
+    add         r3,             r3
+    add         r4,             r4
+.loop:
+    movu        m0,             [r0]
+    movu        m2,             [r1]
+    movu        m1,             [r0 + r3]
+    movu        m3,             [r1 + r4]
+    dec         r6d
+    lea         r0,             [r0 + r3 * 2]
+    lea         r1,             [r1 + r4 * 2]
+    paddw       m0,             m2
+    paddw       m1,             m3
+    pmulhrsw    m0,             m4
+    pmulhrsw    m1,             m4
+    paddw       m0,             m5
+    paddw       m1,             m5
+    packuswb    m0,             m0
+    packuswb    m1,             m1
+    movd        [r2],           m0
+    pextrw      [r2 + 4],       m0, 2
+    movd        [r2 + r5],      m1
+    pextrw      [r2 + r5 + 4],  m1, 2
+    lea         r2,             [r2 + r5 * 2]
+    jnz         .loop
+    RET
+;-----------------------------------------------------------------------------
+
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
 cglobal addAvg_8x2, 6,6,6, pSrc0, src0, src1, dst, src0Stride, src1tride, dstStride
     mova        m4,          [pw_256]
     mova        m5,          [pw_128]
@@ -1392,6 +1563,9 @@
 ADDAVG_W8_H4 16
 ADDAVG_W8_H4 32
 
+ADDAVG_W8_H4 12
+ADDAVG_W8_H4 64
+
 ;-----------------------------------------------------------------------------
 
 
@@ -1485,6 +1659,8 @@
 
 ADDAVG_W12_H4 16
 
+ADDAVG_W12_H4 32
+
 ;-----------------------------------------------------------------------------
 
 
@@ -1580,6 +1756,8 @@
 ADDAVG_W16_H4 32
 ADDAVG_W16_H4 64
 
+ADDAVG_W16_H4 24
+
 ;-----------------------------------------------------------------------------
 
 
@@ -1654,6 +1832,8 @@
 
 ADDAVG_W24_H2 24, 32
 
+ADDAVG_W24_H2 24, 64
+
 ;-----------------------------------------------------------------------------
 
 ;-----------------------------------------------------------------------------
@@ -1743,6 +1923,8 @@
 ADDAVG_W32_H2 32
 ADDAVG_W32_H2 64
 
+ADDAVG_W32_H2 48
+
 ;-----------------------------------------------------------------------------
 
 
@@ -3320,10 +3502,10 @@
     mov   eax, r2m
     and   eax, 0x3f
     cmp   eax, 0x30
-    jb x264_pixel_avg2_w16_sse2
+    jb x265_pixel_avg2_w16_sse2
     or    eax, r4m
     and   eax, 7
-    jz x264_pixel_avg2_w16_sse2
+    jz x265_pixel_avg2_w16_sse2
 %endif
     PROLOGUE 6, 8
     lea    r6, [r4+r2]

x265-hg20140324.tar.bz2/source/common/x86/mc-a2.asm -> x265-hg20140928.tar.bz2/source/common/x86/mc-a2.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/mc.h -> x265-hg20140928.tar.bz2/source/common/x86/mc.h Changed

x265-hg20140324.tar.bz2/source/common/x86/pixel-32.asm -> x265-hg20140928.tar.bz2/source/common/x86/pixel-32.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/pixel-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/pixel-a.asm Changed

@@ -7,7 +7,7 @@
 ;*          Holger Lubitz <holger@lubitz.org>
 ;*          Laurent Aimar <fenrir@via.ecp.fr>
 ;*          Alex Izvorski <aizvorksi@gmail.com>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
+;*          Fiona Glaser <fiona@x264.com>
 ;*          Oskar Arvidsson <oskar@irock.se>
 ;*
 ;* This program is free software; you can redistribute it and/or modify
@@ -25,7 +25,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@x264.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************
 
 %include "x86inc.asm"
@@ -590,7 +590,7 @@
 %macro SATDS_SSE2 0
 %define vertical ((notcpuflag(ssse3) || cpuflag(atom)) || HIGH_BIT_DEPTH)
 
-%if vertical==0 || HIGH_BIT_DEPTH
+%if cpuflag(ssse3) && (vertical==0 || HIGH_BIT_DEPTH)
 cglobal pixel_satd_4x4, 4, 6, 6
     SATD_START_MMX
     mova m4, [hmul_4p]
@@ -666,7 +666,7 @@
     SATD_8x4_1_SSE 0, 4, 8, 5, 9, 6, 3, 10, 12, 13
     ret
 
-cglobal pixel_satd_16x4, 4,6,12
+cglobal pixel_satd_16x4, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -680,14 +680,14 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_16x8, 4,6,12
+cglobal pixel_satd_16x8, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
 %endif
     jmp %%pixel_satd_16x8_internal
 
-cglobal pixel_satd_16x12, 4,6,12
+cglobal pixel_satd_16x12, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -695,7 +695,7 @@
     call pixel_satd_16x4_internal2
     jmp %%pixel_satd_16x8_internal
 
-cglobal pixel_satd_16x32, 4,6,12
+cglobal pixel_satd_16x32, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -708,7 +708,7 @@
     call pixel_satd_16x4_internal2
     jmp %%pixel_satd_16x8_internal
 
-cglobal pixel_satd_16x64, 4,6,12
+cglobal pixel_satd_16x64, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -729,7 +729,7 @@
     call pixel_satd_16x4_internal2
     jmp %%pixel_satd_16x8_internal
 
-cglobal pixel_satd_16x16, 4,6,12
+cglobal pixel_satd_16x16, 4,6,14
     SATD_START_SSE2 m10, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -747,7 +747,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_32x8, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_32x8, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -768,7 +768,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_32x16, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_32x16, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -793,7 +793,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_32x24, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_32x24, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -822,7 +822,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_32x32, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_32x32, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -855,7 +855,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_32x64, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_32x64, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -904,7 +904,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_48x64, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_48x64, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -971,7 +971,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_64x16, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_64x16, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -1008,7 +1008,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_64x32, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_64x32, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -1062,7 +1062,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_64x48, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_64x48, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -1132,7 +1132,7 @@
     movd    eax, m10
     RET
 
-cglobal pixel_satd_64x64, 4,8,11    ;if WIN64 && notcpuflag(avx)
+cglobal pixel_satd_64x64, 4,8,14    ;if WIN64 && notcpuflag(avx)
     SATD_START_SSE2 m10, m7
     mov r6, r0
     mov r7, r2
@@ -1221,7 +1221,7 @@
 %else
 
 %if WIN64
-cglobal pixel_satd_32x8, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_32x8, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1270,7 +1270,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_32x16, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_32x16, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1327,7 +1327,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_32x24, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_32x24, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1392,7 +1392,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_32x32, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_32x32, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1465,7 +1465,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_32x64, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_32x64, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1570,7 +1570,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_48x64, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_48x64, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1718,7 +1718,7 @@
 
 
 %if WIN64
-cglobal pixel_satd_64x16, 4,8,8    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_64x16, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1811,7 +1811,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_64x32, 4,8,9    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_64x32, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -1936,7 +1936,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_64x48, 4,8,9    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_64x48, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -2093,7 +2093,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_64x64, 4,8,9    ;if WIN64 && cpuflag(avx)
+cglobal pixel_satd_64x64, 4,8,14    ;if WIN64 && cpuflag(avx)
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -2281,7 +2281,11 @@
     RET
 %endif
 
+%if WIN64
+cglobal pixel_satd_16x4, 4,6,14
+%else
 cglobal pixel_satd_16x4, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7
     BACKUP_POINTERS
     call %%pixel_satd_8x4_internal2
@@ -2295,7 +2299,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_16x8, 4,6,14
+%else
 cglobal pixel_satd_16x8, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7
     BACKUP_POINTERS
     call pixel_satd_8x8_internal2
@@ -2309,7 +2317,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_16x12, 4,6,14
+%else
 cglobal pixel_satd_16x12, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7, 1
     BACKUP_POINTERS
     call pixel_satd_8x8_internal2
@@ -2325,7 +2337,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_16x16, 4,6,14
+%else
 cglobal pixel_satd_16x16, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7, 1
     BACKUP_POINTERS
     call pixel_satd_8x8_internal2
@@ -2341,7 +2357,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_16x32, 4,6,14
+%else
 cglobal pixel_satd_16x32, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7, 1
     BACKUP_POINTERS
     call pixel_satd_8x8_internal2
@@ -2361,7 +2381,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_16x64, 4,6,14
+%else
 cglobal pixel_satd_16x64, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7, 1
     BACKUP_POINTERS
     call pixel_satd_8x8_internal2
@@ -2513,7 +2537,7 @@
 %endif
 
 %if WIN64
-cglobal pixel_satd_24x32, 4,8,8
+cglobal pixel_satd_24x32, 4,8,14
     SATD_START_SSE2 m6, m7
     mov r6, r0
     mov r7, r2
@@ -2572,7 +2596,11 @@
     RET
 %endif    ;WIN64
 
+%if WIN64
+cglobal pixel_satd_8x32, 4,6,14
+%else
 cglobal pixel_satd_8x32, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7
 %if vertical
     mova m7, [pw_00ff]
@@ -2589,7 +2617,11 @@
     movd   eax, m6
     RET
 
+%if WIN64
+cglobal pixel_satd_8x16, 4,6,14
+%else
 cglobal pixel_satd_8x16, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7
     call pixel_satd_8x8_internal2
     call pixel_satd_8x8_internal2
@@ -2606,7 +2638,11 @@
     call pixel_satd_8x8_internal
     SATD_END_SSE2 m6
 
+%if WIN64
+cglobal pixel_satd_8x4, 4,6,14
+%else
 cglobal pixel_satd_8x4, 4,6,8
+%endif
     SATD_START_SSE2 m6, m7
     call %%pixel_satd_8x4_internal2
     SATD_END_SSE2 m6
@@ -2759,7 +2795,7 @@
     shr  eax, 1
     RET
 
-cglobal pixel_sa8d_8x16, 4,8,12
+cglobal pixel_sa8d_8x16, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2774,7 +2810,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_8x32, 4,8,12
+cglobal pixel_sa8d_8x32, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2795,7 +2831,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_16x8, 4,8,12
+cglobal pixel_sa8d_16x8, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2810,7 +2846,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_16x32, 4,8,12
+cglobal pixel_sa8d_16x32, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2825,7 +2861,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_16x64, 4,8,12
+cglobal pixel_sa8d_16x64, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2846,7 +2882,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_24x32, 4,8,12
+cglobal pixel_sa8d_24x32, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2891,7 +2927,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_32x8, 4,8,12
+cglobal pixel_sa8d_32x8, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2912,7 +2948,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_32x16, 4,8,12
+cglobal pixel_sa8d_32x16, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2923,8 +2959,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -2933,7 +2969,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_32x24, 4,8,12
+cglobal pixel_sa8d_32x24, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2978,7 +3014,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_32x32, 4,8,12
+cglobal pixel_sa8d_32x32, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -2989,8 +3025,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3001,8 +3037,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3011,7 +3047,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_32x64, 4,8,12
+cglobal pixel_sa8d_32x64, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3022,8 +3058,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3034,8 +3070,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3046,8 +3082,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3058,8 +3094,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3068,7 +3104,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_48x64, 4,8,12
+cglobal pixel_sa8d_48x64, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3079,8 +3115,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3088,8 +3124,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3100,8 +3136,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3109,8 +3145,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3121,8 +3157,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3130,8 +3166,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3142,8 +3178,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3151,8 +3187,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3161,7 +3197,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_64x16, 4,8,12
+cglobal pixel_sa8d_64x16, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3172,8 +3208,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3181,8 +3217,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3190,8 +3226,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3200,7 +3236,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_64x32, 4,8,12
+cglobal pixel_sa8d_64x32, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3211,8 +3247,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3220,8 +3256,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3229,8 +3265,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3241,8 +3277,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3250,8 +3286,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3259,8 +3295,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3269,7 +3305,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_64x48, 4,8,12
+cglobal pixel_sa8d_64x48, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3280,8 +3316,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3289,8 +3325,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3298,8 +3334,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3310,8 +3346,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3319,8 +3355,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3328,8 +3364,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3340,8 +3376,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3349,8 +3385,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3358,8 +3394,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3368,7 +3404,7 @@
     movd eax, m12
     RET
 
-cglobal pixel_sa8d_64x64, 4,8,12
+cglobal pixel_sa8d_64x64, 4,8,13
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
     lea  r5, [3*r3]
@@ -3379,8 +3415,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3388,8 +3424,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3397,8 +3433,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3409,8 +3445,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3418,8 +3454,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3427,8 +3463,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3439,8 +3475,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3448,8 +3484,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3457,8 +3493,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     add  r2, 16*SIZEOF_PIXEL
     add  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3469,8 +3505,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3478,8 +3514,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -3487,8 +3523,8 @@
     SA8D_16x16
     lea  r4, [8*r1]
     lea  r5, [8*r3]
-    sub  r2, r4
-    sub  r0, r5
+    sub  r0, r4
+    sub  r2, r5
     sub  r2, 16*SIZEOF_PIXEL
     sub  r0, 16*SIZEOF_PIXEL
     lea  r4, [3*r1]
@@ -6356,3 +6392,190 @@
     RET
 %endif ; HIGH_BIT_DEPTH
 
+; Input 16bpp, Output 8bpp
+;------------------------------------------------------------------------------------------------------------------------
+;void planecopy_sp(uint16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask)
+;------------------------------------------------------------------------------------------------------------------------
+INIT_XMM sse2
+cglobal downShift_16, 7,7,3
+    movd        m0, r6d        ; m0 = shift
+    add         r1, r1
+    dec         r5d
+.loopH:
+    xor         r6, r6
+.loopW:
+    movu        m1, [r0 + r6 * 2]
+    movu        m2, [r0 + r6 * 2 + 16]
+    psrlw       m1, m0
+    psrlw       m2, m0
+    packuswb    m1, m2
+    movu        [r2 + r6], m1
+
+    add         r6, 16
+    cmp         r6d, r4d
+    jl          .loopW
+
+    ; move to next row
+    add         r0, r1
+    add         r2, r3
+    dec         r5d
+    jnz         .loopH
+
+;processing last row of every frame [To handle width which not a multiple of 16]
+
+.loop16:
+    movu        m1, [r0]
+    movu        m2, [r0 + 16]
+    psrlw       m1, m0
+    psrlw       m2, m0
+    packuswb    m1, m2
+    movu        [r2], m1
+
+    add         r0, 2 * mmsize
+    add         r2, mmsize
+    sub         r4d, 16
+    jz          .end
+    cmp         r4d, 15
+    jg          .loop16
+
+    cmp         r4d, 8
+    jl          .process4
+    movu        m1, [r0]
+    psrlw       m1, m0
+    packuswb    m1, m1
+    movh        [r2], m1
+
+    add         r0, mmsize
+    add         r2, 8
+    sub         r4d, 8
+    jz          .end
+
+.process4:
+    cmp         r4d, 4
+    jl          .process2
+    movh        m1,[r0]
+    psrlw       m1, m0
+    packuswb    m1, m1
+    movd        [r2], m1
+
+    add         r0, 8
+    add         r2, 4
+    sub         r4d, 4
+    jz          .end
+
+.process2:
+    cmp         r4d, 2
+    jl          .process1
+    movd        m1, [r0]
+    psrlw       m1, m0
+    packuswb    m1, m1
+    movd        r6, m1
+    mov         [r2], r6w
+
+    add         r0, 4
+    add         r2, 2
+    sub         r4d, 2
+    jz          .end
+
+.process1:
+    movd        m1, [r0]
+    psrlw       m1, m0
+    packuswb    m1, m1
+    movd        r3, m1
+    mov         [r2], r3b
+.end:
+    RET
+
+; Input 8bpp, Output 16bpp
+;---------------------------------------------------------------------------------------------------------------------
+;void planecopy_cp(uint8_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift)
+;---------------------------------------------------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal upShift_8, 7,7,3
+
+    movd        m2, r6d        ; m0 = shift
+    add         r3, r3
+    dec         r5d
+
+.loopH:
+    xor         r6, r6
+.loopW:
+    pmovzxbw    m0,[r0 + r6]
+    pmovzxbw    m1,[r0 + r6 + 8]
+    psllw       m0, m2
+    psllw       m1, m2
+    movu        [r2 + r6 * 2], m0
+    movu        [r2 + r6 * 2 + 16], m1
+
+    add         r6, 16
+    cmp         r6d, r4d
+    jl          .loopW
+
+    ; move to next row
+    add         r0, r1
+    add         r2, r3
+    dec         r5d
+    jnz         .loopH
+
+;processing last row of every frame [To handle width which not a multiple of 16]
+
+.loop16:
+    pmovzxbw    m0,[r0]
+    pmovzxbw    m1,[r0 + 8]
+    psllw       m0, m2
+    psllw       m1, m2
+    movu        [r2], m0
+    movu        [r2 + 16], m1
+
+    add         r0, mmsize
+    add         r2, 2 * mmsize
+    sub         r4d, 16
+    jz          .end
+    cmp         r4d, 15
+    jg          .loop16
+
+    cmp         r4d, 8
+    jl          .process4
+    pmovzxbw    m0,[r0]
+    psllw       m0, m2
+    movu        [r2], m0
+
+    add         r0, 8
+    add         r2, mmsize
+    sub         r4d, 8
+    jz          .end
+
+.process4:
+    cmp         r4d, 4
+    jl          .process2
+    movd        m0,[r0]
+    pmovzxbw    m0,m0
+    psllw       m0, m2
+    movh        [r2], m0
+
+    add         r0, 4
+    add         r2, 8
+    sub         r4d, 4
+    jz          .end
+
+.process2:
+    cmp         r4d, 2
+    jl          .process1
+    movzx       r3d, byte [r0]
+    shl         r3d, 2
+    mov         [r2], r3w
+    movzx       r3d, byte [r0 + 1]
+    shl         r3d, 2
+    mov         [r2 + 2], r3w
+
+    add         r0, 2
+    add         r2, 4
+    sub         r4d, 2
+    jz          .end
+
+.process1:
+    movzx       r3d, byte [r0]
+    shl         r3d, 2
+    mov         [r2], r3w
+.end:
+    RET

x265-hg20140324.tar.bz2/source/common/x86/pixel-util.h -> x265-hg20140928.tar.bz2/source/common/x86/pixel-util.h Changed

@@ -18,18 +18,18 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_PIXEL_UTIL_H
 #define X265_PIXEL_UTIL_H
 
-void x265_calcRecons4_sse2(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons8_sse2(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons16_sse2(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons32_sse2(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons16_sse4(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
-void x265_calcRecons32_sse4(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons4_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons8_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons16_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons32_sse2(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons16_sse4(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
+void x265_calcRecons32_sse4(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred);
 
 void x265_getResidual4_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
 void x265_getResidual8_sse2(pixel *fenc, pixel *pred, int16_t *residual, intptr_t stride);
@@ -44,9 +44,13 @@
 void x265_transpose32_sse2(pixel *dest, pixel *src, intptr_t stride);
 void x265_transpose64_sse2(pixel *dest, pixel *src, intptr_t stride);
 
-uint32_t x265_quant_sse4(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int32_t *qCoef, int qBits, int add, int numCoeff, int32_t* lastPos);
-void x265_dequant_normal_sse4(const int32_t* quantCoef, int32_t* coef, int num, int scale, int shift);
-int x265_count_nonzero_sse2(const int32_t *quantCoeff, int numCoeff);
+uint32_t x265_quant_sse4(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_quant_avx2(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_nquant_sse4(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
+uint32_t x265_nquant_avx2(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
+void x265_dequant_normal_sse4(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift);
+void x265_dequant_normal_avx2(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift);
+int x265_count_nonzero_ssse3(const int16_t *quantCoeff, int numCoeff);
 
 void x265_weight_pp_sse4(pixel *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
 void x265_weight_sp_sse4(int16_t *src, pixel *dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset);
@@ -69,66 +73,34 @@
 
 #define CHROMA_PIXELSUB_DEF(cpu) \
     SETUP_CHROMA_PIXELSUB_PS_FUNC(4, 4, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(4, 2, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(2, 4, cpu); \
     SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 8, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 4, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(4, 8, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 6, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(6, 8, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 2, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(2, 8, cpu); \
     SETUP_CHROMA_PIXELSUB_PS_FUNC(16, 16, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(16, 8, cpu); \
+    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 32, cpu);
+
+#define CHROMA_PIXELSUB_DEF_422(cpu) \
+    SETUP_CHROMA_PIXELSUB_PS_FUNC(4, 8, cpu); \
     SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 16, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(16, 12, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(12, 16, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(16, 4, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(4, 16, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 32, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 16, cpu); \
     SETUP_CHROMA_PIXELSUB_PS_FUNC(16, 32, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 24, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(24, 32, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 8, cpu); \
-    SETUP_CHROMA_PIXELSUB_PS_FUNC(8, 32, cpu);
+    SETUP_CHROMA_PIXELSUB_PS_FUNC(32, 64, cpu);
 
 #define SETUP_LUMA_PIXELSUB_PS_FUNC(W, H, cpu) \
     void x265_pixel_sub_ps_ ## W ## x ## H ## cpu(int16_t * dest, intptr_t destride, pixel * src0, pixel * src1, intptr_t srcstride0, intptr_t srcstride1); \
     void x265_pixel_add_ps_ ## W ## x ## H ## cpu(pixel * dest, intptr_t destride, pixel * src0, int16_t * scr1, intptr_t srcStride0, intptr_t srcStride1);
 
 #define LUMA_PIXELSUB_DEF(cpu) \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(4,   4, cpu); \
     SETUP_LUMA_PIXELSUB_PS_FUNC(8,   8, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(8,   4, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(4,   8, cpu); \
     SETUP_LUMA_PIXELSUB_PS_FUNC(16, 16, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(16,  8, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(8,  16, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(16, 12, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(12, 16, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(16,  4, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(4,  16, cpu); \
     SETUP_LUMA_PIXELSUB_PS_FUNC(32, 32, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(32, 16, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(16, 32, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(32, 24, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(24, 32, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(32,  8, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(8,  32, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(64, 64, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(64, 32, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(32, 64, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(64, 48, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(48, 64, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(64, 16, cpu); \
-    SETUP_LUMA_PIXELSUB_PS_FUNC(16, 64, cpu);
+    SETUP_LUMA_PIXELSUB_PS_FUNC(64, 64, cpu);
 
 CHROMA_PIXELSUB_DEF(_sse4);
 LUMA_PIXELSUB_DEF(_sse4);
 CHROMA_PIXELSUB_DEF(_sse2);
 LUMA_PIXELSUB_DEF(_sse2);
 
+CHROMA_PIXELSUB_DEF_422(_sse4);
+CHROMA_PIXELSUB_DEF_422(_sse2);
+
 #define SETUP_LUMA_PIXELVAR_FUNC(W, H, cpu) \
     uint64_t x265_pixel_var_ ## W ## x ## H ## cpu(pixel * pix, intptr_t pixstride);
 
@@ -141,6 +113,7 @@
 LUMA_PIXELVAR_DEF(_sse2);
 
 #undef CHROMA_PIXELSUB_DEF
+#undef CHROMA_PIXELSUB_DEF_422
 #undef LUMA_PIXELSUB_DEF
 #undef LUMA_PIXELVAR_DEF
 #undef SETUP_CHROMA_PIXELSUB_PS_FUNC

x265-hg20140324.tar.bz2/source/common/x86/pixel-util8.asm -> x265-hg20140928.tar.bz2/source/common/x86/pixel-util8.asm Changed

@@ -19,7 +19,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@multicorewareinc.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************/
 
 %include "x86inc.asm"
@@ -27,8 +27,6 @@
 
 SECTION_RODATA 32
 
-c_d_4:             dd 4, 4, 4, 4
-c_d_1234:          dd 1, 2, 3, 4
 %if BIT_DEPTH == 10
 ssim_c1:   times 4 dd 6697.7856    ; .01*.01*1023*1023*64
 ssim_c2:   times 4 dd 3797644.4352 ; .03*.03*1023*1023*64*63
@@ -56,668 +54,450 @@
 cextern pw_00ff
 cextern pw_2000
 cextern pw_pixel_max
+cextern pd_1
+cextern pd_32767
+cextern pd_n32768
 
-%define NEW_CALCRECON 1 ; TODO: remove recon[] arg
 ;-----------------------------------------------------------------------------
-; void calcrecon(pixel* pred, int16_t* residual, pixel* recon, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred)
+; void calcrecon(pixel* pred, int16_t* residual, int16_t* reconqt, pixel *reconipred, int stride, int strideqt, int strideipred)
 ;-----------------------------------------------------------------------------
 INIT_XMM sse2
-cglobal calcRecons4
 %if HIGH_BIT_DEPTH
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,6
+cglobal calcRecons4, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,6
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons4, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
-
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-    add         t7, t7
-%endif
+    mov         r6d, r6m
+    add         r4d, r4d
+    add         r5d, r5d
+    add         r6d, r6d
 
     pxor        m4, m4
     mova        m5, [pw_pixel_max]
-    add         t5, t5
-    mov         t8d, 4/2
+    mov         t7b, 4/2
 .loop:
-    movh        m0, [t0]
-    movh        m1, [t0 + t5]
+    movh        m0, [r0]
+    movh        m1, [r0 + r4]
     punpcklqdq  m0, m1
-    movh        m2, [t1]
-    movh        m3, [t1 + t5]
+    movh        m2, [r1]
+    movh        m3, [r1 + r4]
     punpcklqdq  m2, m3
     paddw       m0, m2
     CLIPW       m0, m4, m5
 
-    ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movh        [t2], m0
-%endif
-    movh        [t4], m0
-%if ARCH_X86_64 == 0
-    add         t4, t7
-    add         t4, t7
-    movhps      [t4], m0
-    add         t4, t7
-    add         t4, t7
-%else
-    movhps      [t4 + t7], m0
-    lea         t4, [t4 + t7 * 2]
-%endif
-%if NEW_CALCRECON == 0
-    movhps      [t2 + t5], m0
-%endif
+    ; store recipred[]
+    movh        [r3], m0
+    movhps      [r3 + r6], m0
 
     ; store recqt[]
-    movh        [t3], m0
-    add         t3, t6
-    movhps      [t3], m0
-    add         t3, t6
-
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    movh        [r2], m0
+    movhps      [r2 + r5], m0
 
-    dec         t8d
-    jnz        .loop
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 2]
+    lea         r2, [r2 + r5 * 2]
+    lea         r3, [r3 + r6 * 2]
 
+    dec         t7b
+    jnz        .loop
+    RET
 %else          ;HIGH_BIT_DEPTH
+
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,4
+cglobal calcRecons4, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,4
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons4, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
-
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-%endif
+    mov         r6d, r6m
+    add         r5d, r5d
 
     pxor        m0, m0
-    mov         t8d, 4/2
+    mov         t7b, 4/2
 .loop:
-    movd        m1, [t0]
-    movd        m2, [t0 + t5]
+    movd        m1, [r0]
+    movd        m2, [r0 + r4]
     punpckldq   m1, m2
     punpcklbw   m1, m0
-    movh        m2, [t1]
-    movh        m3, [t1 + t5 * 2]
+    movh        m2, [r1]
+    movh        m3, [r1 + r4 * 2]
     punpcklqdq  m2, m3
     paddw       m1, m2
     packuswb    m1, m1
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movd        [t2], m1
-%endif
-    movd        [t4], m1
-    add         t4, t7
+    movd        [r3], m1
     pshufd      m2, m1, 1
-%if NEW_CALCRECON == 0
-    movd        [t2 + t5], m2
-%endif
-    movd        [t4], m2
-    add         t4, t7
+    movd        [r3 + r6], m2
 
     ; store recqt[]
     punpcklbw   m1, m0
-    movlps      [t3], m1
-    add         t3, t6
-    movhps      [t3], m1
-    add         t3, t6
-
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 4]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    movh        [r2], m1
+    movhps      [r2 + r5], m1
+
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 4]
+    lea         r2, [r2 + r5 * 2]
+    lea         r3, [r3 + r6 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
-%endif          ;HIGH_BIT_DEPTH
     RET
+%endif          ;HIGH_BIT_DEPTH
 
 
 INIT_XMM sse2
-cglobal calcRecons8
-%if HIGH_BIT_DEPTH
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,6
+cglobal calcRecons8, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,6
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons8, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
 
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+%if HIGH_BIT_DEPTH
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-    add         t7, t7
-%endif
+    mov         r6d, r6m
+    add         r4d, r4d
+    add         r5d, r5d
+    add         r6d, r6d
 
     pxor        m4, m4
     mova        m5, [pw_pixel_max]
-    add         t5, t5
-    mov         t8d, 8/2
+    mov         t7b, 8/2
 .loop:
-    movu        m0, [t0]
-    movu        m1, [t0 + t5]
-    movu        m2, [t1]
-    movu        m3, [t1 + t5]
+    movu        m0, [r0]
+    movu        m1, [r0 + r4]
+    movu        m2, [r1]
+    movu        m3, [r1 + r4]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
-    ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2], m0
-    movu        [t2 + t5], m1
-%endif
-    movu        [t4], m0
-%if ARCH_X86_64 == 0
-    add         t4, t7
-    add         t4, t7
-    movu        [t4], m1
-    add         t4, t7
-    add         t4, t7
-%else
-    movu        [t4 + t7], m1
-    lea         t4, [t4 + t7 * 2]
-%endif
+    ; store recipred[]
+    movu        [r3], m0
+    movu        [r3 + r6], m1
 
     ; store recqt[]
-    movu        [t3], m0
-    add         t3, t6
-    movu        [t3], m1
-    add         t3, t6
-
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    movu        [r2], m0
+    movu        [r2 + r5], m1
+
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 2]
+    lea         r2, [r2 + r5 * 2]
+    lea         r3, [r3 + r6 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
+    RET
 %else          ;HIGH_BIT_DEPTH
 
-%if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,5
-%else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,5
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
-%endif
-
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-%endif
+    mov         r6d, r6m
+    add         r5d, r5d
 
     pxor        m0, m0
-    mov         t8d, 8/2
+    mov         t7b, 8/2
 .loop:
-    movh        m1, [t0]
-    movh        m2, [t0 + t5]
+    movh        m1, [r0]
+    movh        m2, [r0 + r4]
     punpcklbw   m1, m0
     punpcklbw   m2, m0
-    movu        m3, [t1]
-    movu        m4, [t1 + t5 * 2]
+    movu        m3, [r1]
+    movu        m4, [r1 + r4 * 2]
     paddw       m1, m3
     paddw       m2, m4
     packuswb    m1, m2
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movlps      [t2], m1
-    movhps      [t2 + t5], m1
-%endif
-    movlps      [t4], m1
-%if ARCH_X86_64 == 0
-    add         t4, t7
-    movhps      [t4], m1
-    add         t4, t7
-%else
-    movhps      [t4 + t7], m1
-    lea         t4, [t4 + t7 * 2]
-%endif
+    movh        [r3], m1
+    movhps      [r3 + r6], m1
 
     ; store recqt[]
     punpcklbw   m2, m1, m0
     punpckhbw   m1, m0
-    movu        [t3], m2
-    add         t3, t6
-    movu        [t3], m1
-    add         t3, t6
-
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 4]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    movu        [r2], m2
+    movu        [r2 + r5], m1
 
-    dec         t8d
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 4]
+    lea         r2, [r2 + r5 * 2]
+    lea         r3, [r3 + r6 * 2]
+
+    dec         t7b
     jnz        .loop
-%endif          ;HIGH_BIT_DEPTH
     RET
+%endif          ;HIGH_BIT_DEPTH
 
 
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal calcRecons16
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,6
+cglobal calcRecons16, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,6
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons16, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
 
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-    add         t7, t7
-%endif
+    mov         r6d, r6m
+    add         r4d, r4d
+    add         r5d, r5d
+    add         r6d, r6d
 
     pxor        m4, m4
     mova        m5, [pw_pixel_max]
-    add         t5, t5
-    mov         t8d, 16/2
+    mov         t7b, 16/2
 .loop:
-    movu        m0, [t0]
-    movu        m1, [t0 + 16]
-    movu        m2, [t1]
-    movu        m3, [t1 + 16]
+    movu        m0, [r0]
+    movu        m1, [r0 + 16]
+    movu        m2, [r1]
+    movu        m3, [r1 + 16]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
-    ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2], m0
-    movu        [t2 + 16], m1
-%endif
-    movu        [t4], m0
-    movu        [t4 + 16], m1
-%if ARCH_X86_64 == 0
-    add         t4, t7
-    add         t4, t7
-%endif
+    ; store recipred[]
+    movu        [r3], m0
+    movu        [r3 + 16], m1
 
     ; store recqt[]
-    movu        [t3], m0
-    movu        [t3 + 16], m1
-    add         t3, t6
-
-    movu        m0, [t0 + t5]
-    movu        m1, [t0 + t5 + 16]
-    movu        m2, [t1 + t5]
-    movu        m3, [t1 + t5 + 16]
+    movu        [r2], m0
+    movu        [r2 + 16], m1
+
+    movu        m0, [r0 + r4]
+    movu        m1, [r0 + r4 + 16]
+    movu        m2, [r1 + r4]
+    movu        m3, [r1 + r4 + 16]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2 + t5], m0
-    movu        [t2 + t5 + 16], m1
-%endif
-%if ARCH_X86_64 == 0
-    movu        [t4], m0
-    movu        [t4 + 16], m1
-    add         t4, t7
-    add         t4, t7
-%else
-    movu        [t4 + t7], m0
-    movu        [t4 + t7 + 16], m1
-    lea         t4, [t4 + t7 * 2]
-%endif
+    movu        [r3 + r6], m0
+    movu        [r3 + r6 + 16], m1
 
     ; store recqt[]
-    movu        [t3], m0
-    movu        [t3 + 16], m1
-    add         t3, t6
+    movu        [r2 + r5], m0
+    movu        [r2 + r5 + 16], m1
 
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 2]
+    lea         r2, [r2 + r5 * 2]
+    lea         r3, [r3 + r6 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
+    RET
 %else          ;HIGH_BIT_DEPTH
+
 INIT_XMM sse4
-cglobal calcRecons16
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,3
+cglobal calcRecons16, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,3
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons16, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
 
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-%endif
+    mov         r6d, r6m
+    add         r5d, r5d
 
     pxor        m0, m0
-    mov         t8d, 16
+    mov         t7b, 16
 .loop:
-    movu        m2, [t0]
+    movu        m2, [r0]
     pmovzxbw    m1, m2
     punpckhbw   m2, m0
-    paddw       m1, [t1]
-    paddw       m2, [t1 + 16]
+    paddw       m1, [r1]
+    paddw       m2, [r1 + 16]
     packuswb    m1, m2
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2], m1
-%endif
-    movu        [t4], m1
+    movu        [r3], m1
 
     ; store recqt[]
     pmovzxbw    m2, m1
     punpckhbw   m1, m0
-    movu        [t3], m2
-    movu        [t3 + 16], m1
+    movu        [r2], m2
+    movu        [r2 + 16], m1
 
-    add         t3, t6
-    add         t4, t7
-    add         t0, t5
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    add         t2, t5
-%endif
+    add         r2, r5
+    add         r3, r6
+    add         r0, r4
+    lea         r1, [r1 + r4 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
-%endif          ;HIGH_BIT_DEPTH
     RET
+%endif          ;HIGH_BIT_DEPTH
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal calcRecons32
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,6
+cglobal calcRecons32, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,6
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons32, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
 
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-    add         t7, t7
-%endif
+    mov         r6d, r6m
+    add         r4d, r4d
+    add         r5d, r5d
+    add         r6d, r6d
 
     pxor        m4, m4
     mova        m5, [pw_pixel_max]
-    add         t5, t5
-    mov         t8d, 32/2
+    mov         t7b, 32/2
 .loop:
 
-    movu        m0, [t0]
-    movu        m1, [t0 + 16]
-    movu        m2, [t1]
-    movu        m3, [t1 + 16]
+    movu        m0, [r0]
+    movu        m1, [r0 + 16]
+    movu        m2, [r1]
+    movu        m3, [r1 + 16]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
-    ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2], m0
-    movu        [t2 + 16], m1
-%endif
-    movu        [t4], m0
-    movu        [t4 + 16], m1
+    ; store recipred[]
+    movu        [r3], m0
+    movu        [r3 + 16], m1
 
     ; store recqt[]
-    movu        [t3], m0
-    movu        [t3 + 16], m1
+    movu        [r2], m0
+    movu        [r2 + 16], m1
 
-    movu        m0, [t0 + 32]
-    movu        m1, [t0 + 48]
-    movu        m2, [t1 + 32]
-    movu        m3, [t1 + 48]
+    movu        m0, [r0 + 32]
+    movu        m1, [r0 + 48]
+    movu        m2, [r1 + 32]
+    movu        m3, [r1 + 48]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2 + 32], m0
-    movu        [t2 + 48], m1
-%endif
-    movu        [t4 + 32], m0
-    movu        [t4 + 48], m1
-%if ARCH_X86_64 == 0
-    add         t4, t7
-    add         t4, t7
-%endif
+    movu        [r3 + 32], m0
+    movu        [r3 + 48], m1
 
     ; store recqt[]
-    movu        [t3 + 32], m0
-    movu        [t3 + 48], m1
-    add         t3, t6
-
-    movu        m0, [t0 + t5]
-    movu        m1, [t0 + t5 + 16]
-    movu        m2, [t1 + t5]
-    movu        m3, [t1 + t5 + 16]
+    movu        [r2 + 32], m0
+    movu        [r2 + 48], m1
+    add         r2, r5
+
+    movu        m0, [r0 + r4]
+    movu        m1, [r0 + r4 + 16]
+    movu        m2, [r1 + r4]
+    movu        m3, [r1 + r4 + 16]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2 + t5], m0
-    movu        [t2 + t5 + 16], m1
-%endif
-%if ARCH_X86_64 == 0
-    movu        [t4], m0
-    movu        [t4 + 16], m1
-%else
-    movu        [t4 + t7], m0
-    movu        [t4 + t7 + 16], m1
-%endif
+    movu        [r3 + r6], m0
+    movu        [r3 + r6 + 16], m1
 
     ; store recqt[]
-    movu        [t3], m0
-    movu        [t3 + 16], m1
+    movu        [r2], m0
+    movu        [r2 + 16], m1
 
-    movu        m0, [t0 + t5 + 32]
-    movu        m1, [t0 + t5 + 48]
-    movu        m2, [t1 + t5 + 32]
-    movu        m3, [t1 + t5 + 48]
+    movu        m0, [r0 + r4 + 32]
+    movu        m1, [r0 + r4 + 48]
+    movu        m2, [r1 + r4 + 32]
+    movu        m3, [r1 + r4 + 48]
     paddw       m0, m2
     paddw       m1, m3
-    CLIPW       m0, m4, m5
-    CLIPW       m1, m4, m5
+    CLIPW2      m0, m1, m4, m5
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2 + t5 + 32], m0
-    movu        [t2 + t5 + 48], m1
-%endif
-%if ARCH_X86_64 == 0
-    movu        [t4 + 32], m0
-    movu        [t4 + 48], m1
-    add         t4, t7
-    add         t4, t7
-%else
-    movu        [t4 + t7 + 32], m0
-    movu        [t4 + t7 + 48], m1
-    lea         t4, [t4 + t7 * 2]
-%endif
+    movu        [r3 + r6 + 32], m0
+    movu        [r3 + r6 + 48], m1
+    lea         r3, [r3 + r6 * 2]
 
     ; store recqt[]
-    movu        [t3 + 32], m0
-    movu        [t3 + 48], m1
-    add         t3, t6
+    movu        [r2 + 32], m0
+    movu        [r2 + 48], m1
+    add         r2, r5
 
-    lea         t0, [t0 + t5 * 2]
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    lea         t2, [t2 + t5 * 2]
-%endif
+    lea         r0, [r0 + r4 * 2]
+    lea         r1, [r1 + r4 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
+    RET
 %else          ;HIGH_BIT_DEPTH
 INIT_XMM sse4
-cglobal calcRecons32
 %if ARCH_X86_64 == 1
-    DECLARE_REG_TMP 0,1,2,3,4,5,6,7,8
-    PROLOGUE 6,9,5
+cglobal calcRecons32, 5,8,4
+    %define t7b     r7b
 %else
-    DECLARE_REG_TMP 0,1,2,3,4,5
-    PROLOGUE 6,7,5
-    %define t6      r6m
-    %define t6d     r6d
-    %define t7      r7m
-    %define t8d     r6d
+cglobal calcRecons32, 5,7,4,0-1
+    %define t7b     byte [rsp]
 %endif
 
-    mov         t6d, r6m
-%if ARCH_X86_64 == 0
-    add         t6d, t6d
-    mov         r6m, t6d
-%else
+    mov         r4d, r4m
     mov         r5d, r5m
-    mov         r7d, r7m
-    add         t6d, t6d
-%endif
+    mov         r6d, r6m
+    add         r5d, r5d
 
     pxor        m0, m0
-    mov         t8d, 32
+    mov         t7b, 32
 .loop:
-    movu        m2, [t0]
-    movu        m4, [t0 + 16]
+    movu        m2, [r0]
+    movu        m4, [r0 + 16]
     pmovzxbw    m1, m2
     punpckhbw   m2, m0
     pmovzxbw    m3, m4
     punpckhbw   m4, m0
 
-    paddw       m1, [t1 + 0 * 16]
-    paddw       m2, [t1 + 1 * 16]
+    paddw       m1, [r1 + 0 * 16]
+    paddw       m2, [r1 + 1 * 16]
     packuswb    m1, m2
 
-    paddw       m3, [t1 + 2 * 16]
-    paddw       m4, [t1 + 3 * 16]
+    paddw       m3, [r1 + 2 * 16]
+    paddw       m4, [r1 + 3 * 16]
     packuswb    m3, m4
 
     ; store recon[] and recipred[]
-%if NEW_CALCRECON == 0
-    movu        [t2], m1
-    movu        [t2 + 16], m3
-%endif
-    movu        [t4], m1
-    movu        [t4 + 16], m3
+    movu        [r3], m1
+    movu        [r3 + 16], m3
 
     ; store recqt[]
     pmovzxbw    m2, m1
     punpckhbw   m1, m0
-    movu        [t3 + 0 * 16], m2
-    movu        [t3 + 1 * 16], m1
+    movu        [r2 + 0 * 16], m2
+    movu        [r2 + 1 * 16], m1
     pmovzxbw    m4, m3
     punpckhbw   m3, m0
-    movu        [t3 + 2 * 16], m4
-    movu        [t3 + 3 * 16], m3
+    movu        [r2 + 2 * 16], m4
+    movu        [r2 + 3 * 16], m3
 
-    add         t3, t6
-    add         t4, t7
-    add         t0, t5
-    lea         t1, [t1 + t5 * 2]
-%if NEW_CALCRECON == 0
-    add         t2, t5
-%endif
+    add         r2, r5
+    add         r3, r6
+    add         r0, r4
+    lea         r1, [r1 + r4 * 2]
 
-    dec         t8d
+    dec         t7b
     jnz        .loop
-%endif          ;HIGH_BIT_DEPTH
     RET
+%endif          ;HIGH_BIT_DEPTH
 
 
 ;-----------------------------------------------------------------------------
@@ -768,7 +548,7 @@
     punpckldq   m3, m4
     punpcklbw   m3, m0
     psubw       m1, m3
-    movlps      [r2], m1
+    movh        [r2], m1
     movhps      [r2 + r3 * 2], m1
     lea         r0, [r0 + r3 * 2]
     lea         r1, [r1 + r3 * 2]
@@ -784,7 +564,7 @@
     punpckldq   m3, m4
     punpcklbw   m3, m0
     psubw       m1, m3
-    movlps      [r2], m1
+    movh        [r2], m1
     movhps      [r2 + r3 * 2], m1
 %endif
     RET
@@ -1078,149 +858,350 @@
 
 
 ;-----------------------------------------------------------------------------
-; uint32_t quant(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int32_t *qCoef, int qBits, int add, int numCoeff, int32_t* lastPos);
+; uint32_t quant(int32_t *coef, int32_t *quantCoeff, int32_t *deltaU, int16_t *qCoef, int qBits, int add, int numCoeff);
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
-%if ARCH_X86_64 == 1
-cglobal quant, 5,6,11
-  %define addVec    m8
-  %define qbits     m9
-  %define qbits8    m10
-%else
-cglobal quant, 5,6,8, 0-(3*mmsize)
-  %define addVec    [rsp + 0 * mmsize]
-  %define qbits     [rsp + 1 * mmsize]
-  %define qbits8    [rsp + 2 * mmsize]
-%endif
-
-    ; fill qbits-8
-    movd        m0, r4d
-    mova        qbits, m0
+cglobal quant, 5,6,8
+    ; fill qbits
+    movd        m4, r4d         ; m4 = qbits
 
     ; fill qbits-8
     sub         r4d, 8
-    movd        m0, r4d
-    mova        qbits8, m0
+    movd        m6, r4d         ; m6 = qbits8
 
     ; fill offset
-    mov         r4d, r5m
-    movd        m0, r4d
-    pshufd      m0, m0, 0
-    mova        addVec, m0
+    movd        m5, r5m
+    pshufd      m5, m5, 0       ; m5 = add
+
+    lea         r5, [pd_1]
 
     mov         r4d, r6m
     shr         r4d, 3
-    pxor        m7, m7          ; m7 = acSum4
-    mova        m6, [c_d_1234]  ; m6 = last4
-    pxor        m5, m5          ; m5 = count
-    mova        m4, [c_d_4]     ; m4 = [4 4 4 4]
+    pxor        m7, m7          ; m7 = numZero
 .loop:
     ; 4 coeff
     movu        m0, [r0]        ; m0 = level
-    pxor        m1, m1
-    pcmpgtd     m1, m0          ; m1 = sign
-    movu        m2, [r1]        ; m2 = qcoeff
-    pabsd       m0, m0
-    pmulld      m0, m2          ; m0 = tmpLevel1
-    paddd       m2, m0, addVec
-    psrad       m2, qbits       ; m2 = level1
+    pabsd       m1, m0
+    pmulld      m1, [r1]        ; m0 = tmpLevel1
+    paddd       m2, m1, m5
+    psrad       m2, m4          ; m2 = level1
+
+    pslld       m3, m2, 8
+    psrad       m1, m6
+    psubd       m1, m3          ; m1 = deltaU1
+
+    movu        [r2], m1
+    psignd      m3, m2, m0
+    pminud      m2, [r5]
     paddd       m7, m2
-    pslld       m3, m2, qbits
-    psubd       m0, m3
-    psrad       m0, qbits8      ; m0 = deltaU1
-    movu        [r2], m0
-    pxor        m0, m0
-    pcmpeqd     m0, m2          ; m0 = mask4
-    pand        m5, m0
-    pandn       m0, m6
-    por         m5, m0
-    paddd       m6, m4
-
-    pxor        m2, m1
-    psubd       m2, m1
-    packssdw    m2, m2
-    pmovsxwd    m2, m2
-    movu        [r3], m2
+    packssdw    m3, m3
+    movh        [r3], m3
+
     ; 4 coeff
     movu        m0, [r0 + 16]   ; m0 = level
-    pxor        m1, m1
-    pcmpgtd     m1, m0          ; m1 = sign
-    movu        m2, [r1 + 16]   ; m2 = qcoeff
-    pabsd       m0, m0
-    pmulld      m0, m2          ; m0 = tmpLevel1
-    paddd       m2, m0, addVec
-    psrad       m2, qbits       ; m2 = level1
+    pabsd       m1, m0
+    pmulld      m1, [r1 + 16]   ; m0 = tmpLevel1
+    paddd       m2, m1, m5
+    psrad       m2, m4          ; m2 = level1
+    pslld       m3, m2, 8
+    psrad       m1, m6
+    psubd       m1, m3          ; m1 = deltaU1
+    movu        [r2 + 16], m1
+    psignd      m3, m2, m0
+    pminud      m2, [r5]
     paddd       m7, m2
-    pslld       m3, m2, qbits
-    psubd       m0, m3
-    psrad       m0, qbits8      ; m0 = deltaU1
-    movu        [r2 + 16], m0
-    pxor        m0, m0
-    pcmpeqd     m0, m2          ; m0 = mask4
-    pand        m5, m0
-    pandn       m0, m6
-    por         m5, m0
-    paddd       m6, m4
-
-    pxor        m2, m1
-    psubd       m2, m1
-    packssdw    m2, m2
-    pmovsxwd    m2, m2
-    movu        [r3 + 16], m2
+    packssdw    m3, m3
+    movh        [r3 + 8], m3
 
     add         r0, 32
     add         r1, 32
     add         r2, 32
-    add         r3, 32
+    add         r3, 16
 
     dec         r4d
     jnz        .loop
 
-    movhlps     m4, m5
-    pmaxud      m4, m5
-    pshufd      m5, m4, 1
-    pmaxud      m4, m5
-
-    mov         r4, r7m
-    movd        [r4], m4
-    dec         dword [r4]
-
-    phaddd      m7, m7
-    phaddd      m7, m7
+    pxor        m0, m0
+    psadbw      m7, m0
+    movhlps     m0, m7
+    paddd       m7, m0
     movd        eax, m7
+    RET
+
+
+IACA_START
+%if ARCH_X86_64 == 1
+INIT_YMM avx2
+cglobal quant, 5,5,10
+    ; fill qbits
+    movd            xm4, r4d            ; m4 = qbits
+
+    ; fill qbits-8
+    sub             r4d, 8
+    movd            xm6, r4d            ; m6 = qbits8
+
+    ; fill offset
+    vpbroadcastd    m5, r5m             ; m5 = add
+
+    vpbroadcastw    m9, [pw_1]          ; m9 = word [1]
+
+    mov             r4d, r6m
+    shr             r4d, 4
+    pxor            m7, m7              ; m7 = numZero
+.loop:
+    ; 8 coeff
+    movu            m0, [r0]            ; m0 = level
+    pabsd           m1, m0
+    pmulld          m1, [r1]            ; m0 = tmpLevel1
+    paddd           m2, m1, m5
+    psrad           m2, xm4             ; m2 = level1
+
+    pslld           m3, m2, 8
+    psrad           m1, xm6
+    psubd           m1, m3              ; m1 = deltaU1
+    movu            [r2], m1
+    psignd          m2, m0
+
+    ; 8 coeff
+    movu            m0, [r0 + mmsize]   ; m0 = level
+    pabsd           m1, m0
+    pmulld          m1, [r1 + mmsize]   ; m0 = tmpLevel1
+    paddd           m3, m1, m5
+    psrad           m3, xm4             ; m2 = level1
+
+    pslld           m8, m3, 8
+    psrad           m1, xm6
+    psubd           m1, m8              ; m1 = deltaU1
+    movu            [r2 + mmsize], m1
+    psignd          m3, m0
+
+    packssdw        m2, m3
+    vpermq          m2, m2, q3120
+    movu            [r3], m2
+
+    ; count non-zero coeff
+    ; TODO: popcnt is faster, but some CPU can't support
+    pminuw          m2, m9
+    paddw           m7, m2
+
+    add             r0, mmsize*2
+    add             r1, mmsize*2
+    add             r2, mmsize*2
+    add             r3, mmsize
+
+    dec             r4d
+    jnz            .loop
+
+    ; sum count
+    xorpd           m0, m0
+    psadbw          m7, m0
+    vextracti128    xm1, m7, 1
+    paddd           xm7, xm1
+    movhlps         xm0, xm7
+    paddd           xm7, xm0
+    movd            eax, xm7
+    RET
+
+%else ; ARCH_X86_64 == 1
+INIT_YMM avx2
+cglobal quant, 5,6,8
+    ; fill qbits
+    movd            xm4, r4d        ; m4 = qbits
+
+    ; fill qbits-8
+    sub             r4d, 8
+    movd            xm6, r4d        ; m6 = qbits8
+
+    ; fill offset
+    vpbroadcastd    m5, r5m         ; m5 = ad
 
+    lea             r5, [pd_1]
+
+    mov             r4d, r6m
+    shr             r4d, 4
+    pxor            m7, m7          ; m7 = numZero
+.loop:
+    ; 8 coeff
+    movu            m0, [r0]        ; m0 = level
+    pabsd           m1, m0
+    pmulld          m1, [r1]        ; m0 = tmpLevel1
+    paddd           m2, m1, m5
+    psrad           m2, xm4         ; m2 = level1
+
+    pslld           m3, m2, 8
+    psrad           m1, xm6
+    psubd           m1, m3          ; m1 = deltaU1
+
+    movu            [r2], m1
+    psignd          m3, m2, m0
+    pminud          m2, [r5]
+    paddd           m7, m2
+    packssdw        m3, m3
+    vpermq          m3, m3, q0020
+    movu            [r3], xm3
+
+    ; 8 coeff
+    movu            m0, [r0 + mmsize]        ; m0 = level
+    pabsd           m1, m0
+    pmulld          m1, [r1 + mmsize]        ; m0 = tmpLevel1
+    paddd           m2, m1, m5
+    psrad           m2, xm4         ; m2 = level1
+
+    pslld           m3, m2, 8
+    psrad           m1, xm6
+    psubd           m1, m3          ; m1 = deltaU1
+
+    movu            [r2 + mmsize], m1
+    psignd          m3, m2, m0
+    pminud          m2, [r5]
+    paddd           m7, m2
+    packssdw        m3, m3
+    vpermq          m3, m3, q0020
+    movu            [r3 + mmsize/2], xm3
+
+    add             r0, mmsize*2
+    add             r1, mmsize*2
+    add             r2, mmsize*2
+    add             r3, mmsize
+
+    dec             r4d
+    jnz            .loop
+
+    xorpd           m0, m0
+    psadbw          m7, m0
+    vextracti128    xm1, m7, 1
+    paddd           xm7, xm1
+    movhlps         xm0, xm7
+    paddd           xm7, xm0
+    movd            eax, xm7
+    RET
+%endif ; ARCH_X86_64 == 1
+IACA_END
+
+
+;-----------------------------------------------------------------------------
+; uint32_t nquant(int32_t *coef, int32_t *quantCoeff, int16_t *qCoef, int qBits, int add, int numCoeff);
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal nquant, 3,5,8
+    movd        m6, r4m
+    mov         r4d, r5m
+    pxor        m7, m7          ; m7 = numZero
+    movd        m5, r3m         ; m5 = qbits
+    pshufd      m6, m6, 0       ; m6 = add
+    mov         r3d, r4d        ; r3 = numCoeff
+    shr         r4d, 3
+
+.loop:
+    movu        m0, [r0]        ; m0 = level
+    movu        m1, [r0 + 16]   ; m1 = level
+
+    pabsd       m2, m0
+    pmulld      m2, [r1]        ; m0 = tmpLevel1 * qcoeff
+    paddd       m2, m6
+    psrad       m2, m5          ; m0 = level1
+    psignd      m2, m0
+
+    pabsd       m3, m1
+    pmulld      m3, [r1 + 16]   ; m1 = tmpLevel1 * qcoeff
+    paddd       m3, m6
+    psrad       m3, m5          ; m1 = level1
+    psignd      m3, m1
+
+    packssdw    m2, m3
+
+    movu        [r2], m2
+    add         r0, 32
+    add         r1, 32
+    add         r2, 16
+
+    pxor        m4, m4
+    pcmpeqw     m2, m4
+    psubw       m7, m2
+
+    dec         r4d
+    jnz         .loop
+
+    packuswb    m7, m7
+    psadbw      m7, m4
+    mov         eax, r3d
+    movd        r4d, m7
+    sub         eax, r4d        ; numSig
+    RET
+
+
+INIT_YMM avx2
+cglobal nquant, 3,5,7
+    vpbroadcastd m4, r4m
+    vpbroadcastd m6, [pw_1]
+    mov         r4d, r5m
+    pxor        m5, m5              ; m7 = numZero
+    movd        xm3, r3m            ; m5 = qbits
+    mov         r3d, r4d            ; r3 = numCoeff
+    shr         r4d, 4
+
+.loop:
+    movu        m0, [r0]            ; m0 = level
+    pabsd       m1, m0
+    pmulld      m1, [r1]            ; m0 = tmpLevel1 * qcoeff
+    paddd       m1, m4
+    psrad       m1, xm3             ; m0 = level1
+    psignd      m1, m0
+
+    movu        m0, [r0 + mmsize]   ; m0 = level
+    pabsd       m2, m0
+    pmulld      m2, [r1 + mmsize]   ; m0 = tmpLevel1 * qcoeff
+    paddd       m2, m4
+    psrad       m2, xm3             ; m0 = level1
+    psignd      m2, m0
+
+    packssdw    m1, m2
+    vpermq      m2, m1, q3120
+
+    movu        [r2], m2
+    add         r0, mmsize * 2
+    add         r1, mmsize * 2
+    add         r2, mmsize
+
+    pminuw      m1, m6
+    paddw       m5, m1
+
+    dec         r4d
+    jnz         .loop
+
+    pxor        m0, m0
+    psadbw      m5, m0
+    vextracti128 xm0, m5, 1
+    paddd       xm5, xm0
+    pshufd      xm0, xm5, 2
+    paddd       xm5, xm0
+    movd        eax, xm5
     RET
 
 
 ;-----------------------------------------------------------------------------
-; void dequant_normal(const int32_t* quantCoef, int32_t* coef, int num, int scale, int shift)
+; void dequant_normal(const int16_t* quantCoef, int32_t* coef, int num, int scale, int shift)
 ;-----------------------------------------------------------------------------
 INIT_XMM sse4
-cglobal dequant_normal, 4,5,5
-    movd        m1, r3             ; m1 = word [scale]
+cglobal dequant_normal, 5,5,5
+    mova        m2, [pw_1]
+%if HIGH_BIT_DEPTH
     cmp         r3d, 32767
     jle         .skip
-
-    psrld       m1, 2
-    mov         r4d, r4m
+    shr         r3d, 2
+    sub         r4d, 2
+.skip:
+%endif
     movd        m0, r4d             ; m0 = shift
-    xor         r3d, r3d
-    dec         r4d
+    add         r4d, 15
     bts         r3d, r4d
-    movd        m2, r3d
-    punpcklwd   m1, m2
+    movd        m1, r3d
     pshufd      m1, m1, 0           ; m1 = dword [add scale]
-    mova        m2, [pw_1]
-    mov         r2d, r2m
-
     ; m0 = shift
     ; m1 = scale
     ; m2 = word [1]
 .loop:
     movu        m3, [r0]
-    movu        m4, [r0 + 16]
-    packssdw    m3, m4              ; m3 = clipQCoef
-    psllw       m3, 2
     punpckhwd   m4, m3, m2
     punpcklwd   m3, m2
     pmaddwd     m3, m1              ; m3 = dword (clipQCoef * scale + add)
@@ -1231,81 +1212,86 @@
     pmovsxwd    m3, m3
     packssdw    m4, m4
     pmovsxwd    m4, m4
-    movu        [r1], m3
-    movu        [r1 + 16], m4
+    mova        [r1], m3
+    mova        [r1 + 16], m4
 
-    add         r0, 32
+    add         r0, 16
     add         r1, 32
 
     sub         r2d, 8
     jnz        .loop
-    jz         .end
+    RET
+
 
+INIT_YMM avx2
+cglobal dequant_normal, 5,5,7
+    vpbroadcastd    m2, [pw_1]          ; m2 = word [1]
+    vpbroadcastd    m5, [pd_32767]      ; m5 = dword [32767]
+    vpbroadcastd    m6, [pd_n32768]     ; m6 = dword [-32768]
+%if HIGH_BIT_DEPTH
+    cmp             r3d, 32767
+    jle            .skip
+    shr             r3d, 2
+    sub             r4d, 2
 .skip:
-    mov         r4d, r4m
-    movd        m0, r4d             ; m0 = shift
-    xor         r3d, r3d
-    dec         r4d
-    bts         r3d, r4d
-    movd        m2, r3d
-    punpcklwd   m1, m2
-    pshufd      m1, m1, 0           ; m1 = dword [add scale]
-    mova        m2, [pw_1]
-    mov         r2d, r2m
+%endif
+    movd            xm0, r4d            ; m0 = shift
+    add             r4d, -1+16
+    bts             r3d, r4d
+    vpbroadcastd    m1, r3d             ; m1 = dword [add scale]
+
     ; m0 = shift
     ; m1 = scale
     ; m2 = word [1]
-.sloop:
-    movu        m3, [r0]
-    movu        m4, [r0 + 16]
-    packssdw    m3, m4              ; m3 = clipQCoef
-    punpckhwd   m4, m3, m2
-    punpcklwd   m3, m2
-    pmaddwd     m3, m1              ; m3 = dword (clipQCoef * scale + add)
-    pmaddwd     m4, m1
-    psrad       m3, m0
-    psrad       m4, m0
-    packssdw    m3, m3              ; OPT_ME: store must be 32 bits
-    pmovsxwd    m3, m3
-    packssdw    m4, m4
-    pmovsxwd    m4, m4
-    movu        [r1], m3
-    movu        [r1 + 16], m4
-
-    add         r0, 32
-    add         r1, 32
+    shr             r2d, 4
+.loop:
+    movu            m3, [r0]
+    punpckhwd       m4, m3, m2
+    punpcklwd       m3, m2
+    pmaddwd         m3, m1              ; m3 = dword (clipQCoef * scale + add)
+    pmaddwd         m4, m1
+    psrad           m3, xm0
+    psrad           m4, xm0
+    pminsd          m3, m5
+    pmaxsd          m3, m6
+    pminsd          m4, m5
+    pmaxsd          m4, m6
+    mova            [r1 + 0 * mmsize/2], xm3
+    mova            [r1 + 1 * mmsize/2], xm4
+    vextracti128    [r1 + 2 * mmsize/2], m3, 1
+    vextracti128    [r1 + 3 * mmsize/2], m4, 1
+
+    add             r0, mmsize
+    add             r1, mmsize * 2
 
-    sub         r2d, 8
-    jnz        .sloop
-.end:
+    dec             r2d
+    jnz            .loop
     RET
 
 
 ;-----------------------------------------------------------------------------
-; int count_nonzero(const int32_t *quantCoeff, int numCoeff);
+; int count_nonzero(const int16_t *quantCoeff, int numCoeff);
 ;-----------------------------------------------------------------------------
-INIT_XMM sse2
-cglobal count_nonzero, 2,2,4
+INIT_XMM ssse3
+cglobal count_nonzero, 2,2,3
     pxor        m0, m0
-    shr         r1d, 3
+    shr         r1d, 4
     movd        m1, r1d
-    pshuflw     m1, m1, 0
-    punpcklqdq  m1, m1
+    pshufb      m1, m0
 
 .loop:
-    mova        m2, [r0]
-    mova        m3, [r0 + 16]
+    mova        m2, [r0 +  0]
+    packsswb    m2, [r0 + 16]
     add         r0, 32
-    packssdw    m2, m3
-    pcmpeqw     m2, m0
-    paddw       m1, m2
+    pcmpeqb     m2, m0
+    paddb       m1, m2
     dec         r1d
     jnz         .loop
 
-    packuswb    m1, m1
     psadbw      m1, m0
-    movd        eax, m1
-
+    pshufd      m0, m1, 2
+    paddd       m0, m1
+    movd        eax, m0
     RET
 
 
@@ -1518,9 +1504,9 @@
     punpcklwd    m2,    m3
     punpckhdq    m1,    m0,    m2
     punpckldq    m0,    m2
-    movlps         [r0],             m0
-    movhps         [r0 + %1],        m0
-    movlps         [r0 + 2 * %1],    m1
+    movh         [r0],             m0
+    movhps       [r0 + %1],        m0
+    movh         [r0 + 2 * %1],    m1
     lea            r0,               [r0 + 2 * %1]
     movhps         [r0 + %1],        m1
 %endmacro
@@ -1607,15 +1593,15 @@
     punpckhdq    m3,    m1,    m5
     punpckldq    m1,    m5
 
-    movlps         [r0],             m0
+    movh           [r0],             m0
     movhps         [r0 + %1],        m0
-    movlps         [r0 + 2 * %1],    m2
+    movh           [r0 + 2 * %1],    m2
     lea            r0,               [r0 + 2 * %1]
     movhps         [r0 + %1],        m2
-    movlps         [r0 + 2 * %1],    m1
+    movh           [r0 + 2 * %1],    m1
     lea            r0,               [r0 + 2 * %1]
     movhps         [r0 + %1],        m1
-    movlps         [r0 + 2 * %1],    m3
+    movh           [r0 + 2 * %1],    m3
     lea            r0,               [r0 + 2 * %1]
     movhps         [r0 + %1],        m3
 
@@ -2195,12 +2181,13 @@
 ;-----------------------------------------------------------------------------
 ; float pixel_ssim_end( int sum0[5][4], int sum1[5][4], int width )
 ;-----------------------------------------------------------------------------
-cglobal pixel_ssim_end4, 3,3,7
-    movdqa    m0, [r0+ 0]
-    movdqa    m1, [r0+16]
-    movdqa    m2, [r0+32]
-    movdqa    m3, [r0+48]
-    movdqa    m4, [r0+64]
+cglobal pixel_ssim_end4, 2,3
+    mov       r2d, r2m
+    mova      m0, [r0+ 0]
+    mova      m1, [r0+16]
+    mova      m2, [r0+32]
+    mova      m3, [r0+48]
+    mova      m4, [r0+64]
     paddd     m0, [r1+ 0]
     paddd     m1, [r1+16]
     paddd     m2, [r1+32]
@@ -2210,8 +2197,6 @@
     paddd     m1, m2
     paddd     m2, m3
     paddd     m3, m4
-    movdqa    m5, [ssim_c1]
-    movdqa    m6, [ssim_c2]
     TRANSPOSE4x4D  0, 1, 2, 3, 4
 
 ;   s1=m0, s2=m1, ss=m2, s12=m3
@@ -2220,20 +2205,21 @@
     cvtdq2ps  m1, m1
     cvtdq2ps  m2, m2
     cvtdq2ps  m3, m3
+    mulps     m4, m0, m1  ; s1*s2
+    mulps     m0, m0      ; s1*s1
+    mulps     m1, m1      ; s2*s2
     mulps     m2, [pf_64] ; ss*64
     mulps     m3, [pf_128] ; s12*128
-    movdqa    m4, m1
-    mulps     m4, m0      ; s1*s2
-    mulps     m1, m1      ; s2*s2
-    mulps     m0, m0      ; s1*s1
     addps     m4, m4      ; s1*s2*2
     addps     m0, m1      ; s1*s1 + s2*s2
     subps     m2, m0      ; vars
     subps     m3, m4      ; covar*2
-    addps     m4, m5      ; s1*s2*2 + ssim_c1
-    addps     m0, m5      ; s1*s1 + s2*s2 + ssim_c1
-    addps     m2, m6      ; vars + ssim_c2
-    addps     m3, m6      ; covar*2 + ssim_c2
+    movaps    m1, [ssim_c1]
+    addps     m4, m1      ; s1*s2*2 + ssim_c1
+    addps     m0, m1      ; s1*s1 + s2*s2 + ssim_c1
+    movaps    m1, [ssim_c2]
+    addps     m2, m1      ; vars + ssim_c2
+    addps     m3, m1      ; covar*2 + ssim_c2
 %else
     pmaddwd   m4, m1, m0  ; s1*s2
     pslld     m1, 16
@@ -2244,10 +2230,12 @@
     pslld     m2, 6
     psubd     m3, m4  ; covar*2
     psubd     m2, m0  ; vars
-    paddd     m0, m5
-    paddd     m4, m5
-    paddd     m3, m6
-    paddd     m2, m6
+    mova      m1, [ssim_c1]
+    paddd     m0, m1
+    paddd     m4, m1
+    mova      m1, [ssim_c2]
+    paddd     m3, m1
+    paddd     m2, m1
     cvtdq2ps  m0, m0  ; (float)(s1*s1 + s2*s2 + ssim_c1)
     cvtdq2ps  m4, m4  ; (float)(s1*s2*2 + ssim_c1)
     cvtdq2ps  m3, m3  ; (float)(covar*2 + ssim_c2)
@@ -2260,20 +2248,31 @@
     cmp       r2d, 4
     je .skip ; faster only if this is the common case; remove branch if we use ssim on a macroblock level
     neg       r2
+
 %ifdef PIC
     lea       r3, [mask_ff + 16]
-    movdqu    m1, [r3 + r2*4]
+    %xdefine %%mask r3
+%else
+    %xdefine %%mask mask_ff + 16
+%endif
+%if cpuflag(avx)
+    andps     m4, [%%mask + r2*4]
 %else
-    movdqu    m1, [mask_ff + r2*4 + 16]
+    movups    m0, [%%mask + r2*4]
+    andps     m4, m0
 %endif
-    pand      m4, m1
+
 .skip:
     movhlps   m0, m4
     addps     m0, m4
-    pshuflw   m4, m0, q0032
+%if cpuflag(ssse3)
+    movshdup  m4, m0
+%else
+     pshuflw   m4, m0, q0032
+%endif
     addss     m0, m4
 %if ARCH_X86_64 == 0
-    movd     r0m, m0
+    movss    r0m, m0
     fld     dword r0m
 %endif
     RET
@@ -2720,1588 +2719,747 @@
 
 
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_2x4(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_4x4(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_sub_ps_2x4, 6, 6, 2
-    add      r1, r1
-    add      r4, r4
-    add      r5, r5
-
-    movd     m0, [r2]
-    movd     m1, [r3]
-    psubw    m0, m1
-    movd     [r0], m0
-
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    movd     [r0 + r1], m0
-
-    movd     m0, [r2 + 2 * r4]
-    movd     m1, [r3 + 2 * r5]
-    psubw    m0, m1
-    movd     [r0 + 2 * r1], m0
-
-    lea      r2, [r2 + 2 * r4]
-    lea      r3, [r3 + 2 * r5]
-    lea      r0, [r0 + 2 * r1]
-
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    movd     [r0 + r1], m0
-%else
-INIT_XMM sse4
-%if ARCH_X86_64
-    cglobal pixel_sub_ps_2x4, 6, 8, 0
-
-    %define tmp_r1     r1
-    DECLARE_REG_TMP    6, 7
-%else
-    cglobal pixel_sub_ps_2x4, 6, 7, 0, 0-4
-
-    %define tmp_r1     dword [rsp]
-    DECLARE_REG_TMP    6, 1
-%endif ; ARCH_X86_64
-
-    add    r1,         r1
-
-%if ARCH_X86_64 == 0
-    mov    tmp_r1,     r1
-
-%endif
-
-movzx    t0d,      byte [r2]
-movzx    t1d,      byte [r3]
-
-sub      t0d,      t1d
-
-mov      [r0],     t0w
-
-movzx    t0d,      byte [r2 + 1]
-movzx    t1d,      byte [r3 + 1]
-
-sub      t0d,      t1d
-
-mov      [r0 + 2], t0w
-
-add      r0,       tmp_r1
-
-movzx    t0d,      byte [r2 + r4]
-movzx    t1d,      byte [r3 + r5]
-
-sub      t0d,      t1d
-
-mov      [r0],     t0w
-
-movzx    t0d,      byte [r2 + r4 + 1]
-movzx    t1d,      byte [r3 + r5 + 1]
-
-sub      t0d,      t1d
-
-mov      [r0 + 2], t0w
-
-add      r0,       tmp_r1
-
-movzx    t0d,      byte [r2 + r4 * 2]
-movzx    t1d,      byte [r3 + r5 * 2]
-
-sub      t0d,      t1d
-
-mov      [r0],     t0w
-
-movzx    t0d,      byte [r2 + r4 * 2 + 1]
-movzx    t1d,      byte [r3 + r5 * 2 + 1]
-
-sub      t0d,      t1d
-
-mov      [r0 + 2], t0w
-
-add      r0,       tmp_r1
-
-lea      r2,       [r2 + r4 * 2]
-lea      r3,       [r3 + r5 * 2]
-
-movzx    t0d,      byte [r2 + r4]
-movzx    t1d,      byte [r3 + r5]
-
-sub      t0d,      t1d
-
-mov      [r0],     t0w
-
-movzx    t0d,      byte [r2 + r4 + 1]
-movzx    t1d,      byte [r3 + r5 + 1]
-
-sub      t0d,      t1d
-
-mov      [r0 + 2], t0w
-%endif
-RET
+cglobal pixel_sub_ps_4x4, 6, 6, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
+    movh    m0,     [r2]
+    movh    m2,     [r2 + r4]
+    movh    m1,     [r3]
+    movh    m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+    movh    m4,     [r2]
+    movh    m6,     [r2 + r4]
+    movh    m5,     [r3]
+    movh    m7,     [r3 + r5]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movh    [r0],       m0
+    movh    [r0 + r1],  m2
+    lea     r0,     [r0 + r1 * 2]
+    movh    [r0],       m4
+    movh    [r0 + r1],  m6
 
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_2x8(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_sub_ps_2x8, 6, 6, 2
-    add      r1, r1
-    add      r4, r4
-    add      r5, r5
-
-    movd     m0, [r2]
-    movd     m1, [r3]
-    psubw    m0, m1
-    movd     [r0], m0
-
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    movd     [r0 + r1], m0
-
-    movd     m0, [r2 + 2 * r4]
-    movd     m1, [r3 + 2 * r5]
-    psubw    m0, m1
-    movd     [r0 + 2 * r1], m0
-
-    lea      r2, [r2 + 2 * r4]
-    lea      r3, [r3 + 2 * r5]
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    lea      r0, [r0 + 2 * r1]
-    movd     [r0 + r1], m0
-
-    movd     m0, [r2 + 2 * r4]
-    movd     m1, [r3 + 2 * r5]
-    psubw    m0,    m1
-    movd     [r0 + 2 * r1],  m0
-
-    lea      r2, [r2 + 2 * r4]
-    lea      r3, [r3 + 2 * r5]
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    lea      r0, [r0 + 2 * r1]
-    movd     [r0 + r1], m0
-
-    movd     m0, [r2 + 2 * r4]
-    movd     m1, [r3 + 2 * r5]
-    psubw    m0, m1
-    movd     [r0 + 2 * r1], m0
-
-    lea      r2, [r2 + 2 * r4]
-    lea      r3, [r3 + 2 * r5]
-    movd     m0, [r2 + r4]
-    movd     m1, [r3 + r5]
-    psubw    m0, m1
-    lea      r0, [r0 + 2 * r1]
-    movd     [r0 + r1], m0
+    RET
 %else
 INIT_XMM sse4
-%if ARCH_X86_64
-    cglobal pixel_sub_ps_2x8, 6, 8, 0
-
-    %define tmp_r1     r1
-    DECLARE_REG_TMP    6, 7
-%else
-    cglobal pixel_sub_ps_2x8, 6, 7, 0, 0-4
-
-    %define tmp_r1     dword [rsp]
-    DECLARE_REG_TMP    6, 1
-%endif ; ARCH_X86_64
-
-    add    r1,         r1
-
-%if ARCH_X86_64 == 0
-    mov    tmp_r1,     r1
-
-%endif
-
-    movzx    t0d,      byte [r2]
-    movzx    t1d,      byte [r3]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + 1]
-    movzx    t1d,      byte [r3 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    movzx    t0d,      byte [r2 + r4]
-    movzx    t1d,      byte [r3 + r5]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 + 1]
-    movzx    t1d,      byte [r3 + r5 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    movzx    t0d,      byte [r2 + r4 * 2]
-    movzx    t1d,      byte [r3 + r5 * 2]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 * 2 + 1]
-    movzx    t1d,      byte [r3 + r5 * 2 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    lea      r2,       [r2 + r4 * 2]
-    lea      r3,       [r3 + r5 * 2]
-
-    movzx    t0d,      byte [r2 + r4]
-    movzx    t1d,      byte [r3 + r5]
+cglobal pixel_sub_ps_4x4, 6, 6, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    add         r1,     r1
+    movd        m0,     [r2]
+    movd        m2,     [r2 + r4]
+    movd        m1,     [r3]
+    movd        m3,     [r3 + r5]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    movd        m4,     [r2]
+    movd        m6,     [r2 + r4]
+    movd        m5,     [r3]
+    movd        m7,     [r3 + r5]
+    punpckldq   m0,     m2
+    punpckldq   m1,     m3
+    punpckldq   m4,     m6
+    punpckldq   m5,     m7
+    pmovzxbw    m0,     m0
+    pmovzxbw    m1,     m1
+    pmovzxbw    m4,     m4
+    pmovzxbw    m5,     m5
+
+    psubw       m0,     m1
+    psubw       m4,     m5
+
+    movh        [r0],           m0
+    movhps      [r0 + r1],      m0
+    movh        [r0 + r1 * 2],  m4
+    lea         r0,     [r0 + r1 * 2]
+    movhps      [r0 + r1],      m4
 
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 + 1]
-    movzx    t1d,      byte [r3 + r5 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    movzx    t0d,      byte [r2 + r4 * 2]
-    movzx    t1d,      byte [r3 + r5 * 2]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 * 2 + 1]
-    movzx    t1d,      byte [r3 + r5 * 2 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    lea      r2,       [r2 + r4 * 2]
-    lea      r3,       [r3 + r5 * 2]
-
-    movzx    t0d,      byte [r2 + r4]
-    movzx    t1d,      byte [r3 + r5]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 + 1]
-    movzx    t1d,      byte [r3 + r5 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    movzx    t0d,      byte [r2 + r4 * 2]
-    movzx    t1d,      byte [r3 + r5 * 2]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 * 2 + 1]
-    movzx    t1d,      byte [r3 + r5 * 2 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
-
-    add      r0,       tmp_r1
-
-    lea      r2,       [r2 + r4 * 2]
-    lea      r3,       [r3 + r5 * 2]
-
-    movzx    t0d,      byte [r2 + r4]
-    movzx    t1d,      byte [r3 + r5]
-
-    sub      t0d,      t1d
-
-    mov      [r0],     t0w
-    movzx    t0d,      byte [r2 + r4 + 1]
-    movzx    t1d,      byte [r3 + r5 + 1]
-
-    sub      t0d,      t1d
-
-    mov      [r0 + 2], t0w
+    RET
 %endif
-RET
-
-;-----------------------------------------------------------------------------
-; void pixel_sub_sp_c_4x2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_sub_ps_4x2, 6, 6, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    movh     m0,    [r2]
-    movh     m1,    [r3]
-    movh     m2,    [r2 + r4]
-    movh     m3,    [r3 + r5]
-    psubw    m0,    m1
-    psubw    m2,    m3
-
-    movh     [r0],         m0
-    movh     [r0 + r1],    m2
-%else
-INIT_XMM sse4
-cglobal pixel_sub_ps_4x2, 6, 6, 4, dest, deststride, src0, src1, srcstride0, srcstride1
-
-add          r1,    r1
-
-movd         m0,    [r2]
-movd         m1,    [r3]
-
-movd         m2,    [r2 + r4]
-movd         m3,    [r3 + r5]
-
-punpckldq    m0,    m2
-punpckldq    m1,    m3
-pmovzxbw     m0,    m0
-pmovzxbw     m1,    m1
 
-psubw        m0,    m1
-
-movlps    [r0],         m0
-movhps    [r0 + r1],    m0
-%endif
-RET
 
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_4x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 %macro PIXELSUB_PS_W4_H4 2
-cglobal pixel_sub_ps_%1x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    mov    r6d,    %2/4
-    add          r1,    r1
-%if HIGH_BIT_DEPTH
-    add      r4,    r4
-    add      r5,    r5
-.loop:
-    movh     m0,    [r2]
-    movh     m1,    [r3]
-    movh     m2,    [r2 + r4]
-    movh     m3,    [r3 + r5]
-    movh     m4,    [r2 + 2 * r4]
-    movh     m5,    [r3 + 2 * r5]
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    movh     m6,    [r2 + r4]
-    movh     m7,    [r3 + r5]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    psubw    m6,    m7
-
-    movh     [r0],           m0
-    movh     [r0 + r1],      m2
-    movh     [r0 + 2 * r1],  m4
-    lea      r0,             [r0 + 2 * r1]
-    movh     [r0 + r1],      m6
-%else
-.loop:
-    movd         m0,    [r2]
-    movd         m1,    [r3]
-    movd         m2,    [r2 + r4]
-    movd         m3,    [r3 + r5]
-    movd         m4,    [r2 + 2 * r4]
-    movd         m5,    [r3 + 2 * r5]
-    lea          r2,    [r2 + 2 * r4]
-    lea          r3,    [r3 + 2 * r5]
-    movd         m6,    [r2 + r4]
-    movd         m7,    [r3 + r5]
-    punpckldq    m0,    m2
-    punpckldq    m1,    m3
-    punpckldq    m4,    m6
-    punpckldq    m5,    m7
-    pmovzxbw     m0,    m0
-    pmovzxbw     m1,    m1
-    pmovzxbw     m4,    m4
-    pmovzxbw     m5,    m5
-    psubw        m0,    m1
-    psubw        m4,    m5
-
-    movlps    [r0],             m0
-    movhps    [r0 + r1],        m0
-    movlps    [r0 + 2 * r1],    m4
-    lea       r0,               [r0 + 2 * r1]
-    movhps    [r0 + r1],        m4
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
-RET
-%endmacro
 %if HIGH_BIT_DEPTH
-INIT_XMM sse2
-PIXELSUB_PS_W4_H4 4, 4
-PIXELSUB_PS_W4_H4 4, 8
-PIXELSUB_PS_W4_H4 4, 16
-%else
-INIT_XMM sse4
-PIXELSUB_PS_W4_H4 4, 4
-PIXELSUB_PS_W4_H4 4, 8
-PIXELSUB_PS_W4_H4 4, 16
-%endif
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%macro PIXELSUB_PS_W6_H4 2
-cglobal pixel_sub_ps_%1x%2, 6, 7, 2
-    add    r1,     r1
-    mov    r6d,    %2/4
-%if HIGH_BIT_DEPTH
-    add         r4, r4
-    add         r5, r5
-.loop:
-    movu        m0, [r2]
-    movu        m1, [r3]
-    psubw       m0, m1
-    movh        [r0], m0
-    movhlps     m0, m0
-    movd        [r0 + 8], m0
-
-    movu        m0, [r2 + r4]
-    movu        m1, [r3 + r5]
-    psubw       m0, m1
-    movh        [r0 + r1], m0
-    movhlps     m0, m0
-    movd        [r0 + r1 + 8], m0
-
-    lea         r2, [r2 + 2 * r4]
-    lea         r3, [r3 + 2 * r5]
-    lea         r0, [r0 + 2 * r1]
-
-    movu        m0, [r2]
-    movu        m1, [r3]
-    psubw       m0, m1
-    movh        [r0], m0
-    movhlps     m0, m0
-    movd        [r0 + 8], m0
-
-    movu        m0, [r2 + r4]
-    movu        m1, [r3 + r5]
-    psubw       m0, m1
-    movh        [r0 + r1], m0
-    movhlps     m0, m0
-    movd        [r0 + r1 + 8], m0
-%else
+cglobal pixel_sub_ps_4x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov     r6d,    %2/4
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movh        m0, [r2]
-    pmovzxbw    m0, m0
-    movh        m1, [r3]
-    pmovzxbw    m1, m1
-    psubw       m0, m1
-    movh        [r0], m0
-    pextrd      [r0 + 8], m0, 2
-
-    movh        m0, [r2 + r4]
-    pmovzxbw    m0, m0
-    movh        m1, [r3 + r5]
-    pmovzxbw    m1, m1
-    psubw       m0, m1
-    movh        [r0 + r1], m0
-    pextrd      [r0 + r1 + 8], m0, 2
-
-    movh        m0, [r2 + 2 * r4]
-    pmovzxbw    m0, m0
-    movh        m1, [r3 + 2 * r5]
-    pmovzxbw    m1, m1
-    psubw       m0, m1
-    movh        [r0 + 2* r1], m0
-    pextrd      [r0 + 2 * r1 + 8], m0, 2
-
-    lea         r2, [r2 + 2 * r4]
-    lea         r3, [r3 + 2 * r5]
-    movh        m0, [r2 + r4]
-    pmovzxbw    m0, m0
-    movh        m1, [r3 + r5]
-    pmovzxbw    m1, m1
-    psubw       m0, m1
-    lea         r0, [r0 + 2 * r1]
-    movh        [r0 + r1], m0
-    pextrd      [r0 + r1 + 8], m0, 2
-%endif
+    movh    m0,     [r2]
+    movh    m2,     [r2 + r4]
+    movh    m1,     [r3]
+    movh    m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+    movh    m4,     [r2]
+    movh    m6,     [r2 + r4]
+    movh    m5,     [r3]
+    movh    m7,     [r3 + r5]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movh    [r0],           m0
+    movh    [r0 + r1],      m2
+    movh    [r0 + r1 * 2],  m4
+    lea     r0,     [r0 + r1 * 2]
+    movh    [r0 + r1],      m6
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
+%else
+cglobal pixel_sub_ps_4x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov         r6d,    %2/4
+    add         r1,     r1
+.loop:
+    movd        m0,     [r2]
+    movd        m2,     [r2 + r4]
+    movd        m1,     [r3]
+    movd        m3,     [r3 + r5]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    movd        m4,     [r2]
+    movd        m6,     [r2 + r4]
+    movd        m5,     [r3]
+    movd        m7,     [r3 + r5]
     dec         r6d
-    lea         r2, [r2 + 2 * r4]
-    lea         r3, [r3 + 2 * r5]
-    lea         r0, [r0 + 2 * r1]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    punpckldq   m0,     m2
+    punpckldq   m1,     m3
+    punpckldq   m4,     m6
+    punpckldq   m5,     m7
+    pmovzxbw    m0,     m0
+    pmovzxbw    m1,     m1
+    pmovzxbw    m4,     m4
+    pmovzxbw    m5,     m5
+
+    psubw       m0,     m1
+    psubw       m4,     m5
+
+    movh        [r0],           m0
+    movhps      [r0 + r1],      m0
+    movh        [r0 + r1 * 2],  m4
+    lea         r0,     [r0 + r1 * 2]
+    movhps      [r0 + r1],      m4
+    lea         r0,     [r0 + r1 * 2]
+
     jnz         .loop
     RET
+%endif
 %endmacro
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-PIXELSUB_PS_W6_H4 6, 8
-%else
-INIT_XMM sse4
-PIXELSUB_PS_W6_H4 6, 8
-%endif
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_8x2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_sub_ps_8x2, 6, 6, 4, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,    r1
-    add     r4,    r4
-    add     r5,    r5
-    movu    m0,    [r2]
-    movu    m1,    [r3]
-    movu    m2,    [r2 + r4]
-    movu    m3,    [r3 + r5]
+PIXELSUB_PS_W4_H4 4, 8
 %else
 INIT_XMM sse4
-cglobal pixel_sub_ps_8x2, 6, 6, 4, dest, deststride, src0, src1, srcstride0, srcstride1
-    add         r1,    r1
-    movh        m0,    [r2]
-    movh        m1,    [r3]
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    movh        m2,    [r2 + r4]
-    movh        m3,    [r3 + r5]
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
+PIXELSUB_PS_W4_H4 4, 8
 %endif
-    psubw       m0,    m1
-    psubw       m2,    m3
 
-    movu    [r0],         m0
-    movu    [r0 + r1],    m2
-    RET
 
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_8x6(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_sub_ps_8x6, 6, 6, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    movu     m4,    [r2]
-    movu     m5,    [r3]
-    movu     m6,    [r2 + r4]
-    movu     m7,    [r3 + r5]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    psubw    m6,    m7
-
-    movu    [r0],             m0
-    movu    [r0 + r1],        m2
-    lea     r0,               [r0 + 2 * r1]
-    movu    [r0],             m4
-    movu    [r0 + r1],        m6
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    psubw    m0,    m1
-    psubw    m2,    m3
-
-    lea     r0,               [r0 + 2 * r1]
-    movu    [r0],             m0
-    movu    [r0 + r1],        m2
-%else
-
-INIT_XMM sse4
-cglobal pixel_sub_ps_8x6, 6, 6, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    add         r1,    r1
-    movh        m0,    [r2]
-    movh        m1,    [r3]
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    movh        m2,    [r2 + r4]
-    movh        m3,    [r3 + r5]
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
-    movh        m4,    [r2 + 2 * r4]
-    movh        m5,    [r3 + 2 * r5]
-    pmovzxbw    m4,    m4
-    pmovzxbw    m5,    m5
-    psubw       m0,    m1
-    psubw       m2,    m3
-    psubw       m4,    m5
-    lea         r2,    [r2 + 2 * r4]
-    lea         r3,    [r3 + 2 * r5]
-    movh        m6,    [r2 + r4]
-    movh        m7,    [r3 + r5]
-    pmovzxbw    m6,    m6
-    pmovzxbw    m7,    m7
-    movh        m1,    [r2 + 2 * r4]
-    movh        m3,    [r3 + 2 * r5]
-    pmovzxbw    m1,    m1
-    pmovzxbw    m3,    m3
-    psubw       m6,    m7
-    psubw       m1,    m3
-    lea         r2,    [r2 + 2 * r4]
-    lea         r3,    [r3 + 2 * r5]
-    movh        m3,    [r2 + r4]
-    movh        m5,    [r3 + r5]
-    pmovzxbw    m3,    m3
-    pmovzxbw    m5,    m5
-    psubw       m3,     m5
-
-    movu    [r0],             m0
-    movu    [r0 + r1],        m2
-    movu    [r0 + 2 * r1],    m4
-    lea     r0,               [r0 + 2 * r1]
-    movu    [r0 + r1],        m6
-    movu    [r0 + 2 * r1],    m1
-    lea     r0,               [r0 + 2 * r1]
-    movu    [r0 + r1],        m3
-%endif
-    RET
-
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_8x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 %macro PIXELSUB_PS_W8_H4 2
-cglobal pixel_sub_ps_%1x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    add    r1,     r1
-    mov    r6d,    %2/4
 %if HIGH_BIT_DEPTH
-    add     r4,    r4
-    add     r5,    r5
+cglobal pixel_sub_ps_8x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov     r6d,    %2/4
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu    m0,    [r2]
-    movu    m1,    [r3]
-    movu    m2,    [r2 + r4]
-    movu    m3,    [r3 + r5]
-    lea     r2,    [r2 + 2 * r4]
-    lea     r3,    [r3 + 2 * r5]
-    movu    m4,    [r2]
-    movu    m5,    [r3]
-    movu    m6,    [r2 + r4]
-    movu    m7,    [r3 + r5]
-%else
+    movu    m0,     [r2]
+    movu    m2,     [r2 + r4]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+    movu    m4,     [r2]
+    movu    m6,     [r2 + r4]
+    movu    m5,     [r3]
+    movu    m7,     [r3 + r5]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
 
-.loop:
-    movh        m0,    [r2]
-    movh        m1,    [r3]
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    movh        m2,    [r2 + r4]
-    movh        m3,    [r3 + r5]
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
-    movh        m4,    [r2 + 2 * r4]
-    movh        m5,    [r3 + 2 * r5]
-    pmovzxbw    m4,    m4
-    pmovzxbw    m5,    m5
-    lea         r2,    [r2 + 2 * r4]
-    lea         r3,    [r3 + 2 * r5]
-    movh        m6,    [r2 + r4]
-    movh        m7,    [r3 + r5]
-    pmovzxbw    m6,    m6
-    pmovzxbw    m7,    m7
-%endif
-    psubw       m0,    m1
-    psubw       m2,    m3
-    psubw       m4,    m5
-    psubw       m6,    m7
-
-    movu    [r0],             m0
-    movu    [r0 + r1],        m2
-    movu    [r0 + 2 * r1],    m4
-    lea     r0,               [r0 + 2 * r1]
-    movu    [r0 + r1],        m6
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0],           m0
+    movu    [r0 + r1],      m2
+    movu    [r0 + r1 * 2],  m4
+    lea     r0,     [r0 + r1 * 2]
+    movu    [r0 + r1],      m6
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
+%else
+cglobal pixel_sub_ps_8x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov         r6d,    %2/4
+    add         r1,     r1
+.loop:
+    movh        m0,     [r2]
+    movh        m2,     [r2 + r4]
+    movh        m1,     [r3]
+    movh        m3,     [r3 + r5]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    movh        m4,     [r2]
+    movh        m6,     [r2 + r4]
+    movh        m5,     [r3]
+    movh        m7,     [r3 + r5]
+    dec         r6d
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    pmovzxbw    m0,     m0
+    pmovzxbw    m1,     m1
+    pmovzxbw    m2,     m2
+    pmovzxbw    m3,     m3
+    pmovzxbw    m4,     m4
+    pmovzxbw    m5,     m5
+    pmovzxbw    m6,     m6
+    pmovzxbw    m7,     m7
+
+    psubw       m0,     m1
+    psubw       m2,     m3
+    psubw       m4,     m5
+    psubw       m6,     m7
+
+    movu        [r0],           m0
+    movu        [r0 + r1],      m2
+    movu        [r0 + r1 * 2],  m4
+    lea         r0,     [r0 + r1 * 2]
+    movu        [r0 + r1],      m6
+    lea         r0,     [r0 + r1 * 2]
 
-    dec     r6d
-    lea     r2,               [r2 + 2 * r4]
-    lea     r3,               [r3 + 2 * r5]
-    lea     r0,               [r0 + 2 * r1]
-    jnz    .loop
+    jnz         .loop
     RET
+%endif
 %endmacro
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-PIXELSUB_PS_W8_H4 8, 4
 PIXELSUB_PS_W8_H4 8, 8
 PIXELSUB_PS_W8_H4 8, 16
-PIXELSUB_PS_W8_H4 8, 32
 %else
 INIT_XMM sse4
-PIXELSUB_PS_W8_H4 8, 4
 PIXELSUB_PS_W8_H4 8, 8
 PIXELSUB_PS_W8_H4 8, 16
-PIXELSUB_PS_W8_H4 8, 32
 %endif
 
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%macro PIXELSUB_PS_W12_H4 2
-cglobal pixel_sub_ps_%1x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
-    add    r1,     r1
-    mov    r6d,    %2/4
-%if HIGH_BIT_DEPTH
-    add     r4,    r4
-    add     r5,    r5
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movh     m2,    [r2 + 16]
-    movh     m3,    [r3 + 16]
-    movu     m4,    [r2 + r4]
-    movu     m5,    [r3 + r5]
-    movh     m6,    [r2 + r4 + 16]
-    movh     m7,    [r3 + r5 + 16]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    psubw    m6,    m7
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-
-    movu    [r0],              m0
-    movh    [r0 + 16],         m2
-    movu    [r0 + r1],         m4
-    movh    [r0 + r1 + 16],    m6
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movh     m2,    [r2 + 16]
-    movh     m3,    [r3 + 16]
-    movu     m4,    [r2 + r4]
-    movu     m5,    [r3 + r5]
-    movh     m6,    [r2 + r4 + 16]
-    movh     m7,    [r3 + r5 + 16]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    psubw    m6,    m7
-    lea     r0,     [r0 + 2 * r1]
-
-    movu    [r0],              m0
-    movh    [r0 + 16],         m2
-    movu    [r0 + r1],         m4
-    movh    [r0 + r1 + 16],    m6
-%else
-.loop:
-    movu        m0,    [r2]
-    movu        m1,    [r3]
-    movu        m2,    [r2 + r4]
-    movu        m3,    [r3 + r5]
-    mova        m4,    m0
-    mova        m5,    m1
-    punpckhdq   m4,    m2
-    punpckhdq   m5,    m3
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
-    pmovzxbw    m4,    m4
-    pmovzxbw    m5,    m5
-    psubw       m0,    m1
-    psubw       m2,    m3
-    psubw       m4,    m5
-
-    movu      [r0],              m0
-    movlps    [r0 + 16],         m4
-    movu      [r0 + r1],         m2
-    movhps    [r0 + r1 + 16],    m4
-    movu      m0,    [r2 + 2 * r4]
-    movu      m1,    [r3 + 2 * r5]
-    lea       r2,    [r2 + 2 * r4]
-    lea       r3,    [r3 + 2 * r5]
-
-    movu         m2,    [r2 + r4]
-    movu         m3,    [r3 + r5]
-    mova         m4,    m0
-    mova         m5,    m1
-    punpckhdq    m4,    m2
-    punpckhdq    m5,    m3
-    pmovzxbw     m0,    m0
-    pmovzxbw     m1,    m1
-    pmovzxbw     m2,    m2
-    pmovzxbw     m3,    m3
-    pmovzxbw     m4,    m4
-    pmovzxbw     m5,    m5
-    psubw        m0,    m1
-    psubw        m2,    m3
-    psubw        m4,    m5
-
-    movu      [r0 + 2 * r1],         m0
-    movlps    [r0 + 2 * r1 + 16],    m4
-    lea       r0,                    [r0 + 2 * r1]
-    movu      [r0 + r1],             m2
-    movhps    [r0 + r1 + 16],        m4
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
-    RET
-%endmacro
-
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-PIXELSUB_PS_W12_H4 12, 16
-%else
-INIT_XMM sse4
-PIXELSUB_PS_W12_H4 12, 16
-%endif
 
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_16x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 %macro PIXELSUB_PS_W16_H4 2
 %if HIGH_BIT_DEPTH
-cglobal pixel_sub_ps_%1x%2, 6, 7, 6, dest, deststride, src0, src1, srcstride0, srcstride1
-    add    r1,     r1
-    mov    r6d,    %2/4
-    add     r4,    r4
-    add     r5,    r5
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    movu     m4,    [r2 + r4]
-    movu     m5,    [r3 + r5]
-    movu     m1,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    psubw    m4,    m5
-    psubw    m1,    m3
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-
-    movu    [r0],              m0
-    movu    [r0 + 16],         m2
-    movu    [r0 + r1],         m4
-    movu    [r0 + r1 + 16],    m1
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    movu     m4,    [r2 + r4]
-    movu     m5,    [r3 + r5]
-    movu     m1,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    psubw    m4,    m5
-    psubw    m1,    m3
-    lea     r0,     [r0 + 2 * r1]
-
-    movu    [r0],              m0
-    movu    [r0 + 16],         m2
-    movu    [r0 + r1],         m4
-    movu    [r0 + r1 + 16],    m1
-%else
-cglobal pixel_sub_ps_%1x%2, 6, 7, 7, dest, deststride, src0, src1, srcstride0, srcstride1
-    add    r1,     r1
-    mov    r6d,    %2/4
-    pxor   m6,     m6
+cglobal pixel_sub_ps_16x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov     r6d,    %2/4
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu         m1,    [r2]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu         m5,    [r2 + r4]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    movu         m3,    [r3 + r5]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m2
-    psubw        m5,    m3
-
-    movu    [r0],              m0
-    movu    [r0 + 16],         m1
-    movu    [r0 + r1],         m4
-    movu    [r0 + r1 + 16],    m5
-
-    movu         m1,    [r2 + 2 * r4]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3 + 2 * r5]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    lea          r2,    [r2 + 2 * r4]
-    lea          r3,    [r3 + 2 * r5]
-    psubw        m0,    m2
-    psubw        m1,    m3
-    movu         m5,    [r2 + r4]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    movu         m3,    [r3 + r5]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m2
-    psubw        m5,    m3
-
-    movu    [r0 + 2 * r1],         m0
-    movu    [r0 + 2 * r1 + 16],    m1
-    lea     r0,                    [r0 + 2 * r1]
-    movu    [r0 + r1],             m4
-    movu    [r0 + r1 + 16],        m5
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+    movu    m4,     [r2 + r4]
+    movu    m6,     [r2 + r4 + 16]
+    movu    m5,     [r3 + r5]
+    movu    m7,     [r3 + r5 + 16]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
 
-RET
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0],           m0
+    movu    [r0 + 16],      m2
+    movu    [r0 + r1],      m4
+    movu    [r0 + r1 + 16], m6
+
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+    movu    m4,     [r2 + r4]
+    movu    m5,     [r3 + r5]
+    movu    m6,     [r2 + r4 + 16]
+    movu    m7,     [r3 + r5 + 16]
+    lea     r0,     [r0 + r1 * 2]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0],           m0
+    movu    [r0 + 16],      m2
+    movu    [r0 + r1],      m4
+    movu    [r0 + r1 + 16], m6
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
+%else
+cglobal pixel_sub_ps_16x%2, 6, 7, 7, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov         r6d,    %2/4
+    pxor        m6,     m6
+    add         r1,     r1
+.loop:
+    movu        m1,     [r2]
+    movu        m3,     [r3]
+    pmovzxbw    m0,     m1
+    pmovzxbw    m2,     m3
+    punpckhbw   m1,     m6
+    punpckhbw   m3,     m6
+
+    psubw       m0,     m2
+    psubw       m1,     m3
+
+    movu        m5,     [r2 + r4]
+    movu        m3,     [r3 + r5]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    pmovzxbw    m4,     m5
+    pmovzxbw    m2,     m3
+    punpckhbw   m5,     m6
+    punpckhbw   m3,     m6
+
+    psubw       m4,     m2
+    psubw       m5,     m3
+
+    movu        [r0],           m0
+    movu        [r0 + 16],      m1
+    movu        [r0 + r1],      m4
+    movu        [r0 + r1 + 16], m5
+
+    movu        m1,     [r2]
+    movu        m3,     [r3]
+    pmovzxbw    m0,     m1
+    pmovzxbw    m2,     m3
+    punpckhbw   m1,     m6
+    punpckhbw   m3,     m6
+
+    psubw       m0,     m2
+    psubw       m1,     m3
+
+    movu        m5,     [r2 + r4]
+    movu        m3,     [r3 + r5]
+    dec         r6d
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    lea         r0,     [r0 + r1 * 2]
+    pmovzxbw    m4,     m5
+    pmovzxbw    m2,     m3
+    punpckhbw   m5,     m6
+    punpckhbw   m3,     m6
+
+    psubw       m4,     m2
+    psubw       m5,     m3
+
+    movu        [r0],           m0
+    movu        [r0 + 16],      m1
+    movu        [r0 + r1],      m4
+    movu        [r0 + r1 + 16], m5
+    lea         r0,     [r0 + r1 * 2]
+
+    jnz         .loop
+    RET
+%endif
 %endmacro
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-PIXELSUB_PS_W16_H4 16, 4
-PIXELSUB_PS_W16_H4 16, 8
-PIXELSUB_PS_W16_H4 16, 12
 PIXELSUB_PS_W16_H4 16, 16
 PIXELSUB_PS_W16_H4 16, 32
-PIXELSUB_PS_W16_H4 16, 64
 %else
 INIT_XMM sse4
-PIXELSUB_PS_W16_H4 16, 4
-PIXELSUB_PS_W16_H4 16, 8
-PIXELSUB_PS_W16_H4 16, 12
 PIXELSUB_PS_W16_H4 16, 16
 PIXELSUB_PS_W16_H4 16, 32
-PIXELSUB_PS_W16_H4 16, 64
 %endif
 
+
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_32x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
-%macro PIXELSUB_PS_W24_H2 2
+%macro PIXELSUB_PS_W32_H2 2
 %if HIGH_BIT_DEPTH
-cglobal pixel_sub_ps_%1x%2, 6, 7, 6, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
+cglobal pixel_sub_ps_32x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
     mov     r6d,    %2/2
     add     r4,     r4
     add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    movu     m4,    [r2 + 32]
-    movu     m5,    [r3 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0],              m0
-    movu    [r0 + 16],         m2
-    movu    [r0 + 32],         m4
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    movu     m4,    [r2 + r4 + 32]
-    movu     m5,    [r3 + r5 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m2
-    movu    [r0 + r1 + 32],    m4
-%else
-cglobal pixel_sub_ps_%1x%2, 6, 7, 7, dest, deststride, src0, src1, srcstride0, srcstride1
-    add    r1,     r1
-    mov    r6d,    %2/2
-    pxor   m6,     m6
-.loop:
-    movu         m1,    [r2]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movh         m2,    [r2 + 16]
-    pmovzxbw     m2,    m2
-    movu         m4,    [r3]
-    pmovzxbw     m3,    m4
-    punpckhbw    m4,    m6
-    movh         m5,    [r3 + 16]
-    pmovzxbw     m5,    m5
-    psubw        m0,    m3
-    psubw        m1,    m4
-    psubw        m2,    m5
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m1
-    movu    [r0 + 32],    m2
-
-    movu         m1,    [r2 + r4]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movh         m2,    [r2 + r4 + 16]
-    pmovzxbw     m2,    m2
-    movu         m4,    [r3 + r5]
-    pmovzxbw     m3,    m4
-    punpckhbw    m4,    m6
-    movh         m5,    [r3 + r5 + 16]
-    pmovzxbw     m5,    m5
-    psubw        m0,    m3
-    psubw        m1,    m4
-    psubw        m2,    m5
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m1
-    movu    [r0 + r1 + 32],    m2
-%endif
-
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
-
-RET
-%endmacro
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m4,     [r2 + 32]
+    movu    m6,     [r2 + 48]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+    movu    m5,     [r3 + 32]
+    movu    m7,     [r3 + 48]
+    dec     r6d
 
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-PIXELSUB_PS_W24_H2 24, 32
-%else
-INIT_XMM sse4
-PIXELSUB_PS_W24_H2 24, 32
-%endif
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+    movu    [r0 + 32],  m4
+    movu    [r0 + 48],  m6
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m4,     [r2 + r4 + 32]
+    movu    m6,     [r2 + r4 + 48]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+    movu    m5,     [r3 + r5 + 32]
+    movu    m7,     [r3 + r5 + 48]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+    movu    [r0 + r1 + 32], m4
+    movu    [r0 + r1 + 48], m6
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
+%else
+cglobal pixel_sub_ps_32x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov         r6d,    %2/2
+    add         r1,     r1
+.loop:
+    movh        m0,     [r2]
+    movh        m1,     [r2 + 8]
+    movh        m2,     [r2 + 16]
+    movh        m6,     [r2 + 24]
+    movh        m3,     [r3]
+    movh        m4,     [r3 + 8]
+    movh        m5,     [r3 + 16]
+    movh        m7,     [r3 + 24]
+    dec         r6d
+    pmovzxbw    m0,     m0
+    pmovzxbw    m1,     m1
+    pmovzxbw    m2,     m2
+    pmovzxbw    m6,     m6
+    pmovzxbw    m3,     m3
+    pmovzxbw    m4,     m4
+    pmovzxbw    m5,     m5
+    pmovzxbw    m7,     m7
+
+    psubw       m0,     m3
+    psubw       m1,     m4
+    psubw       m2,     m5
+    psubw       m6,     m7
+
+    movu        [r0],       m0
+    movu        [r0 + 16],  m1
+    movu        [r0 + 32],  m2
+    movu        [r0 + 48],  m6
+
+    movh        m0,     [r2 + r4]
+    movh        m1,     [r2 + r4 + 8]
+    movh        m2,     [r2 + r4 + 16]
+    movh        m6,     [r2 + r4 + 24]
+    movh        m3,     [r3 + r5]
+    movh        m4,     [r3 + r5 + 8]
+    movh        m5,     [r3 + r5 + 16]
+    movh        m7,     [r3 + r5 + 24]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    pmovzxbw    m0,     m0
+    pmovzxbw    m1,     m1
+    pmovzxbw    m2,     m2
+    pmovzxbw    m6,     m6
+    pmovzxbw    m3,     m3
+    pmovzxbw    m4,     m4
+    pmovzxbw    m5,     m5
+    pmovzxbw    m7,     m7
+
+    psubw       m0,     m3
+    psubw       m1,     m4
+    psubw       m2,     m5
+    psubw       m6,     m7
+
+    movu        [r0 + r1],      m0
+    movu        [r0 + r1 + 16], m1
+    movu        [r0 + r1 + 32], m2
+    movu        [r0 + r1 + 48], m6
+    lea         r0,     [r0 + r1 * 2]
 
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%macro PIXELSUB_PS_W32_H2 2
-cglobal pixel_sub_ps_%1x%2, 6, 7, 6, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
-    mov     r6d,    %2/2
-%if HIGH_BIT_DEPTH
-    add     r4,     r4
-    add     r5,     r5
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    movu     m4,    [r2 + 32]
-    movu     m5,    [r3 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + 48]
-    movu     m5,    [r3 + 48]
-    psubw    m3,    m5
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m2
-    movu    [r0 + 32],    m4
-    movu    [r0 + 48],    m3
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    movu     m4,    [r2 + r4 + 32]
-    movu     m5,    [r3 + r5 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + r4 + 48]
-    movu     m5,    [r3 + r5 + 48]
-    psubw    m3,    m5
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m2
-    movu    [r0 + r1 + 32],    m4
-    movu    [r0 + r1 + 48],    m3
-%else
-.loop:
-    movh        m0,    [r2]
-    movh        m1,    [r2 + 8]
-    movh        m2,    [r2 + 16]
-    movh        m3,    [r3]
-    movh        m4,    [r3 + 8]
-    movh        m5,    [r3 + 16]
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
-    pmovzxbw    m4,    m4
-    pmovzxbw    m5,    m5
-    psubw       m0,    m3
-    psubw       m1,    m4
-    psubw       m2,    m5
-    movh        m3,    [r2 + 24]
-    movh        m4,    [r3 + 24]
-    pmovzxbw    m4,    m4
-    pmovzxbw    m3,    m3
-    psubw       m3,    m4
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m1
-    movu    [r0 + 32],    m2
-    movu    [r0 + 48],    m3
-
-    movh        m0,    [r2 + r4]
-    movh        m1,    [r2 + r4 + 8]
-    movh        m2,    [r2 + r4 + 16]
-    movh        m3,    [r3 + r5]
-    movh        m4,    [r3 + r5 + 8]
-    movh        m5,    [r3 + r5 + 16]
-    pmovzxbw    m0,    m0
-    pmovzxbw    m1,    m1
-    pmovzxbw    m2,    m2
-    pmovzxbw    m3,    m3
-    pmovzxbw    m4,    m4
-    pmovzxbw    m5,    m5
-    psubw       m0,    m3
-    psubw       m1,    m4
-    psubw       m2,    m5
-    movh        m3,    [r2 + r4 + 24]
-    movh        m4,    [r3 + r5 + 24]
-    pmovzxbw    m3,    m3
-    pmovzxbw    m4,    m4
-    psubw       m3,    m4
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m1
-    movu    [r0 + r1 + 32],    m2
-    movu    [r0 + r1 + 48],    m3
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
+    jnz         .loop
     RET
+%endif
 %endmacro
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-PIXELSUB_PS_W32_H2 32, 8
-PIXELSUB_PS_W32_H2 32, 16
-PIXELSUB_PS_W32_H2 32, 24
 PIXELSUB_PS_W32_H2 32, 32
 PIXELSUB_PS_W32_H2 32, 64
 %else
 INIT_XMM sse4
-PIXELSUB_PS_W32_H2 32, 8
-PIXELSUB_PS_W32_H2 32, 16
-PIXELSUB_PS_W32_H2 32, 24
 PIXELSUB_PS_W32_H2 32, 32
 PIXELSUB_PS_W32_H2 32, 64
 %endif
 
-;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
-;-----------------------------------------------------------------------------
-%macro PIXELSUB_PS_W48_H2 2
-%if HIGH_BIT_DEPTH
-cglobal pixel_sub_ps_%1x%2, 6, 7, 6, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
-    mov     r6d,    %2/2
-    add     r4,     r4
-    add     r5,     r5
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    movu     m4,    [r2 + 32]
-    movu     m5,    [r3 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m2
-    movu    [r0 + 32],    m4
-
-    movu     m0,    [r2 + 48]
-    movu     m1,    [r3 + 48]
-    movu     m2,    [r2 + 64]
-    movu     m3,    [r3 + 64]
-    movu     m4,    [r2 + 80]
-    movu     m5,    [r3 + 80]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0 + 48],    m0
-    movu    [r0 + 64],    m2
-    movu    [r0 + 80],    m4
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    movu     m4,    [r2 + r4 + 32]
-    movu     m5,    [r3 + r5 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m2
-    movu    [r0 + r1 + 32],    m4
-
-    movu     m0,    [r2 + r4 + 48]
-    movu     m1,    [r3 + r5 + 48]
-    movu     m2,    [r2 + r4 + 64]
-    movu     m3,    [r3 + r5 + 64]
-    movu     m4,    [r2 + r4 + 80]
-    movu     m5,    [r3 + r5 + 80]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-
-    movu    [r0 + r1 + 48],    m0
-    movu    [r0 + r1 + 64],    m2
-    movu    [r0 + r1 + 80],    m4
-%else
-
-cglobal pixel_sub_ps_%1x%2, 6, 7, 7, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
-    mov     r6d,    %2/2
-    pxor    m6,     m6
-.loop:
-    movu         m1,    [r2]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    movu         m5,    [r2 + 16]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m1
-
-    movu         m3,    [r3 + 16]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m2
-    psubw        m5,    m3
-
-    movu    [r0 + 32],    m4
-    movu    [r0 + 48],    m5
-
-    movu         m1,    [r2 + 32]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3 + 32]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0 + 64],    m0
-    movu    [r0 + 80],    m1
-
-    movu         m1,    [r2 + r4]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3 + r5]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    movu         m5,    [r2 + r5 + 16]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m1
-
-    movu         m3,    [r3 + r4 + 16]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m2
-    psubw        m5,    m3
-
-    movu    [r0 + r1 + 32],    m4
-    movu    [r0 + r1 + 48],    m5
-
-    movu         m1,    [r2 + r4 + 32]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3 + r5 + 32]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0 + r1 + 64],    m0
-    movu    [r0 + r1 + 80],    m1
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
-
-RET
-%endmacro
-
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-PIXELSUB_PS_W48_H2 48, 64
-%else
-INIT_XMM sse4
-PIXELSUB_PS_W48_H2 48, 64
-%endif
 
 ;-----------------------------------------------------------------------------
-; void pixel_sub_ps_c_%1x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+; void pixel_sub_ps_64x%2(int16_t *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 %macro PIXELSUB_PS_W64_H2 2
 %if HIGH_BIT_DEPTH
-cglobal pixel_sub_ps_%1x%2, 6, 7, 6, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
+cglobal pixel_sub_ps_64x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
     mov     r6d,    %2/2
     add     r4,     r4
     add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    movu     m4,    [r2 + 32]
-    movu     m5,    [r3 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + 48]
-    movu     m5,    [r3 + 48]
-    psubw    m3,    m5
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m2
-    movu    [r0 + 32],    m4
-    movu    [r0 + 48],    m3
-
-    movu     m0,    [r2 + 64]
-    movu     m1,    [r3 + 64]
-    movu     m2,    [r2 + 80]
-    movu     m3,    [r3 + 80]
-    movu     m4,    [r2 + 96]
-    movu     m5,    [r3 + 96]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + 112]
-    movu     m5,    [r3 + 112]
-    psubw    m3,    m5
-
-    movu    [r0 + 64],     m0
-    movu    [r0 + 80],     m2
-    movu    [r0 + 96],     m4
-    movu    [r0 + 112],    m3
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    movu     m4,    [r2 + r4 + 32]
-    movu     m5,    [r3 + r5 + 32]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + r4 + 48]
-    movu     m5,    [r3 + r5 + 48]
-    psubw    m3,    m5
-
-    movu    [r0 + r1],         m0
-    movu    [r0 + r1 + 16],    m2
-    movu    [r0 + r1 + 32],    m4
-    movu    [r0 + r1 + 48],    m3
-
-    movu     m0,    [r2 + r4 + 64]
-    movu     m1,    [r3 + r5 + 64]
-    movu     m2,    [r2 + r4 + 80]
-    movu     m3,    [r3 + r5 + 80]
-    movu     m4,    [r2 + r4 + 96]
-    movu     m5,    [r3 + r5 + 96]
-    psubw    m0,    m1
-    psubw    m2,    m3
-    psubw    m4,    m5
-    movu     m3,    [r2 + r4 + 112]
-    movu     m5,    [r3 + r5 + 112]
-    psubw    m3,    m5
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m4,     [r2 + 32]
+    movu    m6,     [r2 + 48]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+    movu    m5,     [r3 + 32]
+    movu    m7,     [r3 + 48]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+    movu    [r0 + 32],  m4
+    movu    [r0 + 48],  m6
+
+    movu    m0,     [r2 + 64]
+    movu    m2,     [r2 + 80]
+    movu    m4,     [r2 + 96]
+    movu    m6,     [r2 + 112]
+    movu    m1,     [r3 + 64]
+    movu    m3,     [r3 + 80]
+    movu    m5,     [r3 + 96]
+    movu    m7,     [r3 + 112]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0 + 64],  m0
+    movu    [r0 + 80],  m2
+    movu    [r0 + 96],  m4
+    movu    [r0 + 112], m6
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m4,     [r2 + r4 + 32]
+    movu    m6,     [r2 + r4 + 48]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+    movu    m5,     [r3 + r5 + 32]
+    movu    m7,     [r3 + r5 + 48]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+    movu    [r0 + r1 + 32], m4
+    movu    [r0 + r1 + 48], m6
+
+    movu    m0,     [r2 + r4 + 64]
+    movu    m2,     [r2 + r4 + 80]
+    movu    m4,     [r2 + r4 + 96]
+    movu    m6,     [r2 + r4 + 112]
+    movu    m1,     [r3 + r5 + 64]
+    movu    m3,     [r3 + r5 + 80]
+    movu    m5,     [r3 + r5 + 96]
+    movu    m7,     [r3 + r5 + 112]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    psubw   m0,     m1
+    psubw   m2,     m3
+    psubw   m4,     m5
+    psubw   m6,     m7
 
     movu    [r0 + r1 + 64],     m0
     movu    [r0 + r1 + 80],     m2
     movu    [r0 + r1 + 96],     m4
-    movu    [r0 + r1 + 112],    m3
+    movu    [r0 + r1 + 112],    m6
+    lea     r0,     [r0 + r1 * 2]
 
+    jnz     .loop
+    RET
 %else
+cglobal pixel_sub_ps_64x%2, 6, 7, 8, dest, deststride, src0, src1, srcstride0, srcstride1
+    mov         r6d,    %2/2
+    pxor        m6,     m6
+    add         r1,     r1
+.loop:
+    movu        m1,     [r2]
+    movu        m5,     [r2 + 16]
+    movu        m3,     [r3]
+    movu        m7,     [r3 + 16]
+
+    pmovzxbw    m0,     m1
+    pmovzxbw    m4,     m5
+    pmovzxbw    m2,     m3
+    punpckhbw   m1,     m6
+    punpckhbw   m3,     m6
+    punpckhbw   m5,     m6
+
+    psubw       m0,     m2
+    psubw       m1,     m3
+    pmovzxbw    m2,     m7
+    punpckhbw   m7,     m6
+    psubw       m4,     m2
+    psubw       m5,     m7
+
+    movu        m3,     [r2 + 32]
+    movu        m7,     [r3 + 32]
+    pmovzxbw    m2,     m3
+    punpckhbw   m3,     m6
+
+    movu        [r0],       m0
+    movu        [r0 + 16],  m1
+    movu        [r0 + 32],  m4
+    movu        [r0 + 48],  m5
+
+    movu        m1,     [r2 + 48]
+    movu        m5,     [r3 + 48]
+    pmovzxbw    m0,     m1
+    pmovzxbw    m4,     m7
+    punpckhbw   m1,     m6
+    punpckhbw   m7,     m6
+
+    psubw       m2,     m4
+    psubw       m3,     m7
+
+    movu        [r0 + 64],  m2
+    movu        [r0 + 80],  m3
+
+    movu        m7,     [r2 + r4]
+    movu        m3,     [r3 + r5]
+    pmovzxbw    m2,     m5
+    pmovzxbw    m4,     m7
+    punpckhbw   m5,     m6
+    punpckhbw   m7,     m6
+
+    psubw       m0,     m2
+    psubw       m1,     m5
+
+    movu        [r0 + 96],  m0
+    movu        [r0 + 112], m1
+
+    movu        m2,     [r2 + r4 + 16]
+    movu        m5,     [r3 + r5 + 16]
+    pmovzxbw    m0,     m3
+    pmovzxbw    m1,     m2
+    punpckhbw   m3,     m6
+    punpckhbw   m2,     m6
 
-cglobal pixel_sub_ps_%1x%2, 6, 7, 7, dest, deststride, src0, src1, srcstride0, srcstride1
-    add     r1,     r1
-    mov     r6d,    %2/2
-    pxor    m6,     m6
-.loop:
-    movu         m1,    [r2]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r3]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    movu         m5,    [r2 + 16]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0],         m0
-    movu    [r0 + 16],    m1
-
-    movu         m1,    [r3 + 16]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r2 + 32]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m0
-    psubw        m5,    m1
-
-    movu    [r0 + 32],    m4
-    movu    [r0 + 48],    m5
-
-    movu         m5,    [r3 + 32]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    movu         m1,    [r2 + 48]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    psubw        m2,    m4
-    psubw        m3,    m5
-
-    movu    [r0 + 64],    m2
-    movu    [r0 + 80],    m3
-
-    movu         m3,    [r3 + 48]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    movu         m5,    [r2 + r4]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0 + 96],     m0
-    movu    [r0 + 112],    m1
-
-    movu         m1,    [r3 + r5]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    movu         m3,    [r2 + r4 + 16]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    psubw        m4,    m0
-    psubw        m5,    m1
-
-    movu    [r0 + r1],         m4
-    movu    [r0 + r1 + 16],    m5
-
-    movu         m5,    [r3 + r5 + 16]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    movu         m1,    [r2 + r4 + 32]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    psubw        m2,    m4
-    psubw        m3,    m5
-
-    movu    [r0 + r1 + 32],    m2
-    movu    [r0 + r1 + 48],    m3
-
-    movu         m3,    [r3 + r5 + 32]
-    pmovzxbw     m2,    m3
-    punpckhbw    m3,    m6
-    movu         m5,    [r2 + r4 + 48]
-    pmovzxbw     m4,    m5
-    punpckhbw    m5,    m6
-    psubw        m0,    m2
-    psubw        m1,    m3
-
-    movu    [r0 + r1 + 64],    m0
-    movu    [r0 + r1 + 80],    m1
-
-    movu         m1,    [r3 + r5 + 48]
-    pmovzxbw     m0,    m1
-    punpckhbw    m1,    m6
-    psubw        m4,    m0
-    psubw        m5,    m1
+    psubw       m4,     m0
+    psubw       m7,     m3
 
-    movu    [r0 + r1 + 96],     m4
-    movu    [r0 + r1 + 112],    m5
-%endif
-    dec    r6d
-    lea    r2,    [r2 + 2 * r4]
-    lea    r3,    [r3 + 2 * r5]
-    lea    r0,    [r0 + 2 * r1]
-    jnz    .loop
+    movu        [r0 + r1],      m4
+    movu        [r0 + r1 + 16], m7
+
+    movu        m0,     [r2 + r4 + 32]
+    movu        m3,     [r3 + r5 + 32]
+    dec         r6d
+    pmovzxbw    m4,     m5
+    pmovzxbw    m7,     m0
+    punpckhbw   m5,     m6
+    punpckhbw   m0,     m6
+
+    psubw       m1,     m4
+    psubw       m2,     m5
+
+    movu        [r0 + r1 + 32], m1
+    movu        [r0 + r1 + 48], m2
+
+    movu        m4,     [r2 + r4 + 48]
+    movu        m5,     [r3 + r5 + 48]
+    lea         r2,     [r2 + r4 * 2]
+    lea         r3,     [r3 + r5 * 2]
+    pmovzxbw    m1,     m3
+    pmovzxbw    m2,     m4
+    punpckhbw   m3,     m6
+    punpckhbw   m4,     m6
+
+    psubw       m7,     m1
+    psubw       m0,     m3
+
+    movu        [r0 + r1 + 64], m7
+    movu        [r0 + r1 + 80], m0
+
+    pmovzxbw    m7,     m5
+    punpckhbw   m5,     m6
+    psubw       m2,     m7
+    psubw       m4,     m5
+
+    movu        [r0 + r1 + 96],     m2
+    movu        [r0 + r1 + 112],    m4
+    lea         r0,     [r0 + r1 * 2]
+
+    jnz         .loop
     RET
+%endif
 %endmacro
 
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-PIXELSUB_PS_W64_H2 64, 16
-PIXELSUB_PS_W64_H2 64, 32
-PIXELSUB_PS_W64_H2 64, 48
 PIXELSUB_PS_W64_H2 64, 64
 %else
 INIT_XMM sse4
-PIXELSUB_PS_W64_H2 64, 16
-PIXELSUB_PS_W64_H2 64, 32
-PIXELSUB_PS_W64_H2 64, 48
 PIXELSUB_PS_W64_H2 64, 64
 %endif
 
+
 ;=============================================================================
 ; variance
 ;=============================================================================

x265-hg20140324.tar.bz2/source/common/x86/pixel.h -> x265-hg20140928.tar.bz2/source/common/x86/pixel.h Changed

@@ -5,7 +5,7 @@
  *
  * Authors: Laurent Aimar <fenrir@via.ecp.fr>
  *          Loren Merritt <lorenm@u.washington.edu>
- *          Jason Garrett-Glaser <darkshikari@gmail.com>
+ *          Fiona Glaser <fiona@x264.com>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -22,7 +22,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@x264.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_I386_PIXEL_H
@@ -166,6 +166,12 @@
 int x265_pixel_ssd_64x48_sse4(pixel *, intptr_t, pixel *, intptr_t);
 int x265_pixel_ssd_64x64_sse4(pixel *, intptr_t, pixel *, intptr_t);
 
+int x265_pixel_ssd_s_4_sse2(int16_t *, intptr_t);
+int x265_pixel_ssd_s_8_sse2(int16_t *, intptr_t);
+int x265_pixel_ssd_s_16_sse2(int16_t *, intptr_t);
+int x265_pixel_ssd_s_32_sse2(int16_t *, intptr_t);
+int x265_pixel_ssd_s_32_avx2(int16_t *, intptr_t);
+
 #define ADDAVG(func)  \
     void x265_ ## func ## _sse4(int16_t*, int16_t*, pixel*, intptr_t, intptr_t, intptr_t);
 ADDAVG(addAvg_2x4)
@@ -200,6 +206,19 @@
 ADDAVG(addAvg_64x48)
 ADDAVG(addAvg_64x64)
 
+ADDAVG(addAvg_2x16)
+ADDAVG(addAvg_4x32)
+ADDAVG(addAvg_6x16)
+ADDAVG(addAvg_8x12)
+ADDAVG(addAvg_8x64)
+ADDAVG(addAvg_12x32)
+ADDAVG(addAvg_16x24)
+ADDAVG(addAvg_24x64)
+ADDAVG(addAvg_32x48)
+
+void x265_downShift_16_sse2(uint16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask);
+void x265_upShift_8_sse4(uint8_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift);
+
 #undef DECL_PIXELS
 #undef DECL_HEVC_SSD
 #undef DECL_X1

x265-hg20140324.tar.bz2/source/common/x86/pixeladd8.asm -> x265-hg20140928.tar.bz2/source/common/x86/pixeladd8.asm Changed

@@ -18,7 +18,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@multicorewareinc.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************/
 
 %include "x86inc.asm"
@@ -31,1626 +31,710 @@
 cextern pw_pixel_max
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_2x4(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_4x4(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_2x4, 6, 6, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-
-    movd     m0,    [r2]
-    movd     m1,    [r3]
-    movd     m2,    [r2 + r4]
-    movd     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movd     [r0],           m0
-    movd     [r0 + r1],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movd     m0,    [r2]
-    movd     m1,    [r3]
-    movd     m2,    [r2 + r4]
-    movd     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movd     [r0],           m0
-    movd     [r0 + r1],      m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_2x4, 6, 6, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-pmovzxbw    m0,            [r2]
-movd        m1,            [r3]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-pextrw      [r0],          m0,      0
-
-pmovzxbw    m0,            [r2 + r4]
-movd        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-pextrw      [r0 + r1],     m0,     0
-
-pmovzxbw    m0,            [r2 + 2 * r4]
-movd        m1,            [r3 + 2 * r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
+cglobal pixel_add_ps_4x4, 6, 6, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m1,     [pw_pixel_max]
+    pxor    m0,     m0
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
+    movh    m2,     [r2]
+    movhps  m2,     [r2 + r4]
+    movh    m3,     [r3]
+    movhps  m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+    movh    m4,     [r2]
+    movhps  m4,     [r2 + r4]
+    movh    m5,     [r3]
+    movhps  m5,     [r3 + r5]
+
+    paddw   m2,     m3
+    paddw   m4,     m5
+    CLIPW2  m2, m4, m0, m1
+
+    movh    [r0],       m2
+    movhps  [r0 + r1],  m2
+    lea     r0,     [r0 + r1 * 2]
+    movh    [r0],       m4
+    movhps  [r0 + r1],  m4
 
-pextrw      [r0 + 2 * r1], m0,     0
-
-lea         r0,            [r0 + 2 * r1]
-lea         r2,            [r2 + 2 * r4]
-lea         r3,            [r3 + 2 * r5]
-
-pmovzxbw    m0,            [r2 + r4]
-movd        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-pextrw      [r0 + r1],     m0,     0
-%endif
-RET
-
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%macro PIXEL_ADD_PS_W2_H4 2
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mov      r6d,   %2/4
-    mova     m5,    [pw_pixel_max]
-.loop:
-    movd     m0,    [r2]
-    movd     m1,    [r3]
-    movd     m2,    [r2 + r4]
-    movd     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movd     [r0],           m0
-    movd     [r0 + r1],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movd     m0,    [r2]
-    movd     m1,    [r3]
-    movd     m2,    [r2 + r4]
-    movd     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movd     [r0],           m0
-    movd     [r0 + r1],      m2
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
-.loop:
-      pmovzxbw    m0,            [r2]
-      movd        m1,            [r3]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      pextrw      [r0],          m0,      0
-
-      pmovzxbw    m0,            [r2 + r4]
-      movd        m1,            [r3 + r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      pextrw      [r0 + r1],     m0,      0
+cglobal pixel_add_ps_4x4, 6, 6, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    add         r5,         r5
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m2,         [r2 + r4]
+    movh        m1,         [r3]
+    movh        m3,         [r3 + r5]
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+    pmovzxbw    m4,         [r2]
+    pmovzxbw    m6,         [r2 + r4]
+    movh        m5,         [r3]
+    movh        m7,         [r3 + r5]
+
+    paddw       m0,         m1
+    paddw       m2,         m3
+    paddw       m4,         m5
+    paddw       m6,         m7
+    packuswb    m0,         m0
+    packuswb    m2,         m2
+    packuswb    m4,         m4
+    packuswb    m6,         m6
+
+    movd        [r0],       m0
+    movd        [r0 + r1],  m2
+    lea         r0,         [r0 + r1 * 2]
+    movd        [r0],       m4
+    movd        [r0 + r1],  m6
 
-      pmovzxbw    m0,            [r2 + 2 * r4]
-      movd        m1,            [r3 + 2 * r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      pextrw      [r0 + 2 * r1], m0,      0
-
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      pmovzxbw    m0,            [r2 + r4]
-      movd        m1,            [r3 + r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      pextrw      [r0 + r1],     m0,      0
+    RET
 %endif
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
-%endmacro
 
-PIXEL_ADD_PS_W2_H4   2, 8
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_4x2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_4x2, 6, 6, 4
-    add      r1, r1
-    add      r4, r4
-    add      r5, r5
-    pxor     m0, m0
-    mova     m1, [pw_pixel_max]
-
-    movh     m2, [r2]
-    movhps   m2, [r2 + r4]
-
-    movh     m3, [r3]
-    movhps   m3, [r3 + r5]
-
-    paddw    m2, m3
-    CLIPW    m2, m0, m1
-
-    movh     [r0], m2
-    movhps   [r0 + r1], m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_4x2, 6, 6, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-pmovzxbw    m0,            [r2]
-movh        m1,            [r3]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movd        [r0],          m0
-
-pmovzxbw    m0,            [r2 + r4]
-movh        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movd        [r0 + r1],     m0
-%endif
-RET
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_4x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %macro PIXEL_ADD_PS_W4_H4 2
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 4
-    mov      r6d, %2/4
-    add      r1, r1
-    add      r4, r4
-    add      r5, r5
-    pxor     m0, m0
-    mova     m1, [pw_pixel_max]
-.loop:
-    movh     m2, [r2]
-    movhps   m2, [r2 + r4]
-
-    movh     m3, [r3]
-    movhps   m3, [r3 + r5]
-
-    paddw    m2, m3
-    CLIPW    m2, m0, m1
-
-    movlps   [r0], m2
-    movhps   [r0 + r1], m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movh     m2, [r2]
-    movhps   m2, [r2 + r4]
-
-    movh     m3, [r3]
-    movhps   m3, [r3 + r5]
-
-    paddw    m2, m3
-    CLIPW    m2, m0, m1
-
-    movh     [r0], m2
-    movhps   [r0 + r1], m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
-.loop:
-
-    pmovzxbw    m0,            [r2]
-    movh        m1,            [r3]
-
-    paddw       m0,            m1
-    packuswb    m0,            m0
-
-    movd        [r0],          m0
-
-    pmovzxbw    m0,            [r2 + r4]
-    movh        m1,            [r3 + r5]
-
-    paddw       m0,            m1
-    packuswb    m0,            m0
-
-    movd        [r0 + r1],     m0
-
-    pmovzxbw    m0,            [r2 + 2 * r4]
-    movh        m1,            [r3 + 2 * r5]
-
-    paddw       m0,            m1
-    packuswb    m0,            m0
-
-    movd        [r0 + 2 * r1], m0
-
-    lea         r0,            [r0 + 2 * r1]
-    lea         r2,            [r2 + 2 * r4]
-    lea         r3,            [r3 + 2 * r5]
-
-    pmovzxbw    m0,            [r2 + r4]
-    movh        m1,            [r3 + r5]
-
-    paddw       m0,            m1
-    packuswb    m0,            m0
-
-    movd        [r0 + r1],     m0
-%endif
-    dec         r6d
-    lea         r0,            [r0 + 2 * r1]
-    lea         r2,            [r2 + 2 * r4]
-    lea         r3,            [r3 + 2 * r5]
-    jnz         .loop
-
-RET
-%endmacro
-
-PIXEL_ADD_PS_W4_H4   4,  4
-PIXEL_ADD_PS_W4_H4   4,  8
-PIXEL_ADD_PS_W4_H4   4, 16
-
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%macro PIXEL_ADD_PS_W6_H4 2
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+cglobal pixel_add_ps_4x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m1,     [pw_pixel_max]
+    pxor    m0,     m0
     mov     r6d,    %2/4
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu        m0,    [r2]
-    movu        m1,    [r3]
-    movu        m2,    [r2 + r4]
-    movu        m3,    [r3 + r5]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movh        [r0],             m0
-    pshufd      m1,    m0,        2
-    movd        [r0 + 8],         m1
-    movh        [r0 + r1],        m2
-    pshufd      m3,    m2,        2
-    movd        [r0 + r1 + 8],    m3
-
-    lea         r2,    [r2 + 2 * r4]
-    lea         r3,    [r3 + 2 * r5]
-    lea         r0,    [r0 + 2 * r1]
-    movu        m0,    [r2]
-    movu        m1,    [r3]
-    movu        m2,    [r2 + r4]
-    movu        m3,    [r3 + r5]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movh        [r0],             m0
-    pshufd      m1,    m0,        2
-    movd        [r0 + 8],         m1
-    movh        [r0 + r1],        m2
-    pshufd      m3,    m2,        2
-    movd        [r0 + r1 + 8],    m3
+    movh    m2,     [r2]
+    movhps  m2,     [r2 + r4]
+    movh    m3,     [r3]
+    movhps  m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+    movh    m4,     [r2]
+    movhps  m4,     [r2 + r4]
+    movh    m5,     [r3]
+    movhps  m5,     [r3 + r5]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m2,     m3
+    paddw   m4,     m5
+    CLIPW2  m2, m4, m0, m1
+
+    movh    [r0],       m2
+    movhps  [r0 + r1],  m2
+    lea     r0,     [r0 + r1 * 2]
+    movh    [r0],       m4
+    movhps  [r0 + r1],  m4
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
+cglobal pixel_add_ps_4x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    mov         r6d,        %2/4
+    add         r5,         r5
 .loop:
-      pmovzxbw    m0,                [r2]
-      movu        m1,                [r3]
-
-      paddw       m0,                m1
-      packuswb    m0,                m0
-
-      movd        [r0],              m0
-      pextrw      [r0 + 4],          m0,        2
-
-      pmovzxbw    m0,                [r2 + r4]
-      movu        m1,                [r3 + r5]
-
-      paddw       m0,                m1
-      packuswb    m0,                m0
-
-      movd        [r0 + r1],         m0
-      pextrw      [r0 + r1 + 4],     m0,        2
-
-      pmovzxbw    m0,                [r2 + 2 * r4]
-      movu        m1,                [r3 + 2 * r5]
-
-      paddw       m0,                m1
-      packuswb    m0,                m0
-
-      movd        [r0 + 2 * r1],     m0
-      pextrw      [r0 + 2 * r1 + 4], m0,        2
-
-      lea         r0,                [r0 + 2 * r1]
-      lea         r2,                [r2 + 2 * r4]
-      lea         r3,                [r3 + 2 * r5]
-
-      pmovzxbw    m0,                [r2 + r4]
-      movu        m1,                [r3 + r5]
-
-      paddw       m0,                m1
-      packuswb    m0,                m0
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m2,         [r2 + r4]
+    movh        m1,         [r3]
+    movh        m3,         [r3 + r5]
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+    pmovzxbw    m4,         [r2]
+    pmovzxbw    m6,         [r2 + r4]
+    movh        m5,         [r3]
+    movh        m7,         [r3 + r5]
+    dec         r6d
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+
+    paddw       m0,         m1
+    paddw       m2,         m3
+    paddw       m4,         m5
+    paddw       m6,         m7
+    packuswb    m0,         m0
+    packuswb    m2,         m2
+    packuswb    m4,         m4
+    packuswb    m6,         m6
+
+    movd        [r0],       m0
+    movd        [r0 + r1],  m2
+    lea         r0,         [r0 + r1 * 2]
+    movd        [r0],       m4
+    movd        [r0 + r1],  m6
+    lea         r0,         [r0 + r1 * 2]
 
-      movd        [r0 + r1],         m0
-      pextrw      [r0 + r1 + 4],     m0,        2
+    jnz         .loop
+    RET
 %endif
-      lea         r0,                [r0 + 2 * r1]
-      lea         r2,                [r2 + 2 * r4]
-      lea         r3,                [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
 %endmacro
 
-PIXEL_ADD_PS_W6_H4 6,  8
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_8x2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_8x2, 6, 6, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_8x2, 6, 6, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-pmovzxbw    m0,            [r2]
-movu        m1,            [r3]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0],          m0
-
-pmovzxbw    m0,            [r2 + r4]
-movu        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0 + r1],     m0
-%endif
-RET
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_8x6(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_8x6, 6, 6, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_8x6, 6, 6, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-pmovzxbw    m0,            [r2]
-movu        m1,            [r3]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0],          m0
-
-pmovzxbw    m0,            [r2 + r4]
-movu        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0 + r1],     m0
-
-pmovzxbw    m0,            [r2 + 2 * r4]
-movu        m1,            [r3 + 2 * r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0 + 2 * r1], m0
-
-lea         r0,            [r0 + 2 * r1]
-lea         r2,            [r2 + 2 * r4]
-lea         r3,            [r3 + 2 * r5]
-
-pmovzxbw    m0,            [r2 + r4]
-movu        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0 + r1],     m0
-
-pmovzxbw    m0,            [r2 + 2 * r4]
-movu        m1,            [r3 + 2 * r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
-
-movh        [r0 + 2 * r1], m0
-
-lea         r0,            [r0 + 2 * r1]
-lea         r2,            [r2 + 2 * r4]
-lea         r3,            [r3 + 2 * r5]
-
-pmovzxbw    m0,            [r2 + r4]
-movu        m1,            [r3 + r5]
-
-paddw       m0,            m1
-packuswb    m0,            m0
+PIXEL_ADD_PS_W4_H4   4,  8
 
-movh        [r0 + r1],     m0
-%endif
-RET
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_8x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %macro PIXEL_ADD_PS_W8_H4 2
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+cglobal pixel_add_ps_8x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m5,     [pw_pixel_max]
+    pxor    m4,     m4
     mov     r6d,    %2/4
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + r1],      m2
+    movu    m0,     [r2]
+    movu    m2,     [r2 + r4]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + r5]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + r1],  m2
+
+    movu    m0,     [r2]
+    movu    m2,     [r2 + r4]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + r5]
+    dec     r6d
+    lea     r0,     [r0 + r1 * 2]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + r1],  m2
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 2, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
+cglobal pixel_add_ps_8x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    mov         r6d,        %2/4
+    add         r5,         r5
 .loop:
-      pmovzxbw    m0,            [r2]
-      movu        m1,            [r3]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      movh        [r0],          m0
-
-      pmovzxbw    m0,            [r2 + r4]
-      movu        m1,            [r3 + r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      movh        [r0 + r1],     m0
-
-      pmovzxbw    m0,            [r2 + 2 * r4]
-      movu        m1,            [r3 + 2 * r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
-
-      movh        [r0 + 2 * r1], m0
-
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      pmovzxbw    m0,            [r2 + r4]
-      movu        m1,            [r3 + r5]
-
-      paddw       m0,            m1
-      packuswb    m0,            m0
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m2,         [r2 + r4]
+    movu        m1,         [r3]
+    movu        m3,         [r3 + r5]
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+    pmovzxbw    m4,         [r2]
+    pmovzxbw    m6,         [r2 + r4]
+    movu        m5,         [r3]
+    movu        m7,         [r3 + r5]
+    dec         r6d
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+
+    paddw       m0,         m1
+    paddw       m2,         m3
+    paddw       m4,         m5
+    paddw       m6,         m7
+    packuswb    m0,         m0
+    packuswb    m2,         m2
+    packuswb    m4,         m4
+    packuswb    m6,         m6
+
+    movh        [r0],       m0
+    movh        [r0 + r1],  m2
+    lea         r0,         [r0 + r1 * 2]
+    movh        [r0],       m4
+    movh        [r0 + r1],  m6
+    lea         r0,         [r0 + r1 * 2]
 
-      movh        [r0 + r1],     m0
+    jnz         .loop
+    RET
 %endif
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
 %endmacro
 
-PIXEL_ADD_PS_W8_H4 8,  4
 PIXEL_ADD_PS_W8_H4 8,  8
 PIXEL_ADD_PS_W8_H4 8, 16
-PIXEL_ADD_PS_W8_H4 8, 32
-
-
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%macro PIXEL_ADD_PS_W12_H4 2
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    mov     r6d,    %2/4
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-.loop:
-    movu        m0,    [r2]
-    movu        m1,    [r3]
-    movh        m2,    [r2 + 16]
-    movh        m3,    [r3 + 16]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movu        [r0],         m0
-    movh        [r0 + 16],     m2
-
-    movu        m0,    [r2 + r4]
-    movu        m1,    [r3 + r5]
-    movh        m2,    [r2 + r4 + 16]
-    movh        m3,    [r3 + r5 + 16]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movu        [r0 + r1],        m0
-    movh        [r0 + r1 + 16],    m2
-
-    lea         r2,    [r2 + 2 * r4]
-    lea         r3,    [r3 + 2 * r5]
-    lea         r0,    [r0 + 2 * r1]
-    movu        m0,    [r2]
-    movu        m1,    [r3]
-    movh        m2,    [r2 + 16]
-    movh        m3,    [r3 + 16]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movu        [r0],         m0
-    movh        [r0 + 16],     m2
-
-    movu        m0,    [r2 + r4]
-    movu        m1,    [r3 + r5]
-    movh        m2,    [r2 + r4 + 16]
-    movh        m3,    [r3 + r5 + 16]
-    paddw       m0,    m1
-    paddw       m2,    m3
-    CLIPW       m0,    m4,    m5
-    CLIPW       m2,    m4,    m5
-
-    movu        [r0 + r1],        m0
-    movh        [r0 + r1 + 16],    m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 4, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
-.loop:
-      pmovzxbw    m0,                [r2]
-      pmovzxbw    m1,                [r2 + 8]
-
-      movu        m2,                [r3]
-      movh        m3,                [r3 + 16]
-
-      paddw       m0,                m2
-      paddw       m1,                m3
-
-      packuswb    m0,                m1
-
-      movh        [r0],              m0
-      movhlps     m0,                m0
-      movd        [r0 + 8],          m0
-
-      pmovzxbw    m0,                [r2 + r4]
-      pmovzxbw    m1,                [r2 + r4 + 8]
-
-      movu        m2,                [r3 + r5]
-      movh        m3,                [r3 + r5 + 16]
-
-      paddw       m0,                m2
-      paddw       m1,                m3
-
-      packuswb    m0,                m1
-
-      movh        [r0 + r1],         m0
-      movhlps     m0,                m0
-      movd        [r0 + r1 + 8],     m0
-
-      pmovzxbw    m0,                [r2 + 2 * r4]
-      pmovzxbw    m1,                [r2 + 2 * r4 + 8]
-
-      movu        m2,                [r3 + 2 * r5]
-      movh        m3,                [r3 + 2 * r5 + 16]
-
-      paddw       m0,                m2
-      paddw       m1,                m3
-
-      packuswb    m0,                m1
 
-      movh        [r0 + 2 * r1],     m0
-      movhlps     m0,                m0
-      movd        [r0 + 2 * r1 + 8], m0
-
-      lea         r0,                [r0 + 2 * r1]
-      lea         r2,                [r2 + 2 * r4]
-      lea         r3,                [r3 + 2 * r5]
-
-      pmovzxbw    m0,                [r2 + r4]
-      pmovzxbw    m1,                [r2 + r4 + 8]
-
-      movu        m2,                [r3 + r5]
-      movh        m3,                [r3 + r5 + 16]
-
-      paddw       m0,                m2
-      paddw       m1,                m3
-
-      packuswb    m0,                m1
-
-      movh        [r0 + r1],         m0
-      movhlps     m0,                m0
-      movd        [r0 + r1 + 8],     m0
-%endif
-      lea         r0,                [r0 + 2 * r1]
-      lea         r2,                [r2 + 2 * r4]
-      lea         r3,                [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
-%endmacro
-
-PIXEL_ADD_PS_W12_H4 12, 16
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_16x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %macro PIXEL_ADD_PS_W16_H4 2
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+cglobal pixel_add_ps_16x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m5,     [pw_pixel_max]
+    pxor    m4,     m4
     mov     r6d,    %2/4
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + 16],      m2
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1],           m0
-    movu     [r0 + r1 + 16],      m2
-
-    lea      r2,    [r2 + 2 * r4]
-    lea      r3,    [r3 + 2 * r5]
-    lea      r0,    [r0 + 2 * r1]
-
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],           m0
-    movu     [r0 + 16],      m2
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1],           m0
-    movu     [r0 + r1 + 16],      m2
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+    lea     r0,     [r0 + r1 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/4
-
+cglobal pixel_add_ps_16x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    mov         r6d,        %2/4
+    add         r5,         r5
 .loop:
-      pmovzxbw    m0,            [r2]
-      pmovzxbw    m1,            [r2 + 8]
-
-      movu        m2,            [r3]
-      movu        m3,            [r3 + 16]
-
-      paddw       m0,            m2
-      paddw       m1,            m3
-
-      packuswb    m0,            m1
-
-      movu        [r0],          m0
-
-      pmovzxbw    m0,            [r2 + r4]
-      pmovzxbw    m1,            [r2 + r4 + 8]
-
-      movu        m2,            [r3 + r5]
-      movu        m3,            [r3 + r5 + 16]
-
-      paddw       m0,            m2
-      paddw       m1,            m3
-
-      packuswb    m0,            m1
-
-      movu        [r0 + r1],     m0
-
-      pmovzxbw    m0,            [r2 + 2 * r4]
-      pmovzxbw    m1,            [r2 + 2 * r4 + 8]
-
-      movu        m2,            [r3 + 2 * r5]
-      movu        m3,            [r3 + 2 * r5 + 16]
-
-      paddw       m0,            m2
-      paddw       m1,            m3
-
-      packuswb    m0,            m1
-
-      movu        [r0 + 2 * r1], m0
-
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      pmovzxbw    m0,            [r2 + r4]
-      pmovzxbw    m1,            [r2 + r4 + 8]
-
-      movu        m2,            [r3 + r5]
-      movu        m3,            [r3 + r5 + 16]
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m1,         [r2 + 8]
+    pmovzxbw    m4,         [r2 + r4]
+    pmovzxbw    m5,         [r2 + r4 + 8]
+    movu        m2,         [r3]
+    movu        m3,         [r3 + 16]
+    movu        m6,         [r3 + r5]
+    movu        m7,         [r3 + r5 + 16]
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
+
+    paddw       m0,         m2
+    paddw       m1,         m3
+    paddw       m4,         m6
+    paddw       m5,         m7
+    packuswb    m0,         m1
+    packuswb    m4,         m5
+
+    movu        [r0],       m0
+    movu        [r0 + r1],  m4
+
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m1,         [r2 + 8]
+    pmovzxbw    m4,         [r2 + r4]
+    pmovzxbw    m5,         [r2 + r4 + 8]
+    movu        m2,         [r3]
+    movu        m3,         [r3 + 16]
+    movu        m6,         [r3 + r5]
+    movu        m7,         [r3 + r5 + 16]
+    dec         r6d
+    lea         r0,         [r0 + r1 * 2]
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
 
-      paddw       m0,            m2
-      paddw       m1,            m3
+    paddw       m0,         m2
+    paddw       m1,         m3
+    paddw       m4,         m6
+    paddw       m5,         m7
+    packuswb    m0,         m1
+    packuswb    m4,         m5
 
-      packuswb    m0,            m1
+    movu        [r0],       m0
+    movu        [r0 + r1],  m4
+    lea         r0,         [r0 + r1 * 2]
 
-      movu        [r0 + r1],     m0
+    jnz         .loop
+    RET
 %endif
-      lea         r0,            [r0 + 2 * r1]
-      lea         r2,            [r2 + 2 * r4]
-      lea         r3,            [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
 %endmacro
 
-PIXEL_ADD_PS_W16_H4 16,  4
-PIXEL_ADD_PS_W16_H4 16,  8
-PIXEL_ADD_PS_W16_H4 16, 12
 PIXEL_ADD_PS_W16_H4 16, 16
 PIXEL_ADD_PS_W16_H4 16, 32
-PIXEL_ADD_PS_W16_H4 16, 64
 
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%macro PIXEL_ADD_PS_W24_H2 2
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    mov     r6d,    %2/2
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],         m0
-    movu     [r0 + 16],    m2
-
-    movu     m0,    [r2 + 32]
-    movu     m1,    [r3 + 32]
-    movu     m2,    [r2 + r4]
-    movu     m3,    [r3 + r5]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 32],    m0
-    movu     [r0 + r1],    m2
-
-    movu     m0,    [r2 + r4 + 16]
-    movu     m1,    [r3 + r5 + 16]
-    movu     m2,    [r2 + r4 + 32]
-    movu     m3,    [r3 + r5 + 32]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 16],    m0
-    movu     [r0 + r1 + 32],    m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-
-    add         r5,            r5
-    mov         r6d,           %2/2
-
-.loop:
-    pmovzxbw    m0,             [r2]
-    pmovzxbw    m1,             [r2 + 8]
-    pmovzxbw    m2,             [r2 + 16]
-
-    movu        m3,             [r3]
-    movu        m4,             [r3 + 16]
-    movu        m5,             [r3 + 32]
-
-    paddw       m0,             m3
-    paddw       m1,             m4
-    paddw       m2,             m5
-
-    packuswb    m0,             m1
-    packuswb    m2,             m2
-
-    movu        [r0],           m0
-    movh        [r0 + 16],      m2
-
-    pmovzxbw    m0,             [r2 + r4]
-    pmovzxbw    m1,             [r2 + r4 + 8]
-    pmovzxbw    m2,             [r2 + r4 + 16]
-
-    movu        m3,             [r3 + r5]
-    movu        m4,             [r3 + r5 + 16]
-    movu        m5,             [r3 + r5 + 32]
-
-    paddw       m0,             m3
-    paddw       m1,             m4
-    paddw       m2,             m5
-
-    packuswb    m0,             m1
-    packuswb    m2,             m2
-
-    movu        [r0 + r1],      m0
-    movh        [r0 + r1 + 16], m2
-%endif
-    lea         r0,             [r0 + 2 * r1]
-    lea         r2,             [r2 + 2 * r4]
-    lea         r3,             [r3 + 2 * r5]
-
-    dec         r6d
-    jnz         .loop
-
-    RET
-%endmacro
-
-PIXEL_ADD_PS_W24_H2 24, 32
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_32x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %macro PIXEL_ADD_PS_W32_H2 2
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+cglobal pixel_add_ps_32x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m5,     [pw_pixel_max]
+    pxor    m4,     m4
     mov     r6d,    %2/2
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],         m0
-    movu     [r0 + 16],    m2
-
-    movu     m0,    [r2 + 32]
-    movu     m1,    [r3 + 32]
-    movu     m2,    [r2 + 48]
-    movu     m3,    [r3 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 32],    m0
-    movu     [r0 + 48],    m2
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1],         m0
-    movu     [r0 + r1 + 16],    m2
-
-    movu     m0,    [r2 + r4 + 32]
-    movu     m1,    [r3 + r5 + 32]
-    movu     m2,    [r2 + r4 + 48]
-    movu     m3,    [r3 + r5 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 32],    m0
-    movu     [r0 + r1 + 48],    m2
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+
+    movu    m0,     [r2 + 32]
+    movu    m2,     [r2 + 48]
+    movu    m1,     [r3 + 32]
+    movu    m3,     [r3 + 48]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + 32],  m0
+    movu    [r0 + 48],  m2
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+
+    movu    m0,     [r2 + r4 + 32]
+    movu    m2,     [r2 + r4 + 48]
+    movu    m1,     [r3 + r5 + 32]
+    movu    m3,     [r3 + r5 + 48]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1 + 32], m0
+    movu    [r0 + r1 + 48], m2
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
-
-    add         r5,            r5
-    mov         r6d,           %2/2
-
+cglobal pixel_add_ps_32x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    mov         r6d,        %2/2
+    add         r5,         r5
 .loop:
-    pmovzxbw    m0,             [r2]
-    pmovzxbw    m1,             [r2 + 8]
-    pmovzxbw    m2,             [r2 + 16]
-    pmovzxbw    m3,             [r2 + 24]
-
-    movu        m4,             [r3]
-    movu        m5,             [r3 + 16]
-    movu        m6,             [r3 + 32]
-    movu        m7,             [r3 + 48]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
-
-    packuswb    m0,             m1
-    packuswb    m2,             m3
-
-    movu        [r0],           m0
-    movu        [r0 + 16],      m2
-
-    pmovzxbw    m0,             [r2 + r4]
-    pmovzxbw    m1,             [r2 + r4 + 8]
-    pmovzxbw    m2,             [r2 + r4 + 16]
-    pmovzxbw    m3,             [r2 + r4 + 24]
-
-    movu        m4,             [r3 + r5]
-    movu        m5,             [r3 + r5 + 16]
-    movu        m6,             [r3 + r5 + 32]
-    movu        m7,             [r3 + r5 + 48]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m1,         [r2 + 8]
+    pmovzxbw    m2,         [r2 + 16]
+    pmovzxbw    m3,         [r2 + 24]
+    movu        m4,         [r3]
+    movu        m5,         [r3 + 16]
+    movu        m6,         [r3 + 32]
+    movu        m7,         [r3 + 48]
+
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
+
+    movu        [r0],       m0
+    movu        [r0 + 16],  m2
+
+    pmovzxbw    m0,         [r2 + r4]
+    pmovzxbw    m1,         [r2 + r4 + 8]
+    pmovzxbw    m2,         [r2 + r4 + 16]
+    pmovzxbw    m3,         [r2 + r4 + 24]
+    movu        m4,         [r3 + r5]
+    movu        m5,         [r3 + r5 + 16]
+    movu        m6,         [r3 + r5 + 32]
+    movu        m7,         [r3 + r5 + 48]
+    dec         r6d
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
 
-    packuswb    m0,             m1
-    packuswb    m2,             m3
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
 
     movu        [r0 + r1],      m0
     movu        [r0 + r1 + 16], m2
-%endif
-    lea         r0,             [r0 + 2 * r1]
-    lea         r2,             [r2 + 2 * r4]
-    lea         r3,             [r3 + 2 * r5]
-
-    dec         r6d
-    jnz        .loop
+    lea         r0,         [r0 + r1 * 2]
 
+    jnz         .loop
     RET
+%endif
 %endmacro
 
-PIXEL_ADD_PS_W32_H2 32,  8
-PIXEL_ADD_PS_W32_H2 32, 16
-PIXEL_ADD_PS_W32_H2 32, 24
 PIXEL_ADD_PS_W32_H2 32, 32
 PIXEL_ADD_PS_W32_H2 32, 64
 
-;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
-;-----------------------------------------------------------------------------
-%macro PIXEL_ADD_PS_W48_H2 2
-%if HIGH_BIT_DEPTH
-INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
-    mov     r6d,    %2/2
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
-.loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],         m0
-    movu     [r0 + 16],    m2
-
-    movu     m0,    [r2 + 32]
-    movu     m1,    [r3 + 32]
-    movu     m2,    [r2 + 48]
-    movu     m3,    [r3 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 32],    m0
-    movu     [r0 + 48],    m2
-
-    movu     m0,    [r2 + 64]
-    movu     m1,    [r3 + 64]
-    movu     m2,    [r2 + 80]
-    movu     m3,    [r3 + 80]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 64],    m0
-    movu     [r0 + 80],    m2
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1],         m0
-    movu     [r0 + r1 + 16],    m2
-
-    movu     m0,    [r2 + r4 + 32]
-    movu     m1,    [r3 + r5 + 32]
-    movu     m2,    [r2 + r4 + 48]
-    movu     m3,    [r3 + r5 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 32],    m0
-    movu     [r0 + r1 + 48],    m2
-
-    movu     m0,    [r2 + r4 + 64]
-    movu     m1,    [r3 + r5 + 64]
-    movu     m2,    [r2 + r4 + 80]
-    movu     m3,    [r3 + r5 + 80]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 64],    m0
-    movu     [r0 + r1 + 80],    m2
-%else
-INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
-
-add         r5,            r5
-
-mov         r6d,           %2/2
-
-.loop:
-      pmovzxbw    m0,             [r2]
-      pmovzxbw    m1,             [r2 + 8]
-      pmovzxbw    m2,             [r2 + 16]
-      pmovzxbw    m3,             [r2 + 24]
-
-      movu        m4,             [r3]
-      movu        m5,             [r3 + 16]
-      movu        m6,             [r3 + 32]
-      movu        m7,             [r3 + 48]
-
-      paddw       m0,             m4
-      paddw       m1,             m5
-      paddw       m2,             m6
-      paddw       m3,             m7
-
-      packuswb    m0,             m1
-      packuswb    m2,             m3
-
-      movu        [r0],           m0
-      movu        [r0 + 16],      m2
-
-      pmovzxbw    m0,             [r2 + 32]
-      pmovzxbw    m1,             [r2 + 40]
-
-      movu        m2,             [r3 + 64]
-      movu        m3,             [r3 + 80]
-
-      paddw       m0,             m2
-      paddw       m1,             m3
-
-      packuswb    m0,             m1
-
-      movu        [r0 + 32],      m0
-
-      pmovzxbw    m0,             [r2 + r4]
-      pmovzxbw    m1,             [r2 + r4 + 8]
-      pmovzxbw    m2,             [r2 + r4 + 16]
-      pmovzxbw    m3,             [r2 + r4 + 24]
-
-      movu        m4,             [r3 + r5]
-      movu        m5,             [r3 + r5 + 16]
-      movu        m6,             [r3 + r5 + 32]
-      movu        m7,             [r3 + r5 + 48]
-
-      paddw       m0,             m4
-      paddw       m1,             m5
-      paddw       m2,             m6
-      paddw       m3,             m7
-
-      packuswb    m0,             m1
-      packuswb    m2,             m3
-
-      movu        [r0 + r1],      m0
-      movu        [r0 + r1 + 16], m2
-
-      pmovzxbw    m0,             [r2 + r4 + 32]
-      pmovzxbw    m1,             [r2 + r4 + 40]
-
-      movu        m2,             [r3 + r5 + 64]
-      movu        m3,             [r3 + r5 + 80]
-
-      paddw       m0,             m2
-      paddw       m1,             m3
-
-      packuswb    m0,             m1
-
-      movu        [r0 + r1 + 32], m0
-%endif
-      lea         r0,             [r0 + 2 * r1]
-      lea         r2,             [r2 + 2 * r4]
-      lea         r3,             [r3 + 2 * r5]
-
-      dec         r6d
-      jnz         .loop
-
-RET
-%endmacro
-
-PIXEL_ADD_PS_W48_H2 48, 64
 
 ;-----------------------------------------------------------------------------
-; void pixel_add_ps_%1x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
+; void pixel_add_ps_64x%2(pixel *dest, intptr_t destride, pixel *src0, int16_t *scr1, intptr_t srcStride0, intptr_t srcStride1)
 ;-----------------------------------------------------------------------------
 %macro PIXEL_ADD_PS_W64_H2 2
 %if HIGH_BIT_DEPTH
 INIT_XMM sse2
-cglobal pixel_add_ps_%1x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+cglobal pixel_add_ps_64x%2, 6, 7, 6, dest, destride, src0, scr1, srcStride0, srcStride1
+    mova    m5,     [pw_pixel_max]
+    pxor    m4,     m4
     mov     r6d,    %2/2
-    add      r1,    r1
-    add      r4,    r4
-    add      r5,    r5
-    pxor     m4,    m4
-    mova     m5,    [pw_pixel_max]
+    add     r4,     r4
+    add     r5,     r5
+    add     r1,     r1
 .loop:
-    movu     m0,    [r2]
-    movu     m1,    [r3]
-    movu     m2,    [r2 + 16]
-    movu     m3,    [r3 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0],         m0
-    movu     [r0 + 16],    m2
-
-    movu     m0,    [r2 + 32]
-    movu     m1,    [r3 + 32]
-    movu     m2,    [r2 + 48]
-    movu     m3,    [r3 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 32],    m0
-    movu     [r0 + 48],    m2
-
-    movu     m0,    [r2 + 64]
-    movu     m1,    [r3 + 64]
-    movu     m2,    [r2 + 80]
-    movu     m3,    [r3 + 80]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 64],    m0
-    movu     [r0 + 80],    m2
-
-    movu     m0,    [r2 + 96]
-    movu     m1,    [r3 + 96]
-    movu     m2,    [r2 + 112]
-    movu     m3,    [r3 + 112]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + 96],     m0
-    movu     [r0 + 112],    m2
-
-    movu     m0,    [r2 + r4]
-    movu     m1,    [r3 + r5]
-    movu     m2,    [r2 + r4 + 16]
-    movu     m3,    [r3 + r5 + 16]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1],         m0
-    movu     [r0 + r1 + 16],    m2
-
-    movu     m0,    [r2 + r4 + 32]
-    movu     m1,    [r3 + r5 + 32]
-    movu     m2,    [r2 + r4 + 48]
-    movu     m3,    [r3 + r5 + 48]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 32],    m0
-    movu     [r0 + r1 + 48],    m2
-
-    movu     m0,    [r2 + r4 + 64]
-    movu     m1,    [r3 + r5 + 64]
-    movu     m2,    [r2 + r4 + 80]
-    movu     m3,    [r3 + r5 + 80]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 64],    m0
-    movu     [r0 + r1 + 80],    m2
-
-    movu     m0,    [r2 + r4 + 96]
-    movu     m1,    [r3 + r5 + 96]
-    movu     m2,    [r2 + r4 + 112]
-    movu     m3,    [r3 + r5 + 112]
-    paddw    m0,    m1
-    paddw    m2,    m3
-    CLIPW    m0,    m4,    m5
-    CLIPW    m2,    m4,    m5
-
-    movu     [r0 + r1 + 96],     m0
-    movu     [r0 + r1 + 112],    m2
+    movu    m0,     [r2]
+    movu    m2,     [r2 + 16]
+    movu    m1,     [r3]
+    movu    m3,     [r3 + 16]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0],       m0
+    movu    [r0 + 16],  m2
+
+    movu    m0,     [r2 + 32]
+    movu    m2,     [r2 + 48]
+    movu    m1,     [r3 + 32]
+    movu    m3,     [r3 + 48]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + 32],  m0
+    movu    [r0 + 48],  m2
+
+    movu    m0,     [r2 + 64]
+    movu    m2,     [r2 + 80]
+    movu    m1,     [r3 + 64]
+    movu    m3,     [r3 + 80]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + 64],  m0
+    movu    [r0 + 80],  m2
+
+    movu    m0,     [r2 + 96]
+    movu    m2,     [r2 + 112]
+    movu    m1,     [r3 + 96]
+    movu    m3,     [r3 + 112]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + 96],  m0
+    movu    [r0 + 112], m2
+
+    movu    m0,     [r2 + r4]
+    movu    m2,     [r2 + r4 + 16]
+    movu    m1,     [r3 + r5]
+    movu    m3,     [r3 + r5 + 16]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1],      m0
+    movu    [r0 + r1 + 16], m2
+
+    movu    m0,     [r2 + r4 + 32]
+    movu    m2,     [r2 + r4 + 48]
+    movu    m1,     [r3 + r5 + 32]
+    movu    m3,     [r3 + r5 + 48]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1 + 32], m0
+    movu    [r0 + r1 + 48], m2
+
+    movu    m0,     [r2 + r4 + 64]
+    movu    m2,     [r2 + r4 + 80]
+    movu    m1,     [r3 + r5 + 64]
+    movu    m3,     [r3 + r5 + 80]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1 + 64], m0
+    movu    [r0 + r1 + 80], m2
+
+    movu    m0,     [r2 + r4 + 96]
+    movu    m2,     [r2 + r4 + 112]
+    movu    m1,     [r3 + r5 + 96]
+    movu    m3,     [r3 + r5 + 112]
+    dec     r6d
+    lea     r2,     [r2 + r4 * 2]
+    lea     r3,     [r3 + r5 * 2]
+
+    paddw   m0,     m1
+    paddw   m2,     m3
+    CLIPW2  m0, m2, m4, m5
+
+    movu    [r0 + r1 + 96],     m0
+    movu    [r0 + r1 + 112],    m2
+    lea     r0,     [r0 + r1 * 2]
+
+    jnz     .loop
+    RET
 %else
 INIT_XMM sse4
-cglobal pixel_add_ps_%1x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
-
-    add         r5,            r5
-    mov         r6d,           %2/2
-
+cglobal pixel_add_ps_64x%2, 6, 7, 8, dest, destride, src0, scr1, srcStride0, srcStride1
+    mov         r6d,        %2/2
+    add         r5,         r5
 .loop:
-    pmovzxbw    m0,             [r2]
-    pmovzxbw    m1,             [r2 + 8]
-    pmovzxbw    m2,             [r2 + 16]
-    pmovzxbw    m3,             [r2 + 24]
-
-    movu        m4,             [r3]
-    movu        m5,             [r3 + 16]
-    movu        m6,             [r3 + 32]
-    movu        m7,             [r3 + 48]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
-
-    packuswb    m0,             m1
-    packuswb    m2,             m3
-
-    movu        [r0],           m0
-    movu        [r0 + 16],      m2
-
-    pmovzxbw    m0,             [r2 + 32]
-    pmovzxbw    m1,             [r2 + 40]
-    pmovzxbw    m2,             [r2 + 48]
-    pmovzxbw    m3,             [r2 + 56]
-
-    movu        m4,             [r3 + 64]
-    movu        m5,             [r3 + 80]
-    movu        m6,             [r3 + 96]
-    movu        m7,             [r3 + 112]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
-
-    packuswb    m0,             m1
-    packuswb    m2,             m3
-
-    movu        [r0 + 32],      m0
-    movu        [r0 + 48],      m2
-
-    pmovzxbw    m0,             [r2 + r4]
-    pmovzxbw    m1,             [r2 + r4 + 8]
-    pmovzxbw    m2,             [r2 + r4 + 16]
-    pmovzxbw    m3,             [r2 + r4 + 24]
-
-    movu        m4,             [r3 + r5]
-    movu        m5,             [r3 + r5 + 16]
-    movu        m6,             [r3 + r5 + 32]
-    movu        m7,             [r3 + r5 + 48]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
-
-    packuswb    m0,             m1
-    packuswb    m2,             m3
+    pmovzxbw    m0,         [r2]
+    pmovzxbw    m1,         [r2 + 8]
+    pmovzxbw    m2,         [r2 + 16]
+    pmovzxbw    m3,         [r2 + 24]
+    movu        m4,         [r3]
+    movu        m5,         [r3 + 16]
+    movu        m6,         [r3 + 32]
+    movu        m7,         [r3 + 48]
+
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
+
+    movu        [r0],       m0
+    movu        [r0 + 16],  m2
+
+    pmovzxbw    m0,         [r2 + 32]
+    pmovzxbw    m1,         [r2 + 40]
+    pmovzxbw    m2,         [r2 + 48]
+    pmovzxbw    m3,         [r2 + 56]
+    movu        m4,         [r3 + 64]
+    movu        m5,         [r3 + 80]
+    movu        m6,         [r3 + 96]
+    movu        m7,         [r3 + 112]
+
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
+
+    movu        [r0 + 32],  m0
+    movu        [r0 + 48],  m2
+
+    pmovzxbw    m0,         [r2 + r4]
+    pmovzxbw    m1,         [r2 + r4 + 8]
+    pmovzxbw    m2,         [r2 + r4 + 16]
+    pmovzxbw    m3,         [r2 + r4 + 24]
+    movu        m4,         [r3 + r5]
+    movu        m5,         [r3 + r5 + 16]
+    movu        m6,         [r3 + r5 + 32]
+    movu        m7,         [r3 + r5 + 48]
+
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
 
     movu        [r0 + r1],      m0
     movu        [r0 + r1 + 16], m2
 
-    pmovzxbw    m0,             [r2 + r4 + 32]
-    pmovzxbw    m1,             [r2 + r4 + 40]
-    pmovzxbw    m2,             [r2 + r4 + 48]
-    pmovzxbw    m3,             [r2 + r4 + 56]
-
-    movu        m4,             [r3 + r5 + 64]
-    movu        m5,             [r3 + r5 + 80]
-    movu        m6,             [r3 + r5 + 96]
-    movu        m7,             [r3 + r5 + 112]
-
-    paddw       m0,             m4
-    paddw       m1,             m5
-    paddw       m2,             m6
-    paddw       m3,             m7
+    pmovzxbw    m0,         [r2 + r4 + 32]
+    pmovzxbw    m1,         [r2 + r4 + 40]
+    pmovzxbw    m2,         [r2 + r4 + 48]
+    pmovzxbw    m3,         [r2 + r4 + 56]
+    movu        m4,         [r3 + r5 + 64]
+    movu        m5,         [r3 + r5 + 80]
+    movu        m6,         [r3 + r5 + 96]
+    movu        m7,         [r3 + r5 + 112]
+    dec         r6d
+    lea         r2,         [r2 + r4 * 2]
+    lea         r3,         [r3 + r5 * 2]
 
-    packuswb    m0,             m1
-    packuswb    m2,             m3
+    paddw       m0,         m4
+    paddw       m1,         m5
+    paddw       m2,         m6
+    paddw       m3,         m7
+    packuswb    m0,         m1
+    packuswb    m2,         m3
 
     movu        [r0 + r1 + 32], m0
     movu        [r0 + r1 + 48], m2
-%endif
-    lea         r0,             [r0 + 2 * r1]
-    lea         r2,             [r2 + 2 * r4]
-    lea         r3,             [r3 + 2 * r5]
+    lea         r0,         [r0 + r1 * 2]
 
-    dec         r6d
     jnz         .loop
-
     RET
+%endif
 %endmacro
 
-PIXEL_ADD_PS_W64_H2 64, 16
-PIXEL_ADD_PS_W64_H2 64, 32
-PIXEL_ADD_PS_W64_H2 64, 48
 PIXEL_ADD_PS_W64_H2 64, 64

x265-hg20140324.tar.bz2/source/common/x86/sad-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/sad-a.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/sad16-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/sad16-a.asm Changed

x265-hg20140324.tar.bz2/source/common/x86/ssd-a.asm -> x265-hg20140928.tar.bz2/source/common/x86/ssd-a.asm Changed

@@ -4,7 +4,7 @@
 ;* Copyright (C) 2003-2013 x264 project
 ;*
 ;* Authors: Loren Merritt <lorenm@u.washington.edu>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
+;*          Fiona Glaser <fiona@x264.com>
 ;*          Laurent Aimar <fenrir@via.ecp.fr>
 ;*          Alex Izvorski <aizvorksi@gmail.com>
 ;*
@@ -23,7 +23,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@x264.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************
 
 %include "x86inc.asm"
@@ -435,10 +435,14 @@
     jg .loop
 %endif
 %if %1 == 4
-    phaddd    m0, m0
+  %if notcpuflag(ssse3)
+    pshufd   m1, m0, 1
+    paddd    m0, m1
+  %else
+    phaddd   m0, m0
+  %endif
 %else
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD    m0, m1
 %endif
     movd     eax, m0
     RET
@@ -495,8 +499,7 @@
     lea       r2, [r2 + 2*r3]
     dec      r4d
     jnz .loop
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD     m0, m1
     movd     eax, m0
     RET
 %endmacro
@@ -553,8 +556,7 @@
     lea       r2, [r2 + 2*r3]
     dec      r4d
     jnz .loop
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD     m0, m1
     movd     eax, m0
     RET
 %endmacro
@@ -609,8 +611,7 @@
     lea       r2, [r2 + 2*r3]
     dec      r4d
     jnz .loop
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD     m0, m1
     movd     eax, m0
     RET
 %endmacro
@@ -687,8 +688,7 @@
     lea       r2, [r2 + 2*r3]
     dec      r4d
     jnz .loop
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD     m0, m1
     movd     eax, m0
     RET
 %endmacro
@@ -785,8 +785,7 @@
     lea     r2, [r2 + 2*r3]
     dec     r4d
     jnz .loop
-    phaddd    m0, m0
-    phaddd    m0, m0
+    HADDD     m0, m1
     movd     eax, m0
     RET
 %endmacro
@@ -2395,3 +2394,202 @@
     HADDD    m7,     m1
     movd     eax,    m7
     RET
+
+
+;-----------------------------------------------------------------------------
+; int pixel_ssd_s( int16_t *ref, intptr_t i_stride )
+;-----------------------------------------------------------------------------
+INIT_XMM sse2
+cglobal pixel_ssd_s_4, 2,2,2
+    add     r1, r1
+    movh    m0, [r0]
+    movhps  m0, [r0 + r1]
+
+    lea     r0, [r0 + r1 * 2]
+    movh    m1, [r0]
+    movhps  m1, [r0 + r1]
+
+    pmaddwd m0, m0
+    pmaddwd m1, m1
+    paddd   m0, m1
+
+    ; calculate sum and return
+    HADDD   m0, m1
+    movd    eax, m0
+    RET
+
+
+INIT_XMM sse2
+cglobal pixel_ssd_s_8, 2,3,5
+    add     r1, r1
+    lea     r2, [r1 * 3]
+    movu    m0, [r0]
+    movu    m1, [r0 + r1]
+    movu    m2, [r0 + r1 * 2]
+    movu    m3, [r0 + r2]
+
+    pmaddwd m0, m0
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    paddd   m0, m1
+    paddd   m2, m3
+    paddd   m0, m2
+
+    lea     r0, [r0 + r1 * 4]
+    movu    m4, [r0]
+    movu    m1, [r0 + r1]
+    movu    m2, [r0 + r1 * 2]
+    movu    m3, [r0 + r2]
+
+    pmaddwd m4, m4
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    paddd   m4, m1
+    paddd   m2, m3
+    paddd   m4, m2
+    paddd   m0, m4
+
+    ; calculate sum and return
+    HADDD   m0, m1
+    movd    eax, m0
+    RET
+
+
+INIT_XMM sse2
+cglobal pixel_ssd_s_16, 2,3,5
+    add     r1, r1
+
+    mov     r2d, 4
+    pxor    m0, m0
+.loop:
+    movu    m1, [r0]
+    movu    m2, [r0 + mmsize]
+    movu    m3, [r0 + r1]
+    movu    m4, [r0 + r1 + mmsize]
+    lea     r0, [r0 + r1 * 2]
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    movu    m1, [r0]
+    movu    m2, [r0 + mmsize]
+    movu    m3, [r0 + r1]
+    movu    m4, [r0 + r1 + mmsize]
+    lea     r0, [r0 + r1 * 2]
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    dec     r2d
+    jnz    .loop
+
+    ; calculate sum and return
+    HADDD   m0, m1
+    movd    eax, m0
+    RET
+
+
+INIT_XMM sse2
+cglobal pixel_ssd_s_32, 2,3,5
+    add     r1, r1
+
+    mov     r2d, 16
+    pxor    m0, m0
+.loop:
+    movu    m1, [r0 + 0 * mmsize]
+    movu    m2, [r0 + 1 * mmsize]
+    movu    m3, [r0 + 2 * mmsize]
+    movu    m4, [r0 + 3 * mmsize]
+    add     r0, r1
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    movu    m1, [r0 + 0 * mmsize]
+    movu    m2, [r0 + 1 * mmsize]
+    movu    m3, [r0 + 2 * mmsize]
+    movu    m4, [r0 + 3 * mmsize]
+    add     r0, r1
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    dec     r2d
+    jnz    .loop
+
+    ; calculate sum and return
+    HADDD   m0, m1
+    movd    eax, m0
+    RET
+
+
+INIT_YMM avx2
+cglobal pixel_ssd_s_32, 2,4,5
+    add     r1, r1
+    lea     r3, [r1 * 3]
+
+    mov     r2d, 8
+    pxor    m0, m0
+.loop:
+    movu    m1, [r0 + 0 * mmsize]
+    movu    m2, [r0 + 1 * mmsize]
+    movu    m3, [r0 + r1 + 0 * mmsize]
+    movu    m4, [r0 + r1 + 1 * mmsize]
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    movu    m1, [r0 + r1 * 2 + 0 * mmsize]
+    movu    m2, [r0 + r1 * 2 + 1 * mmsize]
+    movu    m3, [r0 + r3 + 0 * mmsize]
+    movu    m4, [r0 + r3 + 1 * mmsize]
+    lea     r0, [r0 + 4 * r1]
+
+    pmaddwd m1, m1
+    pmaddwd m2, m2
+    pmaddwd m3, m3
+    pmaddwd m4, m4
+    paddd   m1, m2
+    paddd   m3, m4
+    paddd   m1, m3
+    paddd   m0, m1
+
+    dec     r2d
+    jnz    .loop
+
+    ; calculate sum and return
+    HADDD   m0, m1
+    movd    eax, xm0
+    RET

x265-hg20140324.tar.bz2/source/common/x86/x86inc.asm -> x265-hg20140928.tar.bz2/source/common/x86/x86inc.asm Changed

@@ -2,11 +2,13 @@
 ;* x86inc.asm: x264asm abstraction layer
 ;*****************************************************************************
 ;* Copyright (C) 2005-2014 x264 project
+;*               2013-2014 x265 project
 ;*
 ;* Authors: Loren Merritt <lorenm@u.washington.edu>
 ;*          Anton Mitrofanov <BugMaster@narod.ru>
-;*          Jason Garrett-Glaser <darkshikari@gmail.com>
+;*          Fiona Glaser <fiona@x264.com>
 ;*          Henrik Gramner <henrik@gramner.com>
+;*          Min Chen <chenm003@163.com>
 ;*
 ;* Permission to use, copy, modify, and/or distribute this software for any
 ;* purpose with or without fee is hereby granted, provided that the above
@@ -42,6 +44,14 @@
     %define public_prefix private_prefix
 %endif
 
+%ifndef STACK_ALIGNMENT
+    %if ARCH_X86_64
+        %define STACK_ALIGNMENT 16
+    %else
+        %define STACK_ALIGNMENT 4
+    %endif
+%endif
+
 %define WIN64  0
 %define UNIX64 0
 %if ARCH_X86_64
@@ -82,9 +92,6 @@
     default rel
 %endif
 
-; Always use long nops (reduces 0x90 spam in disassembly on x86_32)
-CPU amdnop
-
 ; Macros to eliminate most code duplication between x86_32 and x86_64:
 ; Currently this works only for leaf functions which load all their arguments
 ; into registers at the start, and make no other use of the stack. Luckily that
@@ -94,8 +101,9 @@
 ; %1 = number of arguments. loads them from stack if needed.
 ; %2 = number of registers used. pushes callee-saved regs if needed.
 ; %3 = number of xmm registers used. pushes callee-saved xmm regs if needed.
-; %4 = (optional) stack size to be allocated. If not aligned (x86-32 ICC 10.x,
-;      MSVC or YMM), the stack will be manually aligned (to 16 or 32 bytes),
+; %4 = (optional) stack size to be allocated. The stack will be aligned before
+;      allocating the specified stack size. If the required stack alignment is
+;      larger than the known stack alignment the stack will be manually aligned
 ;      and an extra register will be allocated to hold the original stack
 ;      pointer (to not invalidate r0m etc.). To prevent the use of an extra
 ;      register as stack pointer, request a negative stack size.
@@ -103,8 +111,10 @@
 ; PROLOGUE can also be invoked by adding the same options to cglobal
 
 ; e.g.
-; cglobal foo, 2,3,0, dst, src, tmp
-; declares a function (foo), taking two args (dst and src) and one local variable (tmp)
+; cglobal foo, 2,3,7,0x40, dst, src, tmp
+; declares a function (foo) that automatically loads two arguments (dst and
+; src) into registers, uses one additional register (tmp) plus 7 vector
+; registers (m0-m6) and allocates 0x40 bytes of stack space.
 
 ; TODO Some functions can use some args directly from the stack. If they're the
 ; last args then you can just not declare them, but if they're in the middle
@@ -304,26 +314,28 @@
     %assign n_arg_names %0
 %endmacro
 
+%define required_stack_alignment ((mmsize + 15) & ~15)
+
 %macro ALLOC_STACK 1-2 0 ; stack_size, n_xmm_regs (for win64 only)
     %ifnum %1
         %if %1 != 0
-            %assign %%stack_alignment ((mmsize + 15) & ~15)
+            %assign %%pad 0
             %assign stack_size %1
             %if stack_size < 0
                 %assign stack_size -stack_size
             %endif
-            %assign stack_size_padded stack_size
             %if WIN64
-                %assign stack_size_padded stack_size_padded + 32 ; reserve 32 bytes for shadow space
+                %assign %%pad %%pad + 32 ; shadow space
                 %if mmsize != 8
                     %assign xmm_regs_used %2
                     %if xmm_regs_used > 8
-                        %assign stack_size_padded stack_size_padded + (xmm_regs_used-8)*16
+                        %assign %%pad %%pad + (xmm_regs_used-8)*16 ; callee-saved xmm registers
                     %endif
                 %endif
             %endif
-            %if mmsize <= 16 && HAVE_ALIGNED_STACK
-                %assign stack_size_padded stack_size_padded + %%stack_alignment - gprsize - (stack_offset & (%%stack_alignment - 1))
+            %if required_stack_alignment <= STACK_ALIGNMENT
+                ; maintain the current stack alignment
+                %assign stack_size_padded stack_size + %%pad + ((-%%pad-stack_offset-gprsize) & (STACK_ALIGNMENT-1))
                 SUB rsp, stack_size_padded
             %else
                 %assign %%reg_num (regs_used - 1)
@@ -332,17 +344,17 @@
                 ; it, i.e. in [rsp+stack_size_padded], so we can restore the
                 ; stack in a single instruction (i.e. mov rsp, rstk or mov
                 ; rsp, [rsp+stack_size_padded])
-                mov  rstk, rsp
                 %if %1 < 0 ; need to store rsp on stack
-                    sub  rsp, gprsize+stack_size_padded
-                    and  rsp, ~(%%stack_alignment-1)
-                    %xdefine rstkm [rsp+stack_size_padded]
-                    mov rstkm, rstk
+                    %xdefine rstkm [rsp + stack_size + %%pad]
+                    %assign %%pad %%pad + gprsize
                 %else ; can keep rsp in rstk during whole function
-                    sub  rsp, stack_size_padded
-                    and  rsp, ~(%%stack_alignment-1)
                     %xdefine rstkm rstk
                 %endif
+                %assign stack_size_padded stack_size + ((%%pad + required_stack_alignment-1) & ~(required_stack_alignment-1))
+                mov rstk, rsp
+                and rsp, ~(required_stack_alignment-1)
+                sub rsp, stack_size_padded
+                movifnidn rstkm, rstk
             %endif
             WIN64_PUSH_XMM
         %endif
@@ -351,7 +363,7 @@
 
 %macro SETUP_STACK_POINTER 1
     %ifnum %1
-        %if %1 != 0 && (HAVE_ALIGNED_STACK == 0 || mmsize == 32)
+        %if %1 != 0 && required_stack_alignment > STACK_ALIGNMENT
             %if %1 > 0
                 %assign regs_used (regs_used + 1)
             %elif ARCH_X86_64 && regs_used == num_args && num_args <= 4 + UNIX64 * 2
@@ -425,7 +437,9 @@
     %assign xmm_regs_used %1
     ASSERT xmm_regs_used <= 16
     %if xmm_regs_used > 8
-        %assign stack_size_padded (xmm_regs_used-8)*16 + (~stack_offset&8) + 32
+        ; Allocate stack space for callee-saved xmm registers plus shadow space and align the stack.
+        %assign %%pad (xmm_regs_used-8)*16 + 32
+        %assign stack_size_padded %%pad + ((-%%pad-stack_offset-gprsize) & (STACK_ALIGNMENT-1))
         SUB rsp, stack_size_padded
     %endif
     WIN64_PUSH_XMM
@@ -441,7 +455,7 @@
         %endrep
     %endif
     %if stack_size_padded > 0
-        %if stack_size > 0 && (mmsize == 32 || HAVE_ALIGNED_STACK == 0)
+        %if stack_size > 0 && required_stack_alignment > STACK_ALIGNMENT
             mov rsp, rstkm
         %else
             add %1, stack_size_padded
@@ -507,7 +521,7 @@
 
 %macro RET 0
 %if stack_size_padded > 0
-%if mmsize == 32 || HAVE_ALIGNED_STACK == 0
+%if required_stack_alignment > STACK_ALIGNMENT
     mov rsp, rstkm
 %else
     add rsp, stack_size_padded
@@ -563,7 +577,7 @@
 
 %macro RET 0
 %if stack_size_padded > 0
-%if mmsize == 32 || HAVE_ALIGNED_STACK == 0
+%if required_stack_alignment > STACK_ALIGNMENT
     mov rsp, rstkm
 %else
     add rsp, stack_size_padded
@@ -741,19 +755,26 @@
 %define    cpuflag(x) ((cpuflags & (cpuflags_ %+ x)) == (cpuflags_ %+ x))
 %define notcpuflag(x) ((cpuflags & (cpuflags_ %+ x)) != (cpuflags_ %+ x))
 
-; Takes up to 2 cpuflags from the above list.
+; Takes an arbitrary number of cpuflags from the above list.
 ; All subsequent functions (up to the next INIT_CPUFLAGS) is built for the specified cpu.
 ; You shouldn't need to invoke this macro directly, it's a subroutine for INIT_MMX &co.
-%macro INIT_CPUFLAGS 0-2
-    CPU amdnop
+%macro INIT_CPUFLAGS 0-*
+    %xdefine SUFFIX
+    %undef cpuname
+    %assign cpuflags 0
+
     %if %0 >= 1
-        %xdefine cpuname %1
-        %assign cpuflags cpuflags_%1
-        %if %0 >= 2
-            %xdefine cpuname %1_%2
-            %assign cpuflags cpuflags | cpuflags_%2
-        %endif
+        %rep %0
+            %ifdef cpuname
+                %xdefine cpuname cpuname %+ _%1
+            %else
+                %xdefine cpuname %1
+            %endif
+            %assign cpuflags cpuflags | cpuflags_%1
+            %rotate 1
+        %endrep
         %xdefine SUFFIX _ %+ cpuname
+
         %if cpuflag(avx)
             %assign avx_enabled 1
         %endif
@@ -764,16 +785,15 @@
         %endif
         %if cpuflag(aligned)
             %define movu mova
-        %elifidn %1, sse3
+        %elif cpuflag(sse3) && notcpuflag(ssse3)
             %define movu lddqu
         %endif
-        %if ARCH_X86_64 == 0 && notcpuflag(sse2)
-            CPU basicnop
-        %endif
+    %endif
+
+    %if ARCH_X86_64 || cpuflag(sse2)
+        CPU amdnop
     %else
-        %xdefine SUFFIX
-        %undef cpuname
-        %undef cpuflags
+        CPU basicnop
     %endif
 %endmacro
 
@@ -868,6 +888,8 @@
     %define ymmmm%1   mm%1
     %define ymmxmm%1 xmm%1
     %define ymmymm%1 ymm%1
+    %define ymm%1xmm xmm%1
+    %define xmm%1ymm ymm%1
     %define xm%1 xmm %+ m%1
     %define ym%1 ymm %+ m%1
 %endmacro
@@ -1026,15 +1048,16 @@
 %endmacro
 
 ;%1 == instruction
-;%2 == 1 if float, 0 if int
-;%3 == 1 if non-destructive or 4-operand (xmm, xmm, xmm, imm), 0 otherwise
-;%4 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not
-;%5+: operands
-%macro RUN_AVX_INSTR 5-8+
-    %ifnum sizeof%6
+;%2 == minimal instruction set
+;%3 == 1 if float, 0 if int
+;%4 == 1 if non-destructive or 4-operand (xmm, xmm, xmm, imm), 0 otherwise
+;%5 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not
+;%6+: operands
+%macro RUN_AVX_INSTR 6-9+
+    %ifnum sizeof%7
+        %assign __sizeofreg sizeof%7
+    %elifnum sizeof%6
         %assign __sizeofreg sizeof%6
-    %elifnum sizeof%5
-        %assign __sizeofreg sizeof%5
     %else
         %assign __sizeofreg mmsize
     %endif
@@ -1043,325 +1066,333 @@
         %xdefine __instr v%1
     %else
         %xdefine __instr %1
-        %if %0 >= 7+%3
+        %if %0 >= 8+%4
             %assign __emulate_avx 1
         %endif
     %endif
+    %ifnidn %2, fnord
+        %ifdef cpuname
+            %if notcpuflag(%2)
+                %error use of ``%1'' %2 instruction in cpuname function: current_function
+            %endif
+        %endif
+    %endif
 
     %if __emulate_avx
-        %xdefine __src1 %6
-        %xdefine __src2 %7
-        %ifnidn %5, %6
-            %if %0 >= 8
-                CHECK_AVX_INSTR_EMU {%1 %5, %6, %7, %8}, %5, %7, %8
+        %xdefine __src1 %7
+        %xdefine __src2 %8
+        %ifnidn %6, %7
+            %if %0 >= 9
+                CHECK_AVX_INSTR_EMU {%1 %6, %7, %8, %9}, %6, %8, %9
             %else
-                CHECK_AVX_INSTR_EMU {%1 %5, %6, %7}, %5, %7
+                CHECK_AVX_INSTR_EMU {%1 %6, %7, %8}, %6, %8
             %endif
-            %if %4 && %3 == 0
-                %ifnid %7
+            %if %5 && %4 == 0
+                %ifnid %8
                     ; 3-operand AVX instructions with a memory arg can only have it in src2,
                     ; whereas SSE emulation prefers to have it in src1 (i.e. the mov).
                     ; So, if the instruction is commutative with a memory arg, swap them.
-                    %xdefine __src1 %7
-                    %xdefine __src2 %6
+                    %xdefine __src1 %8
+                    %xdefine __src2 %7
                 %endif
             %endif
             %if __sizeofreg == 8
-                MOVQ %5, __src1
-            %elif %2
-                MOVAPS %5, __src1
+                MOVQ %6, __src1
+            %elif %3
+                MOVAPS %6, __src1
             %else
-                MOVDQA %5, __src1
+                MOVDQA %6, __src1
             %endif
         %endif
-        %if %0 >= 8
-            %1 %5, __src2, %8
+        %if %0 >= 9
+            %1 %6, __src2, %9
         %else
-            %1 %5, __src2
+            %1 %6, __src2
         %endif
-    %elif %0 >= 8
-        __instr %5, %6, %7, %8
+    %elif %0 >= 9
+        __instr %6, %7, %8, %9
+    %elif %0 == 8
+        __instr %6, %7, %8
     %elif %0 == 7
-        __instr %5, %6, %7
-    %elif %0 == 6
-        __instr %5, %6
+        __instr %6, %7
     %else
-        __instr %5
+        __instr %6
     %endif
 %endmacro
 
 ;%1 == instruction
-;%2 == 1 if float, 0 if int
-;%3 == 1 if non-destructive or 4-operand (xmm, xmm, xmm, imm), 0 otherwise
-;%4 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not
-%macro AVX_INSTR 1-4 0, 1, 0
-    %macro %1 1-9 fnord, fnord, fnord, fnord, %1, %2, %3, %4
+;%2 == minimal instruction set
+;%3 == 1 if float, 0 if int
+;%4 == 1 if non-destructive or 4-operand (xmm, xmm, xmm, imm), 0 otherwise
+;%5 == 1 if commutative (i.e. doesn't matter which src arg is which), 0 if not
+%macro AVX_INSTR 1-5 fnord, 0, 1, 0
+    %macro %1 1-10 fnord, fnord, fnord, fnord, %1, %2, %3, %4, %5
         %ifidn %2, fnord
-            RUN_AVX_INSTR %6, %7, %8, %9, %1
+            RUN_AVX_INSTR %6, %7, %8, %9, %10, %1
         %elifidn %3, fnord
-            RUN_AVX_INSTR %6, %7, %8, %9, %1, %2
+            RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2
         %elifidn %4, fnord
-            RUN_AVX_INSTR %6, %7, %8, %9, %1, %2, %3
+            RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3
         %elifidn %5, fnord
-            RUN_AVX_INSTR %6, %7, %8, %9, %1, %2, %3, %4
+            RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3, %4
         %else
-            RUN_AVX_INSTR %6, %7, %8, %9, %1, %2, %3, %4, %5
+            RUN_AVX_INSTR %6, %7, %8, %9, %10, %1, %2, %3, %4, %5
         %endif
     %endmacro
 %endmacro
 
 ; Instructions with both VEX and non-VEX encodings
 ; Non-destructive instructions are written without parameters
-AVX_INSTR addpd, 1, 0, 1
-AVX_INSTR addps, 1, 0, 1
-AVX_INSTR addsd, 1, 0, 1
-AVX_INSTR addss, 1, 0, 1
-AVX_INSTR addsubpd, 1, 0, 0
-AVX_INSTR addsubps, 1, 0, 0
-AVX_INSTR aesdec, 0, 0, 0
-AVX_INSTR aesdeclast, 0, 0, 0
-AVX_INSTR aesenc, 0, 0, 0
-AVX_INSTR aesenclast, 0, 0, 0
+AVX_INSTR addpd, sse2, 1, 0, 1
+AVX_INSTR addps, sse, 1, 0, 1
+AVX_INSTR addsd, sse2, 1, 0, 1
+AVX_INSTR addss, sse, 1, 0, 1
+AVX_INSTR addsubpd, sse3, 1, 0, 0
+AVX_INSTR addsubps, sse3, 1, 0, 0
+AVX_INSTR aesdec, fnord, 0, 0, 0
+AVX_INSTR aesdeclast, fnord, 0, 0, 0
+AVX_INSTR aesenc, fnord, 0, 0, 0
+AVX_INSTR aesenclast, fnord, 0, 0, 0
 AVX_INSTR aesimc
 AVX_INSTR aeskeygenassist
-AVX_INSTR andnpd, 1, 0, 0
-AVX_INSTR andnps, 1, 0, 0
-AVX_INSTR andpd, 1, 0, 1
-AVX_INSTR andps, 1, 0, 1
-AVX_INSTR blendpd, 1, 0, 0
-AVX_INSTR blendps, 1, 0, 0
-AVX_INSTR blendvpd, 1, 0, 0
-AVX_INSTR blendvps, 1, 0, 0
-AVX_INSTR cmppd, 1, 1, 0
-AVX_INSTR cmpps, 1, 1, 0
-AVX_INSTR cmpsd, 1, 1, 0
-AVX_INSTR cmpss, 1, 1, 0
-AVX_INSTR comisd
-AVX_INSTR comiss
-AVX_INSTR cvtdq2pd
-AVX_INSTR cvtdq2ps
-AVX_INSTR cvtpd2dq
-AVX_INSTR cvtpd2ps
-AVX_INSTR cvtps2dq
-AVX_INSTR cvtps2pd
-AVX_INSTR cvtsd2si
-AVX_INSTR cvtsd2ss
-AVX_INSTR cvtsi2sd
-AVX_INSTR cvtsi2ss
-AVX_INSTR cvtss2sd
-AVX_INSTR cvtss2si
-AVX_INSTR cvttpd2dq
-AVX_INSTR cvttps2dq
-AVX_INSTR cvttsd2si
-AVX_INSTR cvttss2si
-AVX_INSTR divpd, 1, 0, 0
-AVX_INSTR divps, 1, 0, 0
-AVX_INSTR divsd, 1, 0, 0
-AVX_INSTR divss, 1, 0, 0
-AVX_INSTR dppd, 1, 1, 0
-AVX_INSTR dpps, 1, 1, 0
-AVX_INSTR extractps
-AVX_INSTR haddpd, 1, 0, 0
-AVX_INSTR haddps, 1, 0, 0
-AVX_INSTR hsubpd, 1, 0, 0
-AVX_INSTR hsubps, 1, 0, 0
-AVX_INSTR insertps, 1, 1, 0
-AVX_INSTR lddqu
-AVX_INSTR ldmxcsr
-AVX_INSTR maskmovdqu
-AVX_INSTR maxpd, 1, 0, 1
-AVX_INSTR maxps, 1, 0, 1
-AVX_INSTR maxsd, 1, 0, 1
-AVX_INSTR maxss, 1, 0, 1
-AVX_INSTR minpd, 1, 0, 1
-AVX_INSTR minps, 1, 0, 1
-AVX_INSTR minsd, 1, 0, 1
-AVX_INSTR minss, 1, 0, 1
-AVX_INSTR movapd
-AVX_INSTR movaps
+AVX_INSTR andnpd, sse2, 1, 0, 0
+AVX_INSTR andnps, sse, 1, 0, 0
+AVX_INSTR andpd, sse2, 1, 0, 1
+AVX_INSTR andps, sse, 1, 0, 1
+AVX_INSTR blendpd, sse4, 1, 0, 0
+AVX_INSTR blendps, sse4, 1, 0, 0
+AVX_INSTR blendvpd, sse4, 1, 0, 0
+AVX_INSTR blendvps, sse4, 1, 0, 0
+AVX_INSTR cmppd, sse2, 1, 1, 0
+AVX_INSTR cmpps, sse, 1, 1, 0
+AVX_INSTR cmpsd, sse2, 1, 1, 0
+AVX_INSTR cmpss, sse, 1, 1, 0
+AVX_INSTR comisd, sse2
+AVX_INSTR comiss, sse
+AVX_INSTR cvtdq2pd, sse2
+AVX_INSTR cvtdq2ps, sse2
+AVX_INSTR cvtpd2dq, sse2
+AVX_INSTR cvtpd2ps, sse2
+AVX_INSTR cvtps2dq, sse2
+AVX_INSTR cvtps2pd, sse2
+AVX_INSTR cvtsd2si, sse2
+AVX_INSTR cvtsd2ss, sse2
+AVX_INSTR cvtsi2sd, sse2
+AVX_INSTR cvtsi2ss, sse
+AVX_INSTR cvtss2sd, sse2
+AVX_INSTR cvtss2si, sse
+AVX_INSTR cvttpd2dq, sse2
+AVX_INSTR cvttps2dq, sse2
+AVX_INSTR cvttsd2si, sse2
+AVX_INSTR cvttss2si, sse
+AVX_INSTR divpd, sse2, 1, 0, 0
+AVX_INSTR divps, sse, 1, 0, 0
+AVX_INSTR divsd, sse2, 1, 0, 0
+AVX_INSTR divss, sse, 1, 0, 0
+AVX_INSTR dppd, sse4, 1, 1, 0
+AVX_INSTR dpps, sse4, 1, 1, 0
+AVX_INSTR extractps, sse4
+AVX_INSTR haddpd, sse3, 1, 0, 0
+AVX_INSTR haddps, sse3, 1, 0, 0
+AVX_INSTR hsubpd, sse3, 1, 0, 0
+AVX_INSTR hsubps, sse3, 1, 0, 0
+AVX_INSTR insertps, sse4, 1, 1, 0
+AVX_INSTR lddqu, sse3
+AVX_INSTR ldmxcsr, sse
+AVX_INSTR maskmovdqu, sse2
+AVX_INSTR maxpd, sse2, 1, 0, 1
+AVX_INSTR maxps, sse, 1, 0, 1
+AVX_INSTR maxsd, sse2, 1, 0, 1
+AVX_INSTR maxss, sse, 1, 0, 1
+AVX_INSTR minpd, sse2, 1, 0, 1
+AVX_INSTR minps, sse, 1, 0, 1
+AVX_INSTR minsd, sse2, 1, 0, 1
+AVX_INSTR minss, sse, 1, 0, 1
+AVX_INSTR movapd, sse2
+AVX_INSTR movaps, sse
 AVX_INSTR movd
-AVX_INSTR movddup
-AVX_INSTR movdqa
-AVX_INSTR movdqu
-AVX_INSTR movhlps, 1, 0, 0
-AVX_INSTR movhpd, 1, 0, 0
-AVX_INSTR movhps, 1, 0, 0
-AVX_INSTR movlhps, 1, 0, 0
-AVX_INSTR movlpd, 1, 0, 0
-AVX_INSTR movlps, 1, 0, 0
-AVX_INSTR movmskpd
-AVX_INSTR movmskps
-AVX_INSTR movntdq
-AVX_INSTR movntdqa
-AVX_INSTR movntpd
-AVX_INSTR movntps
+AVX_INSTR movddup, sse3
+AVX_INSTR movdqa, sse2
+AVX_INSTR movdqu, sse2
+AVX_INSTR movhlps, sse, 1, 0, 0
+AVX_INSTR movhpd, sse2, 1, 0, 0
+AVX_INSTR movhps, sse, 1, 0, 0
+AVX_INSTR movlhps, sse, 1, 0, 0
+AVX_INSTR movlpd, sse2, 1, 0, 0
+AVX_INSTR movlps, sse, 1, 0, 0
+AVX_INSTR movmskpd, sse2
+AVX_INSTR movmskps, sse
+AVX_INSTR movntdq, sse2
+AVX_INSTR movntdqa, sse4
+AVX_INSTR movntpd, sse2
+AVX_INSTR movntps, sse
 AVX_INSTR movq
-AVX_INSTR movsd, 1, 0, 0
-AVX_INSTR movshdup
-AVX_INSTR movsldup
-AVX_INSTR movss, 1, 0, 0
-AVX_INSTR movupd
-AVX_INSTR movups
-AVX_INSTR mpsadbw, 0, 1, 0
-AVX_INSTR mulpd, 1, 0, 1
-AVX_INSTR mulps, 1, 0, 1
-AVX_INSTR mulsd, 1, 0, 1
-AVX_INSTR mulss, 1, 0, 1
-AVX_INSTR orpd, 1, 0, 1
-AVX_INSTR orps, 1, 0, 1
-AVX_INSTR pabsb
-AVX_INSTR pabsd
-AVX_INSTR pabsw
-AVX_INSTR packsswb, 0, 0, 0
-AVX_INSTR packssdw, 0, 0, 0
-AVX_INSTR packuswb, 0, 0, 0
-AVX_INSTR packusdw, 0, 0, 0
-AVX_INSTR paddb, 0, 0, 1
-AVX_INSTR paddw, 0, 0, 1
-AVX_INSTR paddd, 0, 0, 1
-AVX_INSTR paddq, 0, 0, 1
-AVX_INSTR paddsb, 0, 0, 1
-AVX_INSTR paddsw, 0, 0, 1
-AVX_INSTR paddusb, 0, 0, 1
-AVX_INSTR paddusw, 0, 0, 1
-AVX_INSTR palignr, 0, 1, 0
-AVX_INSTR pand, 0, 0, 1
-AVX_INSTR pandn, 0, 0, 0
-AVX_INSTR pavgb, 0, 0, 1
-AVX_INSTR pavgw, 0, 0, 1
-AVX_INSTR pblendvb, 0, 0, 0
-AVX_INSTR pblendw, 0, 1, 0
-AVX_INSTR pclmulqdq, 0, 1, 0
-AVX_INSTR pcmpestri
-AVX_INSTR pcmpestrm
-AVX_INSTR pcmpistri
-AVX_INSTR pcmpistrm
-AVX_INSTR pcmpeqb, 0, 0, 1
-AVX_INSTR pcmpeqw, 0, 0, 1
-AVX_INSTR pcmpeqd, 0, 0, 1
-AVX_INSTR pcmpeqq, 0, 0, 1
-AVX_INSTR pcmpgtb, 0, 0, 0
-AVX_INSTR pcmpgtw, 0, 0, 0
-AVX_INSTR pcmpgtd, 0, 0, 0
-AVX_INSTR pcmpgtq, 0, 0, 0
-AVX_INSTR pextrb
-AVX_INSTR pextrd
-AVX_INSTR pextrq
-AVX_INSTR pextrw
-AVX_INSTR phaddw, 0, 0, 0
-AVX_INSTR phaddd, 0, 0, 0
-AVX_INSTR phaddsw, 0, 0, 0
-AVX_INSTR phminposuw
-AVX_INSTR phsubw, 0, 0, 0
-AVX_INSTR phsubd, 0, 0, 0
-AVX_INSTR phsubsw, 0, 0, 0
-AVX_INSTR pinsrb, 0, 1, 0
-AVX_INSTR pinsrd, 0, 1, 0
-AVX_INSTR pinsrq, 0, 1, 0
-AVX_INSTR pinsrw, 0, 1, 0
-AVX_INSTR pmaddwd, 0, 0, 1
-AVX_INSTR pmaddubsw, 0, 0, 0
-AVX_INSTR pmaxsb, 0, 0, 1
-AVX_INSTR pmaxsw, 0, 0, 1
-AVX_INSTR pmaxsd, 0, 0, 1
-AVX_INSTR pmaxub, 0, 0, 1
-AVX_INSTR pmaxuw, 0, 0, 1
-AVX_INSTR pmaxud, 0, 0, 1
-AVX_INSTR pminsb, 0, 0, 1
-AVX_INSTR pminsw, 0, 0, 1
-AVX_INSTR pminsd, 0, 0, 1
-AVX_INSTR pminub, 0, 0, 1
-AVX_INSTR pminuw, 0, 0, 1
-AVX_INSTR pminud, 0, 0, 1
-AVX_INSTR pmovmskb
-AVX_INSTR pmovsxbw
-AVX_INSTR pmovsxbd
-AVX_INSTR pmovsxbq
-AVX_INSTR pmovsxwd
-AVX_INSTR pmovsxwq
-AVX_INSTR pmovsxdq
-AVX_INSTR pmovzxbw
-AVX_INSTR pmovzxbd
-AVX_INSTR pmovzxbq
-AVX_INSTR pmovzxwd
-AVX_INSTR pmovzxwq
-AVX_INSTR pmovzxdq
-AVX_INSTR pmuldq, 0, 0, 1
-AVX_INSTR pmulhrsw, 0, 0, 1
-AVX_INSTR pmulhuw, 0, 0, 1
-AVX_INSTR pmulhw, 0, 0, 1
-AVX_INSTR pmullw, 0, 0, 1
-AVX_INSTR pmulld, 0, 0, 1
-AVX_INSTR pmuludq, 0, 0, 1
-AVX_INSTR por, 0, 0, 1
-AVX_INSTR psadbw, 0, 0, 1
-AVX_INSTR pshufb, 0, 0, 0
-AVX_INSTR pshufd
-AVX_INSTR pshufhw
-AVX_INSTR pshuflw
-AVX_INSTR psignb, 0, 0, 0
-AVX_INSTR psignw, 0, 0, 0
-AVX_INSTR psignd, 0, 0, 0
-AVX_INSTR psllw, 0, 0, 0
-AVX_INSTR pslld, 0, 0, 0
-AVX_INSTR psllq, 0, 0, 0
-AVX_INSTR pslldq, 0, 0, 0
-AVX_INSTR psraw, 0, 0, 0
-AVX_INSTR psrad, 0, 0, 0
-AVX_INSTR psrlw, 0, 0, 0
-AVX_INSTR psrld, 0, 0, 0
-AVX_INSTR psrlq, 0, 0, 0
-AVX_INSTR psrldq, 0, 0, 0
-AVX_INSTR psubb, 0, 0, 0
-AVX_INSTR psubw, 0, 0, 0
-AVX_INSTR psubd, 0, 0, 0
-AVX_INSTR psubq, 0, 0, 0
-AVX_INSTR psubsb, 0, 0, 0
-AVX_INSTR psubsw, 0, 0, 0
-AVX_INSTR psubusb, 0, 0, 0
-AVX_INSTR psubusw, 0, 0, 0
-AVX_INSTR ptest
-AVX_INSTR punpckhbw, 0, 0, 0
-AVX_INSTR punpckhwd, 0, 0, 0
-AVX_INSTR punpckhdq, 0, 0, 0
-AVX_INSTR punpckhqdq, 0, 0, 0
-AVX_INSTR punpcklbw, 0, 0, 0
-AVX_INSTR punpcklwd, 0, 0, 0
-AVX_INSTR punpckldq, 0, 0, 0
-AVX_INSTR punpcklqdq, 0, 0, 0
-AVX_INSTR pxor, 0, 0, 1
-AVX_INSTR rcpps, 1, 0, 0
-AVX_INSTR rcpss, 1, 0, 0
-AVX_INSTR roundpd
-AVX_INSTR roundps
-AVX_INSTR roundsd
-AVX_INSTR roundss
-AVX_INSTR rsqrtps, 1, 0, 0
-AVX_INSTR rsqrtss, 1, 0, 0
-AVX_INSTR shufpd, 1, 1, 0
-AVX_INSTR shufps, 1, 1, 0
-AVX_INSTR sqrtpd, 1, 0, 0
-AVX_INSTR sqrtps, 1, 0, 0
-AVX_INSTR sqrtsd, 1, 0, 0
-AVX_INSTR sqrtss, 1, 0, 0
-AVX_INSTR stmxcsr
-AVX_INSTR subpd, 1, 0, 0
-AVX_INSTR subps, 1, 0, 0
-AVX_INSTR subsd, 1, 0, 0
-AVX_INSTR subss, 1, 0, 0
-AVX_INSTR ucomisd
-AVX_INSTR ucomiss
-AVX_INSTR unpckhpd, 1, 0, 0
-AVX_INSTR unpckhps, 1, 0, 0
-AVX_INSTR unpcklpd, 1, 0, 0
-AVX_INSTR unpcklps, 1, 0, 0
-AVX_INSTR xorpd, 1, 0, 1
-AVX_INSTR xorps, 1, 0, 1
+AVX_INSTR movsd, sse2, 1, 0, 0
+AVX_INSTR movshdup, sse3
+AVX_INSTR movsldup, sse3
+AVX_INSTR movss, sse, 1, 0, 0
+AVX_INSTR movupd, sse2
+AVX_INSTR movups, sse
+AVX_INSTR mpsadbw, sse4
+AVX_INSTR mulpd, sse2, 1, 0, 1
+AVX_INSTR mulps, sse, 1, 0, 1
+AVX_INSTR mulsd, sse2, 1, 0, 1
+AVX_INSTR mulss, sse, 1, 0, 1
+AVX_INSTR orpd, sse2, 1, 0, 1
+AVX_INSTR orps, sse, 1, 0, 1
+AVX_INSTR pabsb, ssse3
+AVX_INSTR pabsd, ssse3
+AVX_INSTR pabsw, ssse3
+AVX_INSTR packsswb, mmx, 0, 0, 0
+AVX_INSTR packssdw, mmx, 0, 0, 0
+AVX_INSTR packuswb, mmx, 0, 0, 0
+AVX_INSTR packusdw, sse4, 0, 0, 0
+AVX_INSTR paddb, mmx, 0, 0, 1
+AVX_INSTR paddw, mmx, 0, 0, 1
+AVX_INSTR paddd, mmx, 0, 0, 1
+AVX_INSTR paddq, sse2, 0, 0, 1
+AVX_INSTR paddsb, mmx, 0, 0, 1
+AVX_INSTR paddsw, mmx, 0, 0, 1
+AVX_INSTR paddusb, mmx, 0, 0, 1
+AVX_INSTR paddusw, mmx, 0, 0, 1
+AVX_INSTR palignr, ssse3
+AVX_INSTR pand, mmx, 0, 0, 1
+AVX_INSTR pandn, mmx, 0, 0, 0
+AVX_INSTR pavgb, mmx2, 0, 0, 1
+AVX_INSTR pavgw, mmx2, 0, 0, 1
+AVX_INSTR pblendvb, sse4, 0, 0, 0
+AVX_INSTR pblendw, sse4
+AVX_INSTR pclmulqdq
+AVX_INSTR pcmpestri, sse42
+AVX_INSTR pcmpestrm, sse42
+AVX_INSTR pcmpistri, sse42
+AVX_INSTR pcmpistrm, sse42
+AVX_INSTR pcmpeqb, mmx, 0, 0, 1
+AVX_INSTR pcmpeqw, mmx, 0, 0, 1
+AVX_INSTR pcmpeqd, mmx, 0, 0, 1
+AVX_INSTR pcmpeqq, sse4, 0, 0, 1
+AVX_INSTR pcmpgtb, mmx, 0, 0, 0
+AVX_INSTR pcmpgtw, mmx, 0, 0, 0
+AVX_INSTR pcmpgtd, mmx, 0, 0, 0
+AVX_INSTR pcmpgtq, sse42, 0, 0, 0
+AVX_INSTR pextrb, sse4
+AVX_INSTR pextrd, sse4
+AVX_INSTR pextrq, sse4
+AVX_INSTR pextrw, mmx2
+AVX_INSTR phaddw, ssse3, 0, 0, 0
+AVX_INSTR phaddd, ssse3, 0, 0, 0
+AVX_INSTR phaddsw, ssse3, 0, 0, 0
+AVX_INSTR phminposuw, sse4
+AVX_INSTR phsubw, ssse3, 0, 0, 0
+AVX_INSTR phsubd, ssse3, 0, 0, 0
+AVX_INSTR phsubsw, ssse3, 0, 0, 0
+AVX_INSTR pinsrb, sse4
+AVX_INSTR pinsrd, sse4
+AVX_INSTR pinsrq, sse4
+AVX_INSTR pinsrw, mmx2
+AVX_INSTR pmaddwd, mmx, 0, 0, 1
+AVX_INSTR pmaddubsw, ssse3, 0, 0, 0
+AVX_INSTR pmaxsb, sse4, 0, 0, 1
+AVX_INSTR pmaxsw, mmx2, 0, 0, 1
+AVX_INSTR pmaxsd, sse4, 0, 0, 1
+AVX_INSTR pmaxub, mmx2, 0, 0, 1
+AVX_INSTR pmaxuw, sse4, 0, 0, 1
+AVX_INSTR pmaxud, sse4, 0, 0, 1
+AVX_INSTR pminsb, sse4, 0, 0, 1
+AVX_INSTR pminsw, mmx2, 0, 0, 1
+AVX_INSTR pminsd, sse4, 0, 0, 1
+AVX_INSTR pminub, mmx2, 0, 0, 1
+AVX_INSTR pminuw, sse4, 0, 0, 1
+AVX_INSTR pminud, sse4, 0, 0, 1
+AVX_INSTR pmovmskb, mmx2
+AVX_INSTR pmovsxbw, sse4
+AVX_INSTR pmovsxbd, sse4
+AVX_INSTR pmovsxbq, sse4
+AVX_INSTR pmovsxwd, sse4
+AVX_INSTR pmovsxwq, sse4
+AVX_INSTR pmovsxdq, sse4
+AVX_INSTR pmovzxbw, sse4
+AVX_INSTR pmovzxbd, sse4
+AVX_INSTR pmovzxbq, sse4
+AVX_INSTR pmovzxwd, sse4
+AVX_INSTR pmovzxwq, sse4
+AVX_INSTR pmovzxdq, sse4
+AVX_INSTR pmuldq, sse4, 0, 0, 1
+AVX_INSTR pmulhrsw, ssse3, 0, 0, 1
+AVX_INSTR pmulhuw, mmx2, 0, 0, 1
+AVX_INSTR pmulhw, mmx, 0, 0, 1
+AVX_INSTR pmullw, mmx, 0, 0, 1
+AVX_INSTR pmulld, sse4, 0, 0, 1
+AVX_INSTR pmuludq, sse2, 0, 0, 1
+AVX_INSTR por, mmx, 0, 0, 1
+AVX_INSTR psadbw, mmx2, 0, 0, 1
+AVX_INSTR pshufb, ssse3, 0, 0, 0
+AVX_INSTR pshufd, sse2
+AVX_INSTR pshufhw, sse2
+AVX_INSTR pshuflw, sse2
+AVX_INSTR psignb, ssse3, 0, 0, 0
+AVX_INSTR psignw, ssse3, 0, 0, 0
+AVX_INSTR psignd, ssse3, 0, 0, 0
+AVX_INSTR psllw, mmx, 0, 0, 0
+AVX_INSTR pslld, mmx, 0, 0, 0
+AVX_INSTR psllq, mmx, 0, 0, 0
+AVX_INSTR pslldq, sse2, 0, 0, 0
+AVX_INSTR psraw, mmx, 0, 0, 0
+AVX_INSTR psrad, mmx, 0, 0, 0
+AVX_INSTR psrlw, mmx, 0, 0, 0
+AVX_INSTR psrld, mmx, 0, 0, 0
+AVX_INSTR psrlq, mmx, 0, 0, 0
+AVX_INSTR psrldq, sse2, 0, 0, 0
+AVX_INSTR psubb, mmx, 0, 0, 0
+AVX_INSTR psubw, mmx, 0, 0, 0
+AVX_INSTR psubd, mmx, 0, 0, 0
+AVX_INSTR psubq, sse2, 0, 0, 0
+AVX_INSTR psubsb, mmx, 0, 0, 0
+AVX_INSTR psubsw, mmx, 0, 0, 0
+AVX_INSTR psubusb, mmx, 0, 0, 0
+AVX_INSTR psubusw, mmx, 0, 0, 0
+AVX_INSTR ptest, sse4
+AVX_INSTR punpckhbw, mmx, 0, 0, 0
+AVX_INSTR punpckhwd, mmx, 0, 0, 0
+AVX_INSTR punpckhdq, mmx, 0, 0, 0
+AVX_INSTR punpckhqdq, sse2, 0, 0, 0
+AVX_INSTR punpcklbw, mmx, 0, 0, 0
+AVX_INSTR punpcklwd, mmx, 0, 0, 0
+AVX_INSTR punpckldq, mmx, 0, 0, 0
+AVX_INSTR punpcklqdq, sse2, 0, 0, 0
+AVX_INSTR pxor, mmx, 0, 0, 1
+AVX_INSTR rcpps, sse, 1, 0, 0
+AVX_INSTR rcpss, sse, 1, 0, 0
+AVX_INSTR roundpd, sse4
+AVX_INSTR roundps, sse4
+AVX_INSTR roundsd, sse4
+AVX_INSTR roundss, sse4
+AVX_INSTR rsqrtps, sse, 1, 0, 0
+AVX_INSTR rsqrtss, sse, 1, 0, 0
+AVX_INSTR shufpd, sse2, 1, 1, 0
+AVX_INSTR shufps, sse, 1, 1, 0
+AVX_INSTR sqrtpd, sse2, 1, 0, 0
+AVX_INSTR sqrtps, sse, 1, 0, 0
+AVX_INSTR sqrtsd, sse2, 1, 0, 0
+AVX_INSTR sqrtss, sse, 1, 0, 0
+AVX_INSTR stmxcsr, sse
+AVX_INSTR subpd, sse2, 1, 0, 0
+AVX_INSTR subps, sse, 1, 0, 0
+AVX_INSTR subsd, sse2, 1, 0, 0
+AVX_INSTR subss, sse, 1, 0, 0
+AVX_INSTR ucomisd, sse2
+AVX_INSTR ucomiss, sse
+AVX_INSTR unpckhpd, sse2, 1, 0, 0
+AVX_INSTR unpckhps, sse, 1, 0, 0
+AVX_INSTR unpcklpd, sse2, 1, 0, 0
+AVX_INSTR unpcklps, sse, 1, 0, 0
+AVX_INSTR xorpd, sse2, 1, 0, 1
+AVX_INSTR xorps, sse, 1, 0, 1
 
 ; 3DNow instructions, for sharing code between AVX, SSE and 3DN
-AVX_INSTR pfadd, 1, 0, 1
-AVX_INSTR pfsub, 1, 0, 0
-AVX_INSTR pfmul, 1, 0, 1
+AVX_INSTR pfadd, 3dnow, 1, 0, 1
+AVX_INSTR pfsub, 3dnow, 1, 0, 0
+AVX_INSTR pfmul, 3dnow, 1, 0, 1
 
 ; base-4 constants for shuffles
 %assign i 0
@@ -1385,15 +1416,18 @@
     %macro %1 4-7 %1, %2, %3
         %if cpuflag(xop)
             v%5 %1, %2, %3, %4
-        %else
+        %elifnidn %1, %4
             %6 %1, %2, %3
             %7 %1, %4
+        %else
+            %error non-xop emulation of ``%5 %1, %2, %3, %4'' is not supported
         %endif
     %endmacro
 %endmacro
 
-FMA_INSTR  pmacsdd,  pmulld, paddd
 FMA_INSTR  pmacsww,  pmullw, paddw
+FMA_INSTR  pmacsdd,  pmulld, paddd ; sse4 emulation
+FMA_INSTR pmacsdql,  pmuldq, paddq ; sse4 emulation
 FMA_INSTR pmadcswd, pmaddwd, paddd
 
 ; convert FMA4 to FMA3 if possible
@@ -1448,3 +1482,13 @@
 %endif
 %endmacro
 %endif
+
+; workaround: vpbroadcastd with register, the yasm will generate wrong code
+%macro vpbroadcastd 2
+  %ifid %2
+    movd         %1 %+ xmm, %2
+    vpbroadcastd %1, %1 %+ xmm
+  %else
+    vpbroadcastd %1, %2
+  %endif
+%endmacro

x265-hg20140324.tar.bz2/source/common/x86/x86util.asm -> x265-hg20140928.tar.bz2/source/common/x86/x86util.asm Changed

@@ -21,7 +21,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
 ;*
 ;* This program is also available under a commercial proprietary license.
-;* For more information, contact us at licensing@x264.com.
+;* For more information, contact us at license @ x265.com.
 ;*****************************************************************************
 
 %assign FENC_STRIDE 64
@@ -290,6 +290,13 @@
     pminsw %1, %3
 %endmacro
 
+%macro CLIPW2 4 ;(dst0, dst1, min, max)
+    pmaxsw %1, %3
+    pmaxsw %2, %3
+    pminsw %1, %4
+    pminsw %2, %4
+%endmacro
+
 %macro HADDD 2 ; sum junk
 %if sizeof%1 == 32
 %define %2 xmm%2
@@ -298,11 +305,16 @@
     paddd   %1, %2
 %endif
 %if mmsize >= 16
+%if cpuflag(xop) && sizeof%1 == 16
+    vphadddq %1, %1
+%endif
     movhlps %2, %1
     paddd   %1, %2
 %endif
+%if notcpuflag(xop)
     PSHUFLW %2, %1, q0032
     paddd   %1, %2
+%endif
 %undef %1
 %undef %2
 %endmacro
@@ -868,3 +880,14 @@
     SWAP       %2, %3
 %endif
 %endmacro
+
+; IACA support
+%macro IACA_START 0
+    mov ebx, 111
+    db 0x64, 0x67, 0x90
+%endmacro
+
+%macro IACA_END 0
+    mov ebx, 222
+    db 0x64, 0x67, 0x90
+%endmacro

x265-hg20140324.tar.bz2/source/compat/msvc/stdint.h -> x265-hg20140928.tar.bz2/source/compat/msvc/stdint.h Changed

x265-hg20140324.tar.bz2/source/encoder/CMakeLists.txt -> x265-hg20140928.tar.bz2/source/encoder/CMakeLists.txt Changed

@@ -1,60 +1,21 @@
 # vim: syntax=cmake
-set(LIBENCODER_HDR
-    ../Lib/TLibEncoder/NALwrite.h
-    ../Lib/TLibEncoder/SEIwrite.h
-    ../Lib/TLibEncoder/SyntaxElementWriter.h
-    ../Lib/TLibEncoder/TEncBinCoderCABAC.h
-    ../Lib/TLibEncoder/TEncCu.h
-    ../Lib/TLibEncoder/TEncEntropy.h
-    ../Lib/TLibEncoder/TEncSampleAdaptiveOffset.h
-    ../Lib/TLibEncoder/TEncSbac.h
-    ../Lib/TLibEncoder/TEncSearch.h)
-set(LIBENCODER_SRC
-    ../Lib/TLibEncoder/NALwrite.cpp
-    ../Lib/TLibEncoder/SEIwrite.cpp
-    ../Lib/TLibEncoder/SyntaxElementWriter.cpp
-    ../Lib/TLibEncoder/TEncBinCoderCABAC.cpp
-    ../Lib/TLibEncoder/TEncCu.cpp
-    ../Lib/TLibEncoder/TEncEntropy.cpp
-    ../Lib/TLibEncoder/TEncSampleAdaptiveOffset.cpp
-    ../Lib/TLibEncoder/TEncSbac.cpp
-    ../Lib/TLibEncoder/TEncSearch.cpp)
-source_group(TLibEncoder FILES ${LIBENCODER_HDR} ${LIBENCODER_SRC})
-
-if(GCC)
-    # encoder.cpp must include HM headers which are not careful about named parameters
-    set_source_files_properties(encoder.cpp PROPERTIES COMPILE_FLAGS -Wno-unused-parameter)
-
-    # HM code is not careful about signed compares
-    set_source_files_properties(compress.cpp ${LIBENCODER_SRC} PROPERTIES COMPILE_FLAGS 
-        "-Wno-sign-compare")
-endif(GCC)
-if(MSVC)
-    # ignore these warnings from HM source
-    # /wd4244 type conversion, possible loss of data
-    # /wd4389 signed/unsigned mismatch
-    # /wd4018 '<' signed/unsigned mismatch
-    # /wd4800 performance warning: bool coersion
-    set_source_files_properties(${LIBENCODER_SRC} PROPERTIES COMPILE_FLAGS 
-        "/wd4244 /wd4389 /wd4018 /wd4800")
-endif(MSVC)
-
-check_include_files(inttypes.h HAVE_INT_TYPES_H)
-if(HAVE_INT_TYPES_H)
-    add_definitions(-DHAVE_INT_TYPES_H=1)
-endif()
 
 add_library(encoder OBJECT ../x265.h
-    ${LIBENCODER_SRC} ${LIBENCODER_HDR}
-    bitcost.cpp bitcost.h
+    analysis.cpp analysis.h
+    search.cpp search.h
+    predict.cpp  predict.h
+    bitcost.cpp bitcost.h rdcost.h
     motion.cpp motion.h
     slicetype.cpp slicetype.h
     frameencoder.cpp frameencoder.h
     framefilter.cpp framefilter.h
-    cturow.cpp cturow.h
+    level.cpp level.h
+    nal.cpp nal.h
+    sei.cpp sei.h
+    sao.cpp sao.h
+    entropy.cpp entropy.h
     dpb.cpp dpb.h
     ratecontrol.cpp ratecontrol.h
-    compress.cpp
     reference.cpp reference.h
     encoder.cpp encoder.h
     api.cpp

x265-hg20140928.tar.bz2/source/encoder/analysis.cpp Added

@@ -0,0 +1,2274 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Deepthi Nandakumar <deepthi@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "analysis.h"
+#include "primitives.h"
+#include "common.h"
+#include "rdcost.h"
+#include "encoder.h"
+#include "PPA/ppa.h"
+
+using namespace x265;
+
+namespace {
+// TO DO: Remove this function with a table.
+int getDepthScanIdx(int x, int y, int size)
+{
+    if (size == 1)
+        return 0;
+
+    int depth = 0;
+    int h = size >> 1;
+
+    if (x >= h)
+    {
+        x -= h;
+        depth += h * h;
+    }
+
+    if (y >= h)
+    {
+        y -= h;
+        depth += 2 * h * h;
+    }
+
+    return depth + getDepthScanIdx(x, y, h);
+}
+}
+
+Analysis::Analysis()
+{
+    m_bestPredYuv     = NULL;
+    m_bestResiYuv     = NULL;
+    m_bestRecoYuv     = NULL;
+
+    m_tmpPredYuv      = NULL;
+    m_tmpResiYuv      = NULL;
+    m_tmpRecoYuv      = NULL;
+    m_bestMergeRecoYuv = NULL;
+    m_origYuv         = NULL;
+    for (int i = 0; i < MAX_PRED_TYPES; i++)
+        m_modePredYuv[i] = NULL;
+}
+
+bool Analysis::create(uint32_t numCUDepth, uint32_t maxWidth)
+{
+    X265_CHECK(numCUDepth <= NUM_CU_DEPTH, "invalid numCUDepth\n");
+
+    m_bestPredYuv = new TComYuv*[numCUDepth];
+    m_bestResiYuv = new ShortYuv*[numCUDepth];
+    m_bestRecoYuv = new TComYuv*[numCUDepth];
+
+    m_tmpPredYuv     = new TComYuv*[numCUDepth];
+    m_modePredYuv[0] = new TComYuv*[numCUDepth];
+    m_modePredYuv[1] = new TComYuv*[numCUDepth];
+    m_modePredYuv[2] = new TComYuv*[numCUDepth];
+    m_modePredYuv[3] = new TComYuv*[numCUDepth];
+    m_modePredYuv[4] = new TComYuv*[numCUDepth];
+    m_modePredYuv[5] = new TComYuv*[numCUDepth];
+
+    m_tmpResiYuv = new ShortYuv*[numCUDepth];
+    m_tmpRecoYuv = new TComYuv*[numCUDepth];
+
+    m_bestMergeRecoYuv = new TComYuv*[numCUDepth];
+
+    m_origYuv = new TComYuv*[numCUDepth];
+
+    int csp       = m_param->internalCsp;
+    bool tqBypass = m_param->bCULossless || m_param->bLossless;
+
+    m_memPool = new TComDataCU[numCUDepth];
+
+    bool ok = true;
+    for (uint32_t i = 0; i < numCUDepth; i++)
+    {
+        uint32_t numPartitions = 1 << (g_maxFullDepth - i) * 2;
+        uint32_t cuSize = maxWidth >> i;
+
+        uint32_t sizeL = cuSize * cuSize;
+        uint32_t sizeC = sizeL >> (CHROMA_H_SHIFT(csp) + CHROMA_V_SHIFT(csp));
+
+        ok &= m_memPool[i].initialize(numPartitions, sizeL, sizeC, 8, tqBypass);
+
+        m_interCU_2Nx2N[i]  = new TComDataCU;
+        m_interCU_2Nx2N[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 0, tqBypass);
+
+        m_interCU_2NxN[i]   = new TComDataCU;
+        m_interCU_2NxN[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 1, tqBypass);
+
+        m_interCU_Nx2N[i]   = new TComDataCU;
+        m_interCU_Nx2N[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 2, tqBypass);
+
+        m_intraInInterCU[i] = new TComDataCU;
+        m_intraInInterCU[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 3, tqBypass);
+
+        m_mergeCU[i]        = new TComDataCU;
+        m_mergeCU[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 4, tqBypass);
+
+        m_bestMergeCU[i]    = new TComDataCU;
+        m_bestMergeCU[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 5, tqBypass);
+
+        m_bestCU[i]         = new TComDataCU;
+        m_bestCU[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 6, tqBypass);
+
+        m_tempCU[i]         = new TComDataCU;
+        m_tempCU[i]->create(&m_memPool[i], numPartitions, cuSize, csp, 7, tqBypass);
+
+        m_bestPredYuv[i] = new TComYuv;
+        ok &= m_bestPredYuv[i]->create(cuSize, cuSize, csp);
+
+        m_bestResiYuv[i] = new ShortYuv;
+        ok &= m_bestResiYuv[i]->create(cuSize, cuSize, csp);
+
+        m_bestRecoYuv[i] = new TComYuv;
+        ok &= m_bestRecoYuv[i]->create(cuSize, cuSize, csp);
+
+        m_tmpPredYuv[i] = new TComYuv;
+        ok &= m_tmpPredYuv[i]->create(cuSize, cuSize, csp);
+
+        for (int j = 0; j < MAX_PRED_TYPES; j++)
+        {
+            m_modePredYuv[j][i] = new TComYuv;
+            ok &= m_modePredYuv[j][i]->create(cuSize, cuSize, csp);
+        }
+
+        m_tmpResiYuv[i] = new ShortYuv;
+        ok &= m_tmpResiYuv[i]->create(cuSize, cuSize, csp);
+
+        m_tmpRecoYuv[i] = new TComYuv;
+        ok &= m_tmpRecoYuv[i]->create(cuSize, cuSize, csp);
+
+        m_bestMergeRecoYuv[i] = new TComYuv;
+        ok &= m_bestMergeRecoYuv[i]->create(cuSize, cuSize, csp);
+
+        m_origYuv[i] = new TComYuv;
+        ok &= m_origYuv[i]->create(cuSize, cuSize, csp);
+    }
+
+    m_bEncodeDQP = false;
+    return ok;
+}
+
+void Analysis::destroy()
+{
+    uint32_t numCUDepth = g_maxCUDepth + 1;
+    for (uint32_t i = 0; i < numCUDepth; i++)
+    {
+        m_memPool[i].destroy();
+
+        delete m_interCU_2Nx2N[i];
+        delete m_interCU_2NxN[i];
+        delete m_interCU_Nx2N[i];
+        delete m_intraInInterCU[i];
+        delete m_mergeCU[i];
+        delete m_bestMergeCU[i];
+        delete m_bestCU[i];
+        delete m_tempCU[i];
+
+        if (m_bestPredYuv && m_bestPredYuv[i])
+        {
+            m_bestPredYuv[i]->destroy();
+            delete m_bestPredYuv[i];
+        }
+        if (m_bestResiYuv && m_bestResiYuv[i])
+        {
+            m_bestResiYuv[i]->destroy();
+            delete m_bestResiYuv[i];
+        }
+        if (m_bestRecoYuv && m_bestRecoYuv[i])
+        {
+            m_bestRecoYuv[i]->destroy();
+            delete m_bestRecoYuv[i];
+        }
+
+        if (m_tmpPredYuv && m_tmpPredYuv[i])
+        {
+            m_tmpPredYuv[i]->destroy();
+            delete m_tmpPredYuv[i];
+        }
+        for (int j = 0; j < MAX_PRED_TYPES; j++)
+        {
+            if (m_modePredYuv[j] && m_modePredYuv[j][i])
+            {
+                m_modePredYuv[j][i]->destroy();
+                delete m_modePredYuv[j][i];
+            }
+        }
+
+        if (m_tmpResiYuv && m_tmpResiYuv[i])
+        {
+            m_tmpResiYuv[i]->destroy();
+            delete m_tmpResiYuv[i];
+        }
+        if (m_tmpRecoYuv && m_tmpRecoYuv[i])
+        {
+            m_tmpRecoYuv[i]->destroy();
+            delete m_tmpRecoYuv[i];
+        }
+        if (m_bestMergeRecoYuv && m_bestMergeRecoYuv[i])
+        {
+            m_bestMergeRecoYuv[i]->destroy();
+            delete m_bestMergeRecoYuv[i];
+        }
+
+        if (m_origYuv && m_origYuv[i])
+        {
+            m_origYuv[i]->destroy();
+            delete m_origYuv[i];
+        }
+    }
+
+    delete [] m_memPool;
+    delete [] m_bestPredYuv;
+    delete [] m_bestResiYuv;
+    delete [] m_bestRecoYuv;
+    delete [] m_bestMergeRecoYuv;
+    delete [] m_tmpPredYuv;
+
+    for (int i = 0; i < MAX_PRED_TYPES; i++)
+        delete [] m_modePredYuv[i];
+
+    delete [] m_tmpResiYuv;
+    delete [] m_tmpRecoYuv;
+    delete [] m_origYuv;
+}
+
+void Analysis::loadCTUData(TComDataCU* parentCU)
+{
+    uint8_t cuRange[2]= {MIN_LOG2_CU_SIZE, g_log2Size[m_param->maxCUSize]};
+
+    // Initialize the coding blocks inside the CTB
+    for (int rangeIdx = cuRange[1], rangeCUIdx = 0; rangeIdx >= cuRange[0]; rangeIdx--)
+    {
+        uint32_t log2CUSize = rangeIdx;
+        int32_t  blockSize  = 1 << log2CUSize;
+        uint32_t b8Width    = 1 << (cuRange[1] - 3);
+        uint32_t sbWidth    = 1 << (cuRange[1] - rangeIdx);
+        int32_t last_level_flag = rangeIdx == cuRange[0];
+        for (uint32_t sb_y = 0; sb_y < sbWidth; sb_y++)
+        {
+            for (uint32_t sb_x = 0; sb_x < sbWidth; sb_x++)
+            {
+                uint32_t depth_idx = getDepthScanIdx(sb_x, sb_y, sbWidth);
+                uint32_t cuIdx = rangeCUIdx + depth_idx;
+                uint32_t child_idx = rangeCUIdx + sbWidth * sbWidth + (depth_idx << 2);
+                int32_t px = parentCU->getCUPelX() + sb_x * blockSize;
+                int32_t py = parentCU->getCUPelY() + sb_y * blockSize;
+                int32_t present_flag = px < parentCU->m_pic->m_origPicYuv->m_picWidth && py < parentCU->m_pic->m_origPicYuv->m_picHeight;
+                int32_t split_mandatory_flag = present_flag && !last_level_flag && (px + blockSize > parentCU->m_pic->m_origPicYuv->m_picWidth || py + blockSize > parentCU->m_pic->m_origPicYuv->m_picHeight);
+
+                CU *cu = parentCU->m_CULocalData + cuIdx;
+                cu->log2CUSize = log2CUSize;
+                cu->childIdx = child_idx;
+                cu->offset[0] = sb_x * blockSize;
+                cu->offset[1] = sb_y * blockSize;
+                cu->encodeIdx = getDepthScanIdx(cu->offset[0] >> 3, cu->offset[1] >> 3, b8Width);
+                cu->flags = 0;
+
+                CU_SET_FLAG(cu->flags, CU::PRESENT, present_flag);
+                CU_SET_FLAG(cu->flags, CU::SPLIT_MANDATORY | CU::SPLIT, split_mandatory_flag);
+                CU_SET_FLAG(cu->flags, CU::LEAF, last_level_flag);
+            }
+        }
+        rangeCUIdx += sbWidth * sbWidth;
+    }
+}
+
+void Analysis::compressCU(TComDataCU* cu)
+{
+    Frame* pic = cu->m_pic;
+    uint32_t cuAddr = cu->getAddr();
+
+    if (cu->m_slice->m_pps->bUseDQP)
+        m_bEncodeDQP = true;
+
+    // initialize CU data
+    m_bestCU[0]->initCU(pic, cuAddr);
+    m_tempCU[0]->initCU(pic, cuAddr);
+
+    // analysis of CU
+    uint32_t numPartition = cu->getTotalNumPart();
+    if (m_bestCU[0]->m_slice->m_sliceType == I_SLICE)
+    {
+        if (m_param->analysisMode == X265_ANALYSIS_LOAD && pic->m_intraData)
+        {
+            uint32_t zOrder = 0;
+            compressSharedIntraCTU(m_bestCU[0], m_tempCU[0], false, cu->m_CULocalData, 
+                &pic->m_intraData->depth[cuAddr * cu->m_numPartitions],
+                &pic->m_intraData->partSizes[cuAddr * cu->m_numPartitions],
+                &pic->m_intraData->modes[cuAddr * cu->m_numPartitions], zOrder);
+        }
+        else
+        {
+            compressIntraCU(m_bestCU[0], m_tempCU[0], false, cu->m_CULocalData);
+            if (m_param->analysisMode == X265_ANALYSIS_SAVE && pic->m_intraData)
+            {
+                memcpy(&pic->m_intraData->depth[cuAddr * cu->m_numPartitions], m_bestCU[0]->getDepth(), sizeof(uint8_t) * cu->getTotalNumPart());
+                memcpy(&pic->m_intraData->modes[cuAddr * cu->m_numPartitions], m_bestCU[0]->getLumaIntraDir(), sizeof(uint8_t) * cu->getTotalNumPart());
+                memcpy(&pic->m_intraData->partSizes[cuAddr * cu->m_numPartitions], m_bestCU[0]->getPartitionSize(), sizeof(char) * cu->getTotalNumPart());
+                pic->m_intraData->cuAddr[cuAddr] = cuAddr;
+                pic->m_intraData->poc[cuAddr]    = cu->m_pic->m_POC;
+            }
+        }
+        if (m_param->bLogCuStats || m_param->rc.bStatWrite)
+        {
+            uint32_t i = 0;
+            do
+            {
+                m_log->totalCu++;
+                uint32_t depth = cu->getDepth(i);
+                int next = numPartition >> (depth * 2);
+                m_log->qTreeIntraCnt[depth]++;
+                if (depth == g_maxCUDepth && cu->getPartitionSize(i) != SIZE_2Nx2N)
+                    m_log->cntIntraNxN++;
+                else
+                {
+                    m_log->cntIntra[depth]++;
+                    if (cu->getLumaIntraDir(i) > 1)
+                        m_log->cuIntraDistribution[depth][ANGULAR_MODE_ID]++;
+                    else
+                        m_log->cuIntraDistribution[depth][cu->getLumaIntraDir(i)]++;
+                }
+                i += next;
+            }
+            while (i < numPartition);
+        }
+    }
+    else
+    {
+        if (m_param->rdLevel < 5)
+        {
+            TComDataCU* outBestCU = NULL;
+
+            /* At the start of analysis, the best CU is a null pointer
+             * On return, it points to the CU encode with best chosen mode */
+            compressInterCU_rd0_4(outBestCU, m_tempCU[0], cu, 0, cu->m_CULocalData, false, 0, 4);
+        }
+        else
+            compressInterCU_rd5_6(m_bestCU[0], m_tempCU[0], 0, cu->m_CULocalData);
+
+        if (m_param->bLogCuStats || m_param->rc.bStatWrite)
+        {
+            uint32_t i = 0;
+            do
+            {
+                uint32_t depth = cu->getDepth(i);
+                m_log->cntTotalCu[depth]++;
+                int next = numPartition >> (depth * 2);
+                if (cu->isSkipped(i))
+                {
+                    m_log->cntSkipCu[depth]++;
+                    m_log->qTreeSkipCnt[depth]++;
+                }
+                else
+                {
+                    m_log->totalCu++;
+                    if (cu->getPredictionMode(0) == MODE_INTER)
+                    {
+                        m_log->cntInter[depth]++;
+                        m_log->qTreeInterCnt[depth]++;
+                        if (cu->getPartitionSize(0) < AMP_ID)
+                            m_log->cuInterDistribution[depth][cu->getPartitionSize(0)]++;
+                        else
+                            m_log->cuInterDistribution[depth][AMP_ID]++;
+                    }
+                    else if (cu->getPredictionMode(0) == MODE_INTRA)
+                    {
+                        m_log->qTreeIntraCnt[depth]++;
+                        if (depth == g_maxCUDepth && cu->getPartitionSize(0) == SIZE_NxN)
+                        {
+                            m_log->cntIntraNxN++;
+                        }
+                        else
+                        {
+                            m_log->cntIntra[depth]++;
+                            if (cu->getLumaIntraDir(0) > 1)
+                                m_log->cuIntraDistribution[depth][ANGULAR_MODE_ID]++;
+                            else
+                                m_log->cuIntraDistribution[depth][cu->getLumaIntraDir(0)]++;
+                        }
+                    }
+                }
+                i = i + next;
+            }
+            while (i < numPartition);
+        }
+    }
+}
+
+void Analysis::compressIntraCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu)
+{
+    //PPAScopeEvent(CompressIntraCU + depth);
+    Frame* pic = outBestCU->m_pic;
+    uint32_t cuAddr = outBestCU->getAddr();
+    uint32_t absPartIdx = outBestCU->getZorderIdxInCU();
+
+    if (depth == 0)
+        // get original YUV data from picture
+        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), cuAddr, absPartIdx);
+    else
+        // copy partition YUV from depth 0 CTU cache
+        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], absPartIdx);
+    Slice* slice = outTempCU->m_slice;
+    // We need to split, so don't try these modes.
+    int cu_split_flag = !(cu->flags & CU::LEAF);
+    int cu_unsplit_flag = !(cu->flags & CU::SPLIT_MANDATORY);
+
+    if (cu_unsplit_flag)
+    {
+        m_quant.setQPforQuant(outTempCU);
+        checkIntra(outTempCU, SIZE_2Nx2N, cu, NULL);
+        checkBestMode(outBestCU, outTempCU, depth);
+
+        if (depth == g_maxCUDepth)
+        {
+            checkIntra(outTempCU, SIZE_NxN, cu, NULL);
+            checkBestMode(outBestCU, outTempCU, depth);
+        }
+        else
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outBestCU, 0, depth);
+            outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+        }
+        if (m_rdCost.m_psyRd)
+            outBestCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits, outBestCU->m_psyEnergy);
+        else
+            outBestCU->m_totalRDCost  = m_rdCost.calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
+
+        // copy original YUV samples in lossless mode
+        if (outBestCU->isLosslessCoded(0))
+            fillOrigYUVBuffer(outBestCU, m_origYuv[depth]);
+    }
+
+    // further split
+    if (cu_split_flag)
+    {
+        uint32_t    nextDepth     = depth + 1;
+        TComDataCU* subBestPartCU = m_bestCU[nextDepth];
+        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++)
+        {
+            CU *child_cu = pic->getCU(cuAddr)->m_CULocalData + cu->childIdx + partUnitIdx;
+            int qp = outTempCU->getQP(0);
+            subBestPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+            if (child_cu->flags & CU::PRESENT)
+            {
+                subTempPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+                if (0 == partUnitIdx) //initialize RD with previous depth buffer
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+                else
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+
+                compressIntraCU(subBestPartCU, subTempPartCU, nextDepth, child_cu);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth); // Keep best part data to current temporary data.
+                m_bestRecoYuv[nextDepth]->copyToPartYuv(m_tmpRecoYuv[depth], subBestPartCU->getTotalNumPart() * partUnitIdx);
+            }
+            else
+            {
+                subBestPartCU->copyToPic(nextDepth);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth);
+            }
+        }
+        if (cu_unsplit_flag)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outTempCU, 0, depth);
+            outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+        }
+
+        if (m_rdCost.m_psyRd)
+            outTempCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits, outTempCU->m_psyEnergy);
+        else
+            outTempCU->m_totalRDCost = m_rdCost.calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+        if (depth == slice->m_pps->maxCuDQPDepth && slice->m_pps->bUseDQP)
+        {
+            bool hasResidual = false;
+            for (uint32_t blkIdx = 0; blkIdx < outTempCU->getTotalNumPart(); blkIdx++)
+            {
+                if (outTempCU->getCbf(blkIdx, TEXT_LUMA) || outTempCU->getCbf(blkIdx, TEXT_CHROMA_U) ||
+                    outTempCU->getCbf(blkIdx, TEXT_CHROMA_V))
+                {
+                    hasResidual = true;
+                    break;
+                }
+            }
+
+            uint32_t targetPartIdx = 0;
+            if (hasResidual)
+            {
+                bool foundNonZeroCbf = false;
+                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
+                X265_CHECK(foundNonZeroCbf, "expected to find non-zero CBF\n");
+            }
+            else
+                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
+        }
+
+        m_rdEntropyCoders[nextDepth][CI_NEXT_BEST].store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+        checkBestMode(outBestCU, outTempCU, depth); // RD compare current CU against split
+    }
+
+    // TODO: write the best CTU at the end of complete CTU analysis
+    outBestCU->copyToPic(depth); // Copy Best data to Picture for next partition prediction.
+    if (!cu_unsplit_flag)
+        return;
+    // Copy Yuv data to picture Yuv
+    m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+
+#if CHECKED_BUILD || _DEBUG
+    X265_CHECK(outBestCU->getPartitionSize(0) != SIZE_NONE, "no best partition size\n");
+    X265_CHECK(outBestCU->getPredictionMode(0) != MODE_NONE, "no best partition mode\n");
+    if (m_rdCost.m_psyRd)
+    {
+        X265_CHECK(outBestCU->m_totalPsyCost != MAX_INT64, "no best partition cost\n");
+    }
+    else
+    {
+        X265_CHECK(outBestCU->m_totalRDCost != MAX_INT64, "no best partition cost\n");
+    }
+#endif
+}
+
+void Analysis::compressSharedIntraCTU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu, uint8_t* sharedDepth,
+                                      char* sharedPartSizes, uint8_t* sharedModes, uint32_t &zOrder)
+{
+    Frame* pic = outBestCU->m_pic;
+
+    // if current depth == shared depth then skip further splitting.
+    bool bSubBranch = true;
+
+    // index to g_depthInc array to increment zOrder offset to next depth
+    int32_t ctuToDepthIndex = g_maxCUDepth - 1;
+
+    if (depth)
+        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], outBestCU->getZorderIdxInCU());
+    else
+        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU());
+
+    Slice* slice = outTempCU->m_slice;
+    int32_t cu_split_flag = !(cu->flags & CU::LEAF);
+    int32_t cu_unsplit_flag = !(cu->flags & CU::SPLIT_MANDATORY);
+
+    if (cu_unsplit_flag && ((zOrder == outBestCU->getZorderIdxInCU()) && (depth == sharedDepth[zOrder])))
+    {
+        m_quant.setQPforQuant(outTempCU);
+        checkIntra(outTempCU, (PartSize)sharedPartSizes[zOrder], cu, &sharedModes[zOrder]);
+        checkBestMode(outBestCU, outTempCU, depth);
+
+        if (depth != g_maxCUDepth)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outBestCU, 0, depth);
+            outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits();
+        }
+
+        // set current best CU cost to 0 marking as best CU present in shared CU data
+        outBestCU->m_totalRDCost = 0;
+        bSubBranch = false;
+
+        // increment zOrder offset to point to next best depth in sharedDepth buffer
+        zOrder += g_depthInc[ctuToDepthIndex][sharedDepth[zOrder]];
+    }
+
+    // copy original YUV samples in lossless mode
+    if (outBestCU->isLosslessCoded(0))
+        fillOrigYUVBuffer(outBestCU, m_origYuv[depth]);
+
+    // further split
+    if (cu_split_flag && bSubBranch)
+    {
+        uint32_t    nextDepth     = depth + 1;
+        TComDataCU* subBestPartCU = m_bestCU[nextDepth];
+        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++)
+        {
+            CU *child_cu = pic->getCU(outTempCU->getAddr())->m_CULocalData + cu->childIdx + partUnitIdx;
+            int qp = outTempCU->getQP(0);
+            subBestPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+            if (child_cu->flags & CU::PRESENT)
+            {
+                subTempPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+
+                if (partUnitIdx) // initialize RD with previous depth buffer
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+                else
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+                // set current best CU cost to 1 marking as non-best CU by default
+                subTempPartCU->m_totalRDCost = 1;
+
+                compressSharedIntraCTU(subBestPartCU, subTempPartCU, nextDepth, child_cu, sharedDepth, sharedPartSizes, sharedModes, zOrder);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth); // Keep best part data to current temporary data.
+
+                if (!subBestPartCU->m_totalRDCost) // if cost is 0, CU is best CU
+                    outTempCU->m_totalRDCost = 0;  // set outTempCU cost to 0, so later check will use this CU as best CU
+
+                m_bestRecoYuv[nextDepth]->copyToPartYuv(m_tmpRecoYuv[depth], subBestPartCU->getTotalNumPart() * partUnitIdx);
+            }
+            else
+            {
+                subBestPartCU->copyToPic(nextDepth);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth);
+
+                // increment zOrder offset to point to next best depth in sharedDepth buffer
+                zOrder += g_depthInc[ctuToDepthIndex][nextDepth];
+            }
+        }
+
+        if (cu_unsplit_flag)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outTempCU, 0, depth);
+            outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+        }
+        if (depth == slice->m_pps->maxCuDQPDepth && slice->m_pps->bUseDQP)
+        {
+            bool hasResidual = false;
+            for (uint32_t blkIdx = 0; blkIdx < outTempCU->getTotalNumPart(); blkIdx++)
+            {
+                if (outTempCU->getCbf(blkIdx, TEXT_LUMA) || outTempCU->getCbf(blkIdx, TEXT_CHROMA_U) ||
+                    outTempCU->getCbf(blkIdx, TEXT_CHROMA_V))
+                {
+                    hasResidual = true;
+                    break;
+                }
+            }
+
+            uint32_t targetPartIdx = 0;
+            if (hasResidual)
+            {
+                bool foundNonZeroCbf = false;
+                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
+                X265_CHECK(foundNonZeroCbf, "expected to find non-zero CBF\n");
+            }
+            else
+                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
+        }
+        m_rdEntropyCoders[nextDepth][CI_NEXT_BEST].store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+        checkBestMode(outBestCU, outTempCU, depth);
+    }
+    outBestCU->copyToPic(depth);
+    if (!cu_unsplit_flag)
+        return;
+    m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), outBestCU->getAddr(), outBestCU->getZorderIdxInCU());
+
+#if CHECKED_BUILD || _DEBUG
+    X265_CHECK(outBestCU->getPartitionSize(0) != SIZE_NONE, "no best partition size\n");
+    X265_CHECK(outBestCU->getPredictionMode(0) != MODE_NONE, "no best partition mode\n");
+    if (m_rdCost.m_psyRd)
+    {
+        X265_CHECK(outBestCU->m_totalPsyCost != MAX_INT64, "no best partition cost\n");
+    }
+    else
+    {
+        X265_CHECK(outBestCU->m_totalRDCost != MAX_INT64, "no best partition cost\n");
+    }
+#endif
+}
+
+void Analysis::checkIntra(TComDataCU*& outTempCU, PartSize partSize, CU *cu, uint8_t* sharedModes)
+{
+    //PPAScopeEvent(CheckRDCostIntra + depth);
+    uint32_t depth = g_log2Size[m_param->maxCUSize] - cu->log2CUSize;
+    outTempCU->setPartSizeSubParts(partSize, 0, depth);
+    outTempCU->setPredModeSubParts(MODE_INTRA, 0, depth);
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+
+    uint32_t tuDepthRange[2];
+    outTempCU->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+
+    if (sharedModes)
+        sharedEstIntraPredQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth], tuDepthRange, sharedModes);
+    else
+        estIntraPredQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth], tuDepthRange);
+
+    estIntraPredChromaQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
+
+    m_entropyCoder->resetBits();
+    if (outTempCU->m_slice->m_pps->bTransquantBypassEnabled)
+        m_entropyCoder->codeCUTransquantBypassFlag(outTempCU->getCUTransquantBypass(0));
+
+    m_entropyCoder->codePartSize(outTempCU, 0, depth);
+    m_entropyCoder->codePredInfo(outTempCU, 0);
+    outTempCU->m_mvBits = m_entropyCoder->getNumberOfWrittenBits();
+
+    // Encode Coefficients
+    bool bCodeDQP = m_bEncodeDQP;
+    m_entropyCoder->codeCoeff(outTempCU, 0, depth, bCodeDQP, tuDepthRange);
+    m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+    outTempCU->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
+    outTempCU->m_coeffBits = outTempCU->m_totalBits - outTempCU->m_mvBits;
+
+    /* TODO: add chroma psyEnergy also to psyCost */
+    if (m_rdCost.m_psyRd)
+    {
+        int part = outTempCU->getLog2CUSize(0) - 2;
+        outTempCU->m_psyEnergy = m_rdCost.psyCost(part, m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
+                                                  m_tmpRecoYuv[depth]->getLumaAddr(), m_tmpRecoYuv[depth]->getStride());
+        outTempCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits, outTempCU->m_psyEnergy);
+    }
+    else
+        outTempCU->m_totalRDCost = m_rdCost.calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+    checkDQP(outTempCU);
+}
+
+void Analysis::compressInterCU_rd0_4(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComDataCU* parentCU, uint32_t depth, CU *cu, 
+                                     int bInsidePicture, uint32_t PartitionIndex, uint32_t minDepth)
+{
+    Frame* pic = outTempCU->m_pic;
+    uint32_t cuAddr = outTempCU->getAddr();
+    uint32_t absPartIdx = outTempCU->getZorderIdxInCU();
+
+    if (depth)
+        // copy partition YUV from depth 0 CTU cache
+        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], absPartIdx);
+    else
+        // get original YUV data from picture
+        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), cuAddr, absPartIdx);
+
+    // variables for fast encoder decision
+    bool bSubBranch = true;
+    int qp = outTempCU->getQP(0);
+
+    int bInsidePictureParent = bInsidePicture;
+
+    Slice* slice = outTempCU->m_slice;
+    int cu_split_flag = !(cu->flags & CU::LEAF);
+    int cu_unsplit_flag = !(cu->flags & CU::SPLIT_MANDATORY);
+
+    if (!depth && !m_param->rdLevel)
+        m_origYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, 0);
+
+    // We need to split, so don't try these modes.
+    if (cu_unsplit_flag && !bInsidePictureParent)
+    {
+        TComDataCU* colocated0 = slice->m_numRefIdx[0] > 0 ? slice->m_refPicList[0][0]->getCU(cuAddr) : NULL;
+        TComDataCU* colocated1 = slice->m_numRefIdx[1] > 0 ? slice->m_refPicList[1][0]->getCU(cuAddr) : NULL;
+        char currentQP = outTempCU->getQP(0);
+        char previousQP = colocated0->getQP(0);
+        uint32_t delta = 0, minDepth0 = 4, minDepth1 = 4;
+        uint32_t sum0 = 0, sum1 = 0;
+        uint32_t numPartitions = outTempCU->getTotalNumPart();
+        for (uint32_t i = 0; i < numPartitions; i = i + 4)
+        {
+            uint32_t j = absPartIdx + i;
+            if (colocated0 && colocated0->getDepth(j) < minDepth0)
+                minDepth0 = colocated0->getDepth(j);
+            if (colocated1 && colocated1->getDepth(j) < minDepth1)
+                minDepth1 = colocated1->getDepth(j);
+            if (colocated0)
+                sum0 += (colocated0->getDepth(j) * 4);
+            if (colocated1)
+                sum1 += (colocated1->getDepth(j) * 4);
+        }
+
+        uint32_t avgDepth2 = (sum0 + sum1) / numPartitions;
+        minDepth = X265_MIN(minDepth0, minDepth1);
+        if (((currentQP - previousQP) < 0) || (((currentQP - previousQP) >= 0) && ((avgDepth2 - 2 * minDepth) > 1)))
+            delta = 0;
+        else
+            delta = 1;
+        if (minDepth > 0)
+            minDepth = minDepth - delta;
+    }
+    if (depth >= minDepth)
+    {
+        if (cu_unsplit_flag)
+        {
+            /* Initialise all Mode-CUs based on parentCU */
+            if (depth)
+            {
+                m_interCU_2Nx2N[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+                m_interCU_2NxN[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+                m_interCU_Nx2N[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+                m_intraInInterCU[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+                m_mergeCU[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+                m_bestMergeCU[depth]->initSubCU(parentCU, PartitionIndex, depth, qp);
+            }
+            else
+            {
+                m_interCU_2Nx2N[depth]->initCU(pic, cuAddr);
+                m_interCU_Nx2N[depth]->initCU(pic, cuAddr);
+                m_interCU_2NxN[depth]->initCU(pic, cuAddr);
+                m_intraInInterCU[depth]->initCU(pic, cuAddr);
+                m_mergeCU[depth]->initCU(pic, cuAddr);
+                m_bestMergeCU[depth]->initCU(pic, cuAddr);
+            }
+
+            /* Compute Merge Cost */
+            checkMerge2Nx2N_rd0_4(m_bestMergeCU[depth], m_mergeCU[depth], m_modePredYuv[3][depth], m_bestMergeRecoYuv[depth]);
+            bool earlyskip = false;
+            if (m_param->rdLevel >= 1)
+                earlyskip = (m_param->bEnableEarlySkip && m_bestMergeCU[depth]->isSkipped(0));
+
+            if (!earlyskip)
+            {
+                /* Compute 2Nx2N mode costs */
+                checkInter_rd0_4(m_interCU_2Nx2N[depth], m_modePredYuv[0][depth], SIZE_2Nx2N);
+
+                /* initialise outBestCU to 2Nx2N */
+                outBestCU = m_interCU_2Nx2N[depth];
+                std::swap(m_bestPredYuv[depth], m_modePredYuv[0][depth]);
+
+                /* Compute Rect costs */
+                if (m_param->bEnableRectInter)
+                {
+                    checkInter_rd0_4(m_interCU_Nx2N[depth], m_modePredYuv[1][depth], SIZE_Nx2N);
+                    checkInter_rd0_4(m_interCU_2NxN[depth], m_modePredYuv[2][depth], SIZE_2NxN);
+                    if (m_interCU_Nx2N[depth]->m_sa8dCost < outBestCU->m_sa8dCost)
+                    {
+                        outBestCU = m_interCU_Nx2N[depth];
+                        std::swap(m_bestPredYuv[depth], m_modePredYuv[1][depth]);
+                    }
+                    if (m_interCU_2NxN[depth]->m_sa8dCost < outBestCU->m_sa8dCost)
+                    {
+                        outBestCU = m_interCU_2NxN[depth];
+                        std::swap(m_bestPredYuv[depth], m_modePredYuv[2][depth]);
+                    }
+                }
+
+                if (m_param->rdLevel > 2)
+                {
+                    // calculate the motion compensation for chroma for the best mode selected
+                    int numPart = outBestCU->getNumPartInter();
+                    for (int partIdx = 0; partIdx < numPart; partIdx++)
+                    {
+                        prepMotionCompensation(outBestCU, partIdx);
+                        motionCompensation(m_bestPredYuv[depth], false, true);
+                    }
+
+                    encodeResAndCalcRdInterCU(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],
+                                              m_bestResiYuv[depth], m_bestRecoYuv[depth]);
+                    uint64_t bestMergeCost = m_rdCost.m_psyRd ? m_bestMergeCU[depth]->m_totalPsyCost : m_bestMergeCU[depth]->m_totalRDCost;
+                    uint64_t bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+                    if (bestMergeCost < bestCost)
+                    {
+                        outBestCU = m_bestMergeCU[depth];
+                        std::swap(m_bestPredYuv[depth], m_modePredYuv[3][depth]);
+                        std::swap(m_bestRecoYuv[depth], m_bestMergeRecoYuv[depth]);
+                    }
+                    else
+                        m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+                }
+
+                /* Check for Intra in inter frames only if it is a P-slice */
+                if (slice->m_sliceType == P_SLICE)
+                {
+                    /* compute intra cost */
+                    bool bdoIntra = true;
+                    if (m_param->rdLevel > 2)
+                        bdoIntra = outBestCU->getCbf(0, TEXT_LUMA) || outBestCU->getCbf(0, TEXT_CHROMA_U) ||
+                                   outBestCU->getCbf(0, TEXT_CHROMA_V);
+
+                    if (bdoIntra)
+                    {
+                        checkIntraInInter_rd0_4(m_intraInInterCU[depth], SIZE_2Nx2N);
+                        uint64_t intraInInterCost, bestCost;
+                        if (m_param->rdLevel > 2)
+                        {
+                            encodeIntraInInter(m_intraInInterCU[depth], m_origYuv[depth], m_modePredYuv[5][depth],
+                                               m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
+                            intraInInterCost = m_rdCost.m_psyRd ? m_intraInInterCU[depth]->m_totalPsyCost : m_intraInInterCU[depth]->m_totalRDCost;
+                            bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+                        }
+                        else
+                        {
+                            intraInInterCost = m_intraInInterCU[depth]->m_sa8dCost;
+                            bestCost = outBestCU->m_sa8dCost;
+                        }
+                        if (intraInInterCost < bestCost)
+                        {
+                            outBestCU = m_intraInInterCU[depth];
+                            std::swap(m_bestPredYuv[depth], m_modePredYuv[5][depth]);
+                            std::swap(m_bestRecoYuv[depth], m_tmpRecoYuv[depth]);
+                            if (m_param->rdLevel > 2)
+                                m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+                        }
+                    }
+                }
+                if (m_param->rdLevel == 2)
+                {
+                    if (m_bestMergeCU[depth]->m_sa8dCost < outBestCU->m_sa8dCost)
+                    {
+                        outBestCU = m_bestMergeCU[depth];
+                        std::swap(m_bestPredYuv[depth], m_modePredYuv[3][depth]);
+                        std::swap(m_bestRecoYuv[depth], m_bestMergeRecoYuv[depth]);
+                    }
+                    else if (outBestCU->getPredictionMode(0) == MODE_INTER)
+                    {
+                        int numPart = outBestCU->getNumPartInter();
+                        for (int partIdx = 0; partIdx < numPart; partIdx++)
+                        {
+                            prepMotionCompensation(outBestCU, partIdx);
+                            motionCompensation(m_bestPredYuv[depth], false, true);
+                        }
+
+                        encodeResAndCalcRdInterCU(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],
+                                                  m_bestResiYuv[depth], m_bestRecoYuv[depth]);
+                        m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+                    }
+                    else if (outBestCU->getPredictionMode(0) == MODE_INTRA)
+                    {
+                        encodeIntraInInter(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth],  m_bestRecoYuv[depth]);
+                        m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+                    }
+                }
+                else if (m_param->rdLevel == 1)
+                {
+                    if (m_bestMergeCU[depth]->m_sa8dCost < outBestCU->m_sa8dCost)
+                    {
+                        outBestCU = m_bestMergeCU[depth];
+                        std::swap(m_bestPredYuv[depth], m_modePredYuv[3][depth]);
+                        std::swap(m_bestRecoYuv[depth], m_bestMergeRecoYuv[depth]);
+                    }
+                    else if (outBestCU->getPredictionMode(0) == MODE_INTER)
+                    {
+                        int numPart = outBestCU->getNumPartInter();
+                        for (int partIdx = 0; partIdx < numPart; partIdx++)
+                        {
+                            prepMotionCompensation(outBestCU, partIdx);
+                            motionCompensation(m_bestPredYuv[depth], false, true);
+                        }
+
+                        m_tmpResiYuv[depth]->subtract(m_origYuv[depth], m_bestPredYuv[depth], outBestCU->getLog2CUSize(0));
+                        generateCoeffRecon(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth], m_bestRecoYuv[depth]);
+                    }
+                    else
+                        generateCoeffRecon(outBestCU, m_origYuv[depth], m_bestPredYuv[depth], m_tmpResiYuv[depth], m_bestRecoYuv[depth]);
+                }
+                else if (!m_param->rdLevel)
+                {
+                    if (outBestCU->getPredictionMode(0) == MODE_INTER)
+                    {
+                        int numPart = outBestCU->getNumPartInter();
+                        for (int partIdx = 0; partIdx < numPart; partIdx++)
+                        {
+                            prepMotionCompensation(outBestCU, partIdx);
+                            motionCompensation(m_bestPredYuv[depth], false, true);
+                        }
+                    }
+                }
+            }
+            else
+            {
+                outBestCU = m_bestMergeCU[depth];
+                std::swap(m_bestPredYuv[depth], m_modePredYuv[3][depth]);
+                std::swap(m_bestRecoYuv[depth], m_bestMergeRecoYuv[depth]);
+            }
+
+            if (m_param->rdLevel > 0) // checkDQP can be done only after residual encoding is done
+                checkDQP(outBestCU);
+
+            /* Disable recursive analysis for whole CUs temporarily */
+            if ((outBestCU != 0) && (outBestCU->isSkipped(0)))
+                bSubBranch = false;
+            else
+                bSubBranch = true;
+
+            if (m_param->rdLevel > 1)
+            {
+                if (depth < g_maxCUDepth)
+                {
+                    m_entropyCoder->resetBits();
+                    m_entropyCoder->codeSplitFlag(outBestCU, 0, depth);
+                    outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+                }
+                if (m_rdCost.m_psyRd)
+                    outBestCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits, outBestCU->m_psyEnergy);
+                else
+                    outBestCU->m_totalRDCost = m_rdCost.calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
+            }
+
+            // copy original YUV samples in lossless mode
+            if (outBestCU->isLosslessCoded(0))
+                fillOrigYUVBuffer(outBestCU, m_origYuv[depth]);
+        }
+    }
+
+    // further split
+    if (bSubBranch && cu_split_flag)
+    {
+        // early exit when the RD cost of best mode at depth n is less than the sum of avgerage of RD cost of the neighbour
+        // CU's(above, aboveleft, aboveright, left, colocated) and avg cost of that CU at depth "n" with weightage for each quantity
+        if (outBestCU && !(depth < minDepth))
+        {
+            uint64_t totalCostNeigh = 0, totalCostCU = 0, totalCountNeigh = 0, totalCountCU = 0;
+            TComDataCU* above = outTempCU->getCUAbove();
+            TComDataCU* aboveLeft = outTempCU->getCUAboveLeft();
+            TComDataCU* aboveRight = outTempCU->getCUAboveRight();
+            TComDataCU* left = outTempCU->getCULeft();
+            TComDataCU* rootCU = pic->getPicSym()->getCU(cuAddr);
+
+            totalCostCU += rootCU->m_avgCost[depth] * rootCU->m_count[depth];
+            totalCountCU += rootCU->m_count[depth];
+            if (above)
+            {
+                totalCostNeigh += above->m_avgCost[depth] * above->m_count[depth];
+                totalCountNeigh += above->m_count[depth];
+            }
+            if (aboveLeft)
+            {
+                totalCostNeigh += aboveLeft->m_avgCost[depth] * aboveLeft->m_count[depth];
+                totalCountNeigh += aboveLeft->m_count[depth];
+            }
+            if (aboveRight)
+            {
+                totalCostNeigh += aboveRight->m_avgCost[depth] * aboveRight->m_count[depth];
+                totalCountNeigh += aboveRight->m_count[depth];
+            }
+            if (left)
+            {
+                totalCostNeigh += left->m_avgCost[depth] * left->m_count[depth];
+                totalCountNeigh += left->m_count[depth];
+            }
+
+            // give 60% weight to all CU's and 40% weight to neighbour CU's
+            uint64_t avgCost = 0;
+            if (totalCountNeigh + totalCountCU)
+                avgCost = ((3 * totalCostCU) + (2 * totalCostNeigh)) / ((3 * totalCountCU) + (2 * totalCountNeigh));
+            uint64_t bestavgCost = 0;
+            if (m_param->rdLevel > 1)
+                bestavgCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+            else
+                bestavgCost = outBestCU->m_totalRDCost;
+
+            if (bestavgCost < avgCost && avgCost && depth)
+            {
+                /* Copy Best data to Picture for next partition prediction */
+                outBestCU->copyToPic(depth);
+
+                /* Copy Yuv data to picture Yuv */
+                if (m_param->rdLevel)
+                    m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+                return;
+            }
+        }
+        outTempCU->setQPSubParts(qp, 0, depth);
+        uint32_t    nextDepth = depth + 1;
+        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++)
+        {
+            CU *child_cu = pic->getCU(cuAddr)->m_CULocalData + cu->childIdx + partUnitIdx;
+
+            TComDataCU* subBestPartCU = NULL;
+            subTempPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+
+            if (child_cu->flags & CU::PRESENT)
+            {
+                if (partUnitIdx) // initialize RD with previous depth buffer
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+                else
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+                compressInterCU_rd0_4(subBestPartCU, subTempPartCU, outTempCU, nextDepth, child_cu, cu_unsplit_flag, partUnitIdx, minDepth);
+
+                if (subBestPartCU->getPredictionMode(0) != MODE_INTRA)
+                {
+                    uint64_t tempavgCost = 0;
+                    if (m_param->rdLevel > 1)
+                        tempavgCost = m_rdCost.m_psyRd ? subBestPartCU->m_totalPsyCost : subBestPartCU->m_totalRDCost;
+                    else
+                        tempavgCost = subBestPartCU->m_totalRDCost;
+                    TComDataCU* rootCU = pic->getPicSym()->getCU(cuAddr);
+                    uint64_t temp = rootCU->m_avgCost[nextDepth] * rootCU->m_count[nextDepth];
+                    rootCU->m_count[nextDepth] += 1;
+                    rootCU->m_avgCost[nextDepth] = (temp + tempavgCost) / rootCU->m_count[nextDepth];
+                }
+
+                /* Adding costs from best SUbCUs */
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth, true); // Keep best part data to current temporary data.
+                if (m_param->rdLevel)
+                    m_bestRecoYuv[nextDepth]->copyToPartYuv(m_tmpRecoYuv[depth], subBestPartCU->getTotalNumPart() * partUnitIdx);
+                else
+                    m_bestPredYuv[nextDepth]->copyToPartYuv(m_tmpPredYuv[depth], subBestPartCU->getTotalNumPart() * partUnitIdx);
+            }
+            else
+            {
+                subTempPartCU->copyToPic(nextDepth);
+                outTempCU->copyPartFrom(subTempPartCU, partUnitIdx, nextDepth, false);
+            }
+        }
+
+        if (m_param->rdLevel > 1)
+        {
+            if (cu_unsplit_flag)
+            {
+                m_entropyCoder->resetBits();
+                m_entropyCoder->codeSplitFlag(outTempCU, 0, depth);
+                outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+            }
+            if (m_rdCost.m_psyRd)
+                outTempCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits, outTempCU->m_psyEnergy);
+            else
+                outTempCU->m_totalRDCost = m_rdCost.calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+        }
+        else
+            outTempCU->m_sa8dCost = m_rdCost.calcRdSADCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+        if (depth == slice->m_pps->maxCuDQPDepth && slice->m_pps->bUseDQP)
+        {
+            bool hasResidual = false;
+            for (uint32_t blkIdx = 0; blkIdx < outTempCU->getTotalNumPart(); blkIdx++)
+            {
+                if (outTempCU->getCbf(blkIdx, TEXT_LUMA) || outTempCU->getCbf(blkIdx, TEXT_CHROMA_U) ||
+                    outTempCU->getCbf(blkIdx, TEXT_CHROMA_V))
+                {
+                    hasResidual = true;
+                    break;
+                }
+            }
+
+            uint32_t targetPartIdx = 0;
+            if (hasResidual)
+            {
+                bool foundNonZeroCbf = false;
+                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
+                X265_CHECK(foundNonZeroCbf, "setQPSubCUs did not find non-zero Cbf\n");
+            }
+            else
+                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
+        }
+
+        m_rdEntropyCoders[nextDepth][CI_NEXT_BEST].store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+
+        /* If Best Mode is not NULL; then compare costs. Else assign best mode to Sub-CU costs
+         * Copy recon data from Temp structure to Best structure */
+        if (outBestCU)
+        {
+            if (!depth)
+            {
+                uint64_t tempavgCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+                TComDataCU* rootCU = pic->getPicSym()->getCU(cuAddr);
+                uint64_t temp = rootCU->m_avgCost[depth] * rootCU->m_count[depth];
+                rootCU->m_count[depth] += 1;
+                rootCU->m_avgCost[depth] = (temp + tempavgCost) / rootCU->m_count[depth];
+            }
+
+            uint64_t tempCost = m_rdCost.m_psyRd ? outTempCU->m_totalPsyCost : outTempCU->m_totalRDCost;
+            uint64_t bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost; 
+            if (tempCost < bestCost)
+            {
+                outBestCU = outTempCU;
+                std::swap(m_bestRecoYuv[depth], m_tmpRecoYuv[depth]);
+                std::swap(m_bestPredYuv[depth], m_tmpPredYuv[depth]);
+            }
+        }
+        else
+        {
+            outBestCU = outTempCU;
+            std::swap(m_bestRecoYuv[depth], m_tmpRecoYuv[depth]);
+            std::swap(m_bestPredYuv[depth], m_tmpPredYuv[depth]);
+        }
+    }
+
+    /* Copy Best data to Picture for next partition prediction. */
+    outBestCU->copyToPic(depth);
+
+    if (!m_param->rdLevel && !depth)
+        encodeResidue(outBestCU, outBestCU, 0, 0);
+    else if (m_param->rdLevel)
+    {
+        /* Copy Yuv data to picture Yuv */
+        if (cu_unsplit_flag)
+            m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+    }
+
+#if CHECKED_BUILD || _DEBUG
+    /* Assert if Best prediction mode is NONE
+     * Selected mode's RD-cost must be not MAX_INT64 */
+    if (cu_unsplit_flag)
+    {
+        X265_CHECK(outBestCU->getPartitionSize(0) != SIZE_NONE, "no best prediction size\n");
+        X265_CHECK(outBestCU->getPredictionMode(0) != MODE_NONE, "no best prediction mode\n");
+        if (m_param->rdLevel > 1)
+        {
+            if (m_rdCost.m_psyRd)
+            {
+                X265_CHECK(outBestCU->m_totalPsyCost != MAX_INT64, "no best partition cost\n");
+            }
+            else
+            {
+                X265_CHECK(outBestCU->m_totalRDCost != MAX_INT64, "no best partition cost\n");
+            }
+        }
+        else
+        {
+            X265_CHECK(outBestCU->m_sa8dCost != MAX_INT64, "no best partition cost\n");
+        }
+    }
+#endif
+
+    x265_emms();
+}
+
+void Analysis::compressInterCU_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu, PartSize parentSize)
+{
+    //PPAScopeEvent(CompressCU + depth);
+
+    Frame* pic = outBestCU->m_pic;
+    uint32_t cuAddr = outBestCU->getAddr();
+    uint32_t absPartIdx = outBestCU->getZorderIdxInCU();
+
+    if (depth)
+        // copy partition YUV from depth 0 CTU cache
+        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], absPartIdx);
+    else
+        // get original YUV data from picture
+        m_origYuv[depth]->copyFromPicYuv(pic->getPicYuvOrg(), cuAddr, absPartIdx);
+
+    // variable for Cbf fast mode PU decision
+    bool doNotBlockPu = true;
+    bool earlyDetectionSkipMode = false;
+
+    Slice* slice = outTempCU->m_slice;
+    int cu_split_flag = !(cu->flags & CU::LEAF);
+    int cu_unsplit_flag = !(cu->flags & CU::SPLIT_MANDATORY);
+
+    // We need to split, so don't try these modes.
+    if (cu_unsplit_flag)
+    {
+        m_quant.setQPforQuant(outTempCU);
+
+        // do inter modes, SKIP and 2Nx2N
+        if (slice->m_sliceType != I_SLICE)
+        {
+            // by Merge for inter_2Nx2N
+            checkMerge2Nx2N_rd5_6(outBestCU, outTempCU, &earlyDetectionSkipMode, m_bestPredYuv[depth], m_bestRecoYuv[depth]);
+
+            outTempCU->initEstData();
+
+            if (!m_param->bEnableEarlySkip)
+            {
+                // 2Nx2N, NxN
+                checkInter_rd5_6(outBestCU, outTempCU, SIZE_2Nx2N);
+                outTempCU->initEstData();
+                if (m_param->bEnableCbfFastMode)
+                    doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+            }
+        }
+
+        if (!earlyDetectionSkipMode)
+        {
+            outTempCU->initEstData();
+
+            // do inter modes, NxN, 2NxN, and Nx2N
+            if (slice->m_sliceType != I_SLICE)
+            {
+                // 2Nx2N, NxN
+                if (cu->log2CUSize != 3 && depth == g_maxCUDepth && doNotBlockPu)
+                {
+                    checkInter_rd5_6(outBestCU, outTempCU, SIZE_NxN);
+                    outTempCU->initEstData();
+                }
+
+                if (m_param->bEnableRectInter)
+                {
+                    // 2NxN, Nx2N
+                    if (doNotBlockPu)
+                    {
+                        checkInter_rd5_6(outBestCU, outTempCU, SIZE_Nx2N);
+                        outTempCU->initEstData();
+                        if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_Nx2N)
+                            doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                    }
+                    if (doNotBlockPu)
+                    {
+                        checkInter_rd5_6(outBestCU, outTempCU, SIZE_2NxN);
+                        outTempCU->initEstData();
+                        if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxN)
+                            doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                    }
+                }
+
+                // Try AMP (SIZE_2NxnU, SIZE_2NxnD, SIZE_nLx2N, SIZE_nRx2N)
+                if (slice->m_sps->maxAMPDepth > depth)
+                {
+                    bool bTestAMP_Hor = false, bTestAMP_Ver = false;
+                    bool bTestMergeAMP_Hor = false, bTestMergeAMP_Ver = false;
+
+                    deriveTestModeAMP(outBestCU, parentSize, bTestAMP_Hor, bTestAMP_Ver, bTestMergeAMP_Hor, bTestMergeAMP_Ver);
+
+                    // Do horizontal AMP
+                    if (bTestAMP_Hor)
+                    {
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_2NxnU);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnU)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_2NxnD);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnD)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                    }
+                    else if (bTestMergeAMP_Hor)
+                    {
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_2NxnU, true);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnU)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_2NxnD, true);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_2NxnD)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                    }
+
+                    // Do horizontal AMP
+                    if (bTestAMP_Ver)
+                    {
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_nLx2N);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_nLx2N)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_nRx2N);
+                            outTempCU->initEstData();
+                        }
+                    }
+                    else if (bTestMergeAMP_Ver)
+                    {
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_nLx2N, true);
+                            outTempCU->initEstData();
+                            if (m_param->bEnableCbfFastMode && outBestCU->getPartitionSize(0) == SIZE_nLx2N)
+                                doNotBlockPu = outBestCU->getQtRootCbf(0) != 0;
+                        }
+                        if (doNotBlockPu)
+                        {
+                            checkInter_rd5_6(outBestCU, outTempCU, SIZE_nRx2N, true);
+                            outTempCU->initEstData();
+                        }
+                    }
+                }
+            }
+
+            // speedup for inter frames, avoid intra in special cases
+            bool doIntra = slice->m_sliceType == B_SLICE ? !!m_param->bIntraInBFrames : true;
+            if ((outBestCU->getCbf(0, TEXT_LUMA)     != 0   ||
+                 outBestCU->getCbf(0, TEXT_CHROMA_U) != 0   ||
+                 outBestCU->getCbf(0, TEXT_CHROMA_V) != 0)  && doIntra)
+            {
+                checkIntraInInter_rd5_6(outBestCU, outTempCU, SIZE_2Nx2N);
+                outTempCU->initEstData();
+
+                if (depth == g_maxCUDepth && cu->log2CUSize > slice->m_sps->quadtreeTULog2MinSize)
+                {
+                    checkIntraInInter_rd5_6(outBestCU, outTempCU, SIZE_NxN);
+                    outTempCU->initEstData();
+                }
+            }
+        }
+
+        if (depth < g_maxCUDepth)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outBestCU, 0, depth);
+            outBestCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+        }
+        if (m_rdCost.m_psyRd)
+            outBestCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits, outBestCU->m_psyEnergy);
+        else
+            outBestCU->m_totalRDCost = m_rdCost.calcRdCost(outBestCU->m_totalDistortion, outBestCU->m_totalBits);
+
+        // copy original YUV samples in lossless mode
+        if (outBestCU->isLosslessCoded(0))
+            fillOrigYUVBuffer(outBestCU, m_origYuv[depth]);
+    }
+
+    // further split
+    if (cu_split_flag && !outBestCU->isSkipped(0))
+    {
+        uint32_t    nextDepth     = depth + 1;
+        TComDataCU* subBestPartCU = m_bestCU[nextDepth];
+        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++)
+        {
+            CU *child_cu = pic->getCU(cuAddr)->m_CULocalData + cu->childIdx + partUnitIdx;
+
+            int qp = outTempCU->getQP(0);
+            subBestPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+
+            if (child_cu->flags & CU::PRESENT)
+            {
+                subTempPartCU->initSubCU(outTempCU, partUnitIdx, nextDepth, qp); // clear sub partition datas or init.
+
+                if (partUnitIdx) // initialize RD with previous depth buffer
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+                else
+                    m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+                compressInterCU_rd5_6(subBestPartCU, subTempPartCU, nextDepth, child_cu);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth); // Keep best part data to current temporary data.
+                m_bestRecoYuv[nextDepth]->copyToPartYuv(m_tmpRecoYuv[depth], subBestPartCU->getTotalNumPart() * partUnitIdx);
+            }
+            else
+            {
+                subBestPartCU->copyToPic(nextDepth);
+                outTempCU->copyPartFrom(subBestPartCU, partUnitIdx, nextDepth);
+            }
+        }
+
+        if (cu_unsplit_flag)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeSplitFlag(outTempCU, 0, depth);
+            outTempCU->m_totalBits += m_entropyCoder->getNumberOfWrittenBits(); // split bits
+        }
+
+        if (m_rdCost.m_psyRd)
+            outTempCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits, outTempCU->m_psyEnergy);
+        else
+            outTempCU->m_totalRDCost = m_rdCost.calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+        if (depth == slice->m_pps->maxCuDQPDepth && slice->m_pps->bUseDQP)
+        {
+            bool hasResidual = false;
+            for (uint32_t blkIdx = 0; blkIdx < outTempCU->getTotalNumPart(); blkIdx++)
+            {
+                if (outTempCU->getCbf(blkIdx, TEXT_LUMA) || outTempCU->getCbf(blkIdx, TEXT_CHROMA_U) ||
+                    outTempCU->getCbf(blkIdx, TEXT_CHROMA_V))
+                {
+                    hasResidual = true;
+                    break;
+                }
+            }
+
+            uint32_t targetPartIdx = 0;
+            if (hasResidual)
+            {
+                bool foundNonZeroCbf = false;
+                outTempCU->setQPSubCUs(outTempCU->getRefQP(targetPartIdx), outTempCU, 0, depth, foundNonZeroCbf);
+                X265_CHECK(foundNonZeroCbf, "expected to find non-zero CBF\n");
+            }
+            else
+                outTempCU->setQPSubParts(outTempCU->getRefQP(targetPartIdx), 0, depth); // set QP to default QP
+        }
+
+        m_rdEntropyCoders[nextDepth][CI_NEXT_BEST].store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+        checkBestMode(outBestCU, outTempCU, depth); // RD compare current CU against split
+    }
+    outBestCU->copyToPic(depth); // Copy Best data to Picture for next partition prediction.
+
+    // Copy Yuv data to picture Yuv
+    m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+
+#if CHECKED_BUILD || _DEBUG
+    X265_CHECK(outBestCU->getPartitionSize(0) != SIZE_NONE, "no best partition size\n");
+    X265_CHECK(outBestCU->getPredictionMode(0) != MODE_NONE, "no best partition mode\n");
+    if (m_rdCost.m_psyRd)
+    {
+        X265_CHECK(outBestCU->m_totalPsyCost != MAX_INT64, "no best partition cost\n");
+    }
+    else
+    {
+        X265_CHECK(outBestCU->m_totalRDCost != MAX_INT64, "no best partition cost\n");
+    }
+#endif
+}
+
+void Analysis::checkMerge2Nx2N_rd0_4(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComYuv*& bestPredYuv, TComYuv*& yuvReconBest)
+{
+    X265_CHECK(outTempCU->m_slice->m_sliceType != I_SLICE, "Evaluating merge in I slice\n");
+    TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS][2]; // double length for mv of both lists
+    uint8_t interDirNeighbours[MRG_MAX_NUM_CANDS];
+    uint32_t maxNumMergeCand = outTempCU->m_slice->m_maxNumMergeCand;
+
+    uint32_t depth = outTempCU->getDepth(0);
+    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to CTU level
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+    outTempCU->getInterMergeCandidates(0, 0, mvFieldNeighbours, interDirNeighbours, maxNumMergeCand);
+    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
+    outTempCU->setMergeFlag(0, true);
+
+    outBestCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to CTU level
+    outBestCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+    outBestCU->setPredModeSubParts(MODE_INTER, 0, depth);
+    outBestCU->setMergeFlag(0, true);
+
+    int sizeIdx = outTempCU->getLog2CUSize(0) - 2;
+    int bestMergeCand = -1;
+
+    for (uint32_t mergeCand = 0; mergeCand < maxNumMergeCand; ++mergeCand)
+    {
+        if (!m_bFrameParallel ||
+            (mvFieldNeighbours[mergeCand][0].mv.y < (m_param->searchRange + 1) * 4 &&
+             mvFieldNeighbours[mergeCand][1].mv.y < (m_param->searchRange + 1) * 4))
+        {
+            // set MC parameters, interprets depth relative to CTU level
+            outTempCU->setMergeIndex(0, mergeCand);
+            outTempCU->setInterDirSubParts(interDirNeighbours[mergeCand], 0, 0, depth);
+            outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[mergeCand][0], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
+            outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[mergeCand][1], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
+
+            // do MC only for Luma part
+            /* Set CU parameters for motion compensation */
+            prepMotionCompensation(outTempCU, 0);
+            motionCompensation(m_tmpPredYuv[depth], true, false);
+            uint32_t bitsCand = getTUBits(mergeCand, maxNumMergeCand);
+            outTempCU->m_totalBits = bitsCand;
+            outTempCU->m_totalDistortion = primitives.sa8d[sizeIdx](m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
+                                                                    m_tmpPredYuv[depth]->getLumaAddr(), m_tmpPredYuv[depth]->getStride());
+            outTempCU->m_sa8dCost = m_rdCost.calcRdSADCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+            if (outTempCU->m_sa8dCost < outBestCU->m_sa8dCost)
+            {
+                bestMergeCand = mergeCand;
+                std::swap(outBestCU, outTempCU);
+                // Change Prediction data
+                std::swap(bestPredYuv, m_tmpPredYuv[depth]);
+            }
+        }
+    }
+
+    if (bestMergeCand < 0)
+    {
+        outBestCU->setMergeFlag(0, false);
+        outBestCU->setQPSubParts(outBestCU->getQP(0), 0, depth);
+    }
+    else
+    {
+        outTempCU->setMergeIndex(0, bestMergeCand);
+        outTempCU->setInterDirSubParts(interDirNeighbours[bestMergeCand], 0, 0, depth);
+        outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[bestMergeCand][0], SIZE_2Nx2N, 0, 0);
+        outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[bestMergeCand][1], SIZE_2Nx2N, 0, 0);
+        outTempCU->m_totalBits = outBestCU->m_totalBits;
+        outTempCU->m_totalDistortion = outBestCU->m_totalDistortion;
+        outTempCU->m_sa8dCost = outBestCU->m_sa8dCost;
+
+        if (m_param->rdLevel >= 1)
+        {
+            // calculate the motion compensation for chroma for the best mode selected
+            int numPart = outBestCU->getNumPartInter();
+            for (int partIdx = 0; partIdx < numPart; partIdx++)
+            {
+                prepMotionCompensation(outBestCU, partIdx);
+                motionCompensation(bestPredYuv, false, true);
+            }
+
+            if (outTempCU->isLosslessCoded(0))
+                outBestCU->m_totalRDCost = MAX_INT64;
+            else
+            {
+                // No-residue mode
+                encodeResAndCalcRdSkipCU(outBestCU, m_origYuv[depth], bestPredYuv, m_tmpRecoYuv[depth]);
+                std::swap(yuvReconBest, m_tmpRecoYuv[depth]);
+                m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+            }
+
+            // Encode with residue
+            encodeResAndCalcRdInterCU(outTempCU, m_origYuv[depth], bestPredYuv, m_tmpResiYuv[depth], m_bestResiYuv[depth], m_tmpRecoYuv[depth]);
+
+            uint64_t tempCost = m_rdCost.m_psyRd ? outTempCU->m_totalPsyCost : outTempCU->m_totalRDCost;
+            uint64_t bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+            if (tempCost < bestCost) // Choose best from no-residue mode and residue mode
+            {
+                std::swap(outBestCU, outTempCU);
+                std::swap(yuvReconBest, m_tmpRecoYuv[depth]);
+                m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+            }
+        }
+    }
+}
+
+void Analysis::checkMerge2Nx2N_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, bool *earlyDetectionSkipMode, TComYuv*& outBestPredYuv, TComYuv*& rpcYuvReconBest)
+{
+    X265_CHECK(outTempCU->m_slice->m_sliceType != I_SLICE, "I slice not expected\n");
+    TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS][2]; // double length for mv of both lists
+    uint8_t interDirNeighbours[MRG_MAX_NUM_CANDS];
+    uint32_t maxNumMergeCand = outTempCU->m_slice->m_maxNumMergeCand;
+
+    uint32_t depth = outTempCU->getDepth(0);
+    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to CTU level
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+    outTempCU->getInterMergeCandidates(0, 0, mvFieldNeighbours, interDirNeighbours, maxNumMergeCand);
+
+    int mergeCandBuffer[MRG_MAX_NUM_CANDS];
+    for (uint32_t i = 0; i < maxNumMergeCand; ++i)
+        mergeCandBuffer[i] = 0;
+
+    bool bestIsSkip = false;
+
+    uint32_t iteration = outTempCU->isLosslessCoded(0) ? 1 : 2;
+
+    for (uint32_t noResidual = 0; noResidual < iteration; ++noResidual)
+    {
+        for (uint32_t mergeCand = 0; mergeCand < maxNumMergeCand; ++mergeCand)
+        {
+            if (m_bFrameParallel &&
+                (mvFieldNeighbours[mergeCand][0].mv.y >= (m_param->searchRange + 1) * 4 ||
+                 mvFieldNeighbours[mergeCand][1].mv.y >= (m_param->searchRange + 1) * 4))
+            {
+                continue;
+            }
+            if (!(noResidual && mergeCandBuffer[mergeCand] == 1))
+            {
+                if (!(bestIsSkip && !noResidual))
+                {
+                    // set MC parameters
+                    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth); // interprets depth relative to CTU level
+                    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+                    outTempCU->setPartSizeSubParts(SIZE_2Nx2N, 0, depth); // interprets depth relative to CTU level
+                    outTempCU->setMergeFlag(0, true);
+                    outTempCU->setMergeIndex(0, mergeCand);
+                    outTempCU->setInterDirSubParts(interDirNeighbours[mergeCand], 0, 0, depth); // interprets depth relative to CTU level
+                    outTempCU->getCUMvField(REF_PIC_LIST_0)->setAllMvField(mvFieldNeighbours[mergeCand][0], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
+                    outTempCU->getCUMvField(REF_PIC_LIST_1)->setAllMvField(mvFieldNeighbours[mergeCand][1], SIZE_2Nx2N, 0, 0); // interprets depth relative to outTempCU level
+
+                    // do MC
+                    prepMotionCompensation(outTempCU, 0);
+                    motionCompensation(m_tmpPredYuv[depth], true, true);
+
+                    // estimate residual and encode everything
+                    if (noResidual)
+                        encodeResAndCalcRdSkipCU(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpRecoYuv[depth]);
+                    else
+                        encodeResAndCalcRdInterCU(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth],
+                                                  m_bestResiYuv[depth], m_tmpRecoYuv[depth]); 
+
+                    /* TODO: Fix the satd cost estimates. Why is merge being chosen in high motion areas: estimated distortion is too low? */
+                    if (!noResidual && !outTempCU->getQtRootCbf(0))
+                        mergeCandBuffer[mergeCand] = 1;
+
+                    outTempCU->setSkipFlagSubParts(!outTempCU->getQtRootCbf(0), 0, depth);
+                    int origQP = outTempCU->getQP(0);
+                    checkDQP(outTempCU);
+                    uint64_t tempCost = m_rdCost.m_psyRd ? outTempCU->m_totalPsyCost : outTempCU->m_totalRDCost;
+                    uint64_t bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+                    if (tempCost < bestCost)
+                    {
+                        std::swap(outBestCU, outTempCU);
+                        std::swap(outBestPredYuv, m_tmpPredYuv[depth]);
+                        std::swap(rpcYuvReconBest, m_tmpRecoYuv[depth]);
+
+                        m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+                    }
+                    outTempCU->setQPSubParts(origQP, 0, depth);
+                    outTempCU->setSkipFlagSubParts(false, 0, depth);
+                    if (!bestIsSkip)
+                        bestIsSkip = !outBestCU->getQtRootCbf(0);
+                }
+            }
+        }
+
+        if (!noResidual && m_param->bEnableEarlySkip && !outBestCU->getQtRootCbf(0))
+        {
+            if (outBestCU->getMergeFlag(0))
+                *earlyDetectionSkipMode = true;
+            else
+            {
+                bool noMvd = true;
+                for (uint32_t refListIdx = 0; refListIdx < 2; refListIdx++)
+                    if (outBestCU->m_slice->m_numRefIdx[refListIdx] > 0)
+                        noMvd &= !outBestCU->getCUMvField(refListIdx)->getMvd(0).word;
+                if (noMvd)
+                    *earlyDetectionSkipMode = true;
+            }
+        }
+    }
+}
+
+void Analysis::checkInter_rd0_4(TComDataCU* outTempCU, TComYuv* outPredYuv, PartSize partSize, bool bUseMRG)
+{
+    uint32_t depth = outTempCU->getDepth(0);
+
+    outTempCU->setPartSizeSubParts(partSize, 0, depth);
+    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+
+    // do motion compensation only for Luma since luma cost alone is calculated
+    outTempCU->m_totalBits = 0;
+    if (predInterSearch(outTempCU, outPredYuv, bUseMRG, false))
+    {
+        int sizeIdx = outTempCU->getLog2CUSize(0) - 2;
+        uint32_t distortion = primitives.sa8d[sizeIdx](m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
+                                                       outPredYuv->getLumaAddr(), outPredYuv->getStride());
+        outTempCU->m_totalDistortion = distortion;
+        outTempCU->m_sa8dCost = m_rdCost.calcRdSADCost(distortion, outTempCU->m_totalBits);
+    }
+    else
+    {
+        outTempCU->m_totalDistortion = MAX_UINT;
+        outTempCU->m_totalRDCost = MAX_INT64;
+    }
+}
+
+void Analysis::checkInter_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize, bool bUseMRG)
+{
+    uint32_t depth = outTempCU->getDepth(0);
+
+    outTempCU->setSkipFlagSubParts(false, 0, depth);
+    outTempCU->setPartSizeSubParts(partSize, 0, depth);
+    outTempCU->setPredModeSubParts(MODE_INTER, 0, depth);
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+
+    if (predInterSearch(outTempCU, m_tmpPredYuv[depth], bUseMRG, true))
+    {
+        encodeResAndCalcRdInterCU(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_bestResiYuv[depth], m_tmpRecoYuv[depth]);
+        checkDQP(outTempCU);
+        checkBestMode(outBestCU, outTempCU, depth);
+    }
+}
+
+void Analysis::checkIntraInInter_rd0_4(TComDataCU* cu, PartSize partSize)
+{
+    uint32_t depth = cu->getDepth(0);
+
+    cu->setPartSizeSubParts(partSize, 0, depth);
+    cu->setPredModeSubParts(MODE_INTRA, 0, depth);
+    cu->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+
+    uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
+    uint32_t log2TrSize  = cu->getLog2CUSize(0) - initTrDepth;
+    uint32_t tuSize      = 1 << log2TrSize;
+    const uint32_t partOffset  = 0;
+
+    // Reference sample smoothing
+    TComPattern::initAdiPattern(cu, partOffset, initTrDepth, m_predBuf, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt, ALL_IDX);
+
+    pixel* fenc     = m_origYuv[depth]->getLumaAddr();
+    uint32_t stride = m_modePredYuv[5][depth]->getStride();
+
+    pixel *above         = m_refAbove    + tuSize - 1;
+    pixel *aboveFiltered = m_refAboveFlt + tuSize - 1;
+    pixel *left          = m_refLeft     + tuSize - 1;
+    pixel *leftFiltered  = m_refLeftFlt  + tuSize - 1;
+    int sad, bsad;
+    uint32_t bits, bbits, mode, bmode;
+    uint64_t cost, bcost;
+
+    // 33 Angle modes once
+    ALIGN_VAR_32(pixel, buf_trans[32 * 32]);
+    ALIGN_VAR_32(pixel, tmp[33 * 32 * 32]);
+    int scaleTuSize = tuSize;
+    int scaleStride = stride;
+    int costShift = 0;
+    int sizeIdx = log2TrSize - 2;
+
+    if (tuSize > 32)
+    {
+        // origin is 64x64, we scale to 32x32 and setup required parameters
+        ALIGN_VAR_32(pixel, bufScale[32 * 32]);
+        primitives.scale2D_64to32(bufScale, fenc, stride);
+        fenc = bufScale;
+
+        // reserve space in case primitives need to store data in above
+        // or left buffers
+        pixel _above[4 * 32 + 1];
+        pixel _left[4 * 32 + 1];
+        pixel *aboveScale  = _above + 2 * 32;
+        pixel *leftScale   = _left + 2 * 32;
+        aboveScale[0] = leftScale[0] = above[0];
+        primitives.scale1D_128to64(aboveScale + 1, above + 1, 0);
+        primitives.scale1D_128to64(leftScale + 1, left + 1, 0);
+
+        scaleTuSize = 32;
+        scaleStride = 32;
+        costShift = 2;
+        sizeIdx = 5 - 2; // log2(scaleTuSize) - 2
+
+        // Filtered and Unfiltered refAbove and refLeft pointing to above and left.
+        above         = aboveScale;
+        left          = leftScale;
+        aboveFiltered = aboveScale;
+        leftFiltered  = leftScale;
+    }
+
+    pixelcmp_t sa8d = primitives.sa8d[sizeIdx];
+    int predsize = scaleTuSize * scaleTuSize;
+
+    uint32_t preds[3];
+    cu->getIntraDirLumaPredictor(partOffset, preds);
+
+    uint64_t mpms;
+    uint32_t rbits = getIntraRemModeBits(cu, partOffset, depth, preds, mpms);
+
+    // DC
+    primitives.intra_pred[DC_IDX][sizeIdx](tmp, scaleStride, left, above, 0, (scaleTuSize <= 16));
+    bsad = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
+    bmode = mode = DC_IDX;
+    bbits = (mpms & ((uint64_t)1 << mode)) ? getIntraModeBits(cu, mode, partOffset, depth) : rbits;
+    bcost = m_rdCost.calcRdSADCost(bsad, bbits);
+
+    pixel *abovePlanar = above;
+    pixel *leftPlanar  = left;
+
+    if (tuSize & (8 | 16 | 32))
+    {
+        abovePlanar = aboveFiltered;
+        leftPlanar  = leftFiltered;
+    }
+
+    // PLANAR
+    primitives.intra_pred[PLANAR_IDX][sizeIdx](tmp, scaleStride, leftPlanar, abovePlanar, 0, 0);
+    sad = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
+    mode = PLANAR_IDX;
+    bits = (mpms & ((uint64_t)1 << mode)) ? getIntraModeBits(cu, mode, partOffset, depth) : rbits;
+    cost = m_rdCost.calcRdSADCost(sad, bits);
+    COPY4_IF_LT(bcost, cost, bmode, mode, bsad, sad, bbits, bits);
+
+    // Transpose NxN
+    primitives.transpose[sizeIdx](buf_trans, fenc, scaleStride);
+
+    primitives.intra_pred_allangs[sizeIdx](tmp, above, left, aboveFiltered, leftFiltered, (scaleTuSize <= 16));
+
+    bool modeHor;
+    pixel *cmp;
+    intptr_t srcStride;
+
+#define TRY_ANGLE(angle) \
+    modeHor = angle < 18; \
+    cmp = modeHor ? buf_trans : fenc; \
+    srcStride = modeHor ? scaleTuSize : scaleStride; \
+    sad = sa8d(cmp, srcStride, &tmp[(angle - 2) * predsize], scaleTuSize) << costShift; \
+    bits = (mpms & ((uint64_t)1 << angle)) ? getIntraModeBits(cu, angle, partOffset, depth) : rbits; \
+    cost = m_rdCost.calcRdSADCost(sad, bits)
+
+    if (m_param->bEnableFastIntra)
+    {
+        int asad = 0;
+        uint32_t lowmode, highmode, amode = 5, abits = 0;
+        uint64_t acost = MAX_INT64;
+
+        /* pick the best angle, sampling at distance of 5 */
+        for (mode = 5; mode < 35; mode += 5)
+        {
+            TRY_ANGLE(mode);
+            COPY4_IF_LT(acost, cost, amode, mode, asad, sad, abits, bits);
+        }
+
+        /* refine best angle at distance 2, then distance 1 */
+        for (uint32_t dist = 2; dist >= 1; dist--)
+        {
+            lowmode = amode - dist;
+            highmode = amode + dist;
+
+            X265_CHECK(lowmode >= 2 && lowmode <= 34, "low intra mode out of range\n");
+            TRY_ANGLE(lowmode);
+            COPY4_IF_LT(acost, cost, amode, lowmode, asad, sad, abits, bits);
+
+            X265_CHECK(highmode >= 2 && highmode <= 34, "high intra mode out of range\n");
+            TRY_ANGLE(highmode);
+            COPY4_IF_LT(acost, cost, amode, highmode, asad, sad, abits, bits);
+        }
+
+        if (amode == 33)
+        {
+            TRY_ANGLE(34);
+            COPY4_IF_LT(acost, cost, amode, 34, asad, sad, abits, bits);
+        }
+
+        COPY4_IF_LT(bcost, acost, bmode, amode, bsad, asad, bbits, abits);
+    }
+    else // calculate and search all intra prediction angles for lowest cost
+    {
+        for (mode = 2; mode < 35; mode++)
+        {
+            TRY_ANGLE(mode);
+            COPY4_IF_LT(bcost, cost, bmode, mode, bsad, sad, bbits, bits);
+        }
+    }
+    cu->m_totalBits = bbits;
+    cu->m_totalDistortion = bsad;
+    cu->m_sa8dCost = bcost;
+
+    // generate predYuv for the best mode
+    cu->setLumaIntraDirSubParts(bmode, partOffset, depth + initTrDepth);
+}
+
+void Analysis::checkIntraInInter_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize)
+{
+    uint32_t depth = outTempCU->getDepth(0);
+
+    PPAScopeEvent(CheckRDCostIntra + depth);
+
+    m_quant.setQPforQuant(outTempCU);
+    outTempCU->setSkipFlagSubParts(false, 0, depth);
+    outTempCU->setPartSizeSubParts(partSize, 0, depth);
+    outTempCU->setPredModeSubParts(MODE_INTRA, 0, depth);
+    outTempCU->setCUTransquantBypassSubParts(!!m_param->bLossless, 0, depth);
+
+    uint32_t tuDepthRange[2];
+    outTempCU->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+
+    estIntraPredQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth], tuDepthRange);
+
+    estIntraPredChromaQT(outTempCU, m_origYuv[depth], m_tmpPredYuv[depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
+
+    m_entropyCoder->resetBits();
+    if (outTempCU->m_slice->m_pps->bTransquantBypassEnabled)
+        m_entropyCoder->codeCUTransquantBypassFlag(outTempCU->getCUTransquantBypass(0));
+
+    if (!outTempCU->m_slice->isIntra())
+    {
+        m_entropyCoder->codeSkipFlag(outTempCU, 0);
+        m_entropyCoder->codePredMode(outTempCU->getPredictionMode(0));
+    }
+    m_entropyCoder->codePartSize(outTempCU, 0, depth);
+    m_entropyCoder->codePredInfo(outTempCU, 0);
+    outTempCU->m_mvBits = m_entropyCoder->getNumberOfWrittenBits();
+
+    // Encode Coefficients
+    bool bCodeDQP = m_bEncodeDQP;
+    m_entropyCoder->codeCoeff(outTempCU, 0, depth, bCodeDQP, tuDepthRange);
+    m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+    outTempCU->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
+    outTempCU->m_coeffBits = outTempCU->m_totalBits - outTempCU->m_mvBits;
+
+    if (m_rdCost.m_psyRd)
+    {
+        int part = outTempCU->getLog2CUSize(0) - 2;
+        outTempCU->m_psyEnergy = m_rdCost.psyCost(part, m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
+                                                  m_tmpRecoYuv[depth]->getLumaAddr(), m_tmpRecoYuv[depth]->getStride());
+        outTempCU->m_totalPsyCost = m_rdCost.calcPsyRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits, outTempCU->m_psyEnergy);
+    }
+    else
+        outTempCU->m_totalRDCost = m_rdCost.calcRdCost(outTempCU->m_totalDistortion, outTempCU->m_totalBits);
+
+    checkDQP(outTempCU);
+    checkBestMode(outBestCU, outTempCU, depth);
+}
+
+void Analysis::encodeIntraInInter(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv,  ShortYuv* outResiYuv, TComYuv* outReconYuv)
+{
+    uint64_t puCost = 0;
+    uint32_t puBits = 0;
+    uint32_t psyEnergy = 0;
+    uint32_t depth = cu->getDepth(0);
+    uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
+
+    // set context models
+    m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+    m_quant.setQPforQuant(cu);
+
+    uint32_t tuDepthRange[2];
+    cu->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+
+    uint32_t puDistY = xRecurIntraCodingQT(cu, initTrDepth, 0, fencYuv, predYuv, outResiYuv, false, puCost, puBits, psyEnergy, tuDepthRange);
+    xSetIntraResultQT(cu, initTrDepth, 0, outReconYuv);
+
+    // update PU data
+    cu->copyToPic(cu->getDepth(0), 0, initTrDepth);
+
+    // set distortion (rate and r-d costs are determined later)
+    cu->m_totalDistortion = puDistY;
+
+    estIntraPredChromaQT(cu, fencYuv, predYuv, outResiYuv, outReconYuv);
+    m_entropyCoder->resetBits();
+    if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+        m_entropyCoder->codeCUTransquantBypassFlag(cu->getCUTransquantBypass(0));
+
+    if (!cu->m_slice->isIntra())
+    {
+        m_entropyCoder->codeSkipFlag(cu, 0);
+        m_entropyCoder->codePredMode(cu->getPredictionMode(0));
+    }
+    m_entropyCoder->codePartSize(cu, 0, depth);
+    m_entropyCoder->codePredInfo(cu, 0);
+    cu->m_mvBits += m_entropyCoder->getNumberOfWrittenBits();
+
+    // Encode Coefficients
+    bool bCodeDQP = m_bEncodeDQP;
+    m_entropyCoder->codeCoeff(cu, 0, depth, bCodeDQP, tuDepthRange);
+    m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+
+    cu->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
+    cu->m_coeffBits = cu->m_totalBits - cu->m_mvBits;
+    if (m_rdCost.m_psyRd)
+    {
+        int part = cu->getLog2CUSize(0) - 2;
+        cu->m_psyEnergy = m_rdCost.psyCost(part, m_origYuv[depth]->getLumaAddr(), m_origYuv[depth]->getStride(),
+            m_tmpRecoYuv[depth]->getLumaAddr(), m_tmpRecoYuv[depth]->getStride());
+        cu->m_totalPsyCost = m_rdCost.calcPsyRdCost(cu->m_totalDistortion, cu->m_totalBits, cu->m_psyEnergy);
+    }
+    else
+        cu->m_totalRDCost = m_rdCost.calcRdCost(cu->m_totalDistortion, cu->m_totalBits);
+}
+
+void Analysis::encodeResidue(TComDataCU* ctu, TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
+{
+    Frame* pic = cu->m_pic;
+
+    if (depth < ctu->getDepth(absPartIdx) && depth < g_maxCUDepth)
+    {
+        Slice* slice = cu->m_slice;
+        uint32_t nextDepth = depth + 1;
+        TComDataCU* subTempPartCU = m_tempCU[nextDepth];
+        uint32_t qNumParts = (NUM_CU_PARTITIONS >> (depth << 1)) >> 2;
+        uint32_t xmax = slice->m_sps->picWidthInLumaSamples  - ctu->getCUPelX();
+        uint32_t ymax = slice->m_sps->picHeightInLumaSamples - ctu->getCUPelY();
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
+        {
+            if (g_zscanToPelX[absPartIdx] < xmax && g_zscanToPelY[absPartIdx] < ymax)
+            {
+                subTempPartCU->copyToSubCU(cu, partUnitIdx, nextDepth);
+                encodeResidue(ctu, subTempPartCU, absPartIdx, nextDepth);
+            }
+        }
+
+        return;
+    }
+
+    uint32_t cuAddr = cu->getAddr();
+
+    m_quant.setQPforQuant(cu);
+
+    if (ctu->getPredictionMode(absPartIdx) == MODE_INTER)
+    {
+        int log2CUSize = cu->getLog2CUSize(0);
+        if (!ctu->getSkipFlag(absPartIdx))
+        {
+            const int sizeIdx = log2CUSize - 2;
+            // Calculate Residue
+            pixel* src2 = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
+            pixel* src1 = m_origYuv[0]->getLumaAddr(absPartIdx);
+            int16_t* dst = m_tmpResiYuv[depth]->getLumaAddr();
+            uint32_t src2stride = m_bestPredYuv[0]->getStride();
+            uint32_t src1stride = m_origYuv[0]->getStride();
+            uint32_t dststride = m_tmpResiYuv[depth]->m_width;
+            primitives.luma_sub_ps[sizeIdx](dst, dststride, src1, src2, src1stride, src2stride);
+
+            src2 = m_bestPredYuv[0]->getCbAddr(absPartIdx);
+            src1 = m_origYuv[0]->getCbAddr(absPartIdx);
+            dst = m_tmpResiYuv[depth]->getCbAddr();
+            src2stride = m_bestPredYuv[0]->getCStride();
+            src1stride = m_origYuv[0]->getCStride();
+            dststride = m_tmpResiYuv[depth]->m_cwidth;
+            primitives.chroma[m_param->internalCsp].sub_ps[sizeIdx](dst, dststride, src1, src2, src1stride, src2stride);
+
+            src2 = m_bestPredYuv[0]->getCrAddr(absPartIdx);
+            src1 = m_origYuv[0]->getCrAddr(absPartIdx);
+            dst = m_tmpResiYuv[depth]->getCrAddr();
+            primitives.chroma[m_param->internalCsp].sub_ps[sizeIdx](dst, dststride, src1, src2, src1stride, src2stride);
+
+            uint32_t tuDepthRange[2];
+            cu->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+            // Residual encoding
+            residualTransformQuantInter(cu, 0, m_origYuv[0], m_tmpResiYuv[depth], cu->getDepth(0), tuDepthRange);
+            checkDQP(cu);
+
+            if (ctu->getMergeFlag(absPartIdx) && cu->getPartitionSize(0) == SIZE_2Nx2N && !cu->getQtRootCbf(0))
+            {
+                cu->setSkipFlagSubParts(true, 0, depth);
+                cu->copyCodedToPic(depth);
+            }
+            else
+            {
+                cu->copyCodedToPic(depth);
+
+                // Generate Recon
+                pixel* pred = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
+                int16_t* res = m_tmpResiYuv[depth]->getLumaAddr();
+                pixel* reco = m_bestRecoYuv[depth]->getLumaAddr();
+                dststride = m_bestRecoYuv[depth]->getStride();
+                src1stride = m_bestPredYuv[0]->getStride();
+                src2stride = m_tmpResiYuv[depth]->m_width;
+                primitives.luma_add_ps[sizeIdx](reco, dststride, pred, res, src1stride, src2stride);
+
+                pred = m_bestPredYuv[0]->getCbAddr(absPartIdx);
+                res = m_tmpResiYuv[depth]->getCbAddr();
+                reco = m_bestRecoYuv[depth]->getCbAddr();
+                dststride = m_bestRecoYuv[depth]->getCStride();
+                src1stride = m_bestPredYuv[0]->getCStride();
+                src2stride = m_tmpResiYuv[depth]->m_cwidth;
+                primitives.chroma[m_param->internalCsp].add_ps[sizeIdx](reco, dststride, pred, res, src1stride, src2stride);
+
+                pred = m_bestPredYuv[0]->getCrAddr(absPartIdx);
+                res = m_tmpResiYuv[depth]->getCrAddr();
+                reco = m_bestRecoYuv[depth]->getCrAddr();
+                primitives.chroma[m_param->internalCsp].add_ps[sizeIdx](reco, dststride, pred, res, src1stride, src2stride);
+                m_bestRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+                return;
+            }
+        }
+
+        // Generate Recon
+        int part = partitionFromLog2Size(log2CUSize);
+        TComPicYuv* rec = pic->getPicYuvRec();
+        pixel* src = m_bestPredYuv[0]->getLumaAddr(absPartIdx);
+        pixel* dst = rec->getLumaAddr(cuAddr, absPartIdx);
+        uint32_t srcstride = m_bestPredYuv[0]->getStride();
+        uint32_t dststride = rec->getStride();
+        primitives.luma_copy_pp[part](dst, dststride, src, srcstride);
+
+        src = m_bestPredYuv[0]->getCbAddr(absPartIdx);
+        dst = rec->getCbAddr(cuAddr, absPartIdx);
+        srcstride = m_bestPredYuv[0]->getCStride();
+        dststride = rec->getCStride();
+        primitives.chroma[m_param->internalCsp].copy_pp[part](dst, dststride, src, srcstride);
+
+        src = m_bestPredYuv[0]->getCrAddr(absPartIdx);
+        dst = rec->getCrAddr(cuAddr, absPartIdx);
+        primitives.chroma[m_param->internalCsp].copy_pp[part](dst, dststride, src, srcstride);
+    }
+    else
+    {
+        m_origYuv[0]->copyPartToYuv(m_origYuv[depth], absPartIdx);
+        generateCoeffRecon(cu, m_origYuv[depth], m_modePredYuv[5][depth], m_tmpResiYuv[depth], m_tmpRecoYuv[depth]);
+        checkDQP(cu);
+        m_tmpRecoYuv[depth]->copyToPicYuv(pic->getPicYuvRec(), cuAddr, absPartIdx);
+        cu->copyCodedToPic(depth);
+    }
+}
+
+/* check whether current try is the best with identifying the depth of current try */
+void Analysis::checkBestMode(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth)
+{
+    uint64_t tempCost = m_rdCost.m_psyRd ? outTempCU->m_totalPsyCost : outTempCU->m_totalRDCost;
+    uint64_t bestCost = m_rdCost.m_psyRd ? outBestCU->m_totalPsyCost : outBestCU->m_totalRDCost;
+
+    if (tempCost < bestCost)
+    {
+        // Change Information data
+        std::swap(outBestCU, outTempCU);
+
+        // Change Prediction data
+        std::swap(m_bestPredYuv[depth], m_tmpPredYuv[depth]);
+
+        // Change Reconstruction data
+        std::swap(m_bestRecoYuv[depth], m_tmpRecoYuv[depth]);
+
+        m_rdEntropyCoders[depth][CI_TEMP_BEST].store(m_rdEntropyCoders[depth][CI_NEXT_BEST]);
+    }
+}
+
+void Analysis::deriveTestModeAMP(TComDataCU* outBestCU, PartSize parentSize, bool &bTestAMP_Hor, bool &bTestAMP_Ver,
+                                 bool &bTestMergeAMP_Hor, bool &bTestMergeAMP_Ver)
+{
+    if (outBestCU->getPartitionSize(0) == SIZE_2NxN)
+        bTestAMP_Hor = true;
+    else if (outBestCU->getPartitionSize(0) == SIZE_Nx2N)
+        bTestAMP_Ver = true;
+    else if (outBestCU->getPartitionSize(0) == SIZE_2Nx2N && outBestCU->getMergeFlag(0) == false && outBestCU->isSkipped(0) == false)
+    {
+        bTestAMP_Hor = true;
+        bTestAMP_Ver = true;
+    }
+
+    // Utilize the partition size of parent PU
+    if (parentSize >= SIZE_2NxnU && parentSize <= SIZE_nRx2N)
+    {
+        bTestMergeAMP_Hor = true;
+        bTestMergeAMP_Ver = true;
+    }
+
+    if (parentSize == SIZE_NONE) //! if parent is intra
+    {
+        if (outBestCU->getPartitionSize(0) == SIZE_2NxN)
+            bTestMergeAMP_Hor = true;
+        else if (outBestCU->getPartitionSize(0) == SIZE_Nx2N)
+            bTestMergeAMP_Ver = true;
+    }
+
+    if (outBestCU->getPartitionSize(0) == SIZE_2Nx2N && outBestCU->isSkipped(0) == false)
+    {
+        bTestMergeAMP_Hor = true;
+        bTestMergeAMP_Ver = true;
+    }
+
+    if (outBestCU->getLog2CUSize(0) == 6)
+    {
+        bTestAMP_Hor = false;
+        bTestAMP_Ver = false;
+    }
+}
+
+void Analysis::checkDQP(TComDataCU* cu)
+{
+    uint32_t depth = cu->getDepth(0);
+    Slice* slice = cu->m_slice;
+
+    if (slice->m_pps->bUseDQP && depth <= slice->m_pps->maxCuDQPDepth)
+    {
+        if (!cu->getCbf(0, TEXT_LUMA, 0) && !cu->getCbf(0, TEXT_CHROMA_U, 0) && !cu->getCbf(0, TEXT_CHROMA_V, 0))
+            cu->setQPSubParts(cu->getRefQP(0), 0, depth); // set QP to default QP
+    }
+}
+
+/* Function for filling original YUV samples of a CU in lossless mode */
+void Analysis::fillOrigYUVBuffer(TComDataCU* cu, TComYuv* fencYuv)
+{
+    /* TODO: is this extra copy really necessary? the source pixels will still
+     * be available when getLumaOrigYuv() is used */
+
+    uint32_t width  = 1 << cu->getLog2CUSize(0);
+    uint32_t height = 1 << cu->getLog2CUSize(0);
+
+    pixel* srcY = fencYuv->getLumaAddr();
+    pixel* dstY = cu->getLumaOrigYuv();
+    uint32_t srcStride = fencYuv->getStride();
+
+    /* TODO: square block copy primitive */
+    for (uint32_t y = 0; y < height; y++)
+    {
+        for (uint32_t x = 0; x < width; x++)
+            dstY[x] = srcY[x];
+
+        dstY += width;
+        srcY += srcStride;
+    }
+
+    pixel* srcCb = fencYuv->getChromaAddr(1);
+    pixel* srcCr = fencYuv->getChromaAddr(2);
+
+    pixel* dstCb = cu->getChromaOrigYuv(1);
+    pixel* dstCr = cu->getChromaOrigYuv(2);
+
+    uint32_t srcStrideC = fencYuv->getCStride();
+    uint32_t widthC  = width  >> cu->getHorzChromaShift();
+    uint32_t heightC = height >> cu->getVertChromaShift();
+
+    /* TODO: block copy primitives */
+    for (uint32_t y = 0; y < heightC; y++)
+    {
+        for (uint32_t x = 0; x < widthC; x++)
+        {
+            dstCb[x] = srcCb[x];
+            dstCr[x] = srcCr[x];
+        }
+
+        dstCb += widthC;
+        dstCr += widthC;
+        srcCb += srcStrideC;
+        srcCr += srcStrideC;
+    }
+}

x265-hg20140928.tar.bz2/source/encoder/analysis.h Added

@@ -0,0 +1,138 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Deepthi Nandakumar <deepthi@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_ANALYSIS_H
+#define X265_ANALYSIS_H
+
+#define ANGULAR_MODE_ID 2
+#define AMP_ID 3
+#define INTER_MODES 4
+#define INTRA_MODES 3
+
+#include "common.h"
+#include "predict.h"
+#include "quant.h"
+#include "shortyuv.h"
+#include "TLibCommon/TComYuv.h"
+#include "TLibCommon/TComDataCU.h"
+
+#include "entropy.h"
+#include "search.h"
+
+namespace x265 {
+// private namespace
+
+struct StatisticLog
+{
+    uint64_t cntInter[4];
+    uint64_t cntIntra[4];
+    uint64_t cuInterDistribution[4][INTER_MODES];
+    uint64_t cuIntraDistribution[4][INTRA_MODES];
+    uint64_t cntIntraNxN;
+    uint64_t cntSkipCu[4];
+    uint64_t cntTotalCu[4];
+    uint64_t totalCu;
+
+    /* These states store the count of inter,intra and skip ctus within quad tree structure of each CU */
+    uint32_t qTreeInterCnt[4];
+    uint32_t qTreeIntraCnt[4];
+    uint32_t qTreeSkipCnt[4];
+
+    StatisticLog()
+    {
+        memset(this, 0, sizeof(StatisticLog));
+    }
+};
+
+class Encoder;
+class Entropy;
+
+class Analysis : public Search
+{
+public:
+
+    static const int MAX_PRED_TYPES = 6;
+
+    TComDataCU*  m_memPool;
+
+    TComDataCU*  m_interCU_2Nx2N[NUM_CU_DEPTH];
+    TComDataCU*  m_interCU_2NxN[NUM_CU_DEPTH];
+    TComDataCU*  m_interCU_Nx2N[NUM_CU_DEPTH];
+    TComDataCU*  m_intraInInterCU[NUM_CU_DEPTH];
+    TComDataCU*  m_mergeCU[NUM_CU_DEPTH];
+    TComDataCU*  m_bestMergeCU[NUM_CU_DEPTH];
+    TComDataCU*  m_bestCU[NUM_CU_DEPTH]; // Best CUs at each depth
+    TComDataCU*  m_tempCU[NUM_CU_DEPTH]; // Temporary CUs at each depth
+
+    TComYuv**    m_bestPredYuv;          // Best Prediction Yuv for each depth
+    ShortYuv**   m_bestResiYuv;          // Best Residual Yuv for each depth
+    TComYuv**    m_bestRecoYuv;          // Best Reconstruction Yuv for each depth
+
+    TComYuv**    m_tmpPredYuv;           // Temporary Prediction Yuv for each depth
+    ShortYuv**   m_tmpResiYuv;           // Temporary Residual Yuv for each depth
+    TComYuv**    m_tmpRecoYuv;           // Temporary Reconstruction Yuv for each depth
+    TComYuv**    m_modePredYuv[MAX_PRED_TYPES]; // Prediction buffers for inter, intra, rect(2) and merge
+    TComYuv**    m_bestMergeRecoYuv;
+    TComYuv**    m_origYuv;             // Original Yuv at each depth
+
+    bool         m_bEncodeDQP;
+
+    StatisticLog  m_sliceTypeLog[3];
+    StatisticLog* m_log;
+
+    Analysis();
+    bool create(uint32_t totalDepth, uint32_t maxWidth);
+    void destroy();
+    void compressCU(TComDataCU* cu);
+    void loadCTUData(TComDataCU* cu);
+
+protected:
+
+    /* Warning: The interface for these functions will undergo significant changes as a major refactor is under progress */
+    void compressIntraCU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu);
+    void checkIntra(TComDataCU*& outTempCU, PartSize partSize, CU *cu, uint8_t* sharedModes);
+    void compressSharedIntraCTU(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu, uint8_t* sharedDepth,
+                                char* sharedPartSizes, uint8_t* sharedModes, uint32_t &zOrder);
+    void compressInterCU_rd0_4(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComDataCU* parentCU, uint32_t depth, CU *cu,
+                               int bInsidePicture, uint32_t partitionIndex, uint32_t minDepth);
+    void compressInterCU_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth, CU *cu,
+                               PartSize parentSize = SIZE_NONE);
+    void checkMerge2Nx2N_rd0_4(TComDataCU*& outBestCU, TComDataCU*& outTempCU, TComYuv*& bestPredYuv, TComYuv*& tmpPredYuv);
+    void checkMerge2Nx2N_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, bool *earlyDetectionSkipMode,
+                               TComYuv*& outBestPredYuv, TComYuv*& rpcYuvReconBest);
+    void checkInter_rd0_4(TComDataCU* outTempCU, TComYuv* outPredYUV, PartSize partSize, bool bUseMRG = false);
+    void checkInter_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize, bool bUseMRG = false);
+    void checkIntraInInter_rd0_4(TComDataCU* cu, PartSize partSize);
+    void checkIntraInInter_rd5_6(TComDataCU*& outBestCU, TComDataCU*& outTempCU, PartSize partSize);
+
+    void checkBestMode(TComDataCU*& outBestCU, TComDataCU*& outTempCU, uint32_t depth);
+    void encodeIntraInInter(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* outResiYuv, TComYuv* outReconYuv);
+    void encodeResidue(TComDataCU* lcu, TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
+    void checkDQP(TComDataCU* cu);
+    void deriveTestModeAMP(TComDataCU* bestCU, PartSize parentSize, bool &bTestAMP_Hor, bool &bTestAMP_Ver,
+                           bool &bTestMergeAMP_Hor, bool &bTestMergeAMP_Ver);
+    void fillOrigYUVBuffer(TComDataCU* outCU, TComYuv* origYuv);
+};
+}
+
+#endif // ifndef X265_ANALYSIS_H

x265-hg20140324.tar.bz2/source/encoder/api.cpp -> x265-hg20140928.tar.bz2/source/encoder/api.cpp Changed

@@ -18,22 +18,35 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/CommonDef.h"
+#include "common.h"
+#include "bitstream.h"
 #include "param.h"
+
 #include "encoder.h"
-#include "frameencoder.h"
+#include "entropy.h"
+#include "level.h"
+#include "nal.h"
+#include "bitcost.h"
 
 using namespace x265;
 
 extern "C"
-x265_encoder *x265_encoder_open(x265_param *param)
+x265_encoder *x265_encoder_open(x265_param *p)
 {
+    if (!p)
+        return NULL;
+
+    x265_param *param = X265_MALLOC(x265_param, 1);
     if (!param)
         return NULL;
 
+    memcpy(param, p, sizeof(x265_param));
+    x265_log(param, X265_LOG_INFO, "HEVC encoder version %s\n", x265_version_str);
+    x265_log(param, X265_LOG_INFO, "build info %s\n", x265_build_info_str);
+
     x265_setup_primitives(param, param->cpuid);
 
     if (x265_check_params(param))
@@ -43,60 +56,53 @@
         return NULL;
 
     Encoder *encoder = new Encoder;
-    if (encoder)
+    if (!param->rc.bEnableSlowFirstPass)
+        x265_param_apply_fastfirstpass(param);
+    // may change params for auto-detect, etc
+    encoder->configure(param);
+    
+    // may change rate control and CPB params
+    if (!enforceLevel(*param, encoder->m_vps))
     {
-        // these may change params for auto-detect, etc
-        encoder->determineLevelAndProfile(param);
-        encoder->configure(param);
-
-        encoder->param = X265_MALLOC(x265_param, 1);
-        if (!encoder->param)
-        {
-            encoder->destroy();
-            return NULL;
-        }
-
-        memcpy(encoder->param, param, sizeof(x265_param));
-        x265_print_params(param);
-        encoder->create();
-        encoder->init();
+        delete encoder;
+        return NULL;
     }
 
+    // will detect and set profile/tier/level in VPS
+    determineLevel(*param, encoder->m_vps);
+
+    x265_print_params(param);
+    encoder->create();
+    encoder->init();
+
     return encoder;
 }
 
 extern "C"
 int x265_encoder_headers(x265_encoder *enc, x265_nal **pp_nal, uint32_t *pi_nal)
 {
-    if (!pp_nal || !enc)
-        return 0;
-
-    Encoder *encoder = static_cast<Encoder*>(enc);
-
-    int ret = 0;
-    NALUnitEBSP *nalunits[MAX_NAL_UNITS] = { 0, 0, 0, 0, 0 };
-    if (!encoder->getStreamHeaders(nalunits))
+    if (pp_nal && enc)
     {
-        int nalcount = encoder->extractNalData(nalunits);
-        *pp_nal = &encoder->m_nals[0];
-        if (pi_nal) *pi_nal = nalcount;
-    }
-    else if (pi_nal)
-    {
-        *pi_nal = 0;
-        ret = -1;
+        Encoder *encoder = static_cast<Encoder*>(enc);
+        Entropy sbacCoder;
+        Bitstream bs;
+        encoder->getStreamHeaders(encoder->m_nalList, sbacCoder, bs);
+        *pp_nal = &encoder->m_nalList.m_nal[0];
+        if (pi_nal) *pi_nal = encoder->m_nalList.m_numNal;
+        return encoder->m_nalList.m_occupancy;
     }
 
-    for (int i = 0; i < MAX_NAL_UNITS; i++)
+    return -1;
+}
+
+extern "C"
+void x265_encoder_parameters(x265_encoder *enc, x265_param *out)
+{
+    if (enc && out)
     {
-        if (nalunits[i])
-        {
-            free(nalunits[i]->m_nalUnitData);
-            X265_FREE(nalunits[i]);
-        }
+        Encoder *encoder = static_cast<Encoder*>(enc);
+        memcpy(out, encoder->m_param, sizeof(x265_param));
     }
-
-    return ret;
 }
 
 extern "C"
@@ -106,27 +112,31 @@
         return -1;
 
     Encoder *encoder = static_cast<Encoder*>(enc);
-    NALUnitEBSP *nalunits[MAX_NAL_UNITS] = { 0, 0, 0, 0, 0 };
-    int numEncoded = encoder->encode(!pic_in, pic_in, pic_out, nalunits);
+    int numEncoded;
 
-    if (pp_nal && numEncoded > 0)
+    // While flushing, we cannot return 0 until the entire stream is flushed
+    do
     {
-        int nalcount = encoder->extractNalData(nalunits);
-        *pp_nal = &encoder->m_nals[0];
-        if (pi_nal) *pi_nal = nalcount;
+        numEncoded = encoder->encode(pic_in, pic_out);
     }
-    else if (pi_nal)
-        *pi_nal = 0;
+    while (numEncoded == 0 && !pic_in && encoder->m_numDelayedPic);
 
-    for (int i = 0; i < MAX_NAL_UNITS; i++)
+    // do not allow reuse of these buffers for more than one picture. The
+    // encoder now owns these analysisData buffers.
+    if (pic_in)
     {
-        if (nalunits[i])
-        {
-            free(nalunits[i]->m_nalUnitData);
-            X265_FREE(nalunits[i]);
-        }
+        pic_in->analysisData.intraData = NULL;
+        pic_in->analysisData.interData = NULL;
     }
 
+    if (pp_nal && numEncoded > 0)
+    {
+        *pp_nal = &encoder->m_nalList.m_nal[0];
+        if (pi_nal) *pi_nal = encoder->m_nalList.m_numNal;
+    }
+    else if (pi_nal)
+        *pi_nal = 0;
+
     return numEncoded;
 }
 
@@ -183,6 +193,17 @@
 
     pic->bitDepth = param->internalBitDepth;
     pic->colorSpace = param->internalCsp;
+    pic->forceqp = X265_QP_AUTO;
+    if (param->analysisMode)
+    {
+        uint32_t numPartitions   = 1 << (g_maxFullDepth * 2);
+        uint32_t widthInCU       = (param->sourceWidth  + g_maxCUSize - 1) >> g_maxLog2CUSize;
+        uint32_t heightInCU      = (param->sourceHeight + g_maxCUSize - 1) >> g_maxLog2CUSize;
+
+        uint32_t numCUsInFrame   = widthInCU * heightInCU;
+        pic->analysisData.numCUsInFrame = numCUsInFrame;
+        pic->analysisData.numPartitions = numPartitions;
+    }
 }
 
 extern "C"
@@ -190,3 +211,37 @@
 {
     return x265_free(p);
 }
+
+int x265_alloc_analysis_data(x265_picture* pic)
+{
+    CHECKED_MALLOC(pic->analysisData.interData, x265_inter_data, pic->analysisData.numCUsInFrame * 85);
+    CHECKED_MALLOC(pic->analysisData.intraData, x265_intra_data, 1);
+    pic->analysisData.intraData->cuAddr     = NULL;
+    pic->analysisData.intraData->depth      = NULL;
+    pic->analysisData.intraData->modes      = NULL;
+    pic->analysisData.intraData->partSizes  = NULL;
+    pic->analysisData.intraData->poc        = NULL;
+    CHECKED_MALLOC(pic->analysisData.intraData->depth, uint8_t, pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame);
+    CHECKED_MALLOC(pic->analysisData.intraData->modes, uint8_t, pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame);
+    CHECKED_MALLOC(pic->analysisData.intraData->partSizes, char, pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame);
+    CHECKED_MALLOC(pic->analysisData.intraData->cuAddr, uint32_t, pic->analysisData.numCUsInFrame);
+    CHECKED_MALLOC(pic->analysisData.intraData->poc, int, pic->analysisData.numCUsInFrame);
+    return 0;
+
+fail:
+    x265_free_analysis_data(pic);
+    return -1;
+}
+
+void x265_free_analysis_data(x265_picture* pic)
+{
+    X265_FREE(pic->analysisData.interData);
+    pic->analysisData.interData = NULL;
+    X265_FREE(pic->analysisData.intraData->depth);
+    X265_FREE(pic->analysisData.intraData->modes);
+    X265_FREE(pic->analysisData.intraData->partSizes);
+    X265_FREE(pic->analysisData.intraData->cuAddr);
+    X265_FREE(pic->analysisData.intraData->poc);
+    X265_FREE(pic->analysisData.intraData);
+    pic->analysisData.intraData = NULL;
+}

x265-hg20140324.tar.bz2/source/encoder/bitcost.cpp -> x265-hg20140928.tar.bz2/source/encoder/bitcost.cpp Changed

x265-hg20140324.tar.bz2/source/encoder/bitcost.h -> x265-hg20140928.tar.bz2/source/encoder/bitcost.h Changed

x265-hg20140324.tar.bz2/source/encoder/dpb.cpp -> x265-hg20140928.tar.bz2/source/encoder/dpb.cpp Changed

@@ -18,39 +18,56 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComPic.h"
-#include "TLibCommon/TComSlice.h"
-#include "encoder.h"
+#include "common.h"
+#include "frame.h"
+#include "slice.h"
 
-#include "PPA/ppa.h"
 #include "dpb.h"
-#include "frameencoder.h"
 
 using namespace x265;
 
 DPB::~DPB()
 {
+    while (!m_freeList.empty())
+    {
+        Frame* pic = m_freeList.popFront();
+        pic->destroy();
+        delete pic;
+    }
+
     while (!m_picList.empty())
     {
-        TComPic* pic = m_picList.popFront();
+        Frame* pic = m_picList.popFront();
         pic->destroy();
         delete pic;
     }
+
+    while (m_picSymFreeList)
+    {
+        TComPicSym* next = m_picSymFreeList->m_freeListNext;
+        m_picSymFreeList->destroy();
+
+        m_picSymFreeList->m_reconPicYuv->destroy();
+        delete m_picSymFreeList->m_reconPicYuv;
+
+        delete m_picSymFreeList;
+        m_picSymFreeList = next;
+    }
 }
 
 // move unreferenced pictures from picList to freeList for recycle
-void DPB::recycleUnreferenced(PicList& freeList)
+void DPB::recycleUnreferenced()
 {
-    TComPic *iterPic = m_picList.first();
+    Frame *iterPic = m_picList.first();
 
     while (iterPic)
     {
-        TComPic *pic = iterPic;
+        Frame *pic = iterPic;
         iterPic = iterPic->m_next;
-        if (pic->getSlice()->isReferenced() == false && pic->m_countRefEncoders == 0)
+        if (!pic->m_picSym->m_bHasReferences && !pic->m_countRefEncoders)
         {
             pic->m_reconRowCount.set(0);
             pic->m_bChromaPlanesExtended = false;
@@ -59,467 +76,222 @@
             m_picList.remove(*pic);
             iterPic = m_picList.first();
 
-            freeList.pushBack(*pic);
+            m_freeList.pushBack(*pic);
+            pic->m_picSym->m_freeListNext = m_picSymFreeList;
+            m_picSymFreeList = pic->m_picSym;
+            pic->m_picSym = NULL;
+            pic->m_reconPicYuv = NULL;
         }
     }
 }
 
-void DPB::prepareEncode(TComPic *pic)
+void DPB::prepareEncode(Frame *pic)
 {
-    PPAScopeEvent(DPB_prepareEncode);
-
-    int pocCurr = pic->getSlice()->getPOC();
+    Slice* slice = pic->m_picSym->m_slice;
+    slice->m_pic = pic;
+    slice->m_poc = pic->m_POC;
 
-    m_picList.pushFront(*pic);
+    int pocCurr = slice->m_poc;
+    int type = pic->m_lowres.sliceType;
+    bool bIsKeyFrame = pic->m_lowres.bKeyframe;
 
-    TComSlice* slice = pic->getSlice();
-    if (getNalUnitType(pocCurr, m_lastIDR, pic) == NAL_UNIT_CODED_SLICE_IDR_W_RADL ||
-        getNalUnitType(pocCurr, m_lastIDR, pic) == NAL_UNIT_CODED_SLICE_IDR_N_LP)
-    {
+    slice->m_nalUnitType = getNalUnitType(pocCurr, bIsKeyFrame);
+    if (slice->m_nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL)
         m_lastIDR = pocCurr;
-    }
-    slice->setLastIDR(m_lastIDR);
-
-    if (slice->getSliceType() != B_SLICE)
-        slice->setReferenced(true);
-    else
-    {
-        if (pic->m_lowres.sliceType == X265_TYPE_BREF)
-            slice->setReferenced(true);
-        else
-            slice->setReferenced(false);
-    }
-
-    slice->setTemporalLayerNonReferenceFlag(!slice->isReferenced());
-    // Set the nal unit type
-    slice->setNalUnitType(getNalUnitType(pocCurr, m_lastIDR, pic));
+    slice->m_lastIDR = m_lastIDR;
+    slice->m_sliceType = IS_X265_TYPE_B(type) ? B_SLICE : (type == X265_TYPE_P) ? P_SLICE : I_SLICE;
 
-    // If the slice is un-referenced, change from _R "referenced" to _N "non-referenced" NAL unit type
-    if (slice->getTemporalLayerNonReferenceFlag())
+    if (type == X265_TYPE_B)
     {
-        switch (slice->getNalUnitType())
+        // change from _R "referenced" to _N "non-referenced" NAL unit type
+        switch (slice->m_nalUnitType)
         {
         case NAL_UNIT_CODED_SLICE_TRAIL_R:
-            slice->setNalUnitType(NAL_UNIT_CODED_SLICE_TRAIL_N);
+            slice->m_nalUnitType = NAL_UNIT_CODED_SLICE_TRAIL_N;
             break;
         case NAL_UNIT_CODED_SLICE_RADL_R:
-            slice->setNalUnitType(NAL_UNIT_CODED_SLICE_RADL_N);
+            slice->m_nalUnitType = NAL_UNIT_CODED_SLICE_RADL_N;
             break;
         case NAL_UNIT_CODED_SLICE_RASL_R:
-            slice->setNalUnitType(NAL_UNIT_CODED_SLICE_RASL_N);
+            slice->m_nalUnitType = NAL_UNIT_CODED_SLICE_RASL_N;
             break;
         default:
             break;
         }
     }
 
-    // Do decoding refresh marking if any
-    decodingRefreshMarking(pocCurr, slice->getNalUnitType());
+    /* m_bHasReferences starts out as true for non-B pictures, and is set to false
+     * once no more pictures reference it */
+    pic->m_picSym->m_bHasReferences = IS_REFERENCED(slice);
 
-    computeRPS(pocCurr, slice->isIRAP(), slice->getLocalRPS(), slice->getSPS()->getMaxDecPicBuffering(0));
-    slice->setRPS(slice->getLocalRPS());
-    slice->setRPSidx(-1); // Force use of RPS from slice, rather than from SPS
+    m_picList.pushFront(*pic);
 
-    applyReferencePictureSet(slice->getRPS(), pocCurr); // Mark pictures in m_piclist as unreferenced if they are not included in RPS
+    // Do decoding refresh marking if any
+    decodingRefreshMarking(pocCurr, slice->m_nalUnitType);
 
-    arrangeLongtermPicturesInRPS(slice);
-    slice->setNumRefIdx(REF_PIC_LIST_0, X265_MIN(m_maxRefL0, slice->getRPS()->getNumberOfNegativePictures())); // Ensuring L0 contains just the -ve POC
-    slice->setNumRefIdx(REF_PIC_LIST_1, X265_MIN(m_maxRefL1, slice->getRPS()->getNumberOfPositivePictures()));
+    computeRPS(pocCurr, slice->isIRAP(), &slice->m_rps, slice->m_sps->maxDecPicBuffering);
 
+    // Mark pictures in m_piclist as unreferenced if they are not included in RPS
+    applyReferencePictureSet(&slice->m_rps, pocCurr);
+
+    slice->m_numRefIdx[0] = X265_MIN(m_maxRefL0, slice->m_rps.numberOfNegativePictures); // Ensuring L0 contains just the -ve POC
+    slice->m_numRefIdx[1] = X265_MIN(m_maxRefL1, slice->m_rps.numberOfPositivePictures);
     slice->setRefPicList(m_picList);
 
-    // Slice type refinement
-    if ((slice->getSliceType() == B_SLICE) && (slice->getNumRefIdx(REF_PIC_LIST_1) == 0))
-    {
-        slice->setSliceType(P_SLICE);
-    }
+    X265_CHECK(slice->m_sliceType != B_SLICE || slice->m_numRefIdx[1], "B slice without L1 references (non-fatal)\n");
 
-    if (slice->getSliceType() == B_SLICE)
+    if (slice->m_sliceType == B_SLICE)
     {
-        // TODO: Can we estimate this from lookahead?
-        slice->setColFromL0Flag(0);
-
-        bool bLowDelay = true;
-        int curPOC = slice->getPOC();
-        int refIdx = 0;
-
-        for (refIdx = 0; refIdx < slice->getNumRefIdx(REF_PIC_LIST_0) && bLowDelay; refIdx++)
-        {
-            if (slice->getRefPic(REF_PIC_LIST_0, refIdx)->getPOC() > curPOC)
-            {
-                bLowDelay = false;
-            }
-        }
-
-        for (refIdx = 0; refIdx < slice->getNumRefIdx(REF_PIC_LIST_1) && bLowDelay; refIdx++)
-        {
-            if (slice->getRefPic(REF_PIC_LIST_1, refIdx)->getPOC() > curPOC)
-            {
-                bLowDelay = false;
-            }
-        }
-
-        slice->setCheckLDC(bLowDelay);
+        /* TODO: the lookahead should be able to tell which reference picture
+         * had the least motion residual.  We should be able to use that here to
+         * select a colocation reference list and index */
+        slice->m_colFromL0Flag = false;
+        slice->m_colRefIdx = 0;
+        slice->m_bCheckLDC = false;
     }
     else
     {
-        slice->setCheckLDC(true);
+        slice->m_bCheckLDC = true;
+        slice->m_colFromL0Flag = true;
+        slice->m_colRefIdx = 0;
     }
+    slice->m_sLFaseFlag = (SLFASE_CONSTANT & (1 << (pocCurr % 31))) > 0;
 
-    slice->setRefPOCList();
-    slice->setEnableTMVPFlag(1);
-
-    bool bGPBcheck = false;
-    if (slice->getSliceType() == B_SLICE)
-    {
-        if (slice->getNumRefIdx(REF_PIC_LIST_0) == slice->getNumRefIdx(REF_PIC_LIST_1))
-        {
-            bGPBcheck = true;
-            for (int i = 0; i < slice->getNumRefIdx(REF_PIC_LIST_1); i++)
-            {
-                if (slice->getRefPOC(REF_PIC_LIST_1, i) != slice->getRefPOC(REF_PIC_LIST_0, i))
-                {
-                    bGPBcheck = false;
-                    break;
-                }
-            }
-        }
-    }
-
-    slice->setMvdL1ZeroFlag(bGPBcheck);
-    slice->setNextSlice(false);
-
-    /* Increment reference count of all motion-referenced frames.  This serves two purposes. First
-     * it prevents the frame from being recycled, and second the referenced frames know how many
-     * other FrameEncoders are using them for motion reference */
+    /* Increment reference count of all motion-referenced frames to prevent them
+     * from being recycled. These counts are decremented at the end of
+     * compressFrame() */
     int numPredDir = slice->isInterP() ? 1 : slice->isInterB() ? 2 : 0;
     for (int l = 0; l < numPredDir; l++)
     {
-        for (int ref = 0; ref < slice->getNumRefIdx(l); ref++)
+        for (int ref = 0; ref < slice->m_numRefIdx[l]; ref++)
         {
-            TComPic *refpic = slice->getRefPic(l, ref);
+            Frame *refpic = slice->m_refPicList[l][ref];
             ATOMIC_INC(&refpic->m_countRefEncoders);
         }
     }
 }
 
-void DPB::computeRPS(int curPoc, bool isRAP, TComReferencePictureSet * rps, unsigned int maxDecPicBuffer)
+void DPB::computeRPS(int curPoc, bool isRAP, RPS * rps, unsigned int maxDecPicBuffer)
 {
     unsigned int poci = 0, numNeg = 0, numPos = 0;
 
-    TComPic* iterPic = m_picList.first();
+    Frame* iterPic = m_picList.first();
 
     while (iterPic && (poci < maxDecPicBuffer - 1))
     {
-        if ((iterPic->getPOC() != curPoc) && (iterPic->getSlice()->isReferenced()))
+        if ((iterPic->m_POC != curPoc) && iterPic->m_picSym->m_bHasReferences)
         {
-            rps->m_POC[poci] = iterPic->getPOC();
-            rps->m_deltaPOC[poci] = rps->m_POC[poci] - curPoc;
-            (rps->m_deltaPOC[poci] < 0) ? numNeg++ : numPos++;
-            rps->m_used[poci] = !isRAP;
+            rps->poc[poci] = iterPic->m_POC;
+            rps->deltaPOC[poci] = rps->poc[poci] - curPoc;
+            (rps->deltaPOC[poci] < 0) ? numNeg++ : numPos++;
+            rps->bUsed[poci] = !isRAP;
             poci++;
         }
         iterPic = iterPic->m_next;
     }
 
-    rps->m_numberOfPictures = poci;
-    rps->m_numberOfPositivePictures = numPos;
-    rps->m_numberOfNegativePictures = numNeg;
-    rps->m_numberOfLongtermPictures = 0;
-    rps->m_interRPSPrediction = false;          // To be changed later when needed
+    rps->numberOfPictures = poci;
+    rps->numberOfPositivePictures = numPos;
+    rps->numberOfNegativePictures = numNeg;
 
     rps->sortDeltaPOC();
 }
 
-/** Function for marking the reference pictures when an IDR/CRA/CRANT/BLA/BLANT is encountered.
- * \param pocCRA POC of the CRA/CRANT/BLA/BLANT picture
- * \param bRefreshPending flag indicating if a deferred decoding refresh is pending
- * \param picList reference to the reference picture list
- * This function marks the reference pictures as "unused for reference" in the following conditions.
- * If the nal_unit_type is IDR/BLA/BLANT, all pictures in the reference picture list
- * are marked as "unused for reference"
- *    If the nal_unit_type is BLA/BLANT, set the pocCRA to the temporal reference of the current picture.
- * Otherwise
- *    If the bRefreshPending flag is true (a deferred decoding refresh is pending) and the current
- *    temporal reference is greater than the temporal reference of the latest CRA/CRANT/BLA/BLANT picture (pocCRA),
- *    mark all reference pictures except the latest CRA/CRANT/BLA/BLANT picture as "unused for reference" and set
- *    the bRefreshPending flag to false.
- *    If the nal_unit_type is CRA/CRANT, set the bRefreshPending flag to true and pocCRA to the temporal
- *    reference of the current picture.
- * Note that the current picture is already placed in the reference list and its marking is not changed.
- * If the current picture has a nal_ref_idc that is not 0, it will remain marked as "used for reference".
- */
+/* Marking reference pictures when an IDR/CRA is encountered. */
 void DPB::decodingRefreshMarking(int pocCurr, NalUnitType nalUnitType)
 {
-    if (nalUnitType == NAL_UNIT_CODED_SLICE_BLA_W_LP
-        || nalUnitType == NAL_UNIT_CODED_SLICE_BLA_W_RADL
-        || nalUnitType == NAL_UNIT_CODED_SLICE_BLA_N_LP
-        || nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL
-        || nalUnitType == NAL_UNIT_CODED_SLICE_IDR_N_LP) // IDR or BLA picture
+    if (nalUnitType == NAL_UNIT_CODED_SLICE_IDR_W_RADL)
     {
-        // mark all pictures as not used for reference
-        TComPic* iterPic = m_picList.first();
+        /* If the nal_unit_type is IDR, all pictures in the reference picture
+         * list are marked as "unused for reference" */
+        Frame* iterPic = m_picList.first();
         while (iterPic)
         {
             if (iterPic->getPOC() != pocCurr)
-                iterPic->getSlice()->setReferenced(false);
+                iterPic->m_picSym->m_bHasReferences = false;
             iterPic = iterPic->m_next;
         }
-
-        if (nalUnitType == NAL_UNIT_CODED_SLICE_BLA_W_LP
-            || nalUnitType == NAL_UNIT_CODED_SLICE_BLA_W_RADL
-            || nalUnitType == NAL_UNIT_CODED_SLICE_BLA_N_LP)
-        {
-            m_pocCRA = pocCurr;
-        }
     }
     else // CRA or No DR
     {
-        if (m_bRefreshPending == true && pocCurr > m_pocCRA) // CRA reference marking pending
+        if (m_bRefreshPending && pocCurr > m_pocCRA)
         {
-            TComPic* iterPic = m_picList.first();
+            /* If the bRefreshPending flag is true (a deferred decoding refresh
+             * is pending) and the current temporal reference is greater than
+             * the temporal reference of the latest CRA picture (pocCRA), mark
+             * all reference pictures except the latest CRA picture as "unused
+             * for reference" and set the bRefreshPending flag to false */
+            Frame* iterPic = m_picList.first();
             while (iterPic)
             {
                 if (iterPic->getPOC() != pocCurr && iterPic->getPOC() != m_pocCRA)
-                    iterPic->getSlice()->setReferenced(false);
+                    iterPic->m_picSym->m_bHasReferences = false;
                 iterPic = iterPic->m_next;
             }
 
             m_bRefreshPending = false;
         }
-        if (nalUnitType == NAL_UNIT_CODED_SLICE_CRA) // CRA picture found
+        if (nalUnitType == NAL_UNIT_CODED_SLICE_CRA)
         {
+            /* If the nal_unit_type is CRA, set the bRefreshPending flag to true
+             * and pocCRA to the temporal reference of the current picture */
             m_bRefreshPending = true;
             m_pocCRA = pocCurr;
         }
     }
+
+    /* Note that the current picture is already placed in the reference list and
+     * its marking is not changed.  If the current picture has a nal_ref_idc
+     * that is not 0, it will remain marked as "used for reference" */
 }
 
-/** Function for applying picture marking based on the Reference Picture Set in pReferencePictureSet */
-void DPB::applyReferencePictureSet(TComReferencePictureSet *rps, int curPoc)
+/** Function for applying picture marking based on the Reference Picture Set */
+void DPB::applyReferencePictureSet(RPS *rps, int curPoc)
 {
-    TComPic* outPic;
-    int i, isReference;
-
     // loop through all pictures in the reference picture buffer
-    TComPic* iterPic = m_picList.first();
-
+    Frame* iterPic = m_picList.first();
     while (iterPic)
     {
-        outPic = iterPic;
-        iterPic = iterPic->m_next;
-
-        if (!outPic->getSlice()->isReferenced())
+        if (iterPic->m_POC != curPoc && iterPic->m_picSym->m_bHasReferences)
         {
-            continue;
-        }
-
-        isReference = 0;
-        // loop through all pictures in the Reference Picture Set
-        // to see if the picture should be kept as reference picture
-        for (i = 0; i < rps->getNumberOfPositivePictures() + rps->getNumberOfNegativePictures(); i++)
-        {
-            if (!outPic->getIsLongTerm() && outPic->getPicSym()->getSlice()->getPOC() == curPoc + rps->getDeltaPOC(i))
-            {
-                isReference = 1;
-                outPic->setUsedByCurr(rps->getUsed(i) == 1);
-                outPic->setIsLongTerm(0);
-            }
-        }
-
-        for (; i < rps->getNumberOfPictures(); i++)
-        {
-            if (rps->getCheckLTMSBPresent(i) == true)
+            // loop through all pictures in the Reference Picture Set
+            // to see if the picture should be kept as reference picture
+            bool referenced = false;
+            for (int i = 0; i < rps->numberOfPositivePictures + rps->numberOfNegativePictures; i++)
             {
-                if (outPic->getIsLongTerm() && (outPic->getPicSym()->getSlice()->getPOC()) == rps->getPOC(i))
+                if (iterPic->m_POC == curPoc + rps->deltaPOC[i])
                 {
-                    isReference = 1;
-                    outPic->setUsedByCurr(rps->getUsed(i) == 1);
-                }
-            }
-            else
-            {
-                if (outPic->getIsLongTerm() && (outPic->getPicSym()->getSlice()->getPOC() %
-                                                (1 << outPic->getPicSym()->getSlice()->getSPS()->getBitsForPOC())) == rps->getPOC(i) %
-                    (1 << outPic->getPicSym()->getSlice()->getSPS()->getBitsForPOC()))
-                {
-                    isReference = 1;
-                    outPic->setUsedByCurr(rps->getUsed(i) == 1);
+                    referenced = true;
+                    break;
                 }
             }
+            if (!referenced)
+                iterPic->m_picSym->m_bHasReferences = false;
         }
-
-        // mark the picture as "unused for reference" if it is not in
-        // the Reference Picture Set
-        if (outPic->getPicSym()->getSlice()->getPOC() != curPoc && isReference == 0)
-        {
-            outPic->getSlice()->setReferenced(false);
-            outPic->setUsedByCurr(0);
-            outPic->setIsLongTerm(0);
-        }
+        iterPic = iterPic->m_next;
     }
 }
 
-/** Function for deciding the nal_unit_type.
- * \param pocCurr POC of the current picture
- * \returns the nal unit type of the picture
- * This function checks the configuration and returns the appropriate nal_unit_type for the picture.
- */
-NalUnitType DPB::getNalUnitType(int curPOC, int lastIDR, TComPic* pic)
+/* deciding the nal_unit_type */
+NalUnitType DPB::getNalUnitType(int curPOC, bool bIsKeyFrame)
 {
-    if (curPOC == 0)
-    {
+    if (!curPOC)
         return NAL_UNIT_CODED_SLICE_IDR_W_RADL;
-    }
-    if (pic->m_lowres.bKeyframe)
-    {
-        if (m_bOpenGOP)
-        {
-            return NAL_UNIT_CODED_SLICE_CRA;
-        }
-        else
-        {
-            return NAL_UNIT_CODED_SLICE_IDR_W_RADL;
-        }
-    }
-    if (m_pocCRA > 0)
-    {
-        if (curPOC < m_pocCRA)
-        {
-            // All leading pictures are being marked as TFD pictures here since current encoder uses all
-            // reference pictures while encoding leading pictures. An encoder can ensure that a leading
-            // picture can be still decodable when random accessing to a CRA/CRANT/BLA/BLANT picture by
-            // controlling the reference pictures used for encoding that leading picture. Such a leading
-            // picture need not be marked as a TFD picture.
-            return NAL_UNIT_CODED_SLICE_RASL_R;
-        }
-    }
-    if (lastIDR > 0)
-    {
-        if (curPOC < lastIDR)
-        {
-            return NAL_UNIT_CODED_SLICE_RADL_R;
-        }
-    }
-    return NAL_UNIT_CODED_SLICE_TRAIL_R;
-}
 
-static inline int getLSB(int poc, int maxLSB)
-{
-    if (poc >= 0)
-    {
-        return poc % maxLSB;
-    }
-    else
-    {
-        return (maxLSB - ((-poc) % maxLSB)) % maxLSB;
-    }
-}
+    if (bIsKeyFrame)
+        return m_bOpenGOP ? NAL_UNIT_CODED_SLICE_CRA : NAL_UNIT_CODED_SLICE_IDR_W_RADL;
 
-// Function will arrange the long-term pictures in the decreasing order of poc_lsb_lt,
-// and among the pictures with the same lsb, it arranges them in increasing delta_poc_msb_cycle_lt value
-void DPB::arrangeLongtermPicturesInRPS(TComSlice *slice)
-{
-    TComReferencePictureSet *rps = slice->getRPS();
+    if (m_pocCRA && curPOC < m_pocCRA)
+        // All leading pictures are being marked as TFD pictures here since
+        // current encoder uses all reference pictures while encoding leading
+        // pictures. An encoder can ensure that a leading picture can be still
+        // decodable when random accessing to a CRA/CRANT/BLA/BLANT picture by
+        // controlling the reference pictures used for encoding that leading
+        // picture. Such a leading picture need not be marked as a TFD picture.
+        return NAL_UNIT_CODED_SLICE_RASL_R;
 
-    if (!rps->getNumberOfLongtermPictures())
-    {
-        return;
-    }
-
-    // Arrange long-term reference pictures in the correct order of LSB and MSB,
-    // and assign values for pocLSBLT and MSB present flag
-    int longtermPicsPoc[MAX_NUM_REF_PICS], longtermPicsLSB[MAX_NUM_REF_PICS], indices[MAX_NUM_REF_PICS];
-    int longtermPicsMSB[MAX_NUM_REF_PICS];
-    bool mSBPresentFlag[MAX_NUM_REF_PICS];
-    ::memset(longtermPicsPoc, 0, sizeof(longtermPicsPoc));  // Store POC values of LTRP
-    ::memset(longtermPicsLSB, 0, sizeof(longtermPicsLSB));  // Store POC LSB values of LTRP
-    ::memset(longtermPicsMSB, 0, sizeof(longtermPicsMSB));  // Store POC LSB values of LTRP
-    ::memset(indices, 0, sizeof(indices));                  // Indices to aid in tracking sorted LTRPs
-    ::memset(mSBPresentFlag, 0, sizeof(mSBPresentFlag));    // Indicate if MSB needs to be present
-
-    // Get the long-term reference pictures
-    int offset = rps->getNumberOfNegativePictures() + rps->getNumberOfPositivePictures();
-    int i, ctr = 0;
-    int maxPicOrderCntLSB = 1 << slice->getSPS()->getBitsForPOC();
-    for (i = rps->getNumberOfPictures() - 1; i >= offset; i--, ctr++)
-    {
-        longtermPicsPoc[ctr] = rps->getPOC(i);                                  // LTRP POC
-        longtermPicsLSB[ctr] = getLSB(longtermPicsPoc[ctr], maxPicOrderCntLSB); // LTRP POC LSB
-        indices[ctr] = i;
-        longtermPicsMSB[ctr] = longtermPicsPoc[ctr] - longtermPicsLSB[ctr];
-    }
-
-    int numLongPics = rps->getNumberOfLongtermPictures();
-    assert(ctr == numLongPics);
+    if (m_lastIDR && curPOC < m_lastIDR)
+        return NAL_UNIT_CODED_SLICE_RADL_R;
 
-    // Arrange pictures in decreasing order of MSB;
-    for (i = 0; i < numLongPics; i++)
-    {
-        for (int j = 0; j < numLongPics - 1; j++)
-        {
-            if (longtermPicsMSB[j] < longtermPicsMSB[j + 1])
-            {
-                std::swap(longtermPicsPoc[j], longtermPicsPoc[j + 1]);
-                std::swap(longtermPicsLSB[j], longtermPicsLSB[j + 1]);
-                std::swap(longtermPicsMSB[j], longtermPicsMSB[j + 1]);
-                std::swap(indices[j], indices[j + 1]);
-            }
-        }
-    }
-
-    for (i = 0; i < numLongPics; i++)
-    {
-        // Check if MSB present flag should be enabled.
-        // Check if the buffer contains any pictures that have the same LSB.
-        TComPic* iterPic = m_picList.first();
-        while (iterPic)
-        {
-            if ((getLSB(iterPic->getPOC(), maxPicOrderCntLSB) == longtermPicsLSB[i])   && // Same LSB
-                (iterPic->getSlice()->isReferenced()) &&                                  // Reference picture
-                (iterPic->getPOC() != longtermPicsPoc[i]))                                // Not the LTRP itself
-            {
-                mSBPresentFlag[i] = true;
-                break;
-            }
-            iterPic = iterPic->m_next;
-        }
-    }
-
-    // tempArray for usedByCurr flag
-    bool tempArray[MAX_NUM_REF_PICS];
-    ::memset(tempArray, 0, sizeof(tempArray));
-    for (i = 0; i < numLongPics; i++)
-    {
-        tempArray[i] = rps->getUsed(indices[i]) ? true : false;
-    }
-
-    // Now write the final values;
-    ctr = 0;
-    int currMSB = 0, currLSB = 0;
-    // currPicPoc = currMSB + currLSB
-    currLSB = getLSB(slice->getPOC(), maxPicOrderCntLSB);
-    currMSB = slice->getPOC() - currLSB;
-
-    for (i = rps->getNumberOfPictures() - 1; i >= offset; i--, ctr++)
-    {
-        rps->setPOC(i, longtermPicsPoc[ctr]);
-        rps->setDeltaPOC(i, -slice->getPOC() + longtermPicsPoc[ctr]);
-        rps->setUsed(i, tempArray[ctr]);
-        rps->setPocLSBLT(i, longtermPicsLSB[ctr]);
-        rps->setDeltaPocMSBCycleLT(i, (currMSB - (longtermPicsPoc[ctr] - longtermPicsLSB[ctr])) / maxPicOrderCntLSB);
-        rps->setDeltaPocMSBPresentFlag(i, mSBPresentFlag[ctr]);
-
-        assert(rps->getDeltaPocMSBCycleLT(i) >= 0); // Non-negative value
-    }
-
-    for (i = rps->getNumberOfPictures() - 1, ctr = 1; i >= offset; i--, ctr++)
-    {
-        for (int j = rps->getNumberOfPictures() - 1 - ctr; j >= offset; j--)
-        {
-            // Here at the encoder we know that we have set the full POC value for the LTRPs, hence we
-            // don't have to check the MSB present flag values for this constraint.
-            assert(rps->getPOC(i) != rps->getPOC(j)); // If assert fails, LTRP entry repeated in RPS!!!
-        }
-    }
+    return NAL_UNIT_CODED_SLICE_TRAIL_R;
 }

x265-hg20140324.tar.bz2/source/encoder/dpb.h -> x265-hg20140928.tar.bz2/source/encoder/dpb.h Changed

x265-hg20140324.tar.bz2/source/encoder/encoder.cpp -> x265-hg20140928.tar.bz2/source/encoder/encoder.cpp Changed

@@ -18,36 +18,39 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/NAL.h"
-#include "TLibCommon/CommonDef.h"
-#include "TLibCommon/TComPic.h"
-#include "TLibCommon/TComPicYuv.h"
-#include "TLibCommon/TComRom.h"
+#include "common.h"
 #include "primitives.h"
 #include "threadpool.h"
-#include "common.h"
 #include "param.h"
+#include "frame.h"
+
+#include "TLibCommon/TComPicYuv.h"
+#include "TLibCommon/TComRom.h"
 
-#include "TLibEncoder/NALwrite.h"
 #include "bitcost.h"
 #include "encoder.h"
 #include "slicetype.h"
 #include "frameencoder.h"
 #include "ratecontrol.h"
 #include "dpb.h"
+#include "nal.h"
 
 #include "x265.h"
 
-#if HAVE_INT_TYPES_H
-#define __STDC_FORMAT_MACROS
-#include <inttypes.h>
-#define LL "%" PRIu64
-#else
-#define LL "%lld"
-#endif
+namespace x265 {
+const char g_sliceTypeToChar[] = {'B', 'P', 'I'};
+}
+
+static const char *summaryCSVHeader =
+    "Command, Date/Time, Elapsed Time, FPS, Bitrate, "
+    "Y PSNR, U PSNR, V PSNR, Global PSNR, SSIM, SSIM (dB), "
+    "I count, I ave-QP, I kpbs, I-PSNR Y, I-PSNR U, I-PSNR V, I-SSIM (dB), "
+    "P count, P ave-QP, P kpbs, P-PSNR Y, P-PSNR U, P-PSNR V, P-SSIM (dB), "
+    "B count, B ave-QP, B kpbs, B-PSNR Y, B-PSNR U, B-PSNR V, B-SSIM (dB), "
+    "Version\n";
 
 using namespace x265;
 
@@ -56,32 +59,20 @@
     m_aborted = false;
     m_encodedFrameNum = 0;
     m_pocLast = -1;
-    m_maxRefPicNum = 0;
     m_curEncoder = 0;
     m_numLumaWPFrames = 0;
     m_numChromaWPFrames = 0;
+    m_numLumaWPBiFrames = 0;
+    m_numChromaWPBiFrames = 0;
     m_lookahead = NULL;
     m_frameEncoder = NULL;
     m_rateControl = NULL;
     m_dpb = NULL;
-    m_nals = NULL;
-    m_packetData = NULL;
+    m_exportedPic = NULL;
+    m_numDelayedPic = 0;
     m_outputCount = 0;
     m_csvfpt = NULL;
-    param = NULL;
-
-#if ENC_DEC_TRACE
-    g_hTrace = fopen("TraceEnc.txt", "wb");
-    g_bJustDoIt = g_bEncDecTraceDisable;
-    g_nSymbolCounter = 0;
-#endif
-}
-
-Encoder::~Encoder()
-{
-#if ENC_DEC_TRACE
-    fclose(g_hTrace);
-#endif
+    m_param = NULL;
 }
 
 void Encoder::create()
@@ -89,67 +80,106 @@
     if (!primitives.sad[0])
     {
         // this should be an impossible condition when using our public API, and indicates a serious bug.
-        x265_log(param, X265_LOG_ERROR, "Primitives must be initialized before encoder is created\n");
+        x265_log(m_param, X265_LOG_ERROR, "Primitives must be initialized before encoder is created\n");
         abort();
     }
 
-    m_frameEncoder = new FrameEncoder[param->frameNumThreads];
-    if (m_frameEncoder)
+    m_frameEncoder = new FrameEncoder[m_param->frameNumThreads];
+    for (int i = 0; i < m_param->frameNumThreads; i++)
+        m_frameEncoder[i].setThreadPool(m_threadPool);
+
+    if (!m_scalingList.init())
     {
-        for (int i = 0; i < param->frameNumThreads; i++)
-        {
-            m_frameEncoder[i].setThreadPool(m_threadPool);
-        }
+        x265_log(m_param, X265_LOG_ERROR, "Unable to allocate scaling list arrays\n");
+        m_aborted = true;
     }
-    m_lookahead = new Lookahead(this, m_threadPool);
-    m_dpb = new DPB(this);
-    m_rateControl = new RateControl(this);
+    else if (!m_param->scalingLists || !strcmp(m_param->scalingLists, "off"))
+        m_scalingList.m_bEnabled = false;
+    else if (!strcmp(m_param->scalingLists, "default"))
+        m_scalingList.setDefaultScalingList();
+    else if (m_scalingList.parseScalingList(m_param->scalingLists))
+        m_aborted = true;
+    m_scalingList.setupQuantMatrices();
+
+    /* Allocate thread local data shared by all frame encoders */
+    const int poolThreadCount = ThreadPool::getThreadPool()->getThreadCount();
+    int numLocalData = m_param->frameNumThreads;
+    if (m_param->bEnableWavefront)
+        numLocalData = poolThreadCount;
+    m_threadLocalData = new ThreadLocalData[numLocalData];
+    for (int i = 0; i < numLocalData; i++)
+    {
+        m_threadLocalData[i].analysis.initSearch(m_param, m_scalingList);
+        m_threadLocalData[i].analysis.create(g_maxCUDepth + 1, g_maxCUSize);
+    }
+
+    for (int i = 0; i < m_param->frameNumThreads; i++)
+        m_frameEncoder[i].m_tld = &m_threadLocalData[i];
+
+    m_lookahead = new Lookahead(m_param, m_threadPool, this);
+    m_dpb = new DPB(m_param);
+    m_rateControl = new RateControl(m_param);
+
+    initSPS(&m_sps);
+    initPPS(&m_pps);
 
     /* Try to open CSV file handle */
-    if (param->csvfn)
+    if (m_param->csvfn)
     {
-        m_csvfpt = fopen(param->csvfn, "r");
+        m_csvfpt = fopen(m_param->csvfn, "r");
         if (m_csvfpt)
         {
             // file already exists, re-open for append
             fclose(m_csvfpt);
-            m_csvfpt = fopen(param->csvfn, "ab");
+            m_csvfpt = fopen(m_param->csvfn, "ab");
         }
         else
         {
             // new CSV file, write header
-            m_csvfpt = fopen(param->csvfn, "wb");
+            m_csvfpt = fopen(m_param->csvfn, "wb");
             if (m_csvfpt)
             {
-                if (param->logLevel >= X265_LOG_DEBUG)
-                    fprintf(m_csvfpt, "Encode Order, Type, POC, QP, Bits, Y PSNR, U PSNR, V PSNR, YUV PSNR, SSIM, SSIM (dB), Encoding time, Elapsed time, List 0, List 1\n");
+                if (m_param->logLevel >= X265_LOG_DEBUG)
+                {
+                    fprintf(m_csvfpt, "Encode Order, Type, POC, QP, Bits, ");
+                    if (m_param->rc.rateControlMode == X265_RC_CRF)
+                        fprintf(m_csvfpt, "RateFactor, ");
+                    fprintf(m_csvfpt, "Y PSNR, U PSNR, V PSNR, YUV PSNR, SSIM, SSIM (dB), "
+                                      "Encoding time, Elapsed time, List 0, List 1\n");
+                }
                 else
-                    fprintf(m_csvfpt, "Command, Date/Time, Elapsed Time, FPS, Bitrate, Y PSNR, U PSNR, V PSNR, Global PSNR, SSIM, SSIM (dB), Version\n");
+                    fputs(summaryCSVHeader, m_csvfpt);
             }
         }
     }
+
+    m_aborted |= parseLambdaFile(m_param);
 }
 
 void Encoder::destroy()
 {
+    if (m_exportedPic)
+    {
+        ATOMIC_DEC(&m_exportedPic->m_countRefEncoders);
+        m_exportedPic = NULL;
+    }
+
+    if (m_rateControl)
+        m_rateControl->terminate(); // unblock all blocked RC calls
+
     if (m_frameEncoder)
     {
-        for (int i = 0; i < param->frameNumThreads; i++)
+        for (int i = 0; i < m_param->frameNumThreads; i++)
         {
             // Ensure frame encoder is idle before destroying it
-            m_frameEncoder[i].getEncodedPicture(NULL);
+            m_frameEncoder[i].getEncodedPicture(m_nalList);
             m_frameEncoder[i].destroy();
         }
 
         delete [] m_frameEncoder;
     }
 
-    while (!m_freeList.empty())
-    {
-        TComPic* pic = m_freeList.popFront();
-        pic->destroy();
-        delete pic;
-    }
+    delete [] m_threadLocalData;
 
     if (m_lookahead)
     {
@@ -158,15 +188,17 @@
     }
 
     delete m_dpb;
-    delete m_rateControl;
-
+    if (m_rateControl)
+    {
+        m_rateControl->destroy();
+        delete m_rateControl;
+    }
     // thread pool release should always happen last
     if (m_threadPool)
         m_threadPool->release();
 
-    X265_FREE(m_nals);
-    X265_FREE(m_packetData);
-    X265_FREE(param);
+    free(m_param->rc.statFileName); // alloc'd by strdup
+    X265_FREE(m_param);
     if (m_csvfpt)
         fclose(m_csvfpt);
 }
@@ -175,141 +207,104 @@
 {
     if (m_frameEncoder)
     {
-        int numRows = (param->sourceHeight + g_maxCUSize - 1) / g_maxCUSize;
-        for (int i = 0; i < param->frameNumThreads; i++)
+        int numRows = (m_param->sourceHeight + g_maxCUSize - 1) / g_maxCUSize;
+        int numCols = (m_param->sourceWidth  + g_maxCUSize - 1) / g_maxCUSize;
+        for (int i = 0; i < m_param->frameNumThreads; i++)
         {
-            if (!m_frameEncoder[i].init(this, numRows))
+            if (!m_frameEncoder[i].init(this, numRows, numCols))
             {
-                x265_log(param, X265_LOG_ERROR, "Unable to initialize frame encoder, aborting\n");
+                x265_log(m_param, X265_LOG_ERROR, "Unable to initialize frame encoder, aborting\n");
                 m_aborted = true;
             }
         }
     }
+    if (m_param->bEmitHRDSEI)
+        m_rateControl->initHRD(&m_sps);
+    if (!m_rateControl->init(&m_sps))
+        m_aborted = true;
     m_lookahead->init();
     m_encodeStartTime = x265_mdate();
 }
 
-int Encoder::getStreamHeaders(NALUnitEBSP **nalunits)
-{
-    return m_frameEncoder->getStreamHeaders(nalunits);
-}
-
 void Encoder::updateVbvPlan(RateControl* rc)
 {
-    int encIdx, curIdx;
-
-    curIdx = (m_curEncoder + param->frameNumThreads - 1) % param->frameNumThreads;
-    encIdx = (curIdx + 1) % param->frameNumThreads;
-    while (encIdx != curIdx)
+    for (int i = 0; i < m_param->frameNumThreads; i++)
     {
-        FrameEncoder *encoder = &m_frameEncoder[encIdx];
-        double bits;
-        bits = encoder->m_rce.frameSizePlanned;
-        if (!encoder->m_rce.isActive)
+        FrameEncoder *encoder = &m_frameEncoder[i];
+        if (encoder->m_rce.isActive && encoder->m_rce.poc != rc->m_curSlice->m_poc)
         {
-            encIdx = (encIdx + 1) % param->frameNumThreads;
-            continue;
+            int64_t bits = (int64_t) X265_MAX(encoder->m_rce.frameSizeEstimated, encoder->m_rce.frameSizePlanned);
+            rc->m_bufferFill -= bits;
+            rc->m_bufferFill = X265_MAX(rc->m_bufferFill, 0);
+            rc->m_bufferFill += encoder->m_rce.bufferRate;
+            rc->m_bufferFill = X265_MIN(rc->m_bufferFill, rc->m_bufferSize);
+            if (rc->m_2pass)
+                rc->m_predictedBits += bits;
         }
-        bits = X265_MAX(bits, encoder->m_rce.frameSizeEstimated);
-        rc->bufferFill -= bits;
-        rc->bufferFill = X265_MAX(rc->bufferFill, 0);
-        rc->bufferFill += encoder->m_rce.bufferRate;
-        rc->bufferFill = X265_MIN(rc->bufferFill, rc->bufferSize);
-        encIdx = (encIdx + 1) % param->frameNumThreads;
     }
 }
 
-#define VERBOSE_RATE 0
-#if VERBOSE_RATE
-static const char* nalUnitTypeToString(NalUnitType type)
-{
-    switch (type)
-    {
-    case NAL_UNIT_CODED_SLICE_TRAIL_R:    return "TRAIL_R";
-    case NAL_UNIT_CODED_SLICE_TRAIL_N:    return "TRAIL_N";
-    case NAL_UNIT_CODED_SLICE_TLA_R:      return "TLA_R";
-    case NAL_UNIT_CODED_SLICE_TSA_N:      return "TSA_N";
-    case NAL_UNIT_CODED_SLICE_STSA_R:     return "STSA_R";
-    case NAL_UNIT_CODED_SLICE_STSA_N:     return "STSA_N";
-    case NAL_UNIT_CODED_SLICE_BLA_W_LP:   return "BLA_W_LP";
-    case NAL_UNIT_CODED_SLICE_BLA_W_RADL: return "BLA_W_RADL";
-    case NAL_UNIT_CODED_SLICE_BLA_N_LP:   return "BLA_N_LP";
-    case NAL_UNIT_CODED_SLICE_IDR_W_RADL: return "IDR_W_RADL";
-    case NAL_UNIT_CODED_SLICE_IDR_N_LP:   return "IDR_N_LP";
-    case NAL_UNIT_CODED_SLICE_CRA:        return "CRA";
-    case NAL_UNIT_CODED_SLICE_RADL_R:     return "RADL_R";
-    case NAL_UNIT_CODED_SLICE_RASL_R:     return "RASL_R";
-    case NAL_UNIT_VPS:                    return "VPS";
-    case NAL_UNIT_SPS:                    return "SPS";
-    case NAL_UNIT_PPS:                    return "PPS";
-    case NAL_UNIT_ACCESS_UNIT_DELIMITER:  return "AUD";
-    case NAL_UNIT_EOS:                    return "EOS";
-    case NAL_UNIT_EOB:                    return "EOB";
-    case NAL_UNIT_FILLER_DATA:            return "FILLER";
-    case NAL_UNIT_PREFIX_SEI:             return "SEI";
-    case NAL_UNIT_SUFFIX_SEI:             return "SEI";
-    default:                              return "UNK";
-    }
-}
-#endif // if VERBOSE_RATE
-
 /**
- \param   flush               force encoder to encode a frame
- \param   pic_in              input original YUV picture or NULL
- \param   pic_out             pointer to reconstructed picture struct
- \param   nalunits            output NAL packets
- \retval                      number of encoded pictures
- */
-int Encoder::encode(bool flush, const x265_picture* pic_in, x265_picture *pic_out, NALUnitEBSP **nalunits)
+ * Feed one new input frame into the encoder, get one frame out. If pic_in is
+ * NULL, a flush condition is implied and pic_in must be NULL for all subsequent
+ * calls for this encoder instance.
+ *
+ * pic_in  input original YUV picture or NULL
+ * pic_out pointer to reconstructed picture struct
+ *
+ * returns 0 if no frames are currently available for output
+ *         1 if frame was output, m_nalList contains access unit
+ *         negative on malloc error or abort */
+int Encoder::encode(const x265_picture* pic_in, x265_picture* pic_out)
 {
     if (m_aborted)
         return -1;
 
+    if (m_exportedPic)
+    {
+        ATOMIC_DEC(&m_exportedPic->m_countRefEncoders);
+        m_exportedPic = NULL;
+        m_dpb->recycleUnreferenced();
+    }
+
     if (pic_in)
     {
-        if (pic_in->colorSpace != param->internalCsp)
+        if (pic_in->colorSpace != m_param->internalCsp)
         {
-            x265_log(param, X265_LOG_ERROR, "Unsupported color space (%d) on input\n",
+            x265_log(m_param, X265_LOG_ERROR, "Unsupported color space (%d) on input\n",
                      pic_in->colorSpace);
             return -1;
         }
         if (pic_in->bitDepth < 8 || pic_in->bitDepth > 16)
         {
-            x265_log(param, X265_LOG_ERROR, "Input bit depth (%d) must be between 8 and 16\n",
+            x265_log(m_param, X265_LOG_ERROR, "Input bit depth (%d) must be between 8 and 16\n",
                      pic_in->bitDepth);
             return -1;
         }
 
-        TComPic *pic;
-        if (m_freeList.empty())
+        Frame *pic;
+        if (m_dpb->m_freeList.empty())
         {
-            pic = new TComPic;
-            if (!pic || !pic->create(this))
+            pic = new Frame;
+            if (!pic->create(m_param, m_sps.vuiParameters.defaultDisplayWindow, m_conformanceWindow))
             {
                 m_aborted = true;
-                x265_log(param, X265_LOG_ERROR, "memory allocation failure, aborting encode\n");
-                if (pic)
-                {
-                    pic->destroy();
-                    delete pic;
-                }
+                x265_log(m_param, X265_LOG_ERROR, "memory allocation failure, aborting encode\n");
+                pic->destroy();
+                delete pic;
                 return -1;
             }
-            if (param->bEnableSAO)
-            {
-                // TODO: these should be allocated on demand within the encoder
-                // NOTE: the SAO pointer from m_frameEncoder for read m_maxSplitLevel, etc, we can remove it later
-                pic->getPicSym()->allocSaoParam(m_frameEncoder->getSAO());
-            }
         }
         else
-            pic = m_freeList.popBack();
+            pic = m_dpb->m_freeList.popBack();
+
         /* Copy input picture into a TComPic, send to lookahead */
-        pic->getSlice()->setPOC(++m_pocLast);
-        pic->reInit(this);
-        pic->getPicYuvOrg()->copyFromPicture(*pic_in, m_pad);
+        pic->m_POC = ++m_pocLast;
+        pic->reinit(m_param);
+        pic->getPicYuvOrg()->copyFromPicture(*pic_in, m_sps.conformanceWindow.rightOffset, m_sps.conformanceWindow.bottomOffset);
         pic->m_userData = pic_in->userData;
         pic->m_pts = pic_in->pts;
+        pic->m_forceqp = pic_in->forceqp;
 
         if (m_pocLast == 0)
             m_firstPts = pic->m_pts;
@@ -318,53 +313,55 @@
 
         // Encoder holds a reference count until collecting stats
         ATOMIC_INC(&pic->m_countRefEncoders);
-        if (param->rc.aqMode || param->bEnableWeightedPred)
-            m_rateControl->calcAdaptiveQuantFrame(pic);
+        bool bEnableWP = m_param->bEnableWeightedPred || m_param->bEnableWeightedBiPred;
+        if (m_param->rc.aqMode || bEnableWP)
+        {
+            if (m_param->rc.cuTree && m_param->rc.bStatRead)
+            {
+                if (!m_rateControl->cuTreeReadFor2Pass(pic))
+                {
+                    m_aborted = 1;
+                    return -1;
+                }
+            }
+            else
+                m_rateControl->calcAdaptiveQuantFrame(pic);
+        }
+        if (pic_in->analysisData.intraData)
+        {
+            pic->m_intraData = pic_in->analysisData.intraData;
+            pic->m_interData = pic_in->analysisData.interData;
+        }
         m_lookahead->addPicture(pic, pic_in->sliceType);
+        m_numDelayedPic++;
     }
-
-    if (flush)
+    else
         m_lookahead->flush();
 
     FrameEncoder *curEncoder = &m_frameEncoder[m_curEncoder];
-    m_curEncoder = (m_curEncoder + 1) % param->frameNumThreads;
+    m_curEncoder = (m_curEncoder + 1) % m_param->frameNumThreads;
     int ret = 0;
 
     // getEncodedPicture() should block until the FrameEncoder has completed
     // encoding the frame.  This is how back-pressure through the API is
     // accomplished when the encoder is full.
-    TComPic *out = curEncoder->getEncodedPicture(nalunits);
+    Frame *out = curEncoder->getEncodedPicture(m_nalList);
 
-    if (!out && flush)
-    {
-        // if the current encoder did not return an output picture and we are
-        // flushing, check all the other encoders in logical order until
-        // we find an output picture or have cycled around.  We cannot return
-        // 0 until the entire stream is flushed
-        // (can only be an issue when --frames < --frame-threads)
-        int flushed = m_curEncoder;
-        do
-        {
-            curEncoder = &m_frameEncoder[m_curEncoder];
-            m_curEncoder = (m_curEncoder + 1) % param->frameNumThreads;
-            out = curEncoder->getEncodedPicture(nalunits);
-        }
-        while (!out && flushed != m_curEncoder);
-    }
     if (out)
     {
+        Slice *slice = out->m_picSym->m_slice;
         if (pic_out)
         {
             TComPicYuv *recpic = out->getPicYuvRec();
-            pic_out->poc = out->getSlice()->getPOC();
+            pic_out->poc = slice->m_poc;
             pic_out->bitDepth = X265_DEPTH;
             pic_out->userData = out->m_userData;
-            pic_out->colorSpace = param->internalCsp;
+            pic_out->colorSpace = m_param->internalCsp;
 
             pic_out->pts = out->m_pts;
             pic_out->dts = out->m_dts;
 
-            switch (out->getSlice()->getSliceType())
+            switch (slice->m_sliceType)
             {
             case I_SLICE:
                 pic_out->sliceType = out->m_lowres.bKeyframe ? X265_TYPE_IDR : X265_TYPE_I;
@@ -385,48 +382,80 @@
             pic_out->stride[2] = recpic->getCStride() * sizeof(pixel);
         }
 
-        if (out->getSlice()->getSliceType() == P_SLICE)
+        if (m_param->analysisMode)
+        {
+            pic_out->analysisData.interData = out->m_interData;
+            pic_out->analysisData.intraData = out->m_intraData;
+            pic_out->analysisData.numCUsInFrame = out->m_picSym->getNumberOfCUsInFrame();
+            pic_out->analysisData.numPartitions = out->m_picSym->getNumPartition();
+        }
+
+        if (slice->m_sliceType == P_SLICE)
         {
-            if (out->getSlice()->m_weightPredTable[0][0][0].bPresentFlag)
+            if (slice->m_weightPredTable[0][0][0].bPresentFlag)
                 m_numLumaWPFrames++;
-            if (out->getSlice()->m_weightPredTable[0][0][1].bPresentFlag ||
-                out->getSlice()->m_weightPredTable[0][0][2].bPresentFlag)
+            if (slice->m_weightPredTable[0][0][1].bPresentFlag ||
+                slice->m_weightPredTable[0][0][2].bPresentFlag)
                 m_numChromaWPFrames++;
         }
-
-        /* calculate the size of the access unit, excluding:
-         *  - any AnnexB contributions (start_code_prefix, zero_byte, etc.,)
-         *  - SEI NAL units
-         */
-        uint32_t numRBSPBytes = 0;
-        for (int count = 0; nalunits[count] != NULL; count++)
+        else if (slice->m_sliceType == B_SLICE)
         {
-            uint32_t numRBSPBytes_nal = nalunits[count]->m_packetSize;
-#if VERBOSE_RATE
-            printf("*** %6s numBytesInNALunit: %u\n", nalUnitTypeToString(nalunits[count]->m_nalUnitType), numRBSPBytes_nal);
-#endif
-            if (nalunits[count]->m_nalUnitType != NAL_UNIT_PREFIX_SEI && nalunits[count]->m_nalUnitType != NAL_UNIT_SUFFIX_SEI)
+            bool bLuma = false, bChroma = false;
+            for (int l = 0; l < 2; l++)
             {
-                numRBSPBytes += numRBSPBytes_nal;
+                if (slice->m_weightPredTable[l][0][0].bPresentFlag)
+                    bLuma = true;
+                if (slice->m_weightPredTable[l][0][1].bPresentFlag ||
+                    slice->m_weightPredTable[l][0][2].bPresentFlag)
+                    bChroma = true;
             }
+
+            if (bLuma)
+                m_numLumaWPBiFrames++;
+            if (bChroma)
+                m_numChromaWPBiFrames++;
         }
-        uint64_t bits = numRBSPBytes * 8;
-        m_rateControl->rateControlEnd(out, bits, &curEncoder->m_rce);
-        finishFrameStats(out, curEncoder, bits);
+        if (m_aborted)
+            return -1;
 
+        finishFrameStats(out, curEncoder, curEncoder->m_accessUnitBits);
         // Allow this frame to be recycled if no frame encoders are using it for reference
-        ATOMIC_DEC(&out->m_countRefEncoders);
-        m_dpb->recycleUnreferenced(m_freeList);
+        if (!pic_out)
+        {
+            ATOMIC_DEC(&out->m_countRefEncoders);
+            m_dpb->recycleUnreferenced();
+        }
+        else
+            m_exportedPic = out;
+
+        m_numDelayedPic--;
+
         ret = 1;
     }
 
-    if (!m_lookahead->outputQueue.empty())
+    // pop a single frame from decided list, then provide to frame encoder
+    // curEncoder is guaranteed to be idle at this point
+    Frame* fenc = m_lookahead->getDecidedPicture();
+    if (fenc)
     {
-        // pop a single frame from decided list, then provide to frame encoder
-        // curEncoder is guaranteed to be idle at this point
-        TComPic *fenc = m_lookahead->outputQueue.popFront();
-
-        m_encodedFrameNum++;
+        // give this picture a TComPicSym instance before encoding
+        if (m_dpb->m_picSymFreeList)
+        {
+            fenc->m_picSym = m_dpb->m_picSymFreeList;
+            m_dpb->m_picSymFreeList = m_dpb->m_picSymFreeList->m_freeListNext;
+            fenc->m_reconPicYuv = fenc->m_picSym->m_reconPicYuv;
+        }
+        else
+        {
+            fenc->allocPicSym(m_param);
+            Slice* slice = fenc->m_picSym->m_slice;
+            slice->m_pic = fenc;
+            slice->m_sps = &m_sps;
+            slice->m_pps = &m_pps;
+            slice->m_maxNumMergeCand = m_param->maxNumMergeCand;
+            slice->m_endCUAddr = slice->realEndAddress(fenc->getNumCUsInFrame() * NUM_CU_PARTITIONS);
+        }
+        curEncoder->m_rce.encodeOrder = m_encodedFrameNum++;
         if (m_bframeDelay)
         {
             int64_t *prevReorderedPts = m_prevReorderedPts;
@@ -438,18 +467,17 @@
         else
             fenc->m_dts = fenc->m_reorderedPts;
 
-        // Initialize slice for encoding with this FrameEncoder
-        curEncoder->initSlice(fenc);
-
         // determine references, setup RPS, etc
         m_dpb->prepareEncode(fenc);
 
-        // set slice QP
-        m_rateControl->rateControlStart(fenc, m_lookahead, &curEncoder->m_rce, this);
+        if (m_param->rc.rateControlMode != X265_RC_CQP)
+            m_lookahead->getEstimatedPictureCost(fenc);
 
         // Allow FrameEncoder::compressFrame() to start in a worker thread
-        curEncoder->m_enable.trigger();
+        curEncoder->startCompressFrame(fenc);
     }
+    else if (m_encodedFrameNum)
+        m_rateControl->setFinalFrameCount(m_encodedFrameNum);
 
     return ret;
 }
@@ -477,24 +505,55 @@
     m_totalQp += aveQp;
 }
 
+char* Encoder::statsCSVString(EncStats& stat, char* buffer)
+{
+    if (!stat.m_numPics)
+    {
+        sprintf(buffer, "-, -, -, -, -, -, -, ");
+        return buffer;
+    }
+
+    double fps = (double)m_param->fpsNum / m_param->fpsDenom;
+    double scale = fps / 1000 / (double)stat.m_numPics;
+
+    int len = sprintf(buffer, "%-6u, ", stat.m_numPics);
+
+    len += sprintf(buffer + len, "%2.2lf, ", stat.m_totalQp / (double)stat.m_numPics);
+    len += sprintf(buffer + len, "%-8.2lf, ", stat.m_accBits * scale);
+    if (m_param->bEnablePsnr)
+    {
+        len += sprintf(buffer + len, "%.3lf, %.3lf, %.3lf, ",
+                       stat.m_psnrSumY / (double)stat.m_numPics,
+                       stat.m_psnrSumU / (double)stat.m_numPics,
+                       stat.m_psnrSumV / (double)stat.m_numPics);
+    }
+    else
+        len += sprintf(buffer + len, "-, -, -, ");
+
+    if (m_param->bEnableSsim)
+        sprintf(buffer + len, "%.3lf, ", x265_ssim2dB(stat.m_globalSsim / (double)stat.m_numPics));
+    else
+        sprintf(buffer + len, "-, ");
+    return buffer;
+}
 
 char* Encoder::statsString(EncStats& stat, char* buffer)
 {
-    double fps = (double)param->fpsNum / param->fpsDenom;
+    double fps = (double)m_param->fpsNum / m_param->fpsDenom;
     double scale = fps / 1000 / (double)stat.m_numPics;
 
-    int len = sprintf(buffer, "%-6d ", stat.m_numPics);
+    int len = sprintf(buffer, "%6u, ", stat.m_numPics);
 
     len += sprintf(buffer + len, "Avg QP:%2.2lf", stat.m_totalQp / (double)stat.m_numPics);
     len += sprintf(buffer + len, "  kb/s: %-8.2lf", stat.m_accBits * scale);
-    if (param->bEnablePsnr)
+    if (m_param->bEnablePsnr)
     {
         len += sprintf(buffer + len, "  PSNR Mean: Y:%.3lf U:%.3lf V:%.3lf",
                        stat.m_psnrSumY / (double)stat.m_numPics,
                        stat.m_psnrSumU / (double)stat.m_numPics,
                        stat.m_psnrSumV / (double)stat.m_numPics);
     }
-    if (param->bEnableSsim)
+    if (m_param->bEnableSsim)
     {
         sprintf(buffer + len, "  SSIM Mean: %.6lf (%.3lfdB)",
                 stat.m_globalSsim / (double)stat.m_numPics,
@@ -505,7 +564,56 @@
 
 void Encoder::printSummary()
 {
-#if LOG_CU_STATISTICS
+    if (m_param->logLevel < X265_LOG_INFO)
+        return;
+
+    char buffer[200];
+    if (m_analyzeI.m_numPics)
+        x265_log(m_param, X265_LOG_INFO, "frame I: %s\n", statsString(m_analyzeI, buffer));
+    if (m_analyzeP.m_numPics)
+        x265_log(m_param, X265_LOG_INFO, "frame P: %s\n", statsString(m_analyzeP, buffer));
+    if (m_analyzeB.m_numPics)
+        x265_log(m_param, X265_LOG_INFO, "frame B: %s\n", statsString(m_analyzeB, buffer));
+    if (m_analyzeAll.m_numPics)
+        x265_log(m_param, X265_LOG_INFO, "global : %s\n", statsString(m_analyzeAll, buffer));
+    if (m_param->bEnableWeightedPred && m_analyzeP.m_numPics)
+    {
+        x265_log(m_param, X265_LOG_INFO, "Weighted P-Frames: Y:%.1f%% UV:%.1f%%\n",
+            (float)100.0 * m_numLumaWPFrames / m_analyzeP.m_numPics,
+            (float)100.0 * m_numChromaWPFrames / m_analyzeP.m_numPics);
+    }
+    if (m_param->bEnableWeightedBiPred && m_analyzeB.m_numPics)
+    {
+        x265_log(m_param, X265_LOG_INFO, "Weighted B-Frames: Y:%.1f%% UV:%.1f%%\n",
+            (float)100.0 * m_numLumaWPBiFrames / m_analyzeB.m_numPics,
+            (float)100.0 * m_numChromaWPBiFrames / m_analyzeB.m_numPics);
+    }
+    int pWithB = 0;
+    for (int i = 0; i <= m_param->bframes; i++)
+        pWithB += m_lookahead->m_histogram[i];
+
+    if (pWithB)
+    {
+        int p = 0;
+        for (int i = 0; i <= m_param->bframes; i++)
+            p += sprintf(buffer + p, "%.1f%% ", 100. * m_lookahead->m_histogram[i] / pWithB);
+
+        x265_log(m_param, X265_LOG_INFO, "consecutive B-frames: %s\n", buffer);
+    }
+    if (m_param->bLossless)
+    {
+        float frameSize = (float)(m_param->sourceWidth - m_sps.conformanceWindow.rightOffset) *
+                                 (m_param->sourceHeight - m_sps.conformanceWindow.bottomOffset);
+        float uncompressed = frameSize * X265_DEPTH * m_analyzeAll.m_numPics;
+
+        x265_log(m_param, X265_LOG_INFO, "lossless compression ratio %.2f::1\n", uncompressed / m_analyzeAll.m_accBits);
+    }
+
+    if (!m_param->bLogCuStats)
+        return;
+
+    const int poolThreadCount = m_threadPool ? m_threadPool->getThreadCount() : 1;
+
     for (int sliceType = 2; sliceType >= 0; sliceType--)
     {
         if (sliceType == P_SLICE && !m_analyzeP.m_numPics)
@@ -514,33 +622,28 @@
             continue;
 
         StatisticLog finalLog;
-        for (int depth = 0; depth < (int)g_maxCUDepth; depth++)
+        for (uint32_t depth = 0; depth <= g_maxCUDepth; depth++)
         {
-            for (int j = 0; j < param->frameNumThreads; j++)
+            for (int i = 0; i < poolThreadCount; i++)
             {
-                for (int row = 0; row < m_frameEncoder[0].m_numRows; row++)
+                StatisticLog& enclog = m_threadLocalData[i].analysis.m_sliceTypeLog[sliceType];
+                if (depth == 0)
+                    finalLog.totalCu += enclog.totalCu;
+                finalLog.cntIntra[depth] += enclog.cntIntra[depth];
+                for (int m = 0; m < INTER_MODES; m++)
                 {
-                    StatisticLog& enclog = m_frameEncoder[j].m_rows[row].m_cuCoder.m_sliceTypeLog[sliceType];
-                    if (depth == 0)
-                        finalLog.totalCu += enclog.totalCu;
-                    finalLog.cntIntra[depth] += enclog.cntIntra[depth];
-                    for (int m = 0; m < INTER_MODES; m++)
-                    {
-                        if (m < INTRA_MODES)
-                        {
-                            finalLog.cuIntraDistribution[depth][m] += enclog.cuIntraDistribution[depth][m];
-                        }
-                        finalLog.cuInterDistribution[depth][m] += enclog.cuInterDistribution[depth][m];
-                    }
+                    if (m < INTRA_MODES)
+                        finalLog.cuIntraDistribution[depth][m] += enclog.cuIntraDistribution[depth][m];
+                    finalLog.cuInterDistribution[depth][m] += enclog.cuInterDistribution[depth][m];
+                }
 
-                    if (depth == (int)g_maxCUDepth - 1)
-                        finalLog.cntIntraNxN += enclog.cntIntraNxN;
-                    if (sliceType != I_SLICE)
-                    {
-                        finalLog.cntTotalCu[depth] += enclog.cntTotalCu[depth];
-                        finalLog.cntInter[depth] += enclog.cntInter[depth];
-                        finalLog.cntSkipCu[depth] += enclog.cntSkipCu[depth];
-                    }
+                if (depth == g_maxCUDepth)
+                    finalLog.cntIntraNxN += enclog.cntIntraNxN;
+                if (sliceType != I_SLICE)
+                {
+                    finalLog.cntTotalCu[depth] += enclog.cntTotalCu[depth];
+                    finalLog.cntInter[depth] += enclog.cntInter[depth];
+                    finalLog.cntSkipCu[depth] += enclog.cntSkipCu[depth];
                 }
             }
 
@@ -611,74 +714,44 @@
             int len = 0;
             if (sliceType != I_SLICE)
             {
-                len += sprintf(stats + len, "EncCU "LL "%% Merge "LL "%%", encCu, cntSkipCu);
+                len += sprintf(stats + len, " EncCU "X265_LL "%% Merge "X265_LL "%%", encCu, cntSkipCu);
             }
             if (cntInter)
             {
-                len += sprintf(stats + len, " Inter "LL "%%", cntInter);
-                if (param->bEnableAMP)
-                    len += sprintf(stats + len, "(%dx%d "LL "%% %dx%d "LL "%% %dx%d "LL "%% AMP "LL "%%)",
+                len += sprintf(stats + len, " Inter "X265_LL "%%", cntInter);
+                if (m_param->bEnableAMP)
+                    len += sprintf(stats + len, "(%dx%d "X265_LL "%% %dx%d "X265_LL "%% %dx%d "X265_LL "%% AMP "X265_LL "%%)",
                                    cuSize, cuSize, cuInterDistribution[0],
                                    cuSize / 2, cuSize, cuInterDistribution[2],
                                    cuSize, cuSize / 2, cuInterDistribution[1],
                                    cuInterDistribution[3]);
-                else if (param->bEnableRectInter)
-                    len += sprintf(stats + len, "(%dx%d "LL "%% %dx%d "LL "%% %dx%d "LL "%%)",
+                else if (m_param->bEnableRectInter)
+                    len += sprintf(stats + len, "(%dx%d "X265_LL "%% %dx%d "X265_LL "%% %dx%d "X265_LL "%%)",
                                    cuSize, cuSize, cuInterDistribution[0],
                                    cuSize / 2, cuSize, cuInterDistribution[2],
                                    cuSize, cuSize / 2, cuInterDistribution[1]);
             }
             if (cntIntra)
             {
-                len += sprintf(stats + len, " Intra "LL "%%(DC "LL "%% P "LL "%% Ang "LL "%%",
+                len += sprintf(stats + len, " Intra "X265_LL "%%(DC "X265_LL "%% P "X265_LL "%% Ang "X265_LL "%%",
                                cntIntra, cuIntraDistribution[0],
                                cuIntraDistribution[1], cuIntraDistribution[2]);
                 if (sliceType != I_SLICE)
                 {
-                    if (depth == (int)g_maxCUDepth - 1)
-                        len += sprintf(stats + len, " %dx%d "LL "%%", cuSize / 2, cuSize / 2, cntIntraNxN);
+                    if (depth == g_maxCUDepth)
+                        len += sprintf(stats + len, " %dx%d "X265_LL "%%", cuSize / 2, cuSize / 2, cntIntraNxN);
                 }
 
                 len += sprintf(stats + len, ")");
                 if (sliceType == I_SLICE)
                 {
-                    if (depth == (int)g_maxCUDepth - 1)
-                        len += sprintf(stats + len, " %dx%d: "LL "%%", cuSize / 2, cuSize / 2, cntIntraNxN);
+                    if (depth == g_maxCUDepth)
+                        len += sprintf(stats + len, " %dx%d: "X265_LL "%%", cuSize / 2, cuSize / 2, cntIntraNxN);
                 }
             }
             const char slicechars[] = "BPI";
             if (stats[0])
-                x265_log(param, X265_LOG_INFO, "%c%-2d: %s\n", slicechars[sliceType], cuSize, stats);
-        }
-    }
-
-#endif // if LOG_CU_STATISTICS
-    if (param->logLevel >= X265_LOG_INFO)
-    {
-        char buffer[200];
-        if (m_analyzeI.m_numPics)
-            x265_log(param, X265_LOG_INFO, "frame I: %s\n", statsString(m_analyzeI, buffer));
-        if (m_analyzeP.m_numPics)
-            x265_log(param, X265_LOG_INFO, "frame P: %s\n", statsString(m_analyzeP, buffer));
-        if (m_analyzeB.m_numPics)
-            x265_log(param, X265_LOG_INFO, "frame B: %s\n", statsString(m_analyzeB, buffer));
-        if (m_analyzeAll.m_numPics)
-            x265_log(param, X265_LOG_INFO, "global : %s\n", statsString(m_analyzeAll, buffer));
-        if (param->bEnableWeightedPred && m_analyzeP.m_numPics)
-        {
-            x265_log(param, X265_LOG_INFO, "Weighted P-Frames: Y:%.1f%% UV:%.1f%%\n",
-                     (float)100.0 * m_numLumaWPFrames / m_analyzeP.m_numPics,
-                     (float)100.0 * m_numChromaWPFrames / m_analyzeP.m_numPics);
-        }
-        int pWithB = 0;
-        for (int i = 0; i <= param->bframes; i++)
-            pWithB += m_lookahead->histogram[i];
-        if (pWithB)
-        {
-            int p = 0;
-            for (int i = 0; i <= param->bframes; i++)
-                p += sprintf(buffer + p, "%.1f%% ", 100. * m_lookahead->histogram[i] / pWithB);
-            x265_log(param, X265_LOG_INFO, "consecutive B-frames: %s\n", buffer);
+                x265_log(m_param, X265_LOG_INFO, "%c%-2d:%s\n", slicechars[sliceType], cuSize, stats);
         }
     }
 }
@@ -698,7 +771,7 @@
         {
             stats->globalSsim = m_analyzeAll.m_globalSsim / stats->encodedPictureCount;
             stats->globalPsnr = (stats->globalPsnrY * 6 + stats->globalPsnrU + stats->globalPsnrV) / (8 * stats->encodedPictureCount);
-            stats->elapsedVideoTime = (double)stats->encodedPictureCount * param->fpsDenom / param->fpsNum;
+            stats->elapsedVideoTime = (double)stats->encodedPictureCount * m_param->fpsDenom / m_param->fpsNum;
             stats->bitrate = (0.001f * stats->accBits) / stats->elapsedVideoTime;
         }
         else
@@ -717,12 +790,13 @@
 
 void Encoder::writeLog(int argc, char **argv)
 {
-    if (param->logLevel <= X265_LOG_DEBUG && m_csvfpt)
+    if (m_csvfpt)
     {
-        if (param->logLevel == X265_LOG_DEBUG)
+        if (m_param->logLevel >= X265_LOG_DEBUG)
         {
-            fprintf(m_csvfpt, "Summary\n");
-            fprintf(m_csvfpt, "Command, Date/Time, Elapsed Time, FPS, Bitrate, Y PSNR, U PSNR, V PSNR, Global PSNR, SSIM, SSIM (dB), Version\n");
+            // adding summary to a per-frame csv log file needs a summary header
+            fprintf(m_csvfpt, "\nSummary\n");
+            fputs(summaryCSVHeader, m_csvfpt);
         }
         // CLI arguments or other
         for (int i = 1; i < argc; i++)
@@ -736,7 +810,7 @@
         struct tm* timeinfo;
         time(&now);
         timeinfo = localtime(&now);
-        char buffer[128];
+        char buffer[200];
         strftime(buffer, 128, "%c", timeinfo);
         fprintf(m_csvfpt, ", %s, ", buffer);
 
@@ -747,17 +821,20 @@
         fprintf(m_csvfpt, "%.2f, %.2f, %.2f,",
                 stats.elapsedEncodeTime, stats.encodedPictureCount / stats.elapsedEncodeTime, stats.bitrate);
 
-        if (param->bEnablePsnr)
+        if (m_param->bEnablePsnr)
             fprintf(m_csvfpt, " %.3lf, %.3lf, %.3lf, %.3lf,",
                     stats.globalPsnrY / stats.encodedPictureCount, stats.globalPsnrU / stats.encodedPictureCount,
                     stats.globalPsnrV / stats.encodedPictureCount, stats.globalPsnr);
         else
             fprintf(m_csvfpt, " -, -, -, -,");
-        if (param->bEnableSsim)
+        if (m_param->bEnableSsim)
             fprintf(m_csvfpt, " %.6f, %6.3f,", stats.globalSsim, x265_ssim2dB(stats.globalSsim));
         else
             fprintf(m_csvfpt, " -, -,");
 
+        fputs(statsCSVString(m_analyzeI, buffer), m_csvfpt);
+        fputs(statsCSVString(m_analyzeP, buffer), m_csvfpt);
+        fputs(statsCSVString(m_analyzeB, buffer), m_csvfpt);
         fprintf(m_csvfpt, " %s\n", x265_version_str);
     }
 }
@@ -788,13 +865,13 @@
     return string;
 }
 
-void Encoder::finishFrameStats(TComPic* pic, FrameEncoder *curEncoder, uint64_t bits)
+void Encoder::finishFrameStats(Frame* pic, FrameEncoder *curEncoder, uint64_t bits)
 {
     TComPicYuv* recon = pic->getPicYuvRec();
 
     //===== calculate PSNR =====
-    int width  = recon->getWidth() - m_pad[0];
-    int height = recon->getHeight() - m_pad[1];
+    int width  = recon->getWidth() - m_sps.conformanceWindow.rightOffset;
+    int height = recon->getHeight() - m_sps.conformanceWindow.bottomOffset;
     int size = width * height;
 
     int maxvalY = 255 << (X265_DEPTH - 8);
@@ -803,72 +880,74 @@
     double refValueC = (double)maxvalC * maxvalC * size / 4.0;
     uint64_t ssdY, ssdU, ssdV;
 
-    ssdY = pic->m_SSDY;
-    ssdU = pic->m_SSDU;
-    ssdV = pic->m_SSDV;
+    ssdY = curEncoder->m_SSDY;
+    ssdU = curEncoder->m_SSDU;
+    ssdV = curEncoder->m_SSDV;
     double psnrY = (ssdY ? 10.0 * log10(refValueY / (double)ssdY) : 99.99);
     double psnrU = (ssdU ? 10.0 * log10(refValueC / (double)ssdU) : 99.99);
     double psnrV = (ssdV ? 10.0 * log10(refValueC / (double)ssdV) : 99.99);
 
-    TComSlice*  slice = pic->getSlice();
+    Slice*  slice = pic->m_picSym->m_slice;
 
     //===== add bits, psnr and ssim =====
     m_analyzeAll.addBits(bits);
     m_analyzeAll.addQP(pic->m_avgQpAq);
 
-    if (param->bEnablePsnr)
+    if (m_param->bEnablePsnr)
     {
         m_analyzeAll.addPsnr(psnrY, psnrU, psnrV);
     }
 
     double ssim = 0.0;
-    if (param->bEnableSsim && pic->m_ssimCnt > 0)
+    if (m_param->bEnableSsim && curEncoder->m_ssimCnt)
     {
-        ssim = pic->m_ssim / pic->m_ssimCnt;
+        ssim = curEncoder->m_ssim / curEncoder->m_ssimCnt;
         m_analyzeAll.addSsim(ssim);
     }
     if (slice->isIntra())
     {
         m_analyzeI.addBits(bits);
         m_analyzeI.addQP(pic->m_avgQpAq);
-        if (param->bEnablePsnr)
+        if (m_param->bEnablePsnr)
             m_analyzeI.addPsnr(psnrY, psnrU, psnrV);
-        if (param->bEnableSsim)
+        if (m_param->bEnableSsim)
             m_analyzeI.addSsim(ssim);
     }
     else if (slice->isInterP())
     {
         m_analyzeP.addBits(bits);
         m_analyzeP.addQP(pic->m_avgQpAq);
-        if (param->bEnablePsnr)
+        if (m_param->bEnablePsnr)
             m_analyzeP.addPsnr(psnrY, psnrU, psnrV);
-        if (param->bEnableSsim)
+        if (m_param->bEnableSsim)
             m_analyzeP.addSsim(ssim);
     }
     else if (slice->isInterB())
     {
         m_analyzeB.addBits(bits);
         m_analyzeB.addQP(pic->m_avgQpAq);
-        if (param->bEnablePsnr)
+        if (m_param->bEnablePsnr)
             m_analyzeB.addPsnr(psnrY, psnrU, psnrV);
-        if (param->bEnableSsim)
+        if (m_param->bEnableSsim)
             m_analyzeB.addSsim(ssim);
     }
 
     // if debug log level is enabled, per frame logging is performed
-    if (param->logLevel >= X265_LOG_DEBUG)
+    if (m_param->logLevel >= X265_LOG_DEBUG)
     {
         char c = (slice->isIntra() ? 'I' : slice->isInterP() ? 'P' : 'B');
-        int poc = slice->getPOC();
-        if (!slice->isReferenced())
+        int poc = slice->m_poc;
+        if (!IS_REFERENCED(slice))
             c += 32; // lower case if unreferenced
 
         char buf[1024];
         int p;
-        p = sprintf(buf, "POC:%d %c QP %2.2lf(%d) %10d bits", poc, c, pic->m_avgQpAq, slice->getSliceQp(), (int)bits);
-        if (param->bEnablePsnr)
+        p = sprintf(buf, "POC:%d %c QP %2.2lf(%d) %10d bits", poc, c, pic->m_avgQpAq, slice->m_sliceQp, (int)bits);
+        if (m_param->rc.rateControlMode == X265_RC_CRF)
+            p += sprintf(buf + p, " RF:%.3lf", pic->m_rateFactor);
+        if (m_param->bEnablePsnr)
             p += sprintf(buf + p, " [Y:%6.2lf U:%6.2lf V:%6.2lf]", psnrY, psnrU, psnrV);
-        if (param->bEnableSsim)
+        if (m_param->bEnableSsim)
             p += sprintf(buf + p, " [SSIM: %.3lfdB]", x265_ssim2dB(ssim));
 
         if (!slice->isIntra())
@@ -877,9 +956,9 @@
             for (int list = 0; list < numLists; list++)
             {
                 p += sprintf(buf + p, " [L%d ", list);
-                for (int ref = 0; ref < slice->getNumRefIdx(list); ref++)
+                for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
                 {
-                    int k = slice->getRefPOC(list, ref) - slice->getLastIDR();
+                    int k = slice->m_refPOCList[list][ref] - slice->m_lastIDR;
                     p += sprintf(buf + p, "%d ", k);
                 }
 
@@ -891,25 +970,27 @@
         if (m_csvfpt)
         {
             fprintf(m_csvfpt, "%d, %c-SLICE, %4d, %2.2lf, %10d,", m_outputCount++, c, poc, pic->m_avgQpAq, (int)bits);
+            if (m_param->rc.rateControlMode == X265_RC_CRF)
+                fprintf(m_csvfpt, "%.3lf,", pic->m_rateFactor);
             double psnr = (psnrY * 6 + psnrU + psnrV) / 8;
-            if (param->bEnablePsnr)
+            if (m_param->bEnablePsnr)
                 fprintf(m_csvfpt, "%.3lf, %.3lf, %.3lf, %.3lf,", psnrY, psnrU, psnrV, psnr);
             else
                 fprintf(m_csvfpt, " -, -, -, -,");
-            if (param->bEnableSsim)
+            if (m_param->bEnableSsim)
                 fprintf(m_csvfpt, " %.6f, %6.3f,", ssim, x265_ssim2dB(ssim));
             else
                 fprintf(m_csvfpt, " -, -,");
-            fprintf(m_csvfpt, " %.3lf, %.3lf", pic->m_frameTime, pic->m_elapsedCompressTime);
+            fprintf(m_csvfpt, " %.3lf, %.3lf", curEncoder->m_frameTime, curEncoder->m_elapsedCompressTime);
             if (!slice->isIntra())
             {
                 int numLists = slice->isInterP() ? 1 : 2;
                 for (int list = 0; list < numLists; list++)
                 {
                     fprintf(m_csvfpt, ", ");
-                    for (int ref = 0; ref < slice->getNumRefIdx(list); ref++)
+                    for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
                     {
-                        int k = slice->getRefPOC(list, ref) - slice->getLastIDR();
+                        int k = slice->m_refPOCList[list][ref] - slice->m_lastIDR;
                         fprintf(m_csvfpt, " %d", k);
                     }
                 }
@@ -922,26 +1003,26 @@
             fprintf(m_csvfpt, "\n");
         }
 
-        if (param->decodedPictureHashSEI && param->logLevel >= X265_LOG_FULL)
+        if (m_param->decodedPictureHashSEI && m_param->logLevel >= X265_LOG_FULL)
         {
             const char* digestStr = NULL;
-            if (param->decodedPictureHashSEI == 1)
+            if (m_param->decodedPictureHashSEI == 1)
             {
-                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.digest, 16);
+                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.m_digest, 16);
                 p += sprintf(buf + p, " [MD5:%s]", digestStr);
             }
-            else if (param->decodedPictureHashSEI == 2)
+            else if (m_param->decodedPictureHashSEI == 2)
             {
-                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.digest, 2);
+                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.m_digest, 2);
                 p += sprintf(buf + p, " [CRC:%s]", digestStr);
             }
-            else if (param->decodedPictureHashSEI == 3)
+            else if (m_param->decodedPictureHashSEI == 3)
             {
-                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.digest, 4);
+                digestStr = digestToString(curEncoder->m_seiReconPictureDigest.m_digest, 4);
                 p += sprintf(buf + p, " [Checksum:%s]", digestStr);
             }
         }
-        x265_log(param, X265_LOG_DEBUG, "%s\n", buf);
+        x265_log(m_param, X265_LOG_DEBUG, "%s\n", buf);
         fflush(stderr);
     }
 }
@@ -951,349 +1032,196 @@
 #pragma warning(disable: 4127) // conditional expression is constant
 #endif
 
-void Encoder::initSPS(TComSPS *sps)
+void Encoder::getStreamHeaders(NALList& list, Entropy& sbacCoder, Bitstream& bs)
 {
-    ProfileTierLevel& profileTierLevel = *sps->getPTL()->getGeneralPTL();
-
-    profileTierLevel.setLevelIdc(m_level);
-    profileTierLevel.setTierFlag(m_levelTier ? true : false);
-    profileTierLevel.setProfileIdc(m_profile);
-    profileTierLevel.setProfileCompatibilityFlag(m_profile, 1);
-    profileTierLevel.setProgressiveSourceFlag(!param->interlaceMode);
-    profileTierLevel.setInterlacedSourceFlag(!!param->interlaceMode);
-    profileTierLevel.setNonPackedConstraintFlag(m_nonPackedConstraintFlag);
-    profileTierLevel.setFrameOnlyConstraintFlag(m_frameOnlyConstraintFlag);
-
-    if (m_profile == Profile::MAIN10 && X265_DEPTH == 8)
-    {
-        /* The above constraint is equal to Profile::MAIN */
-        profileTierLevel.setProfileCompatibilityFlag(Profile::MAIN, 1);
-    }
-    if (m_profile == Profile::MAIN)
-    {
-        /* A Profile::MAIN10 decoder can always decode Profile::MAIN */
-        profileTierLevel.setProfileCompatibilityFlag(Profile::MAIN10, 1);
-    }
-    /* XXX: should Main be marked as compatible with still picture? */
-
-    /* XXX: may be a good idea to refactor the above into a function
-     * that chooses the actual compatibility based upon options */
-
-    sps->setPicWidthInLumaSamples(param->sourceWidth);
-    sps->setPicHeightInLumaSamples(param->sourceHeight);
-    sps->setConformanceWindow(m_conformanceWindow);
-    sps->setChromaFormatIdc(param->internalCsp);
-    sps->setMaxCUSize(g_maxCUSize);
-    sps->setMaxCUDepth(g_maxCUDepth);
-
-    int minCUSize = sps->getMaxCUSize() >> (sps->getMaxCUDepth() - g_addCUDepth);
-    int log2MinCUSize = 0;
-    while (minCUSize > 1)
-    {
-        minCUSize >>= 1;
-        log2MinCUSize++;
-    }
-
-    sps->setLog2MinCodingBlockSize(log2MinCUSize);
-    sps->setLog2DiffMaxMinCodingBlockSize(sps->getMaxCUDepth() - g_addCUDepth);
+    sbacCoder.setBitstream(&bs);
 
-    sps->setPCMLog2MinSize(m_pcmLog2MinSize);
-    sps->setUsePCM(m_usePCM);
-    sps->setPCMLog2MaxSize(m_pcmLog2MaxSize);
+    /* headers for start of bitstream */
+    bs.resetBits();
+    sbacCoder.codeVPS(&m_vps);
+    bs.writeByteAlignment();
+    list.serialize(NAL_UNIT_VPS, bs);
 
-    sps->setQuadtreeTULog2MaxSize(m_quadtreeTULog2MaxSize);
-    sps->setQuadtreeTULog2MinSize(m_quadtreeTULog2MinSize);
-    sps->setQuadtreeTUMaxDepthInter(param->tuQTMaxInterDepth);
-    sps->setQuadtreeTUMaxDepthIntra(param->tuQTMaxIntraDepth);
+    bs.resetBits();
+    sbacCoder.codeSPS(&m_sps, &m_scalingList, &m_vps.ptl);
+    bs.writeByteAlignment();
+    list.serialize(NAL_UNIT_SPS, bs);
 
-    sps->setTMVPFlagsPresent(false);
-    sps->setUseLossless(m_useLossless);
+    bs.resetBits();
+    sbacCoder.codePPS(&m_pps);
+    bs.writeByteAlignment();
+    list.serialize(NAL_UNIT_PPS, bs);
 
-    sps->setMaxTrSize(1 << m_quadtreeTULog2MaxSize);
-
-    for (uint32_t i = 0; i < g_maxCUDepth - g_addCUDepth; i++)
+    if (m_param->bEmitInfoSEI)
     {
-        sps->setAMPAcc(i, param->bEnableAMP);
-    }
+        char *opts = x265_param2string(m_param);
+        if (opts)
+        {
+            char *buffer = X265_MALLOC(char, strlen(opts) + strlen(x265_version_str) +
+                                             strlen(x265_build_info_str) + 200);
+            if (buffer)
+            {
+                sprintf(buffer, "x265 (build %d) - %s:%s - H.265/HEVC codec - "
+                        "Copyright 2013-2014 (c) Multicoreware Inc - "
+                        "http://x265.org - options: %s",
+                        X265_BUILD, x265_version_str, x265_build_info_str, opts);
+                
+                bs.resetBits();
+                SEIuserDataUnregistered idsei;
+                idsei.m_userData = (uint8_t*)buffer;
+                idsei.m_userDataLength = (uint32_t)strlen(buffer);
+                idsei.write(bs, m_sps);
+                bs.writeByteAlignment();
+                list.serialize(NAL_UNIT_PREFIX_SEI, bs);
+
+                X265_FREE(buffer);
+            }
 
-    sps->setUseAMP(param->bEnableAMP);
+            X265_FREE(opts);
+        }
+    }
 
-    for (uint32_t i = g_maxCUDepth - g_addCUDepth; i < g_maxCUDepth; i++)
+    if (m_param->bEmitHRDSEI)
     {
-        sps->setAMPAcc(i, 0);
+        /* Picture Timing and Buffering Period SEI require the SPS to be "activated" */
+        SEIActiveParameterSets sei;
+        sei.m_selfContainedCvsFlag = true;
+        sei.m_noParamSetUpdateFlag = true;
+
+        bs.resetBits();
+        sei.write(bs, m_sps);
+        bs.writeByteAlignment();
+        list.serialize(NAL_UNIT_PREFIX_SEI, bs);
     }
+}
 
-    sps->setBitDepthY(X265_DEPTH);
-    sps->setBitDepthC(X265_DEPTH);
+void Encoder::initSPS(SPS *sps)
+{
+    m_vps.ptl.progressiveSourceFlag = !m_param->interlaceMode;
+    m_vps.ptl.interlacedSourceFlag = !!m_param->interlaceMode;
+    m_vps.ptl.nonPackedConstraintFlag = false;
+    m_vps.ptl.frameOnlyConstraintFlag = false;
 
-    sps->setQpBDOffsetY(QP_BD_OFFSET);
-    sps->setQpBDOffsetC(QP_BD_OFFSET);
+    sps->conformanceWindow = m_conformanceWindow;
+    sps->chromaFormatIdc = m_param->internalCsp;
+    sps->picWidthInLumaSamples = m_param->sourceWidth;
+    sps->picHeightInLumaSamples = m_param->sourceHeight;
 
-    sps->setUseSAO(param->bEnableSAO);
+    sps->log2MinCodingBlockSize = g_maxLog2CUSize - g_maxCUDepth;
+    sps->log2DiffMaxMinCodingBlockSize = g_maxCUDepth;
 
-    // TODO: hard-code these values in SPS code
-    sps->setMaxTLayers(1);
-    sps->setTemporalIdNestingFlag(true);
-    for (uint32_t i = 0; i < sps->getMaxTLayers(); i++)
-    {
-        sps->setMaxDecPicBuffering(m_maxDecPicBuffering[i], i);
-        sps->setNumReorderPics(m_numReorderPics[i], i);
-    }
+    sps->quadtreeTULog2MaxSize = g_log2Size[m_param->maxCUSize] - 1;
+    sps->quadtreeTULog2MinSize = 2;
+    sps->quadtreeTUMaxDepthInter = m_param->tuQTMaxInterDepth;
+    sps->quadtreeTUMaxDepthIntra = m_param->tuQTMaxIntraDepth;
 
-    // TODO: it is recommended for this to match the input bit depth
-    sps->setPCMBitDepthLuma(X265_DEPTH);
-    sps->setPCMBitDepthChroma(X265_DEPTH);
+    sps->bUseSAO = m_param->bEnableSAO;
 
-    sps->setPCMFilterDisableFlag(m_bPCMFilterDisableFlag);
+    sps->bUseAMP = m_param->bEnableAMP;
+    sps->maxAMPDepth = m_param->bEnableAMP ? g_maxCUDepth : 0;
 
-    sps->setScalingListFlag((m_useScalingListId == 0) ? 0 : 1);
+    sps->maxDecPicBuffering = m_vps.maxDecPicBuffering;
+    sps->numReorderPics = m_vps.numReorderPics;
 
-    sps->setUseStrongIntraSmoothing(param->bEnableStrongIntraSmoothing);
+    sps->bUseStrongIntraSmoothing = m_param->bEnableStrongIntraSmoothing;
 
-    sps->setVuiParametersPresentFlag(param->vui.bEnableVuiParametersPresentFlag);
-    if (sps->getVuiParametersPresentFlag())
-    {
-        TComVUI* vui = sps->getVuiParameters();
-        vui->setAspectRatioInfoPresentFlag(param->vui.bEnableAspectRatioIdc);
-        vui->setAspectRatioIdc(param->vui.aspectRatioIdc);
-        vui->setSarWidth(param->vui.sarWidth);
-        vui->setSarHeight(param->vui.sarHeight);
-        vui->setOverscanInfoPresentFlag(param->vui.bEnableOverscanInfoPresentFlag);
-        vui->setOverscanAppropriateFlag(param->vui.bEnableOverscanAppropriateFlag);
-        vui->setVideoSignalTypePresentFlag(param->vui.bEnableVideoSignalTypePresentFlag);
-        vui->setVideoFormat(param->vui.videoFormat);
-        vui->setVideoFullRangeFlag(param->vui.bEnableVideoFullRangeFlag);
-        vui->setColourDescriptionPresentFlag(param->vui.bEnableColorDescriptionPresentFlag);
-        vui->setColourPrimaries(param->vui.colorPrimaries);
-        vui->setTransferCharacteristics(param->vui.transferCharacteristics);
-        vui->setMatrixCoefficients(param->vui.matrixCoeffs);
-        vui->setChromaLocInfoPresentFlag(param->vui.bEnableChromaLocInfoPresentFlag);
-        vui->setChromaSampleLocTypeTopField(param->vui.chromaSampleLocTypeTopField);
-        vui->setChromaSampleLocTypeBottomField(param->vui.chromaSampleLocTypeBottomField);
-        vui->setNeutralChromaIndicationFlag(m_neutralChromaIndicationFlag);
-        vui->setDefaultDisplayWindow(m_defaultDisplayWindow);
-        vui->setFrameFieldInfoPresentFlag(!!param->interlaceMode);
-        vui->setFieldSeqFlag(!!param->interlaceMode);
-        vui->setHrdParametersPresentFlag(param->vui.bEnableVuiHrdParametersPresentFlag);
-        vui->getHrdParameters()->setNalHrdParametersPresentFlag(param->vui.bEnableNalHrdParametersPresentFlag);
-        vui->getHrdParameters()->setSubPicHrdParamsPresentFlag(param->vui.bEnableSubPicHrdParamsPresentFlag);
-        vui->getTimingInfo()->setTimingInfoPresentFlag(param->vui.bEnableVuiTimingInfoPresentFlag);
-        vui->getTimingInfo()->setNumUnitsInTick(param->fpsDenom);
-        vui->getTimingInfo()->setTimeScale(param->fpsNum);
-        vui->getTimingInfo()->setPocProportionalToTimingFlag(m_pocProportionalToTimingFlag);
-        vui->getTimingInfo()->setNumTicksPocDiffOneMinus1(m_numTicksPocDiffOneMinus1);
-        vui->setBitstreamRestrictionFlag(param->vui.bEnableBitstreamRestrictionFlag);
-        vui->setTilesFixedStructureFlag(m_tilesFixedStructureFlag);
-        vui->setMotionVectorsOverPicBoundariesFlag(m_motionVectorsOverPicBoundariesFlag);
-        vui->setRestrictedRefPicListsFlag(m_restrictedRefPicListsFlag);
-        vui->setMinSpatialSegmentationIdc(m_minSpatialSegmentationIdc);
-        vui->setMaxBytesPerPicDenom(m_maxBytesPerPicDenom);
-        vui->setMaxBitsPerMinCuDenom(m_maxBitsPerMinCuDenom);
-        vui->setLog2MaxMvLengthHorizontal(m_log2MaxMvLengthHorizontal);
-        vui->setLog2MaxMvLengthVertical(m_log2MaxMvLengthVertical);
-    }
+    VUI& vui = sps->vuiParameters;
+    vui.aspectRatioInfoPresentFlag = !!m_param->vui.aspectRatioIdc;
+    vui.aspectRatioIdc = m_param->vui.aspectRatioIdc;
+    vui.sarWidth = m_param->vui.sarWidth;
+    vui.sarHeight = m_param->vui.sarHeight;
 
-    /* set the VPS profile information */
-    *m_vps.getPTL() = *sps->getPTL();
-    TimingInfo *t = m_vps.getTimingInfo();
-    t->setTimingInfoPresentFlag(true);
-    t->setNumUnitsInTick(param->fpsDenom);
-    t->setTimeScale(param->fpsNum);
-}
+    vui.overscanInfoPresentFlag = m_param->vui.bEnableOverscanInfoPresentFlag;
+    vui.overscanAppropriateFlag = m_param->vui.bEnableOverscanAppropriateFlag;
 
-void Encoder::initPPS(TComPPS *pps)
-{
-    pps->setConstrainedIntraPred(param->bEnableConstrainedIntra);
-    bool isVbv = param->rc.vbvBufferSize > 0 && param->rc.vbvMaxBitrate > 0;
+    vui.videoSignalTypePresentFlag = m_param->vui.bEnableVideoSignalTypePresentFlag;
+    vui.videoFormat = m_param->vui.videoFormat;
+    vui.videoFullRangeFlag = m_param->vui.bEnableVideoFullRangeFlag;
 
-    /* TODO: This variable m_maxCuDQPDepth needs to be a CLI option to allow us to choose AQ granularity */
-    bool bUseDQP = (m_maxCuDQPDepth > 0 || param->rc.aqMode || isVbv) ? true : false;
+    vui.colourDescriptionPresentFlag = m_param->vui.bEnableColorDescriptionPresentFlag;
+    vui.colourPrimaries = m_param->vui.colorPrimaries;
+    vui.transferCharacteristics = m_param->vui.transferCharacteristics;
+    vui.matrixCoefficients = m_param->vui.matrixCoeffs;
 
-    int lowestQP = -QP_BD_OFFSET; 
+    vui.chromaLocInfoPresentFlag = m_param->vui.bEnableChromaLocInfoPresentFlag;
+    vui.chromaSampleLocTypeTopField = m_param->vui.chromaSampleLocTypeTopField;
+    vui.chromaSampleLocTypeBottomField = m_param->vui.chromaSampleLocTypeBottomField;
 
-    if (m_useLossless)
-    {
-        if ((m_maxCuDQPDepth == 0) && (param->rc.qp == lowestQP))
-        {
-            bUseDQP = false;
-        }
-        else
-        {
-            bUseDQP = true;
-        }
-    }
+    vui.defaultDisplayWindow.bEnabled = m_param->vui.bEnableDefaultDisplayWindowFlag;
+    vui.defaultDisplayWindow.rightOffset = m_param->vui.defDispWinRightOffset;
+    vui.defaultDisplayWindow.topOffset = m_param->vui.defDispWinTopOffset;
+    vui.defaultDisplayWindow.bottomOffset = m_param->vui.defDispWinBottomOffset;
+    vui.defaultDisplayWindow.leftOffset = m_param->vui.defDispWinLeftOffset;
 
-    if (bUseDQP)
-    {
-        pps->setUseDQP(true);
-        pps->setMaxCuDQPDepth(m_maxCuDQPDepth);
-        pps->setMinCuDQPSize(pps->getSPS()->getMaxCUSize() >> (pps->getMaxCuDQPDepth()));
-    }
-    else
-    {
-        pps->setUseDQP(false);
-        pps->setMaxCuDQPDepth(0);
-        pps->setMinCuDQPSize(pps->getSPS()->getMaxCUSize() >> (pps->getMaxCuDQPDepth()));
-    }
+    vui.frameFieldInfoPresentFlag = !!m_param->interlaceMode;
+    vui.fieldSeqFlag = !!m_param->interlaceMode;
 
-    pps->setChromaCbQpOffset(param->cbQpOffset);
-    pps->setChromaCrQpOffset(param->crQpOffset);
-
-    pps->setEntropyCodingSyncEnabledFlag(param->bEnableWavefront);
-    pps->setUseWP(param->bEnableWeightedPred);
-    pps->setWPBiPred(param->bEnableWeightedBiPred);
-    pps->setOutputFlagPresentFlag(false);
-    pps->setSignHideFlag(param->bEnableSignHiding);
-    pps->setDeblockingFilterControlPresentFlag(!param->bEnableLoopFilter);
-    pps->setDeblockingFilterOverrideEnabledFlag(!m_loopFilterOffsetInPPS);
-    pps->setPicDisableDeblockingFilterFlag(!param->bEnableLoopFilter);
-    pps->setLog2ParallelMergeLevelMinus2(m_log2ParallelMergeLevelMinus2);
-    pps->setCabacInitPresentFlag(param->frameNumThreads > 1 ? 0 : CABAC_INIT_PRESENT_FLAG);
-
-    pps->setNumRefIdxL0DefaultActive(1);
-    pps->setNumRefIdxL1DefaultActive(1);
-
-    pps->setTransquantBypassEnableFlag(m_TransquantBypassEnableFlag);
-    pps->setUseTransformSkip(param->bEnableTransformSkip);
-    pps->setLoopFilterAcrossTilesEnabledFlag(m_loopFilterAcrossTilesEnabledFlag);
+    vui.hrdParametersPresentFlag = m_param->bEmitHRDSEI;
+
+    vui.timingInfo.numUnitsInTick = m_param->fpsDenom;
+    vui.timingInfo.timeScale = m_param->fpsNum;
 }
 
-void Encoder::determineLevelAndProfile(x265_param *_param)
+void Encoder::initPPS(PPS *pps)
 {
-    // this is all based on the table at on Wikipedia at
-    // http://en.wikipedia.org/wiki/High_Efficiency_Video_Coding#Profiles
-
-    // TODO: there are minimum CTU sizes for higher levels, needs to be enforced
+    bool bIsVbv = m_param->rc.vbvBufferSize > 0 && m_param->rc.vbvMaxBitrate > 0;
 
-    uint32_t lumaSamples = _param->sourceWidth * _param->sourceHeight;
-    uint32_t samplesPerSec = (uint32_t)(lumaSamples * ((double)_param->fpsNum / _param->fpsDenom));
-    uint32_t bitrate = _param->rc.bitrate;
-
-    m_level = Level::LEVEL1;
-    const char *level = "1";
-    if (samplesPerSec > 552960 || lumaSamples > 36864 || bitrate > 128)
-    {
-        m_level = Level::LEVEL2;
-        level = "2";
-    }
-    if (samplesPerSec > 3686400 || lumaSamples > 122880 || bitrate > 1500)
-    {
-        m_level = Level::LEVEL2_1;
-        level = "2.1";
-    }
-    if (samplesPerSec > 7372800 || lumaSamples > 245760 || bitrate > 3000)
-    {
-        m_level = Level::LEVEL3;
-        level = "3";
-    }
-    if (samplesPerSec > 16588800 || lumaSamples > 552960 || bitrate > 6000)
-    {
-        m_level = Level::LEVEL3_1;
-        level = "3.1";
-    }
-    if (samplesPerSec > 33177600 || lumaSamples > 983040 || bitrate > 10000)
-    {
-        m_level = Level::LEVEL4;
-        level = "4";
-    }
-    if (samplesPerSec > 66846720 || bitrate > 30000)
-    {
-        m_level = Level::LEVEL4_1;
-        level = "4.1";
-    }
-    if (samplesPerSec > 133693440 || lumaSamples > 2228224 || bitrate > 50000)
-    {
-        m_level = Level::LEVEL5;
-        level = "5";
-    }
-    if (samplesPerSec > 267386880 || bitrate > 100000)
-    {
-        m_level = Level::LEVEL5_1;
-        level = "5.1";
-    }
-    if (samplesPerSec > 534773760 || bitrate > 160000)
-    {
-        m_level = Level::LEVEL5_2;
-        level = "5.2";
-    }
-    if (samplesPerSec > 1069547520 || lumaSamples > 8912896 || bitrate > 240000)
-    {
-        m_level = Level::LEVEL6;
-        level = "6";
-    }
-    if (samplesPerSec > 1069547520 || bitrate > 240000)
+    if (!m_param->bLossless && (m_param->rc.aqMode || bIsVbv))
     {
-        m_level = Level::LEVEL6_1;
-        level = "6.1";
+        pps->bUseDQP = true;
+        pps->maxCuDQPDepth = 0; /* TODO: make configurable? */
     }
-    if (samplesPerSec > 2139095040 || bitrate > 480000)
+    else
     {
-        m_level = Level::LEVEL6_2;
-        level = "6.2";
+        pps->bUseDQP = false;
+        pps->maxCuDQPDepth = 0;
     }
-    if (samplesPerSec > 4278190080U || lumaSamples > 35651584 || bitrate > 800000)
-        x265_log(_param, X265_LOG_WARNING, "video size or bitrate out of scope for HEVC\n");
 
-    /* Within a given level, we might be at a high tier, depending on bitrate */
-    m_levelTier = Level::MAIN;
-    switch (m_level)
-    {
-    case Level::LEVEL4:
-        if (bitrate > 12000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL4_1:
-        if (bitrate > 20000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL5:
-        if (bitrate > 25000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL5_1:
-        if (bitrate > 40000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL5_2:
-        if (bitrate > 60000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL6:
-        if (bitrate > 60000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL6_1:
-        if (bitrate > 120000) m_levelTier = Level::HIGH;
-        break;
-    case Level::LEVEL6_2:
-        if (bitrate > 240000) m_levelTier = Level::HIGH;
-        break;
-    default:
-        break;
-    }
+    pps->chromaCbQpOffset = m_param->cbQpOffset;
+    pps->chromaCrQpOffset = m_param->crQpOffset;
 
-    if (_param->internalBitDepth > 8)
-        m_profile = Profile::MAIN10;
-    else if (_param->keyframeMax == 1)
-        m_profile = Profile::MAINSTILLPICTURE;
-    else
-        m_profile = Profile::MAIN;
+    pps->bConstrainedIntraPred = m_param->bEnableConstrainedIntra;
+    pps->bUseWeightPred = m_param->bEnableWeightedPred;
+    pps->bUseWeightedBiPred = m_param->bEnableWeightedBiPred;
+    pps->bTransquantBypassEnabled = m_param->bCULossless || m_param->bLossless;
+    pps->bTransformSkipEnabled = m_param->bEnableTransformSkip;
+    pps->bSignHideEnabled = m_param->bEnableSignHiding;
+
+    /* If offsets are ever configured, enable bDeblockingFilterControlPresent and set
+     * deblockingFilterBetaOffsetDiv2 / deblockingFilterTcOffsetDiv2 */
+    bool bDeblockOffsetInPPS = 0;
+    pps->bDeblockingFilterControlPresent = !m_param->bEnableLoopFilter || bDeblockOffsetInPPS;
+    pps->bPicDisableDeblockingFilter = !m_param->bEnableLoopFilter;
+    pps->deblockingFilterBetaOffsetDiv2 = 0;
+    pps->deblockingFilterTcOffsetDiv2 = 0;
 
-    static const char *profiles[] = { "None", "Main", "Main10", "Mainstillpicture" };
-    static const char *tiers[]    = { "Main", "High" };
-    x265_log(_param, X265_LOG_INFO, "%s profile, Level-%s (%s tier)\n", profiles[m_profile], level, tiers[m_levelTier]);
+    pps->bEntropyCodingSyncEnabled = m_param->bEnableWavefront;
 }
 
 void Encoder::configure(x265_param *p)
 {
+    this->m_param = p;
+
+    uint32_t maxLog2CUSize = g_log2Size[p->maxCUSize] - 1;
+    int rows = (p->sourceHeight + p->maxCUSize - 1) >> maxLog2CUSize;
+
+    // Do not allow WPP if only one row, it is pointless and unstable
+    if (rows == 1)
+        p->bEnableWavefront = 0;
+
     // Trim the thread pool if WPP is disabled
     if (!p->bEnableWavefront)
         p->poolNumThreads = 1;
 
     setThreadPool(ThreadPool::allocThreadPool(p->poolNumThreads));
     int poolThreadCount = ThreadPool::getThreadPool()->getThreadCount();
-    int rows = (p->sourceHeight + p->maxCUSize - 1) / p->maxCUSize;
 
-    if (p->frameNumThreads == 0)
+    if (!p->frameNumThreads)
     {
         // auto-detect frame threads
         int cpuCount = getCpuCount();
         if (poolThreadCount <= 1)
-            p->frameNumThreads = X265_MIN(cpuCount, rows / 2);
+            p->frameNumThreads = X265_MIN(cpuCount, (rows + 1) / 2);
         else if (cpuCount > 32)
             p->frameNumThreads = 6; // dual-socket 10-core IvyBridge or higher
         else if (cpuCount >= 16)
@@ -1323,6 +1251,7 @@
     {
         x265_log(p, X265_LOG_INFO, "Warning: picture-based SAO used with frame parallelism\n");
     }
+
     if (p->keyframeMax < 0)
     {
         /* A negative max GOP size indicates the user wants only one I frame at
@@ -1343,43 +1272,34 @@
         p->keyframeMin = X265_MIN((int)fps, p->keyframeMax / 10);
     }
     p->keyframeMin = X265_MAX(1, X265_MIN(p->keyframeMin, p->keyframeMax / 2 + 1));
+
     if (!p->bEnableRectInter)
-    {
         p->bEnableAMP = false;
-    }
+
     if (p->bBPyramid && !p->bframes)
-    {
         p->bBPyramid = 0;
-    }
-    /* Set flags according to RDLevel specified - check_params has verified that RDLevel is within range */
-    switch (p->rdLevel)
+
+    // psy-rd is not supported in RD levels below 2
+    if (p->rdLevel < 2)
+        p->psyRd = 0.0;
+
+    /* In 444, chroma gets twice as much resolution, so halve quality when psy-rd is enabled */
+    if (p->internalCsp == X265_CSP_I444 && p->psyRd)
     {
-    case 6:
-        bEnableRDOQ = bEnableRDOQTS = 1;
-        break;
-    case 5:
-        bEnableRDOQ = bEnableRDOQTS = 1;
-        break;
-    case 4:
-        bEnableRDOQ = bEnableRDOQTS = 1;
-        break;
-    case 3:
-        bEnableRDOQ = bEnableRDOQTS = 0;
-        break;
-    case 2:
-        bEnableRDOQ = bEnableRDOQTS = 0;
-        break;
-    case 1:
-        bEnableRDOQ = bEnableRDOQTS = 0;
-        break;
-    case 0:
-        bEnableRDOQ = bEnableRDOQTS = 0;
-        break;
+        p->cbQpOffset += 6;
+        p->crQpOffset += 6;
     }
 
-    if (!(bEnableRDOQ && p->bEnableTransformSkip))
+    /* disable psy-rdoq if RDOQ is not enabled (do not show in logs as in-use) */
+    if (p->rdLevel < 4)
+        p->psyRdoq = 0;
+
+    if (p->bLossless)
     {
-        bEnableRDOQTS = 0;
+        p->rc.rateControlMode = X265_RC_CQP;
+        p->rc.qp = 4; // An oddity, QP=4 is more lossless than QP=0 and gives better lambdas
+        p->bEnableSsim = 0;
+        p->bEnablePsnr = 0;
     }
 
     if (p->rc.rateControlMode == X265_RC_CQP)
@@ -1387,6 +1307,7 @@
         p->rc.aqMode = X265_AQ_NONE;
         p->rc.bitrate = 0;
         p->rc.cuTree = 0;
+        p->rc.aqStrength = 0;
     }
 
     if (p->rc.aqMode == 0 && p->rc.cuTree)
@@ -1395,192 +1316,102 @@
         p->rc.aqStrength = 0.0;
     }
 
-    if (p->lookaheadDepth == 0 && p->rc.cuTree)
+    if (p->lookaheadDepth == 0 && p->rc.cuTree && !p->rc.bStatRead)
     {
         x265_log(p, X265_LOG_WARNING, "cuTree disabled, requires lookahead to be enabled\n");
         p->rc.cuTree = 0;
     }
 
     if (p->rc.aqStrength == 0 && p->rc.cuTree == 0)
-    {
         p->rc.aqMode = X265_AQ_NONE;
-    }
+
+    if (p->rc.aqMode == X265_AQ_NONE && p->rc.cuTree == 0)
+        p->rc.aqStrength = 0;
 
     if (p->internalCsp != X265_CSP_I420)
     {
         x265_log(p, X265_LOG_WARNING, "!! HEVC Range Extension specifications are not finalized !!\n");
         x265_log(p, X265_LOG_WARNING, "!! This output bitstream may not be compliant with the final spec !!\n");
     }
-    if (p->internalCsp == X265_CSP_I444 && p->bEnableWeightedPred)
+
+    if (p->scalingLists && p->internalCsp == X265_CSP_I444)
     {
-        x265_log(p, X265_LOG_WARNING, "Weightp not supported for 4:4:4 internal color space, weightp disabled\n");
-        p->bEnableWeightedPred = false;
+        x265_log(p, X265_LOG_WARNING, "Scaling lists are not yet supported for 4:4:4 color space\n");
+        p->scalingLists = 0;
     }
+
     if (p->interlaceMode)
-    {
         x265_log(p, X265_LOG_WARNING, "Support for interlaced video is experimental\n");
+
+    if (p->rc.rfConstantMin > p->rc.rfConstant)
+    {
+        x265_log(m_param, X265_LOG_WARNING, "CRF min must be less than CRF\n");
+        p->rc.rfConstantMin = 0;
     }
 
     m_bframeDelay = p->bframes ? (p->bBPyramid ? 2 : 1) : 0;
 
-    //====== Coding Tools ========
+    p->bFrameBias = X265_MIN(X265_MAX(-90, p->bFrameBias), 100);
 
-    uint32_t tuQTMaxLog2Size = g_convertToBit[p->maxCUSize] + 2 - 1;
-    m_quadtreeTULog2MaxSize = tuQTMaxLog2Size;
-    uint32_t tuQTMinLog2Size = 2; //log2(4)
-    m_quadtreeTULog2MinSize = tuQTMinLog2Size;
+    if (p->logLevel < X265_LOG_INFO)
+    {
+        /* don't measure these metrics if they will not be reported */
+        p->bEnablePsnr = 0;
+        p->bEnableSsim = 0;
+    }
+    /* Warn users trying to measure PSNR/SSIM with psy opts on. */
+    if (p->bEnablePsnr || p->bEnableSsim)
+    {
+        const char *s = NULL;
+
+        if (p->psyRd || p->psyRdoq)
+        {
+            s = p->bEnablePsnr ? "psnr" : "ssim";
+            x265_log(p, X265_LOG_WARNING, "--%s used with psy on: results will be invalid!\n", s);
+        }
+        else if (!p->rc.aqMode && p->bEnableSsim)
+        {
+            x265_log(p, X265_LOG_WARNING, "--ssim used with AQ off: results will be invalid!\n");
+            s = "ssim";
+        }
+        else if (p->rc.aqStrength > 0 && p->bEnablePsnr)
+        {
+            x265_log(p, X265_LOG_WARNING, "--psnr used with AQ on: results will be invalid!\n");
+            s = "psnr";
+        }
+        if (s)
+            x265_log(p, X265_LOG_WARNING, "--tune %s should be used if attempting to benchmark %s!\n", s, s);
+    }
 
     //========= set default display window ==================================
-    m_defaultDisplayWindow.m_enabledFlag = p->vui.bEnableDefaultDisplayWindowFlag;
-    m_defaultDisplayWindow.m_winRightOffset = p->vui.defDispWinRightOffset;
-    m_defaultDisplayWindow.m_winTopOffset = p->vui.defDispWinTopOffset;
-    m_defaultDisplayWindow.m_winBottomOffset = p->vui.defDispWinBottomOffset;
-    m_defaultDisplayWindow.m_winLeftOffset = p->vui.defDispWinLeftOffset;
-    m_pad[0] = m_pad[1] = 0;
+    m_conformanceWindow.bEnabled = false;
+    m_conformanceWindow.rightOffset = 0;
+    m_conformanceWindow.topOffset = 0;
+    m_conformanceWindow.bottomOffset = 0;
+    m_conformanceWindow.leftOffset = 0;
 
     //======== set pad size if width is not multiple of the minimum CU size =========
-    uint32_t maxCUDepth = (uint32_t)g_convertToBit[p->maxCUSize];
-    uint32_t minCUDepth = (p->maxCUSize >> (maxCUDepth - 1));
-    if ((p->sourceWidth % minCUDepth) != 0)
+    const uint32_t minCUSize = MIN_CU_SIZE;
+    if (p->sourceWidth & (minCUSize - 1))
     {
-        uint32_t padsize = 0;
-        uint32_t rem = p->sourceWidth % minCUDepth;
-        padsize = minCUDepth - rem;
+        uint32_t rem = p->sourceWidth & (minCUSize - 1);
+        uint32_t padsize = minCUSize - rem;
         p->sourceWidth += padsize;
-        m_pad[0] = padsize; //pad width
 
         /* set the confirmation window offsets  */
-        m_conformanceWindow.m_enabledFlag = true;
-        m_conformanceWindow.m_winRightOffset = m_pad[0];
+        m_conformanceWindow.bEnabled = true;
+        m_conformanceWindow.rightOffset = padsize;
     }
 
     //======== set pad size if height is not multiple of the minimum CU size =========
-    if ((p->sourceHeight % minCUDepth) != 0)
+    if (p->sourceHeight & (minCUSize - 1))
     {
-        uint32_t padsize = 0;
-        uint32_t rem = p->sourceHeight % minCUDepth;
-        padsize = minCUDepth - rem;
+        uint32_t rem = p->sourceHeight & (minCUSize - 1);
+        uint32_t padsize = minCUSize - rem;
         p->sourceHeight += padsize;
-        m_pad[1] = padsize; //pad height
 
         /* set the confirmation window offsets  */
-        m_conformanceWindow.m_enabledFlag = true;
-        m_conformanceWindow.m_winBottomOffset = m_pad[1];
-    }
-
-    //====== HM Settings not exposed for configuration ======
-    m_loopFilterOffsetInPPS = 0;
-    m_loopFilterBetaOffsetDiv2 = 0;
-    m_loopFilterTcOffsetDiv2 = 0;
-    m_loopFilterAcrossTilesEnabledFlag = 1;
-
-    m_vps.setMaxTLayers(1);
-    m_vps.setTemporalNestingFlag(true);
-    m_vps.setMaxLayers(1);
-    for (int i = 0; i < MAX_TLAYER; i++)
-    {
-        /* Increase the DPB size and reorder picture if bpyramid is enabled */
-        m_numReorderPics[i] = (p->bBPyramid && p->bframes > 1) ? 2 : 1;
-        m_maxDecPicBuffering[i] = X265_MIN(MAX_NUM_REF, X265_MAX(m_numReorderPics[i] + 1, p->maxNumReferences) + m_numReorderPics[i]);
-
-        m_vps.setNumReorderPics(m_numReorderPics[i], i);
-        m_vps.setMaxDecPicBuffering(m_maxDecPicBuffering[i], i);
-    }
-
-    m_maxCuDQPDepth = 0;
-    m_maxNumOffsetsPerPic = 2048;
-    m_log2ParallelMergeLevelMinus2 = 0;
-
-    m_nonPackedConstraintFlag = false;
-    m_frameOnlyConstraintFlag = false;
-    m_recoveryPointSEIEnabled = 0;
-    m_bufferingPeriodSEIEnabled = 0;
-    m_displayOrientationSEIAngle = 0;
-    m_gradualDecodingRefreshInfoEnabled = 0;
-    m_decodingUnitInfoSEIEnabled = 0;
-    m_useScalingListId = 0;
-    m_activeParameterSetsSEIEnabled = 0;
-    m_minSpatialSegmentationIdc = 0;
-    m_neutralChromaIndicationFlag = false;
-    m_pocProportionalToTimingFlag = false;
-    m_numTicksPocDiffOneMinus1 = 0;
-    m_motionVectorsOverPicBoundariesFlag = false;
-    m_maxBytesPerPicDenom = 2;
-    m_maxBitsPerMinCuDenom = 1;
-    m_log2MaxMvLengthHorizontal = 15;
-    m_log2MaxMvLengthVertical = 15;
-    m_usePCM = 0;
-    m_pcmLog2MinSize = 3;
-    m_pcmLog2MaxSize = 5;
-    m_bPCMInputBitDepthFlag = true;
-    m_bPCMFilterDisableFlag = false;
-
-    m_useLossless = false;  // x264 configures this via --qp=0
-    m_TransquantBypassEnableFlag = false;
-    m_CUTransquantBypassFlagValue = false;
-}
-
-int Encoder::extractNalData(NALUnitEBSP **nalunits)
-{
-    uint32_t memsize = 0;
-    uint32_t offset = 0;
-    int nalcount = 0;
-
-    int num = 0;
-
-    for (; num < MAX_NAL_UNITS && nalunits[num] != NULL; num++)
-    {
-        const NALUnitEBSP& temp = *nalunits[num];
-        memsize += temp.m_packetSize + 4;
+        m_conformanceWindow.bEnabled = true;
+        m_conformanceWindow.bottomOffset = padsize;
     }
-
-    X265_FREE(m_packetData);
-    X265_FREE(m_nals);
-    CHECKED_MALLOC(m_packetData, char, memsize);
-    CHECKED_MALLOC(m_nals, x265_nal, num);
-
-    memsize = 0;
-
-    /* Copy NAL output packets into x265_nal_t structures */
-    for (; nalcount < num; nalcount++)
-    {
-        const NALUnitEBSP& nalu = *nalunits[nalcount];
-        uint32_t size; /* size of annexB unit in bytes */
-
-        static const char start_code_prefix[] = { 0, 0, 0, 1 };
-        if (nalcount == 0 || nalu.m_nalUnitType == NAL_UNIT_SPS || nalu.m_nalUnitType == NAL_UNIT_PPS)
-        {
-            /* From AVC, When any of the following conditions are fulfilled, the
-             * zero_byte syntax element shall be present:
-             *  - the nal_unit_type within the nal_unit() is equal to 7 (sequence
-             *    parameter set) or 8 (picture parameter set),
-             *  - the byte stream NAL unit syntax structure contains the first NAL
-             *    unit of an access unit in decoding order, as specified by subclause
-             *    7.4.1.2.3.
-             */
-            ::memcpy(m_packetData + memsize, start_code_prefix, 4);
-            size = 4;
-        }
-        else
-        {
-            ::memcpy(m_packetData + memsize, start_code_prefix + 1, 3);
-            size = 3;
-        }
-        memsize += size;
-        ::memcpy(m_packetData + memsize, nalu.m_nalUnitData, nalu.m_packetSize);
-        memsize += nalu.m_packetSize;
-
-        m_nals[nalcount].type = nalu.m_nalUnitType;
-        m_nals[nalcount].sizeBytes = size + nalu.m_packetSize;
-    }
-
-    /* Setup payload pointers, now that we're done adding content to m_packetData */
-    for (int i = 0; i < nalcount; i++)
-    {
-        m_nals[i].payload = (uint8_t*)m_packetData + offset;
-        offset += m_nals[i].sizeBytes;
-    }
-
-fail:
-    return nalcount;
 }

x265-hg20140324.tar.bz2/source/encoder/encoder.h -> x265-hg20140928.tar.bz2/source/encoder/encoder.h Changed

@@ -18,19 +18,25 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_ENCODER_H
 #define X265_ENCODER_H
 
+#include "common.h"
+#include "slice.h"
+#include "scalinglist.h"
 #include "x265.h"
+#include "nal.h"
 
-#include "TLibCommon/TComSlice.h"
+struct x265_encoder {};
 
-#include "piclist.h"
+namespace x265 {
+// private namespace
+extern const char g_sliceTypeToChar[3];
 
-struct x265_encoder {};
+class Entropy;
 
 struct EncStats
 {
@@ -59,38 +65,35 @@
     void addSsim(double ssim);
 };
 
-namespace x265 {
-// private namespace
-
 class FrameEncoder;
 class DPB;
-struct Lookahead;
-struct RateControl;
+class Lookahead;
+class RateControl;
 class ThreadPool;
-struct NALUnitEBSP;
+struct ThreadLocalData;
 
 class Encoder : public x265_encoder
 {
 private:
 
-    bool               m_aborted;          // fatal error detected
     int                m_pocLast;          ///< time index (POC)
+    int                m_encodedFrameNum;
     int                m_outputCount;
-    PicList            m_freeList;
 
     int                m_bframeDelay;
     int64_t            m_firstPts;
     int64_t            m_bframeDelayTime;
     int64_t            m_prevReorderedPts[2];
-    int64_t            m_encodedFrameNum;
 
     ThreadPool*        m_threadPool;
-    Lookahead*         m_lookahead;
     FrameEncoder*      m_frameEncoder;
     DPB*               m_dpb;
-    /* frame parallelism */
+
+    Frame*             m_exportedPic;
+
     int                m_curEncoder;
 
+
     /* Collect statistics globally */
     EncStats           m_analyzeAll;
     EncStats           m_analyzeI;
@@ -99,114 +102,43 @@
     FILE*              m_csvfpt;
     int64_t            m_encodeStartTime;
 
-    // quality control
-    TComScalingList    m_scalingList;      ///< quantization matrix information
-
     // weighted prediction
     int                m_numLumaWPFrames;    // number of P frames with weighted luma reference
     int                m_numChromaWPFrames;  // number of P frames with weighted chroma reference
+    int                m_numLumaWPBiFrames;  // number of B frames with weighted luma reference
+    int                m_numChromaWPBiFrames; // number of B frames with weighted chroma reference
 
 public:
 
     int                m_conformanceMode;
-    TComVPS            m_vps;
-
-    /* profile & level */
-    Profile::Name      m_profile;
-    Level::Tier        m_levelTier;
-    Level::Name        m_level;
-
-    bool               m_nonPackedConstraintFlag;
-    bool               m_frameOnlyConstraintFlag;
-
-    //====== Coding Structure ========
-    int                m_maxDecPicBuffering[MAX_TLAYER];
-    int                m_numReorderPics[MAX_TLAYER];
-    int                m_maxRefPicNum;
-
-    //======= Transform =============
-    uint32_t           m_quadtreeTULog2MaxSize;
-    uint32_t           m_quadtreeTULog2MinSize;
-
-    //====== Loop/Deblock Filter ========
-    bool               m_loopFilterOffsetInPPS;
-    int                m_loopFilterBetaOffset;
-    int                m_loopFilterBetaOffsetDiv2;
-    int                m_loopFilterTcOffset;
-    int                m_loopFilterTcOffsetDiv2;
-    int                m_maxNumOffsetsPerPic;
-
-    //====== Lossless ========
-    bool               m_useLossless;
-
-    //====== Quality control ========
-    int                m_maxCuDQPDepth;    //  Max. depth for a minimum CuDQP (0:default)
-
-    //====== Tool list ========
-    bool               m_usePCM;
-    uint32_t           m_pcmLog2MaxSize;
-    uint32_t           m_pcmLog2MinSize;
-
-    bool               m_bPCMInputBitDepthFlag; //unused field
-    uint32_t           m_pcmBitDepthLuma;  // unused field, TComSPS has it's own version defaulted to 8
-    uint32_t           m_pcmBitDepthChroma;// unused field, TComSPS has it's own version defaulted to 8
-
-    bool               m_bPCMFilterDisableFlag;
-    bool               m_loopFilterAcrossTilesEnabledFlag;
-
-    int                m_bufferingPeriodSEIEnabled;
-    int                m_recoveryPointSEIEnabled;
-    int                m_displayOrientationSEIAngle;
-    int                m_gradualDecodingRefreshInfoEnabled;
-    int                m_decodingUnitInfoSEIEnabled;
-
-    uint32_t           m_log2ParallelMergeLevelMinus2; ///< Parallel merge estimation region
-
-    int                m_useScalingListId; ///< Using quantization matrix i.e. 0=off, 1=default.
-
-    bool               m_TransquantBypassEnableFlag;   ///< transquant_bypass_enable_flag setting in PPS.
-    bool               m_CUTransquantBypassFlagValue;  ///< if transquant_bypass_enable_flag, the fixed value to use for the per-CU cu_transquant_bypass_flag.
-    int                m_activeParameterSetsSEIEnabled;///< enable active parameter set SEI message
-
-    bool               m_neutralChromaIndicationFlag;
-    bool               m_pocProportionalToTimingFlag;
-    int                m_numTicksPocDiffOneMinus1;
-    bool               m_tilesFixedStructureFlag;
-    bool               m_motionVectorsOverPicBoundariesFlag;
-    bool               m_restrictedRefPicListsFlag;
-    int                m_minSpatialSegmentationIdc;
-    int                m_maxBytesPerPicDenom;
-    int                m_maxBitsPerMinCuDenom;
-    int                m_log2MaxMvLengthHorizontal;
-    int                m_log2MaxMvLengthVertical;
-
-    x265_param*        param;
-    RateControl*       m_rateControl;
+    VPS                m_vps;
+    SPS                m_sps;
+    PPS                m_pps;
+    NALList            m_nalList;
+    ScalingList        m_scalingList;      // quantization matrix information
 
-    int                bEnableRDOQ;
-    int                bEnableRDOQTS;
+    int                m_lastBPSEI;
+    uint32_t           m_numDelayedPic;
 
-    int                m_pad[2];
+    x265_param*        m_param;
+    RateControl*       m_rateControl;
+    ThreadLocalData*   m_threadLocalData;
+    Lookahead*         m_lookahead;
     Window             m_conformanceWindow;
-    Window             m_defaultDisplayWindow;
 
-    x265_nal*          m_nals;
-    char*              m_packetData;
+    bool               m_aborted;          // fatal error detected
 
     Encoder();
 
-    virtual ~Encoder();
+    ~Encoder() {}
 
     void create();
     void destroy();
     void init();
 
-    void initSPS(TComSPS *sps);
-    void initPPS(TComPPS *pps);
-
-    int encode(bool bEos, const x265_picture* pic, x265_picture *pic_out, NALUnitEBSP **nalunits);
+    int encode(const x265_picture* pic, x265_picture *pic_out);
 
-    int getStreamHeaders(NALUnitEBSP **nalunits);
+    void getStreamHeaders(NALList& list, Entropy& sbacCoder, Bitstream& bs);
 
     void fetchStats(x265_stats* stats, size_t statsSizeBytes);
 
@@ -216,21 +148,20 @@
 
     char* statsString(EncStats&, char*);
 
-    TComScalingList* getScalingList() { return &m_scalingList; }
+    char* statsCSVString(EncStats& stat, char* buffer);
 
     void setThreadPool(ThreadPool* p) { m_threadPool = p; }
 
     void configure(x265_param *param);
 
-    void determineLevelAndProfile(x265_param *param);
-
-    int  extractNalData(NALUnitEBSP **nalunits);
-
     void updateVbvPlan(RateControl* rc);
 
 protected:
 
-    void finishFrameStats(TComPic* pic, FrameEncoder *curEncoder, uint64_t bits);
+    void initSPS(SPS *sps);
+    void initPPS(PPS *pps);
+
+    void finishFrameStats(Frame* pic, FrameEncoder *curEncoder, uint64_t bits);
 };
 }

x265-hg20140928.tar.bz2/source/encoder/entropy.cpp Added

@@ -0,0 +1,2180 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "common.h"
+#include "scalinglist.h"
+#include "quant.h"
+#include "sao.h"
+#include "entropy.h"
+
+#define CU_DQP_TU_CMAX 5 // max number bins for truncated unary
+#define CU_DQP_EG_k    0 // exp-golomb order
+#define START_VALUE    8 // start value for dpcm mode
+
+namespace x265 {
+
+Entropy::Entropy()
+{
+    m_fracBits = 0;
+    X265_CHECK(sizeof(m_contextState) >= sizeof(m_contextState[0]) * MAX_OFF_CTX_MOD, "context state table is too small\n");
+}
+
+void Entropy::codeVPS(VPS* vps)
+{
+    WRITE_CODE(0,       4, "vps_video_parameter_set_id");
+    WRITE_CODE(3,       2, "vps_reserved_three_2bits");
+    WRITE_CODE(0,       6, "vps_reserved_zero_6bits");
+    WRITE_CODE(0,       3, "vps_max_sub_layers_minus1");
+    WRITE_FLAG(1,          "vps_temporal_id_nesting_flag");
+    WRITE_CODE(0xffff, 16, "vps_reserved_ffff_16bits");
+
+    codeProfileTier(vps->ptl);
+
+    WRITE_FLAG(true, "vps_sub_layer_ordering_info_present_flag");
+    WRITE_UVLC(vps->maxDecPicBuffering - 1, "vps_max_dec_pic_buffering_minus1[i]");
+    WRITE_UVLC(vps->numReorderPics,         "vps_num_reorder_pics[i]");
+
+    WRITE_UVLC(0,    "vps_max_latency_increase_plus1[i]");
+    WRITE_CODE(0, 6, "vps_max_nuh_reserved_zero_layer_id");
+    WRITE_UVLC(0,    "vps_max_op_sets_minus1");
+    WRITE_FLAG(0,    "vps_timing_info_present_flag"); /* we signal timing info in SPS-VUI */
+    WRITE_FLAG(0,    "vps_extension_flag");
+}
+
+void Entropy::codeSPS(SPS* sps, ScalingList *scalingList, ProfileTierLevel *ptl)
+{
+    WRITE_CODE(0, 4, "sps_video_parameter_set_id");
+    WRITE_CODE(0, 3, "sps_max_sub_layers_minus1");
+    WRITE_FLAG(1,    "sps_temporal_id_nesting_flag");
+
+    codeProfileTier(*ptl);
+
+    WRITE_UVLC(0, "sps_seq_parameter_set_id");
+    WRITE_UVLC(sps->chromaFormatIdc, "chroma_format_idc");
+
+    if (sps->chromaFormatIdc == X265_CSP_I444)
+        WRITE_FLAG(0,                        "separate_colour_plane_flag");
+
+    WRITE_UVLC(sps->picWidthInLumaSamples,   "pic_width_in_luma_samples");
+    WRITE_UVLC(sps->picHeightInLumaSamples,  "pic_height_in_luma_samples");
+
+    Window& conf = sps->conformanceWindow;
+    WRITE_FLAG(conf.bEnabled, "conformance_window_flag");
+    if (conf.bEnabled)
+    {
+        WRITE_UVLC(conf.leftOffset   / g_winUnitX[sps->chromaFormatIdc], "conf_win_left_offset");
+        WRITE_UVLC(conf.rightOffset  / g_winUnitX[sps->chromaFormatIdc], "conf_win_right_offset");
+        WRITE_UVLC(conf.topOffset    / g_winUnitY[sps->chromaFormatIdc], "conf_win_top_offset");
+        WRITE_UVLC(conf.bottomOffset / g_winUnitY[sps->chromaFormatIdc], "conf_win_bottom_offset");
+    }
+
+    WRITE_UVLC(X265_DEPTH - 8,   "bit_depth_luma_minus8");
+    WRITE_UVLC(X265_DEPTH - 8,   "bit_depth_chroma_minus8");
+    WRITE_UVLC(BITS_FOR_POC - 4, "log2_max_pic_order_cnt_lsb_minus4");
+    WRITE_FLAG(true,             "sps_sub_layer_ordering_info_present_flag");
+
+    WRITE_UVLC(sps->maxDecPicBuffering - 1, "sps_max_dec_pic_buffering_minus1[i]");
+    WRITE_UVLC(sps->numReorderPics,         "sps_num_reorder_pics[i]");
+    WRITE_UVLC(0,                           "sps_max_latency_increase_plus1[i]");
+
+    WRITE_UVLC(sps->log2MinCodingBlockSize - 3,    "log2_min_coding_block_size_minus3");
+    WRITE_UVLC(sps->log2DiffMaxMinCodingBlockSize, "log2_diff_max_min_coding_block_size");
+    WRITE_UVLC(sps->quadtreeTULog2MinSize - 2,     "log2_min_transform_block_size_minus2");
+    WRITE_UVLC(sps->quadtreeTULog2MaxSize - sps->quadtreeTULog2MinSize, "log2_diff_max_min_transform_block_size");
+    WRITE_UVLC(sps->quadtreeTUMaxDepthInter - 1,   "max_transform_hierarchy_depth_inter");
+    WRITE_UVLC(sps->quadtreeTUMaxDepthIntra - 1,   "max_transform_hierarchy_depth_intra");
+    WRITE_FLAG(scalingList->m_bEnabled,            "scaling_list_enabled_flag");
+    if (scalingList->m_bEnabled)
+    {
+        WRITE_FLAG(scalingList->m_bDataPresent,    "sps_scaling_list_data_present_flag");
+        if (scalingList->m_bDataPresent)
+            codeScalingList(scalingList);
+    }
+    WRITE_FLAG(sps->bUseAMP, "amp_enabled_flag");
+    WRITE_FLAG(sps->bUseSAO, "sample_adaptive_offset_enabled_flag");
+
+    WRITE_FLAG(0, "pcm_enabled_flag");
+    WRITE_UVLC(0, "num_short_term_ref_pic_sets");
+    WRITE_FLAG(0, "long_term_ref_pics_present_flag");
+
+    WRITE_FLAG(1, "sps_temporal_mvp_enable_flag");
+    WRITE_FLAG(sps->bUseStrongIntraSmoothing, "sps_strong_intra_smoothing_enable_flag");
+
+    WRITE_FLAG(1, "vui_parameters_present_flag");
+    codeVUI(&sps->vuiParameters);
+
+    WRITE_FLAG(0, "sps_extension_flag");
+}
+
+void Entropy::codePPS(PPS* pps)
+{
+    WRITE_UVLC(0,                          "pps_pic_parameter_set_id");
+    WRITE_UVLC(0,                          "pps_seq_parameter_set_id");
+    WRITE_FLAG(0,                          "dependent_slice_segments_enabled_flag");
+    WRITE_FLAG(0,                          "output_flag_present_flag");
+    WRITE_CODE(0, 3,                       "num_extra_slice_header_bits");
+    WRITE_FLAG(pps->bSignHideEnabled,      "sign_data_hiding_flag");
+    WRITE_FLAG(0,                          "cabac_init_present_flag");
+    WRITE_UVLC(0,                          "num_ref_idx_l0_default_active_minus1");
+    WRITE_UVLC(0,                          "num_ref_idx_l1_default_active_minus1");
+
+    WRITE_SVLC(0, "init_qp_minus26");
+    WRITE_FLAG(pps->bConstrainedIntraPred, "constrained_intra_pred_flag");
+    WRITE_FLAG(pps->bTransformSkipEnabled, "transform_skip_enabled_flag");
+
+    WRITE_FLAG(pps->bUseDQP,               "cu_qp_delta_enabled_flag");
+    if (pps->bUseDQP)
+        WRITE_UVLC(pps->maxCuDQPDepth,     "diff_cu_qp_delta_depth");
+
+    WRITE_SVLC(pps->chromaCbQpOffset,      "pps_cb_qp_offset");
+    WRITE_SVLC(pps->chromaCrQpOffset,      "pps_cr_qp_offset");
+    WRITE_FLAG(0,                          "pps_slice_chroma_qp_offsets_present_flag");
+
+    WRITE_FLAG(pps->bUseWeightPred,            "weighted_pred_flag");
+    WRITE_FLAG(pps->bUseWeightedBiPred,        "weighted_bipred_flag");
+    WRITE_FLAG(pps->bTransquantBypassEnabled,  "transquant_bypass_enable_flag");
+    WRITE_FLAG(0,                              "tiles_enabled_flag");
+    WRITE_FLAG(pps->bEntropyCodingSyncEnabled, "entropy_coding_sync_enabled_flag");
+    WRITE_FLAG(1,                              "loop_filter_across_slices_enabled_flag");
+
+    WRITE_FLAG(pps->bDeblockingFilterControlPresent, "deblocking_filter_control_present_flag");
+    if (pps->bDeblockingFilterControlPresent)
+    {
+        WRITE_FLAG(0,                                "deblocking_filter_override_enabled_flag");
+        WRITE_FLAG(pps->bPicDisableDeblockingFilter, "pps_disable_deblocking_filter_flag");
+        if (!pps->bPicDisableDeblockingFilter)
+        {
+            WRITE_SVLC(pps->deblockingFilterBetaOffsetDiv2, "pps_beta_offset_div2");
+            WRITE_SVLC(pps->deblockingFilterTcOffsetDiv2,   "pps_tc_offset_div2");
+        }
+    }
+
+    WRITE_FLAG(0, "pps_scaling_list_data_present_flag");
+    WRITE_FLAG(0, "lists_modification_present_flag");
+    WRITE_UVLC(0, "log2_parallel_merge_level_minus2");
+    WRITE_FLAG(0, "slice_segment_header_extension_present_flag");
+    WRITE_FLAG(0, "pps_extension_flag");
+}
+
+void Entropy::codeProfileTier(ProfileTierLevel& ptl)
+{
+    WRITE_CODE(0, 2,                "XXX_profile_space[]");
+    WRITE_FLAG(ptl.tierFlag,        "XXX_tier_flag[]");
+    WRITE_CODE(ptl.profileIdc, 5,   "XXX_profile_idc[]");
+    for (int j = 0; j < 32; j++)
+        WRITE_FLAG(ptl.profileCompatibilityFlag[j], "XXX_profile_compatibility_flag[][j]");
+
+    WRITE_FLAG(ptl.progressiveSourceFlag,   "general_progressive_source_flag");
+    WRITE_FLAG(ptl.interlacedSourceFlag,    "general_interlaced_source_flag");
+    WRITE_FLAG(ptl.nonPackedConstraintFlag, "general_non_packed_constraint_flag");
+    WRITE_FLAG(ptl.frameOnlyConstraintFlag, "general_frame_only_constraint_flag");
+
+    WRITE_CODE(0, 16, "XXX_reserved_zero_44bits[0..15]");
+    WRITE_CODE(0, 16, "XXX_reserved_zero_44bits[16..31]");
+    WRITE_CODE(0, 12, "XXX_reserved_zero_44bits[32..43]");
+
+    WRITE_CODE(ptl.levelIdc, 8, "general_level_idc");
+}
+
+void Entropy::codeVUI(VUI *vui)
+{
+    WRITE_FLAG(vui->aspectRatioInfoPresentFlag,  "aspect_ratio_info_present_flag");
+    if (vui->aspectRatioInfoPresentFlag)
+    {
+        WRITE_CODE(vui->aspectRatioIdc, 8,       "aspect_ratio_idc");
+        if (vui->aspectRatioIdc == 255)
+        {
+            WRITE_CODE(vui->sarWidth, 16,        "sar_width");
+            WRITE_CODE(vui->sarHeight, 16,       "sar_height");
+        }
+    }
+
+    WRITE_FLAG(vui->overscanInfoPresentFlag,     "overscan_info_present_flag");
+    if (vui->overscanInfoPresentFlag)
+        WRITE_FLAG(vui->overscanAppropriateFlag, "overscan_appropriate_flag");
+
+    WRITE_FLAG(vui->videoSignalTypePresentFlag,  "video_signal_type_present_flag");
+    if (vui->videoSignalTypePresentFlag)
+    {
+        WRITE_CODE(vui->videoFormat, 3,          "video_format");
+        WRITE_FLAG(vui->videoFullRangeFlag,      "video_full_range_flag");
+        WRITE_FLAG(vui->colourDescriptionPresentFlag, "colour_description_present_flag");
+        if (vui->colourDescriptionPresentFlag)
+        {
+            WRITE_CODE(vui->colourPrimaries, 8,         "colour_primaries");
+            WRITE_CODE(vui->transferCharacteristics, 8, "transfer_characteristics");
+            WRITE_CODE(vui->matrixCoefficients, 8,      "matrix_coefficients");
+        }
+    }
+
+    WRITE_FLAG(vui->chromaLocInfoPresentFlag,           "chroma_loc_info_present_flag");
+    if (vui->chromaLocInfoPresentFlag)
+    {
+        WRITE_UVLC(vui->chromaSampleLocTypeTopField,    "chroma_sample_loc_type_top_field");
+        WRITE_UVLC(vui->chromaSampleLocTypeBottomField, "chroma_sample_loc_type_bottom_field");
+    }
+
+    WRITE_FLAG(0,                                     "neutral_chroma_indication_flag");
+    WRITE_FLAG(vui->fieldSeqFlag,                     "field_seq_flag");
+    WRITE_FLAG(vui->frameFieldInfoPresentFlag,        "frame_field_info_present_flag");
+
+    Window defaultDisplayWindow = vui->defaultDisplayWindow;
+    WRITE_FLAG(defaultDisplayWindow.bEnabled,         "default_display_window_flag");
+    if (defaultDisplayWindow.bEnabled)
+    {
+        WRITE_UVLC(defaultDisplayWindow.leftOffset,   "def_disp_win_left_offset");
+        WRITE_UVLC(defaultDisplayWindow.rightOffset,  "def_disp_win_right_offset");
+        WRITE_UVLC(defaultDisplayWindow.topOffset,    "def_disp_win_top_offset");
+        WRITE_UVLC(defaultDisplayWindow.bottomOffset, "def_disp_win_bottom_offset");
+    }
+
+    TimingInfo *timingInfo = &vui->timingInfo;
+    WRITE_FLAG(1,                              "vui_timing_info_present_flag");
+    WRITE_CODE(timingInfo->numUnitsInTick, 32, "vui_num_units_in_tick");
+    WRITE_CODE(timingInfo->timeScale,      32, "vui_time_scale");
+    WRITE_FLAG(0,                              "vui_poc_proportional_to_timing_flag");
+
+    WRITE_FLAG(vui->hrdParametersPresentFlag,  "vui_hrd_parameters_present_flag");
+    if (vui->hrdParametersPresentFlag)
+        codeHrdParameters(&vui->hrdParameters);
+
+    WRITE_FLAG(0, "bitstream_restriction_flag");
+}
+
+void Entropy::codeScalingList(ScalingList* scalingList)
+{
+    for (int sizeId = 0; sizeId < ScalingList::NUM_SIZES; sizeId++)
+    {
+        for (int listId = 0; listId < ScalingList::NUM_LISTS; listId++)
+        {
+            int predList = scalingList->checkPredMode(sizeId, listId);
+            WRITE_FLAG(predList < 0, "scaling_list_pred_mode_flag");
+            if (predList >= 0)
+                WRITE_UVLC(listId - predList, "scaling_list_pred_matrix_id_delta");
+            else // DPCM Mode
+                codeScalingList(scalingList, sizeId, listId);
+        }
+    }
+}
+
+void Entropy::codeScalingList(ScalingList* scalingList, uint32_t sizeId, uint32_t listId)
+{
+    int coefNum = X265_MIN(ScalingList::MAX_MATRIX_COEF_NUM, (int)ScalingList::s_numCoefPerSize[sizeId]);
+    const uint16_t* scan = (sizeId == 0 ? g_scan4x4[SCAN_DIAG] : g_scan8x8diag);
+    int nextCoef = START_VALUE;
+    int32_t *src = scalingList->m_scalingListCoef[sizeId][listId];
+    int data;
+
+    if (sizeId > BLOCK_8x8)
+    {
+        WRITE_SVLC(scalingList->m_scalingListDC[sizeId][listId] - 8, "scaling_list_dc_coef_minus8");
+        nextCoef = scalingList->m_scalingListDC[sizeId][listId];
+    }
+    for (int i = 0; i < coefNum; i++)
+    {
+        data = src[scan[i]] - nextCoef;
+        nextCoef = src[scan[i]];
+        if (data > 127)
+            data = data - 256;
+        if (data < -128)
+            data = data + 256;
+
+        WRITE_SVLC(data,  "scaling_list_delta_coef");
+    }
+}
+
+void Entropy::codeHrdParameters(HRDInfo *hrd)
+{
+    WRITE_FLAG(1, "nal_hrd_parameters_present_flag");
+    WRITE_FLAG(0, "vcl_hrd_parameters_present_flag");
+    WRITE_FLAG(0, "sub_pic_hrd_params_present_flag");
+
+    WRITE_CODE(hrd->bitRateScale, 4, "bit_rate_scale");
+    WRITE_CODE(hrd->cpbSizeScale, 4, "cpb_size_scale");
+
+    WRITE_CODE(hrd->initialCpbRemovalDelayLength - 1, 5, "initial_cpb_removal_delay_length_minus1");
+    WRITE_CODE(hrd->cpbRemovalDelayLength - 1,        5, "au_cpb_removal_delay_length_minus1");
+    WRITE_CODE(hrd->dpbOutputDelayLength - 1,         5, "dpb_output_delay_length_minus1");
+
+    WRITE_FLAG(1, "fixed_pic_rate_general_flag");
+    WRITE_UVLC(0, "elemental_duration_in_tc_minus1");
+    WRITE_UVLC(0, "cpb_cnt_minus1");
+
+    WRITE_UVLC(hrd->bitRateValue - 1, "bit_rate_value_minus1");
+    WRITE_UVLC(hrd->cpbSizeValue - 1, "cpb_size_value_minus1");
+    WRITE_FLAG(hrd->cbrFlag, "cbr_flag");
+}
+
+void Entropy::codeAUD(Slice* slice)
+{
+    int picType;
+
+    switch (slice->m_sliceType)
+    {
+    case I_SLICE:
+        picType = 0;
+        break;
+    case P_SLICE:
+        picType = 1;
+        break;
+    case B_SLICE:
+        picType = 2;
+        break;
+    default:
+        picType = 7;
+        break;
+    }
+
+    WRITE_CODE(picType, 3, "pic_type");
+}
+
+void Entropy::codeSliceHeader(Slice* slice)
+{
+    WRITE_FLAG(1, "first_slice_segment_in_pic_flag");
+    if (slice->getRapPicFlag())
+        WRITE_FLAG(0, "no_output_of_prior_pics_flag");
+
+    WRITE_UVLC(0, "slice_pic_parameter_set_id");
+
+    /* x265 does not use dependent slices, so always write all this data */
+
+    WRITE_UVLC(slice->m_sliceType, "slice_type");
+
+    if (!slice->getIdrPicFlag())
+    {
+        int picOrderCntLSB = (slice->m_poc - slice->m_lastIDR + (1 << BITS_FOR_POC)) % (1 << BITS_FOR_POC);
+        WRITE_CODE(picOrderCntLSB, BITS_FOR_POC, "pic_order_cnt_lsb");
+
+#if _DEBUG || CHECKED_BUILD
+        // check for bitstream restriction stating that:
+        // If the current picture is a BLA or CRA picture, the value of NumPocTotalCurr shall be equal to 0.
+        // Ideally this process should not be repeated for each slice in a picture
+        if (slice->isIRAP())
+            for (int picIdx = 0; picIdx < slice->m_rps.numberOfPictures; picIdx++)
+                X265_CHECK(!slice->m_rps.bUsed[picIdx], "pic unused failure\n");
+#endif
+
+        WRITE_FLAG(0, "short_term_ref_pic_set_sps_flag");
+        codeShortTermRefPicSet(&slice->m_rps);
+
+        WRITE_FLAG(1, "slice_temporal_mvp_enable_flag");
+    }
+    SAOParam *saoParam = slice->m_pic->getPicSym()->m_saoParam;
+    if (slice->m_sps->bUseSAO)
+    {
+        WRITE_FLAG(saoParam->bSaoFlag[0], "slice_sao_luma_flag");
+        WRITE_FLAG(saoParam->bSaoFlag[1], "slice_sao_chroma_flag");
+    }
+
+    // check if numRefIdx match the defaults (1, hard-coded in PPS). If not, override
+    // TODO: this might be a place to optimize a few bits per slice, by using param->refs for L0 default
+
+    if (!slice->isIntra())
+    {
+        bool overrideFlag = (slice->m_numRefIdx[0] != 1 || (slice->isInterB() && slice->m_numRefIdx[1] != 1));
+        WRITE_FLAG(overrideFlag, "num_ref_idx_active_override_flag");
+        if (overrideFlag)
+        {
+            WRITE_UVLC(slice->m_numRefIdx[0] - 1, "num_ref_idx_l0_active_minus1");
+            if (slice->isInterB())
+                WRITE_UVLC(slice->m_numRefIdx[1] - 1, "num_ref_idx_l1_active_minus1");
+            else
+                slice->m_numRefIdx[1] = 0;
+        }
+    }
+    else
+    {
+        // TODO: why reset these here and above?
+        slice->m_numRefIdx[0] = 0;
+        slice->m_numRefIdx[1] = 0;
+    }
+
+    if (slice->isInterB())
+        WRITE_FLAG(0, "mvd_l1_zero_flag");
+
+    // TMVP always enabled
+    {
+        if (slice->m_sliceType == B_SLICE)
+            WRITE_FLAG(slice->m_colFromL0Flag, "collocated_from_l0_flag");
+
+        if (slice->m_sliceType != I_SLICE &&
+            ((slice->m_colFromL0Flag && slice->m_numRefIdx[0] > 1) ||
+            (!slice->m_colFromL0Flag && slice->m_numRefIdx[1] > 1)))
+        {
+            WRITE_UVLC(slice->m_colRefIdx, "collocated_ref_idx");
+        }
+    }
+    if ((slice->m_pps->bUseWeightPred && slice->m_sliceType == P_SLICE) || (slice->m_pps->bUseWeightedBiPred && slice->m_sliceType == B_SLICE))
+        codePredWeightTable(slice);
+
+    X265_CHECK(slice->m_maxNumMergeCand <= MRG_MAX_NUM_CANDS, "too many merge candidates\n");
+    if (!slice->isIntra())
+        WRITE_UVLC(MRG_MAX_NUM_CANDS - slice->m_maxNumMergeCand, "five_minus_max_num_merge_cand");
+
+    int code = slice->m_sliceQp - 26;
+    WRITE_SVLC(code, "slice_qp_delta");
+
+    bool isSAOEnabled = slice->m_sps->bUseSAO ? saoParam->bSaoFlag[0] || saoParam->bSaoFlag[1] : false;
+    bool isDBFEnabled = !slice->m_pps->bPicDisableDeblockingFilter;
+
+    if (isSAOEnabled || isDBFEnabled)
+        WRITE_FLAG(slice->m_sLFaseFlag, "slice_loop_filter_across_slices_enabled_flag");
+}
+
+/** write wavefront substreams sizes for the slice header */
+void Entropy::codeSliceHeaderWPPEntryPoints(Slice* slice, uint32_t *substreamSizes, uint32_t maxOffset)
+{
+    uint32_t offsetLen = 1;
+    while (maxOffset >= (1U << offsetLen))
+    {
+        offsetLen++;
+        X265_CHECK(offsetLen < 32, "offsetLen is too large\n");
+    }
+
+    uint32_t numRows = slice->m_pic->getFrameHeightInCU() - 1;
+    WRITE_UVLC(numRows, "num_entry_point_offsets");
+    if (numRows > 0)
+        WRITE_UVLC(offsetLen - 1, "offset_len_minus1");
+
+    for (uint32_t i = 0; i < numRows; i++)
+        WRITE_CODE(substreamSizes[i] - 1, offsetLen, "entry_point_offset_minus1");
+}
+
+void Entropy::codeShortTermRefPicSet(RPS* rps)
+{
+    WRITE_UVLC(rps->numberOfNegativePictures, "num_negative_pics");
+    WRITE_UVLC(rps->numberOfPositivePictures, "num_positive_pics");
+    int prev = 0;
+    for (int j = 0; j < rps->numberOfNegativePictures; j++)
+    {
+        WRITE_UVLC(prev - rps->deltaPOC[j] - 1, "delta_poc_s0_minus1");
+        prev = rps->deltaPOC[j];
+        WRITE_FLAG(rps->bUsed[j], "used_by_curr_pic_s0_flag");
+    }
+
+    prev = 0;
+    for (int j = rps->numberOfNegativePictures; j < rps->numberOfNegativePictures + rps->numberOfPositivePictures; j++)
+    {
+        WRITE_UVLC(rps->deltaPOC[j] - prev - 1, "delta_poc_s1_minus1");
+        prev = rps->deltaPOC[j];
+        WRITE_FLAG(rps->bUsed[j], "used_by_curr_pic_s1_flag");
+    }
+}
+
+void Entropy::encodeCTU(TComDataCU* cu)
+{
+    bool bEncodeDQP = cu->m_slice->m_pps->bUseDQP;
+    encodeCU(cu, 0, 0, bEncodeDQP, cu->m_CULocalData);
+}
+
+/* encode a CU block recursively */
+void Entropy::encodeCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool& bEncodeDQP, CU* cuData)
+{
+    Slice* slice = cu->m_slice;
+
+    if (depth <= slice->m_pps->maxCuDQPDepth && slice->m_pps->bUseDQP)
+        bEncodeDQP = true;
+
+    int cuSplitFlag = !(cuData->flags & CU::LEAF);
+    int cuUnsplitFlag = !(cuData->flags & CU::SPLIT_MANDATORY);
+
+    if (!cuUnsplitFlag)
+    {
+        uint32_t qNumParts = (NUM_CU_PARTITIONS >> (depth << 1)) >> 2;
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
+        {
+            CU *childCU = cu->m_CULocalData + cuData->childIdx + partUnitIdx;
+            int cuPresentFlagChild = !(childCU->flags & CU::PRESENT);
+            if (!cuPresentFlagChild)
+                encodeCU(cu, absPartIdx, depth + 1, bEncodeDQP, childCU);
+        }
+        return;
+    }
+
+    // We need to split, so don't try these modes.
+    if (cuSplitFlag) 
+        codeSplitFlag(cu, absPartIdx, depth);
+
+    if (depth < cu->getDepth(absPartIdx) && depth < g_maxCUDepth)
+    {
+        uint32_t qNumParts = (NUM_CU_PARTITIONS >> (depth << 1)) >> 2;
+
+        for (uint32_t partUnitIdx = 0; partUnitIdx < 4; partUnitIdx++, absPartIdx += qNumParts)
+        {
+            CU *childCU = cu->m_CULocalData + cuData->childIdx + partUnitIdx;
+            encodeCU(cu, absPartIdx, depth + 1, bEncodeDQP, childCU);
+        }
+        return;
+    }
+
+    if (slice->m_pps->bTransquantBypassEnabled)
+        codeCUTransquantBypassFlag(cu->getCUTransquantBypass(absPartIdx));
+
+    if (!slice->isIntra())
+        codeSkipFlag(cu, absPartIdx);
+
+    if (cu->isSkipped(absPartIdx))
+    {
+        codeMergeIndex(cu, absPartIdx);
+        finishCU(cu, absPartIdx, depth);
+        return;
+    }
+
+    if (!slice->isIntra())
+        codePredMode(cu->getPredictionMode(absPartIdx));
+
+    codePartSize(cu, absPartIdx, depth);
+
+    // prediction Info ( Intra : direction mode, Inter : Mv, reference idx )
+    codePredInfo(cu, absPartIdx);
+
+    uint32_t tuDepthRange[2];
+    cu->getQuadtreeTULog2MinSizeInCU(tuDepthRange, absPartIdx);
+
+    // Encode Coefficients, allow codeCoeff() to modify m_bEncodeDQP
+    codeCoeff(cu, absPartIdx, depth, bEncodeDQP, tuDepthRange);
+
+    // --- write terminating bit ---
+    finishCU(cu, absPartIdx, depth);
+}
+
+/* finish encoding a cu and handle end-of-slice conditions */
+void Entropy::finishCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
+{
+    Slice* slice = cu->m_slice;
+
+    // Calculate end address
+    X265_CHECK(slice->m_endCUAddr == slice->realEndAddress(slice->m_endCUAddr), "real end address expected\n");
+    uint32_t realEndAddress = slice->m_endCUAddr;
+    uint32_t cuAddr = cu->getSCUAddr() + absPartIdx;
+
+    uint32_t granularityMask = g_maxCUSize - 1;
+    uint32_t cuSize = 1 << cu->getLog2CUSize(absPartIdx);
+    uint32_t rpelx = cu->getCUPelX() + g_zscanToPelX[absPartIdx] + cuSize;
+    uint32_t bpely = cu->getCUPelY() + g_zscanToPelY[absPartIdx] + cuSize;
+    bool granularityBoundary = (((rpelx & granularityMask) == 0 || (rpelx == slice->m_sps->picWidthInLumaSamples )) &&
+                                ((bpely & granularityMask) == 0 || (bpely == slice->m_sps->picHeightInLumaSamples)));
+
+    if (granularityBoundary)
+    {
+        // Encode slice finish
+        bool bTerminateSlice = false;
+        if (cuAddr + (NUM_CU_PARTITIONS >> (depth << 1)) == realEndAddress)
+            bTerminateSlice = true;
+
+        // The 1-terminating bit is added to all streams, so don't add it here when it's 1.
+        if (!bTerminateSlice)
+            encodeBinTrm(0);
+
+        if (!m_bitIf)
+            resetBits(); // TODO: most likely unnecessary
+    }
+}
+
+void Entropy::encodeTransform(TComDataCU* cu, CoeffCodeState& state, uint32_t offsetLuma, uint32_t offsetChroma, uint32_t absPartIdx,
+                              uint32_t absPartIdxStep, uint32_t depth, uint32_t log2TrSize, uint32_t trIdx, bool& bCodeDQP, uint32_t* depthRange)
+{
+    const bool subdiv = cu->getTransformIdx(absPartIdx) + cu->getDepth(absPartIdx) > (uint8_t)depth;
+    uint32_t hChromaShift = cu->getHorzChromaShift();
+    uint32_t vChromaShift = cu->getVertChromaShift();
+    uint32_t cbfY = cu->getCbf(absPartIdx, TEXT_LUMA, trIdx);
+    uint32_t cbfU = cu->getCbf(absPartIdx, TEXT_CHROMA_U, trIdx);
+    uint32_t cbfV = cu->getCbf(absPartIdx, TEXT_CHROMA_V, trIdx);
+
+    if (!trIdx)
+        state.bakAbsPartIdxCU = absPartIdx;
+
+    if ((log2TrSize == 2) && !(cu->getChromaFormat() == X265_CSP_I444))
+    {
+        uint32_t partNum = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+        if ((absPartIdx & (partNum - 1)) == 0)
+        {
+            state.bakAbsPartIdx   = absPartIdx;
+            state.bakChromaOffset = offsetChroma;
+        }
+        else if ((absPartIdx & (partNum - 1)) == (partNum - 1))
+        {
+            cbfU = cu->getCbf(state.bakAbsPartIdx, TEXT_CHROMA_U, trIdx);
+            cbfV = cu->getCbf(state.bakAbsPartIdx, TEXT_CHROMA_V, trIdx);
+        }
+    }
+
+    if (cu->getPredictionMode(absPartIdx) == MODE_INTRA && cu->getPartitionSize(absPartIdx) == SIZE_NxN && depth == cu->getDepth(absPartIdx))
+    {
+        X265_CHECK(subdiv, "subdivision state failure\n");
+    }
+    else if (cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N) && depth == cu->getDepth(absPartIdx) &&
+             (cu->m_slice->m_sps->quadtreeTUMaxDepthInter == 1))
+    {
+        if (log2TrSize > *depthRange)
+        {
+            X265_CHECK(subdiv, "subdivision state failure\n");
+        }
+        else
+        {
+            X265_CHECK(!subdiv, "subdivision state failure\n");
+        }
+    }
+    else if (log2TrSize > *(depthRange + 1))
+    {
+        X265_CHECK(subdiv, "subdivision state failure\n");
+    }
+    else if (log2TrSize == cu->m_slice->m_sps->quadtreeTULog2MinSize)
+    {
+        X265_CHECK(!subdiv, "subdivision state failure\n");
+    }
+    else if (log2TrSize == *depthRange)
+    {
+        X265_CHECK(!subdiv, "subdivision state failure\n");
+    }
+    else
+    {
+        X265_CHECK(log2TrSize > *depthRange, "transform size failure\n");
+        codeTransformSubdivFlag(subdiv, 5 - log2TrSize);
+    }
+
+    const uint32_t trDepthCurr = depth - cu->getDepth(absPartIdx);
+    const bool bFirstCbfOfCU = trDepthCurr == 0;
+
+    bool mCodeAll = true;
+    const uint32_t numPels = 1 << (log2TrSize * 2 - hChromaShift - vChromaShift);
+    if (numPels < (MIN_TU_SIZE * MIN_TU_SIZE))
+        mCodeAll = false;
+
+    if (bFirstCbfOfCU || mCodeAll)
+    {
+        uint32_t tuSize = 1 << log2TrSize;
+        if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr - 1))
+            codeQtCbf(cu, absPartIdx, absPartIdxStep, (tuSize >> hChromaShift), (tuSize >> vChromaShift), TEXT_CHROMA_U, trDepthCurr, (subdiv == 0));
+        if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr - 1))
+            codeQtCbf(cu, absPartIdx, absPartIdxStep, (tuSize >> hChromaShift), (tuSize >> vChromaShift), TEXT_CHROMA_V, trDepthCurr, (subdiv == 0));
+    }
+    else
+    {
+        X265_CHECK(cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr) == cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepthCurr - 1), "chroma xform size match failure\n");
+        X265_CHECK(cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr) == cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepthCurr - 1), "chroma xform size match failure\n");
+    }
+
+    if (subdiv)
+    {
+        log2TrSize--;
+        uint32_t numCoeff  = 1 << (log2TrSize * 2);
+        uint32_t numCoeffC = (numCoeff >> (hChromaShift + vChromaShift));
+        trIdx++;
+        ++depth;
+        absPartIdxStep >>= 2;
+        const uint32_t partNum = NUM_CU_PARTITIONS >> (depth << 1);
+
+        encodeTransform(cu, state, offsetLuma, offsetChroma, absPartIdx, absPartIdxStep, depth, log2TrSize, trIdx, bCodeDQP, depthRange);
+
+        absPartIdx += partNum;
+        offsetLuma += numCoeff;
+        offsetChroma += numCoeffC;
+        encodeTransform(cu, state, offsetLuma, offsetChroma, absPartIdx, absPartIdxStep, depth, log2TrSize, trIdx, bCodeDQP, depthRange);
+
+        absPartIdx += partNum;
+        offsetLuma += numCoeff;
+        offsetChroma += numCoeffC;
+        encodeTransform(cu, state, offsetLuma, offsetChroma, absPartIdx, absPartIdxStep, depth, log2TrSize, trIdx, bCodeDQP, depthRange);
+
+        absPartIdx += partNum;
+        offsetLuma += numCoeff;
+        offsetChroma += numCoeffC;
+        encodeTransform(cu, state, offsetLuma, offsetChroma, absPartIdx, absPartIdxStep, depth, log2TrSize, trIdx, bCodeDQP, depthRange);
+    }
+    else
+    {
+        if (cu->getPredictionMode(absPartIdx) != MODE_INTRA && depth == cu->getDepth(absPartIdx) && !cu->getCbf(absPartIdx, TEXT_CHROMA_U, 0) && !cu->getCbf(absPartIdx, TEXT_CHROMA_V, 0))
+        {
+            X265_CHECK(cu->getCbf(absPartIdx, TEXT_LUMA, 0), "CBF should have been set\n");
+        }
+        else
+            codeQtCbf(cu, absPartIdx, TEXT_LUMA, cu->getTransformIdx(absPartIdx));
+
+        if (cbfY || cbfU || cbfV)
+        {
+            // dQP: only for LCU once
+            if (cu->m_slice->m_pps->bUseDQP)
+            {
+                if (bCodeDQP)
+                {
+                    codeDeltaQP(cu, state.bakAbsPartIdxCU);
+                    bCodeDQP = false;
+                }
+            }
+        }
+        if (cbfY)
+            codeCoeffNxN(cu, (cu->getCoeffY() + offsetLuma), absPartIdx, log2TrSize, TEXT_LUMA);
+
+        int chFmt = cu->getChromaFormat();
+        if ((log2TrSize == 2) && !(chFmt == X265_CSP_I444))
+        {
+            uint32_t partNum = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+            if ((absPartIdx & (partNum - 1)) == (partNum - 1))
+            {
+                const uint32_t log2TrSizeC = 2;
+                const bool splitIntoSubTUs = (chFmt == X265_CSP_I422);
+
+                uint32_t curPartNum = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+
+                for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+                {
+                    TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, curPartNum, state.bakAbsPartIdx);
+                    coeff_t* coeffChroma = cu->getCoeff((TextType)chromaId);
+                    do
+                    {
+                        uint32_t cbf = cu->getCbf(tuIterator.absPartIdxTURelCU, (TextType)chromaId, trIdx + splitIntoSubTUs);
+                        if (cbf)
+                        {
+                            uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+                            codeCoeffNxN(cu, (coeffChroma + state.bakChromaOffset + subTUOffset), tuIterator.absPartIdxTURelCU, log2TrSizeC, (TextType)chromaId);
+                        }
+                    }
+                    while (tuIterator.isNextSection());
+                }
+            }
+        }
+        else
+        {
+            uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+            const bool splitIntoSubTUs = (chFmt == X265_CSP_I422);
+            uint32_t curPartNum = NUM_CU_PARTITIONS >> (depth << 1);
+            for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+            {
+                TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, curPartNum, absPartIdx);
+                coeff_t* coeffChroma = cu->getCoeff((TextType)chromaId);
+                do
+                {
+                    uint32_t cbf = cu->getCbf(tuIterator.absPartIdxTURelCU, (TextType)chromaId, trIdx + splitIntoSubTUs);
+                    if (cbf)
+                    {
+                        uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+                        codeCoeffNxN(cu, (coeffChroma + offsetChroma + subTUOffset), tuIterator.absPartIdxTURelCU, log2TrSizeC, (TextType)chromaId);
+                    }
+                }
+                while (tuIterator.isNextSection());
+            }
+        }
+    }
+}
+
+void Entropy::codePredInfo(TComDataCU* cu, uint32_t absPartIdx)
+{
+    if (cu->isIntra(absPartIdx)) // If it is intra mode, encode intra prediction mode.
+    {
+        codeIntraDirLumaAng(cu, absPartIdx, true);
+        int chFmt = cu->getChromaFormat();
+        if (chFmt != X265_CSP_I400)
+        {
+            codeIntraDirChroma(cu, absPartIdx);
+
+            if ((chFmt == X265_CSP_I444) && (cu->getPartitionSize(absPartIdx) == SIZE_NxN))
+            {
+                uint32_t partOffset = (NUM_CU_PARTITIONS >> (cu->getDepth(absPartIdx) << 1)) >> 2;
+                codeIntraDirChroma(cu, absPartIdx + partOffset);
+                codeIntraDirChroma(cu, absPartIdx + partOffset * 2);
+                codeIntraDirChroma(cu, absPartIdx + partOffset * 3);
+            }
+        }
+    }
+    else                        // if it is inter mode, encode motion vector and reference index
+    {
+        codePUWise(cu, absPartIdx);
+    }
+}
+
+/** encode motion information for every PU block */
+void Entropy::codePUWise(TComDataCU* cu, uint32_t absPartIdx)
+{
+    PartSize partSize = cu->getPartitionSize(absPartIdx);
+    uint32_t numPU = (partSize == SIZE_2Nx2N ? 1 : (partSize == SIZE_NxN ? 4 : 2));
+    uint32_t depth = cu->getDepth(absPartIdx);
+    uint32_t puOffset = (g_puOffset[uint32_t(partSize)] << (g_maxFullDepth - depth) * 2) >> 4;
+
+    for (uint32_t partIdx = 0, subPartIdx = absPartIdx; partIdx < numPU; partIdx++, subPartIdx += puOffset)
+    {
+        codeMergeFlag(cu, subPartIdx);
+        if (cu->getMergeFlag(subPartIdx))
+            codeMergeIndex(cu, subPartIdx);
+        else
+        {
+            if (cu->m_slice->isInterB())
+                codeInterDir(cu, subPartIdx);
+
+            uint32_t interDir = cu->getInterDir(subPartIdx);
+            for (uint32_t list = 0; list < 2; list++)
+            {
+                if (interDir & (1 << list))
+                {
+                    X265_CHECK(cu->m_slice->m_numRefIdx[list] > 0, "numRefs should have been > 0\n");
+
+                    codeRefFrmIdxPU(cu, subPartIdx, list);
+                    codeMvd(cu, subPartIdx, list);
+                    codeMVPIdx(cu->getMVPIdx(list, subPartIdx));
+                }
+            }
+        }
+    }
+}
+
+/** encode reference frame index for a PU block */
+void Entropy::codeRefFrmIdxPU(TComDataCU* cu, uint32_t absPartIdx, int list)
+{
+    X265_CHECK(!cu->isIntra(absPartIdx), "intra block not expected\n");
+
+    if (cu->m_slice->m_numRefIdx[list] > 1)
+        codeRefFrmIdx(cu, absPartIdx, list);
+}
+
+void Entropy::codeCoeff(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool& bCodeDQP, uint32_t* depthRange)
+{
+    if (!cu->isIntra(absPartIdx))
+    {
+        if (!(cu->getMergeFlag(absPartIdx) && cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N))
+            codeQtRootCbf(cu->getQtRootCbf(absPartIdx));
+        if (!cu->getQtRootCbf(absPartIdx))
+            return;
+    }
+
+    uint32_t log2CUSize   = cu->getLog2CUSize(absPartIdx);
+    uint32_t lumaOffset   = absPartIdx << (LOG2_UNIT_SIZE * 2);
+    uint32_t chromaOffset = lumaOffset >> (cu->getHorzChromaShift() + cu->getVertChromaShift());
+    uint32_t absPartIdxStep = NUM_CU_PARTITIONS >> (depth << 1);
+    CoeffCodeState state;
+    encodeTransform(cu, state, lumaOffset, chromaOffset, absPartIdx, absPartIdxStep, depth, log2CUSize, 0, bCodeDQP, depthRange);
+}
+
+void Entropy::codeSaoOffset(SaoLcuParam* saoLcuParam, uint32_t compIdx)
+{
+    uint32_t symbol;
+    int i;
+
+    symbol = saoLcuParam->typeIdx + 1;
+    if (compIdx != 2)
+        codeSaoTypeIdx(symbol);
+
+    if (symbol)
+    {
+        if (saoLcuParam->typeIdx < SAO_BO && compIdx != 2)
+            saoLcuParam->subTypeIdx = saoLcuParam->typeIdx;
+
+        int offsetTh = 1 << X265_MIN(X265_DEPTH - 5, 5);
+        if (saoLcuParam->typeIdx == SAO_BO)
+        {
+            for (i = 0; i < SAO_BO_LEN; i++)
+            {
+                uint32_t absOffset = ((saoLcuParam->offset[i] < 0) ? -saoLcuParam->offset[i] : saoLcuParam->offset[i]);
+                codeSaoMaxUvlc(absOffset, offsetTh - 1);
+            }
+
+            for (i = 0; i < SAO_BO_LEN; i++)
+            {
+                if (saoLcuParam->offset[i] != 0)
+                {
+                    uint32_t sign = (saoLcuParam->offset[i] < 0) ? 1 : 0;
+                    codeSAOSign(sign);
+                }
+            }
+
+            symbol = (uint32_t)(saoLcuParam->subTypeIdx);
+            codeSaoUflc(5, symbol);
+        }
+        else // if (saoLcuParam->typeIdx < SAO_BO)
+        {
+            codeSaoMaxUvlc(saoLcuParam->offset[0], offsetTh - 1);
+            codeSaoMaxUvlc(saoLcuParam->offset[1], offsetTh - 1);
+            codeSaoMaxUvlc(-saoLcuParam->offset[2], offsetTh - 1);
+            codeSaoMaxUvlc(-saoLcuParam->offset[3], offsetTh - 1);
+            if (compIdx != 2)
+            {
+                symbol = (uint32_t)(saoLcuParam->subTypeIdx);
+                codeSaoUflc(2, symbol);
+            }
+        }
+    }
+}
+
+void Entropy::codeSaoUnitInterleaving(int compIdx, bool saoFlag, int rx, int ry, SaoLcuParam* saoLcuParam, int cuAddrInSlice, int cuAddrUpInSlice, int allowMergeLeft, int allowMergeUp)
+{
+    if (saoFlag)
+    {
+        if (rx > 0 && cuAddrInSlice != 0 && allowMergeLeft)
+            codeSaoMerge(saoLcuParam->mergeLeftFlag);
+        else
+            saoLcuParam->mergeLeftFlag = 0;
+
+        if (!saoLcuParam->mergeLeftFlag)
+        {
+            if ((ry > 0) && (cuAddrUpInSlice >= 0) && allowMergeUp)
+                codeSaoMerge(saoLcuParam->mergeUpFlag);
+            else
+                saoLcuParam->mergeUpFlag = 0;
+
+            if (!saoLcuParam->mergeUpFlag)
+                codeSaoOffset(saoLcuParam, compIdx);
+        }
+    }
+}
+
+/** initialize context model with respect to QP and initialization value */
+uint8_t sbacInit(int qp, int initValue)
+{
+    qp = Clip3(0, 51, qp);
+
+    int  slope      = (initValue >> 4) * 5 - 45;
+    int  offset     = ((initValue & 15) << 3) - 16;
+    int  initState  =  X265_MIN(X265_MAX(1, (((slope * qp) >> 4) + offset)), 126);
+    uint32_t mpState = (initState >= 64);
+    uint32_t state = ((mpState ? (initState - 64) : (63 - initState)) << 1) + mpState;
+
+    return (uint8_t)state;
+}
+
+static void initBuffer(uint8_t* contextModel, SliceType sliceType, int qp, uint8_t* ctxModel, int size)
+{
+    ctxModel += sliceType * size;
+
+    for (int n = 0; n < size; n++)
+        contextModel[n] = sbacInit(qp, ctxModel[n]);
+}
+
+void Entropy::resetEntropy(Slice *slice)
+{
+    int  qp              = slice->m_sliceQp;
+    SliceType sliceType  = slice->m_sliceType;
+
+    initBuffer(&m_contextState[OFF_SPLIT_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_SPLIT_FLAG, NUM_SPLIT_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_SKIP_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_SKIP_FLAG, NUM_SKIP_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_MERGE_FLAG_EXT_CTX], sliceType, qp, (uint8_t*)INIT_MERGE_FLAG_EXT, NUM_MERGE_FLAG_EXT_CTX);
+    initBuffer(&m_contextState[OFF_MERGE_IDX_EXT_CTX], sliceType, qp, (uint8_t*)INIT_MERGE_IDX_EXT, NUM_MERGE_IDX_EXT_CTX);
+    initBuffer(&m_contextState[OFF_PART_SIZE_CTX], sliceType, qp, (uint8_t*)INIT_PART_SIZE, NUM_PART_SIZE_CTX);
+    initBuffer(&m_contextState[OFF_PRED_MODE_CTX], sliceType, qp, (uint8_t*)INIT_PRED_MODE, NUM_PRED_MODE_CTX);
+    initBuffer(&m_contextState[OFF_ADI_CTX], sliceType, qp, (uint8_t*)INIT_INTRA_PRED_MODE, NUM_ADI_CTX);
+    initBuffer(&m_contextState[OFF_CHROMA_PRED_CTX], sliceType, qp, (uint8_t*)INIT_CHROMA_PRED_MODE, NUM_CHROMA_PRED_CTX);
+    initBuffer(&m_contextState[OFF_DELTA_QP_CTX], sliceType, qp, (uint8_t*)INIT_DQP, NUM_DELTA_QP_CTX);
+    initBuffer(&m_contextState[OFF_INTER_DIR_CTX], sliceType, qp, (uint8_t*)INIT_INTER_DIR, NUM_INTER_DIR_CTX);
+    initBuffer(&m_contextState[OFF_REF_NO_CTX], sliceType, qp, (uint8_t*)INIT_REF_PIC, NUM_REF_NO_CTX);
+    initBuffer(&m_contextState[OFF_MV_RES_CTX], sliceType, qp, (uint8_t*)INIT_MVD, NUM_MV_RES_CTX);
+    initBuffer(&m_contextState[OFF_QT_CBF_CTX], sliceType, qp, (uint8_t*)INIT_QT_CBF, NUM_QT_CBF_CTX);
+    initBuffer(&m_contextState[OFF_TRANS_SUBDIV_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_TRANS_SUBDIV_FLAG, NUM_TRANS_SUBDIV_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_QT_ROOT_CBF_CTX], sliceType, qp, (uint8_t*)INIT_QT_ROOT_CBF, NUM_QT_ROOT_CBF_CTX);
+    initBuffer(&m_contextState[OFF_SIG_CG_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_SIG_CG_FLAG, 2 * NUM_SIG_CG_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_SIG_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_SIG_FLAG, NUM_SIG_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_CTX_LAST_FLAG_X], sliceType, qp, (uint8_t*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
+    initBuffer(&m_contextState[OFF_CTX_LAST_FLAG_Y], sliceType, qp, (uint8_t*)INIT_LAST, NUM_CTX_LAST_FLAG_XY);
+    initBuffer(&m_contextState[OFF_ONE_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_ONE_FLAG, NUM_ONE_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_ABS_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_ABS_FLAG, NUM_ABS_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_MVP_IDX_CTX], sliceType, qp, (uint8_t*)INIT_MVP_IDX, NUM_MVP_IDX_CTX);
+    initBuffer(&m_contextState[OFF_SAO_MERGE_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_SAO_MERGE_FLAG, NUM_SAO_MERGE_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_SAO_TYPE_IDX_CTX], sliceType, qp, (uint8_t*)INIT_SAO_TYPE_IDX, NUM_SAO_TYPE_IDX_CTX);
+    initBuffer(&m_contextState[OFF_TRANSFORMSKIP_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_TRANSFORMSKIP_FLAG, 2 * NUM_TRANSFORMSKIP_FLAG_CTX);
+    initBuffer(&m_contextState[OFF_CU_TRANSQUANT_BYPASS_FLAG_CTX], sliceType, qp, (uint8_t*)INIT_CU_TRANSQUANT_BYPASS_FLAG, NUM_CU_TRANSQUANT_BYPASS_FLAG_CTX);
+    // new structure
+
+    start();
+}
+
+/* code explicit wp tables */
+void Entropy::codePredWeightTable(Slice* slice)
+{
+    WeightParam *wp;
+    bool            bChroma      = true; // 4:0:0 not yet supported
+    bool            bDenomCoded  = false;
+    int             numRefDirs   = slice->m_sliceType == B_SLICE ? 2 : 1;
+    uint32_t        totalSignalledWeightFlags = 0;
+
+    if ((slice->m_sliceType == P_SLICE && slice->m_pps->bUseWeightPred) ||
+        (slice->m_sliceType == B_SLICE && slice->m_pps->bUseWeightedBiPred))
+    {
+        for (int list = 0; list < numRefDirs; list++)
+        {
+            for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
+            {
+                wp = slice->m_weightPredTable[list][ref];
+                if (!bDenomCoded)
+                {
+                    WRITE_UVLC(wp[0].log2WeightDenom, "luma_log2_weight_denom");
+
+                    if (bChroma)
+                    {
+                        int deltaDenom = wp[1].log2WeightDenom - wp[0].log2WeightDenom;
+                        WRITE_SVLC(deltaDenom, "delta_chroma_log2_weight_denom");
+                    }
+                    bDenomCoded = true;
+                }
+                WRITE_FLAG(wp[0].bPresentFlag, "luma_weight_lX_flag");
+                totalSignalledWeightFlags += wp[0].bPresentFlag;
+            }
+
+            if (bChroma)
+            {
+                for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
+                {
+                    wp = slice->m_weightPredTable[list][ref];
+                    WRITE_FLAG(wp[1].bPresentFlag, "chroma_weight_lX_flag");
+                    totalSignalledWeightFlags += 2 * wp[1].bPresentFlag;
+                }
+            }
+
+            for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
+            {
+                wp = slice->m_weightPredTable[list][ref];
+                if (wp[0].bPresentFlag)
+                {
+                    int deltaWeight = (wp[0].inputWeight - (1 << wp[0].log2WeightDenom));
+                    WRITE_SVLC(deltaWeight, "delta_luma_weight_lX");
+                    WRITE_SVLC(wp[0].inputOffset, "luma_offset_lX");
+                }
+
+                if (bChroma)
+                {
+                    if (wp[1].bPresentFlag)
+                    {
+                        for (int plane = 1; plane < 3; plane++)
+                        {
+                            int deltaWeight = (wp[plane].inputWeight - (1 << wp[1].log2WeightDenom));
+                            WRITE_SVLC(deltaWeight, "delta_chroma_weight_lX");
+
+                            int pred = (128 - ((128 * wp[plane].inputWeight) >> (wp[plane].log2WeightDenom)));
+                            int deltaChroma = (wp[plane].inputOffset - pred);
+                            WRITE_SVLC(deltaChroma, "delta_chroma_offset_lX");
+                        }
+                    }
+                }
+            }
+        }
+
+        X265_CHECK(totalSignalledWeightFlags <= 24, "total weights must be <= 24\n");
+    }
+}
+
+void Entropy::writeUnaryMaxSymbol(uint32_t symbol, uint8_t* scmModel, int offset, uint32_t maxSymbol)
+{
+    X265_CHECK(maxSymbol > 0, "maxSymbol too small\n");
+
+    encodeBin(symbol ? 1 : 0, scmModel[0]);
+
+    if (!symbol)
+        return;
+
+    bool bCodeLast = (maxSymbol > symbol);
+
+    while (--symbol)
+        encodeBin(1, scmModel[offset]);
+
+    if (bCodeLast)
+        encodeBin(0, scmModel[offset]);
+}
+
+void Entropy::writeEpExGolomb(uint32_t symbol, uint32_t count)
+{
+    uint32_t bins = 0;
+    int numBins = 0;
+
+    while (symbol >= (uint32_t)(1 << count))
+    {
+        bins = 2 * bins + 1;
+        numBins++;
+        symbol -= 1 << count;
+        count++;
+    }
+
+    bins = 2 * bins + 0;
+    numBins++;
+
+    bins = (bins << count) | symbol;
+    numBins += count;
+
+    X265_CHECK(numBins <= 32, "numBins too large\n");
+    encodeBinsEP(bins, numBins);
+}
+
+/** Coding of coeff_abs_level_minus3 */
+void Entropy::writeCoefRemainExGolomb(uint32_t codeNumber, uint32_t absGoRice)
+{
+    uint32_t length;
+    const uint32_t codeRemain = codeNumber & ((1 << absGoRice) - 1);
+
+    if ((codeNumber >> absGoRice) < COEF_REMAIN_BIN_REDUCTION)
+    {
+        length = codeNumber >> absGoRice;
+
+        X265_CHECK(codeNumber - (length << absGoRice) == (codeNumber & ((1 << absGoRice) - 1)), "codeNumber failure\n");
+        X265_CHECK(length + 1 + absGoRice < 32, "length failure\n");
+        encodeBinsEP((((1 << (length + 1)) - 2) << absGoRice) + codeRemain, length + 1 + absGoRice);
+    }
+    else
+    {
+        length = 0;
+        codeNumber = (codeNumber >> absGoRice) - COEF_REMAIN_BIN_REDUCTION;
+        if (codeNumber != 0)
+        {
+            unsigned long idx;
+            CLZ32(idx, codeNumber + 1);
+            length = idx;
+            codeNumber -= (1 << idx) - 1;
+        }
+        codeNumber = (codeNumber << absGoRice) + codeRemain;
+
+        encodeBinsEP((1 << (COEF_REMAIN_BIN_REDUCTION + length + 1)) - 2, COEF_REMAIN_BIN_REDUCTION + length + 1);
+        encodeBinsEP(codeNumber, length + absGoRice);
+    }
+}
+
+// SBAC RD
+void  Entropy::loadIntraDirModeLuma(Entropy& src)
+{
+    copyState(src);
+
+    ::memcpy(&m_contextState[OFF_ADI_CTX], &src.m_contextState[OFF_ADI_CTX], sizeof(uint8_t) * NUM_ADI_CTX);
+}
+
+void Entropy::copyFrom(Entropy& src)
+{
+    copyState(src);
+
+    memcpy(m_contextState, src.m_contextState, MAX_OFF_CTX_MOD * sizeof(uint8_t));
+}
+
+void Entropy::codeMVPIdx(uint32_t symbol)
+{
+    encodeBin(symbol, m_contextState[OFF_MVP_IDX_CTX]);
+}
+
+void Entropy::codePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
+{
+    PartSize partSize = cu->getPartitionSize(absPartIdx);
+
+    if (cu->isIntra(absPartIdx))
+    {
+        if (depth == g_maxCUDepth)
+            encodeBin(partSize == SIZE_2Nx2N ? 1 : 0, m_contextState[OFF_PART_SIZE_CTX]);
+        return;
+    }
+
+    switch (partSize)
+    {
+    case SIZE_2Nx2N:
+        encodeBin(1, m_contextState[OFF_PART_SIZE_CTX]);
+        break;
+
+    case SIZE_2NxN:
+    case SIZE_2NxnU:
+    case SIZE_2NxnD:
+        encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 0]);
+        encodeBin(1, m_contextState[OFF_PART_SIZE_CTX + 1]);
+        if (cu->m_slice->m_sps->maxAMPDepth > depth)
+        {
+            encodeBin((partSize == SIZE_2NxN) ? 1 : 0, m_contextState[OFF_PART_SIZE_CTX + 3]);
+            if (partSize != SIZE_2NxN)
+                encodeBinEP((partSize == SIZE_2NxnU ? 0 : 1));
+        }
+        break;
+
+    case SIZE_Nx2N:
+    case SIZE_nLx2N:
+    case SIZE_nRx2N:
+        encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 0]);
+        encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 1]);
+        if (depth == g_maxCUDepth && !(cu->getLog2CUSize(absPartIdx) == 3))
+            encodeBin(1, m_contextState[OFF_PART_SIZE_CTX + 2]);
+        if (cu->m_slice->m_sps->maxAMPDepth > depth)
+        {
+            encodeBin((partSize == SIZE_Nx2N) ? 1 : 0, m_contextState[OFF_PART_SIZE_CTX + 3]);
+            if (partSize != SIZE_Nx2N)
+                encodeBinEP((partSize == SIZE_nLx2N ? 0 : 1));
+        }
+        break;
+
+    case SIZE_NxN:
+        if (depth == g_maxCUDepth && !(cu->getLog2CUSize(absPartIdx) == 3))
+        {
+            encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 0]);
+            encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 1]);
+            encodeBin(0, m_contextState[OFF_PART_SIZE_CTX + 2]);
+        }
+        break;
+
+    default:
+        X265_CHECK(0, "invalid CU partition\n");
+        break;
+    }
+}
+
+void Entropy::codePredMode(int predMode)
+{
+    encodeBin(predMode == MODE_INTER ? 0 : 1, m_contextState[OFF_PRED_MODE_CTX]);
+}
+
+void Entropy::codeCUTransquantBypassFlag(uint32_t symbol)
+{
+    encodeBin(symbol, m_contextState[OFF_CU_TRANSQUANT_BYPASS_FLAG_CTX]);
+}
+
+void Entropy::codeSkipFlag(TComDataCU* cu, uint32_t absPartIdx)
+{
+    // get context function is here
+    uint32_t symbol = cu->isSkipped(absPartIdx) ? 1 : 0;
+    uint32_t ctxSkip = cu->getCtxSkipFlag(absPartIdx);
+
+    encodeBin(symbol, m_contextState[OFF_SKIP_FLAG_CTX + ctxSkip]);
+}
+
+void Entropy::codeMergeFlag(TComDataCU* cu, uint32_t absPartIdx)
+{
+    const uint32_t symbol = cu->getMergeFlag(absPartIdx) ? 1 : 0;
+
+    encodeBin(symbol, m_contextState[OFF_MERGE_FLAG_EXT_CTX]);
+}
+
+void Entropy::codeMergeIndex(TComDataCU* cu, uint32_t absPartIdx)
+{
+    uint32_t numCand = cu->m_slice->m_maxNumMergeCand;
+
+    if (numCand > 1)
+    {
+        uint32_t unaryIdx = cu->getMergeIndex(absPartIdx);
+        encodeBin((unaryIdx != 0), m_contextState[OFF_MERGE_IDX_EXT_CTX]);
+
+        X265_CHECK(unaryIdx < numCand, "unaryIdx out of range\n");
+
+        if (unaryIdx != 0)
+        {
+            uint32_t mask = (1 << unaryIdx) - 2;
+            mask >>= (unaryIdx == numCand - 1) ? 1 : 0;
+            encodeBinsEP(mask, unaryIdx - (unaryIdx == numCand - 1));
+        }
+    }
+}
+
+void Entropy::codeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth)
+{
+    X265_CHECK(depth < g_maxCUDepth, "invalid depth\n");
+
+    uint32_t ctx           = cu->getCtxSplitFlag(absPartIdx, depth);
+    uint32_t currSplitFlag = (cu->getDepth(absPartIdx) > depth) ? 1 : 0;
+
+    X265_CHECK(ctx < 3, "ctx out of range\n");
+    encodeBin(currSplitFlag, m_contextState[OFF_SPLIT_FLAG_CTX + ctx]);
+}
+
+void Entropy::codeTransformSubdivFlag(uint32_t symbol, uint32_t ctx)
+{
+    encodeBin(symbol, m_contextState[OFF_TRANS_SUBDIV_FLAG_CTX + ctx]);
+}
+
+void Entropy::codeIntraDirLumaAng(TComDataCU* cu, uint32_t absPartIdx, bool isMultiple)
+{
+    uint32_t dir[4], j;
+    uint32_t preds[4][3];
+    int predIdx[4];
+    PartSize mode = cu->getPartitionSize(absPartIdx);
+    uint32_t partNum = isMultiple ? (mode == SIZE_NxN ? 4 : 1) : 1;
+    uint32_t partOffset = (NUM_CU_PARTITIONS >> (cu->getDepth(absPartIdx) << 1)) >> 2;
+
+    for (j = 0; j < partNum; j++)
+    {
+        dir[j] = cu->getLumaIntraDir(absPartIdx + partOffset * j);
+        cu->getIntraDirLumaPredictor(absPartIdx + partOffset * j, preds[j]);
+        predIdx[j] = -1;
+        for (uint32_t i = 0; i < 3; i++)
+            if (dir[j] == preds[j][i])
+                predIdx[j] = i;
+
+        encodeBin((predIdx[j] != -1) ? 1 : 0, m_contextState[OFF_ADI_CTX]);
+    }
+
+    for (j = 0; j < partNum; j++)
+    {
+        if (predIdx[j] != -1)
+        {
+            X265_CHECK((predIdx[j] >= 0) && (predIdx[j] <= 2), "predIdx out of range\n");
+            // NOTE: Mapping
+            //       0 = 0
+            //       1 = 10
+            //       2 = 11
+            int nonzero = (!!predIdx[j]);
+            encodeBinsEP(predIdx[j] + nonzero, 1 + nonzero);
+        }
+        else
+        {
+            if (preds[j][0] > preds[j][1])
+                std::swap(preds[j][0], preds[j][1]);
+
+            if (preds[j][0] > preds[j][2])
+                std::swap(preds[j][0], preds[j][2]);
+
+            if (preds[j][1] > preds[j][2])
+                std::swap(preds[j][1], preds[j][2]);
+
+            dir[j] += (dir[j] > preds[j][2]) ? -1 : 0;
+            dir[j] += (dir[j] > preds[j][1]) ? -1 : 0;
+            dir[j] += (dir[j] > preds[j][0]) ? -1 : 0;
+
+            encodeBinsEP(dir[j], 5);
+        }
+    }
+}
+
+void Entropy::codeIntraDirChroma(TComDataCU* cu, uint32_t absPartIdx)
+{
+    uint32_t intraDirChroma = cu->getChromaIntraDir(absPartIdx);
+
+    if (intraDirChroma == DM_CHROMA_IDX)
+        encodeBin(0, m_contextState[OFF_CHROMA_PRED_CTX]);
+    else
+    {
+        uint32_t allowedChromaDir[NUM_CHROMA_MODE];
+        cu->getAllowedChromaDir(absPartIdx, allowedChromaDir);
+
+        for (int i = 0; i < NUM_CHROMA_MODE - 1; i++)
+        {
+            if (intraDirChroma == allowedChromaDir[i])
+            {
+                intraDirChroma = i;
+                break;
+            }
+        }
+
+        encodeBin(1, m_contextState[OFF_CHROMA_PRED_CTX]);
+        encodeBinsEP(intraDirChroma, 2);
+    }
+}
+
+void Entropy::codeInterDir(TComDataCU* cu, uint32_t absPartIdx)
+{
+    const uint32_t interDir = cu->getInterDir(absPartIdx) - 1;
+    const uint32_t ctx      = cu->getCtxInterDir(absPartIdx);
+
+    if (cu->getPartitionSize(absPartIdx) == SIZE_2Nx2N || cu->getLog2CUSize(absPartIdx) != 3)
+        encodeBin(interDir == 2 ? 1 : 0, m_contextState[OFF_INTER_DIR_CTX + ctx]);
+    if (interDir < 2)
+        encodeBin(interDir, m_contextState[OFF_INTER_DIR_CTX + 4]);
+}
+
+void Entropy::codeRefFrmIdx(TComDataCU* cu, uint32_t absPartIdx, int list)
+{
+    uint32_t refFrame = cu->getCUMvField(list)->getRefIdx(absPartIdx);
+
+    encodeBin(refFrame > 0, m_contextState[OFF_REF_NO_CTX]);
+
+    if (refFrame > 0)
+    {
+        uint32_t refNum = cu->m_slice->m_numRefIdx[list] - 2;
+        if (refNum == 0)
+            return;
+
+        refFrame--;
+        encodeBin(refFrame > 0, m_contextState[OFF_REF_NO_CTX + 1]);
+        if (refFrame > 0)
+        {
+            uint32_t mask = (1 << refFrame) - 2;
+            mask >>= (refFrame == refNum) ? 1 : 0;
+            encodeBinsEP(mask, refFrame - (refFrame == refNum));
+        }
+    }
+}
+
+void Entropy::codeMvd(TComDataCU* cu, uint32_t absPartIdx, int list)
+{
+    const TComCUMvField* cuMvField = cu->getCUMvField(list);
+    const int hor = cuMvField->getMvd(absPartIdx).x;
+    const int ver = cuMvField->getMvd(absPartIdx).y;
+
+    encodeBin(hor != 0 ? 1 : 0, m_contextState[OFF_MV_RES_CTX]);
+    encodeBin(ver != 0 ? 1 : 0, m_contextState[OFF_MV_RES_CTX]);
+
+    const bool bHorAbsGr0 = hor != 0;
+    const bool bVerAbsGr0 = ver != 0;
+    const uint32_t horAbs   = 0 > hor ? -hor : hor;
+    const uint32_t verAbs   = 0 > ver ? -ver : ver;
+
+    if (bHorAbsGr0)
+        encodeBin(horAbs > 1 ? 1 : 0, m_contextState[OFF_MV_RES_CTX + 1]);
+
+    if (bVerAbsGr0)
+        encodeBin(verAbs > 1 ? 1 : 0, m_contextState[OFF_MV_RES_CTX + 1]);
+
+    if (bHorAbsGr0)
+    {
+        if (horAbs > 1)
+            writeEpExGolomb(horAbs - 2, 1);
+
+        encodeBinEP(0 > hor ? 1 : 0);
+    }
+
+    if (bVerAbsGr0)
+    {
+        if (verAbs > 1)
+            writeEpExGolomb(verAbs - 2, 1);
+
+        encodeBinEP(0 > ver ? 1 : 0);
+    }
+}
+
+void Entropy::codeDeltaQP(TComDataCU* cu, uint32_t absPartIdx)
+{
+    int dqp = cu->getQP(absPartIdx) - cu->getRefQP(absPartIdx);
+
+    int qpBdOffsetY = QP_BD_OFFSET;
+
+    dqp = (dqp + 78 + qpBdOffsetY + (qpBdOffsetY / 2)) % (52 + qpBdOffsetY) - 26 - (qpBdOffsetY / 2);
+
+    uint32_t absDQp = (uint32_t)((dqp > 0) ? dqp  : (-dqp));
+    uint32_t TUValue = X265_MIN((int)absDQp, CU_DQP_TU_CMAX);
+    writeUnaryMaxSymbol(TUValue, &m_contextState[OFF_DELTA_QP_CTX], 1, CU_DQP_TU_CMAX);
+    if (absDQp >= CU_DQP_TU_CMAX)
+        writeEpExGolomb(absDQp - CU_DQP_TU_CMAX, CU_DQP_EG_k);
+
+    if (absDQp > 0)
+    {
+        uint32_t sign = (dqp > 0 ? 0 : 1);
+        encodeBinEP(sign);
+    }
+}
+
+void Entropy::codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, uint32_t absPartIdxStep, uint32_t width, uint32_t height, TextType ttype, uint32_t trDepth, bool lowestLevel)
+{
+    uint32_t ctx = ctxCbf[ttype][trDepth];
+
+    bool canQuadSplit       = (width >= (MIN_TU_SIZE * 2)) && (height >= (MIN_TU_SIZE * 2));
+    uint32_t lowestTUDepth  = trDepth + ((!lowestLevel && !canQuadSplit) ? 1 : 0); // unsplittable TUs inherit their parent's CBF
+
+    if ((width != height) && (lowestLevel || !canQuadSplit)) // if sub-TUs are present
+    {
+        uint32_t subTUDepth        = lowestTUDepth + 1;   // if this is the lowest level of the TU-tree, the sub-TUs are directly below.
+                                                          // Otherwise, this must be the level above the lowest level (as specified above)
+        uint32_t partIdxesPerSubTU = absPartIdxStep >> 1;
+
+        for (uint32_t subTU = 0; subTU < 2; subTU++)
+        {
+            uint32_t subTUAbsPartIdx = absPartIdx + (subTU * partIdxesPerSubTU);
+            uint32_t cbf = cu->getCbf(subTUAbsPartIdx, ttype, subTUDepth);
+
+            encodeBin(cbf, m_contextState[OFF_QT_CBF_CTX + ctx]);
+        }
+    }
+    else
+    {
+        uint32_t cbf = cu->getCbf(absPartIdx, ttype, lowestTUDepth);
+
+        encodeBin(cbf, m_contextState[OFF_QT_CBF_CTX + ctx]);
+    }
+}
+
+void Entropy::codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth)
+{
+    uint32_t ctx = ctxCbf[ttype][trDepth];
+    uint32_t cbf = cu->getCbf(absPartIdx, ttype, trDepth);
+    encodeBin(cbf, m_contextState[OFF_QT_CBF_CTX + ctx]);
+}
+
+void Entropy::codeTransformSkipFlags(TComDataCU* cu, uint32_t absPartIdx, uint32_t trSize, TextType ttype)
+{
+    if (cu->getCUTransquantBypass(absPartIdx))
+        return;
+    if (trSize != 4)
+        return;
+
+    uint32_t useTransformSkip = cu->getTransformSkip(absPartIdx, ttype);
+    encodeBin(useTransformSkip, m_contextState[OFF_TRANSFORMSKIP_FLAG_CTX + (ttype ? NUM_TRANSFORMSKIP_FLAG_CTX : 0)]);
+}
+
+void Entropy::codeQtRootCbf(uint32_t cbf)
+{
+    encodeBin(cbf, m_contextState[OFF_QT_ROOT_CBF_CTX]);
+}
+
+void Entropy::codeQtCbfZero(TextType ttype, uint32_t trDepth)
+{
+    // this function is only used to estimate the bits when cbf is 0
+    // and will never be called when writing the bistream. do not need to write log
+    uint32_t ctx = ctxCbf[ttype][trDepth];
+    encodeBin(0, m_contextState[OFF_QT_CBF_CTX + ctx]);
+}
+
+void Entropy::codeQtRootCbfZero()
+{
+    // this function is only used to estimate the bits when cbf is 0
+    // and will never be called when writing the bistream. do not need to write log
+    encodeBin(0, m_contextState[OFF_QT_ROOT_CBF_CTX]);
+}
+
+/** Encode (X,Y) position of the last significant coefficient
+ * \param posx X component of last coefficient
+ * \param posy Y component of last coefficient
+ * \param log2TrSize
+ * \param bIsLuma
+ * \param scanIdx scan type (zig-zag, hor, ver)
+ * This method encodes the X and Y component within a block of the last significant coefficient.
+ */
+void Entropy::codeLastSignificantXY(uint32_t posx, uint32_t posy, uint32_t log2TrSize, bool bIsLuma, uint32_t scanIdx)
+{
+    // swap
+    if (scanIdx == SCAN_VER)
+        std::swap(posx, posy);
+
+    uint32_t ctxLast;
+    uint32_t groupIdxX = getGroupIdx(posx);
+    uint32_t groupIdxY = getGroupIdx(posy);
+
+    int blkSizeOffset = bIsLuma ? ((log2TrSize - 2) * 3 + ((log2TrSize - 1) >> 2)) : NUM_CTX_LAST_FLAG_XY_LUMA;
+    int ctxShift = bIsLuma ? ((log2TrSize + 1) >> 2) : log2TrSize - 2;
+    uint32_t maxGroupIdx = log2TrSize * 2 - 1;
+
+    // posX
+    uint8_t *ctxX = &m_contextState[OFF_CTX_LAST_FLAG_X];
+    for (ctxLast = 0; ctxLast < groupIdxX; ctxLast++)
+        encodeBin(1, *(ctxX + blkSizeOffset + (ctxLast >> ctxShift)));
+
+    if (groupIdxX < maxGroupIdx)
+        encodeBin(0, *(ctxX + blkSizeOffset + (ctxLast >> ctxShift)));
+
+    // posY
+    uint8_t *ctxY = &m_contextState[OFF_CTX_LAST_FLAG_Y];
+    for (ctxLast = 0; ctxLast < groupIdxY; ctxLast++)
+        encodeBin(1, *(ctxY + blkSizeOffset + (ctxLast >> ctxShift)));
+
+    if (groupIdxY < maxGroupIdx)
+        encodeBin(0, *(ctxY + blkSizeOffset + (ctxLast >> ctxShift)));
+
+    if (groupIdxX > 3)
+    {
+        uint32_t count = (groupIdxX - 2) >> 1;
+        posx = posx - g_minInGroup[groupIdxX];
+        encodeBinsEP(posx, count);
+    }
+    if (groupIdxY > 3)
+    {
+        uint32_t count = (groupIdxY - 2) >> 1;
+        posy = posy - g_minInGroup[groupIdxY];
+        encodeBinsEP(posy, count);
+    }
+}
+
+void Entropy::codeCoeffNxN(TComDataCU* cu, coeff_t* coeff, uint32_t absPartIdx, uint32_t log2TrSize, TextType ttype)
+{
+    uint32_t trSize = 1 << log2TrSize;
+
+    // compute number of significant coefficients
+    uint32_t numSig = primitives.count_nonzero(coeff, (1 << (log2TrSize << 1)));
+
+    X265_CHECK(numSig > 0, "cbf check fail\n");
+
+    bool bHideFirstSign = cu->m_slice->m_pps->bSignHideEnabled && !cu->getCUTransquantBypass(absPartIdx);
+
+    if (cu->m_slice->m_pps->bTransformSkipEnabled)
+        codeTransformSkipFlags(cu, absPartIdx, trSize, ttype);
+    
+    bool bIsLuma = ttype == TEXT_LUMA;
+
+    // select scans
+    TUEntropyCodingParameters codingParameters;
+    cu->getTUEntropyCodingParameters(codingParameters, absPartIdx, log2TrSize, bIsLuma);
+
+    //----- encode significance map -----
+
+    // Find position of last coefficient
+    int scanPosLast = 0;
+    uint32_t posLast;
+    uint64_t sigCoeffGroupFlag64 = 0;
+    const uint32_t maskPosXY = ((uint32_t)~0 >> (31 - log2TrSize + MLS_CG_LOG2_SIZE)) >> 1;
+    assert((uint32_t)((1 << (log2TrSize - MLS_CG_LOG2_SIZE)) - 1) == (((uint32_t)~0 >> (31 - log2TrSize + MLS_CG_LOG2_SIZE)) >> 1));
+    do
+    {
+        posLast = codingParameters.scan[scanPosLast++];
+
+        const uint32_t isNZCoeff = (coeff[posLast] != 0);
+        // get L1 sig map
+        // NOTE: the new algorithm is complicated, so I keep reference code here
+        //uint32_t posy   = posLast >> log2TrSize;
+        //uint32_t posx   = posLast - (posy << log2TrSize);
+        //uint32_t blkIdx0 = ((posy >> MLS_CG_LOG2_SIZE) << codingParameters.log2TrSizeCG) + (posx >> MLS_CG_LOG2_SIZE);
+        const uint32_t blkIdx = ((posLast >> (2 * MLS_CG_LOG2_SIZE)) & ~maskPosXY) + ((posLast >> MLS_CG_LOG2_SIZE) & maskPosXY);
+        sigCoeffGroupFlag64 |= ((uint64_t)isNZCoeff << blkIdx);
+        numSig -= isNZCoeff;
+    }
+    while (numSig > 0);
+    scanPosLast--;
+
+    // Code position of last coefficient
+    int posLastY = posLast >> log2TrSize;
+    int posLastX = posLast & (trSize - 1);
+    codeLastSignificantXY(posLastX, posLastY, log2TrSize, bIsLuma, codingParameters.scanType);
+
+    //===== code significance flag =====
+    uint8_t * const baseCoeffGroupCtx = &m_contextState[OFF_SIG_CG_FLAG_CTX + (bIsLuma ? 0 : NUM_SIG_CG_FLAG_CTX)];
+    uint8_t * const baseCtx = bIsLuma ? &m_contextState[OFF_SIG_FLAG_CTX] : &m_contextState[OFF_SIG_FLAG_CTX + NUM_SIG_FLAG_CTX_LUMA];
+    const int lastScanSet = scanPosLast >> MLS_CG_SIZE;
+    uint32_t c1 = 1;
+    uint32_t goRiceParam = 0;
+    int scanPosSig = scanPosLast;
+
+    for (int subSet = lastScanSet; subSet >= 0; subSet--)
+    {
+        int numNonZero = 0;
+        int subPos     = subSet << MLS_CG_SIZE;
+        goRiceParam    = 0;
+        int absCoeff[1 << MLS_CG_SIZE];
+        uint32_t coeffSigns = 0;
+        int lastNZPosInCG = -1;
+        int firstNZPosInCG = 1 << MLS_CG_SIZE;
+        if (scanPosSig == scanPosLast)
+        {
+            absCoeff[0] = int(abs(coeff[posLast]));
+            coeffSigns  = (coeff[posLast] < 0);
+            numNonZero  = 1;
+            lastNZPosInCG  = scanPosSig;
+            firstNZPosInCG = scanPosSig;
+            scanPosSig--;
+        }
+        // encode significant_coeffgroup_flag
+        const int cgBlkPos = codingParameters.scanCG[subSet];
+        const int cgPosY   = cgBlkPos >> codingParameters.log2TrSizeCG;
+        const int cgPosX   = cgBlkPos - (cgPosY << codingParameters.log2TrSizeCG);
+        const uint64_t cgBlkPosMask = ((uint64_t)1 << cgBlkPos);
+
+        if (subSet == lastScanSet || !subSet)
+            sigCoeffGroupFlag64 |= cgBlkPosMask;
+        else
+        {
+            uint32_t sigCoeffGroup = ((sigCoeffGroupFlag64 & cgBlkPosMask) != 0);
+            uint32_t ctxSig = Quant::getSigCoeffGroupCtxInc(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
+            encodeBin(sigCoeffGroup, baseCoeffGroupCtx[ctxSig]);
+        }
+
+        // encode significant_coeff_flag
+        if (sigCoeffGroupFlag64 & cgBlkPosMask)
+        {
+            const int patternSigCtx = Quant::calcPatternSigCtx(sigCoeffGroupFlag64, cgPosX, cgPosY, codingParameters.log2TrSizeCG);
+            uint32_t blkPos, sig, ctxSig;
+            for (; scanPosSig >= subPos; scanPosSig--)
+            {
+                blkPos  = codingParameters.scan[scanPosSig];
+                sig     = (coeff[blkPos] != 0);
+                if (scanPosSig > subPos || subSet == 0 || numNonZero)
+                {
+                    ctxSig = Quant::getSigCtxInc(patternSigCtx, log2TrSize, trSize, blkPos, bIsLuma, codingParameters.firstSignificanceMapContext);
+                    encodeBin(sig, baseCtx[ctxSig]);
+                }
+                if (sig)
+                {
+                    absCoeff[numNonZero] = int(abs(coeff[blkPos]));
+                    coeffSigns = 2 * coeffSigns + ((uint32_t)coeff[blkPos] >> 31);
+                    numNonZero++;
+                    if (lastNZPosInCG < 0)
+                        lastNZPosInCG = scanPosSig;
+                    firstNZPosInCG = scanPosSig;
+                }
+            }
+        }
+        else
+            scanPosSig = subPos - 1;
+
+        if (numNonZero > 0)
+        {
+            bool signHidden = (lastNZPosInCG - firstNZPosInCG >= SBH_THRESHOLD);
+            uint32_t ctxSet = (subSet > 0 && bIsLuma) ? 2 : 0;
+
+            if (c1 == 0)
+                ctxSet++;
+
+            c1 = 1;
+            uint8_t *baseCtxMod = bIsLuma ? &m_contextState[OFF_ONE_FLAG_CTX + 4 * ctxSet] : &m_contextState[OFF_ONE_FLAG_CTX + NUM_ONE_FLAG_CTX_LUMA + 4 * ctxSet];
+
+            int numC1Flag = X265_MIN(numNonZero, C1FLAG_NUMBER);
+            int firstC2FlagIdx = -1;
+            for (int idx = 0; idx < numC1Flag; idx++)
+            {
+                uint32_t symbol = absCoeff[idx] > 1;
+                encodeBin(symbol, baseCtxMod[c1]);
+                if (symbol)
+                {
+                    c1 = 0;
+
+                    if (firstC2FlagIdx == -1)
+                        firstC2FlagIdx = idx;
+                }
+                else if ((c1 < 3) && (c1 > 0))
+                    c1++;
+            }
+
+            if (!c1)
+            {
+                baseCtxMod = bIsLuma ? &m_contextState[OFF_ABS_FLAG_CTX + ctxSet] : &m_contextState[OFF_ABS_FLAG_CTX + NUM_ABS_FLAG_CTX_LUMA + ctxSet];
+                if (firstC2FlagIdx != -1)
+                {
+                    uint32_t symbol = absCoeff[firstC2FlagIdx] > 2;
+                    encodeBin(symbol, baseCtxMod[0]);
+                }
+            }
+
+            if (bHideFirstSign && signHidden)
+                encodeBinsEP((coeffSigns >> 1), numNonZero - 1);
+            else
+                encodeBinsEP(coeffSigns, numNonZero);
+
+            int firstCoeff2 = 1;
+            if (!c1 || numNonZero > C1FLAG_NUMBER)
+            {
+                for (int idx = 0; idx < numNonZero; idx++)
+                {
+                    int baseLevel = (idx < C1FLAG_NUMBER) ? (2 + firstCoeff2) : 1;
+
+                    if (absCoeff[idx] >= baseLevel)
+                    {
+                        writeCoefRemainExGolomb(absCoeff[idx] - baseLevel, goRiceParam);
+                        if (absCoeff[idx] > 3 * (1 << goRiceParam))
+                            goRiceParam = std::min<uint32_t>(goRiceParam + 1, 4);
+                    }
+                    if (absCoeff[idx] >= 2)
+                        firstCoeff2 = 0;
+                }
+            }
+        }
+    }
+}
+
+void Entropy::codeSaoMaxUvlc(uint32_t code, uint32_t maxSymbol)
+{
+    X265_CHECK(maxSymbol > 0, "maxSymbol too small\n");
+
+    uint32_t isCodeLast = (maxSymbol > code) ? 1 : 0;
+    uint32_t isCodeNonZero = (code != 0) ? 1 : 0;
+
+    encodeBinEP(isCodeNonZero);
+    if (isCodeNonZero)
+    {
+        uint32_t mask = (1 << (code - 1)) - 1;
+        uint32_t len = code - 1 + isCodeLast;
+        mask <<= isCodeLast;
+
+        encodeBinsEP(mask, len);
+    }
+}
+
+/** Code SAO type index */
+void Entropy::codeSaoTypeIdx(uint32_t code)
+{
+    encodeBin((code == 0) ? 0 : 1, m_contextState[OFF_SAO_TYPE_IDX_CTX]);
+    if (code)
+        encodeBinEP(code <= 4 ? 1 : 0);
+}
+
+/* estimate bit cost for CBP, significant map and significant coefficients */
+void Entropy::estBit(EstBitsSbac& estBitsSbac, uint32_t log2TrSize, bool bIsLuma)
+{
+    estCBFBit(estBitsSbac);
+
+    estSignificantCoeffGroupMapBit(estBitsSbac, bIsLuma);
+
+    // encode significance map
+    estSignificantMapBit(estBitsSbac, log2TrSize, bIsLuma);
+
+    // encode significant coefficients
+    estSignificantCoefficientsBit(estBitsSbac, bIsLuma);
+}
+
+/* estimate bit cost for each CBP bit */
+void Entropy::estCBFBit(EstBitsSbac& estBitsSbac)
+{
+    uint8_t *ctx = &m_contextState[OFF_QT_CBF_CTX];
+
+    for (uint32_t ctxInc = 0; ctxInc < NUM_QT_CBF_CTX; ctxInc++)
+    {
+        estBitsSbac.blockCbpBits[ctxInc][0] = sbacGetEntropyBits(ctx[ctxInc], 0);
+        estBitsSbac.blockCbpBits[ctxInc][1] = sbacGetEntropyBits(ctx[ctxInc], 1);
+    }
+
+    ctx = &m_contextState[OFF_QT_ROOT_CBF_CTX];
+
+    estBitsSbac.blockRootCbpBits[0] = sbacGetEntropyBits(ctx[0], 0);
+    estBitsSbac.blockRootCbpBits[1] = sbacGetEntropyBits(ctx[0], 1);
+}
+
+/* estimate SAMBAC bit cost for significant coefficient group map */
+void Entropy::estSignificantCoeffGroupMapBit(EstBitsSbac& estBitsSbac, bool bIsLuma)
+{
+    int firstCtx = 0, numCtx = NUM_SIG_CG_FLAG_CTX;
+
+    for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
+        for (uint32_t bin = 0; bin < 2; bin++)
+            estBitsSbac.significantCoeffGroupBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextState[OFF_SIG_CG_FLAG_CTX + ((bIsLuma ? 0 : NUM_SIG_CG_FLAG_CTX) + ctxIdx)], bin);
+}
+
+/* estimate SAMBAC bit cost for significant coefficient map */
+void Entropy::estSignificantMapBit(EstBitsSbac& estBitsSbac, uint32_t log2TrSize, bool bIsLuma)
+{
+    int firstCtx = 1, numCtx = 8;
+
+    if (log2TrSize >= 4)
+    {
+        firstCtx = bIsLuma ? 21 : 12;
+        numCtx = bIsLuma ? 6 : 3;
+    }
+    else if (log2TrSize == 3)
+    {
+        firstCtx = 9;
+        numCtx = bIsLuma ? 12 : 3;
+    }
+
+    if (bIsLuma)
+    {
+        for (uint32_t bin = 0; bin < 2; bin++)
+            estBitsSbac.significantBits[0][bin] = sbacGetEntropyBits(m_contextState[OFF_SIG_FLAG_CTX], bin);
+
+        for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
+            for (uint32_t bin = 0; bin < 2; bin++)
+                estBitsSbac.significantBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextState[OFF_SIG_FLAG_CTX + ctxIdx], bin);
+    }
+    else
+    {
+        for (uint32_t bin = 0; bin < 2; bin++)
+            estBitsSbac.significantBits[0][bin] = sbacGetEntropyBits(m_contextState[OFF_SIG_FLAG_CTX + (NUM_SIG_FLAG_CTX_LUMA + 0)], bin);
+
+        for (int ctxIdx = firstCtx; ctxIdx < firstCtx + numCtx; ctxIdx++)
+            for (uint32_t bin = 0; bin < 2; bin++)
+                estBitsSbac.significantBits[ctxIdx][bin] = sbacGetEntropyBits(m_contextState[OFF_SIG_FLAG_CTX + (NUM_SIG_FLAG_CTX_LUMA + ctxIdx)], bin);
+    }
+    int bitsX = 0, bitsY = 0;
+
+    int blkSizeOffset = bIsLuma ? ((log2TrSize - 2) * 3 + ((log2TrSize - 1) >> 2)) : NUM_CTX_LAST_FLAG_XY_LUMA;
+    int ctxShift = bIsLuma ? ((log2TrSize + 1) >> 2) : log2TrSize - 2;
+    uint32_t maxGroupIdx = log2TrSize * 2 - 1;
+
+    uint32_t ctx;
+    const uint8_t *ctxX = &m_contextState[OFF_CTX_LAST_FLAG_X];
+    for (ctx = 0; ctx < maxGroupIdx; ctx++)
+    {
+        int ctxOffset = blkSizeOffset + (ctx >> ctxShift);
+        estBitsSbac.lastXBits[ctx] = bitsX + sbacGetEntropyBits(ctxX[ctxOffset], 0);
+        bitsX += sbacGetEntropyBits(ctxX[ctxOffset], 1);
+    }
+
+    estBitsSbac.lastXBits[ctx] = bitsX;
+
+    const uint8_t *ctxY = &m_contextState[OFF_CTX_LAST_FLAG_Y];
+    for (ctx = 0; ctx < maxGroupIdx; ctx++)
+    {
+        int ctxOffset = blkSizeOffset + (ctx >> ctxShift);
+        estBitsSbac.lastYBits[ctx] = bitsY + sbacGetEntropyBits(ctxY[ctxOffset], 0);
+        bitsY += sbacGetEntropyBits(ctxY[ctxOffset], 1);
+    }
+
+    estBitsSbac.lastYBits[ctx] = bitsY;
+}
+
+/* estimate bit cost of significant coefficient */
+void Entropy::estSignificantCoefficientsBit(EstBitsSbac& estBitsSbac, bool bIsLuma)
+{
+    if (bIsLuma)
+    {
+        uint8_t *ctxOne = &m_contextState[OFF_ONE_FLAG_CTX];
+        uint8_t *ctxAbs = &m_contextState[OFF_ABS_FLAG_CTX];
+
+        for (int ctxIdx = 0; ctxIdx < NUM_ONE_FLAG_CTX_LUMA; ctxIdx++)
+        {
+            estBitsSbac.greaterOneBits[ctxIdx][0] = sbacGetEntropyBits(ctxOne[ctxIdx], 0);
+            estBitsSbac.greaterOneBits[ctxIdx][1] = sbacGetEntropyBits(ctxOne[ctxIdx], 1);
+        }
+
+        for (int ctxIdx = 0; ctxIdx < NUM_ABS_FLAG_CTX_LUMA; ctxIdx++)
+        {
+            estBitsSbac.levelAbsBits[ctxIdx][0] = sbacGetEntropyBits(ctxAbs[ctxIdx], 0);
+            estBitsSbac.levelAbsBits[ctxIdx][1] = sbacGetEntropyBits(ctxAbs[ctxIdx], 1);
+        }
+    }
+    else
+    {
+        uint8_t *ctxOne = &m_contextState[OFF_ONE_FLAG_CTX + NUM_ONE_FLAG_CTX_LUMA];
+        uint8_t *ctxAbs = &m_contextState[OFF_ABS_FLAG_CTX + NUM_ABS_FLAG_CTX_LUMA];
+
+        for (int ctxIdx = 0; ctxIdx < NUM_ONE_FLAG_CTX_CHROMA; ctxIdx++)
+        {
+            estBitsSbac.greaterOneBits[ctxIdx][0] = sbacGetEntropyBits(ctxOne[ctxIdx], 0);
+            estBitsSbac.greaterOneBits[ctxIdx][1] = sbacGetEntropyBits(ctxOne[ctxIdx], 1);
+        }
+
+        for (int ctxIdx = 0; ctxIdx < NUM_ABS_FLAG_CTX_CHROMA; ctxIdx++)
+        {
+            estBitsSbac.levelAbsBits[ctxIdx][0] = sbacGetEntropyBits(ctxAbs[ctxIdx], 0);
+            estBitsSbac.levelAbsBits[ctxIdx][1] = sbacGetEntropyBits(ctxAbs[ctxIdx], 1);
+        }
+    }
+}
+
+/* Initialize our context information from the nominated source */
+void Entropy::copyContextsFrom(Entropy& src)
+{
+    memcpy(m_contextState, src.m_contextState, MAX_OFF_CTX_MOD * sizeof(m_contextState[0]));
+}
+
+void Entropy::start()
+{
+    m_low = 0;
+    m_range = 510;
+    m_bitsLeft = -12;
+    m_numBufferedBytes = 0;
+    m_bufferedByte = 0xff;
+}
+
+void Entropy::finish()
+{
+    if (m_low >> (21 + m_bitsLeft))
+    {
+        m_bitIf->writeByte(m_bufferedByte + 1);
+        while (m_numBufferedBytes > 1)
+        {
+            m_bitIf->writeByte(0x00);
+            m_numBufferedBytes--;
+        }
+
+        m_low -= 1 << (21 + m_bitsLeft);
+    }
+    else
+    {
+        if (m_numBufferedBytes > 0)
+            m_bitIf->writeByte(m_bufferedByte);
+
+        while (m_numBufferedBytes > 1)
+        {
+            m_bitIf->writeByte(0xff);
+            m_numBufferedBytes--;
+        }
+    }
+    m_bitIf->write(m_low >> 8, 13 + m_bitsLeft);
+}
+
+void Entropy::copyState(Entropy& other)
+{
+    m_low = other.m_low;
+    m_range = other.m_range;
+    m_bitsLeft = other.m_bitsLeft;
+    m_bufferedByte = other.m_bufferedByte;
+    m_numBufferedBytes = other.m_numBufferedBytes;
+    m_fracBits = other.m_fracBits;
+}
+
+void Entropy::resetBits()
+{
+    m_low = 0;
+    m_bitsLeft = -12;
+    m_numBufferedBytes = 0;
+    m_bufferedByte = 0xff;
+    m_fracBits &= 32767;
+    if (m_bitIf)
+        m_bitIf->resetBits();
+}
+
+/** Encode bin */
+void Entropy::encodeBin(uint32_t binValue, uint8_t &ctxModel)
+{
+    uint32_t mstate = ctxModel;
+
+    ctxModel = sbacNext(mstate, binValue);
+
+    if (!m_bitIf)
+    {
+        m_fracBits += sbacGetEntropyBits(mstate, binValue);
+        return;
+    }
+
+    uint32_t range = m_range;
+    uint32_t state = sbacGetState(mstate);
+    uint32_t lps = g_lpsTable[state][((uint8_t)range >> 6)];
+    range -= lps;
+
+    X265_CHECK(lps >= 2, "lps is too small\n");
+
+    int numBits = (uint32_t)(range - 256) >> 31;
+    uint32_t low = m_low;
+
+    // NOTE: MPS must be LOWEST bit in mstate
+    X265_CHECK((uint32_t)((binValue ^ mstate) & 1) == (uint32_t)(binValue != sbacGetMps(mstate)), "binValue failure\n");
+    if ((binValue ^ mstate) & 1)
+    {
+        // NOTE: lps is non-zero and the maximum of idx is 8 because lps less than 256
+        //numBits   = g_renormTable[lps >> 3];
+        unsigned long idx;
+        CLZ32(idx, lps);
+        X265_CHECK(state != 63 || idx == 1, "state failure\n");
+
+        numBits = 8 - idx;
+        if (state >= 63)
+            numBits = 6;
+        X265_CHECK(numBits <= 6, "numBits failure\n");
+
+        low += range;
+        range = lps;
+    }
+    m_low = (low << numBits);
+    m_range = (range << numBits);
+    m_bitsLeft += numBits;
+
+    if (m_bitsLeft >= 0)
+        writeOut();
+}
+
+/** Encode equiprobable bin */
+void Entropy::encodeBinEP(uint32_t binValue)
+{
+    if (!m_bitIf)
+    {
+        m_fracBits += 32768;
+        return;
+    }
+    m_low <<= 1;
+    if (binValue)
+        m_low += m_range;
+    m_bitsLeft++;
+
+    if (m_bitsLeft >= 0)
+        writeOut();
+}
+
+/** Encode equiprobable bins */
+void Entropy::encodeBinsEP(uint32_t binValues, int numBins)
+{
+    if (!m_bitIf)
+    {
+        m_fracBits += 32768 * numBins;
+        return;
+    }
+
+    while (numBins > 8)
+    {
+        numBins -= 8;
+        uint32_t pattern = binValues >> numBins;
+        m_low <<= 8;
+        m_low += m_range * pattern;
+        binValues -= pattern << numBins;
+        m_bitsLeft += 8;
+
+        if (m_bitsLeft >= 0)
+            writeOut();
+    }
+
+    m_low <<= numBins;
+    m_low += m_range * binValues;
+    m_bitsLeft += numBins;
+
+    if (m_bitsLeft >= 0)
+        writeOut();
+}
+
+/** Encode terminating bin */
+void Entropy::encodeBinTrm(uint32_t binValue)
+{
+    if (!m_bitIf)
+    {
+        m_fracBits += sbacGetEntropyBitsTrm(binValue);
+        return;
+    }
+
+    m_range -= 2;
+    if (binValue)
+    {
+        m_low += m_range;
+        m_low <<= 7;
+        m_range = 2 << 7;
+        m_bitsLeft += 7;
+    }
+    else if (m_range >= 256)
+        return;
+    else
+    {
+        m_low <<= 1;
+        m_range <<= 1;
+        m_bitsLeft++;
+    }
+
+    if (m_bitsLeft >= 0)
+        writeOut();
+}
+
+/** Move bits from register into bitstream */
+void Entropy::writeOut()
+{
+    uint32_t leadByte = m_low >> (13 + m_bitsLeft);
+    uint32_t low_mask = (uint32_t)(~0) >> (11 + 8 - m_bitsLeft);
+
+    m_bitsLeft -= 8;
+    m_low &= low_mask;
+
+    if (leadByte == 0xff)
+        m_numBufferedBytes++;
+    else
+    {
+        uint32_t numBufferedBytes = m_numBufferedBytes;
+        if (numBufferedBytes > 0)
+        {
+            uint32_t carry = leadByte >> 8;
+            uint32_t byteTowrite = m_bufferedByte + carry;
+            m_bitIf->writeByte(byteTowrite);
+
+            byteTowrite = (0xff + carry) & 0xff;
+            while (numBufferedBytes > 1)
+            {
+                m_bitIf->writeByte(byteTowrite);
+                numBufferedBytes--;
+            }
+        }
+        m_numBufferedBytes = 1;
+        m_bufferedByte = (uint8_t)leadByte;
+    }
+}
+
+const uint32_t g_entropyBits[128] =
+{
+    // Corrected table, most notably for last state
+    0x07b23, 0x085f9, 0x074a0, 0x08cbc, 0x06ee4, 0x09354, 0x067f4, 0x09c1b, 0x060b0, 0x0a62a, 0x05a9c, 0x0af5b, 0x0548d, 0x0b955, 0x04f56, 0x0c2a9,
+    0x04a87, 0x0cbf7, 0x045d6, 0x0d5c3, 0x04144, 0x0e01b, 0x03d88, 0x0e937, 0x039e0, 0x0f2cd, 0x03663, 0x0fc9e, 0x03347, 0x10600, 0x03050, 0x10f95,
+    0x02d4d, 0x11a02, 0x02ad3, 0x12333, 0x0286e, 0x12cad, 0x02604, 0x136df, 0x02425, 0x13f48, 0x021f4, 0x149c4, 0x0203e, 0x1527b, 0x01e4d, 0x15d00,
+    0x01c99, 0x166de, 0x01b18, 0x17017, 0x019a5, 0x17988, 0x01841, 0x18327, 0x016df, 0x18d50, 0x015d9, 0x19547, 0x0147c, 0x1a083, 0x0138e, 0x1a8a3,
+    0x01251, 0x1b418, 0x01166, 0x1bd27, 0x01068, 0x1c77b, 0x00f7f, 0x1d18e, 0x00eda, 0x1d91a, 0x00e19, 0x1e254, 0x00d4f, 0x1ec9a, 0x00c90, 0x1f6e0,
+    0x00c01, 0x1fef8, 0x00b5f, 0x208b1, 0x00ab6, 0x21362, 0x00a15, 0x21e46, 0x00988, 0x2285d, 0x00934, 0x22ea8, 0x008a8, 0x239b2, 0x0081d, 0x24577,
+    0x007c9, 0x24ce6, 0x00763, 0x25663, 0x00710, 0x25e8f, 0x006a0, 0x26a26, 0x00672, 0x26f23, 0x005e8, 0x27ef8, 0x005ba, 0x284b5, 0x0055e, 0x29057,
+    0x0050c, 0x29bab, 0x004c1, 0x2a674, 0x004a7, 0x2aa5e, 0x0046f, 0x2b32f, 0x0041f, 0x2c0ad, 0x003e7, 0x2ca8d, 0x003ba, 0x2d323, 0x0010c, 0x3bfbb
+};
+
+const uint8_t g_nextState[128][2] =
+{
+    { 2, 1 }, { 0, 3 }, { 4, 0 }, { 1, 5 }, { 6, 2 }, { 3, 7 }, { 8, 4 }, { 5, 9 },
+    { 10, 4 }, { 5, 11 }, { 12, 8 }, { 9, 13 }, { 14, 8 }, { 9, 15 }, { 16, 10 }, { 11, 17 },
+    { 18, 12 }, { 13, 19 }, { 20, 14 }, { 15, 21 }, { 22, 16 }, { 17, 23 }, { 24, 18 }, { 19, 25 },
+    { 26, 18 }, { 19, 27 }, { 28, 22 }, { 23, 29 }, { 30, 22 }, { 23, 31 }, { 32, 24 }, { 25, 33 },
+    { 34, 26 }, { 27, 35 }, { 36, 26 }, { 27, 37 }, { 38, 30 }, { 31, 39 }, { 40, 30 }, { 31, 41 },
+    { 42, 32 }, { 33, 43 }, { 44, 32 }, { 33, 45 }, { 46, 36 }, { 37, 47 }, { 48, 36 }, { 37, 49 },
+    { 50, 38 }, { 39, 51 }, { 52, 38 }, { 39, 53 }, { 54, 42 }, { 43, 55 }, { 56, 42 }, { 43, 57 },
+    { 58, 44 }, { 45, 59 }, { 60, 44 }, { 45, 61 }, { 62, 46 }, { 47, 63 }, { 64, 48 }, { 49, 65 },
+    { 66, 48 }, { 49, 67 }, { 68, 50 }, { 51, 69 }, { 70, 52 }, { 53, 71 }, { 72, 52 }, { 53, 73 },
+    { 74, 54 }, { 55, 75 }, { 76, 54 }, { 55, 77 }, { 78, 56 }, { 57, 79 }, { 80, 58 }, { 59, 81 },
+    { 82, 58 }, { 59, 83 }, { 84, 60 }, { 61, 85 }, { 86, 60 }, { 61, 87 }, { 88, 60 }, { 61, 89 },
+    { 90, 62 }, { 63, 91 }, { 92, 64 }, { 65, 93 }, { 94, 64 }, { 65, 95 }, { 96, 66 }, { 67, 97 },
+    { 98, 66 }, { 67, 99 }, { 100, 66 }, { 67, 101 }, { 102, 68 }, { 69, 103 }, { 104, 68 }, { 69, 105 },
+    { 106, 70 }, { 71, 107 }, { 108, 70 }, { 71, 109 }, { 110, 70 }, { 71, 111 }, { 112, 72 }, { 73, 113 },
+    { 114, 72 }, { 73, 115 }, { 116, 72 }, { 73, 117 }, { 118, 74 }, { 75, 119 }, { 120, 74 }, { 75, 121 },
+    { 122, 74 }, { 75, 123 }, { 124, 76 }, { 77, 125 }, { 124, 76 }, { 77, 125 }, { 126, 126 }, { 127, 127 }
+};
+
+}

x265-hg20140928.tar.bz2/source/encoder/entropy.h Added

@@ -0,0 +1,241 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_ENTROPY_H
+#define X265_ENTROPY_H
+
+#include "common.h"
+#include "bitstream.h"
+#include "frame.h"
+#include "slice.h"
+
+#include "TLibCommon/ContextTables.h"
+
+namespace x265 {
+// private namespace
+
+struct SaoLcuParam;
+struct EstBitsSbac;
+class TComDataCU;
+class ScalingList;
+
+
+enum SplitType
+{
+    DONT_SPLIT            = 0,
+    VERTICAL_SPLIT        = 1,
+    QUAD_SPLIT            = 2,
+    NUMBER_OF_SPLIT_MODES = 3
+};
+
+struct TURecurse
+{
+    uint32_t section;
+    uint32_t splitMode;
+    uint32_t absPartIdxTURelCU;
+    uint32_t absPartIdxStep;
+
+    TURecurse(SplitType splitType, uint32_t _absPartIdxStep, uint32_t _absPartIdxTU)
+    {
+        static const uint32_t partIdxStepShift[NUMBER_OF_SPLIT_MODES] = { 0, 1, 2 };
+        section           = 0;
+        absPartIdxTURelCU = _absPartIdxTU;
+        splitMode         = (uint32_t)splitType;
+        absPartIdxStep    = _absPartIdxStep >> partIdxStepShift[splitMode];
+    }
+
+    bool isNextSection()
+    {
+        if (splitMode == DONT_SPLIT)
+        {
+            section++;
+            return false;
+        }
+        else
+        {
+            absPartIdxTURelCU += absPartIdxStep;
+
+            section++;
+            return section < (uint32_t)(1 << splitMode);
+        }
+    }
+
+    bool isLastSection() const
+    {
+        return (section + 1) >= (uint32_t)(1 << splitMode);
+    }
+};
+
+struct EstBitsSbac
+{
+    int significantCoeffGroupBits[NUM_SIG_CG_FLAG_CTX][2];
+    int significantBits[NUM_SIG_FLAG_CTX][2];
+    int lastXBits[10];
+    int lastYBits[10];
+    int greaterOneBits[NUM_ONE_FLAG_CTX][2];
+    int levelAbsBits[NUM_ABS_FLAG_CTX][2];
+    int blockCbpBits[NUM_QT_CBF_CTX][2];
+    int blockRootCbpBits[2];
+};
+
+class Entropy : public SyntaxElementWriter
+{
+public:
+
+    uint64_t      m_pad;
+    uint8_t       m_contextState[160]; // MAX_OFF_CTX_MOD + padding
+
+    /* CABAC state */
+    uint32_t      m_low;
+    uint32_t      m_range;
+    uint32_t      m_bufferedByte;
+    int           m_numBufferedBytes;
+    int           m_bitsLeft;
+    uint64_t      m_fracBits;
+    EstBitsSbac   m_estBitsSbac;
+
+    Entropy();
+
+    void setBitstream(Bitstream* p)    { m_bitIf = p; }
+
+    uint32_t getNumberOfWrittenBits()
+    {
+        X265_CHECK(!m_bitIf, "bit counting mode expected\n");
+        return uint32_t(m_fracBits >> 15);
+    }
+
+    void zeroFract()                   { m_fracBits = 0; }
+    void resetBits();
+    void resetEntropy(Slice *slice);
+
+    // SBAC RD
+    void load(Entropy& src)            { copyFrom(src); }
+
+    void loadIntraDirModeLuma(Entropy& src);
+    void store(Entropy& dest)          { dest.copyFrom(*this); }
+    void loadContexts(Entropy& src)    { copyContextsFrom(src); }
+    void copyState(Entropy& other);
+
+    void codeVPS(VPS* vps);
+    void codeSPS(SPS* sps, ScalingList *scalingList, ProfileTierLevel *ptl);
+    void codePPS(PPS* pps);
+    void codeVUI(VUI* vui);
+    void codeAUD(Slice *slice);
+    void codeHrdParameters(HRDInfo* hrd);
+
+    void codeSliceHeader(Slice* slice);
+    void codeSliceHeaderWPPEntryPoints(Slice* slice, uint32_t *substreamSizes, uint32_t maxOffset);
+    void codeShortTermRefPicSet(RPS* rps);
+    void finishSlice()                 { encodeBinTrm(1); finish(); dynamic_cast<Bitstream*>(m_bitIf)->writeByteAlignment(); }
+
+    void encodeCTU(TComDataCU* cu);
+    void codeSaoOffset(SaoLcuParam* saoLcuParam, uint32_t compIdx);
+    void codeSaoUnitInterleaving(int compIdx, bool saoFlag, int rx, int ry, SaoLcuParam* saoLcuParam, int cuAddrInSlice, int cuAddrUpInSlice, int allowMergeLeft, int allowMergeUp);
+    void codeSaoMerge(uint32_t code)   { encodeBin(code, m_contextState[OFF_SAO_MERGE_FLAG_CTX]); }
+
+    void codeCUTransquantBypassFlag(uint32_t symbol);
+    void codeSkipFlag(TComDataCU* cu, uint32_t absPartIdx);
+    void codeMergeFlag(TComDataCU* cu, uint32_t absPartIdx);
+    void codeMergeIndex(TComDataCU* cu, uint32_t absPartIdx);
+    void codeSplitFlag(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
+    void codeMVPIdx(uint32_t symbol);
+    void codeMvd(TComDataCU* cu, uint32_t absPartIdx, int list);
+
+    void codePartSize(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
+    void codePredMode(int predMode);
+    void codePredInfo(TComDataCU* cu, uint32_t absPartIdx);
+    void codeTransformSubdivFlag(uint32_t symbol, uint32_t ctx);
+    void codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, uint32_t absPartIdxStep, uint32_t width, uint32_t height, TextType ttype, uint32_t trDepth, bool lowestLevel);
+    void codeQtCbf(TComDataCU* cu, uint32_t absPartIdx, TextType ttype, uint32_t trDepth);
+    void codeQtCbfZero(TextType ttype, uint32_t trDepth);
+    void codeQtRootCbfZero();
+    void codeCoeff(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool& bCodeDQP, uint32_t* depthRange);
+    void codeCoeffNxN(TComDataCU* cu, coeff_t* coef, uint32_t absPartIdx, uint32_t log2TrSize, TextType ttype);
+
+    void codeIntraDirLumaAng(TComDataCU* cu, uint32_t absPartIdx, bool isMultiple);
+    void codeIntraDirChroma(TComDataCU* cu, uint32_t absPartIdx);
+
+    // RDO functions
+    void estBit(EstBitsSbac& estBitsSbac, uint32_t log2TrSize, bool bIsLuma);
+    void estCBFBit(EstBitsSbac& estBitsSbac);
+    void estSignificantCoeffGroupMapBit(EstBitsSbac& estBitsSbac, bool bIsLuma);
+    void estSignificantMapBit(EstBitsSbac& estBitsSbac, uint32_t log2TrSize, bool bIsLuma);
+    void estSignificantCoefficientsBit(EstBitsSbac& estBitsSbac, bool bIsLuma);
+
+private:
+
+    /* CABAC private methods */
+    void start();
+    void finish();
+
+    void encodeBin(uint32_t binValue, uint8_t& ctxModel);
+    void encodeBinEP(uint32_t binValue);
+    void encodeBinsEP(uint32_t binValues, int numBins);
+    void encodeBinTrm(uint32_t binValue);
+
+    void encodeCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool& bEncodeDQP, CU *cuData);
+    void finishCU(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth);
+
+    void writeOut();
+
+    /* SBac private methods */
+    void writeUnaryMaxSymbol(uint32_t symbol, uint8_t* scmModel, int offset, uint32_t maxSymbol);
+    void writeEpExGolomb(uint32_t symbol, uint32_t count);
+    void writeCoefRemainExGolomb(uint32_t symbol, const uint32_t absGoRice);
+
+    void codeProfileTier(ProfileTierLevel& ptl);
+    void codeScalingList(ScalingList*);
+    void codeScalingList(ScalingList* scalingList, uint32_t sizeId, uint32_t listId);
+
+    void codePredWeightTable(Slice* slice);
+    void codeInterDir(TComDataCU* cu, uint32_t absPartIdx);
+    void codePUWise(TComDataCU* cu, uint32_t absPartIdx);
+    void codeQtRootCbf(uint32_t cbf);
+    void codeRefFrmIdxPU(TComDataCU* cu, uint32_t absPartIdx, int list);
+    void codeRefFrmIdx(TComDataCU* cu, uint32_t absPartIdx, int list);
+
+    void codeSaoMaxUvlc(uint32_t code, uint32_t maxSymbol);
+    void codeSaoTypeIdx(uint32_t code);
+    void codeSaoUflc(uint32_t length, uint32_t code) { encodeBinsEP(code, length); }
+    void codeSAOSign(uint32_t code)                  { encodeBinEP(code); }
+
+    void codeDeltaQP(TComDataCU* cu, uint32_t absPartIdx);
+    void codeLastSignificantXY(uint32_t posx, uint32_t posy, uint32_t log2TrSize, bool bIsLuma, uint32_t scanIdx);
+    void codeTransformSkipFlags(TComDataCU* cu, uint32_t absPartIdx, uint32_t trSize, TextType ttype);
+
+    struct CoeffCodeState
+    {
+        uint32_t bakAbsPartIdx;
+        uint32_t bakChromaOffset;
+        uint32_t bakAbsPartIdxCU;
+    };
+
+    void encodeTransform(TComDataCU* cu, CoeffCodeState& state, uint32_t offsetLumaOffset, uint32_t offsetChroma, 
+                         uint32_t absPartIdx, uint32_t absPartIdxStep, uint32_t depth, uint32_t log2TrSize, uint32_t uiTrIdx, bool& bCodeDQP, uint32_t* depthRange);
+
+    void copyFrom(Entropy& src);
+    void copyContextsFrom(Entropy& src);
+};
+}
+
+#endif // ifndef X265_ENTROPY_H

x265-hg20140324.tar.bz2/source/encoder/frameencoder.cpp -> x265-hg20140928.tar.bz2/source/encoder/frameencoder.cpp Changed

@@ -20,51 +20,40 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "PPA/ppa.h"
 #include "wavefront.h"
+#include "param.h"
 
-#include "TLibEncoder/NALwrite.h"
 #include "encoder.h"
 #include "frameencoder.h"
-#include "cturow.h"
 #include "common.h"
 #include "slicetype.h"
+#include "nal.h"
 
 namespace x265 {
-void weightAnalyse(TComSlice& slice, x265_param& param);
-
-enum SCALING_LIST_PARAMETER
-{
-    SCALING_LIST_OFF,
-    SCALING_LIST_DEFAULT,
-};
+void weightAnalyse(Slice& slice, x265_param& param);
 
 FrameEncoder::FrameEncoder()
     : WaveFront(NULL)
     , m_threadActive(true)
     , m_rows(NULL)
     , m_top(NULL)
-    , m_cfg(NULL)
-    , m_pic(NULL)
+    , m_param(NULL)
+    , m_frame(NULL)
 {
-    for (int i = 0; i < MAX_NAL_UNITS; i++)
-    {
-        m_nalList[i] = NULL;
-    }
-
-    m_nalCount = 0;
     m_totalTime = 0;
+    m_bAllRowsStop = false;
+    m_vbvResetTriggerRow = -1;
+    m_outStreams = NULL;
+    m_substreamSizes = NULL;
+    m_nr = NULL;
+    memset(&m_frameStats, 0, sizeof(m_frameStats));
     memset(&m_rce, 0, sizeof(RateControlEntry));
 }
 
-void FrameEncoder::setThreadPool(ThreadPool *p)
-{
-    m_pool = p;
-}
-
 void FrameEncoder::destroy()
 {
     JobProvider::flush();  // ensure no worker threads are using this frame
@@ -72,230 +61,77 @@
     m_threadActive = false;
     m_enable.trigger();
 
-    // flush condition, release queued NALs
-    for (int i = 0; i < m_nalCount; i++)
-    {
-        NALUnitEBSP *nalu = m_nalList[i];
-        free(nalu->m_nalUnitData);
-        X265_FREE(nalu);
-    }
+    delete[] m_rows;
 
-    if (m_rows)
+    if (m_param->bEmitHRDSEI)
     {
-        for (int i = 0; i < m_numRows; ++i)
-        {
-            m_rows[i].destroy();
-        }
-
-        delete[] m_rows;
+        delete m_rce.picTimingSEI;
+        delete m_rce.hrdTiming;
     }
 
+    delete[] m_outStreams;
+    X265_FREE(m_substreamSizes);
     m_frameFilter.destroy();
+
+    X265_FREE(m_nr);
+
     // wait for worker thread to exit
     stop();
 }
 
-bool FrameEncoder::init(Encoder *top, int numRows)
+bool FrameEncoder::init(Encoder *top, int numRows, int numCols)
 {
-    bool ok = true;
     m_top = top;
-    m_cfg = top;
+    m_param = top->m_param;
     m_numRows = numRows;
-    m_filterRowDelay = (m_cfg->param->saoLcuBasedOptimization && m_cfg->param->saoLcuBoundary) ?
-        2 : (m_cfg->param->bEnableSAO || m_cfg->param->bEnableLoopFilter ? 1 : 0);
+    m_numCols = numCols;
+    m_filterRowDelay = (m_param->bEnableSAO && m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary) ?
+                        2 : (m_param->bEnableSAO || m_param->bEnableLoopFilter ? 1 : 0);
+    m_filterRowDelayCus = m_filterRowDelay * numCols;
 
     m_rows = new CTURow[m_numRows];
-    for (int i = 0; i < m_numRows; ++i)
-    {
-        ok &= m_rows[i].create(top);
+    bool ok = !!m_numRows;
 
-        for (int list = 0; list <= 1; list++)
-        {
-            for (int ref = 0; ref <= MAX_NUM_REF; ref++)
-            {
-                m_rows[i].m_search.m_mref[list][ref] = &m_mref[list][ref];
-            }
-        }
-    }
+    int range  = m_param->searchRange; /* fpel search */
+        range += 1;                    /* diamond search range check lag */
+        range += 2;                    /* subpel refine */
+        range += NTAPS_LUMA / 2;       /* subpel filter half-length */
+    m_refLagRows = 1 + ((range + g_maxCUSize - 1) / g_maxCUSize);
 
     // NOTE: 2 times of numRows because both Encoder and Filter in same queue
     if (!WaveFront::init(m_numRows * 2))
     {
-        x265_log(m_cfg->param, X265_LOG_ERROR, "unable to initialize wavefront queue\n");
+        x265_log(m_param, X265_LOG_ERROR, "unable to initialize wavefront queue\n");
         m_pool = NULL;
     }
 
-    m_frameFilter.init(top, numRows, getRDGoOnSbacCoder(0));
-
-    // initialize SPS
-    top->initSPS(&m_sps);
-
-    // initialize PPS
-    m_pps.setSPS(&m_sps);
-
-    top->initPPS(&m_pps);
-
-    m_sps.setNumLongTermRefPicSPS(0);
-    if (m_cfg->m_decodingUnitInfoSEIEnabled)
-    {
-        m_sps.setHrdParameters(m_cfg->param->fpsNum, m_cfg->param->fpsDenom, 0, m_cfg->param->rc.bitrate, m_cfg->param->bframes > 0);
-    }
-    if (m_cfg->m_bufferingPeriodSEIEnabled || m_cfg->m_decodingUnitInfoSEIEnabled)
-    {
-        m_sps.getVuiParameters()->setHrdParametersPresentFlag(true);
-    }
-
-    m_sps.setTMVPFlagsPresent(true);
+    m_frameFilter.init(top, this, numRows);
 
-    // set default slice level flag to the same as SPS level flag
-    if (m_cfg->m_useScalingListId == SCALING_LIST_OFF)
+    // initialize HRD parameters of SPS
+    if (m_param->bEmitHRDSEI)
     {
-        for (int i = 0; i < m_numRows; i++)
-        {
-            m_rows[i].m_trQuant.setFlatScalingList();
-            m_rows[i].m_trQuant.setUseScalingList(false);
-        }
+        m_rce.picTimingSEI = new SEIPictureTiming;
+        m_rce.hrdTiming = new HRDTiming;
 
-        m_sps.setScalingListPresentFlag(false);
-        m_pps.setScalingListPresentFlag(false);
+        ok &= m_rce.picTimingSEI && m_rce.hrdTiming;
     }
-    else if (m_cfg->m_useScalingListId == SCALING_LIST_DEFAULT)
-    {
-        for (int i = 0; i < m_numRows; i++)
-        {
-            m_rows[i].m_trQuant.setScalingList(m_top->getScalingList());
-            m_rows[i].m_trQuant.setUseScalingList(true);
-        }
 
-        m_sps.setScalingListPresentFlag(false);
-        m_pps.setScalingListPresentFlag(false);
-    }
+    if (m_param->noiseReduction)
+        m_nr = X265_MALLOC(NoiseReduction, 1);
+    if (m_nr)
+        memset(m_nr, 0, sizeof(NoiseReduction));
     else
-    {
-        x265_log(m_cfg->param, X265_LOG_ERROR, "ScalingList == %d not supported\n", m_top->m_useScalingListId);
-        ok = false;
-    }
+        m_param->noiseReduction = 0;
 
     start();
     return ok;
 }
 
-int FrameEncoder::getStreamHeaders(NALUnitEBSP **nalunits)
+void FrameEncoder::startCompressFrame(Frame* pic)
 {
-    TEncEntropy* entropyCoder = getEntropyCoder(0);
-
-    entropyCoder->setEntropyCoder(&m_sbacCoder, NULL);
-    int count = 0;
-
-    /* headers for start of bitstream */
-    OutputNALUnit nalu(NAL_UNIT_VPS);
-    entropyCoder->setBitstream(&nalu.m_bitstream);
-    entropyCoder->encodeVPS(&m_cfg->m_vps);
-    writeRBSPTrailingBits(nalu.m_bitstream);
-    CHECKED_MALLOC(nalunits[count], NALUnitEBSP, 1);
-    nalunits[count]->init(nalu);
-    count++;
-
-    nalu.resetToType(NAL_UNIT_SPS);
-    entropyCoder->setBitstream(&nalu.m_bitstream);
-    entropyCoder->encodeSPS(&m_sps);
-    writeRBSPTrailingBits(nalu.m_bitstream);
-    CHECKED_MALLOC(nalunits[count], NALUnitEBSP, 1);
-    nalunits[count]->init(nalu);
-    count++;
-
-    nalu.resetToType(NAL_UNIT_PPS);
-    entropyCoder->setBitstream(&nalu.m_bitstream);
-    entropyCoder->encodePPS(&m_pps);
-    writeRBSPTrailingBits(nalu.m_bitstream);
-    CHECKED_MALLOC(nalunits[count], NALUnitEBSP, 1);
-    nalunits[count]->init(nalu);
-    count++;
-
-    if (m_cfg->m_activeParameterSetsSEIEnabled)
-    {
-        SEIActiveParameterSets sei;
-        sei.activeVPSId = m_cfg->m_vps.getVPSId();
-        sei.m_fullRandomAccessFlag = false;
-        sei.m_noParamSetUpdateFlag = false;
-        sei.numSpsIdsMinus1 = 0;
-        sei.activeSeqParamSetId = m_sps.getSPSId();
-
-        entropyCoder->setBitstream(&nalu.m_bitstream);
-        m_seiWriter.writeSEImessage(nalu.m_bitstream, sei, &m_sps);
-        writeRBSPTrailingBits(nalu.m_bitstream);
-        CHECKED_MALLOC(nalunits[count], NALUnitEBSP, 1);
-        nalunits[count]->init(nalu);
-        count++;
-    }
-
-    if (m_cfg->m_displayOrientationSEIAngle)
-    {
-        SEIDisplayOrientation sei;
-        sei.cancelFlag = false;
-        sei.horFlip = false;
-        sei.verFlip = false;
-        sei.anticlockwiseRotation = m_cfg->m_displayOrientationSEIAngle;
-
-        nalu.resetToType(NAL_UNIT_PREFIX_SEI);
-        entropyCoder->setBitstream(&nalu.m_bitstream);
-        m_seiWriter.writeSEImessage(nalu.m_bitstream, sei, &m_sps);
-        writeRBSPTrailingBits(nalu.m_bitstream);
-        CHECKED_MALLOC(nalunits[count], NALUnitEBSP, 1);
-        nalunits[count]->init(nalu);
-    }
-    return 0;
-
-fail:
-    return -1;
-}
-
-void FrameEncoder::initSlice(TComPic* pic)
-{
-    m_pic = pic;
-    TComSlice* slice = pic->getSlice();
-    slice->setSPS(&m_sps);
-    slice->setPPS(&m_pps);
-    slice->setSliceBits(0);
-    slice->setPic(pic);
-    slice->initSlice();
-    slice->setPicOutputFlag(true);
-    int type = pic->m_lowres.sliceType;
-    SliceType sliceType = IS_X265_TYPE_B(type) ? B_SLICE : ((type == X265_TYPE_P) ? P_SLICE : I_SLICE);
-    slice->setSliceType(sliceType);
-    slice->setReferenced(true);
-    slice->setScalingList(m_top->getScalingList());
-    slice->getScalingList()->setUseTransformSkip(m_pps.getUseTransformSkip());
-#if LOG_CU_STATISTICS
-    for (int i = 0; i < m_numRows; i++)
-    {
-        m_rows[i].m_cuCoder.m_log = &m_rows[i].m_cuCoder.m_sliceTypeLog[sliceType];
-    }
-
-#endif
-    if (slice->getPPS()->getDeblockingFilterControlPresentFlag())
-    {
-        slice->getPPS()->setDeblockingFilterOverrideEnabledFlag(!m_cfg->m_loopFilterOffsetInPPS);
-        slice->setDeblockingFilterOverrideFlag(!m_cfg->m_loopFilterOffsetInPPS);
-        slice->getPPS()->setPicDisableDeblockingFilterFlag(!m_cfg->param->bEnableLoopFilter);
-        slice->setDeblockingFilterDisable(!m_cfg->param->bEnableLoopFilter);
-        if (!slice->getDeblockingFilterDisable())
-        {
-            slice->getPPS()->setDeblockingFilterBetaOffsetDiv2(m_cfg->m_loopFilterBetaOffset);
-            slice->getPPS()->setDeblockingFilterTcOffsetDiv2(m_cfg->m_loopFilterTcOffset);
-            slice->setDeblockingFilterBetaOffsetDiv2(m_cfg->m_loopFilterBetaOffset);
-            slice->setDeblockingFilterTcOffsetDiv2(m_cfg->m_loopFilterTcOffset);
-        }
-    }
-    else
-    {
-        slice->setDeblockingFilterOverrideFlag(false);
-        slice->setDeblockingFilterDisable(false);
-        slice->setDeblockingFilterBetaOffsetDiv2(0);
-        slice->setDeblockingFilterTcOffsetDiv2(0);
-    }
-
-    slice->setMaxNumMergeCand(m_cfg->param->maxNumMergeCand);
+    m_frame = pic;
+    pic->m_picSym->m_slice->m_mref = m_mref;
+    m_enable.trigger();
 }
 
 void FrameEncoder::threadMain()
@@ -313,636 +149,360 @@
     while (m_threadActive);
 }
 
-void FrameEncoder::setLambda(int qp, int row)
-{
-    TComSlice*  slice = m_pic->getSlice();
-    TComPicYuv* fenc  = slice->getPic()->getPicYuvOrg();
-    int         chFmt = slice->getSPS()->getChromaFormatIdc();
-
-    double lambda = 0;
-
-    if (m_pic->getSlice()->getSliceType() == I_SLICE)
-    {
-        lambda = X265_MAX(1, x265_lambda2_tab_I[qp]);
-    }
-    else
-    {
-        lambda = X265_MAX(1, x265_lambda2_non_I[qp]);
-    }
-
-    // for RDO
-    // in RdCost there is only one lambda because the luma and chroma bits are not separated,
-    // instead we weight the distortion of chroma.
-    int chromaQPOffset = slice->getPPS()->getChromaCbQpOffset() + slice->getSliceQpDeltaCb();
-    int qpc = Clip3(0, MAX_MAX_QP, qp + chromaQPOffset);
-    double cbWeight = pow(2.0, (qp - g_chromaScale[chFmt][qpc])); // takes into account of the chroma qp mapping and chroma qp Offset
-
-    chromaQPOffset = slice->getPPS()->getChromaCrQpOffset() + slice->getSliceQpDeltaCr();
-    qpc = Clip3(0, MAX_MAX_QP, qp + chromaQPOffset);
-    double crWeight = pow(2.0, (qp - g_chromaScale[chFmt][qpc])); // takes into account of the chroma qp mapping and chroma qp Offset
-    double chromaLambda = lambda / crWeight;
-
-    m_rows[row].m_search.setQPLambda(qp, lambda, chromaLambda);
-    m_rows[row].m_search.m_me.setSourcePlane(fenc->getLumaAddr(), fenc->getStride());
-    m_rows[row].m_rdCost.setLambda(lambda);
-    m_rows[row].m_rdCost.setCbDistortionWeight(cbWeight);
-    m_rows[row].m_rdCost.setCrDistortionWeight(crWeight);
-}
-
 void FrameEncoder::compressFrame()
 {
     PPAScopeEvent(FrameEncoder_compressFrame);
-    int64_t      startCompressTime = x265_mdate();
-    TEncEntropy* entropyCoder      = getEntropyCoder(0);
-    TComSlice*   slice             = m_pic->getSlice();
-    int          chFmt             = slice->getSPS()->getChromaFormatIdc();
-
-    m_nalCount = 0;
-
-    int qp = slice->getSliceQp();
-    double lambda = 0;
-    if (slice->getSliceType() == I_SLICE)
-    {
-        lambda = X265_MAX(1, x265_lambda2_tab_I[qp]);
-    }
+    int64_t startCompressTime = x265_mdate();
+    Slice* slice = m_frame->m_picSym->m_slice;
+
+    /* Emit access unit delimiter unless this is the first frame and the user is
+     * not repeating headers (since AUD is supposed to be the first NAL in the access
+     * unit) */
+    if (m_param->bEnableAccessUnitDelimiters && (m_frame->getPOC() || m_param->bRepeatHeaders))
+    {
+        m_bs.resetBits();
+        m_entropyCoder.setBitstream(&m_bs);
+        m_entropyCoder.codeAUD(slice);
+        m_bs.writeByteAlignment();
+        m_nalList.serialize(NAL_UNIT_ACCESS_UNIT_DELIMITER, m_bs);
+    }
+    if (m_frame->m_lowres.bKeyframe && m_param->bRepeatHeaders)
+        m_top->getStreamHeaders(m_nalList, m_entropyCoder, m_bs);
+
+    // Weighted Prediction parameters estimation.
+    bool bUseWeightP = slice->m_sliceType == P_SLICE && slice->m_pps->bUseWeightPred;
+    bool bUseWeightB = slice->m_sliceType == B_SLICE && slice->m_pps->bUseWeightedBiPred;
+    if (bUseWeightP || bUseWeightB)
+        weightAnalyse(*slice, *m_param);
     else
-    {
-        lambda = X265_MAX(1, x265_lambda2_non_I[qp]);
-    }
-
-    // for RDO
-    // in RdCost there is only one lambda because the luma and chroma bits are not separated,
-    // instead we weight the distortion of chroma.
-    int qpc;
-    int chromaQPOffset = slice->getPPS()->getChromaCbQpOffset() + slice->getSliceQpDeltaCb();
-    qpc = Clip3(0, MAX_MAX_QP, qp + chromaQPOffset);
-    double cbWeight = pow(2.0, (qp - g_chromaScale[chFmt][qpc])); // takes into account of the chroma qp mapping and chroma qp Offset
-    chromaQPOffset = slice->getPPS()->getChromaCrQpOffset() + slice->getSliceQpDeltaCr();
-    qpc = Clip3(0, MAX_MAX_QP, qp + chromaQPOffset);
-    double crWeight = pow(2.0, (qp - g_chromaScale[chFmt][qpc])); // takes into account of the chroma qp mapping and chroma qp Offset
-    double chromaLambda = lambda / crWeight;
-
-    // NOTE: set SAO lambda every Frame
-    m_frameFilter.m_sao.lumaLambda = lambda;
-    m_frameFilter.m_sao.chromaLambda = chromaLambda;
-
-    TComPicYuv *fenc = slice->getPic()->getPicYuvOrg();
-    for (int i = 0; i < m_numRows; i++)
-    {
-        m_rows[i].m_search.setQPLambda(qp, lambda, chromaLambda);
-        m_rows[i].m_search.m_me.setSourcePlane(fenc->getLumaAddr(), fenc->getStride());
-        m_rows[i].m_rdCost.setLambda(lambda);
-        m_rows[i].m_rdCost.setCbDistortionWeight(cbWeight);
-        m_rows[i].m_rdCost.setCrDistortionWeight(crWeight);
-    }
-
-    m_frameFilter.m_sao.lumaLambda = lambda;
-    m_frameFilter.m_sao.chromaLambda = chromaLambda;
-
-    switch (slice->getSliceType())
-    {
-    case I_SLICE:
-        m_frameFilter.m_sao.depth = 0;
-        break;
-    case P_SLICE:
-        m_frameFilter.m_sao.depth = 1;
-        break;
-    case B_SLICE:
-        m_frameFilter.m_sao.depth = 2 + !slice->isReferenced();
-        break;
-    }
-
-    /* Clip qps back to 0-51 range before encoding */
-    qp = Clip3(-QP_BD_OFFSET, MAX_QP, qp);
-    slice->setSliceQp(qp);
-    m_pic->m_avgQpAq = qp;
-    slice->setSliceQpDelta(0);
-    slice->setSliceQpDeltaCb(0);
-    slice->setSliceQpDeltaCr(0);
-
-    int numSubstreams = m_cfg->param->bEnableWavefront ? m_pic->getPicSym()->getFrameHeightInCU() : 1;
-    // TODO: these two items can likely be FrameEncoder member variables to avoid re-allocs
-    TComOutputBitstream*  bitstreamRedirect = new TComOutputBitstream;
-    TComOutputBitstream*  outStreams = new TComOutputBitstream[numSubstreams];
-
-    slice->setSliceSegmentBits(0);
-    determineSliceBounds();
-    slice->setNextSlice(false);
-
-    //------------------------------------------------------------------------------
-    //  Weighted Prediction parameters estimation.
-    //------------------------------------------------------------------------------
-    bool bUseWeightP = slice->getSliceType() == P_SLICE && slice->getPPS()->getUseWP();
-    bool bUseweightB = slice->getSliceType() == B_SLICE && slice->getPPS()->getWPBiPred();
-    if (bUseWeightP || bUseweightB)
-    {
-        assert(slice->getPPS()->getUseWP());
-        weightAnalyse(*slice, *m_cfg->param);
-    }
+        slice->disableWeights();
 
     // Generate motion references
     int numPredDir = slice->isInterP() ? 1 : slice->isInterB() ? 2 : 0;
     for (int l = 0; l < numPredDir; l++)
     {
-        for (int ref = 0; ref < slice->getNumRefIdx(l); ref++)
+        for (int ref = 0; ref < slice->m_numRefIdx[l]; ref++)
         {
-            wpScalingParam *w = NULL;
-            if (bUseWeightP && slice->m_weightPredTable[l][ref][0].bPresentFlag)
+            WeightParam *w = NULL;
+            if ((bUseWeightP || bUseWeightB) && slice->m_weightPredTable[l][ref][0].bPresentFlag)
                 w = slice->m_weightPredTable[l][ref];
-            m_mref[l][ref].init(slice->getRefPic(l, ref)->getPicYuvRec(), w);
+            m_mref[l][ref].init(slice->m_refPicList[l][ref]->getPicYuvRec(), w);
         }
     }
 
-    // Analyze CTU rows, most of the hard work is done here
-    // frame is compressed in a wave-front pattern if WPP is enabled. Loop filter runs as a
-    // wave-front behind the CU compression and reconstruction
-    compressCTURows();
+    /* Get the QP for this frame from rate control. This call may block until
+     * frames ahead of it in encode order have called rateControlEnd() */
+    int qp = m_top->m_rateControl->rateControlStart(m_frame, &m_rce, m_top);
+    m_rce.newQp = qp;
 
-    if (m_cfg->param->bEnableWavefront)
-    {
-        slice->setNextSlice(true);
-    }
+    /* Clip slice QP to 0-51 spec range before encoding */
+    slice->m_sliceQp = Clip3(-QP_BD_OFFSET, QP_MAX_SPEC, qp);
 
-    if ((m_cfg->m_recoveryPointSEIEnabled) && (slice->getSliceType() == I_SLICE))
-    {
-        if (m_cfg->m_gradualDecodingRefreshInfoEnabled && !slice->getRapPicFlag())
-        {
-            // Gradual decoding refresh SEI
-            OutputNALUnit nalu(NAL_UNIT_PREFIX_SEI);
+    m_initSliceContext.resetEntropy(slice);
 
-            SEIGradualDecodingRefreshInfo seiGradualDecodingRefreshInfo;
-            seiGradualDecodingRefreshInfo.m_gdrForegroundFlag = true; // Indicating all "foreground"
+    m_frameFilter.start(m_frame, m_initSliceContext, qp);
 
-            m_seiWriter.writeSEImessage(nalu.m_bitstream, seiGradualDecodingRefreshInfo, slice->getSPS());
-            writeRBSPTrailingBits(nalu.m_bitstream);
-            m_nalList[m_nalCount] = X265_MALLOC(NALUnitEBSP, 1);
-            if (m_nalList[m_nalCount])
-            {
-                m_nalList[m_nalCount]->init(nalu);
-                m_nalCount++;
-            }
-        }
-        // Recovery point SEI
-        OutputNALUnit nalu(NAL_UNIT_PREFIX_SEI);
-
-        SEIRecoveryPoint sei_recovery_point;
-        sei_recovery_point.m_recoveryPocCnt    = 0;
-        sei_recovery_point.m_exactMatchingFlag = (slice->getPOC() == 0) ? (true) : (false);
-        sei_recovery_point.m_brokenLinkFlag    = false;
-
-        m_seiWriter.writeSEImessage(nalu.m_bitstream, sei_recovery_point, slice->getSPS());
-        writeRBSPTrailingBits(nalu.m_bitstream);
-        m_nalList[m_nalCount] = X265_MALLOC(NALUnitEBSP, 1);
-        if (m_nalList[m_nalCount])
-        {
-            m_nalList[m_nalCount]->init(nalu);
-            m_nalCount++;
-        }
-    }
+    // reset entropy coders
+    m_entropyCoder.load(m_initSliceContext);
+    for (int i = 0; i < m_numRows; i++)
+        m_rows[i].init(m_initSliceContext);
 
-    if (!!m_cfg->param->interlaceMode)
+    uint32_t numSubstreams = m_param->bEnableWavefront ? m_frame->getPicSym()->getFrameHeightInCU() : 1;
+    if (!m_outStreams)
     {
-        OutputNALUnit nalu(NAL_UNIT_PREFIX_SEI);
-
-        SEIPictureTiming sei;
-        sei.m_picStruct = (slice->getPOC() & 1) && m_cfg->param->interlaceMode == 2 ? 1 /* top */ : 2 /* bot */;
-        sei.m_sourceScanType = 1;
-        sei.m_duplicateFlag = 0;
-
-        entropyCoder->setBitstream(&nalu.m_bitstream);
-        m_seiWriter.writeSEImessage(nalu.m_bitstream, sei, &m_sps);
-        writeRBSPTrailingBits(nalu.m_bitstream);
-        m_nalList[m_nalCount] = X265_MALLOC(NALUnitEBSP, 1);
-        if (m_nalList[m_nalCount])
-        {
-            m_nalList[m_nalCount]->init(nalu);
-            m_nalCount++;
-        }
+        m_outStreams = new Bitstream[numSubstreams];
+        m_substreamSizes = X265_MALLOC(uint32_t, numSubstreams);
+        if (!m_param->bEnableSAO)
+            for (uint32_t i = 0; i < numSubstreams; i++)
+                m_rows[i].rdEntropyCoders[0][CI_CURR_BEST].setBitstream(&m_outStreams[i]);
     }
+    else
+        for (uint32_t i = 0; i < numSubstreams; i++)
+            m_outStreams[i].resetBits();
 
-    /* use the main bitstream buffer for storing the marshaled picture */
-    if (m_sps.getUseSAO())
+    if (m_frame->m_lowres.bKeyframe)
     {
-        SAOParam* saoParam = m_pic->getPicSym()->getSaoParam();
-
-        if (!getSAO()->getSaoLcuBasedOptimization())
+        if (m_param->bEmitHRDSEI)
         {
-            getSAO()->SAOProcess(saoParam);
-            getSAO()->endSaoEnc();
-            PCMLFDisableProcess(m_pic);
+            SEIBufferingPeriod* bpSei = &m_top->m_rateControl->m_bufPeriodSEI;
 
-            // Extend border after whole-frame SAO is finished
-            for (int row = 0; row < m_numRows; row++)
-            {
-                m_frameFilter.processRowPost(row, m_cfg);
-            }
-        }
+            // since the temporal layer HRD is not ready, we assumed it is fixed
+            bpSei->m_auCpbRemovalDelayDelta = 1;
+            bpSei->m_cpbDelayOffset = 0;
+            bpSei->m_dpbDelayOffset = 0;
 
-        slice->setSaoEnabledFlag((saoParam->bSaoFlag[0] == 1) ? true : false);
-    }
+            // hrdFullness() calculates the initial CPB removal delay and offset
+            m_top->m_rateControl->hrdFullness(bpSei);
 
-    entropyCoder->setBitstream(NULL);
+            m_bs.resetBits();
+            bpSei->write(m_bs, *slice->m_sps);
+            m_bs.writeByteAlignment();
 
-    // Reconstruction slice
-    slice->setNextSlice(true);
-    determineSliceBounds();
+            m_nalList.serialize(NAL_UNIT_PREFIX_SEI, m_bs);
 
-    slice->allocSubstreamSizes(numSubstreams);
-    for (int i = 0; i < numSubstreams; i++)
-    {
-        outStreams[i].clear();
-    }
+            m_top->m_lastBPSEI = m_rce.encodeOrder;
+        }
 
-    entropyCoder->setEntropyCoder(&m_sbacCoder, slice);
-    entropyCoder->resetEntropy();
+        // The recovery point SEI message assists a decoder in determining when the decoding
+        // process will produce acceptable pictures for display after the decoder initiates
+        // random access. The m_recoveryPocCnt is in units of POC(picture order count) which
+        // means pictures encoded after the CRA but precede it in display order(leading) are
+        // implicitly discarded after a random access seek regardless of the value of
+        // m_recoveryPocCnt. Our encoder does not use references prior to the most recent CRA,
+        // so all pictures following the CRA in POC order are guaranteed to be displayable,
+        // so m_recoveryPocCnt is always 0.
+        SEIRecoveryPoint sei_recovery_point;
+        sei_recovery_point.m_recoveryPocCnt = 0;
+        sei_recovery_point.m_exactMatchingFlag = true;
+        sei_recovery_point.m_brokenLinkFlag = false;
 
-    /* start slice NALunit */
-    bool sliceSegment = !slice->isNextSlice();
-    OutputNALUnit nalu(slice->getNalUnitType());
-    entropyCoder->setBitstream(&nalu.m_bitstream);
-    entropyCoder->encodeSliceHeader(slice);
+        m_bs.resetBits();
+        sei_recovery_point.write(m_bs, *slice->m_sps);
+        m_bs.writeByteAlignment();
 
-    // is it needed?
-    if (!sliceSegment)
-    {
-        bitstreamRedirect->writeAlignOne();
-    }
-    else
-    {
-        // We've not completed our slice header info yet, do the alignment later.
+        m_nalList.serialize(NAL_UNIT_PREFIX_SEI, m_bs);
     }
 
-    m_sbacCoder.init(&m_binCoderCABAC);
-    entropyCoder->setEntropyCoder(&m_sbacCoder, slice);
-    entropyCoder->resetEntropy();
-    resetEntropy(slice);
-
-    if (slice->isNextSlice())
+    if (m_param->bEmitHRDSEI || !!m_param->interlaceMode)
     {
-        // set entropy coder for writing
-        m_sbacCoder.init(&m_binCoderCABAC);
-        resetEntropy(slice);
-        getSbacCoder(0)->load(&m_sbacCoder);
-
-        //ALF is written in substream #0 with CABAC coder #0 (see ALF param encoding below)
-        entropyCoder->setEntropyCoder(getSbacCoder(0), slice);
-        entropyCoder->resetEntropy();
+        SEIPictureTiming *sei = m_rce.picTimingSEI;
+        const VUI *vui = &slice->m_sps->vuiParameters;
+        const HRDInfo *hrd = &vui->hrdParameters;
+        int poc = slice->m_poc;
 
-        // File writing
-        if (!sliceSegment)
+        if (vui->frameFieldInfoPresentFlag)
         {
-            entropyCoder->setBitstream(bitstreamRedirect);
+            if (m_param->interlaceMode == 2)
+                sei->m_picStruct = (poc & 1) ? 1 /* top */ : 2 /* bottom */;
+            else if (m_param->interlaceMode == 1)
+                sei->m_picStruct = (poc & 1) ? 2 /* bottom */ : 1 /* top */;
+            else
+                sei->m_picStruct = 0;
+            sei->m_sourceScanType = 0;
+            sei->m_duplicateFlag = false;
         }
-        else
+
+        if (vui->hrdParametersPresentFlag)
         {
-            entropyCoder->setBitstream(&nalu.m_bitstream);
+            // The m_aucpbremoval delay specifies how many clock ticks the
+            // access unit associated with the picture timing SEI message has to
+            // wait after removal of the access unit with the most recent
+            // buffering period SEI message
+            sei->m_auCpbRemovalDelay = X265_MIN(X265_MAX(1, m_rce.encodeOrder - m_top->m_lastBPSEI), (1 << hrd->cpbRemovalDelayLength));
+            sei->m_picDpbOutputDelay = slice->m_sps->numReorderPics + poc - m_rce.encodeOrder;
         }
 
-        // for now, override the TILES_DECODER setting in order to write substreams.
-        entropyCoder->setBitstream(&outStreams[0]);
+        m_bs.resetBits();
+        sei->write(m_bs, *slice->m_sps);
+        m_bs.writeByteAlignment();
+        m_nalList.serialize(NAL_UNIT_PREFIX_SEI, m_bs);
     }
-    slice->setFinalized(true);
-
-    m_sbacCoder.load(getSbacCoder(0));
 
-    slice->setTileOffstForMultES(0);
-    encodeSlice(outStreams);
+    // Analyze CTU rows, most of the hard work is done here
+    // frame is compressed in a wave-front pattern if WPP is enabled. Loop filter runs as a
+    // wave-front behind the CU compression and reconstruction
+    compressCTURows();
 
+    if (m_param->rc.bStatWrite)
     {
-        // Construct the final bitstream by flushing and concatenating substreams.
-        // The final bitstream is either nalu.m_bitstream or pcBitstreamRedirect;
-        uint32_t* substreamSizes = slice->getSubstreamSizes();
-        for (int i = 0; i < numSubstreams; i++)
-        {
-            // Flush all substreams -- this includes empty ones.
-            // Terminating bit and flush.
-            entropyCoder->setEntropyCoder(getSbacCoder(i), slice);
-            entropyCoder->setBitstream(&outStreams[i]);
-            entropyCoder->encodeTerminatingBit(1);
-            entropyCoder->encodeSliceFinish();
-
-            outStreams[i].writeByteAlignment(); // Byte-alignment in slice_data() at end of sub-stream
+        int totalI = 0, totalP = 0, totalSkip = 0;
 
-            // Byte alignment is necessary between tiles when tiles are independent.
-            if (i + 1 < numSubstreams)
-            {
-                substreamSizes[i] = outStreams[i].getNumberOfWrittenBits() + (outStreams[i].countStartCodeEmulations() << 3);
-            }
-        }
-
-        // Complete the slice header info.
-        entropyCoder->setEntropyCoder(&m_sbacCoder, slice);
-        entropyCoder->setBitstream(&nalu.m_bitstream);
-        entropyCoder->encodeTilesWPPEntryPoint(slice);
-
-        // Substreams...
-        int nss = m_pps.getEntropyCodingSyncEnabledFlag() ? slice->getNumEntryPointOffsets() + 1 : numSubstreams;
-        for (int i = 0; i < nss; i++)
+        // accumulate intra,inter,skip cu count per frame for 2 pass
+        for (int i = 0; i < m_numRows; i++)
         {
-            bitstreamRedirect->addSubstream(&outStreams[i]);
+            m_frameStats.mvBits    += m_rows[i].rowStats.mvBits;
+            m_frameStats.coeffBits += m_rows[i].rowStats.coeffBits;
+            m_frameStats.miscBits  += m_rows[i].rowStats.miscBits;
+            totalI                 += m_rows[i].rowStats.iCuCnt;
+            totalP                 += m_rows[i].rowStats.pCuCnt;
+            totalSkip              += m_rows[i].rowStats.skipCuCnt;
         }
+        int totalCuCount = totalI + totalP + totalSkip;
+        m_frameStats.percentIntra = (double)totalI / totalCuCount;
+        m_frameStats.percentInter = (double)totalP / totalCuCount;
+        m_frameStats.percentSkip  = (double)totalSkip / totalCuCount;
     }
 
-    // If current NALU is the first NALU of slice (containing slice header) and
-    // more NALUs exist (due to multiple dependent slices) then buffer it.  If
-    // current NALU is the last NALU of slice and a NALU was buffered, then (a)
-    // Write current NALU (b) Update an write buffered NALU at appropriate
-    // location in NALU list.
-    nalu.m_bitstream.writeByteAlignment(); // Slice header byte-alignment
-
-    // Perform bitstream concatenation
-    if (bitstreamRedirect->getNumberOfWrittenBits() > 0)
+    if (slice->m_sps->bUseSAO && !m_param->saoLcuBasedOptimization)
     {
-        nalu.m_bitstream.addSubstream(bitstreamRedirect);
-    }
-    entropyCoder->setBitstream(&nalu.m_bitstream);
-    bitstreamRedirect->clear();
-    m_nalList[m_nalCount] = X265_MALLOC(NALUnitEBSP, 1);
-    if (m_nalList[m_nalCount])
-    {
-        m_nalList[m_nalCount]->init(nalu);
-        m_nalCount++;
+        /* frame based SAO */
+        m_frameFilter.m_sao.SAOProcess(m_frame->getPicSym()->m_saoParam);
+        restoreLFDisabledOrigYuv(m_frame);
+
+        // Extend border after whole-frame SAO is finished
+        for (int row = 0; row < m_numRows; row++)
+            m_frameFilter.processRowPost(row);
     }
 
-    /* write decoded picture hash SEI messages */
-    if (m_cfg->param->decodedPictureHashSEI)
+    m_bs.resetBits();
+    m_entropyCoder.load(m_initSliceContext);
+    m_entropyCoder.setBitstream(&m_bs);
+    m_entropyCoder.codeSliceHeader(slice);
+
+    // finish encode of each CTU row, only required when SAO is enabled
+    if (m_param->bEnableSAO)
+        encodeSlice();
+
+    // serialize each row, record final lengths in slice header
+    uint32_t maxStreamSize = m_nalList.serializeSubstreams(m_substreamSizes, numSubstreams, m_outStreams);
+
+    // complete the slice header by writing WPP row-starts
+    m_entropyCoder.setBitstream(&m_bs);
+    if (slice->m_pps->bEntropyCodingSyncEnabled)
+        m_entropyCoder.codeSliceHeaderWPPEntryPoints(slice, m_substreamSizes, maxStreamSize);
+    m_bs.writeByteAlignment();
+
+    m_nalList.serialize(slice->m_nalUnitType, m_bs);
+
+    if (m_param->decodedPictureHashSEI)
     {
-        if (m_cfg->param->decodedPictureHashSEI == 1)
+        if (m_param->decodedPictureHashSEI == 1)
         {
-            m_seiReconPictureDigest.method = SEIDecodedPictureHash::MD5;
+            m_seiReconPictureDigest.m_method = SEIDecodedPictureHash::MD5;
             for (int i = 0; i < 3; i++)
-            {
-                MD5Final(&(m_pic->m_state[i]), m_seiReconPictureDigest.digest[i]);
-            }
+                MD5Final(&m_state[i], m_seiReconPictureDigest.m_digest[i]);
         }
-        else if (m_cfg->param->decodedPictureHashSEI == 2)
+        else if (m_param->decodedPictureHashSEI == 2)
         {
-            m_seiReconPictureDigest.method = SEIDecodedPictureHash::CRC;
+            m_seiReconPictureDigest.m_method = SEIDecodedPictureHash::CRC;
             for (int i = 0; i < 3; i++)
-            {
-                crcFinish((m_pic->m_crc[i]), m_seiReconPictureDigest.digest[i]);
-            }
+                crcFinish(m_crc[i], m_seiReconPictureDigest.m_digest[i]);
         }
-        else if (m_cfg->param->decodedPictureHashSEI == 3)
+        else if (m_param->decodedPictureHashSEI == 3)
         {
-            m_seiReconPictureDigest.method = SEIDecodedPictureHash::CHECKSUM;
+            m_seiReconPictureDigest.m_method = SEIDecodedPictureHash::CHECKSUM;
             for (int i = 0; i < 3; i++)
-            {
-                checksumFinish(m_pic->m_checksum[i], m_seiReconPictureDigest.digest[i]);
-            }
+                checksumFinish(m_checksum[i], m_seiReconPictureDigest.m_digest[i]);
         }
-        OutputNALUnit onalu(NAL_UNIT_SUFFIX_SEI);
-        m_seiWriter.writeSEImessage(onalu.m_bitstream, m_seiReconPictureDigest, slice->getSPS());
-        writeRBSPTrailingBits(onalu.m_bitstream);
 
-        m_nalList[m_nalCount] = X265_MALLOC(NALUnitEBSP, 1);
-        if (m_nalList[m_nalCount])
-        {
-            m_nalList[m_nalCount]->init(onalu);
-            m_nalCount++;
-        }
+        m_bs.resetBits();
+        m_seiReconPictureDigest.write(m_bs, *slice->m_sps);
+        m_bs.writeByteAlignment();
+
+        m_nalList.serialize(NAL_UNIT_SUFFIX_SEI, m_bs);
     }
 
-    if (m_sps.getUseSAO())
+    uint64_t bytes = 0;
+    for (uint32_t i = 0; i < m_nalList.m_numNal; i++)
     {
-        m_frameFilter.end();
+        int type = m_nalList.m_nal[i].type;
+
+        // exclude SEI
+        if (type != NAL_UNIT_PREFIX_SEI && type != NAL_UNIT_SUFFIX_SEI)
+        {
+            bytes += m_nalList.m_nal[i].sizeBytes;
+            // and exclude start code prefix
+            bytes -= (!i || type == NAL_UNIT_SPS || type == NAL_UNIT_PPS) ? 4 : 3;
+        }
     }
+    m_accessUnitBits = bytes << 3;
+
+    m_elapsedCompressTime = (double)(x265_mdate() - startCompressTime) / 1000000;
+    /* rateControlEnd may also block for earlier frames to call rateControlUpdateStats */
+    if (m_top->m_rateControl->rateControlEnd(m_frame, m_accessUnitBits, &m_rce, &m_frameStats) < 0)
+        m_top->m_aborted = true;
 
-    /* Decrement referenced frame reference counts, allow them to be recycled */
+    noiseReductionUpdate();
+
+    // Decrement referenced frame reference counts, allow them to be recycled
     for (int l = 0; l < numPredDir; l++)
     {
-        for (int ref = 0; ref < slice->getNumRefIdx(l); ref++)
+        for (int ref = 0; ref < slice->m_numRefIdx[l]; ref++)
         {
-            TComPic *refpic = slice->getRefPic(l, ref);
+            Frame *refpic = slice->m_refPicList[l][ref];
             ATOMIC_DEC(&refpic->m_countRefEncoders);
         }
     }
-
-    m_pic->m_elapsedCompressTime = (double)(x265_mdate() - startCompressTime) / 1000000;
-    delete[] outStreams;
-    delete bitstreamRedirect;
 }
 
-void FrameEncoder::encodeSlice(TComOutputBitstream* substreams)
+void FrameEncoder::encodeSlice()
 {
-    // choose entropy coder
-    TEncEntropy *entropyCoder = getEntropyCoder(0);
-    TComSlice* slice = m_pic->getSlice();
-
-    // Initialize slice singletons
-    m_sbacCoder.init(&m_binCoderCABAC);
-    getCuEncoder(0)->setBitCounter(NULL);
-    entropyCoder->setEntropyCoder(&m_sbacCoder, slice);
-
-    uint32_t cuAddr;
-    uint32_t startCUAddr = 0;
-    uint32_t boundingCUAddr = slice->getSliceCurEndCUAddr();
-
-    // Appropriate substream bitstream is switched later.
-    // for every CU
-#if ENC_DEC_TRACE
-    g_bJustDoIt = g_bEncDecTraceEnable;
-#endif
-    DTRACE_CABAC_VL(g_nSymbolCounter++);
-    DTRACE_CABAC_T("\tPOC: ");
-    DTRACE_CABAC_V(m_pic->getPOC());
-    DTRACE_CABAC_T("\n");
-#if ENC_DEC_TRACE
-    g_bJustDoIt = g_bEncDecTraceDisable;
-#endif
-
-    const int  bWaveFrontsynchro = m_cfg->param->bEnableWavefront;
-    const uint32_t heightInLCUs = m_pic->getPicSym()->getFrameHeightInCU();
-    const int  numSubstreams = (bWaveFrontsynchro ? heightInLCUs : 1);
-    uint32_t bitsOriginallyInSubstreams = 0;
-
-    for (int substrmIdx = 0; substrmIdx < numSubstreams; substrmIdx++)
-    {
-        getBufferSBac(substrmIdx)->loadContexts(&m_sbacCoder); //init. state
-        bitsOriginallyInSubstreams += substreams[substrmIdx].getNumberOfWrittenBits();
-    }
+    Slice* slice = m_frame->m_picSym->m_slice;
+    const uint32_t widthInLCUs = m_frame->getPicSym()->getFrameWidthInCU();
+    const uint32_t lastCUAddr = (slice->m_endCUAddr + NUM_CU_PARTITIONS - 1) / NUM_CU_PARTITIONS;
+    const int numSubstreams = m_param->bEnableWavefront ? m_frame->getPicSym()->getFrameHeightInCU() : 1;
 
-    uint32_t widthInLCUs = m_pic->getPicSym()->getFrameWidthInCU();
-    uint32_t col = 0, lin = 0, subStrm = 0;
-    cuAddr = (startCUAddr / m_pic->getNumPartInCU()); /* for tiles, startCUAddr is NOT the real raster scan address, it is actually
-                                                       an encoding order index, so we need to convert the index (startCUAddr)
-                                                       into the real raster scan address (cuAddr) via the CUOrderMap */
-    uint32_t encCUOrder;
-    for (encCUOrder = startCUAddr / m_pic->getNumPartInCU();
-         encCUOrder < (boundingCUAddr + m_pic->getNumPartInCU() - 1) / m_pic->getNumPartInCU();
-         cuAddr = ++encCUOrder)
+    SAOParam *saoParam = slice->m_pic->getPicSym()->m_saoParam;
+    for (uint32_t cuAddr = 0; cuAddr < lastCUAddr; cuAddr++)
     {
-        col     = cuAddr % widthInLCUs;
-        lin     = cuAddr / widthInLCUs;
-        subStrm = lin % numSubstreams;
+        uint32_t col = cuAddr % widthInLCUs;
+        uint32_t lin = cuAddr / widthInLCUs;
+        uint32_t subStrm = lin % numSubstreams;
+        TComDataCU* cu = m_frame->getCU(cuAddr);
 
-        entropyCoder->setBitstream(&substreams[subStrm]);
+        m_entropyCoder.setBitstream(&m_outStreams[subStrm]);
 
         // Synchronize cabac probabilities with upper-right LCU if it's available and we're at the start of a line.
-        if ((numSubstreams > 1) && (col == 0) && bWaveFrontsynchro)
+        if (m_param->bEnableWavefront && !col && lin)
         {
-            // We'll sync if the TR is available.
-            TComDataCU *cuUp = m_pic->getCU(cuAddr)->getCUAbove();
-            uint32_t widthInCU = m_pic->getFrameWidthInCU();
-            TComDataCU *cuTr = NULL;
-
-            // CHECK_ME: here can be optimize a little, do it later
-            if (cuUp && ((cuAddr % widthInCU + 1) < widthInCU))
-            {
-                cuTr = m_pic->getCU(cuAddr - widthInCU + 1);
-            }
-            if ( /*bEnforceSliceRestriction &&*/ ((cuTr == NULL) || (cuTr->getSlice() == NULL)))
-            {
-                // TR not available.
-            }
-            else
-            {
-                // TR is available, we use it.
-                getSbacCoder(subStrm)->loadContexts(getBufferSBac(lin - 1));
-            }
+            m_entropyCoder.copyState(m_initSliceContext);
+            m_entropyCoder.loadContexts(m_rows[lin - 1].bufferEntropyCoder);
         }
-        m_sbacCoder.load(getSbacCoder(subStrm)); //this load is used to simplify the code (avoid to change all the call to m_sbacCoder)
 
-        TComDataCU* cu = m_pic->getCU(cuAddr);
-        if (slice->getSPS()->getUseSAO() && (slice->getSaoEnabledFlag() || slice->getSaoEnabledFlagChroma()))
+        if (slice->m_sps->bUseSAO)
         {
-            SAOParam *saoParam = slice->getPic()->getPicSym()->getSaoParam();
-            int numCuInWidth     = saoParam->numCuInWidth;
-            int cuAddrInSlice    = cuAddr;
-            int rx = cuAddr % numCuInWidth;
-            int ry = cuAddr / numCuInWidth;
-            int allowMergeLeft = 1;
-            int allowMergeUp   = 1;
-            int addr = cu->getAddr();
-            allowMergeLeft = (rx > 0) && (cuAddrInSlice != 0);
-            allowMergeUp = (ry > 0) && (cuAddrInSlice >= 0);
             if (saoParam->bSaoFlag[0] || saoParam->bSaoFlag[1])
             {
-                int mergeLeft = saoParam->saoLcuParam[0][addr].mergeLeftFlag;
-                int mergeUp = saoParam->saoLcuParam[0][addr].mergeUpFlag;
-                if (allowMergeLeft)
+                int mergeLeft = saoParam->saoLcuParam[0][cuAddr].mergeLeftFlag && col;
+                int mergeUp = saoParam->saoLcuParam[0][cuAddr].mergeUpFlag && lin;
+                if (col)
+                    m_entropyCoder.codeSaoMerge(mergeLeft);
+                if (lin && !mergeLeft)
+                    m_entropyCoder.codeSaoMerge(mergeUp);
+                if (!mergeLeft && !mergeUp)
                 {
-                    entropyCoder->m_entropyCoderIf->codeSaoMerge(mergeLeft);
-                }
-                else
-                {
-                    mergeLeft = 0;
-                }
-                if (mergeLeft == 0)
-                {
-                    if (allowMergeUp)
-                    {
-                        entropyCoder->m_entropyCoderIf->codeSaoMerge(mergeUp);
-                    }
-                    else
+                    if (saoParam->bSaoFlag[0])
+                        m_entropyCoder.codeSaoOffset(&saoParam->saoLcuParam[0][cuAddr], 0);
+                    if (saoParam->bSaoFlag[1])
                     {
-                        mergeUp = 0;
-                    }
-                    if (mergeUp == 0)
-                    {
-                        for (int compIdx = 0; compIdx < 3; compIdx++)
-                        {
-                            if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
-                            {
-                                entropyCoder->encodeSaoOffset(&saoParam->saoLcuParam[compIdx][addr], compIdx);
-                            }
-                        }
+                        m_entropyCoder.codeSaoOffset(&saoParam->saoLcuParam[1][cuAddr], 1);
+                        m_entropyCoder.codeSaoOffset(&saoParam->saoLcuParam[2][cuAddr], 2);
                     }
                 }
             }
-        }
-        else if (slice->getSPS()->getUseSAO())
-        {
-            int addr = cu->getAddr();
-            SAOParam *saoParam = slice->getPic()->getPicSym()->getSaoParam();
-            for (int cIdx = 0; cIdx < 3; cIdx++)
+            else
             {
-                SaoLcuParam *saoLcuParam = &(saoParam->saoLcuParam[cIdx][addr]);
-                if (((cIdx == 0) && !slice->getSaoEnabledFlag()) || ((cIdx == 1 || cIdx == 2) && !slice->getSaoEnabledFlagChroma()))
-                {
-                    saoLcuParam->mergeUpFlag   = 0;
-                    saoLcuParam->mergeLeftFlag = 0;
-                    saoLcuParam->subTypeIdx    = 0;
-                    saoLcuParam->typeIdx       = -1;
-                    saoLcuParam->offset[0]     = 0;
-                    saoLcuParam->offset[1]     = 0;
-                    saoLcuParam->offset[2]     = 0;
-                    saoLcuParam->offset[3]     = 0;
-                }
+                for (int i = 0; i < 3; i++)
+                    saoParam->saoLcuParam[i][cuAddr].reset();
             }
         }
 
-#if ENC_DEC_TRACE
-        g_bJustDoIt = g_bEncDecTraceEnable;
-#endif
-        getCuEncoder(0)->setEntropyCoder(entropyCoder);
-        getCuEncoder(0)->encodeCU(cu);
-
-#if ENC_DEC_TRACE
-        g_bJustDoIt = g_bEncDecTraceDisable;
-#endif
-
-        // load back status of the entropy coder after encoding the LCU into relevant bitstream entropy coder
-        getSbacCoder(subStrm)->load(&m_sbacCoder);
+        // final coding (bitstream generation) for this CU
+        m_entropyCoder.encodeCTU(cu);
 
-        // Store probabilities of second LCU in line into buffer
-        if ((numSubstreams > 1) && (col == 1) && bWaveFrontsynchro)
+        if (m_param->bEnableWavefront)
         {
-            getBufferSBac(lin)->loadContexts(getSbacCoder(subStrm));
-        }
-    }
+            if (col == 1)
+                // Store probabilities of second LCU in line into buffer
+                m_rows[lin].bufferEntropyCoder.loadContexts(m_entropyCoder);
 
-    if (slice->getPPS()->getCabacInitPresentFlag())
-    {
-        entropyCoder->determineCabacInitIdx();
+            if (col == widthInLCUs - 1)
+                m_entropyCoder.finishSlice();
+        }
     }
-}
-
-/** Determines the starting and bounding LCU address of current slice / dependent slice
- * \returns Updates startCUAddr, boundingCUAddr with appropriate LCU address
- */
-void FrameEncoder::determineSliceBounds()
-{
-    uint32_t numberOfCUsInFrame = m_pic->getNumCUsInFrame();
-    uint32_t boundingCUAddrSlice = numberOfCUsInFrame * m_pic->getNumPartInCU();
-
-    // WPP: if a slice does not start at the beginning of a CTB row, it must end within the same CTB row
-    m_pic->getSlice()->setSliceCurEndCUAddr(boundingCUAddrSlice);
+    if (!m_param->bEnableWavefront)
+        m_entropyCoder.finishSlice();
 }
 
 void FrameEncoder::compressCTURows()
 {
     PPAScopeEvent(FrameEncoder_compressRows);
-    TComSlice* slice = m_pic->getSlice();
-
-    // reset entropy coders
-    m_sbacCoder.init(&m_binCoderCABAC);
-    for (int i = 0; i < this->m_numRows; i++)
-    {
-        m_rows[i].init(slice);
-        m_rows[i].m_entropyCoder.setEntropyCoder(&m_sbacCoder, slice);
-        m_rows[i].m_entropyCoder.resetEntropy();
-
-        m_rows[i].m_rdSbacCoders[0][CI_CURR_BEST]->load(&m_sbacCoder);
-        m_rows[i].m_rdGoOnBinCodersCABAC.m_fracBits = 0;
-        m_rows[i].m_completed = 0;
-        m_rows[i].m_busy = false;
-    }
+    Slice* slice = m_frame->m_picSym->m_slice;
 
-    int range = m_cfg->param->searchRange; /* fpel search */
-    range    += 1;                        /* diamond search range check lag */
-    range    += 2;                        /* subpel refine */
-    range    += NTAPS_LUMA / 2;           /* subpel filter half-length */
-    int refLagRows = 1 + ((range + g_maxCUSize - 1) / g_maxCUSize);
-    int numPredDir = slice->isInterP() ? 1 : slice->isInterB() ? 2 : 0;
+    m_bAllRowsStop = false;
+    m_vbvResetTriggerRow = -1;
 
-    m_pic->m_SSDY = 0;
-    m_pic->m_SSDU = 0;
-    m_pic->m_SSDV = 0;
+    m_SSDY = m_SSDU = m_SSDV = 0;
+    m_ssim = 0;
+    m_ssimCnt = 0;
+    memset(&m_frameStats, 0, sizeof(m_frameStats));
 
-    m_frameFilter.start(m_pic);
+    bool bUseWeightP = slice->m_pps->bUseWeightPred && slice->m_sliceType == P_SLICE;
+    bool bUseWeightB = slice->m_pps->bUseWeightedBiPred && slice->m_sliceType == B_SLICE;
+    int numPredDir = slice->isInterP() ? 1 : slice->isInterB() ? 2 : 0;
 
-    if (m_pool && m_cfg->param->bEnableWavefront)
+    m_rows[0].active = true;
+    if (m_pool && m_param->bEnableWavefront)
     {
         WaveFront::clearEnabledRowMask();
         WaveFront::enqueue();
@@ -952,18 +512,16 @@
             // block until all reference frames have reconstructed the rows we need
             for (int l = 0; l < numPredDir; l++)
             {
-                for (int ref = 0; ref < slice->getNumRefIdx(l); ref++)
+                for (int ref = 0; ref < slice->m_numRefIdx[l]; ref++)
                 {
-                    TComPic *refpic = slice->getRefPic(l, ref);
+                    Frame *refpic = slice->m_refPicList[l][ref];
 
                     int reconRowCount = refpic->m_reconRowCount.get();
-                    while ((reconRowCount != m_numRows) && (reconRowCount < row + refLagRows))
+                    while ((reconRowCount != m_numRows) && (reconRowCount < row + m_refLagRows))
                         reconRowCount = refpic->m_reconRowCount.waitForChange(reconRowCount);
 
-                    if (slice->getPPS()->getUseWP() && slice->getSliceType() == P_SLICE && m_mref[l][ref].isWeighted)
-                    {
-                        m_mref[l][ref].applyWeight(row + refLagRows, m_numRows);
-                    }
+                    if ((bUseWeightP || bUseWeightB) && m_mref[l][ref].isWeighted)
+                        m_mref[l][ref].applyWeight(row + m_refLagRows, m_numRows);
                 }
             }
 
@@ -982,143 +540,329 @@
     {
         for (int i = 0; i < this->m_numRows + m_filterRowDelay; i++)
         {
-            // Encoder
+            // Encode
             if (i < m_numRows)
             {
                 // block until all reference frames have reconstructed the rows we need
                 for (int l = 0; l < numPredDir; l++)
                 {
                     int list = l;
-                    for (int ref = 0; ref < slice->getNumRefIdx(list); ref++)
+                    for (int ref = 0; ref < slice->m_numRefIdx[list]; ref++)
                     {
-                        TComPic *refpic = slice->getRefPic(list, ref);
+                        Frame *refpic = slice->m_refPicList[list][ref];
 
                         int reconRowCount = refpic->m_reconRowCount.get();
-                        while ((reconRowCount != m_numRows) && (reconRowCount < i + refLagRows))
+                        while ((reconRowCount != m_numRows) && (reconRowCount < i + m_refLagRows))
                             reconRowCount = refpic->m_reconRowCount.waitForChange(reconRowCount);
 
-                        if (slice->getPPS()->getUseWP() && slice->getSliceType() == P_SLICE && m_mref[l][ref].isWeighted)
-                        {
-                            m_mref[list][ref].applyWeight(i + refLagRows, m_numRows);
-                        }
+                        if ((bUseWeightP || bUseWeightB) && m_mref[l][ref].isWeighted)
+                            m_mref[list][ref].applyWeight(i + m_refLagRows, m_numRows);
                     }
                 }
 
-                processRow(i * 2 + 0);
+                processRow(i * 2 + 0, -1);
             }
 
             // Filter
             if (i >= m_filterRowDelay)
-            {
-                processRow((i - m_filterRowDelay) * 2 + 1);
-            }
+                processRow((i - m_filterRowDelay) * 2 + 1, -1);
         }
     }
-    m_pic->m_frameTime = (double)m_totalTime / 1000000;
+    m_frameTime = (double)m_totalTime / 1000000;
     m_totalTime = 0;
 }
 
-void FrameEncoder::processRowEncoder(int row)
+void FrameEncoder::processRow(int row, int threadId)
+{
+    const int realRow = row >> 1;
+    const int typeNum = row & 1;
+
+    ThreadLocalData& tld = threadId >= 0 ? m_top->m_threadLocalData[threadId] : *m_tld;
+    
+    if (!typeNum)
+        processRowEncoder(realRow, tld);
+    else
+    {
+        processRowFilter(realRow);
+
+        // NOTE: Active next row
+        if (realRow != m_numRows - 1)
+            enqueueRowFilter(realRow + 1);
+        else
+            m_completionEvent.trigger();
+    }
+}
+
+// Called by worker threads
+void FrameEncoder::processRowEncoder(int row, ThreadLocalData& tld)
 {
     PPAScopeEvent(Thread_ProcessRow);
 
-    // Called by worker threads
-    CTURow& curRow  = m_rows[row];
-    if (curRow.m_busy)
+    CTURow& curRow = m_rows[row];
+
     {
-        /* On multi-socket Windows servers, we have seen problems with
-         * ATOMIC_CAS which resulted in multiple worker threads processing
-         * the same CU row, which often resulted in bad pointer accesses. We
-         * believe the problem is fixed, but are leaving this check in place
-         * to prevent crashes in case it is not. */
-        x265_log(m_cfg->param, X265_LOG_WARNING,
-                 "internal error - simulaneous row access detected. Please report HW to x265-devel@videolan.org");
-        return;
+        ScopedLock self(curRow.lock);
+        if (!curRow.active)
+            /* VBV restart is in progress, exit out */
+            return;
+        if (curRow.busy)
+        {
+            /* On multi-socket Windows servers, we have seen problems with
+             * ATOMIC_CAS which resulted in multiple worker threads processing
+             * the same CU row, which often resulted in bad pointer accesses. We
+             * believe the problem is fixed, but are leaving this check in place
+             * to prevent crashes in case it is not */
+            x265_log(m_param, X265_LOG_WARNING,
+                     "internal error - simultaneous row access detected. Please report HW to x265-devel@videolan.org\n");
+            return;
+        }
+        curRow.busy = true;
     }
-    curRow.m_busy = true;
+
+    /* When WPP is enabled, every row has its own row coder instance. Otherwise
+     * they share row 0 */
+    Entropy& rowCoder = m_param->bEnableWavefront ? m_rows[row].rdEntropyCoders[0][CI_CURR_BEST] :
+                                                    m_rows[0].rdEntropyCoders[0][CI_CURR_BEST];
+    // setup thread-local data
+    TComPicYuv* fenc = m_frame->getPicYuvOrg();
+    tld.analysis.m_quant.m_nr = m_nr;
+    tld.analysis.m_me.setSourcePlane(fenc->getLumaAddr(), fenc->getStride());
+    tld.analysis.m_log = &tld.analysis.m_sliceTypeLog[m_frame->m_picSym->m_slice->m_sliceType];
+    tld.analysis.m_rdEntropyCoders = curRow.rdEntropyCoders;
+    tld.analysis.m_entropyCoder = &curRow.entropyCoder;
+    tld.analysis.m_quant.m_entropyCoder = &curRow.entropyCoder;
+    setLambda(m_frame->m_picSym->m_slice->m_sliceQp, tld);
 
     int64_t startTime = x265_mdate();
-    CTURow& codeRow = m_rows[m_cfg->param->bEnableWavefront ? row : 0];
-    const uint32_t numCols = m_pic->getPicSym()->getFrameWidthInCU();
+    assert(m_frame->getPicSym()->getFrameWidthInCU() == m_numCols);
+    const uint32_t numCols = m_numCols;
     const uint32_t lineStartCUAddr = row * numCols;
-    bool isVbv = m_cfg->param->rc.vbvBufferSize > 0 && m_cfg->param->rc.vbvMaxBitrate > 0;
-    for (uint32_t col = curRow.m_completed; col < numCols; col++)
+    bool bIsVbv = m_param->rc.vbvBufferSize > 0 && m_param->rc.vbvMaxBitrate > 0;
+
+    while (curRow.completed < numCols)
     {
+        int col = curRow.completed;
         const uint32_t cuAddr = lineStartCUAddr + col;
-        TComDataCU* cu = m_pic->getCU(cuAddr);
-        cu->initCU(m_pic, cuAddr);
-        cu->setQPSubParts(m_pic->getSlice()->getSliceQp(), 0, 0);
-
-        codeRow.m_entropyCoder.setEntropyCoder(&m_sbacCoder, m_pic->getSlice());
-        codeRow.m_entropyCoder.resetEntropy();
-        TEncSbac *bufSbac = (m_cfg->param->bEnableWavefront && col == 0 && row > 0) ? &m_rows[row - 1].m_bufferSbacCoder : NULL;
-        if (isVbv)
+        TComDataCU* cu = m_frame->getCU(cuAddr);
+        cu->initCU(m_frame, cuAddr);
+        cu->setQPSubParts(m_frame->m_picSym->m_slice->m_sliceQp, 0, 0);
+
+        if (bIsVbv)
         {
             if (!row)
-                m_pic->m_rowDiagQp[row] = m_pic->m_avgQpRc;
+            {
+                m_frame->m_rowDiagQp[row] = m_frame->m_avgQpRc;
+                m_frame->m_rowDiagQScale[row] = x265_qp2qScale(m_frame->m_avgQpRc);
+            }
 
-            if ((uint32_t)row >= col && (row != 0))
-                cu->m_baseQp = m_pic->getCU(cuAddr - numCols + 1)->m_baseQp;
+            if (row >= col && row && m_vbvResetTriggerRow != row)
+                cu->m_baseQp = m_frame->getCU(cuAddr - numCols + 1)->m_baseQp;
             else
-                cu->m_baseQp = m_pic->m_rowDiagQp[row];
+                cu->m_baseQp = m_frame->m_rowDiagQp[row];
         }
         else
-            cu->m_baseQp = m_pic->m_avgQpRc;
+            cu->m_baseQp = m_frame->m_avgQpRc;
 
-        if (m_cfg->param->rc.aqMode || isVbv)
+        if (m_param->rc.aqMode || bIsVbv)
         {
-            int qp = calcQpForCu(m_pic, cuAddr, cu->m_baseQp);
-            setLambda(qp, row);
-            qp = Clip3(-QP_BD_OFFSET, MAX_QP, qp);
+            int qp = calcQpForCu(cuAddr, cu->m_baseQp);
+            setLambda(qp, tld);
+            qp = Clip3(QP_MIN, QP_MAX_SPEC, qp);
             cu->setQPSubParts(char(qp), 0, 0);
-            if (m_cfg->param->rc.aqMode)
-                m_pic->m_qpaAq[row] += qp;
+            if (m_param->rc.aqMode)
+                m_frame->m_qpaAq[row] += qp;
         }
-        codeRow.processCU(cu, m_pic->getSlice(), bufSbac, m_cfg->param->bEnableWavefront && col == 1);
-        if (isVbv)
+
+        if (m_param->bEnableWavefront)
+        {
+            if (!col && row)
+            {
+                // Load SBAC coder context from previous row and initialize row state.
+                rowCoder.copyState(m_initSliceContext);
+                rowCoder.loadContexts(m_rows[row - 1].bufferEntropyCoder);
+            }
+        }
+        else if (row)
+            // load current best state from go-on entropy coder
+            curRow.rdEntropyCoders[0][CI_CURR_BEST].load(rowCoder);
+
+        tld.analysis.loadCTUData(cu);
+        tld.analysis.m_quant.setQPforQuant(cu);
+        tld.analysis.compressCU(cu); // Does all the CU analysis
+
+        /* advance top-level row coder to include the context of this CTU.
+         * if SAO is disabled, rowCoder writes the final CTU bitstream */
+        rowCoder.encodeCTU(cu);
+
+        if (m_param->bEnableWavefront && col == 1)
+            // Save CABAC state for next row
+            curRow.bufferEntropyCoder.loadContexts(rowCoder);
+
+        // Completed CU processing
+        curRow.completed++;
+
+        // copy no. of intra, inter Cu cnt per row into frame stats for 2 pass
+        if (m_param->rc.bStatWrite)
+        {
+            curRow.rowStats.mvBits += cu->m_mvBits;
+            curRow.rowStats.coeffBits += cu->m_coeffBits;
+            curRow.rowStats.miscBits += cu->m_totalBits - (cu->m_mvBits + cu->m_coeffBits);
+
+            for (uint32_t depth = 0; depth <= g_maxCUDepth; depth++)
+            {
+                /* 1 << shift == number of 8x8 blocks at current depth */
+                int shift = 2 * (g_maxCUDepth - depth);
+                curRow.rowStats.iCuCnt += tld.analysis.m_log->qTreeIntraCnt[depth] << shift;
+                curRow.rowStats.pCuCnt += tld.analysis.m_log->qTreeInterCnt[depth] << shift;
+                curRow.rowStats.skipCuCnt += tld.analysis.m_log->qTreeSkipCnt[depth] << shift;
+
+                // clear the row cu data from thread local object
+                tld.analysis.m_log->qTreeIntraCnt[depth] = tld.analysis.m_log->qTreeInterCnt[depth] = tld.analysis.m_log->qTreeSkipCnt[depth] = 0;
+            }
+        }
+
+        if (bIsVbv)
         {
             // Update encoded bits, satdCost, baseQP for each CU
-            m_pic->m_rowDiagSatd[row] += m_pic->m_cuCostsForVbv[cuAddr];
-            m_pic->m_rowEncodedBits[row] += cu->m_totalBits;
-            m_pic->m_numEncodedCusPerRow[row] = cuAddr;
-            m_pic->m_qpaRc[row] += cu->m_baseQp;
+            m_frame->m_rowDiagSatd[row] += m_frame->m_cuCostsForVbv[cuAddr];
+            m_frame->m_rowDiagIntraSatd[row] += m_frame->m_intraCuCostsForVbv[cuAddr];
+            m_frame->m_rowEncodedBits[row] += cu->m_totalBits;
+            m_frame->m_numEncodedCusPerRow[row] = cuAddr;
+            m_frame->m_qpaRc[row] += cu->m_baseQp;
 
             // If current block is at row diagonal checkpoint, call vbv ratecontrol.
- 
-            if ((uint32_t)row == col && row != 0)
+
+            if (row == col && row)
             {
-                m_pic->m_rowDiagQp[row] = cu->m_baseQp;
-                m_top->m_rateControl->rowDiagonalVbvRateControl(m_pic, row, &m_rce, m_pic->m_rowDiagQp[row]);
-                m_pic->m_rowDiagQScale[row] = Clip3((double)MIN_QP, (double)MAX_QP, m_pic->m_rowDiagQScale[row]);
-                m_pic->m_rowDiagQScale[row] =  x265_qp2qScale(m_pic->m_rowDiagQp[row]);
+                x265_emms();
+                double qpBase = cu->m_baseQp;
+                int reEncode = m_top->m_rateControl->rowDiagonalVbvRateControl(m_frame, row, &m_rce, qpBase);
+                qpBase = Clip3((double)QP_MIN, (double)QP_MAX_MAX, qpBase);
+                m_frame->m_rowDiagQp[row] = qpBase;
+                m_frame->m_rowDiagQScale[row] =  x265_qp2qScale(qpBase);
+
+                if (reEncode < 0)
+                {
+                    x265_log(m_param, X265_LOG_DEBUG, "POC %d row %d - encode restart required for VBV, to %.2f from %.2f\n",
+                             m_frame->getPOC(), row, qpBase, cu->m_baseQp);
+
+                    // prevent the WaveFront::findJob() method from providing new jobs
+                    m_vbvResetTriggerRow = row;
+                    m_bAllRowsStop = true;
+
+                    for (int r = m_numRows - 1; r >= row; r--)
+                    {
+                        CTURow& stopRow = m_rows[r];
+
+                        if (r != row)
+                        {
+                            /* if row was active (ready to be run) clear active bit and bitmap bit for this row */
+                            stopRow.lock.acquire();
+                            while (stopRow.active)
+                            {
+                                if (dequeueRow(r * 2))
+                                    stopRow.active = false;
+                                else
+                                    GIVE_UP_TIME();
+                            }
+
+                            stopRow.lock.release();
+
+                            bool bRowBusy = true;
+                            do
+                            {
+                                stopRow.lock.acquire();
+                                bRowBusy = stopRow.busy;
+                                stopRow.lock.release();
+
+                                if (bRowBusy)
+                                {
+                                    GIVE_UP_TIME();
+                                }
+                            }
+                            while (bRowBusy);
+                        }
+
+                        m_outStreams[r].resetBits();
+                        stopRow.completed = 0;
+                        memset(&stopRow.rowStats, 0, sizeof(stopRow.rowStats));
+                        if (m_frame->m_qpaAq)
+                            m_frame->m_qpaAq[r] = 0;
+                        m_frame->m_qpaRc[r] = 0;
+                        m_frame->m_rowEncodedBits[r] = 0;
+                        m_frame->m_numEncodedCusPerRow[r] = 0;
+                        m_frame->m_rowDiagSatd[r] = 0;
+                        m_frame->m_rowDiagIntraSatd[r] = 0;
+                    }
+
+                    m_bAllRowsStop = false;
+                }
             }
         }
-        // Completed CU processing
-        m_rows[row].m_completed++;
-        if (m_rows[row].m_completed >= 2 && row < m_numRows - 1)
+
+        // NOTE: do CU level Filter
+        if (m_param->bEnableSAO && m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+            // SAO parameter estimation using non-deblocked pixels for LCU bottom and right boundary areas
+            m_frameFilter.m_sao.calcSaoStatsCu_BeforeDblk(m_frame, col, row);
+
+        // NOTE: active next row
+        if (curRow.completed >= 2 && row < m_numRows - 1)
         {
-            ScopedLock below(m_rows[row + 1].m_lock);
-            if (m_rows[row + 1].m_active == false &&
-                m_rows[row + 1].m_completed + 2 <= m_rows[row].m_completed)
+            ScopedLock below(m_rows[row + 1].lock);
+            if (m_rows[row + 1].active == false &&
+                m_rows[row + 1].completed + 2 <= curRow.completed &&
+                (!m_bAllRowsStop || row + 1 < m_vbvResetTriggerRow))
             {
-                m_rows[row + 1].m_active = true;
+                m_rows[row + 1].active = true;
                 enqueueRowEncoder(row + 1);
             }
         }
 
-        ScopedLock self(curRow.m_lock);
-        if (row > 0 && m_rows[row].m_completed < numCols - 1 && m_rows[row - 1].m_completed < m_rows[row].m_completed + 2)
+        ScopedLock self(curRow.lock);
+        if ((m_bAllRowsStop && row > m_vbvResetTriggerRow) ||
+            (row > 0 && curRow.completed < numCols - 1 && m_rows[row - 1].completed < m_rows[row].completed + 2))
         {
-            curRow.m_active = false;
-            curRow.m_busy = false;
-            m_totalTime = m_totalTime + (x265_mdate() - startTime);
+            curRow.active = false;
+            curRow.busy = false;
+            m_totalTime += x265_mdate() - startTime;
             return;
         }
     }
 
-    // this row of CTUs has been encoded
+    /* *this row of CTUs has been encoded* */
+
+    /* flush row bitstream (if WPP and no SAO) or flush frame if no WPP and no SAO */
+    if (!m_param->bEnableSAO && (m_param->bEnableWavefront || row == m_numRows - 1))
+        rowCoder.finishSlice();
+
+    /* If encoding with ABR, update update bits and complexity in rate control
+     * after a number of rows so the next frame's rateControlStart has more
+     * accurate data for estimation. At the start of the encode we update stats
+     * after half the frame is encoded, but after this initial period we update
+     * after refLagRows (the number of rows reference frames must have completed
+     * before referencees may begin encoding) */
+    int rowCount = 0;
+    if (m_param->rc.rateControlMode == X265_RC_ABR)
+    {
+        if ((uint32_t)m_rce.encodeOrder <= 2 * (m_param->fpsNum / m_param->fpsDenom))
+            rowCount = X265_MIN((m_numRows + 1) / 2, m_numRows - 1);
+        else
+            rowCount = X265_MIN(m_refLagRows, m_numRows - 1);
+    }
+    if (row == rowCount)
+    {
+        int64_t bits = 0;
+        for (uint32_t col = 0; col < rowCount * numCols; col++)
+        {
+            TComDataCU* cu = m_frame->getCU(col);
+            bits += cu->m_totalBits;
+        }
+
+        m_rce.rowTotalBits = bits;
+        m_top->m_rateControl->rateControlUpdateStats(&m_rce);
+    }
 
-    // Run row-wise loop filters
+    // trigger row-wise loop filters
     if (row >= m_filterRowDelay)
     {
         enableRowFilter(row - m_filterRowDelay);
@@ -1130,40 +874,103 @@
     if (row == m_numRows - 1)
     {
         for (int i = m_numRows - m_filterRowDelay; i < m_numRows; i++)
-        {
             enableRowFilter(i);
+    }
+
+    m_totalTime += x265_mdate() - startTime;
+    curRow.busy = false;
+}
+
+void FrameEncoder::setLambda(int qp, ThreadLocalData &tld)
+{
+    Slice* slice = m_frame->m_picSym->m_slice;
+    int qpCb = Clip3(0, QP_MAX_MAX, qp + slice->m_pps->chromaCbQpOffset);
+    int qpCr = Clip3(0, QP_MAX_MAX, qp + slice->m_pps->chromaCrQpOffset);
+    qp = Clip3(0, QP_MAX_MAX, qp);
+
+    tld.analysis.m_me.setQP(qp);
+    tld.analysis.m_rdCost.setLambda(x265_lambda2_tab[qp], x265_lambda_tab[qp]);
+
+    int chroma_offset_idx = X265_MIN(qp - qpCb + 12, MAX_CHROMA_LAMBDA_OFFSET);
+    uint16_t lambdaOffset = tld.analysis.m_rdCost.m_psyRd ? x265_chroma_lambda2_offset_tab[chroma_offset_idx] : 256;
+    tld.analysis.m_rdCost.setCbDistortionWeight(lambdaOffset);
+
+    chroma_offset_idx = X265_MIN(qp - qpCr + 12, MAX_CHROMA_LAMBDA_OFFSET);
+    lambdaOffset = tld.analysis.m_rdCost.m_psyRd ? x265_chroma_lambda2_offset_tab[chroma_offset_idx] : 256;
+    tld.analysis.m_rdCost.setCrDistortionWeight(lambdaOffset);
+}
+
+/* DCT-domain noise reduction / adaptive deadzone from libavcodec */
+void FrameEncoder::noiseReductionUpdate()
+{
+    if (!m_nr)
+        return;
+
+    static const uint32_t maxBlocksPerTrSize[4] = {1 << 18, 1 << 16, 1 << 14, 1 << 12};
+
+    for (int cat = 0; cat < 8; cat++)
+    {
+        int trSize = cat & 3;
+        int coefCount = 1 << ((trSize + 2) * 2);
+
+        if (m_nr->count[cat] > maxBlocksPerTrSize[trSize])
+        {
+            for (int i = 0; i < coefCount; i++)
+                m_nr->residualSum[cat][i] >>= 1;
+            m_nr->count[cat] >>= 1;
         }
+
+        uint64_t scaledCount = (uint64_t)m_param->noiseReduction * m_nr->count[cat];
+
+        for (int i = 0; i < coefCount; i++)
+        {
+            uint64_t value = scaledCount + m_nr->residualSum[cat][i] / 2;
+            uint64_t denom = m_nr->residualSum[cat][i] + 1;
+            m_nr->offsetDenoise[cat][i] = (uint16_t)(value / denom);
+        }
+
+        // Don't denoise DC coefficients
+        m_nr->offsetDenoise[cat][0] = 0;
     }
-    m_totalTime = m_totalTime + (x265_mdate() - startTime);
-    curRow.m_busy = false;
 }
 
-int FrameEncoder::calcQpForCu(TComPic *pic, uint32_t cuAddr, double baseQp)
+int FrameEncoder::calcQpForCu(uint32_t cuAddr, double baseQp)
 {
     x265_emms();
     double qp = baseQp;
 
+    /* clear cuCostsForVbv from when vbv row reset was triggered */
+    bool bIsVbv = m_param->rc.vbvBufferSize > 0 && m_param->rc.vbvMaxBitrate > 0;
+    if (bIsVbv)
+    {
+        m_frame->m_cuCostsForVbv[cuAddr] = 0;
+        m_frame->m_intraCuCostsForVbv[cuAddr] = 0;
+    }
+
     /* Derive qpOffet for each CU by averaging offsets for all 16x16 blocks in the cu. */
     double qp_offset = 0;
-    int maxBlockCols = (pic->getPicYuvOrg()->getWidth() + (16 - 1)) / 16;
-    int maxBlockRows = (pic->getPicYuvOrg()->getHeight() + (16 - 1)) / 16;
+    int maxBlockCols = (m_frame->getPicYuvOrg()->getWidth() + (16 - 1)) / 16;
+    int maxBlockRows = (m_frame->getPicYuvOrg()->getHeight() + (16 - 1)) / 16;
     int noOfBlocks = g_maxCUSize / 16;
-    int block_y = (cuAddr / pic->getPicSym()->getFrameWidthInCU()) * noOfBlocks;
-    int block_x = (cuAddr * noOfBlocks) - block_y * pic->getPicSym()->getFrameWidthInCU();
+    int block_y = (cuAddr / m_frame->getPicSym()->getFrameWidthInCU()) * noOfBlocks;
+    int block_x = (cuAddr * noOfBlocks) - block_y * m_frame->getPicSym()->getFrameWidthInCU();
+
+    /* Use cuTree offsets if cuTree enabled and frame is referenced, else use AQ offsets */
+    bool isReferenced = IS_REFERENCED(m_frame->m_picSym->m_slice);
+    double *qpoffs = (isReferenced && m_param->rc.cuTree) ? m_frame->m_lowres.qpCuTreeOffset : m_frame->m_lowres.qpAqOffset;
 
-    double *qpoffs = (pic->getSlice()->isReferenced() && m_cfg->param->rc.cuTree) ? pic->m_lowres.qpOffset : pic->m_lowres.qpAqOffset;
     int cnt = 0, idx = 0;
     for (int h = 0; h < noOfBlocks && block_y < maxBlockRows; h++, block_y++)
     {
         for (int w = 0; w < noOfBlocks && (block_x + w) < maxBlockCols; w++)
         {
             idx = block_x + w + (block_y * maxBlockCols);
-            if (m_cfg->param->rc.aqMode)
+            if (m_param->rc.aqMode)
                 qp_offset += qpoffs[idx];
-            if (m_cfg->param->rc.vbvBufferSize > 0 && m_cfg->param->rc.vbvMaxBitrate > 0)
+            if (bIsVbv)
             {
-                m_pic->m_cuCostsForVbv[cuAddr] += m_pic->m_lowres.lowresCostForRc[idx];
-                m_pic->m_intraCuCostsForVbv[cuAddr] += m_pic->m_lowres.intraCost[idx];
+                m_frame->m_cuCostsForVbv[cuAddr] += m_frame->m_lowres.lowresCostForRc[idx] & LOWRES_COST_MASK;
+                m_frame->m_intraCuCostsForVbv[cuAddr] += m_frame->m_lowres.intraCost[idx];
             }
             cnt++;
         }
@@ -1172,26 +979,19 @@
     qp_offset /= cnt;
     qp += qp_offset;
 
-    return Clip3(MIN_QP, MAX_MAX_QP, (int)(qp + 0.5));
+    return Clip3(QP_MIN, QP_MAX_MAX, (int)(qp + 0.5));
 }
 
-TComPic *FrameEncoder::getEncodedPicture(NALUnitEBSP **nalunits)
+Frame *FrameEncoder::getEncodedPicture(NALList& output)
 {
-    if (m_pic)
+    if (m_frame)
     {
         /* block here until worker thread completes */
         m_done.wait();
 
-        TComPic *ret = m_pic;
-        m_pic = NULL;
-
-        if (nalunits)
-        {
-            // move NALs from member variable to user's container
-            assert(m_nalCount <= MAX_NAL_UNITS);
-            ::memcpy(nalunits, m_nalList, sizeof(NALUnitEBSP*) * m_nalCount);
-            m_nalCount = 0;
-        }
+        Frame *ret = m_frame;
+        m_frame = NULL;
+        output.takeContents(m_nalList);
         return ret;
     }

x265-hg20140324.tar.bz2/source/encoder/frameencoder.h -> x265-hg20140928.tar.bz2/source/encoder/frameencoder.h Changed

@@ -1,7 +1,7 @@
 /*****************************************************************************
  * Copyright (C) 2013 x265 project
  *
- * Authors: Chung Shin Yee <shinyee@multicorewareinc.com>
+ * Authors: Shin Yee <shinyee@multicorewareinc.com>
  *          Min Chen <chenm003@163.com>
  *
  * This program is free software; you can redistribute it and/or modify
@@ -19,28 +19,25 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_FRAMEENCODER_H
 #define X265_FRAMEENCODER_H
 
-#include "TLibCommon/TComBitCounter.h"
-#include "TLibCommon/TComPic.h"
-#include "TLibCommon/NAL.h"
+#include "common.h"
+#include "wavefront.h"
+#include "bitstream.h"
+#include "frame.h"
 
-#include "TLibEncoder/TEncCu.h"
-#include "TLibEncoder/TEncSearch.h"
-#include "TLibEncoder/TEncSbac.h"
-#include "TLibEncoder/TEncBinCoderCABAC.h"
-#include "TLibEncoder/TEncSampleAdaptiveOffset.h"
-#include "TLibEncoder/SEIwrite.h"
+#include "analysis.h"
+#include "sao.h"
 
-#include "wavefront.h"
+#include "entropy.h"
 #include "framefilter.h"
-#include "cturow.h"
 #include "ratecontrol.h"
 #include "reference.h"
+#include "nal.h"
 
 namespace x265 {
 // private x265 namespace
@@ -48,145 +45,148 @@
 class ThreadPool;
 class Encoder;
 
-// Manages the wave-front processing of a single encoding frame
-class FrameEncoder : public WaveFront, public Thread
+struct ThreadLocalData
 {
-public:
+    Analysis analysis;
 
-    FrameEncoder();
+    ~ThreadLocalData() { analysis.destroy(); }
+};
 
-    virtual ~FrameEncoder() {}
+/* manages the state of encoding one row of CTU blocks.  When
+ * WPP is active, several rows will be simultaneously encoded. */
+struct CTURow
+{
+    Entropy           entropyCoder;
+    Entropy           bufferEntropyCoder;  /* store context for next row */
+    Entropy           rdEntropyCoders[NUM_FULL_DEPTH][CI_NUM];
 
-    void setThreadPool(ThreadPool *p);
+    FrameStats        rowStats;
 
-    bool init(Encoder *top, int numRows);
+    /* Threading variables */
 
-    void destroy();
+    /* This lock must be acquired when reading or writing m_active or m_busy */
+    Lock              lock;
 
-    void processRowEncoder(int row);
+    /* row is ready to run, has no neighbor dependencies. The row may have
+     * external dependencies (reference frame pixels) that prevent it from being
+     * processed, so it may stay with m_active=true for some time before it is
+     * encoded by a worker thread. */
+    volatile bool     active;
 
-    void processRowFilter(int row)
-    {
-        m_frameFilter.processRow(row, m_cfg);
-    }
+    /* row is being processed by a worker thread.  This flag is only true when a
+     * worker thread is within the context of FrameEncoder::processRow(). This
+     * flag is used to detect multiple possible wavefront problems. */
+    volatile bool     busy;
 
-    void enqueueRowEncoder(int row)
-    {
-        WaveFront::enqueueRow(row * 2 + 0);
-    }
+    /* count of completed CUs in this row */
+    volatile uint32_t completed;
 
-    void enqueueRowFilter(int row)
+    /* called at the start of each frame to initialize state */
+    void init(Entropy& initContext)
     {
-        WaveFront::enqueueRow(row * 2 + 1);
-    }
+        active = false;
+        busy = false;
+        completed = 0;
+        memset(&rowStats, 0, sizeof(rowStats));
 
-    void enableRowEncoder(int row)
-    {
-        WaveFront::enableRow(row * 2 + 0);
-    }
+        entropyCoder.load(initContext);
 
-    void enableRowFilter(int row)
-    {
-        WaveFront::enableRow(row * 2 + 1);
+        for (uint32_t depth = 0; depth <= g_maxFullDepth; depth++)
+            for (int ciIdx = 0; ciIdx < CI_NUM; ciIdx++)
+                rdEntropyCoders[depth][ciIdx].load(initContext);
     }
+};
 
-    void processRow(int row)
-    {
-        const int realRow = row >> 1;
-        const int typeNum = row & 1;
-
-        // TODO: use switch when more type
-        if (typeNum == 0)
-        {
-            processRowEncoder(realRow);
-        }
-        else
-        {
-            processRowFilter(realRow);
-
-            // NOTE: Active next row
-            if (realRow != m_numRows - 1)
-                enqueueRowFilter(realRow + 1);
-            else
-                m_completionEvent.trigger();
-        }
-    }
-
-    TEncEntropy* getEntropyCoder(int row)      { return &this->m_rows[row].m_entropyCoder; }
-
-    TEncSbac*    getSbacCoder(int row)         { return &this->m_rows[row].m_sbacCoder; }
-
-    TEncSbac*    getRDGoOnSbacCoder(int row)   { return &this->m_rows[row].m_rdGoOnSbacCoder; }
-
-    TEncSbac*    getBufferSBac(int row)        { return &this->m_rows[row].m_bufferSbacCoder; }
-
-    TEncCu*      getCuEncoder(int row)         { return &this->m_rows[row].m_cuCoder; }
-
-    /* Frame singletons, last the life of the encoder */
-    TEncSampleAdaptiveOffset* getSAO()         { return &m_frameFilter.m_sao; }
-
-    void resetEntropy(TComSlice *slice)
-    {
-        for (int i = 0; i < this->m_numRows; i++)
-        {
-            this->m_rows[i].m_entropyCoder.setEntropyCoder(&this->m_rows[i].m_sbacCoder, slice);
-            this->m_rows[i].m_entropyCoder.resetEntropy();
-        }
-    }
-
-    int getStreamHeaders(NALUnitEBSP **nalunits);
-
-    void initSlice(TComPic* pic);
+// Manages the wave-front processing of a single encoding frame
+class FrameEncoder : public WaveFront, public Thread
+{
+public:
 
-    /* analyze / compress frame, can be run in parallel within reference constraints */
-    void compressFrame();
+    FrameEncoder();
 
-    /* called by compressFrame to perform wave-front compression analysis */
-    void compressCTURows();
+    virtual ~FrameEncoder() {}
 
-    void encodeSlice(TComOutputBitstream* substreams);
+    bool init(Encoder *top, int numRows, int numCols);
 
-    /* blocks until worker thread is done, returns encoded picture and bitstream */
-    TComPic *getEncodedPicture(NALUnitEBSP **nalunits);
+    void destroy();
 
-    void setLambda(int qp, int row);
+    /* triggers encode of a new frame by the worker thread */
+    void startCompressFrame(Frame* pic);
 
-    // worker thread
-    void threadMain();
+    /* blocks until worker thread is done, returns access unit */
+    Frame *getEncodedPicture(NALList& list);
 
     Event                    m_enable;
     Event                    m_done;
     bool                     m_threadActive;
 
     int                      m_numRows;
+    uint32_t                 m_numCols;
+    int                      m_refLagRows;
     CTURow*                  m_rows;
-    SEIWriter                m_seiWriter;
-    TComSPS                  m_sps;
-    TComPPS                  m_pps;
     RateControlEntry         m_rce;
     SEIDecodedPictureHash    m_seiReconPictureDigest;
 
-protected:
+    uint64_t                 m_SSDY;
+    uint64_t                 m_SSDU;
+    uint64_t                 m_SSDV;
+    double                   m_ssim;
+    uint32_t                 m_ssimCnt;
+    MD5Context               m_state[3];
+    uint32_t                 m_crc[3];
+    uint32_t                 m_checksum[3];
+    double                   m_elapsedCompressTime; // elapsed time spent in worker threads
+    double                   m_frameTime;           // wall time from frame start to finish
+    FrameStats               m_frameStats;          // stats of current frame for multipass encodes
+    volatile bool            m_bAllRowsStop;
+    volatile int             m_vbvResetTriggerRow;
+    uint64_t                 m_accessUnitBits;
 
-    void determineSliceBounds();
-    int calcQpForCu(TComPic *pic, uint32_t cuAddr, double baseQp);
     Encoder*                 m_top;
-    Encoder*                 m_cfg;
+    x265_param*              m_param;
+    Frame*                   m_frame;
 
     MotionReference          m_mref[2][MAX_NUM_REF + 1];
-    TEncSbac                 m_sbacCoder;
-    TEncBinCABAC             m_binCoderCABAC;
+    Entropy                  m_entropyCoder;
+    Entropy                  m_initSliceContext;
     FrameFilter              m_frameFilter;
-    TComBitCounter           m_bitCounter;
-
-    /* Picture being encoded, and its output NAL list */
-    TComPic*                 m_pic;
-    NALUnitEBSP*             m_nalList[MAX_NAL_UNITS];
-    int                      m_nalCount;
+    Bitstream                m_bs;
+    Bitstream*               m_outStreams;
+    uint32_t*                m_substreamSizes;
+    NoiseReduction*          m_nr;
+    NALList                  m_nalList;
+    ThreadLocalData*         m_tld; /* for --no-wpp */
 
     int                      m_filterRowDelay;
+    int                      m_filterRowDelayCus;
     Event                    m_completionEvent;
     int64_t                  m_totalTime;
+
+protected:
+
+    /* analyze / compress frame, can be run in parallel within reference constraints */
+    void compressFrame();
+
+    /* called by compressFrame to perform wave-front compression analysis */
+    void compressCTURows();
+
+    /* called by compressFrame to generate final per-row bitstreams */
+    void encodeSlice();
+
+    void threadMain();
+    void setLambda(int qp, ThreadLocalData& tld);
+    int calcQpForCu(uint32_t cuAddr, double baseQp);
+    void noiseReductionUpdate();
+
+    /* Called by WaveFront::findJob() */
+    void processRow(int row, int threadId);
+    void processRowEncoder(int row, ThreadLocalData& tld);
+    void processRowFilter(int row) { m_frameFilter.processRow(row); }
+
+    void enqueueRowEncoder(int row) { WaveFront::enqueueRow(row * 2 + 0); }
+    void enqueueRowFilter(int row)  { WaveFront::enqueueRow(row * 2 + 1); }
+    void enableRowEncoder(int row)  { WaveFront::enableRow(row * 2 + 0); }
+    void enableRowFilter(int row)   { WaveFront::enableRow(row * 2 + 1); }
 };
 }

x265-hg20140324.tar.bz2/source/encoder/framefilter.cpp -> x265-hg20140928.tar.bz2/source/encoder/framefilter.cpp Changed

@@ -19,201 +19,142 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "encoder.h"
 #include "PPA/ppa.h"
 #include "framefilter.h"
+#include "frameencoder.h"
 #include "wavefront.h"
 
 using namespace x265;
 
 static uint64_t computeSSD(pixel *fenc, pixel *rec, int stride, int width, int height);
-static float calculateSSIM(pixel *pix1, intptr_t stride1, pixel *pix2, intptr_t stride2, int width, int height, void *buf, int32_t *cnt);
+static float calculateSSIM(pixel *pix1, intptr_t stride1, pixel *pix2, intptr_t stride2, int width, int height, void *buf, uint32_t& cnt);
 
-// **************************************************************************
-// * LoopFilter
-// **************************************************************************
 FrameFilter::FrameFilter()
     : m_param(NULL)
-    , m_rdGoOnBinCodersCABAC(true)
+    , m_frame(NULL)
+    , m_frameEncoder(NULL)
     , m_ssimBuf(NULL)
 {
 }
 
 void FrameFilter::destroy()
 {
-    if (m_param->bEnableLoopFilter)
-    {
-        m_loopFilter.destroy();
-    }
-
     if (m_param->bEnableSAO)
-    {
-        // NOTE: I don't check sao flag since loopfilter and sao have same control status
         m_sao.destroy();
-        m_sao.destroyEncBuffer();
-    }
+
     X265_FREE(m_ssimBuf);
 }
 
-void FrameFilter::init(Encoder *top, int numRows, TEncSbac* rdGoOnSbacCoder)
+void FrameFilter::init(Encoder *top, FrameEncoder *frame, int numRows)
 {
-    m_param = top->param;
+    m_param = top->m_param;
+    m_frameEncoder = frame;
     m_numRows = numRows;
     m_hChromaShift = CHROMA_H_SHIFT(m_param->internalCsp);
     m_vChromaShift = CHROMA_V_SHIFT(m_param->internalCsp);
+    m_pad[0] = top->m_sps.conformanceWindow.rightOffset;
+    m_pad[1] = top->m_sps.conformanceWindow.bottomOffset;
+    m_saoRowDelay = m_param->bEnableLoopFilter ? 1 : 0;
 
-    // NOTE: for sao only, I write this code because I want to exact match with HM's bug bitstream
-    m_rdGoOnSbacCoderRow0 = rdGoOnSbacCoder;
-
-    if (top->param->bEnableLoopFilter)
-    {
-        m_loopFilter.create(g_maxCUDepth);
-    }
+    m_deblock.init();
 
-    if (top->param->bEnableSAO)
-    {
-        m_sao.setSaoLcuBoundary(top->param->saoLcuBoundary);
-        m_sao.setSaoLcuBasedOptimization(top->param->saoLcuBasedOptimization);
-        m_sao.setMaxNumOffsetsPerPic(top->m_maxNumOffsetsPerPic);
-        m_sao.create(top->param->sourceWidth, top->param->sourceHeight, g_maxCUSize, g_maxCUSize, m_param->internalCsp);
-        m_sao.createEncBuffer();
-    }
+    if (m_param->bEnableSAO)
+        if (!m_sao.create(m_param))
+            m_param->bEnableSAO = 0;
 
     if (m_param->bEnableSsim)
-        m_ssimBuf = (int*)x265_malloc(sizeof(int) * 8 * (m_param->sourceWidth / 4 + 3));
+        m_ssimBuf = X265_MALLOC(int, 8 * (m_param->sourceWidth / 4 + 3));
 }
 
-void FrameFilter::start(TComPic *pic)
+void FrameFilter::start(Frame *pic, Entropy& initState, int qp)
 {
-    m_pic = pic;
-
-    m_saoRowDelay = m_param->bEnableLoopFilter ? 1 : 0;
-    m_loopFilter.setCfg(pic->getSlice()->getPPS()->getLoopFilterAcrossTilesEnabledFlag());
-    m_rdGoOnSbacCoder.init(&m_rdGoOnBinCodersCABAC);
-    m_entropyCoder.setEntropyCoder(&m_rdGoOnSbacCoder, pic->getSlice());
-    m_entropyCoder.setBitstream(&m_bitCounter);
-    m_rdGoOnBinCodersCABAC.m_fracBits = 0;
+    m_frame = pic;
 
     if (m_param->bEnableSAO)
-    {
-        m_sao.resetStats();
-        m_sao.createPicSaoInfo(pic);
-
-        SAOParam* saoParam = pic->getPicSym()->getSaoParam();
-        m_sao.resetSAOParam(saoParam);
-        m_sao.rdoSaoUnitRowInit(saoParam);
-
-        // NOTE: Disable SAO automatic turn-off when frame parallelism is
-        // enabled for output exact independent of frame thread count
-        if (m_param->frameNumThreads > 1)
-        {
-            saoParam->bSaoFlag[0] = true;
-            saoParam->bSaoFlag[1] = true;
-        }
-    }
-}
-
-void FrameFilter::end()
-{
+        m_sao.startSlice(pic, initState, qp);
 }
 
-void FrameFilter::processRow(int row, Encoder* cfg)
+void FrameFilter::processRow(int row)
 {
     PPAScopeEvent(Thread_filterCU);
 
     if (!m_param->bEnableLoopFilter && !m_param->bEnableSAO)
     {
-        processRowPost(row, cfg);
+        processRowPost(row);
         return;
     }
 
-    // NOTE: We are here only active both of loopfilter and sao, the row 0 always finished, so we can safe to copy row[0]'s data
-    if (row == 0 && m_param->bEnableSAO)
-    {
-        // NOTE: not need, seems HM's bug, I want to keep output exact matched.
-        m_rdGoOnBinCodersCABAC.m_fracBits = ((TEncBinCABAC*)((TEncSbac*)m_rdGoOnSbacCoderRow0->m_binIf))->m_fracBits;
-        m_sao.startSaoEnc(m_pic, &m_entropyCoder, &m_rdGoOnSbacCoder);
-    }
-
-    const uint32_t numCols = m_pic->getPicSym()->getFrameWidthInCU();
+    const uint32_t numCols = m_frame->getPicSym()->getFrameWidthInCU();
     const uint32_t lineStartCUAddr = row * numCols;
 
-    // SAO parameter estimation using non-deblocked pixels for LCU bottom and right boundary areas
-    if (m_param->bEnableSAO && m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
-    {
-        m_sao.calcSaoStatsRowCus_BeforeDblk(m_pic, row);
-    }
-
     if (m_param->bEnableLoopFilter)
     {
         for (uint32_t col = 0; col < numCols; col++)
         {
             const uint32_t cuAddr = lineStartCUAddr + col;
-            TComDataCU* cu = m_pic->getCU(cuAddr);
+            TComDataCU* cu = m_frame->getCU(cuAddr);
 
-            m_loopFilter.loopFilterCU(cu, EDGE_VER);
+            m_deblock.deblockCTU(cu, Deblock::EDGE_VER);
 
             if (col > 0)
             {
-                TComDataCU* cu_prev = m_pic->getCU(cuAddr - 1);
-                m_loopFilter.loopFilterCU(cu_prev, EDGE_HOR);
+                TComDataCU* cu_prev = m_frame->getCU(cuAddr - 1);
+                m_deblock.deblockCTU(cu_prev, Deblock::EDGE_HOR);
             }
         }
 
-        {
-            TComDataCU* cu_prev = m_pic->getCU(lineStartCUAddr + numCols - 1);
-            m_loopFilter.loopFilterCU(cu_prev, EDGE_HOR);
-        }
+        TComDataCU* cu_prev = m_frame->getCU(lineStartCUAddr + numCols - 1);
+        m_deblock.deblockCTU(cu_prev, Deblock::EDGE_HOR);
     }
 
     // SAO
-    SAOParam* saoParam = m_pic->getPicSym()->getSaoParam();
-    if (m_param->bEnableSAO && m_sao.getSaoLcuBasedOptimization())
+    SAOParam* saoParam = m_frame->getPicSym()->m_saoParam;
+    if (m_param->bEnableSAO)
     {
-        m_sao.rdoSaoUnitRow(saoParam, row);
-
-        // NOTE: Delay a row because SAO decide need top row pixels at next row, is it HM's bug?
-        if (row >= m_saoRowDelay)
+        if (m_param->saoLcuBasedOptimization)
         {
-            processSao(row - m_saoRowDelay);
+            m_sao.m_entropyCoder.load(m_frameEncoder->m_initSliceContext);
+            m_sao.m_rdEntropyCoders[0][CI_NEXT_BEST].load(m_frameEncoder->m_initSliceContext);
+            m_sao.m_rdEntropyCoders[0][CI_CURR_BEST].load(m_frameEncoder->m_initSliceContext);
+
+            m_sao.rdoSaoUnitRow(saoParam, row);
+
+            // NOTE: Delay a row because SAO decide need top row pixels at next row, is it HM's bug?
+            if (row >= m_saoRowDelay)
+                processSao(row - m_saoRowDelay);
         }
+        else
+            return;
     }
 
     // this row of CTUs has been encoded
 
-    // NOTE: in --sao-lcu-opt=0 mode, we do it later
-    if (m_param->bEnableSAO && !m_sao.getSaoLcuBasedOptimization())
-        return;
-
     if (row > 0)
-    {
-        processRowPost(row - 1, cfg);
-    }
+        processRowPost(row - 1);
 
     if (row == m_numRows - 1)
     {
-        if (m_param->bEnableSAO && m_sao.getSaoLcuBasedOptimization())
+        if (m_param->bEnableSAO && m_param->saoLcuBasedOptimization)
         {
-            m_sao.rdoSaoUnitRowEnd(saoParam, m_pic->getNumCUsInFrame());
+            m_sao.rdoSaoUnitRowEnd(saoParam, m_frame->getNumCUsInFrame());
 
             for (int i = m_numRows - m_saoRowDelay; i < m_numRows; i++)
-            {
                 processSao(i);
-            }
         }
 
-        processRowPost(row, cfg);
+        processRowPost(row);
     }
 }
 
-void FrameFilter::processRowPost(int row, Encoder* cfg)
+void FrameFilter::processRowPost(int row)
 {
-    const uint32_t numCols = m_pic->getPicSym()->getFrameWidthInCU();
+    const uint32_t numCols = m_frame->getPicSym()->getFrameWidthInCU();
     const uint32_t lineStartCUAddr = row * numCols;
-    TComPicYuv *recon = m_pic->getPicYuvRec();
+    TComPicYuv *recon = m_frame->getPicYuvRec();
     const int lastH = ((recon->getHeight() % g_maxCUSize) ? (recon->getHeight() % g_maxCUSize) : g_maxCUSize);
     const int realH = (row != m_numRows - 1) ? g_maxCUSize : lastH;
 
@@ -221,8 +162,9 @@
     primitives.extendRowBorder(recon->getLumaAddr(lineStartCUAddr), recon->getStride(), recon->getWidth(), realH, recon->getLumaMarginX());
     primitives.extendRowBorder(recon->getCbAddr(lineStartCUAddr), recon->getCStride(), recon->getWidth() >> m_hChromaShift, realH >> m_vChromaShift, recon->getChromaMarginX());
     primitives.extendRowBorder(recon->getCrAddr(lineStartCUAddr), recon->getCStride(), recon->getWidth() >> m_hChromaShift, realH >> m_vChromaShift, recon->getChromaMarginX());
+
     // Border extend Top
-    if (row == 0)
+    if (!row)
     {
         const intptr_t stride = recon->getStride();
         const intptr_t strideC = recon->getCStride();
@@ -231,9 +173,7 @@
         pixel *pixV = recon->getCrAddr(lineStartCUAddr) - recon->getChromaMarginX();
 
         for (int y = 0; y < recon->getLumaMarginY(); y++)
-        {
             memcpy(pixY - (y + 1) * stride, pixY, stride * sizeof(pixel));
-        }
 
         for (int y = 0; y < recon->getChromaMarginY(); y++)
         {
@@ -251,9 +191,7 @@
         pixel *pixU = recon->getCbAddr(lineStartCUAddr) - recon->getChromaMarginX() + ((realH >> m_vChromaShift) - 1) * strideC;
         pixel *pixV = recon->getCrAddr(lineStartCUAddr) - recon->getChromaMarginX() + ((realH >> m_vChromaShift) - 1) * strideC;
         for (int y = 0; y < recon->getLumaMarginY(); y++)
-        {
             memcpy(pixY + (y + 1) * stride, pixY, stride * sizeof(pixel));
-        }
 
         for (int y = 0; y < recon->getChromaMarginY(); y++)
         {
@@ -263,15 +201,15 @@
     }
 
     // Notify other FrameEncoders that this row of reconstructed pixels is available
-    m_pic->m_reconRowCount.incr();
+    m_frame->m_reconRowCount.incr();
 
     int cuAddr = lineStartCUAddr;
     if (m_param->bEnablePsnr)
     {
-        TComPicYuv* orig  = m_pic->getPicYuvOrg();
+        TComPicYuv* orig  = m_frame->getPicYuvOrg();
 
         int stride = recon->getStride();
-        int width  = recon->getWidth() - cfg->m_pad[0];
+        int width  = recon->getWidth() - m_pad[0];
         int height;
 
         if (row == m_numRows - 1)
@@ -287,29 +225,29 @@
         uint64_t ssdU = computeSSD(orig->getCbAddr(cuAddr), recon->getCbAddr(cuAddr), stride, width, height);
         uint64_t ssdV = computeSSD(orig->getCrAddr(cuAddr), recon->getCrAddr(cuAddr), stride, width, height);
 
-        m_pic->m_SSDY += ssdY;
-        m_pic->m_SSDU += ssdU;
-        m_pic->m_SSDV += ssdV;
+        m_frameEncoder->m_SSDY += ssdY;
+        m_frameEncoder->m_SSDU += ssdU;
+        m_frameEncoder->m_SSDV += ssdV;
     }
     if (m_param->bEnableSsim && m_ssimBuf)
     {
-        pixel *rec = (pixel*)m_pic->getPicYuvRec()->getLumaAddr();
-        pixel *org = (pixel*)m_pic->getPicYuvOrg()->getLumaAddr();
-        int stride1 = m_pic->getPicYuvOrg()->getStride();
-        int stride2 = m_pic->getPicYuvRec()->getStride();
+        pixel *rec = m_frame->getPicYuvRec()->getLumaAddr();
+        pixel *org = m_frame->getPicYuvOrg()->getLumaAddr();
+        int stride1 = m_frame->getPicYuvOrg()->getStride();
+        int stride2 = m_frame->getPicYuvRec()->getStride();
         int bEnd = ((row + 1) == (this->m_numRows - 1));
         int bStart = (row == 0);
         int minPixY = row * g_maxCUSize - 4 * !bStart;
         int maxPixY = (row + 1) * g_maxCUSize - 4 * !bEnd;
-        int ssim_cnt;
+        uint32_t ssim_cnt;
         x265_emms();
 
         /* SSIM is done for each row in blocks of 4x4 . The First blocks are offset by 2 pixels to the right
         * to avoid alignment of ssim blocks with DCT blocks. */
         minPixY += bStart ? 2 : -6;
-        m_pic->m_ssim += calculateSSIM(rec + 2 + minPixY * stride1, stride1, org + 2 + minPixY * stride2, stride2,
-                                       m_param->sourceWidth - 2, maxPixY - minPixY, m_ssimBuf, &ssim_cnt);
-        m_pic->m_ssimCnt += ssim_cnt;
+        m_frameEncoder->m_ssim += calculateSSIM(rec + 2 + minPixY * stride1, stride1, org + 2 + minPixY * stride2, stride2,
+                                         m_param->sourceWidth - 2, maxPixY - minPixY, m_ssimBuf, ssim_cnt);
+        m_frameEncoder->m_ssimCnt += ssim_cnt;
     }
     if (m_param->decodedPictureHashSEI == 1)
     {
@@ -317,39 +255,34 @@
         uint32_t height = recon->getCUHeight(row);
         uint32_t stride = recon->getStride();
 
-        if (row == 0)
+        if (!row)
         {
             for (int i = 0; i < 3; i++)
-            {
-                MD5Init(&(m_pic->m_state[i]));
-            }
+                MD5Init(&m_frameEncoder->m_state[i]);
         }
 
-        updateMD5Plane(m_pic->m_state[0], recon->getLumaAddr(cuAddr), width, height, stride);
+        updateMD5Plane(m_frameEncoder->m_state[0], recon->getLumaAddr(cuAddr), width, height, stride);
         width  >>= m_hChromaShift;
         height >>= m_vChromaShift;
         stride = recon->getCStride();
 
-        updateMD5Plane(m_pic->m_state[1], recon->getCbAddr(cuAddr), width, height, stride);
-
-        updateMD5Plane(m_pic->m_state[2], recon->getCrAddr(cuAddr), width, height, stride);
+        updateMD5Plane(m_frameEncoder->m_state[1], recon->getCbAddr(cuAddr), width, height, stride);
+        updateMD5Plane(m_frameEncoder->m_state[2], recon->getCrAddr(cuAddr), width, height, stride);
     }
     else if (m_param->decodedPictureHashSEI == 2)
     {
         uint32_t width = recon->getWidth();
         uint32_t height = recon->getCUHeight(row);
         uint32_t stride = recon->getStride();
-        if (row == 0)
-        {
-            m_pic->m_crc[0] = m_pic->m_crc[1] = m_pic->m_crc[2] = 0xffff;
-        }
-        updateCRC(recon->getLumaAddr(cuAddr), m_pic->m_crc[0], height, width, stride);
+        if (!row)
+            m_frameEncoder->m_crc[0] = m_frameEncoder->m_crc[1] = m_frameEncoder->m_crc[2] = 0xffff;
+        updateCRC(recon->getLumaAddr(cuAddr), m_frameEncoder->m_crc[0], height, width, stride);
         width  >>= m_hChromaShift;
         height >>= m_vChromaShift;
         stride = recon->getCStride();
 
-        updateCRC(recon->getCbAddr(cuAddr), m_pic->m_crc[1], height, width, stride);
-        updateCRC(recon->getCrAddr(cuAddr), m_pic->m_crc[2], height, width, stride);
+        updateCRC(recon->getCbAddr(cuAddr), m_frameEncoder->m_crc[1], height, width, stride);
+        updateCRC(recon->getCrAddr(cuAddr), m_frameEncoder->m_crc[2], height, width, stride);
     }
     else if (m_param->decodedPictureHashSEI == 3)
     {
@@ -357,17 +290,16 @@
         uint32_t height = recon->getCUHeight(row);
         uint32_t stride = recon->getStride();
         uint32_t cuHeight = g_maxCUSize;
-        if (row == 0)
-        {
-            m_pic->m_checksum[0] = m_pic->m_checksum[1] = m_pic->m_checksum[2] = 0;
-        }
-        updateChecksum(recon->getLumaAddr(), m_pic->m_checksum[0], height, width, stride, row, cuHeight);
+        if (!row)
+            m_frameEncoder->m_checksum[0] = m_frameEncoder->m_checksum[1] = m_frameEncoder->m_checksum[2] = 0;
+        updateChecksum(recon->getLumaAddr(), m_frameEncoder->m_checksum[0], height, width, stride, row, cuHeight);
         width  >>= m_hChromaShift;
         height >>= m_vChromaShift;
         stride = recon->getCStride();
         cuHeight >>= m_vChromaShift;
-        updateChecksum(recon->getCbAddr(), m_pic->m_checksum[1], height, width, stride, row, cuHeight);
-        updateChecksum(recon->getCrAddr(), m_pic->m_checksum[2], height, width, stride, row, cuHeight);
+
+        updateChecksum(recon->getCbAddr(), m_frameEncoder->m_checksum[1], height, width, stride, row, cuHeight);
+        updateChecksum(recon->getCrAddr(), m_frameEncoder->m_checksum[2], height, width, stride, row, cuHeight);
     }
 }
 
@@ -401,15 +333,11 @@
 
         if (!(stride & 31))
             for (; x + 64 <= width; x += 64)
-            {
                 ssd += primitives.sse_pp[LUMA_64x64](fenc + x, stride, rec + x, stride);
-            }
 
         if (!(stride & 15))
             for (; x + 16 <= width; x += 16)
-            {
                 ssd += primitives.sse_pp[LUMA_16x64](fenc + x, stride, rec + x, stride);
-            }
 
         for (; x + 4 <= width; x += 4)
         {
@@ -430,20 +358,14 @@
 
         if (!(stride & 31))
             for (; x + 64 <= width; x += 64)
-            {
                 ssd += primitives.sse_pp[LUMA_64x16](fenc + x, stride, rec + x, stride);
-            }
 
         if (!(stride & 15))
             for (; x + 16 <= width; x += 16)
-            {
                 ssd += primitives.sse_pp[LUMA_16x16](fenc + x, stride, rec + x, stride);
-            }
 
         for (; x + 4 <= width; x += 4)
-        {
             ssd += primitives.sse_pp[LUMA_4x16](fenc + x, stride, rec + x, stride);
-        }
 
         fenc += stride * 16;
         rec += stride * 16;
@@ -456,14 +378,10 @@
 
         if (!(stride & 15))
             for (; x + 16 <= width; x += 16)
-            {
                 ssd += primitives.sse_pp[LUMA_16x4](fenc + x, stride, rec + x, stride);
-            }
 
         for (; x + 4 <= width; x += 4)
-        {
             ssd += primitives.sse_pp[LUMA_4x4](fenc + x, stride, rec + x, stride);
-        }
 
         fenc += stride * 4;
         rec += stride * 4;
@@ -473,7 +391,7 @@
 }
 
 /* Function to calculate SSIM for each row */
-static float calculateSSIM(pixel *pix1, intptr_t stride1, pixel *pix2, intptr_t stride2, int width, int height, void *buf, int32_t* cnt)
+static float calculateSSIM(pixel *pix1, intptr_t stride1, pixel *pix2, intptr_t stride2, int width, int height, void *buf, uint32_t& cnt)
 {
     int z = 0;
     float ssim = 0.0;
@@ -487,56 +405,45 @@
     {
         for (; z <= y; z++)
         {
-            void* swap = sum0;
-            sum0 = sum1;
-            sum1 = (int(*)[4])swap;
+            std::swap(sum0, sum1);
             for (int x = 0; x < width; x += 2)
-            {
                 primitives.ssim_4x4x2_core(&pix1[(4 * x + (z * stride1))], stride1, &pix2[(4 * x + (z * stride2))], stride2, &sum0[x]);
-            }
         }
 
         for (int x = 0; x < width - 1; x += 4)
-        {
             ssim += primitives.ssim_end_4(sum0 + x, sum1 + x, X265_MIN(4, width - x - 1));
-        }
     }
 
-    *cnt = (height - 1) * (width - 1);
+    cnt = (height - 1) * (width - 1);
     return ssim;
 }
 
 void FrameFilter::processSao(int row)
 {
-    const uint32_t numCols = m_pic->getPicSym()->getFrameWidthInCU();
+    const uint32_t numCols = m_frame->getPicSym()->getFrameWidthInCU();
     const uint32_t lineStartCUAddr = row * numCols;
-    SAOParam* saoParam = m_pic->getPicSym()->getSaoParam();
+    SAOParam* saoParam = m_frame->getPicSym()->m_saoParam;
 
-    // NOTE: these flag is not use in this mode
-    assert(saoParam->oneUnitFlag[0] == false);
-    assert(saoParam->oneUnitFlag[1] == false);
-    assert(saoParam->oneUnitFlag[2] == false);
+    // NOTE: these flags are not used in this mode
+    X265_CHECK(!saoParam->oneUnitFlag[0] && !saoParam->oneUnitFlag[1] && !saoParam->oneUnitFlag[2], "invalid SAO flag");
 
     if (saoParam->bSaoFlag[0])
-    {
         m_sao.processSaoUnitRow(saoParam->saoLcuParam[0], row, 0);
-    }
+
     if (saoParam->bSaoFlag[1])
     {
         m_sao.processSaoUnitRow(saoParam->saoLcuParam[1], row, 1);
         m_sao.processSaoUnitRow(saoParam->saoLcuParam[2], row, 2);
     }
 
-    // TODO: this code is NOT VERIFIED because TransformSkip and PCM modes have some bugs, they are never enabled
-    bool  bPCMFilter = (m_pic->getSlice()->getSPS()->getUsePCM() && m_pic->getSlice()->getSPS()->getPCMFilterDisableFlag()) ? true : false;
-    if (bPCMFilter || m_pic->getSlice()->getPPS()->getTransquantBypassEnableFlag())
+    if (m_frame->m_picSym->m_slice->m_pps->bTransquantBypassEnabled)
     {
         for (uint32_t col = 0; col < numCols; col++)
         {
             const uint32_t cuAddr = lineStartCUAddr + col;
-            TComDataCU* cu = m_pic->getCU(cuAddr);
+            TComDataCU* cu = m_frame->getCU(cuAddr);
 
-            xPCMCURestoration(cu, 0, 0);
+            origCUSampleRestoration(cu, 0, 0);
         }
     }
 }

x265-hg20140324.tar.bz2/source/encoder/framefilter.h -> x265-hg20140928.tar.bz2/source/encoder/framefilter.h Changed

@@ -19,63 +19,54 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_FRAMEFILTER_H
 #define X265_FRAMEFILTER_H
 
-#include "TLibCommon/TComPic.h"
-#include "TLibCommon/TComLoopFilter.h"
-#include "TLibEncoder/TEncSampleAdaptiveOffset.h"
+#include "common.h"
+#include "frame.h"
+#include "deblock.h"
+#include "sao.h"
 
 namespace x265 {
 // private x265 namespace
 
 class Encoder;
+class Entropy;
+class FrameEncoder;
+struct ThreadLocalData;
 
 // Manages the processing of a single frame loopfilter
 class FrameFilter
 {
 public:
 
-    FrameFilter();
+    x265_param*   m_param;
+    Frame*        m_frame;
+    FrameEncoder* m_frameEncoder;
+    int           m_hChromaShift;
+    int           m_vChromaShift;
+    int           m_pad[2];
 
-    virtual ~FrameFilter() {}
+    Deblock       m_deblock;
+    SAO           m_sao;
+    int           m_numRows;
+    int           m_saoRowDelay;
+    
+    void*         m_ssimBuf; /* Temp storage for ssim computation */
 
-    void init(Encoder *top, int numRows, TEncSbac* rdGoOnSbacCoder);
+    FrameFilter();
 
+    void init(Encoder *top, FrameEncoder *frame, int numRows);
     void destroy();
 
-    void start(TComPic *pic);
-    void end();
+    void start(Frame *pic, Entropy& initState, int qp);
 
-    void processRow(int row, Encoder* cfg);
-    void processRowPost(int row, Encoder* cfg);
+    void processRow(int row);
+    void processRowPost(int row);
     void processSao(int row);
-
-protected:
-
-    x265_param*                 m_param;
-    TComPic*                    m_pic;
-    int                         m_hChromaShift;
-    int                         m_vChromaShift;
-
-public:
-
-    TComLoopFilter              m_loopFilter;
-    TEncSampleAdaptiveOffset    m_sao;
-    int                         m_numRows;
-    int                         m_saoRowDelay;
-
-    // SAO
-    TEncEntropy                 m_entropyCoder;
-    TEncSbac                    m_rdGoOnSbacCoder;
-    TEncBinCABAC                m_rdGoOnBinCodersCABAC;
-    TComBitCounter              m_bitCounter;
-    TEncSbac*                   m_rdGoOnSbacCoderRow0;  // for bitstream exact only, depends on HM's bug
-    /* Temp storage for ssim computation that doesn't need repeated malloc */
-    void*                       m_ssimBuf;
 };
 }

x265-hg20140928.tar.bz2/source/encoder/level.cpp Added

@@ -0,0 +1,364 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "common.h"
+#include "slice.h"
+#include "level.h"
+
+namespace x265 {
+typedef struct
+{
+    uint32_t maxLumaSamples;
+    uint32_t maxLumaSamplesPerSecond;
+    uint32_t maxBitrateMain;
+    uint32_t maxBitrateHigh;
+    uint32_t maxCpbSizeMain;
+    uint32_t maxCpbSizeHigh;
+    uint32_t minCompressionRatio;
+    Level::Name levelEnum;
+    const char* name;
+    int levelIdc;
+} LevelSpec;
+
+LevelSpec levels[] =
+{
+    { 36864,    552960,     128,      MAX_UINT, 350,    MAX_UINT, 2, Level::LEVEL1,   "1",   10 },
+    { 122880,   3686400,    1500,     MAX_UINT, 1500,   MAX_UINT, 2, Level::LEVEL2,   "2",   20 },
+    { 245760,   7372800,    3000,     MAX_UINT, 3000,   MAX_UINT, 2, Level::LEVEL2_1, "2.1", 21 },
+    { 552960,   16588800,   6000,     MAX_UINT, 6000,   MAX_UINT, 2, Level::LEVEL3,   "3",   30 },
+    { 983040,   33177600,   10000,    MAX_UINT, 10000,  MAX_UINT, 2, Level::LEVEL3_1, "3.1", 31 },
+    { 2228224,  66846720,   12000,    30000,    12000,  30000,    4, Level::LEVEL4,   "4",   40 },
+    { 2228224,  133693440,  20000,    50000,    20000,  50000,    4, Level::LEVEL4_1, "4.1", 41 },
+    { 8912896,  267386880,  25000,    100000,   25000,  100000,   6, Level::LEVEL5,   "5",   50 },
+    { 8912896,  534773760,  40000,    160000,   40000,  160000,   8, Level::LEVEL5_1, "5.1", 51 },
+    { 8912896,  1069547520, 60000,    240000,   60000,  240000,   8, Level::LEVEL5_2, "5.2", 52 },
+    { 35651584, 1069547520, 60000,    240000,   60000,  240000,   8, Level::LEVEL6,   "6",   60 },
+    { 35651584, 2139095040, 120000,   480000,   120000, 480000,   8, Level::LEVEL6_1, "6.1", 61 },
+    { 35651584, 4278190080U, 240000,  800000,   240000, 800000,   6, Level::LEVEL6_2, "6.2", 62 },
+};
+
+/* determine minimum decoder level required to decode the described video */
+void determineLevel(const x265_param &param, VPS& vps)
+{
+    if (param.bLossless)
+        vps.ptl.profileIdc = Profile::NONE;
+    else if (param.internalCsp == X265_CSP_I420)
+    {
+        if (param.internalBitDepth == 8)
+        {
+            if (param.keyframeMax == 1 && param.maxNumReferences == 1)
+                vps.ptl.profileIdc = Profile::MAINSTILLPICTURE;
+            else 
+                vps.ptl.profileIdc = Profile::MAIN;
+        }
+        else if (param.internalBitDepth == 10)
+            vps.ptl.profileIdc = Profile::MAIN10;
+    }
+    else
+    {
+        /* TODO: Range extension profiles */
+        vps.ptl.profileIdc = Profile::NONE;
+    }
+
+    /* determine which profiles are compatible with this stream */
+
+    memset(vps.ptl.profileCompatibilityFlag, 0, sizeof(vps.ptl.profileCompatibilityFlag));
+    vps.ptl.profileCompatibilityFlag[vps.ptl.profileIdc] = true;
+    if (vps.ptl.profileIdc == Profile::MAIN10 && param.internalBitDepth == 8)
+        vps.ptl.profileCompatibilityFlag[Profile::MAIN] = true;
+    else if (vps.ptl.profileIdc == Profile::MAIN)
+        vps.ptl.profileCompatibilityFlag[Profile::MAIN10] = true;
+    else if (vps.ptl.profileIdc == Profile::MAINSTILLPICTURE)
+    {
+        vps.ptl.profileCompatibilityFlag[Profile::MAIN] = true;
+        vps.ptl.profileCompatibilityFlag[Profile::MAIN10] = true;
+    }
+
+    uint32_t lumaSamples = param.sourceWidth * param.sourceHeight;
+    uint32_t samplesPerSec = (uint32_t)(lumaSamples * ((double)param.fpsNum / param.fpsDenom));
+    uint32_t bitrate = param.rc.vbvMaxBitrate ? param.rc.vbvMaxBitrate : param.rc.bitrate;
+
+    const uint32_t MaxDpbPicBuf = 6;
+    vps.ptl.levelIdc = Level::NONE;
+    vps.ptl.tierFlag = Level::MAIN;
+
+    const size_t NumLevels = sizeof(levels) / sizeof(levels[0]);
+    uint32_t i;
+    for (i = 0; i < NumLevels; i++)
+    {
+        if (lumaSamples > levels[i].maxLumaSamples)
+            continue;
+        else if (samplesPerSec > levels[i].maxLumaSamplesPerSecond)
+            continue;
+        else if (bitrate > levels[i].maxBitrateMain && levels[i].maxBitrateHigh == MAX_UINT)
+            continue;
+        else if (bitrate > levels[i].maxBitrateHigh)
+            continue;
+        else if (param.sourceWidth > sqrt(levels[i].maxLumaSamples * 8.0f))
+            continue;
+        else if (param.sourceHeight > sqrt(levels[i].maxLumaSamples * 8.0f))
+            continue;
+
+        uint32_t maxDpbSize = MaxDpbPicBuf;
+        if (lumaSamples <= (levels[i].maxLumaSamples >> 2))
+            maxDpbSize = X265_MIN(4 * MaxDpbPicBuf, 16);
+        else if (lumaSamples <= (levels[i].maxLumaSamples >> 1))
+            maxDpbSize = X265_MIN(2 * MaxDpbPicBuf, 16);
+        else if (lumaSamples <= ((3 * levels[i].maxLumaSamples) >> 2))
+            maxDpbSize = X265_MIN((4 * MaxDpbPicBuf) / 3, 16);
+
+        /* The value of sps_max_dec_pic_buffering_minus1[ HighestTid ] + 1 shall be less than
+         * or equal to MaxDpbSize */
+        if (vps.maxDecPicBuffering > maxDpbSize)
+            continue;
+
+        /* For level 5 and higher levels, the value of CtbSizeY shall be equal to 32 or 64 */
+        if (levels[i].levelEnum >= Level::LEVEL5 && param.maxCUSize < 32)
+        {
+            x265_log(&param, X265_LOG_WARNING, "level %s detected, but CTU size 16 is non-compliant\n", levels[i].name);
+            vps.ptl.profileIdc = Profile::NONE;
+            vps.ptl.levelIdc = Level::NONE;
+            vps.ptl.tierFlag = Level::MAIN;
+            x265_log(&param, X265_LOG_INFO, "NONE profile, Level-NONE (Main tier)\n");
+            return;
+        }
+
+        /* The value of NumPocTotalCurr shall be less than or equal to 8 */
+        int numPocTotalCurr = param.maxNumReferences + vps.numReorderPics;
+        if (numPocTotalCurr > 8)
+        {
+            x265_log(&param, X265_LOG_WARNING, "level %s detected, but NumPocTotalCurr (total references) is non-compliant\n", levels[i].name);
+            vps.ptl.profileIdc = Profile::NONE;
+            vps.ptl.levelIdc = Level::NONE;
+            vps.ptl.tierFlag = Level::MAIN;
+            x265_log(&param, X265_LOG_INFO, "NONE profile, Level-NONE (Main tier)\n");
+            return;
+        }
+
+        vps.ptl.levelIdc = levels[i].levelEnum;
+        if (bitrate > levels[i].maxBitrateMain && bitrate <= levels[i].maxBitrateHigh &&
+            levels[i].maxBitrateHigh != MAX_UINT)
+            vps.ptl.tierFlag = Level::HIGH;
+        else
+            vps.ptl.tierFlag = Level::MAIN;
+        break;
+    }
+
+    static const char *profiles[] = { "None", "Main", "Main10", "Mainstillpicture" };
+    static const char *tiers[]    = { "Main", "High" };
+    x265_log(&param, X265_LOG_INFO, "%s profile, Level-%s (%s tier)\n",
+             profiles[vps.ptl.profileIdc], levels[i].name, tiers[vps.ptl.tierFlag]);
+}
+
+/* enforce a maximum decoder level requirement, in other words assure that a
+ * decoder of the specified level may decode the video about to be created.
+ * Lower parameters where necessary to ensure the video will be decodable by a
+ * decoder meeting this level of requirement.  Some parameters (resolution and
+ * frame rate) are non-negotiable and thus this function may fail. In those
+ * circumstances it will be quite noisy */
+bool enforceLevel(x265_param& param, VPS& vps)
+{
+    vps.numReorderPics = (param.bBPyramid && param.bframes > 1) ? 2 : 1;
+    vps.maxDecPicBuffering = X265_MIN(MAX_NUM_REF, X265_MAX(vps.numReorderPics + 1, (uint32_t)param.maxNumReferences) + vps.numReorderPics);
+
+    /* no level specified by user, just auto-detect from the configuration */
+    if (param.levelIdc <= 0)
+        return true;
+
+    uint32_t level = 0;
+    while (levels[level].levelIdc != param.levelIdc && level + 1 < sizeof(levels) / sizeof(levels[0]))
+        level++;
+    if (levels[level].levelIdc != param.levelIdc)
+    {
+        x265_log(&param, X265_LOG_WARNING, "specified level %d does not exist\n", param.levelIdc);
+        return false;
+    }
+
+    LevelSpec& l = levels[level];
+    bool highTier = !!param.bHighTier;
+    if (highTier && l.maxBitrateHigh == MAX_UINT)
+    {
+        highTier = false;
+        x265_log(&param, X265_LOG_WARNING, "Level %s has no High tier, using Main tier\n", l.name);
+    }
+
+    uint32_t lumaSamples = param.sourceWidth * param.sourceHeight;
+    uint32_t samplesPerSec = (uint32_t)(lumaSamples * ((double)param.fpsNum / param.fpsDenom));
+    bool ok = true;
+    if (lumaSamples > l.maxLumaSamples)
+        ok = false;
+    else if (param.sourceWidth > sqrt(l.maxLumaSamples * 8.0f))
+        ok = false;
+    else if (param.sourceHeight > sqrt(l.maxLumaSamples * 8.0f))
+        ok = false;
+    if (!ok)
+    {
+        x265_log(&param, X265_LOG_WARNING, "picture dimensions are out of range for specified level\n");
+        return false;
+    }
+    else if (samplesPerSec > l.maxLumaSamplesPerSecond)
+    {
+        x265_log(&param, X265_LOG_WARNING, "frame rate is out of range for specified level\n");
+        return false;
+    }
+
+    if ((uint32_t)param.rc.vbvMaxBitrate > (highTier ? l.maxBitrateHigh : l.maxBitrateMain))
+    {
+        param.rc.vbvMaxBitrate = highTier ? l.maxBitrateHigh : l.maxBitrateMain;
+        x265_log(&param, X265_LOG_INFO, "lowering VBV max bitrate to %dKbps\n", param.rc.vbvMaxBitrate);
+    }
+    if ((uint32_t)param.rc.vbvBufferSize > (highTier ? l.maxCpbSizeHigh : l.maxCpbSizeMain))
+    {
+        param.rc.vbvMaxBitrate = highTier ? l.maxCpbSizeHigh : l.maxCpbSizeMain;
+        x265_log(&param, X265_LOG_INFO, "lowering VBV buffer size to %dKb\n", param.rc.vbvBufferSize);
+    }
+
+    switch (param.rc.rateControlMode)
+    {
+    case X265_RC_ABR:
+        if ((uint32_t)param.rc.bitrate > (highTier ? l.maxBitrateHigh : l.maxBitrateMain))
+        {
+            param.rc.bitrate = l.maxBitrateHigh;
+            x265_log(&param, X265_LOG_INFO, "lowering target bitrate to High tier limit of %dKbps\n", param.rc.bitrate);
+        }
+        break;
+
+    case X265_RC_CQP:
+        x265_log(&param, X265_LOG_WARNING, "Constant QP is inconsistent with specifying a decoder level, no bitrate guarantee is possible.\n");
+        return false;
+
+    case X265_RC_CRF:
+        if (!param.rc.vbvBufferSize || !param.rc.vbvMaxBitrate)
+        {
+            if (!param.rc.vbvMaxBitrate)
+                param.rc.vbvMaxBitrate = highTier ? l.maxBitrateHigh : l.maxBitrateMain;
+            if (!param.rc.vbvBufferSize)
+                param.rc.vbvBufferSize = highTier ? l.maxCpbSizeHigh : l.maxCpbSizeMain;
+            x265_log(&param, X265_LOG_WARNING, "Specifying a decoder level with constant rate factor rate-control requires\n");
+            x265_log(&param, X265_LOG_WARNING, "enabling VBV with vbv-bufsize=%dkb vbv-maxrate=%dkbps. VBV outputs are non-deterministic!\n",
+                     param.rc.vbvBufferSize, param.rc.vbvMaxBitrate);
+        }
+        break;
+
+    default:
+        x265_log(&param, X265_LOG_ERROR, "Unknown rate control mode is inconsistent with specifying a decoder level\n");
+        return false;
+    }
+
+    /* The value of sps_max_dec_pic_buffering_minus1[ HighestTid ] + 1 shall be less than or equal to MaxDpbSize */
+    const uint32_t MaxDpbPicBuf = 6;
+    uint32_t maxDpbSize = MaxDpbPicBuf;
+    if (lumaSamples <= (l.maxLumaSamples >> 2))
+        maxDpbSize = X265_MIN(4 * MaxDpbPicBuf, 16);
+    else if (lumaSamples <= (l.maxLumaSamples >> 1))
+        maxDpbSize = X265_MIN(2 * MaxDpbPicBuf, 16);
+    else if (lumaSamples <= ((3 * l.maxLumaSamples) >> 2))
+        maxDpbSize = X265_MIN((4 * MaxDpbPicBuf) / 3, 16);
+
+    int savedRefCount = param.maxNumReferences;
+    while (vps.maxDecPicBuffering > maxDpbSize && param.maxNumReferences > 1)
+    {
+        param.maxNumReferences--;
+        vps.maxDecPicBuffering = X265_MIN(MAX_NUM_REF, X265_MAX(vps.numReorderPics + 1, (uint32_t)param.maxNumReferences) + vps.numReorderPics);
+    }
+    if (param.maxNumReferences != savedRefCount)
+        x265_log(&param, X265_LOG_INFO, "Lowering max references to %d to meet level requirement\n", param.maxNumReferences);
+
+    /* For level 5 and higher levels, the value of CtbSizeY shall be equal to 32 or 64 */
+    if (param.levelIdc >= 50 && param.maxCUSize < 32)
+    {
+        param.maxCUSize = 32;
+        x265_log(&param, X265_LOG_INFO, "Levels 5.0 and above require a maximum CTU size of at least 32, using --ctu 32\n");
+    }
+
+    /* The value of NumPocTotalCurr shall be less than or equal to 8 */
+    int numPocTotalCurr = param.maxNumReferences + !!param.bframes;
+    if (numPocTotalCurr > 8)
+    {
+        param.maxNumReferences = 8 - !!param.bframes;
+        x265_log(&param, X265_LOG_INFO, "Lowering max references to %d to meet numPocTotalCurr requirement\n", param.maxNumReferences);
+    }
+
+    return true;
+}
+
+extern "C"
+int x265_param_apply_profile(x265_param *param, const char *profile)
+{
+    if (!profile)
+        return 0;
+    if (!strcmp(profile, "main"))
+    {
+        /* SPSs shall have chroma_format_idc equal to 1 only */
+        param->internalCsp = X265_CSP_I420;
+
+#if HIGH_BIT_DEPTH
+        /* SPSs shall have bit_depth_luma_minus8 equal to 0 only */
+        x265_log(param, X265_LOG_ERROR, "Main profile not supported, compiled for Main10.\n");
+        return -1;
+#endif
+    }
+    else if (!strcmp(profile, "main10"))
+    {
+        /* SPSs shall have chroma_format_idc equal to 1 only */
+        param->internalCsp = X265_CSP_I420;
+
+        /* SPSs shall have bit_depth_luma_minus8 in the range of 0 to 2, inclusive 
+         * this covers all builds of x265, currently */
+    }
+    else if (!strcmp(profile, "mainstillpicture") || !strcmp(profile, "msp"))
+    {
+        /* SPSs shall have chroma_format_idc equal to 1 only */
+        param->internalCsp = X265_CSP_I420;
+
+        /* SPSs shall have sps_max_dec_pic_buffering_minus1[ sps_max_sub_layers_minus1 ] equal to 0 only */
+        param->maxNumReferences = 1;
+
+        /* The bitstream shall contain only one picture (we do not enforce this) */
+        /* just in case the user gives us more than one picture: */
+        param->keyframeMax = 1;
+        param->bOpenGOP = 0;
+        param->bRepeatHeaders = 1;
+        param->lookaheadDepth = 0;
+        param->bframes = 0;
+        param->scenecutThreshold = 0;
+        param->bFrameAdaptive = 0;
+        param->rc.cuTree = 0;
+        param->bEnableWeightedPred = 0;
+        param->bEnableWeightedBiPred = 0;
+
+#if HIGH_BIT_DEPTH
+        /* SPSs shall have bit_depth_luma_minus8 equal to 0 only */
+        x265_log(param, X265_LOG_ERROR, "Mainstillpicture profile not supported, compiled for Main10.\n");
+        return -1;
+#endif
+    }
+    else
+    {
+        x265_log(param, X265_LOG_ERROR, "unknown profile <%s>\n", profile);
+        return -1;
+    }
+
+    return 0;
+}
+}

x265-hg20140928.tar.bz2/source/encoder/level.h Added

@@ -0,0 +1,39 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_LEVEL_H
+#define X265_LEVEL_H 1
+
+#include "common.h"
+#include "x265.h"
+
+namespace x265 {
+// encoder private namespace
+
+struct VPS;
+void determineLevel(const x265_param &param, VPS& vps);
+bool enforceLevel(x265_param& param, VPS& vps);
+
+}
+
+#endif // ifndef X265_LEVEL_H

x265-hg20140324.tar.bz2/source/encoder/motion.cpp -> x265-hg20140928.tar.bz2/source/encoder/motion.cpp Changed

x265-hg20140324.tar.bz2/source/encoder/motion.h -> x265-hg20140928.tar.bz2/source/encoder/motion.h Changed

x265-hg20140928.tar.bz2/source/encoder/nal.cpp Added

@@ -0,0 +1,218 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "common.h"
+#include "bitstream.h"
+#include "nal.h"
+
+using namespace x265;
+
+NALList::NALList()
+    : m_numNal(0)
+    , m_buffer(NULL)
+    , m_occupancy(0)
+    , m_allocSize(0)
+    , m_extraBuffer(NULL)
+    , m_extraOccupancy(0)
+    , m_extraAllocSize(0)
+{}
+
+void NALList::takeContents(NALList& other)
+{
+    /* take other NAL buffer, discard our old one */
+    X265_FREE(m_buffer);
+    m_buffer = other.m_buffer;
+    m_allocSize = other.m_allocSize;
+    m_occupancy = other.m_occupancy;
+
+    /* copy packet data */
+    m_numNal = other.m_numNal;
+    memcpy(m_nal, other.m_nal, sizeof(x265_nal) * m_numNal);
+
+    /* reset other list, re-allocate their buffer with same size */
+    other.m_numNal = 0;
+    other.m_occupancy = 0;
+    other.m_buffer = X265_MALLOC(uint8_t, m_allocSize);
+}
+
+void NALList::serialize(NalUnitType nalUnitType, const Bitstream& bs)
+{
+    static const char startCodePrefix[] = { 0, 0, 0, 1 };
+
+    uint32_t payloadSize = bs.getNumberOfWrittenBytes();
+    const uint8_t* bpayload = bs.getFIFO();
+    if (!bpayload)
+        return;
+
+    uint32_t nextSize = m_occupancy + sizeof(startCodePrefix) + 2 + payloadSize + (payloadSize >> 1) + m_extraOccupancy;
+    if (nextSize > m_allocSize)
+    {
+        uint8_t *temp = X265_MALLOC(uint8_t, nextSize);
+        if (temp)
+        {
+            memcpy(temp, m_buffer, m_occupancy);
+
+            /* fixup existing payload pointers */
+            for (uint32_t i = 0; i < m_numNal; i++)
+                m_nal[i].payload = temp + (m_nal[i].payload - m_buffer);
+
+            X265_FREE(m_buffer);
+            m_buffer = temp;
+            m_allocSize = nextSize;
+        }
+        else
+        {
+            x265_log(NULL, X265_LOG_ERROR, "Unable to realloc access unit buffer\n");
+            return;
+        }
+    }
+
+    uint8_t *out = m_buffer + m_occupancy;
+    uint32_t bytes = 0;
+
+    if (!m_numNal || nalUnitType == NAL_UNIT_SPS || nalUnitType == NAL_UNIT_PPS)
+    {
+        memcpy(out, startCodePrefix, 4);
+        bytes += 4;
+    }
+    else
+    {
+        memcpy(out, startCodePrefix + 1, 3);
+        bytes += 3;
+    }
+
+    /* 16 bit NAL header:
+     * forbidden_zero_bit       1-bit
+     * nal_unit_type            6-bits
+     * nuh_reserved_zero_6bits  6-bits
+     * nuh_temporal_id_plus1    3-bits */
+    out[bytes++] = (uint8_t)nalUnitType << 1;
+    out[bytes++] = 1;
+
+    /* 7.4.1 ...
+     * Within the NAL unit, the following three-byte sequences shall not occur at
+     * any byte-aligned position:
+     *  - 0x000000
+     *  - 0x000001
+     *  - 0x000002 */
+    for (uint32_t i = 0; i < payloadSize; i++)
+    {
+        if (i > 2 && !out[bytes - 2] && !out[bytes - 3] && out[bytes - 1] <= 0x03)
+        {
+            /* inject 0x03 to prevent emulating a start code */
+            out[bytes] = out[bytes - 1];
+            out[bytes - 1] = 0x03;
+            bytes++;
+        }
+
+        out[bytes++] = bpayload[i];
+    }
+
+    X265_CHECK(bytes <= 4 + 2 + payloadSize + (payloadSize >> 1), "NAL buffer overflow\n");
+
+    if (m_extraOccupancy)
+    {
+        /* these bytes were escaped by serializeSubstreams */
+        memcpy(out + bytes, m_extraBuffer, m_extraOccupancy);
+        bytes += m_extraOccupancy;
+        m_extraOccupancy = 0;
+    }
+
+    /* 7.4.1.1
+     * ... when the last byte of the RBSP data is equal to 0x00 (which can
+     * only occur when the RBSP ends in a cabac_zero_word), a final byte equal
+     * to 0x03 is appended to the end of the data.  */
+    if (!out[bytes - 1])
+        out[bytes++] = 0x03;
+    m_occupancy += bytes;
+
+    X265_CHECK(m_numNal < (uint32_t)MAX_NAL_UNITS, "NAL count overflow\n");
+
+    x265_nal& nal = m_nal[m_numNal++];
+    nal.type = nalUnitType;
+    nal.sizeBytes = bytes;
+    nal.payload = out;
+}
+
+/* concatenate and escape WPP sub-streams, return escaped row lengths.
+ * These streams will be appended to the next serialized NAL */
+uint32_t NALList::serializeSubstreams(uint32_t* streamSizeBytes, uint32_t streamCount, const Bitstream* streams)
+{
+    uint32_t maxStreamSize = 0;
+    uint32_t estSize = 0;
+    for (uint32_t s = 0; s < streamCount; s++)
+        estSize += streams[s].getNumberOfWrittenBytes();
+    estSize += estSize >> 1;
+
+    if (estSize > m_extraAllocSize)
+    {
+        uint8_t *temp = X265_MALLOC(uint8_t, estSize);
+        if (temp)
+        {
+            X265_FREE(m_extraBuffer);
+            m_extraBuffer = temp;
+            m_extraAllocSize = estSize;
+        }
+        else
+        {
+            x265_log(NULL, X265_LOG_ERROR, "Unable to realloc WPP substream concatenation buffer\n");
+            return 0;
+        }
+    }
+
+    uint32_t bytes = 0;
+    uint8_t *out = m_extraBuffer;
+    for (uint32_t s = 0; s < streamCount; s++)
+    {
+        const Bitstream& stream = streams[s];
+        uint32_t inSize = stream.getNumberOfWrittenBytes();
+        const uint8_t *inBytes = stream.getFIFO();
+        uint32_t prevBufSize = bytes;
+
+        if (inBytes)
+        {
+            for (uint32_t i = 0; i < inSize; i++)
+            {
+                if (bytes > 2 && !out[bytes - 2] && !out[bytes - 3] && out[bytes - 1] <= 0x03)
+                {
+                    /* inject 0x03 to prevent emulating a start code */
+                    out[bytes] = out[bytes - 1];
+                    out[bytes - 1] = 0x03;
+                    bytes++;
+                }
+
+                out[bytes++] = inBytes[i];
+            }
+        }
+
+        if (s < streamCount - 1)
+        {
+            streamSizeBytes[s] = bytes - prevBufSize;
+            if (streamSizeBytes[s] > maxStreamSize)
+                maxStreamSize = streamSizeBytes[s];
+        }
+    }
+
+    m_extraOccupancy = bytes;
+    return maxStreamSize;
+}

x265-hg20140928.tar.bz2/source/encoder/nal.h Added

@@ -0,0 +1,64 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_NAL_H
+#define X265_NAL_H
+
+#include "common.h"
+#include "x265.h"
+
+namespace x265 {
+// private namespace
+
+class Bitstream;
+
+class NALList
+{
+    static const int MAX_NAL_UNITS = 16;
+
+public:
+
+    x265_nal    m_nal[MAX_NAL_UNITS];
+    uint32_t    m_numNal;
+
+    uint8_t*    m_buffer;
+    uint32_t    m_occupancy;
+    uint32_t    m_allocSize;
+
+    uint8_t*    m_extraBuffer;
+    uint32_t    m_extraOccupancy;
+    uint32_t    m_extraAllocSize;
+
+    NALList();
+    ~NALList() { X265_FREE(m_buffer); X265_FREE(m_extraBuffer); }
+
+    void takeContents(NALList& other);
+
+    void serialize(NalUnitType nalUnitType, const Bitstream& bs);
+
+    uint32_t serializeSubstreams(uint32_t* streamSizeBytes, uint32_t streamCount, const Bitstream* streams);
+};
+
+}
+
+#endif // ifndef X265_NAL_H

x265-hg20140928.tar.bz2/source/encoder/predict.cpp Added

@@ -0,0 +1,658 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Deepthi Nandakumar <deepthi@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "predict.h"
+#include "primitives.h"
+#include "common.h"
+
+using namespace x265;
+
+namespace
+{
+inline pixel weightBidir(int w0, int16_t P0, int w1, int16_t P1, int round, int shift, int offset)
+{
+    return Clip((w0 * (P0 + IF_INTERNAL_OFFS) + w1 * (P1 + IF_INTERNAL_OFFS) + round + (offset << (shift - 1))) >> shift);
+}
+}
+
+Predict::Predict()
+{
+    m_predBuf = NULL;
+    m_refAbove = NULL;
+    m_refAboveFlt = NULL;
+    m_refLeft = NULL;
+    m_refLeftFlt = NULL;
+    m_immedVals = NULL;
+}
+
+Predict::~Predict()
+{
+    X265_FREE(m_predBuf);
+    X265_FREE(m_refAbove);
+    X265_FREE(m_immedVals);
+    m_predShortYuv[0].destroy();
+    m_predShortYuv[1].destroy();
+}
+
+bool Predict::allocBuffers(int csp)
+{
+    m_csp = csp;
+
+    int predBufHeight = ((MAX_CU_SIZE + 2) << 4);
+    int predBufStride = ((MAX_CU_SIZE + 8) << 4);
+    CHECKED_MALLOC(m_predBuf, pixel, predBufStride * predBufHeight);
+    CHECKED_MALLOC(m_immedVals, int16_t, 64 * (64 + NTAPS_LUMA - 1));
+    CHECKED_MALLOC(m_refAbove, pixel, 12 * MAX_CU_SIZE);
+
+    m_refAboveFlt = m_refAbove + 3 * MAX_CU_SIZE;
+    m_refLeft = m_refAboveFlt + 3 * MAX_CU_SIZE;
+    m_refLeftFlt = m_refLeft + 3 * MAX_CU_SIZE;
+
+    return m_predShortYuv[0].create(MAX_CU_SIZE, MAX_CU_SIZE, csp) &&
+           m_predShortYuv[1].create(MAX_CU_SIZE, MAX_CU_SIZE, csp);
+
+fail:
+    return false;
+}
+
+void Predict::predIntraLumaAng(uint32_t dirMode, pixel* dst, intptr_t stride, uint32_t log2TrSize)
+{
+    int tuSize = 1 << log2TrSize;
+
+    pixel *refLft, *refAbv;
+
+    if (!(g_intraFilterFlags[dirMode] & tuSize))
+    {
+        refLft = m_refLeft + tuSize - 1;
+        refAbv = m_refAbove + tuSize - 1;
+    }
+    else
+    {
+        refLft = m_refLeftFlt + tuSize - 1;
+        refAbv = m_refAboveFlt + tuSize - 1;
+    }
+
+    bool bFilter = log2TrSize <= 4;
+    int sizeIdx = log2TrSize - 2;
+    X265_CHECK(sizeIdx >= 0 && sizeIdx < 4, "intra block size is out of range\n");
+    primitives.intra_pred[dirMode][sizeIdx](dst, stride, refLft, refAbv, dirMode, bFilter);
+}
+
+void Predict::predIntraChromaAng(pixel* src, uint32_t dirMode, pixel* dst, intptr_t stride, uint32_t log2TrSizeC, int chFmt)
+{
+    int tuSize = 1 << log2TrSizeC;
+    int tuSize2 = tuSize << 1;
+
+    // Create the prediction
+    const int bufOffset = tuSize - 1;
+    pixel buf0[3 * MAX_CU_SIZE];
+    pixel buf1[3 * MAX_CU_SIZE];
+    pixel* above;
+    pixel* left = buf0 + bufOffset;
+
+    int limit = (dirMode <= 25 && dirMode >= 11) ? (tuSize + 1 + 1) : (tuSize2 + 1);
+    for (int k = 0; k < limit; k++)
+        left[k] = src[k * ADI_BUF_STRIDE];
+
+    bool bUseFilteredPredictions = (chFmt == X265_CSP_I444 && (g_intraFilterFlags[dirMode] & tuSize));
+
+    if (bUseFilteredPredictions)
+    {
+        // generate filtered intra prediction samples
+        buf0[bufOffset - 1] = src[1];
+        left = buf1 + bufOffset;
+        for (int i = 0; i < tuSize2; i++)
+            left[i] = (buf0[bufOffset + i - 1] + 2 * buf0[bufOffset + i] + buf0[bufOffset + i + 1] + 2) >> 2;
+        left[tuSize2] = buf0[bufOffset + tuSize2];
+
+        above = buf0 + bufOffset;
+        above[0] = left[0];
+        for (int i = 1; i < tuSize2; i++)
+            above[i] = (src[i - 1] + 2 * src[i] + src[i + 1] + 2) >> 2;
+        above[tuSize2] = src[tuSize2];
+    }
+    else
+    {
+        above = buf1 + bufOffset;
+        memcpy(above, src, (tuSize2 + 1) * sizeof(pixel));
+    }
+
+    int sizeIdx = log2TrSizeC - 2;
+    X265_CHECK(sizeIdx >= 0 && sizeIdx < 4, "intra block size is out of range\n");
+    primitives.intra_pred[dirMode][sizeIdx](dst, stride, left, above, dirMode, 0);
+}
+
+void Predict::prepMotionCompensation(TComDataCU* cu, int partIdx)
+{
+    m_slice = cu->m_slice;
+    cu->getPartIndexAndSize(partIdx, m_partAddr, m_width, m_height);
+    m_cuAddr = cu->getAddr();
+    m_zOrderIdxinCU = cu->getZorderIdxInCU();
+
+    m_mvField[0] = cu->getCUMvField(REF_PIC_LIST_0);
+    m_mvField[1] = cu->getCUMvField(REF_PIC_LIST_1);
+
+    m_clippedMv[0] = m_mvField[0]->getMv(m_partAddr);
+    m_clippedMv[1] = m_mvField[1]->getMv(m_partAddr);
+    cu->clipMv(m_clippedMv[0]);
+    cu->clipMv(m_clippedMv[1]);
+}
+
+void Predict::motionCompensation(TComYuv* predYuv, bool bLuma, bool bChroma)
+{
+    if (m_slice->isInterP())
+    {
+        /* P Slice */
+
+        WeightValues wv0[3];
+        int refIdx0 = m_mvField[0]->getRefIdx(m_partAddr);
+        X265_CHECK(refIdx0 >= 0, "invalid P refidx\n");
+        X265_CHECK(refIdx0 < m_slice->m_numRefIdx[0], "P refidx out of range\n");
+        WeightParam *wp0 = m_slice->m_weightPredTable[0][refIdx0];
+
+        if (m_slice->m_pps->bUseWeightPred && wp0->bPresentFlag)
+        {
+            for (int plane = 0; plane < 3; plane++)
+            {
+                wv0[plane].w      = wp0[plane].inputWeight;
+                wv0[plane].offset = wp0[plane].inputOffset * (1 << (X265_DEPTH - 8));
+                wv0[plane].shift  = wp0[plane].log2WeightDenom;
+                wv0[plane].round  = wp0[plane].log2WeightDenom >= 1 ? 1 << (wp0[plane].log2WeightDenom - 1) : 0;
+            }
+
+            ShortYuv* shortYuv = &m_predShortYuv[0];
+
+            if (bLuma)
+                predInterLumaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), shortYuv, &m_clippedMv[0]);
+            if (bChroma)
+                predInterChromaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), shortYuv, &m_clippedMv[0]);
+
+            addWeightUni(shortYuv, wv0, predYuv, bLuma, bChroma);
+        }
+        else
+        {
+            if (bLuma)
+                predInterLumaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), predYuv, &m_clippedMv[0]);
+            if (bChroma)
+                predInterChromaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), predYuv, &m_clippedMv[0]);
+        }
+    }
+    else
+    {
+        /* B Slice */
+
+        WeightValues wv0[3], wv1[3];
+        WeightParam *pwp0, *pwp1;
+
+        int refIdx0 = m_mvField[0]->getRefIdx(m_partAddr);
+        int refIdx1 = m_mvField[1]->getRefIdx(m_partAddr);
+
+        if (m_slice->m_pps->bUseWeightedBiPred)
+        {
+            pwp0 = refIdx0 >= 0 ? m_slice->m_weightPredTable[0][refIdx0] : NULL;
+            pwp1 = refIdx1 >= 0 ? m_slice->m_weightPredTable[1][refIdx1] : NULL;
+
+            if (pwp0 && pwp1 && (pwp0->bPresentFlag || pwp1->bPresentFlag))
+            {
+                /* biprediction weighting */
+                for (int plane = 0; plane < 3; plane++)
+                {
+                    wv0[plane].w = pwp0[plane].inputWeight;
+                    wv0[plane].o = pwp0[plane].inputOffset * (1 << (X265_DEPTH - 8));
+                    wv0[plane].shift = pwp0[plane].log2WeightDenom;
+                    wv0[plane].round = 1 << pwp0[plane].log2WeightDenom;
+
+                    wv1[plane].w = pwp1[plane].inputWeight;
+                    wv1[plane].o = pwp1[plane].inputOffset * (1 << (X265_DEPTH - 8));
+                    wv1[plane].shift = wv0[plane].shift;
+                    wv1[plane].round = wv0[plane].round;
+                }
+            }
+            else
+            {
+                /* uniprediction weighting, always outputs to wv0 */
+                WeightParam* pwp = (refIdx0 >= 0) ? pwp0 : pwp1;
+                for (int plane = 0; plane < 3; plane++)
+                {
+                    wv0[plane].w = pwp[plane].inputWeight;
+                    wv0[plane].offset = pwp[plane].inputOffset * (1 << (X265_DEPTH - 8));
+                    wv0[plane].shift = pwp[plane].log2WeightDenom;
+                    wv0[plane].round = pwp[plane].log2WeightDenom >= 1 ? 1 << (pwp[plane].log2WeightDenom - 1) : 0;
+                }
+            }
+        }
+        else
+            pwp0 = pwp1 = NULL;
+
+        if (refIdx0 >= 0 && refIdx1 >= 0)
+        {
+            /* Biprediction */
+            X265_CHECK(refIdx0 < m_slice->m_numRefIdx[0], "bidir refidx0 out of range\n");
+            X265_CHECK(refIdx1 < m_slice->m_numRefIdx[1], "bidir refidx1 out of range\n");
+
+            if (bLuma)
+            {
+                predInterLumaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), &m_predShortYuv[0], &m_clippedMv[0]);
+                predInterLumaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), &m_predShortYuv[1], &m_clippedMv[1]);
+            }
+            if (bChroma)
+            {
+                predInterChromaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), &m_predShortYuv[0], &m_clippedMv[0]);
+                predInterChromaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), &m_predShortYuv[1], &m_clippedMv[1]);
+            }
+
+            if (pwp0 && pwp1 && (pwp0->bPresentFlag || pwp1->bPresentFlag))
+                addWeightBi(&m_predShortYuv[0], &m_predShortYuv[1], wv0, wv1, predYuv, bLuma, bChroma);
+            else
+                predYuv->addAvg(&m_predShortYuv[0], &m_predShortYuv[1], m_partAddr, m_width, m_height, bLuma, bChroma);
+        }
+        else if (refIdx0 >= 0)
+        {
+            /* uniprediction to L0 */
+            X265_CHECK(refIdx0 < m_slice->m_numRefIdx[0], "unidir refidx0 out of range\n");
+
+            if (pwp0 && pwp0->bPresentFlag)
+            {
+                ShortYuv* shortYuv = &m_predShortYuv[0];
+
+                if (bLuma)
+                    predInterLumaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), shortYuv, &m_clippedMv[0]);
+                if (bChroma)
+                    predInterChromaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), shortYuv, &m_clippedMv[0]);
+
+                addWeightUni(shortYuv, wv0, predYuv, bLuma, bChroma);
+            }
+            else
+            {
+                if (bLuma)
+                    predInterLumaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), predYuv, &m_clippedMv[0]);
+                if (bChroma)
+                    predInterChromaBlk(m_slice->m_refPicList[0][refIdx0]->getPicYuvRec(), predYuv, &m_clippedMv[0]);
+            }
+        }
+        else
+        {
+            /* uniprediction to L1 */
+            X265_CHECK(refIdx1 >= 0, "refidx1 was not positive\n");
+            X265_CHECK(refIdx1 < m_slice->m_numRefIdx[1], "unidir refidx1 out of range\n");
+
+            if (pwp1 && pwp1->bPresentFlag)
+            {
+                ShortYuv* shortYuv = &m_predShortYuv[0];
+
+                if (bLuma)
+                    predInterLumaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), shortYuv, &m_clippedMv[1]);
+                if (bChroma)
+                    predInterChromaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), shortYuv, &m_clippedMv[1]);
+
+                addWeightUni(shortYuv, wv0, predYuv, bLuma, bChroma);
+            }
+            else
+            {
+                if (bLuma)
+                    predInterLumaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), predYuv, &m_clippedMv[1]);
+                if (bChroma)
+                    predInterChromaBlk(m_slice->m_refPicList[1][refIdx1]->getPicYuvRec(), predYuv, &m_clippedMv[1]);
+            }
+        }
+    }
+}
+
+void Predict::predInterLumaBlk(TComPicYuv *refPic, TComYuv *dstPic, MV *mv)
+{
+    int dstStride = dstPic->getStride();
+    pixel *dst    = dstPic->getLumaAddr(m_partAddr);
+
+    int srcStride = refPic->getStride();
+    int srcOffset = (mv->x >> 2) + (mv->y >> 2) * srcStride;
+    int partEnum = partitionFromSizes(m_width, m_height);
+    pixel* src = refPic->getLumaAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + srcOffset;
+
+    int xFrac = mv->x & 0x3;
+    int yFrac = mv->y & 0x3;
+
+    if (!(yFrac | xFrac))
+        primitives.luma_copy_pp[partEnum](dst, dstStride, src, srcStride);
+    else if (!yFrac)
+        primitives.luma_hpp[partEnum](src, srcStride, dst, dstStride, xFrac);
+    else if (!xFrac)
+        primitives.luma_vpp[partEnum](src, srcStride, dst, dstStride, yFrac);
+    else
+    {
+        int tmpStride = m_width;
+        int filterSize = NTAPS_LUMA;
+        int halfFilterSize = (filterSize >> 1);
+        primitives.luma_hps[partEnum](src, srcStride, m_immedVals, tmpStride, xFrac, 1);
+        primitives.luma_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * tmpStride, tmpStride, dst, dstStride, yFrac);
+    }
+}
+
+void Predict::predInterLumaBlk(TComPicYuv *refPic, ShortYuv *dstPic, MV *mv)
+{
+    int refStride = refPic->getStride();
+    int refOffset = (mv->x >> 2) + (mv->y >> 2) * refStride;
+    pixel *ref    = refPic->getLumaAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + refOffset;
+
+    int dstStride = dstPic->m_width;
+    int16_t *dst  = dstPic->getLumaAddr(m_partAddr);
+
+    int xFrac = mv->x & 0x3;
+    int yFrac = mv->y & 0x3;
+
+    int partEnum = partitionFromSizes(m_width, m_height);
+
+    X265_CHECK((m_width % 4) + (m_height % 4) == 0, "width or height not divisible by 4\n");
+    X265_CHECK(dstStride == MAX_CU_SIZE, "stride expected to be max cu size\n");
+
+    if (!(yFrac | xFrac))
+        primitives.luma_p2s(ref, refStride, dst, m_width, m_height);
+    else if (!yFrac)
+        primitives.luma_hps[partEnum](ref, refStride, dst, dstStride, xFrac, 0);
+    else if (!xFrac)
+        primitives.luma_vps[partEnum](ref, refStride, dst, dstStride, yFrac);
+    else
+    {
+        int tmpStride = m_width;
+        int filterSize = NTAPS_LUMA;
+        int halfFilterSize = (filterSize >> 1);
+        primitives.luma_hps[partEnum](ref, refStride, m_immedVals, tmpStride, xFrac, 1);
+        primitives.luma_vss[partEnum](m_immedVals + (halfFilterSize - 1) * tmpStride, tmpStride, dst, dstStride, yFrac);
+    }
+}
+
+void Predict::predInterChromaBlk(TComPicYuv *refPic, TComYuv *dstPic, MV *mv)
+{
+    int refStride = refPic->getCStride();
+    int dstStride = dstPic->getCStride();
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    int shiftHor = (2 + hChromaShift);
+    int shiftVer = (2 + vChromaShift);
+
+    int refOffset = (mv->x >> shiftHor) + (mv->y >> shiftVer) * refStride;
+
+    pixel* refCb = refPic->getCbAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + refOffset;
+    pixel* refCr = refPic->getCrAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + refOffset;
+
+    pixel* dstCb = dstPic->getCbAddr(m_partAddr);
+    pixel* dstCr = dstPic->getCrAddr(m_partAddr);
+
+    int xFrac = mv->x & ((1 << shiftHor) - 1);
+    int yFrac = mv->y & ((1 << shiftVer) - 1);
+
+    int partEnum = partitionFromSizes(m_width, m_height);
+    
+    if (!(yFrac | xFrac))
+    {
+        primitives.chroma[m_csp].copy_pp[partEnum](dstCb, dstStride, refCb, refStride);
+        primitives.chroma[m_csp].copy_pp[partEnum](dstCr, dstStride, refCr, refStride);
+    }
+    else if (!yFrac)
+    {
+        primitives.chroma[m_csp].filter_hpp[partEnum](refCb, refStride, dstCb, dstStride, xFrac << (1 - hChromaShift));
+        primitives.chroma[m_csp].filter_hpp[partEnum](refCr, refStride, dstCr, dstStride, xFrac << (1 - hChromaShift));
+    }
+    else if (!xFrac)
+    {
+        primitives.chroma[m_csp].filter_vpp[partEnum](refCb, refStride, dstCb, dstStride, yFrac << (1 - vChromaShift));
+        primitives.chroma[m_csp].filter_vpp[partEnum](refCr, refStride, dstCr, dstStride, yFrac << (1 - vChromaShift));
+    }
+    else
+    {
+        int extStride = m_width >> hChromaShift;
+        int filterSize = NTAPS_CHROMA;
+        int halfFilterSize = (filterSize >> 1);
+
+        primitives.chroma[m_csp].filter_hps[partEnum](refCb, refStride, m_immedVals, extStride, xFrac << (1 - hChromaShift), 1);
+        primitives.chroma[m_csp].filter_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCb, dstStride, yFrac << (1 - vChromaShift));
+
+        primitives.chroma[m_csp].filter_hps[partEnum](refCr, refStride, m_immedVals, extStride, xFrac << (1 - hChromaShift), 1);
+        primitives.chroma[m_csp].filter_vsp[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCr, dstStride, yFrac << (1 - vChromaShift));
+    }
+}
+
+void Predict::predInterChromaBlk(TComPicYuv *refPic, ShortYuv *dstPic, MV *mv)
+{
+    int refStride = refPic->getCStride();
+    int dstStride = dstPic->m_cwidth;
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    int shiftHor = (2 + hChromaShift);
+    int shiftVer = (2 + vChromaShift);
+
+    int refOffset = (mv->x >> shiftHor) + (mv->y >> shiftVer) * refStride;
+
+    pixel* refCb = refPic->getCbAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + refOffset;
+    pixel* refCr = refPic->getCrAddr(m_cuAddr, m_zOrderIdxinCU + m_partAddr) + refOffset;
+
+    int16_t* dstCb = dstPic->getCbAddr(m_partAddr);
+    int16_t* dstCr = dstPic->getCrAddr(m_partAddr);
+
+    int xFrac = mv->x & ((1 << shiftHor) - 1);
+    int yFrac = mv->y & ((1 << shiftVer) - 1);
+
+    int partEnum = partitionFromSizes(m_width, m_height);
+    
+    uint32_t cxWidth  = m_width >> hChromaShift;
+    uint32_t cxHeight = m_height >> vChromaShift;
+
+    X265_CHECK(((cxWidth | cxHeight) % 2) == 0, "chroma block size expected to be multiple of 2\n");
+
+    if (!(yFrac | xFrac))
+    {
+        primitives.chroma_p2s[m_csp](refCb, refStride, dstCb, cxWidth, cxHeight);
+        primitives.chroma_p2s[m_csp](refCr, refStride, dstCr, cxWidth, cxHeight);
+    }
+    else if (!yFrac)
+    {
+        primitives.chroma[m_csp].filter_hps[partEnum](refCb, refStride, dstCb, dstStride, xFrac << (1 - hChromaShift), 0);
+        primitives.chroma[m_csp].filter_hps[partEnum](refCr, refStride, dstCr, dstStride, xFrac << (1 - hChromaShift), 0);
+    }
+    else if (!xFrac)
+    {
+        primitives.chroma[m_csp].filter_vps[partEnum](refCb, refStride, dstCb, dstStride, yFrac << (1 - vChromaShift));
+        primitives.chroma[m_csp].filter_vps[partEnum](refCr, refStride, dstCr, dstStride, yFrac << (1 - vChromaShift));
+    }
+    else
+    {
+        int extStride = cxWidth;
+        int filterSize = NTAPS_CHROMA;
+        int halfFilterSize = (filterSize >> 1);
+        primitives.chroma[m_csp].filter_hps[partEnum](refCb, refStride, m_immedVals, extStride, xFrac << (1 - hChromaShift), 1);
+        primitives.chroma[m_csp].filter_vss[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCb, dstStride, yFrac << (1 - vChromaShift));
+        primitives.chroma[m_csp].filter_hps[partEnum](refCr, refStride, m_immedVals, extStride, xFrac << (1 - hChromaShift), 1);
+        primitives.chroma[m_csp].filter_vss[partEnum](m_immedVals + (halfFilterSize - 1) * extStride, extStride, dstCr, dstStride, yFrac << (1 - vChromaShift));
+    }
+}
+
+/* weighted averaging for bi-pred */
+void Predict::addWeightBi(ShortYuv* srcYuv0, ShortYuv* srcYuv1, const WeightValues wp0[3], const WeightValues wp1[3], TComYuv* predYuv, bool bLuma, bool bChroma)
+{
+    int x, y;
+
+    int w0, w1, offset, shiftNum, shift, round;
+    uint32_t src0Stride, src1Stride, dststride;
+
+    int16_t* srcY0 = srcYuv0->getLumaAddr(m_partAddr);
+    int16_t* srcU0 = srcYuv0->getCbAddr(m_partAddr);
+    int16_t* srcV0 = srcYuv0->getCrAddr(m_partAddr);
+
+    int16_t* srcY1 = srcYuv1->getLumaAddr(m_partAddr);
+    int16_t* srcU1 = srcYuv1->getCbAddr(m_partAddr);
+    int16_t* srcV1 = srcYuv1->getCrAddr(m_partAddr);
+
+    pixel* dstY    = predYuv->getLumaAddr(m_partAddr);
+    pixel* dstU    = predYuv->getCbAddr(m_partAddr);
+    pixel* dstV    = predYuv->getCrAddr(m_partAddr);
+
+    if (bLuma)
+    {
+        // Luma
+        w0      = wp0[0].w;
+        offset  = wp0[0].o + wp1[0].o;
+        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
+        shift   = wp0[0].shift + shiftNum + 1;
+        round   = shift ? (1 << (shift - 1)) : 0;
+        w1      = wp1[0].w;
+
+        src0Stride = srcYuv0->m_width;
+        src1Stride = srcYuv1->m_width;
+        dststride  = predYuv->getStride();
+
+        // TODO: can we use weight_sp here?
+        for (y = m_height - 1; y >= 0; y--)
+        {
+            for (x = m_width - 1; x >= 0; )
+            {
+                // note: luma min width is 4
+                dstY[x] = weightBidir(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
+                x--;
+                dstY[x] = weightBidir(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
+                x--;
+                dstY[x] = weightBidir(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
+                x--;
+                dstY[x] = weightBidir(w0, srcY0[x], w1, srcY1[x], round, shift, offset);
+                x--;
+            }
+
+            srcY0 += src0Stride;
+            srcY1 += src1Stride;
+            dstY  += dststride;
+        }
+    }
+
+    if (bChroma)
+    {
+        // Chroma U
+        w0      = wp0[1].w;
+        offset  = wp0[1].o + wp1[1].o;
+        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
+        shift   = wp0[1].shift + shiftNum + 1;
+        round   = shift ? (1 << (shift - 1)) : 0;
+        w1      = wp1[1].w;
+
+        src0Stride = srcYuv0->m_cwidth;
+        src1Stride = srcYuv1->m_cwidth;
+        dststride  = predYuv->getCStride();
+
+        m_width  >>= srcYuv0->getHorzChromaShift();
+        m_height >>= srcYuv0->getVertChromaShift();
+
+        // TODO: can we use weight_sp here?
+        for (y = m_height - 1; y >= 0; y--)
+        {
+            for (x = m_width - 1; x >= 0; )
+            {
+                // note: chroma min width is 2
+                dstU[x] = weightBidir(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
+                x--;
+                dstU[x] = weightBidir(w0, srcU0[x], w1, srcU1[x], round, shift, offset);
+                x--;
+            }
+
+            srcU0 += src0Stride;
+            srcU1 += src1Stride;
+            dstU  += dststride;
+        }
+
+        // Chroma V
+        w0     = wp0[2].w;
+        offset = wp0[2].o + wp1[2].o;
+        shift  = wp0[2].shift + shiftNum + 1;
+        round  = shift ? (1 << (shift - 1)) : 0;
+        w1     = wp1[2].w;
+
+        for (y = m_height - 1; y >= 0; y--)
+        {
+            for (x = m_width - 1; x >= 0; )
+            {
+                // note: chroma min width is 2
+                dstV[x] = weightBidir(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
+                x--;
+                dstV[x] = weightBidir(w0, srcV0[x], w1, srcV1[x], round, shift, offset);
+                x--;
+            }
+
+            srcV0 += src0Stride;
+            srcV1 += src1Stride;
+            dstV  += dststride;
+        }
+    }
+}
+
+/* weighted averaging for uni-pred */
+void Predict::addWeightUni(ShortYuv* srcYuv0, const WeightValues wp[3], TComYuv* predYuv, bool bLuma, bool bChroma)
+{
+    int16_t* srcY0 = srcYuv0->getLumaAddr(m_partAddr);
+    int16_t* srcU0 = srcYuv0->getCbAddr(m_partAddr);
+    int16_t* srcV0 = srcYuv0->getCrAddr(m_partAddr);
+
+    pixel* dstY = predYuv->getLumaAddr(m_partAddr);
+    pixel* dstU = predYuv->getCbAddr(m_partAddr);
+    pixel* dstV = predYuv->getCrAddr(m_partAddr);
+
+    int w0, offset, shiftNum, shift, round;
+    uint32_t srcStride, dstStride;
+
+    if (bLuma)
+    {
+        // Luma
+        w0      = wp[0].w;
+        offset  = wp[0].offset;
+        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
+        shift   = wp[0].shift + shiftNum;
+        round   = shift ? (1 << (shift - 1)) : 0;
+        srcStride = srcYuv0->m_width;
+        dstStride  = predYuv->getStride();
+
+        primitives.weight_sp(srcY0, dstY, srcStride, dstStride, m_width, m_height, w0, round, shift, offset);
+    }
+
+    if (bChroma)
+    {
+        // Chroma U
+        w0      = wp[1].w;
+        offset  = wp[1].offset;
+        shiftNum = IF_INTERNAL_PREC - X265_DEPTH;
+        shift   = wp[1].shift + shiftNum;
+        round   = shift ? (1 << (shift - 1)) : 0;
+
+        srcStride = srcYuv0->m_cwidth;
+        dstStride = predYuv->getCStride();
+
+        m_width  >>= srcYuv0->getHorzChromaShift();
+        m_height >>= srcYuv0->getVertChromaShift();
+
+        primitives.weight_sp(srcU0, dstU, srcStride, dstStride, m_width, m_height, w0, round, shift, offset);
+
+        // Chroma V
+        w0     = wp[2].w;
+        offset = wp[2].offset;
+        shift  = wp[2].shift + shiftNum;
+        round  = shift ? (1 << (shift - 1)) : 0;
+
+        primitives.weight_sp(srcV0, dstV, srcStride, dstStride, m_width, m_height, w0, round, shift, offset);
+    }
+}

x265-hg20140928.tar.bz2/source/encoder/predict.h Added

@@ -0,0 +1,101 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Deepthi Nandakumar <deepthi@multicorewareinc.com>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_PREDICT_H
+#define X265_PREDICT_H
+
+#include "common.h"
+#include "frame.h"
+#include "quant.h"
+#include "shortyuv.h"
+#include "TLibCommon/TComYuv.h"
+#include "TLibCommon/TComMotionInfo.h"
+#include "TLibCommon/TComPattern.h"
+
+namespace x265 {
+
+class Predict
+{
+protected:
+
+    ShortYuv  m_predShortYuv[2]; /* temporary storage for weighted prediction */
+    int16_t*  m_immedVals;
+
+    /* Slice information */
+    Slice*    m_slice;
+    int       m_csp;
+
+    /* CU information for prediction */
+    int       m_width;
+    int       m_height; 
+    uint32_t  m_partAddr;
+    uint32_t  m_cuAddr;
+    uint32_t  m_zOrderIdxinCU;
+
+    /* Motion information */
+    TComCUMvField* m_mvField[2];
+
+    /* TODO: Need to investigate clipping while writing into the TComDataCU fields itself */
+    MV             m_clippedMv[2];
+
+    // motion compensation functions
+    void predInterLumaBlk(TComPicYuv *refPic, TComYuv *dstPic, MV *mv);
+    void predInterLumaBlk(TComPicYuv *refPic, ShortYuv *dstPic, MV *mv);
+
+    void predInterChromaBlk(TComPicYuv *refPic, TComYuv *dstPic, MV *mv);
+    void predInterChromaBlk(TComPicYuv *refPic, ShortYuv *dstPic, MV *mv);
+
+    // Weighted prediction scaling values built from slice parameters (bitdepth scaled)
+    struct WeightValues
+    {
+        int w, o, offset, shift, round;
+    };
+
+    void addWeightBi(ShortYuv* srcYuv0, ShortYuv* srcYuv1, const WeightValues wp0[3], const WeightValues wp1[3], TComYuv* predYuv, bool bLuma, bool bChroma);
+    void addWeightUni(ShortYuv* srcYuv0, const WeightValues wp[3], TComYuv* predYuv, bool bLuma, bool bChroma);
+
+public:
+
+    /* Intra prediction buffers */
+    pixel*    m_predBuf;
+    pixel*    m_refAbove;
+    pixel*    m_refAboveFlt;
+    pixel*    m_refLeft;
+    pixel*    m_refLeftFlt;
+
+    Predict();
+    ~Predict();
+
+    bool allocBuffers(int csp);
+
+    /* prepMotionCompensation needs to be called to prepare MC with CU-relevant data */
+    void prepMotionCompensation(TComDataCU* cu, int partIdx);
+    void motionCompensation(TComYuv* predYuv, bool bLuma, bool bChroma);
+
+    /* Angular Intra */
+    void predIntraLumaAng(uint32_t dirMode, pixel* pred, intptr_t stride, uint32_t log2TrSize);
+    void predIntraChromaAng(pixel* src, uint32_t dirMode, pixel* pred, intptr_t stride, uint32_t log2TrSizeC, int chFmt);
+};
+}
+
+#endif // ifndef X265_PREDICT_H

x265-hg20140324.tar.bz2/source/encoder/ratecontrol.cpp -> x265-hg20140928.tar.bz2/source/encoder/ratecontrol.cpp Changed

@@ -20,19 +20,131 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComPic.h"
+#include "common.h"
+#include "param.h"
+#include "frame.h"
+
 #include "encoder.h"
 #include "slicetype.h"
 #include "ratecontrol.h"
+#include "sei.h"
+
+#define BR_SHIFT  6
+#define CPB_SHIFT 4
 
 using namespace x265;
 
+/* Amortize the partial cost of I frames over the next N frames */
+const double RateControl::s_amortizeFraction = 0.85;
+const int RateControl::s_amortizeFrames = 75;
+const char *RateControl::s_defaultStatFileName = "x265_2pass.log";
+
+namespace {
+#define CMP_OPT_FIRST_PASS(opt, param_val)\
+{\
+    bErr = 0;\
+    p = strstr(opts, opt "=");\
+    char* q = strstr(opts, "no-"opt);\
+    if (p && sscanf(p, opt "=%d" , &i) && param_val != i)\
+        bErr = 1;\
+    else if (!param_val && !q && !p)\
+        bErr = 1;\
+    else if (param_val && (q || !strstr(opts, opt)))\
+        bErr = 1;\
+    if (bErr)\
+    {\
+        x265_log(m_param, X265_LOG_ERROR, "different " opt " setting than first pass (%d vs %d)\n", param_val, i);\
+        return false;\
+    }\
+}
+
+inline int calcScale(uint32_t x)
+{
+    static uint8_t lut[16] = {4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
+    int y, z = (((x & 0xffff) - 1) >> 27) & 16;
+    x >>= z;
+    z += y = (((x & 0xff) - 1) >> 28) & 8;
+    x >>= y;
+    z += y = (((x & 0xf) - 1) >> 29) & 4;
+    x >>= y;
+    return z + lut[x&0xf];
+}
+
+inline int calcLength(uint32_t x)
+{
+    static uint8_t lut[16] = {4, 3, 2, 2, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0};
+    int y, z = (((x >> 16) - 1) >> 27) & 16;
+    x >>= z ^ 16;
+    z += y = ((x - 0x100) >> 28) & 8;
+    x >>= y ^ 8;
+    z += y = ((x - 0x10) >> 29) & 4;
+    x >>= y ^ 4;
+    return z + lut[x];
+}
+
+inline void reduceFraction(int* n, int* d)
+{
+    int a = *n;
+    int b = *d;
+    int c;
+    if (!a || !b)
+        return;
+    c = a % b;
+    while (c)
+    {
+        a = b;
+        b = c;
+        c = a % b;
+    }
+    *n /= b;
+    *d /= b;
+}
+
+inline char *strcatFilename(const char *input, const char *suffix)
+{
+    char *output = X265_MALLOC(char, strlen(input) + strlen(suffix) + 1);
+    if (!output)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "unable to allocate memory for filename\n");
+        return NULL;
+    }
+    strcpy(output, input);
+    strcat(output, suffix);
+    return output;
+}
+
+inline double qScale2bits(RateControlEntry *rce, double qScale)
+{
+    if (qScale < 0.1)
+        qScale = 0.1;
+    return (rce->coeffBits + .1) * pow(rce->qScale / qScale, 1.1)
+           + rce->mvBits * pow(X265_MAX(rce->qScale, 1) / X265_MAX(qScale, 1), 0.5)
+           + rce->miscBits;
+}
 
+inline void copyRceData(RateControlEntry* rce, RateControlEntry* rce2Pass)
+{
+    rce->coeffBits = rce2Pass->coeffBits;
+    rce->mvBits = rce2Pass->mvBits;
+    rce->miscBits = rce2Pass->miscBits;
+    rce->iCuCount = rce2Pass->iCuCount;
+    rce->pCuCount = rce2Pass->pCuCount;
+    rce->skipCuCount = rce2Pass->skipCuCount;
+    rce->keptAsRef = rce2Pass->keptAsRef;
+    rce->qScale = rce2Pass->qScale;
+    rce->newQScale = rce2Pass->newQScale;
+    rce->expectedBits = rce2Pass->expectedBits;
+    rce->expectedVbv = rce2Pass->expectedVbv;
+    rce->blurredComplexity = rce2Pass->blurredComplexity;
+    rce->sliceType = rce2Pass->sliceType;
+}
+
+}  // end anonymous namespace
 /* Compute variance to derive AC energy of each block */
-static inline uint32_t acEnergyVar(TComPic *pic, uint64_t sum_ssd, int shift, int i)
+static inline uint32_t acEnergyVar(Frame *pic, uint64_t sum_ssd, int shift, int i)
 {
     uint32_t sum = (uint32_t)sum_ssd;
     uint32_t ssd = (uint32_t)(sum_ssd >> 32);
@@ -43,10 +155,9 @@
 }
 
 /* Find the energy of each block in Y/Cb/Cr plane */
-static inline uint32_t acEnergyPlane(TComPic *pic, pixel* src, int srcStride, int bChroma, int colorFormat)
+static inline uint32_t acEnergyPlane(Frame *pic, pixel* src, int srcStride, int bChroma, int colorFormat)
 {
-    /* Support only 420 and 444 color spaces */
-    if (colorFormat == X265_CSP_I420 && bChroma)
+    if ((colorFormat != X265_CSP_I444) && bChroma)
     {
         ALIGN_VAR_8(pixel, pix[8 * 8]);
         primitives.luma_copy_pp[LUMA_8x8](pix, 8, src, srcStride);
@@ -57,12 +168,12 @@
 }
 
 /* Find the total AC energy of each block in all planes */
-uint32_t RateControl::acEnergyCu(TComPic* pic, uint32_t block_x, uint32_t block_y)
+uint32_t RateControl::acEnergyCu(Frame* pic, uint32_t block_x, uint32_t block_y)
 {
     int stride = pic->getPicYuvOrg()->getStride();
     int cStride = pic->getPicYuvOrg()->getCStride();
     uint32_t blockOffsetLuma = block_x + (block_y * stride);
-    int colorFormat = param->internalCsp;
+    int colorFormat = m_param->internalCsp;
     int hShift = CHROMA_H_SHIFT(colorFormat);
     int vShift = CHROMA_V_SHIFT(colorFormat);
     uint32_t blockOffsetChroma = (block_x >> hShift) + ((block_y >> vShift) * cStride);
@@ -76,7 +187,7 @@
     return var;
 }
 
-void RateControl::calcAdaptiveQuantFrame(TComPic *pic)
+void RateControl::calcAdaptiveQuantFrame(Frame *pic)
 {
     /* Actual adaptive quantization */
     int maxCol = pic->getPicYuvOrg()->getWidth();
@@ -92,40 +203,34 @@
     int block_xy = 0;
     int block_x = 0, block_y = 0;
     double strength = 0.f;
-    if (param->rc.aqMode == X265_AQ_NONE || param->rc.aqStrength == 0)
+    if (m_param->rc.aqMode == X265_AQ_NONE || m_param->rc.aqStrength == 0)
     {
         /* Need to init it anyways for CU tree */
         int cuWidth = ((maxCol / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
         int cuHeight = ((maxRow / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
         int cuCount = cuWidth * cuHeight;
 
-        if (param->rc.aqMode && param->rc.aqStrength == 0)
+        if (m_param->rc.aqMode && m_param->rc.aqStrength == 0)
         {
-            memset(pic->m_lowres.qpOffset, 0, cuCount * sizeof(double));
+            memset(pic->m_lowres.qpCuTreeOffset, 0, cuCount * sizeof(double));
             memset(pic->m_lowres.qpAqOffset, 0, cuCount * sizeof(double));
             for (int cuxy = 0; cuxy < cuCount; cuxy++)
-            {
                 pic->m_lowres.invQscaleFactor[cuxy] = 256;
-            }
         }
 
         /* Need variance data for weighted prediction */
-        if (param->bEnableWeightedPred)
+        if (m_param->bEnableWeightedPred || m_param->bEnableWeightedBiPred)
         {
             for (block_y = 0; block_y < maxRow; block_y += 16)
-            {
                 for (block_x = 0; block_x < maxCol; block_x += 16)
-                {
                     acEnergyCu(pic, block_x, block_y);
-                }
-            }
         }
     }
     else
     {
         block_xy = 0;
         double avg_adj_pow2 = 0, avg_adj = 0, qp_adj = 0;
-        if (param->rc.aqMode == X265_AQ_AUTO_VARIANCE)
+        if (m_param->rc.aqMode == X265_AQ_AUTO_VARIANCE)
         {
             double bit_depth_correction = pow(1 << (X265_DEPTH - 8), 0.5);
             for (block_y = 0; block_y < maxRow; block_y += 16)
@@ -133,30 +238,30 @@
                 for (block_x = 0; block_x < maxCol; block_x += 16)
                 {
                     uint32_t energy = acEnergyCu(pic, block_x, block_y);
-                    qp_adj = pow(energy + 1, 0.125);
-                    pic->m_lowres.qpOffset[block_xy] = qp_adj;
+                    qp_adj = pow(energy + 1, 0.1);
+                    pic->m_lowres.qpCuTreeOffset[block_xy] = qp_adj;
                     avg_adj += qp_adj;
                     avg_adj_pow2 += qp_adj * qp_adj;
                     block_xy++;
                 }
             }
 
-            avg_adj /= ncu;
-            avg_adj_pow2 /= ncu;
-            strength = param->rc.aqStrength * avg_adj / bit_depth_correction;
-            avg_adj = avg_adj - 0.5f * (avg_adj_pow2 - (14.f * bit_depth_correction)) / avg_adj;
+            avg_adj /= m_ncu;
+            avg_adj_pow2 /= m_ncu;
+            strength = m_param->rc.aqStrength * avg_adj / bit_depth_correction;
+            avg_adj = avg_adj - 0.5f * (avg_adj_pow2 - (11.f * bit_depth_correction)) / avg_adj;
         }
         else
-            strength = param->rc.aqStrength * 1.0397f;
+            strength = m_param->rc.aqStrength * 1.0397f;
 
         block_xy = 0;
         for (block_y = 0; block_y < maxRow; block_y += 16)
         {
             for (block_x = 0; block_x < maxCol; block_x += 16)
             {
-                if (param->rc.aqMode == X265_AQ_AUTO_VARIANCE)
+                if (m_param->rc.aqMode == X265_AQ_AUTO_VARIANCE)
                 {
-                    qp_adj = pic->m_lowres.qpOffset[block_xy];
+                    qp_adj = pic->m_lowres.qpCuTreeOffset[block_xy];
                     qp_adj = strength * (qp_adj - avg_adj);
                 }
                 else
@@ -165,17 +270,17 @@
                     qp_adj = strength * (X265_LOG2(X265_MAX(energy, 1)) - (14.427f + 2 * (X265_DEPTH - 8)));
                 }
                 pic->m_lowres.qpAqOffset[block_xy] = qp_adj;
-                pic->m_lowres.qpOffset[block_xy] = qp_adj;
+                pic->m_lowres.qpCuTreeOffset[block_xy] = qp_adj;
                 pic->m_lowres.invQscaleFactor[block_xy] = x265_exp2fix8(qp_adj);
                 block_xy++;
             }
         }
     }
 
-    if (param->bEnableWeightedPred)
+    if (m_param->bEnableWeightedPred || m_param->bEnableWeightedBiPred)
     {
-        int hShift = CHROMA_H_SHIFT(param->internalCsp);
-        int vShift = CHROMA_V_SHIFT(param->internalCsp);
+        int hShift = CHROMA_H_SHIFT(m_param->internalCsp);
+        int vShift = CHROMA_V_SHIFT(m_param->internalCsp);
         maxCol = ((maxCol + 8) >> 4) << 4;
         maxRow = ((maxRow + 8) >> 4) << 4;
         int width[3]  = { maxCol, maxCol >> hShift, maxCol >> hShift };
@@ -191,293 +296,1062 @@
     }
 }
 
-RateControl::RateControl(Encoder * _cfg)
+RateControl::RateControl(x265_param *p)
 {
-    param = _cfg->param;
-    int lowresCuWidth = ((param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
-    int lowresCuHeight = ((param->sourceHeight / 2)  + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
-    ncu = lowresCuWidth * lowresCuHeight;
-    if (param->rc.cuTree)
-    {
-        qCompress = 1;
-        param->rc.pbFactor = 1;
-    }
+    m_param = p;
+    int lowresCuWidth = ((m_param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    int lowresCuHeight = ((m_param->sourceHeight / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    m_ncu = lowresCuWidth * lowresCuHeight;
+
+    if (m_param->rc.cuTree)
+        m_qCompress = 1;
     else
-        qCompress = param->rc.qCompress;
+        m_qCompress = m_param->rc.qCompress;
 
     // validate for param->rc, maybe it is need to add a function like x265_parameters_valiate()
-    param->rc.rfConstant = Clip3((double)-QP_BD_OFFSET, (double)51, param->rc.rfConstant);
-    param->rc.rfConstantMax = Clip3((double)-QP_BD_OFFSET, (double)51, param->rc.rfConstantMax);
-    rateFactorMaxIncrement = 0;
-    vbvMinRate = 0;
-
-    if (param->rc.rateControlMode == X265_RC_CRF)
+    m_residualFrames = 0;
+    m_partialResidualFrames = 0;
+    m_residualCost = 0;
+    m_partialResidualCost = 0;
+    m_rateFactorMaxIncrement = 0;
+    m_rateFactorMaxDecrement = 0;
+    m_fps = m_param->fpsNum / m_param->fpsDenom;
+    m_startEndOrder.set(0);
+    m_bTerminated = false;
+    m_finalFrameCount = 0;
+    m_numEntries = 0;
+    if (m_param->rc.rateControlMode == X265_RC_CRF)
     {
-        param->rc.qp = (int)param->rc.rfConstant + QP_BD_OFFSET;
-        param->rc.bitrate = 0;
-
-        double baseCplx = ncu * (param->bframes ? 120 : 80);
-        double mbtree_offset = param->rc.cuTree ? (1.0 - param->rc.qCompress) * 13.5 : 0;
-        rateFactorConstant = pow(baseCplx, 1 - qCompress) /
-            x265_qp2qScale(param->rc.rfConstant + mbtree_offset);
-        if (param->rc.rfConstantMax)
+        m_param->rc.qp = (int)m_param->rc.rfConstant;
+        m_param->rc.bitrate = 0;
+
+        double baseCplx = m_ncu * (m_param->bframes ? 120 : 80);
+        double mbtree_offset = m_param->rc.cuTree ? (1.0 - m_param->rc.qCompress) * 13.5 : 0;
+        m_rateFactorConstant = pow(baseCplx, 1 - m_qCompress) /
+            x265_qp2qScale(m_param->rc.rfConstant + mbtree_offset);
+        if (m_param->rc.rfConstantMax)
         {
-            rateFactorMaxIncrement = param->rc.rfConstantMax - param->rc.rfConstant;
-            if (rateFactorMaxIncrement <= 0)
+            m_rateFactorMaxIncrement = m_param->rc.rfConstantMax - m_param->rc.rfConstant;
+            if (m_rateFactorMaxIncrement <= 0)
             {
-                x265_log(param, X265_LOG_WARNING, "CRF max must be greater than CRF\n");
-                rateFactorMaxIncrement = 0;
+                x265_log(m_param, X265_LOG_WARNING, "CRF max must be greater than CRF\n");
+                m_rateFactorMaxIncrement = 0;
             }
         }
+        if (m_param->rc.rfConstantMin)
+            m_rateFactorMaxDecrement = m_param->rc.rfConstant - m_param->rc.rfConstantMin;
     }
+    m_isAbr = m_param->rc.rateControlMode != X265_RC_CQP && !m_param->rc.bStatRead;
+    m_2pass = m_param->rc.rateControlMode == X265_RC_ABR && m_param->rc.bStatRead;
+    m_bitrate = m_param->rc.bitrate * 1000;
+    m_frameDuration = (double)m_param->fpsDenom / m_param->fpsNum;
+    m_qp = m_param->rc.qp;
+    m_lastRceq = 1; /* handles the cmplxrsum when the previous frame cost is zero */
+    m_shortTermCplxSum = 0;
+    m_shortTermCplxCount = 0;
+    m_lastNonBPictType = I_SLICE;
+    m_isAbrReset = false;
+    m_lastAbrResetPoc = -1;
+    m_statFileOut = NULL;
+    m_cutreeStatFileOut = m_cutreeStatFileIn = NULL;
+    m_rce2Pass = NULL;
 
-    isAbr = param->rc.rateControlMode != X265_RC_CQP; // later add 2pass option
-    bitrate = param->rc.bitrate * 1000;
-    frameDuration = (double)param->fpsDenom / param->fpsNum;
-    qp = param->rc.qp;
-    lastRceq = 1; /* handles the cmplxrsum when the previous frame cost is zero */
-    shortTermCplxSum = 0;
-    shortTermCplxCount = 0;
-    lastNonBPictType = I_SLICE;
-    isAbrReset = false;
-    lastAbrResetPoc = -1;
     // vbv initialization
-    param->rc.vbvBufferSize = Clip3(0, 2000000, param->rc.vbvBufferSize);
-    param->rc.vbvMaxBitrate = Clip3(0, 2000000, param->rc.vbvMaxBitrate);
-    param->rc.vbvBufferInit = Clip3(0.0, 2000000.0, param->rc.vbvBufferInit);
-    vbvMinRate = 0;
-    if (param->rc.vbvBufferSize)
+    m_param->rc.vbvBufferSize = Clip3(0, 2000000, m_param->rc.vbvBufferSize);
+    m_param->rc.vbvMaxBitrate = Clip3(0, 2000000, m_param->rc.vbvMaxBitrate);
+    m_param->rc.vbvBufferInit = Clip3(0.0, 2000000.0, m_param->rc.vbvBufferInit);
+    m_singleFrameVbv = 0;
+    if (m_param->rc.vbvBufferSize)
     {
-        if (param->rc.rateControlMode == X265_RC_CQP)
+        if (m_param->rc.rateControlMode == X265_RC_CQP)
         {
-            x265_log(param, X265_LOG_WARNING, "VBV is incompatible with constant QP, ignored.\n");
-            param->rc.vbvBufferSize = 0;
-            param->rc.vbvMaxBitrate = 0;
+            x265_log(m_param, X265_LOG_WARNING, "VBV is incompatible with constant QP, ignored.\n");
+            m_param->rc.vbvBufferSize = 0;
+            m_param->rc.vbvMaxBitrate = 0;
         }
-        else if (param->rc.vbvMaxBitrate == 0)
+        else if (m_param->rc.vbvMaxBitrate == 0)
         {
-            if (param->rc.rateControlMode == X265_RC_ABR)
+            if (m_param->rc.rateControlMode == X265_RC_ABR)
             {
-                x265_log(param, X265_LOG_WARNING, "VBV maxrate unspecified, assuming CBR\n");
-                param->rc.vbvMaxBitrate = param->rc.bitrate;
+                x265_log(m_param, X265_LOG_WARNING, "VBV maxrate unspecified, assuming CBR\n");
+                m_param->rc.vbvMaxBitrate = m_param->rc.bitrate;
             }
             else
             {
-                x265_log(param, X265_LOG_WARNING, "VBV bufsize set but maxrate unspecified, ignored\n");
-                param->rc.vbvBufferSize = 0;
+                x265_log(m_param, X265_LOG_WARNING, "VBV bufsize set but maxrate unspecified, ignored\n");
+                m_param->rc.vbvBufferSize = 0;
             }
         }
-        else if (param->rc.vbvMaxBitrate < param->rc.bitrate &&
-                 param->rc.rateControlMode == X265_RC_ABR)
+        else if (m_param->rc.vbvMaxBitrate < m_param->rc.bitrate &&
+                 m_param->rc.rateControlMode == X265_RC_ABR)
         {
-            x265_log(param, X265_LOG_WARNING, "max bitrate less than average bitrate, assuming CBR\n");
-            param->rc.bitrate = param->rc.vbvMaxBitrate;
+            x265_log(m_param, X265_LOG_WARNING, "max bitrate less than average bitrate, assuming CBR\n");
+            m_param->rc.bitrate = m_param->rc.vbvMaxBitrate;
         }
     }
-    else if (param->rc.vbvMaxBitrate)
+    else if (m_param->rc.vbvMaxBitrate)
+    {
+        x265_log(m_param, X265_LOG_WARNING, "VBV maxrate specified, but no bufsize, ignored\n");
+        m_param->rc.vbvMaxBitrate = 0;
+    }
+    m_isVbv = m_param->rc.vbvMaxBitrate > 0 && m_param->rc.vbvBufferSize > 0;
+    if (m_param->bEmitHRDSEI && !m_isVbv)
+    {
+        x265_log(m_param, X265_LOG_WARNING, "NAL HRD parameters require VBV parameters, ignored\n");
+        m_param->bEmitHRDSEI = 0;
+    }
+
+    m_isCbr = m_param->rc.rateControlMode == X265_RC_ABR && m_isVbv && !m_2pass && m_param->rc.vbvMaxBitrate <= m_param->rc.bitrate;
+    m_leadingBframes = m_param->bframes;
+    m_bframeBits = 0;
+    m_leadingNoBSatd = 0;
+    m_ipOffset = 6.0 * X265_LOG2(m_param->rc.ipFactor);
+    m_pbOffset = 6.0 * X265_LOG2(m_param->rc.pbFactor);
+
+    /* Adjust the first frame in order to stabilize the quality level compared to the rest */
+#define ABR_INIT_QP_MIN (24)
+#define ABR_INIT_QP_MAX (40)
+#define CRF_INIT_QP (int)m_param->rc.rfConstant
+    for (int i = 0; i < 3; i++)
+        m_lastQScaleFor[i] = x265_qp2qScale(m_param->rc.rateControlMode == X265_RC_CRF ? CRF_INIT_QP : ABR_INIT_QP_MIN);
+
+    if (m_param->rc.rateControlMode == X265_RC_CQP)
     {
-        x265_log(param, X265_LOG_WARNING, "VBV maxrate specified, but no bufsize, ignored\n");
-        param->rc.vbvMaxBitrate = 0;
+        if (m_qp && !m_param->bLossless)
+        {
+            m_qpConstant[P_SLICE] = m_qp;
+            m_qpConstant[I_SLICE] = Clip3(0, QP_MAX_MAX, (int)(m_qp - m_ipOffset + 0.5));
+            m_qpConstant[B_SLICE] = Clip3(0, QP_MAX_MAX, (int)(m_qp + m_pbOffset + 0.5));
+        }
+        else
+        {
+            m_qpConstant[P_SLICE] = m_qpConstant[I_SLICE] = m_qpConstant[B_SLICE] = m_qp;
+        }
     }
 
-    isVbv = param->rc.vbvMaxBitrate > 0 && param->rc.vbvBufferSize > 0;
-    double fps = (double)param->fpsNum / param->fpsDenom;
-    if (isVbv)
+    /* qstep - value set as encoder specific */
+    m_lstep = pow(2, m_param->rc.qpStep / 6.0);
+
+    for (int i = 0; i < 2; i++)
+        m_cuTreeStats.qpBuffer[i] = NULL;
+}
+
+bool RateControl::init(const SPS *sps)
+{
+    if (m_isVbv)
     {
         /* We don't support changing the ABR bitrate right now,
-           so if the stream starts as CBR, keep it CBR. */
-        if (param->rc.vbvBufferSize < (int)(param->rc.vbvMaxBitrate / fps))
+         * so if the stream starts as CBR, keep it CBR. */
+        if (m_param->rc.vbvBufferSize < (int)(m_param->rc.vbvMaxBitrate / m_fps))
+        {
+            m_param->rc.vbvBufferSize = (int)(m_param->rc.vbvMaxBitrate / m_fps);
+            x265_log(m_param, X265_LOG_WARNING, "VBV buffer size cannot be smaller than one frame, using %d kbit\n",
+                     m_param->rc.vbvBufferSize);
+        }
+        int vbvBufferSize = m_param->rc.vbvBufferSize * 1000;
+        int vbvMaxBitrate = m_param->rc.vbvMaxBitrate * 1000;
+
+        if (m_param->bEmitHRDSEI)
+        {
+            const HRDInfo* hrd = &sps->vuiParameters.hrdParameters;
+            vbvBufferSize = hrd->cpbSizeValue << (hrd->cpbSizeScale + CPB_SHIFT);
+            vbvMaxBitrate = hrd->bitRateValue << (hrd->bitRateScale + BR_SHIFT);
+        }
+        m_bufferRate = vbvMaxBitrate / m_fps;
+        m_vbvMaxRate = vbvMaxBitrate;
+        m_bufferSize = vbvBufferSize;
+        m_singleFrameVbv = m_bufferRate * 1.1 > m_bufferSize;
+
+        if (m_param->rc.vbvBufferInit > 1.)
+            m_param->rc.vbvBufferInit = Clip3(0.0, 1.0, m_param->rc.vbvBufferInit / m_param->rc.vbvBufferSize);
+        m_param->rc.vbvBufferInit = Clip3(0.0, 1.0, X265_MAX(m_param->rc.vbvBufferInit, m_bufferRate / m_bufferSize));
+        m_bufferFillFinal = m_bufferSize * m_param->rc.vbvBufferInit;
+    }
+
+    m_totalBits = 0;
+    m_framesDone = 0;
+    m_residualCost = 0;
+    m_partialResidualCost = 0;
+
+    /* 720p videos seem to be a good cutoff for cplxrSum */
+    double tuneCplxFactor = (m_param->rc.cuTree && m_ncu > 3600) ? 2.5 : 1;
+
+    /* estimated ratio that produces a reasonable QP for the first I-frame */
+    m_cplxrSum = .01 * pow(7.0e5, m_qCompress) * pow(m_ncu, 0.5) * tuneCplxFactor;
+    m_wantedBitsWindow = m_bitrate * m_frameDuration;
+    m_accumPNorm = .01;
+    m_accumPQp = (m_param->rc.rateControlMode == X265_RC_CRF ? CRF_INIT_QP : ABR_INIT_QP_MIN) * m_accumPNorm;
+
+    /* Frame Predictors and Row predictors used in vbv */
+    for (int i = 0; i < 5; i++)
+    {
+        m_pred[i].coeff = 2.0;
+        m_pred[i].count = 1.0;
+        m_pred[i].decay = 0.5;
+        m_pred[i].offset = 0.0;
+    }
+    m_predBfromP = m_pred[0];
+    if (!m_statFileOut && (m_param->rc.bStatWrite || m_param->rc.bStatRead))
+    {
+        /* If the user hasn't defined the stat filename, use the default value */
+        const char *fileName = m_param->rc.statFileName;
+        if (!fileName)
+            fileName = s_defaultStatFileName;
+        /* Load stat file and init 2pass algo */
+        if (m_param->rc.bStatRead)
+        {
+            m_expectedBitsSum = 0;
+            char *p, *statsIn, *statsBuf;
+            /* read 1st pass stats */
+            statsIn = statsBuf = x265_slurp_file(fileName);
+            if (!statsBuf)
+                return false;
+            if (m_param->rc.cuTree)
+            {
+                char *tmpFile = strcatFilename(fileName, ".cutree");
+                if (!tmpFile)
+                    return false;
+                m_cutreeStatFileIn = fopen(tmpFile, "rb");
+                X265_FREE(tmpFile);
+                if (!m_cutreeStatFileIn)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "can't open stats file %s\n", tmpFile);
+                    return false;
+                }
+            }
+
+            /* check whether 1st pass options were compatible with current options */
+            if (strncmp(statsBuf, "#options:", 9))
+            {
+                x265_log(m_param, X265_LOG_ERROR,"options list in stats file not valid\n");
+                return false;
+            }
+            {
+                int i, j;
+                uint32_t k , l;
+                bool bErr = false;
+                char *opts = statsBuf;
+                statsIn = strchr(statsBuf, '\n');
+                if (!statsIn)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "Malformed stats file\n");
+                    return false;
+                }
+                *statsIn = '\0';
+                statsIn++;
+                if (sscanf(opts, "#options: %dx%d", &i, &j) != 2)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "Resolution specified in stats file not valid\n");
+                    return false;
+                }
+                if ((p = strstr(opts, " fps=")) == 0 || sscanf(p, " fps=%u/%u", &k, &l) != 2)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "fps specified in stats file not valid\n");
+                    return false;
+                }
+                if (k != m_param->fpsNum || l != m_param->fpsDenom)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "fps mismatch with 1st pass (%u/%u vs %u/%u)\n",
+                              m_param->fpsNum, m_param->fpsDenom, k, l);
+                    return false;
+                }
+                CMP_OPT_FIRST_PASS("bitdepth", m_param->internalBitDepth);
+                CMP_OPT_FIRST_PASS("weightp", m_param->bEnableWeightedPred);
+                CMP_OPT_FIRST_PASS("bframes", m_param->bframes);
+                CMP_OPT_FIRST_PASS("b-pyramid", m_param->bBPyramid);
+                CMP_OPT_FIRST_PASS("open-gop", m_param->bOpenGOP);
+                CMP_OPT_FIRST_PASS("keyint", m_param->keyframeMax);
+                CMP_OPT_FIRST_PASS("scenecut", m_param->scenecutThreshold);
+
+                if ((p = strstr(opts, "b-adapt=")) != 0 && sscanf(p, "b-adapt=%d", &i) && i >= X265_B_ADAPT_NONE && i <= X265_B_ADAPT_TRELLIS)
+                {
+                    m_param->bFrameAdaptive = i;
+                }
+                else if (m_param->bframes)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "b-adapt method specified in stats file not valid\n");
+                    return false;
+                }
+
+                if ((p = strstr(opts, "rc-lookahead=")) != 0 && sscanf(p, "rc-lookahead=%d", &i))
+                    m_param->lookaheadDepth = i;
+            }
+            /* find number of pics */
+            p = statsIn;
+            int numEntries;
+            for (numEntries = -1; p; numEntries++)
+                p = strchr(p + 1, ';');
+            if (!numEntries)
+            {
+                x265_log(m_param, X265_LOG_ERROR, "empty stats file\n");
+                return false;
+            }
+            m_numEntries = numEntries;
+
+            if (m_param->totalFrames < m_numEntries && m_param->totalFrames > 0)
+            {
+                x265_log(m_param, X265_LOG_WARNING, "2nd pass has fewer frames than 1st pass (%d vs %d)\n",
+                         m_param->totalFrames, m_numEntries);
+            }
+            if (m_param->totalFrames > m_numEntries)
+            {
+                x265_log(m_param, X265_LOG_ERROR, "2nd pass has more frames than 1st pass (%d vs %d)\n",
+                         m_param->totalFrames, m_numEntries);
+                return false;
+            }
+
+            m_rce2Pass = X265_MALLOC(RateControlEntry, m_numEntries);
+            if (!m_rce2Pass)
+            {
+                 x265_log(m_param, X265_LOG_ERROR, "Rce Entries for 2 pass cannot be allocated\n");
+                 return false;
+            }
+            /* init all to skipped p frames */
+            for (int i = 0; i < m_numEntries; i++)
+            {
+                RateControlEntry *rce = &m_rce2Pass[i];
+                rce->sliceType = P_SLICE;
+                rce->qScale = rce->newQScale = x265_qp2qScale(20);
+                rce->miscBits = m_ncu + 10;
+                rce->newQp = 0;
+            }
+            /* read stats */
+            p = statsIn;
+            double totalQpAq = 0;
+            for (int i = 0; i < m_numEntries; i++)
+            {
+                RateControlEntry *rce;
+                int frameNumber;
+                char picType;
+                int e;
+                char *next;
+                double qpRc, qpAq;
+                next = strstr(p, ";");
+                if (next)
+                    *next++ = 0;
+                e = sscanf(p, " in:%d ", &frameNumber);
+                if (frameNumber < 0 || frameNumber >= m_numEntries)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "bad frame number (%d) at stats line %d\n", frameNumber, i);
+                    return false;
+                }
+                rce = &m_rce2Pass[frameNumber];
+                e += sscanf(p, " in:%*d out:%*d type:%c q:%lf q-aq:%lf tex:%d mv:%d misc:%d icu:%lf pcu:%lf scu:%lf",
+                       &picType, &qpRc, &qpAq, &rce->coeffBits,
+                       &rce->mvBits, &rce->miscBits, &rce->iCuCount, &rce->pCuCount,
+                       &rce->skipCuCount);
+                rce->keptAsRef = true;
+                if (picType == 'b' || picType == 'p')
+                    rce->keptAsRef = false;
+                if (picType == 'I' || picType == 'i')
+                    rce->sliceType = I_SLICE;
+                else if (picType == 'P' || picType == 'p')
+                    rce->sliceType = P_SLICE;
+                else if (picType == 'B' || picType == 'b')
+                    rce->sliceType = B_SLICE;
+                else
+                    e = -1;
+                if (e < 10)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "statistics are damaged at line %d, parser out=%d\n", i, e);
+                    return false;
+                }
+                rce->qScale = x265_qp2qScale(qpRc);
+                totalQpAq += qpAq;
+                p = next;
+            }
+            X265_FREE(statsBuf);
+
+            if (m_param->rc.rateControlMode == X265_RC_ABR)
+            {
+                if (!initPass2())
+                    return false;
+            } /* else we're using constant quant, so no need to run the bitrate allocation */
+        }
+        /* Open output file */
+        /* If input and output files are the same, output to a temp file
+         * and move it to the real name only when it's complete */
+        if (m_param->rc.bStatWrite)
+        {
+            char *p, *statFileTmpname;
+            statFileTmpname = strcatFilename(fileName, ".temp");
+            if (!statFileTmpname)
+                return false;
+            m_statFileOut = fopen(statFileTmpname, "wb");
+            X265_FREE(statFileTmpname);
+            if (!m_statFileOut)
+            {
+                x265_log(m_param, X265_LOG_ERROR, "can't open stats file %s\n, statFileTmpname");
+                return false;
+            }
+            p = x265_param2string(m_param);
+            if (p)
+                fprintf(m_statFileOut, "#options: %s\n", p);
+            X265_FREE(p);
+            if (m_param->rc.cuTree && !m_param->rc.bStatRead)
+            {
+                statFileTmpname = strcatFilename(fileName, ".cutree.temp");
+                if (!statFileTmpname)
+                    return false;
+                m_cutreeStatFileOut = fopen(statFileTmpname, "wb");
+                X265_FREE(statFileTmpname);
+                if (!m_cutreeStatFileOut)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "can't open mbtree stats file %s\n", statFileTmpname);
+                    return false;
+                }
+            }
+        }
+        if (m_param->rc.cuTree)
         {
-            param->rc.vbvBufferSize = (int)(param->rc.vbvMaxBitrate / fps);
-            x265_log(param, X265_LOG_WARNING, "VBV buffer size cannot be smaller than one frame, using %d kbit\n",
-                     param->rc.vbvBufferSize);
+            m_cuTreeStats.qpBuffer[0] = X265_MALLOC(uint16_t, m_ncu * sizeof(uint16_t));
+            if (m_param->bBPyramid && m_param->rc.bStatRead)
+                m_cuTreeStats.qpBuffer[1] = X265_MALLOC(uint16_t, m_ncu * sizeof(uint16_t));
+            m_cuTreeStats.qpBufPos = -1;
         }
-        int vbvBufferSize = param->rc.vbvBufferSize * 1000;
-        int vbvMaxBitrate = param->rc.vbvMaxBitrate * 1000;
+    }
+    return true;
+}
 
-        bufferRate = vbvMaxBitrate / fps;
-        vbvMaxRate = vbvMaxBitrate;
-        bufferSize = vbvBufferSize;
-        singleFrameVbv = bufferRate * 1.1 > bufferSize;
+void RateControl::initHRD(SPS *sps)
+{
+    int vbvBufferSize = m_param->rc.vbvBufferSize * 1000;
+    int vbvMaxBitrate = m_param->rc.vbvMaxBitrate * 1000;
+
+    // Init HRD
+    HRDInfo* hrd = &sps->vuiParameters.hrdParameters;
+    hrd->cbrFlag = m_isCbr;
+
+    // normalize HRD size and rate to the value / scale notation
+    hrd->bitRateScale = Clip3(0, 15, calcScale(vbvMaxBitrate) - BR_SHIFT);
+    hrd->bitRateValue = (vbvMaxBitrate >> (hrd->bitRateScale + BR_SHIFT));
+
+    hrd->cpbSizeScale = Clip3(0, 15, calcScale(vbvBufferSize) - CPB_SHIFT);
+    hrd->cpbSizeValue = (vbvBufferSize >> (hrd->cpbSizeScale + CPB_SHIFT));
+    int bitRateUnscale = hrd->bitRateValue << (hrd->bitRateScale + BR_SHIFT);
+    int cpbSizeUnscale = hrd->cpbSizeValue << (hrd->cpbSizeScale + CPB_SHIFT);
+
+    // arbitrary
+    #define MAX_DURATION 0.5
+
+    TimingInfo *time = &sps->vuiParameters.timingInfo;
+    int maxCpbOutputDelay = (int)(X265_MIN(m_param->keyframeMax * MAX_DURATION * time->timeScale / time->numUnitsInTick, INT_MAX));
+    int maxDpbOutputDelay = (int)(sps->maxDecPicBuffering * MAX_DURATION * time->timeScale / time->numUnitsInTick);
+    int maxDelay = (int)(90000.0 * cpbSizeUnscale / bitRateUnscale + 0.5);
+
+    hrd->initialCpbRemovalDelayLength = 2 + Clip3(4, 22, 32 - calcLength(maxDelay));
+    hrd->cpbRemovalDelayLength = Clip3(4, 31, 32 - calcLength(maxCpbOutputDelay));
+    hrd->dpbOutputDelayLength = Clip3(4, 31, 32 - calcLength(maxDpbOutputDelay));
 
-        if (param->rc.vbvBufferInit > 1.)
-            param->rc.vbvBufferInit = Clip3(0.0, 1.0, param->rc.vbvBufferInit / param->rc.vbvBufferSize);
-        param->rc.vbvBufferInit = Clip3(0.0, 1.0, X265_MAX(param->rc.vbvBufferInit, bufferRate / bufferSize));
-        bufferFillFinal = bufferSize * param->rc.vbvBufferInit;
-        vbvMinRate = /*!rc->b_2pass && */ param->rc.rateControlMode == X265_RC_ABR
-            && param->rc.vbvMaxBitrate <= param->rc.bitrate;
+    #undef MAX_DURATION
+}
+
+bool RateControl::initPass2()
+{
+    uint64_t allConstBits = 0;
+    uint64_t allAvailableBits = uint64_t(m_param->rc.bitrate * 1000. * m_numEntries * m_frameDuration);
+    double rateFactor, stepMult;
+    double qBlur = m_param->rc.qblur;
+    double cplxBlur = m_param->rc.complexityBlur;
+    const int filterSize = (int)(qBlur * 4) | 1;
+    double expectedBits;
+    double *qScale, *blurredQscale;
+    double baseCplx = m_ncu * (m_param->bframes ? 120 : 80);
+    double clippedDuration = CLIP_DURATION(m_frameDuration) / BASE_FRAME_DURATION;
+
+    /* find total/average complexity & const_bits */
+    for (int i = 0; i < m_numEntries; i++)
+        allConstBits += m_rce2Pass[i].miscBits;
+
+    if (allAvailableBits < allConstBits)
+    {
+        x265_log(m_param, X265_LOG_ERROR, "requested bitrate is too low. estimated minimum is %d kbps\n",
+                 (int)(allConstBits * m_fps / m_numEntries * 1000.));
+        return false;
     }
 
-    bframes = param->bframes;
-    bframeBits = 0;
-    leadingNoBSatd = 0;
-    if (param->rc.rateControlMode == X265_RC_ABR)
+    /* Blur complexities, to reduce local fluctuation of QP.
+     * We don't blur the QPs directly, because then one very simple frame
+     * could drag down the QP of a nearby complex frame and give it more
+     * bits than intended. */
+    for (int i = 0; i < m_numEntries; i++)
     {
-        /* Adjust the first frame in order to stabilize the quality level compared to the rest */
-#define ABR_INIT_QP_MIN (24 + QP_BD_OFFSET)
-#define ABR_INIT_QP_MAX (34 + QP_BD_OFFSET)
+        double weightSum = 0;
+        double cplxSum = 0;
+        double weight = 1.0;
+        double gaussianWeight;
+        /* weighted average of cplx of future frames */
+        for (int j = 1; j < cplxBlur * 2 && j < m_numEntries - i; j++)
+        {
+            RateControlEntry *rcj = &m_rce2Pass[i + j];
+            weight *= 1 - pow(rcj->iCuCount / m_ncu, 2);
+            if (weight < 0.0001)
+                break;
+            gaussianWeight = weight * exp(-j * j / 200.0);
+            weightSum += gaussianWeight;
+            cplxSum += gaussianWeight * (qScale2bits(rcj, 1) - rcj->miscBits) / clippedDuration;
+        }
+        /* weighted average of cplx of past frames */
+        weight = 1.0;
+        for (int j = 0; j <= cplxBlur * 2 && j <= i; j++)
+        {
+            RateControlEntry *rcj = &m_rce2Pass[i - j];
+            gaussianWeight = weight * exp(-j * j / 200.0);
+            weightSum += gaussianWeight;
+            cplxSum += gaussianWeight * (qScale2bits(rcj, 1) - rcj->miscBits) / clippedDuration;
+            weight *= 1 - pow(rcj->iCuCount / m_ncu, 2);
+            if (weight < .0001)
+                break;
+        }
+        m_rce2Pass[i].blurredComplexity = cplxSum / weightSum;
     }
-    else if (param->rc.rateControlMode == X265_RC_CRF)
+
+    CHECKED_MALLOC(qScale, double, m_numEntries);
+    if (filterSize > 1)
     {
-#define ABR_INIT_QP ((int)param->rc.rfConstant)
+        CHECKED_MALLOC(blurredQscale, double, m_numEntries);
     }
-    reInit();
+    else
+        blurredQscale = qScale;
 
-    ipOffset = 6.0 * X265_LOG2(param->rc.ipFactor);
-    pbOffset = 6.0 * X265_LOG2(param->rc.pbFactor);
-    for (int i = 0; i < 3; i++)
+    /* Search for a factor which, when multiplied by the RCEQ values from
+     * each frame, adds up to the desired total size.
+     * There is no exact closed-form solution because of VBV constraints and
+     * because qscale2bits is not invertible, but we can start with the simple
+     * approximation of scaling the 1st pass by the ratio of bitrates.
+     * The search range is probably overkill, but speed doesn't matter here. */
+
+    expectedBits = 1;
+    for (int i = 0; i < m_numEntries; i++)
     {
-        lastQScaleFor[i] = x265_qp2qScale(param->rc.rateControlMode == X265_RC_CRF ? ABR_INIT_QP : ABR_INIT_QP_MIN);
-        lmin[i] = x265_qp2qScale(MIN_QP);
-        lmax[i] = x265_qp2qScale(MAX_MAX_QP);
+        RateControlEntry* rce = &m_rce2Pass[i];
+        double q = getQScale(rce, 1.0);
+        expectedBits += qScale2bits(rce, q);
+        m_lastQScaleFor[rce->sliceType] = q;
     }
+    stepMult = allAvailableBits / expectedBits;
 
-    if (param->rc.rateControlMode == X265_RC_CQP)
+    rateFactor = 0;
+    for (double step = 1E4 * stepMult; step > 1E-7 * stepMult; step *= 0.5)
     {
-        qpConstant[P_SLICE] = qp;
-        qpConstant[I_SLICE] = Clip3(0, MAX_MAX_QP, (int)(qp - ipOffset + 0.5));
-        qpConstant[B_SLICE] = Clip3(0, MAX_MAX_QP, (int)(qp + pbOffset + 0.5));
+        expectedBits = 0;
+        rateFactor += step;
+
+        m_lastNonBPictType = -1;
+        m_lastAccumPNorm = 1;
+        m_accumPNorm = 0;
+
+        m_lastQScaleFor[0] = m_lastQScaleFor[1] =
+        m_lastQScaleFor[2] = pow(baseCplx, 1 - m_qCompress) / rateFactor;
+
+        /* find qscale */
+        for (int i = 0; i < m_numEntries; i++)
+        {
+            RateControlEntry *rce = &m_rce2Pass[i];
+            qScale[i] = getQScale(rce, rateFactor);
+            m_lastQScaleFor[rce->sliceType] = qScale[i];
+        }
+
+        /* fixed I/B qscale relative to P */
+        for (int i = m_numEntries - 1; i >= 0; i--)
+        {
+            qScale[i] = getDiffLimitedQScale(&m_rce2Pass[i], qScale[i]);
+            X265_CHECK(qScale[i] >= 0, "qScale became negative\n");
+        }
+
+        /* smooth curve */
+        if (filterSize > 1)
+        {
+            X265_CHECK(filterSize % 2 == 1, "filterSize not an odd number\n");
+            for (int i = 0; i < m_numEntries; i++)
+            {
+                double q = 0.0, sum = 0.0;
+
+                for (int j = 0; j < filterSize; j++)
+                {
+                    int idx = i + j - filterSize / 2;
+                    double d = idx - i;
+                    double coeff = qBlur == 0 ? 1.0 : exp(-d * d / (qBlur * qBlur));
+                    if (idx < 0 || idx >= m_numEntries)
+                        continue;
+                    if (m_rce2Pass[i].sliceType != m_rce2Pass[idx].sliceType)
+                        continue;
+                    q += qScale[idx] * coeff;
+                    sum += coeff;
+                }
+                blurredQscale[i] = q / sum;
+            }
+        }
+
+        /* find expected bits */
+        for (int i = 0; i < m_numEntries; i++)
+        {
+            RateControlEntry *rce = &m_rce2Pass[i];
+            rce->newQScale = clipQscale(NULL, blurredQscale[i]); // check if needed
+            X265_CHECK(rce->newQScale >= 0, "new Qscale is negative\n");
+            expectedBits += qScale2bits(rce, rce->newQScale);
+        }
+
+        if (expectedBits > allAvailableBits)
+            rateFactor -= step;
     }
 
-    /* qstep - value set as encoder specific */
-    lstep = pow(2, param->rc.qpStep / 6.0);
+    X265_FREE(qScale);
+    if (filterSize > 1)
+        X265_FREE(blurredQscale);
+
+    if (m_isVbv)
+        if (!vbv2Pass(allAvailableBits))
+            return false;
+    expectedBits = countExpectedBits();
+
+    if (fabs(expectedBits / allAvailableBits - 1.0) > 0.01)
+    {
+        double avgq = 0;
+        for (int i = 0; i < m_numEntries; i++)
+            avgq += m_rce2Pass[i].newQScale;
+        avgq = x265_qScale2qp(avgq / m_numEntries);
+
+        if (expectedBits > allAvailableBits || !m_isVbv)
+            x265_log(m_param, X265_LOG_WARNING, "Error: 2pass curve failed to converge\n");
+        x265_log(m_param, X265_LOG_WARNING, "target: %.2f kbit/s, expected: %.2f kbit/s, avg QP: %.4f\n",
+                 (double)m_param->rc.bitrate,
+                 expectedBits * m_fps / (m_numEntries * 1000.),
+                 avgq);
+        if (expectedBits < allAvailableBits && avgq < QP_MIN + 2)
+        {
+            x265_log(m_param, X265_LOG_WARNING, "try reducing target bitrate\n");
+        }
+        else if (expectedBits > allAvailableBits && avgq > QP_MAX_SPEC - 2)
+        {
+            x265_log(m_param, X265_LOG_WARNING, "try increasing target bitrate\n");
+        }
+        else if (!(m_2pass && m_isVbv))
+            x265_log(m_param, X265_LOG_WARNING, "internal error\n");
+    }
+
+    return true;
+
+fail:
+    x265_log(m_param, X265_LOG_WARNING, "two-pass ABR initialization failed\n");
+    return false;
 }
 
-void RateControl::reInit()
+bool RateControl::vbv2Pass(uint64_t allAvailableBits)
 {
-    totalBits = 0;
-    framesDone = 0;
+    /* for each interval of bufferFull .. underflow, uniformly increase the qp of all
+     * frames in the interval until either buffer is full at some intermediate frame or the
+     * last frame in the interval no longer underflows.  Recompute intervals and repeat.
+     * Then do the converse to put bits back into overflow areas until target size is met */
+
+    double *fills;
+    double expectedBits = 0;
+    double adjustment;
+    double prevBits = 0;
+    int t0, t1;
+    int iterations = 0 , adjMin, adjMax;
+    CHECKED_MALLOC(fills, double, m_numEntries + 1);
+    fills++;
+
+    /* adjust overall stream size */
+    do
+    {
+        iterations++;
+        prevBits = expectedBits;
+
+        if (expectedBits)
+        {   /* not first iteration */
+            adjustment = X265_MAX(X265_MIN(expectedBits / allAvailableBits, 0.999), 0.9);
+            fills[-1] = m_bufferSize * m_param->rc.vbvBufferInit;
+            t0 = 0;
+            /* fix overflows */
+            adjMin = 1;
+            while (adjMin && findUnderflow(fills, &t0, &t1, 1))
+            {
+                adjMin = fixUnderflow(t0, t1, adjustment, MIN_QPSCALE, MAX_MAX_QPSCALE);
+                t0 = t1;
+            }
+        }
 
-    /* estimated ratio that produces a reasonable QP for the first I-frame */
-    cplxrSum = .01 * pow(7.0e5, qCompress) * pow(ncu, 0.5);
-    wantedBitsWindow = bitrate * frameDuration;
-    accumPNorm = .01;
-    accumPQp = (param->rc.rateControlMode == X265_RC_CRF ? ABR_INIT_QP : ABR_INIT_QP_MIN) * accumPNorm;
+        fills[-1] = m_bufferSize * (1. - m_param->rc.vbvBufferInit);
+        t0 = 0;
+        /* fix underflows -- should be done after overflow, as we'd better undersize target than underflowing VBV */
+        adjMax = 1;
+        while (adjMax && findUnderflow(fills, &t0, &t1, 0))
+            adjMax = fixUnderflow(t0, t1, 1.001, MIN_QPSCALE, MAX_MAX_QPSCALE );
 
-    /* Frame Predictors and Row predictors used in vbv */
-    for (int i = 0; i < 5; i++)
-    {
-        pred[i].coeff = 2.0;
-        pred[i].count = 1.0;
-        pred[i].decay = 0.5;
-        pred[i].offset = 0.0;
+        expectedBits = countExpectedBits();
     }
-   
-    predBfromP = pred[0];
+    while ((expectedBits < .995 * allAvailableBits) && ((int64_t)(expectedBits+.5) > (int64_t)(prevBits+.5)));
+
+    if (!adjMax)
+        x265_log(m_param, X265_LOG_WARNING, "vbv-maxrate issue, qpmax or vbv-maxrate too low\n");
+
+    /* store expected vbv filling values for tracking when encoding */
+    for (int i = 0; i < m_numEntries; i++)
+        m_rce2Pass[i].expectedVbv = m_bufferSize - fills[i];
+
+    X265_FREE(fills - 1);
+    return true;
+
+fail:
+    x265_log(m_param, X265_LOG_ERROR, "malloc failure in two-pass VBV init\n");
+    return false;
 }
 
+/* In 2pass, force the same frame types as in the 1st pass */
+int RateControl::rateControlSliceType(int frameNum)
+{
+    if (m_param->rc.bStatRead)
+    {
+        if (frameNum >= m_numEntries)
+        {
+            /* We could try to initialize everything required for ABR and
+             * adaptive B-frames, but that would be complicated.
+             * So just calculate the average QP used so far. */
+            m_param->rc.qp = (m_accumPQp < 1) ? ABR_INIT_QP_MAX : (int)(m_accumPQp + 0.5);
+            m_qpConstant[P_SLICE] = Clip3(0, QP_MAX_MAX, m_param->rc.qp);
+            m_qpConstant[I_SLICE] = Clip3(0, QP_MAX_MAX, (int)(m_param->rc.qp - m_ipOffset + 0.5));
+            m_qpConstant[B_SLICE] = Clip3(0, QP_MAX_MAX, (int)(m_param->rc.qp + m_pbOffset + 0.5));
+
+            x265_log(m_param, X265_LOG_ERROR, "2nd pass has more frames than 1st pass (%d)\n", m_numEntries);
+            x265_log(m_param, X265_LOG_ERROR, "continuing anyway, at constant QP=%d\n", m_param->rc.qp);
+            if (m_param->bFrameAdaptive)
+                x265_log(m_param, X265_LOG_ERROR, "disabling adaptive B-frames\n");
+
+            m_isAbr = 0;
+            m_2pass = 0;
+            m_param->rc.rateControlMode = X265_RC_CQP;
+            m_param->rc.bStatRead = 0;
+            m_param->bFrameAdaptive = 0;
+            m_param->scenecutThreshold = 0;
+            m_param->rc.cuTree = 0;
+            if (m_param->bframes > 1)
+                m_param->bframes = 1;
+            return X265_TYPE_AUTO;
+        }
+        int frameType = m_rce2Pass[frameNum].sliceType == I_SLICE ? (frameNum > 0 && m_param->bOpenGOP ? X265_TYPE_I : X265_TYPE_IDR)
+                            : m_rce2Pass[frameNum].sliceType == P_SLICE ? X265_TYPE_P
+                            : (m_rce2Pass[frameNum].sliceType == B_SLICE && m_rce2Pass[frameNum].keptAsRef? X265_TYPE_BREF : X265_TYPE_B);
+        return frameType;
+    }
+    else
+        return X265_TYPE_AUTO;
+}
 
-void RateControl::rateControlStart(TComPic* pic, Lookahead *l, RateControlEntry* rce, Encoder* enc)
+int RateControl::rateControlStart(Frame* pic, RateControlEntry* rce, Encoder* enc)
 {
-    curSlice = pic->getSlice();
-    sliceType = curSlice->getSliceType();
-    rce->sliceType = sliceType;
+    int orderValue = m_startEndOrder.get();
+    int startOrdinal = rce->encodeOrder * 2;
+
+    while (orderValue < startOrdinal && !m_bTerminated)
+       orderValue = m_startEndOrder.waitForChange(orderValue);
+
+    if (!pic)
+    {
+        // faked rateControlStart calls when the encoder is flushing
+        m_startEndOrder.incr();
+        return 0;
+    }
+
+    m_curSlice = pic->m_picSym->m_slice;
+    m_sliceType = m_curSlice->m_sliceType;
+    rce->sliceType = m_sliceType;
+    rce->poc = m_curSlice->m_poc;
+    if (m_param->rc.bStatRead)
+    {
+        X265_CHECK(rce->poc >= 0 && rce->poc < m_numEntries, "bad encode ordinal\n");
+        copyRceData(rce, &m_rce2Pass[rce->poc]);
+    }
     rce->isActive = true;
-    if (sliceType == B_SLICE)
-        rce->bframes = bframes;
+    if (m_sliceType == B_SLICE)
+        rce->bframes = m_leadingBframes;
     else
-        bframes = pic->m_lowres.leadingBframes;
+        m_leadingBframes = pic->m_lowres.leadingBframes;
 
     rce->bLastMiniGopBFrame = pic->m_lowres.bLastMiniGopBFrame;
-    rce->bufferRate = bufferRate;
-    rce->poc = curSlice->getPOC();
-    if (isVbv)
+    rce->bufferRate = m_bufferRate;
+    rce->rowCplxrSum = 0.0;
+    rce->rowTotalBits = 0;
+    if (m_isVbv)
     {
         if (rce->rowPreds[0][0].count == 0)
         {
-             for (int i = 0; i < 3; i++)
-             {
-                 for(int j = 0; j < 2; j++)
+            for (int i = 0; i < 3; i++)
+            {
+                for (int j = 0; j < 2; j++)
                 {
                     rce->rowPreds[i][j].coeff = 0.25;
                     rce->rowPreds[i][j].count = 1.0;
                     rce->rowPreds[i][j].decay = 0.5;
                     rce->rowPreds[i][j].offset = 0.0;
                 }
-             }
+            }
         }
-        rce->rowPred[0] = &rce->rowPreds[sliceType][0];
-        rce->rowPred[1] = &rce->rowPreds[sliceType][1];
+        rce->rowPred[0] = &rce->rowPreds[m_sliceType][0];
+        rce->rowPred[1] = &rce->rowPreds[m_sliceType][1];
+        m_predictedBits = m_totalBits;
         updateVbvPlan(enc);
-        rce->bufferFill = bufferFill;
+        rce->bufferFill = m_bufferFill;
     }
-    if (isAbr) //ABR,CRF
+    if (m_isAbr || m_2pass) // ABR,CRF
     {
-        currentSatd = l->getEstimatedPictureCost(pic) >> (X265_DEPTH - 8);
-        /* Update rce for use in rate control VBV later */
-        rce->lastSatd = currentSatd;
+        if (m_isAbr || m_isVbv)
+        {
+            m_currentSatd = pic->m_lowres.satdCost >> (X265_DEPTH - 8);
+            /* Update rce for use in rate control VBV later */
+            rce->lastSatd = m_currentSatd;
+        }
         double q = x265_qScale2qp(rateEstimateQscale(pic, rce));
-        q = Clip3((double)MIN_QP, (double)MAX_MAX_QP, q);
-        qp = int(q + 0.5);
+        q = Clip3((double)QP_MIN, (double)QP_MAX_MAX, q);
+        m_qp = int(q + 0.5);
         rce->qpaRc = pic->m_avgQpRc = pic->m_avgQpAq = q;
         /* copy value of lastRceq into thread local rce struct *to be used in RateControlEnd() */
-        rce->qRceq = lastRceq;
-        rce->qpNoVbv = qpNoVbv;
+        rce->qRceq = m_lastRceq;
         accumPQpUpdate();
     }
-    else //CQP
+    else // CQP
     {
-        if (sliceType == B_SLICE && curSlice->isReferenced())
-            qp = (qpConstant[B_SLICE] + qpConstant[P_SLICE]) / 2;
+        if (m_sliceType == B_SLICE && IS_REFERENCED(m_curSlice))
+            m_qp = (m_qpConstant[B_SLICE] + m_qpConstant[P_SLICE]) / 2;
         else
-            qp = qpConstant[sliceType];
-        pic->m_avgQpAq = pic->m_avgQpRc = qp;
+            m_qp = m_qpConstant[m_sliceType];
+        pic->m_avgQpAq = pic->m_avgQpRc = m_qp;
+    }
+    if (m_sliceType != B_SLICE)
+    {
+        m_lastNonBPictType = m_sliceType;
+        m_leadingNoBSatd = m_currentSatd;
     }
-    if (sliceType != B_SLICE)
+    rce->leadingNoBSatd = m_leadingNoBSatd;
+    if (pic->m_forceqp)
     {
-        lastNonBPictType = sliceType;
-        leadingNoBSatd = currentSatd;
+        m_qp = int32_t(pic->m_forceqp + 0.5) - 1;
+        m_qp = Clip3(QP_MIN, QP_MAX_MAX, m_qp);
+        rce->qpaRc = pic->m_avgQpRc = pic->m_avgQpAq = m_qp;
     }
-    rce->leadingNoBSatd = leadingNoBSatd;
-    framesDone++;
-    /* set the final QP to slice structure */
-    curSlice->setSliceQp(qp);
+    // Do not increment m_startEndOrder here. Make rateControlEnd of previous thread
+    // to wait until rateControlUpdateStats of this frame is called
+    m_framesDone++;
+    return m_qp;
 }
 
 void RateControl::accumPQpUpdate()
 {
-    accumPQp   *= .95;
-    accumPNorm *= .95;
-    accumPNorm += 1;
-    if (sliceType == I_SLICE)
-        accumPQp += qp + ipOffset;
+    m_accumPQp   *= .95;
+    m_accumPNorm *= .95;
+    m_accumPNorm += 1;
+    if (m_sliceType == I_SLICE)
+        m_accumPQp += m_qp + m_ipOffset;
     else
-        accumPQp += qp;
+        m_accumPQp += m_qp;
 }
 
-double RateControl::rateEstimateQscale(TComPic* pic, RateControlEntry *rce)
+double RateControl::getDiffLimitedQScale(RateControlEntry *rce, double q)
+{
+    // force I/B quants as a function of P quants
+    const double lastPqScale    = m_lastQScaleFor[P_SLICE];
+    const double lastNonBqScale= m_lastQScaleFor[m_lastNonBPictType];
+    if (rce->sliceType == I_SLICE)
+    {
+        double iq = q;
+        double pq = x265_qp2qScale(m_accumPQp / m_accumPNorm);
+        double ipFactor = fabs(m_param->rc.ipFactor);
+        /* don't apply ipFactor if the following frame is also I */
+        if (m_accumPNorm <= 0)
+            q = iq;
+        else if (m_param->rc.ipFactor < 0)
+            q = iq / ipFactor;
+        else if (m_accumPNorm >= 1)
+            q = pq / ipFactor;
+        else
+            q = m_accumPNorm * pq / ipFactor + (1 - m_accumPNorm) * iq;
+    }
+    else if (rce->sliceType == B_SLICE)
+    {
+        if (m_param->rc.pbFactor > 0)
+            q = lastNonBqScale;
+        if (!rce->keptAsRef)
+            q *= fabs(m_param->rc.pbFactor);
+    }
+    else if (rce->sliceType == P_SLICE
+             && m_lastNonBPictType == P_SLICE
+             && rce->coeffBits == 0)
+    {
+        q = lastPqScale;
+    }
+
+    /* last qscale / qdiff stuff */
+    if (m_lastNonBPictType == rce->sliceType &&
+        (rce->sliceType != I_SLICE || m_lastAccumPNorm < 1))
+    {
+        double maxQscale = m_lastQScaleFor[rce->sliceType] * m_lstep;
+        double minQscale = m_lastQScaleFor[rce->sliceType] / m_lstep;
+        q = Clip3(minQscale, maxQscale, q);
+    }
+
+    m_lastQScaleFor[rce->sliceType] = q;
+    if (rce->sliceType != B_SLICE)
+        m_lastNonBPictType = rce->sliceType;
+    if (rce->sliceType == I_SLICE)
+    {
+        m_lastAccumPNorm = m_accumPNorm;
+        m_accumPNorm = 0;
+        m_accumPQp = 0;
+    }
+    if (rce->sliceType == P_SLICE)
+    {
+        double mask = 1 - pow(rce->iCuCount / m_ncu, 2);
+        m_accumPQp   = mask * (x265_qScale2qp(q) + m_accumPQp);
+        m_accumPNorm = mask * (1 + m_accumPNorm);
+    }
+
+    return q;
+}
+
+double RateControl::countExpectedBits()
+{
+    double expectedBits = 0;
+    for( int i = 0; i < m_numEntries; i++ )
+    {
+        RateControlEntry *rce = &m_rce2Pass[i];
+        rce->expectedBits = (uint64_t)expectedBits;
+        expectedBits += qScale2bits(rce, rce->newQScale);
+    }
+    return expectedBits;
+}
+
+bool RateControl::findUnderflow(double *fills, int *t0, int *t1, int over)
+{
+    /* find an interval ending on an overflow or underflow (depending on whether
+     * we're adding or removing bits), and starting on the earliest frame that
+     * can influence the buffer fill of that end frame. */
+    const double bufferMin = .1 * m_bufferSize;
+    const double bufferMax = .9 * m_bufferSize;
+    double fill = fills[*t0 - 1];
+    double parity = over ? 1. : -1.;
+    int start = -1, end = -1;
+    for (int i = *t0; i < m_numEntries; i++)
+    {
+        fill += (m_frameDuration * m_vbvMaxRate -
+                 qScale2bits(&m_rce2Pass[i], m_rce2Pass[i].newQScale)) * parity;
+        fill = Clip3(0.0, m_bufferSize, fill);
+        fills[i] = fill;
+        if (fill <= bufferMin || i == 0)
+        {
+            if (end >= 0)
+                break;
+            start = i;
+        }
+        else if (fill >= bufferMax && start >= 0)
+            end = i;
+    }
+    *t0 = start;
+    *t1 = end;
+    return start >= 0 && end >= 0;
+}
+
+bool RateControl::fixUnderflow(int t0, int t1, double adjustment, double qscaleMin, double qscaleMax)
+{
+    double qscaleOrig, qscaleNew;
+    bool adjusted = false;
+    if (t0 > 0)
+        t0++;
+    for (int i = t0; i <= t1; i++)
+    {
+        qscaleOrig = m_rce2Pass[i].newQScale;
+        qscaleOrig = Clip3(qscaleMin, qscaleMax, qscaleOrig);
+        qscaleNew  = qscaleOrig * adjustment;
+        qscaleNew  = Clip3(qscaleMin, qscaleMax, qscaleNew);
+        m_rce2Pass[i].newQScale = qscaleNew;
+        adjusted = adjusted || (qscaleNew != qscaleOrig);
+    }
+    return adjusted;
+}
+
+bool RateControl::cuTreeReadFor2Pass(Frame* frame)
+{
+    uint8_t sliceTypeActual = (uint8_t)m_rce2Pass[frame->m_POC].sliceType;
+
+    if (m_rce2Pass[frame->m_POC].keptAsRef)
+    {
+        uint8_t type;
+        if (m_cuTreeStats.qpBufPos < 0)
+        {
+            do
+            {
+                m_cuTreeStats.qpBufPos++;
+
+                if (!fread(&type, 1, 1, m_cutreeStatFileIn))
+                    goto fail;
+                if (fread(m_cuTreeStats.qpBuffer[m_cuTreeStats.qpBufPos], sizeof(uint16_t), m_ncu, m_cutreeStatFileIn) != (size_t)m_ncu)
+                    goto fail;
+
+                if (type != sliceTypeActual && m_cuTreeStats.qpBufPos == 1)
+                {
+                    x265_log(m_param, X265_LOG_ERROR, "CU-tree frametype %d doesn't match actual frametype %d.\n", type, sliceTypeActual);
+                    return false;
+                }
+            }
+            while(type != sliceTypeActual);
+        }
+        for (int i = 0; i < m_ncu; i++)
+        {
+            int16_t qpFix8 = m_cuTreeStats.qpBuffer[m_cuTreeStats.qpBufPos][i];
+            frame->m_lowres.qpCuTreeOffset[i] = (double)(qpFix8) / 256.0;
+            frame->m_lowres.invQscaleFactor[i] = x265_exp2fix8(frame->m_lowres.qpCuTreeOffset[i]);
+        }
+        m_cuTreeStats.qpBufPos--;
+    }
+    else
+        calcAdaptiveQuantFrame(frame);
+    return true;
+fail:
+   x265_log(m_param, X265_LOG_ERROR, "Incomplete CU-tree stats file.\n");
+    return false;
+}
+
+double RateControl::rateEstimateQscale(Frame* pic, RateControlEntry *rce)
 {
     double q;
 
-    if (sliceType == B_SLICE)
+    if (m_2pass)
+    {
+        if (m_sliceType != rce->sliceType)
+        {
+            x265_log(m_param, X265_LOG_ERROR, "slice=%c but 2pass stats say %c\n",
+                     g_sliceTypeToChar[m_sliceType], g_sliceTypeToChar[rce->sliceType]);
+        }
+    }
+    if (m_sliceType == B_SLICE)
     {
         /* B-frames don't have independent rate control, but rather get the
          * average QP of the two adjacent P-frames + an offset */
-        TComSlice* prevRefSlice = curSlice->getRefPic(REF_PIC_LIST_0, 0)->getSlice();
-        TComSlice* nextRefSlice = curSlice->getRefPic(REF_PIC_LIST_1, 0)->getSlice();
-        double q0 = curSlice->getRefPic(REF_PIC_LIST_0, 0)->m_avgQpRc;
-        double q1 = curSlice->getRefPic(REF_PIC_LIST_1, 0)->m_avgQpRc;
-        bool i0 = prevRefSlice->getSliceType() == I_SLICE;
-        bool i1 = nextRefSlice->getSliceType() == I_SLICE;
-        int dt0 = abs(curSlice->getPOC() - prevRefSlice->getPOC());
-        int dt1 = abs(curSlice->getPOC() - nextRefSlice->getPOC());
+        Slice* prevRefSlice = m_curSlice->m_refPicList[0][0]->m_picSym->m_slice;
+        Slice* nextRefSlice = m_curSlice->m_refPicList[1][0]->m_picSym->m_slice;
+        double q0 = m_curSlice->m_refPicList[0][0]->m_avgQpRc;
+        double q1 = m_curSlice->m_refPicList[1][0]->m_avgQpRc;
+        bool i0 = prevRefSlice->m_sliceType == I_SLICE;
+        bool i1 = nextRefSlice->m_sliceType == I_SLICE;
+        int dt0 = abs(m_curSlice->m_poc - prevRefSlice->m_poc);
+        int dt1 = abs(m_curSlice->m_poc - nextRefSlice->m_poc);
 
         // Skip taking a reference frame before the Scenecut if ABR has been reset.
-        if (lastAbrResetPoc >= 0)
+        if (m_lastAbrResetPoc >= 0)
         {
-            if (prevRefSlice->getSliceType() == P_SLICE && prevRefSlice->getPOC() < lastAbrResetPoc)
+            if (prevRefSlice->m_sliceType == P_SLICE && prevRefSlice->m_poc < m_lastAbrResetPoc)
             {
                 i0 = i1;
                 dt0 = dt1;
                 q0 = q1;
             }
         }
-        if (prevRefSlice->getSliceType() == B_SLICE && prevRefSlice->isReferenced())
-            q0 -= pbOffset / 2;
-        if (nextRefSlice->getSliceType() == B_SLICE && nextRefSlice->isReferenced())
-            q1 -= pbOffset / 2;
+        if (prevRefSlice->m_sliceType == B_SLICE && IS_REFERENCED(prevRefSlice))
+            q0 -= m_pbOffset / 2;
+        if (nextRefSlice->m_sliceType == B_SLICE && IS_REFERENCED(nextRefSlice))
+            q1 -= m_pbOffset / 2;
         if (i0 && i1)
-            q = (q0 + q1) / 2 + ipOffset;
+            q = (q0 + q1) / 2 + m_ipOffset;
         else if (i0)
             q = q1;
         else if (i1)
@@ -485,152 +1359,281 @@
         else
             q = (q0 * dt1 + q1 * dt0) / (dt0 + dt1);
 
-        if (curSlice->isReferenced())
-            q += pbOffset / 2;
+        if (IS_REFERENCED(m_curSlice))
+            q += m_pbOffset / 2;
         else
-            q += pbOffset;
-        qpNoVbv = q;
-        double qScale = x265_qp2qScale(qpNoVbv);
-        rce->frameSizePlanned = predictSize(&predBfromP, qScale, (double)leadingNoBSatd);
+            q += m_pbOffset;
+        rce->qpNoVbv = q;
+        double qScale = x265_qp2qScale(q);
+
+        if (!m_2pass && m_isVbv)
+        {
+            if (m_leadingBframes > 5)
+            {
+                qScale = clipQscale(pic, qScale);
+                m_lastQScaleFor[m_sliceType] = qScale;
+            }
+            rce->frameSizePlanned = predictSize(&m_predBfromP, qScale, (double)m_leadingNoBSatd);
+        }
+        else if (m_2pass && m_isVbv)
+        {
+            rce->frameSizePlanned = qScale2bits(rce, qScale);
+        }
         rce->frameSizeEstimated = rce->frameSizePlanned;
+        rce->newQScale = qScale;
         return qScale;
     }
     else
     {
-        double abrBuffer = 2 * param->rc.rateTolerance * bitrate;
-
-        /* 1pass ABR */
-
-        /* Calculate the quantizer which would have produced the desired
-         * average bitrate if it had been applied to all frames so far.
-         * Then modulate that quant based on the current frame's complexity
-         * relative to the average complexity so far (using the 2pass RCEQ).
-         * Then bias the quant up or down if total size so far was far from
-         * the target.
-         * Result: Depending on the value of rate_tolerance, there is a
-         * tradeoff between quality and bitrate precision. But at large
-         * tolerances, the bit distribution approaches that of 2pass. */
-
-        double wantedBits, overflow = 1;
-        rce->movingAvgSum = shortTermCplxSum;
-        shortTermCplxSum *= 0.5;
-        shortTermCplxCount *= 0.5;
-        shortTermCplxSum += currentSatd / (CLIP_DURATION(frameDuration) / BASE_FRAME_DURATION);
-        shortTermCplxCount++;
-        /* texBits to be used in 2-pass */
-        rce->texBits = currentSatd;
-        rce->blurredComplexity = shortTermCplxSum / shortTermCplxCount;
-        rce->mvBits = 0;
-        rce->sliceType = sliceType;
-
-        if (param->rc.rateControlMode == X265_RC_CRF)
+        double abrBuffer = 2 * m_param->rc.rateTolerance * m_bitrate;
+        if (m_2pass)
         {
-            q = getQScale(rce, rateFactorConstant);
+            int64_t diff;
+            if (!m_isVbv)
+            {
+                m_predictedBits = m_totalBits;
+                if (rce->encodeOrder < m_param->frameNumThreads)
+                    m_predictedBits += (int64_t)(rce->encodeOrder * m_bitrate / m_fps);
+                else
+                    m_predictedBits += (int64_t)(m_param->frameNumThreads * m_bitrate / m_fps);
+            }
+            /* Adjust ABR buffer based on distance to the end of the video. */
+            if (m_numEntries > rce->encodeOrder)
+            {
+                uint64_t finalBits = m_rce2Pass[m_numEntries - 1].expectedBits;
+                double videoPos = (double)rce->expectedBits / finalBits;
+                double scaleFactor = sqrt((1 - videoPos) * m_numEntries);
+                abrBuffer *= 0.5 * X265_MAX(scaleFactor, 0.5);
+            }
+            diff = m_predictedBits - (int64_t)rce->expectedBits;
+            q = rce->newQScale;
+            q /= Clip3(0.5, 2.0, (double)(abrBuffer - diff) / abrBuffer);
+            if (m_expectedBitsSum > 0)
+            {
+                /* Adjust quant based on the difference between
+                 * achieved and expected bitrate so far */
+                double curTime = (double)rce->encodeOrder / m_numEntries;
+                double w = Clip3(0.0, 1.0, curTime * 100);
+                q *= pow((double)m_totalBits / m_expectedBitsSum, w);
+            }
+            rce->qpNoVbv = x265_qScale2qp(q);
+            if (m_isVbv)
+            {
+                /* Do not overflow vbv */
+                double expectedSize = qScale2bits(rce, q);
+                double expectedVbv = m_bufferFill + m_bufferRate - expectedSize;
+                double expectedFullness = rce->expectedVbv / m_bufferSize;
+                double qmax = q * (2 - expectedFullness);
+                double sizeConstraint = 1 + expectedFullness;
+                qmax = X265_MAX(qmax, rce->newQScale);
+                if (expectedFullness < .05)
+                    qmax = MAX_MAX_QPSCALE;
+                qmax = X265_MIN(qmax, MAX_MAX_QPSCALE);
+                while (((expectedVbv < rce->expectedVbv/sizeConstraint) && (q < qmax)) ||
+                        ((expectedVbv < 0) && (q < MAX_MAX_QPSCALE)))
+                {
+                    q *= 1.05;
+                    expectedSize = qScale2bits(rce, q);
+                    expectedVbv = m_bufferFill + m_bufferRate - expectedSize;
+                }
+            }
+            q = Clip3(MIN_QPSCALE, MAX_MAX_QPSCALE, q);
         }
         else
         {
-
-            checkAndResetABR(rce, false);
-            q = getQScale(rce, wantedBitsWindow / cplxrSum);
-
-            /* ABR code can potentially be counterproductive in CBR, so just
-             * don't bother.  Don't run it if the frame complexity is zero
-             * either. */
-            if (!vbvMinRate && currentSatd)
+            /* 1pass ABR */
+
+            /* Calculate the quantizer which would have produced the desired
+             * average bitrate if it had been applied to all frames so far.
+             * Then modulate that quant based on the current frame's complexity
+             * relative to the average complexity so far (using the 2pass RCEQ).
+             * Then bias the quant up or down if total size so far was far from
+             * the target.
+             * Result: Depending on the value of rate_tolerance, there is a
+             * tradeoff between quality and bitrate precision. But at large
+             * tolerances, the bit distribution approaches that of 2pass. */
+
+            double wantedBits, overflow = 1;
+            rce->movingAvgSum = m_shortTermCplxSum;
+            m_shortTermCplxSum *= 0.5;
+            m_shortTermCplxCount *= 0.5;
+            m_shortTermCplxSum += m_currentSatd / (CLIP_DURATION(m_frameDuration) / BASE_FRAME_DURATION);
+            m_shortTermCplxCount++;
+            /* coeffBits to be used in 2-pass */
+            rce->coeffBits = (int)m_currentSatd;
+            rce->blurredComplexity = m_shortTermCplxSum / m_shortTermCplxCount;
+            rce->mvBits = 0;
+            rce->sliceType = m_sliceType;
+
+            if (m_param->rc.rateControlMode == X265_RC_CRF)
+            {
+                q = getQScale(rce, m_rateFactorConstant);
+            }
+            else
             {
-                /* use framesDone instead of POC as poc count is not serial with bframes enabled */
-                double timeDone = (double)(framesDone - param->frameNumThreads + 1) * frameDuration;
-                wantedBits = timeDone * bitrate;
-                if (wantedBits > 0 && totalBits > 0)
+                if (!m_param->rc.bStatRead)
+                    checkAndResetABR(rce, false);
+                q = getQScale(rce, m_wantedBitsWindow / m_cplxrSum);
+
+                /* ABR code can potentially be counterproductive in CBR, so just
+                 * don't bother.  Don't run it if the frame complexity is zero
+                 * either. */
+                if (!m_isCbr && m_currentSatd)
                 {
-                    abrBuffer *= X265_MAX(1, sqrt(timeDone));
-                    overflow = Clip3(.5, 2.0, 1.0 + (totalBits - wantedBits) / abrBuffer);
-                    q *= overflow;
+                    /* use framesDone instead of POC as poc count is not serial with bframes enabled */
+                    double timeDone = (double)(m_framesDone - m_param->frameNumThreads + 1) * m_frameDuration;
+                    wantedBits = timeDone * m_bitrate;
+                    if (wantedBits > 0 && m_totalBits > 0 && !m_partialResidualFrames)
+                    {
+                        abrBuffer *= X265_MAX(1, sqrt(timeDone));
+                        overflow = Clip3(.5, 2.0, 1.0 + (m_totalBits - wantedBits) / abrBuffer);
+                        q *= overflow;
+                    }
                 }
             }
-        }
 
-        if (sliceType == I_SLICE && param->keyframeMax > 1
-            && lastNonBPictType != I_SLICE && !isAbrReset)
-        {
-            q = x265_qp2qScale(accumPQp / accumPNorm);
-            q /= fabs(param->rc.ipFactor);
-        }
-
-        if (param->rc.rateControlMode != X265_RC_CRF)
-        {
-            double lqmin = 0, lqmax = 0;
-            if (totalBits == 0 && !isVbv)
+            if (m_sliceType == I_SLICE && m_param->keyframeMax > 1
+                && m_lastNonBPictType != I_SLICE && !m_isAbrReset)
             {
-                lqmin = x265_qp2qScale(ABR_INIT_QP_MIN) / lstep;
-                lqmax = x265_qp2qScale(ABR_INIT_QP_MAX) * lstep;
-                q = Clip3(lqmin, lqmax, q);
+                q = x265_qp2qScale(m_accumPQp / m_accumPNorm);
+                q /= fabs(m_param->rc.ipFactor);
             }
-            else if (totalBits > 0 || (isVbv && framesDone > 0))
+            else if (m_framesDone > 0)
             {
-                lqmin = lastQScaleFor[sliceType] / lstep;
-                lqmax = lastQScaleFor[sliceType] * lstep;
-                if (overflow > 1.1 && framesDone > 3)
-                    lqmax *= lstep;
-                else if (overflow < 0.9)
-                    lqmin /= lstep;
-                q = Clip3(lqmin, lqmax, q);
+                if (m_param->rc.rateControlMode != X265_RC_CRF)
+                {
+                    double lqmin = 0, lqmax = 0;
+                    lqmin = m_lastQScaleFor[m_sliceType] / m_lstep;
+                    lqmax = m_lastQScaleFor[m_sliceType] * m_lstep;
+                    if (!m_partialResidualFrames)
+                    {
+                        if (overflow > 1.1 && m_framesDone > 3)
+                            lqmax *= m_lstep;
+                        else if (overflow < 0.9)
+                            lqmin /= m_lstep;
+                    }
+                    q = Clip3(lqmin, lqmax, q);
+                }
+            }
+            else if (m_qCompress != 1 && m_param->rc.rateControlMode == X265_RC_CRF)
+            {
+                q = x265_qp2qScale(CRF_INIT_QP) / fabs(m_param->rc.ipFactor);
+            }
+            else if (m_framesDone == 0 && !m_isVbv)
+            {
+                /* for ABR alone, clip the first I frame qp */
+                double lqmax = x265_qp2qScale(ABR_INIT_QP_MAX) * m_lstep;
+                q = X265_MIN(lqmax, q);
             }
+            q = Clip3(MIN_QPSCALE, MAX_MAX_QPSCALE, q);
+            rce->qpNoVbv = x265_qScale2qp(q);
+            q = clipQscale(pic, q);
         }
+        m_lastQScaleFor[m_sliceType] = q;
+        if ((m_curSlice->m_poc == 0 || m_lastQScaleFor[P_SLICE] < q) && !(m_2pass && !m_isVbv))
+            m_lastQScaleFor[P_SLICE] = q * fabs(m_param->rc.ipFactor);
+
+        if (m_2pass && m_isVbv)
+            rce->frameSizePlanned = qScale2bits(rce, q);
         else
-        {
-            if (qCompress != 1 && framesDone == 0)
-                q = x265_qp2qScale(ABR_INIT_QP) / fabs(param->rc.ipFactor);
-        }
-        double lmin1 = lmin[sliceType];
-        double lmax1 = lmax[sliceType];
-        q = Clip3(lmin1, lmax1, q);
-        qpNoVbv = x265_qScale2qp(q);
+            rce->frameSizePlanned = predictSize(&m_pred[m_sliceType], q, (double)m_currentSatd);
+        rce->frameSizeEstimated = rce->frameSizePlanned;
+        /* Always use up the whole VBV in this case. */
+        if (m_singleFrameVbv)
+            rce->frameSizePlanned = m_bufferRate;
 
-        q = clipQscale(pic, q);
+        rce->newQScale = q;
+        return q;
+    }
+}
 
-        lastQScaleFor[sliceType] = q;
+void RateControl::rateControlUpdateStats(RateControlEntry* rce)
+{
+    if (!m_param->rc.bStatWrite && !m_param->rc.bStatRead)
+    {
+        if (rce->sliceType == I_SLICE)
+        {
+            /* previous I still had a residual; roll it into the new loan */
+            if (m_partialResidualFrames)
+                rce->rowTotalBits += m_partialResidualCost * m_partialResidualFrames;
 
-        if (curSlice->getPOC() == 0 || (isAbrReset && sliceType == I_SLICE))
-            lastQScaleFor[P_SLICE] = q * fabs(param->rc.ipFactor);
+            m_partialResidualFrames = X265_MIN(s_amortizeFrames, m_param->keyframeMax);
+            m_partialResidualCost = (int)((rce->rowTotalBits * s_amortizeFraction) /m_partialResidualFrames);
+            rce->rowTotalBits -= m_partialResidualCost * m_partialResidualFrames;
+        }
+        else if (m_partialResidualFrames)
+        {
+             rce->rowTotalBits += m_partialResidualCost;
+             m_partialResidualFrames--;
+        }
+    }
+    if (rce->sliceType != B_SLICE)
+        rce->rowCplxrSum = rce->rowTotalBits * x265_qp2qScale(rce->qpaRc) / rce->qRceq;
+    else
+        rce->rowCplxrSum = rce->rowTotalBits * x265_qp2qScale(rce->qpaRc) / (rce->qRceq * fabs(m_param->rc.pbFactor));
 
-        rce->frameSizePlanned = predictSize(&pred[sliceType], q, (double)currentSatd);
+    m_cplxrSum += rce->rowCplxrSum;
+    m_totalBits += rce->rowTotalBits;
 
-        return q;
-    }
+    /* do not allow the next frame to enter rateControlStart() until this
+     * frame has updated its mid-frame statistics */
+    m_startEndOrder.incr();
+
+    if (rce->encodeOrder < m_param->frameNumThreads - 1)
+        m_startEndOrder.incr(); // faked rateControlEnd calls for negative frames
 }
 
 void RateControl::checkAndResetABR(RateControlEntry* rce, bool isFrameDone)
 {
-    double abrBuffer = 2 * param->rc.rateTolerance * bitrate;
+    double abrBuffer = 2 * m_param->rc.rateTolerance * m_bitrate;
 
     // Check if current Slice is a scene cut that follows low detailed/blank frames
     if (rce->lastSatd > 4 * rce->movingAvgSum)
     {
-        if (!isAbrReset && rce->movingAvgSum > 0)
+        if (!m_isAbrReset && rce->movingAvgSum > 0)
         {
             // Reset ABR if prev frames are blank to prevent further sudden overflows/ high bit rate spikes.
-            double underflow = 1.0 + (totalBits - wantedBitsWindow) / abrBuffer;
+            double underflow = 1.0 + (m_totalBits - m_wantedBitsWindow) / abrBuffer;
             if (underflow < 0.9 && !isFrameDone)
             {
-                reInit();
-                shortTermCplxSum = rce->lastSatd / (CLIP_DURATION(frameDuration) / BASE_FRAME_DURATION);
-                shortTermCplxCount = 1;
-                isAbrReset = true;
-                lastAbrResetPoc = rce->poc;
+                init(m_curSlice->m_sps);
+                m_shortTermCplxSum = rce->lastSatd / (CLIP_DURATION(m_frameDuration) / BASE_FRAME_DURATION);
+                m_shortTermCplxCount = 1;
+                m_isAbrReset = true;
+                m_lastAbrResetPoc = rce->poc;
             }
         }
         else
         {
             // Clear flag to reset ABR and continue as usual.
-            isAbrReset = false;
+            m_isAbrReset = false;
         }
     }
 }
 
+void RateControl::hrdFullness(SEIBufferingPeriod *seiBP)
+{
+    const VUI* vui = &m_curSlice->m_sps->vuiParameters;
+    const HRDInfo* hrd = &vui->hrdParameters;
+    int num = 90000;
+    int denom = hrd->bitRateValue << (hrd->bitRateScale + BR_SHIFT);
+    reduceFraction(&num, &denom);
+    int64_t cpbState = (int64_t)m_bufferFillFinal;
+    int64_t cpbSize = (int64_t)hrd->cpbSizeValue << (hrd->cpbSizeScale + CPB_SHIFT);
+
+    if (cpbState < 0 || cpbState > cpbSize)
+    {
+         x265_log(m_param, X265_LOG_WARNING, "CPB %s: %.0lf bits in a %.0lf-bit buffer\n",
+                   cpbState < 0 ? "underflow" : "overflow", (float)cpbState/denom, (float)cpbSize/denom);
+    }
+
+    seiBP->m_initialCpbRemovalDelay = (uint32_t)(num * cpbState + denom) / denom;
+    seiBP->m_initialCpbRemovalDelayOffset = (uint32_t)(num * cpbSize + denom) / denom - seiBP->m_initialCpbRemovalDelay;
+}
+
 void RateControl::updateVbvPlan(Encoder* enc)
 {
-    bufferFill = bufferFillFinal;
+    m_bufferFill = m_bufferFillFinal;
     enc->updateVbvPlan(this);
 }
 
@@ -639,59 +1642,64 @@
     return (p->coeff * var + p->offset) / (q * p->count);
 }
 
-double RateControl::clipQscale(TComPic* pic, double q)
+double RateControl::clipQscale(Frame* pic, double q)
 {
-    double lmin1 = lmin[sliceType];
-    double lmax1 = lmax[sliceType];
-    double q0 = q;
-
     // B-frames are not directly subject to VBV,
-    // since they are controlled by the P-frames' QPs.
-    if (isVbv && currentSatd > 0)
+    // since they are controlled by referenced P-frames' QPs.
+    double q0 = q;
+    if (m_isVbv && m_currentSatd > 0)
     {
-        if (param->lookaheadDepth)
+        if (m_param->lookaheadDepth || m_param->rc.cuTree ||
+            m_param->scenecutThreshold ||
+            (m_param->bFrameAdaptive && m_param->bframes))
         {
-            int terminate = 0;
-
+           /* Lookahead VBV: If lookahead is done, raise the quantizer as necessary
+            * such that no frames in the lookahead overflow and such that the buffer
+            * is in a reasonable state by the end of the lookahead. */
+            int loopTerminate = 0;
             /* Avoid an infinite loop. */
-            for (int iterations = 0; iterations < 1000 && terminate != 3; iterations++)
+            for (int iterations = 0; iterations < 1000 && loopTerminate != 3; iterations++)
             {
                 double frameQ[3];
-                double curBits = predictSize(&pred[sliceType], q, (double)currentSatd);
-                double bufferFillCur = bufferFill - curBits;
+                double curBits;
+                if (m_sliceType == B_SLICE)
+                    curBits = predictSize(&m_predBfromP, q, (double)m_currentSatd);
+                else
+                    curBits = predictSize(&m_pred[m_sliceType], q, (double)m_currentSatd);
+                double bufferFillCur = m_bufferFill - curBits;
                 double targetFill;
                 double totalDuration = 0;
-                frameQ[P_SLICE] = sliceType == I_SLICE ? q * param->rc.ipFactor : q;
-                frameQ[B_SLICE] = frameQ[P_SLICE] * param->rc.pbFactor;
-                frameQ[I_SLICE] = frameQ[P_SLICE] / param->rc.ipFactor;
+                frameQ[P_SLICE] = m_sliceType == I_SLICE ? q * m_param->rc.ipFactor : (m_sliceType == B_SLICE ? q / m_param->rc.pbFactor : q);
+                frameQ[B_SLICE] = frameQ[P_SLICE] * m_param->rc.pbFactor;
+                frameQ[I_SLICE] = frameQ[P_SLICE] / m_param->rc.ipFactor;
                 /* Loop over the planned future frames. */
-                for (int j = 0; bufferFillCur >= 0 && bufferFillCur <= bufferSize; j++)
+                for (int j = 0; bufferFillCur >= 0 && bufferFillCur <= m_bufferSize; j++)
                 {
-                    totalDuration += frameDuration;
-                    bufferFillCur += vbvMaxRate * frameDuration;
+                    totalDuration += m_frameDuration;
+                    bufferFillCur += m_vbvMaxRate * m_frameDuration;
                     int type = pic->m_lowres.plannedType[j];
                     int64_t satd = pic->m_lowres.plannedSatd[j] >> (X265_DEPTH - 8);
                     if (type == X265_TYPE_AUTO)
                         break;
                     type = IS_X265_TYPE_I(type) ? I_SLICE : IS_X265_TYPE_B(type) ? B_SLICE : P_SLICE;
-                    curBits = predictSize(&pred[type], frameQ[type], (double)satd);
+                    curBits = predictSize(&m_pred[type], frameQ[type], (double)satd);
                     bufferFillCur -= curBits;
                 }
 
                 /* Try to get the buffer at least 50% filled, but don't set an impossible goal. */
-                targetFill = X265_MIN(bufferFill + totalDuration * vbvMaxRate * 0.5, bufferSize * 0.5);
+                targetFill = X265_MIN(m_bufferFill + totalDuration * m_vbvMaxRate * 0.5, m_bufferSize * 0.5);
                 if (bufferFillCur < targetFill)
                 {
                     q *= 1.01;
-                    terminate |= 1;
+                    loopTerminate |= 1;
                     continue;
                 }
                 /* Try to get the buffer no more than 80% filled, but don't set an impossible goal. */
-                targetFill = Clip3(bufferSize * 0.8, bufferSize, bufferFill - totalDuration * vbvMaxRate * 0.5);
-                if (vbvMinRate && bufferFillCur > targetFill)
+                targetFill = Clip3(m_bufferSize * 0.8, m_bufferSize, m_bufferFill - totalDuration * m_vbvMaxRate * 0.5);
+                if (m_isCbr && bufferFillCur > targetFill)
                 {
                     q /= 1.01;
-                    terminate |= 2;
+                    loopTerminate |= 2;
                     continue;
                 }
                 break;
@@ -699,75 +1707,87 @@
         }
         else
         {
-            if ((sliceType == P_SLICE ||
-                 (sliceType == I_SLICE && lastNonBPictType == I_SLICE)) &&
-                bufferFill / bufferSize < 0.5)
+            /* Fallback to old purely-reactive algorithm: no lookahead. */
+            if ((m_sliceType == P_SLICE || m_sliceType == B_SLICE ||
+                    (m_sliceType == I_SLICE && m_lastNonBPictType == I_SLICE)) &&
+                m_bufferFill / m_bufferSize < 0.5)
             {
-                q /= Clip3(0.5, 1.0, 2.0 * bufferFill / bufferSize);
+                q /= Clip3(0.5, 1.0, 2.0 * m_bufferFill / m_bufferSize);
             }
-
             // Now a hard threshold to make sure the frame fits in VBV.
             // This one is mostly for I-frames.
-            double bits = predictSize(&pred[sliceType], q, (double)currentSatd);
+            double bits = predictSize(&m_pred[m_sliceType], q, (double)m_currentSatd);
 
             // For small VBVs, allow the frame to use up the entire VBV.
             double maxFillFactor;
-            maxFillFactor = bufferSize >= 5 * bufferRate ? 2 : 1;
+            maxFillFactor = m_bufferSize >= 5 * m_bufferRate ? 2 : 1;
             // For single-frame VBVs, request that the frame use up the entire VBV.
-            double minFillFactor = singleFrameVbv ? 1 : 2;
+            double minFillFactor = m_singleFrameVbv ? 1 : 2;
 
             for (int iterations = 0; iterations < 10; iterations++)
             {
                 double qf = 1.0;
-                if (bits > bufferFill / maxFillFactor)
-                    qf = Clip3(0.2, 1.0, bufferFill / (maxFillFactor * bits));
+                if (bits > m_bufferFill / maxFillFactor)
+                    qf = Clip3(0.2, 1.0, m_bufferFill / (maxFillFactor * bits));
                 q /= qf;
                 bits *= qf;
-                if (bits < bufferRate / minFillFactor)
-                    q *= bits * minFillFactor / bufferRate;
-                bits = predictSize(&pred[sliceType], q, (double)currentSatd);
+                if (bits < m_bufferRate / minFillFactor)
+                    q *= bits * minFillFactor / m_bufferRate;
+                bits = predictSize(&m_pred[m_sliceType], q, (double)m_currentSatd);
             }
 
             q = X265_MAX(q0, q);
         }
+         // Check B-frame complexity, and use up any bits that would
+         // overflow before the next P-frame.
+         if (m_leadingBframes <= 5 && m_sliceType == P_SLICE && !m_singleFrameVbv)
+         {
+             int nb = m_leadingBframes;
+             double bits = predictSize(&m_pred[m_sliceType], q, (double)m_currentSatd);
+             double bbits = predictSize(&m_predBfromP, q * m_param->rc.pbFactor, (double)m_currentSatd);
+             double space;
+             if (bbits > m_bufferRate)
+                 nb = 0;
+             double pbbits = nb * bbits;
+
+             space = m_bufferFill + (1 + nb) * m_bufferRate - m_bufferSize;
+             if (pbbits < space)
+             {
+                 q *= X265_MAX(pbbits / space, bits / (0.5 * m_bufferSize));
+             }
+             q = X265_MAX(q0 / 2, q);
+         }
 
-        // Check B-frame complexity, and use up any bits that would
-        // overflow before the next P-frame.
-        if (sliceType == P_SLICE)
-        {
-            int nb = bframes;
-            double bits = predictSize(&pred[sliceType], q, (double)currentSatd);
-            double bbits = predictSize(&predBfromP, q * param->rc.pbFactor, (double)currentSatd);
-            double space;
-            if (bbits > bufferRate)
-                nb = 0;
-            double pbbits = nb * bbits;
+        if (!m_isCbr)
+            q = X265_MAX(q0, q);
 
-            space = bufferFill + (1 + nb) * bufferRate - bufferSize;
-            if (pbbits < space)
-            {
-                q *= X265_MAX(pbbits / space, bits / (0.5 * bufferSize));
-            }
-            q = X265_MAX(q0 / 2, q);
+        if (m_rateFactorMaxIncrement)
+        {
+            double qpNoVbv = x265_qScale2qp(q0);
+            double qmax = X265_MIN(MAX_MAX_QPSCALE,x265_qp2qScale(qpNoVbv + m_rateFactorMaxIncrement));
+            return Clip3(MIN_QPSCALE, qmax, q);
         }
-        if (!vbvMinRate)
-            q = X265_MAX(q0, q);
     }
-
-    if (lmin1 == lmax1)
-        return lmin1;
-
-    return Clip3(lmin1, lmax1, q);
+    if (m_2pass)
+    {
+        double min = log(MIN_QPSCALE);
+        double max = log(MAX_MAX_QPSCALE);
+        q = (log(q) - min) / (max - min) - 0.5;
+        q = 1.0 / (1.0 + exp(-4 * q));
+        q = q*(max - min) + min;
+        return exp(q);
+    }
+    return Clip3(MIN_QPSCALE, MAX_MAX_QPSCALE, q);
 }
 
-double RateControl::predictRowsSizeSum(TComPic* pic, RateControlEntry* rce, double qpVbv, int32_t & encodedBitsSoFar)
+double RateControl::predictRowsSizeSum(Frame* pic, RateControlEntry* rce, double qpVbv, int32_t & encodedBitsSoFar)
 {
     uint32_t rowSatdCostSoFar = 0, totalSatdBits = 0;
 
     encodedBitsSoFar = 0;
     double qScale = x265_qp2qScale(qpVbv);
-    int picType = pic->getSlice()->getSliceType();
-    TComPic* refPic = pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0);
+    int picType = pic->m_picSym->m_slice->m_sliceType;
+    Frame* refPic = pic->m_picSym->m_slice->m_refPicList[0][0];
     int maxRows = pic->getPicSym()->getFrameHeightInCU();
     for (int row = 0; row < maxRows; row++)
     {
@@ -798,7 +1818,7 @@
             if (picType == I_SLICE || qScale >= refQScale)
             {
                 if (picType == P_SLICE
-                    && refPic->getSlice()->getSliceType() == picType
+                    && refPic->m_picSym->m_slice->m_sliceType == picType
                     && refQScale > 0
                     && refRowSatdCost > 0)
                 {
@@ -825,11 +1845,12 @@
     return totalSatdBits + encodedBitsSoFar;
 }
 
-int RateControl::rowDiagonalVbvRateControl(TComPic* pic, uint32_t row, RateControlEntry* rce, double& qpVbv)
+int RateControl::rowDiagonalVbvRateControl(Frame* pic, uint32_t row, RateControlEntry* rce, double& qpVbv)
 {
     double qScaleVbv = x265_qp2qScale(qpVbv);
     uint64_t rowSatdCost = pic->m_rowDiagSatd[row];
     double encodedBits = pic->m_rowEncodedBits[row];
+
     if (row == 1)
     {
         rowSatdCost += pic->m_rowDiagSatd[0];
@@ -837,16 +1858,15 @@
     }
     rowSatdCost >>= X265_DEPTH - 8;
     updatePredictor(rce->rowPred[0], qScaleVbv, (double)rowSatdCost, encodedBits);
-    if (pic->getSlice()->getSliceType() == P_SLICE)
+    if (pic->m_picSym->m_slice->m_sliceType == P_SLICE)
     {
-        TComPic* refSlice = pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0);
+        Frame* refSlice = pic->m_picSym->m_slice->m_refPicList[0][0];
         if (qpVbv < refSlice->m_rowDiagQp[row])
         {
             uint64_t intraRowSatdCost = pic->m_rowDiagIntraSatd[row];
             if (row == 1)
-            {
                 intraRowSatdCost += pic->m_rowDiagIntraSatd[0];
-            }
+
             updatePredictor(rce->rowPred[1], qScaleVbv, (double)intraRowSatdCost, encodedBits);
         }
     }
@@ -854,11 +1874,16 @@
     int canReencodeRow = 1;
     /* tweak quality based on difference from predicted size */
     double prevRowQp = qpVbv;
-    double qpAbsoluteMax = MAX_MAX_QP;
-    if (rateFactorMaxIncrement)
-        qpAbsoluteMax = X265_MIN(qpAbsoluteMax, rce->qpNoVbv + rateFactorMaxIncrement);
-    double qpMax = X265_MIN(prevRowQp + param->rc.qpStep, qpAbsoluteMax);
-    double qpMin = X265_MAX(prevRowQp - param->rc.qpStep, MIN_QP);
+    double qpAbsoluteMax = QP_MAX_MAX;
+    double qpAbsoluteMin = QP_MIN;
+    if (m_rateFactorMaxIncrement)
+        qpAbsoluteMax = X265_MIN(qpAbsoluteMax, rce->qpNoVbv + m_rateFactorMaxIncrement);
+
+    if (m_rateFactorMaxDecrement)
+        qpAbsoluteMin = X265_MAX(qpAbsoluteMin, rce->qpNoVbv - m_rateFactorMaxDecrement);
+
+    double qpMax = X265_MIN(prevRowQp + m_param->rc.qpStep, qpAbsoluteMax);
+    double qpMin = X265_MAX(prevRowQp - m_param->rc.qpStep, qpAbsoluteMin);
     double stepSize = 0.5;
     double bufferLeftPlanned = rce->bufferFill - rce->frameSizePlanned;
 
@@ -869,13 +1894,13 @@
         /* B-frames shouldn't use lower QP than their reference frames. */
         if (rce->sliceType == B_SLICE)
         {
-            TComPic* refSlice1 = pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0);
-            TComPic* refSlice2 = pic->getSlice()->getRefPic(REF_PIC_LIST_1, 0);
+            Frame* refSlice1 = pic->m_picSym->m_slice->m_refPicList[0][0];
+            Frame* refSlice2 = pic->m_picSym->m_slice->m_refPicList[1][0];
             qpMin = X265_MAX(qpMin, X265_MAX(refSlice1->m_rowDiagQp[row], refSlice2->m_rowDiagQp[row]));
             qpVbv = X265_MAX(qpVbv, qpMin);
         }
         /* More threads means we have to be more cautious in letting ratecontrol use up extra bits. */
-        double rcTol = bufferLeftPlanned / param->frameNumThreads * param->rc.rateTolerance;
+        double rcTol = bufferLeftPlanned / m_param->frameNumThreads * m_param->rc.rateTolerance;
         int32_t encodedBitsSoFar = 0;
         double accFrameBits = predictRowsSizeSum(pic, rce, qpVbv, encodedBitsSoFar);
 
@@ -888,7 +1913,7 @@
         if (rce->sliceType != I_SLICE)
             rcTol *= 0.5;
 
-        if (!vbvMinRate)
+        if (!m_isCbr)
             qpMin = X265_MAX(qpMin, rce->qpNoVbv);
 
         while (qpVbv < qpMax
@@ -901,9 +1926,9 @@
         }
 
         while (qpVbv > qpMin
-               && (qpVbv > pic->m_rowDiagQp[0] || singleFrameVbv)
+               && (qpVbv > pic->m_rowDiagQp[0] || m_singleFrameVbv)
                && ((accFrameBits < rce->frameSizePlanned * 0.8f && qpVbv <= prevRowQp)
-                   || accFrameBits < (rce->bufferFill - bufferSize + bufferRate) * 1.1))
+                   || accFrameBits < (rce->bufferFill - m_bufferSize + m_bufferRate) * 1.1))
         {
             qpVbv -= stepSize;
             accFrameBits = predictRowsSizeSum(pic, rce, qpVbv, encodedBitsSoFar);
@@ -911,7 +1936,7 @@
 
         /* avoid VBV underflow */
         while ((qpVbv < qpAbsoluteMax)
-               && (rce->bufferFill - accFrameBits < bufferRate * maxFrameError))
+               && (rce->bufferFill - accFrameBits < m_bufferRate * maxFrameError))
         {
             qpVbv += stepSize;
             accFrameBits = predictRowsSizeSum(pic, rce, qpVbv, encodedBitsSoFar);
@@ -926,6 +1951,15 @@
             qpVbv = Clip3(prevRowQp + 1.0f, qpMax, (prevRowQp + qpVbv) * 0.5);
             return -1;
         }
+
+        if (m_param->rc.rfConstantMin)
+        {
+            if (qpVbv < qpMin && prevRowQp > qpMin && canReencodeRow)
+            {
+                qpVbv = Clip3(qpMin, prevRowQp, (prevRowQp + qpVbv) * 0.5);
+                return -1;
+            }
+        }
     }
     else
     {
@@ -934,7 +1968,7 @@
 
         /* Last-ditch attempt: if the last row of the frame underflowed the VBV,
          * try again. */
-        if ((rce->frameSizeEstimated > (rce->bufferFill - bufferRate * maxFrameError) &&
+        if ((rce->frameSizeEstimated > (rce->bufferFill - m_bufferRate * maxFrameError) &&
              qpVbv < qpMax && canReencodeRow))
         {
             qpVbv = qpMax;
@@ -949,21 +1983,20 @@
 {
     double q;
 
-    if (param->rc.cuTree)
+    if (m_param->rc.cuTree)
     {
         // Scale and units are obtained from rateNum and rateDenom for videos with fixed frame rates.
-        double timescale = (double)param->fpsDenom / (2 * param->fpsNum);
-        q = pow(BASE_FRAME_DURATION / CLIP_DURATION(2 * timescale), 1 - param->rc.qCompress);
+        double timescale = (double)m_param->fpsDenom / (2 * m_param->fpsNum);
+        q = pow(BASE_FRAME_DURATION / CLIP_DURATION(2 * timescale), 1 - m_param->rc.qCompress);
     }
     else
-        q = pow(rce->blurredComplexity, 1 - param->rc.qCompress);
-
-    // avoid NaN's in the rc_eq
-    if (rce->texBits + rce->mvBits == 0)
-        q = lastQScaleFor[rce->sliceType];
+        q = pow(rce->blurredComplexity, 1 - m_param->rc.qCompress);
+    // avoid NaN's in the Rceq
+    if (rce->coeffBits + rce->mvBits == 0)
+        q = m_lastQScaleFor[rce->sliceType];
     else
     {
-        lastRceq = q;
+        m_lastRceq = q;
         q /= rateFactor;
     }
     return q;
@@ -992,81 +2025,275 @@
 
 void RateControl::updateVbv(int64_t bits, RateControlEntry* rce)
 {
-    if (rce->lastSatd >= ncu)
-        updatePredictor(&pred[rce->sliceType], x265_qp2qScale(rce->qpaRc), (double)rce->lastSatd, (double)bits);
-    if (!isVbv)
+    if (rce->lastSatd >= m_ncu)
+        updatePredictor(&m_pred[rce->sliceType], x265_qp2qScale(rce->qpaRc), (double)rce->lastSatd, (double)bits);
+    if (!m_isVbv)
         return;
 
-    bufferFillFinal -= bits;
+    m_bufferFillFinal -= bits;
 
-    if (bufferFillFinal < 0)
-        x265_log(param, X265_LOG_WARNING, "poc:%d, VBV underflow (%.0f bits)\n", rce->poc, bufferFillFinal);
+    if (m_bufferFillFinal < 0)
+        x265_log(m_param, X265_LOG_WARNING, "poc:%d, VBV underflow (%.0f bits)\n", rce->poc, m_bufferFillFinal);
 
-    bufferFillFinal = X265_MAX(bufferFillFinal, 0);
-    bufferFillFinal += bufferRate;
-    bufferFillFinal = X265_MIN(bufferFillFinal, bufferSize);
+    m_bufferFillFinal = X265_MAX(m_bufferFillFinal, 0);
+    m_bufferFillFinal += m_bufferRate;
+    m_bufferFillFinal = X265_MIN(m_bufferFillFinal, m_bufferSize);
 }
 
 /* After encoding one frame, update rate control state */
-int RateControl::rateControlEnd(TComPic* pic, int64_t bits, RateControlEntry* rce)
+int RateControl::rateControlEnd(Frame* pic, int64_t bits, RateControlEntry* rce, FrameStats* stats)
 {
-    if (isAbr)
+    int orderValue = m_startEndOrder.get();
+    int endOrdinal = (rce->encodeOrder + m_param->frameNumThreads) * 2 - 1;
+    while (orderValue < endOrdinal && !m_bTerminated)
     {
-        if (param->rc.rateControlMode == X265_RC_ABR)
-        {
+        /* no more frames are being encoded, so fake the start event if we would
+         * have blocked on it. Note that this does not enforce rateControlEnd()
+         * ordering during flush, but this has no impact on the outputs */
+        if (m_finalFrameCount && orderValue >= 2 * m_finalFrameCount)
+            break;
+        orderValue = m_startEndOrder.waitForChange(orderValue);
+    }
+
+    int64_t actualBits = bits;
+    Slice *slice = pic->m_picSym->m_slice;
+    if (m_isAbr)
+    {
+        if (m_param->rc.rateControlMode == X265_RC_ABR && !m_param->rc.bStatRead)
             checkAndResetABR(rce, true);
+
+        if (m_param->rc.rateControlMode == X265_RC_CRF)
+        {
+            if (int(pic->m_avgQpRc + 0.5) == slice->m_sliceQp)
+                pic->m_rateFactor = m_rateFactorConstant;
+            else
+            {
+                /* If vbv changed the frame QP recalculate the rate-factor */
+                double baseCplx = m_ncu * (m_param->bframes ? 120 : 80);
+                double mbtree_offset = m_param->rc.cuTree ? (1.0 - m_param->rc.qCompress) * 13.5 : 0;
+                pic->m_rateFactor = pow(baseCplx, 1 - m_qCompress) /
+                    x265_qp2qScale(int(pic->m_avgQpRc + 0.5) + mbtree_offset);
+            }
         }
-        if (!isAbrReset)
+    }
+    if (!m_isAbrReset)
+    {
+        if (m_param->rc.aqMode || m_isVbv)
         {
-            if (param->rc.aqMode || isVbv)
+            if (pic->m_qpaRc)
             {
-                if (pic->m_qpaRc)
-                {
-                    for (uint32_t i = 0; i < pic->getFrameHeightInCU(); i++)
-                        pic->m_avgQpRc += pic->m_qpaRc[i];
+                for (uint32_t i = 0; i < pic->getFrameHeightInCU(); i++)
+                    pic->m_avgQpRc += pic->m_qpaRc[i];
 
-                    pic->m_avgQpRc /= (pic->getFrameHeightInCU() * pic->getFrameWidthInCU());
-                    rce->qpaRc = pic->m_avgQpRc;
-                }
+                pic->m_avgQpRc /= (pic->getFrameHeightInCU() * pic->getFrameWidthInCU());
+                rce->qpaRc = pic->m_avgQpRc;
+                // copy avg RC qp to m_avgQpAq. To print out the correct qp when aq/cutree is disabled.
+                pic->m_avgQpAq = pic->m_avgQpRc;
+            }
 
-                if (pic->m_qpaAq)
-                {
-                    for (uint32_t i = 0; i < pic->getFrameHeightInCU(); i++)
-                        pic->m_avgQpAq += pic->m_qpaAq[i];
+            if (pic->m_qpaAq)
+            {
+                for (uint32_t i = 0; i < pic->getFrameHeightInCU(); i++)
+                    pic->m_avgQpAq += pic->m_qpaAq[i];
 
-                    pic->m_avgQpAq /= (pic->getFrameHeightInCU() * pic->getFrameWidthInCU());
-                }
+                pic->m_avgQpAq /= (pic->getFrameHeightInCU() * pic->getFrameWidthInCU());
             }
-
-            if (rce->sliceType != B_SLICE)
-                /* The factor 1.5 is to tune up the actual bits, otherwise the cplxrSum is scaled too low
-                 * to improve short term compensation for next frame. */
-                cplxrSum += bits * x265_qp2qScale(rce->qpaRc) / rce->qRceq;
-            else
+        }
+    }
+    // Write frame stats into the stats file if 2 pass is enabled.
+    if (m_param->rc.bStatWrite)
+    {
+        char cType = rce->sliceType == I_SLICE ? (rce->poc > 0 && m_param->bOpenGOP ? 'i' : 'I')
+            : rce->sliceType == P_SLICE ? 'P'
+            : IS_REFERENCED(slice) ? 'B' : 'b';
+        if (fprintf(m_statFileOut,
+                    "in:%d out:%d type:%c q:%.2f q-aq:%.2f tex:%d mv:%d misc:%d icu:%.2f pcu:%.2f scu:%.2f ;\n",
+                    rce->poc, rce->encodeOrder,
+                    cType, pic->m_avgQpRc, pic->m_avgQpAq,
+                    stats->coeffBits,
+                    stats->mvBits,
+                    stats->miscBits,
+                    stats->percentIntra * m_ncu,
+                    stats->percentInter * m_ncu,
+                    stats->percentSkip  * m_ncu) < 0)
+            goto writeFailure;
+        /* Don't re-write the data in multi-pass mode. */
+        if (m_param->rc.cuTree && IS_REFERENCED(slice) && !m_param->rc.bStatRead)
+        {
+            uint8_t sliceType = (uint8_t)rce->sliceType;
+            for (int i = 0; i < m_ncu; i++)
+                    m_cuTreeStats.qpBuffer[0][i] = (uint16_t)(pic->m_lowres.qpCuTreeOffset[i] * 256.0);
+            if (fwrite(&sliceType, 1, 1, m_cutreeStatFileOut) < 1)
+                goto writeFailure;
+            if (fwrite(m_cuTreeStats.qpBuffer[0], sizeof(uint16_t), m_ncu, m_cutreeStatFileOut) < (size_t)m_ncu)
+                goto writeFailure;
+        }
+    }
+    if (m_isAbr && !m_isAbrReset)
+    {
+        /* amortize part of each I slice over the next several frames, up to
+         * keyint-max, to avoid over-compensating for the large I slice cost */
+        if (!m_param->rc.bStatWrite && !m_param->rc.bStatRead)
+        {
+            if (rce->sliceType == I_SLICE)
             {
-                /* Depends on the fact that B-frame's QP is an offset from the following P-frame's.
-                 * Not perfectly accurate with B-refs, but good enough. */
-                cplxrSum += bits * x265_qp2qScale(rce->qpaRc) / (rce->qRceq * fabs(param->rc.pbFactor));
+                /* previous I still had a residual; roll it into the new loan */
+                if (m_residualFrames)
+                    bits += m_residualCost * m_residualFrames;
+                m_residualFrames = X265_MIN(s_amortizeFrames, m_param->keyframeMax);
+                m_residualCost = (int)((bits * s_amortizeFraction) / m_residualFrames);
+                bits -= m_residualCost * m_residualFrames;
             }
-            wantedBitsWindow += frameDuration * bitrate;
-            totalBits += bits;
+            else if (m_residualFrames)
+            {
+                bits += m_residualCost;
+                m_residualFrames--;
+            }
+        }
+        if (rce->sliceType != B_SLICE)
+        {
+            /* The factor 1.5 is to tune up the actual bits, otherwise the cplxrSum is scaled too low
+                * to improve short term compensation for next frame. */
+            m_cplxrSum += (bits * x265_qp2qScale(rce->qpaRc) / rce->qRceq) - (rce->rowCplxrSum);
+        }
+        else
+        {
+            /* Depends on the fact that B-frame's QP is an offset from the following P-frame's.
+                * Not perfectly accurate with B-refs, but good enough. */
+            m_cplxrSum += (bits * x265_qp2qScale(rce->qpaRc) / (rce->qRceq * fabs(m_param->rc.pbFactor))) - (rce->rowCplxrSum);
         }
+        m_wantedBitsWindow += m_frameDuration * m_bitrate;
+        m_totalBits += bits - rce->rowTotalBits;
     }
 
-    if (isVbv)
+    if (m_2pass)
+    {
+        m_expectedBitsSum += qScale2bits(rce, x265_qp2qScale(rce->newQp));
+        m_totalBits += bits - rce->rowTotalBits;
+    }
+
+    if (m_isVbv)
     {
         if (rce->sliceType == B_SLICE)
         {
-            bframeBits += (int)bits;
+            m_bframeBits += actualBits;
             if (rce->bLastMiniGopBFrame)
             {
                 if (rce->bframes != 0)
-                    updatePredictor(&predBfromP, x265_qp2qScale(rce->qpaRc), (double)rce->leadingNoBSatd, (double)bframeBits / rce->bframes);
-                bframeBits = 0;
+                    updatePredictor(&m_predBfromP, x265_qp2qScale(rce->qpaRc), (double)rce->leadingNoBSatd, (double)m_bframeBits / rce->bframes);
+                m_bframeBits = 0;
+            }
+        }
+        updateVbv(actualBits, rce);
+
+        if (m_param->bEmitHRDSEI)
+        {
+            const VUI *vui = &pic->m_picSym->m_slice->m_sps->vuiParameters;
+            const HRDInfo *hrd = &vui->hrdParameters;
+            const TimingInfo *time = &vui->timingInfo;
+            if (!pic->m_POC)
+            {
+                // first access unit initializes the HRD
+                rce->hrdTiming->cpbInitialAT = 0;
+                rce->hrdTiming->cpbRemovalTime = m_nominalRemovalTime = (double)m_bufPeriodSEI.m_initialCpbRemovalDelay / 90000;
             }
+            else
+            {
+                rce->hrdTiming->cpbRemovalTime = m_nominalRemovalTime + (double)rce->picTimingSEI->m_auCpbRemovalDelay * time->numUnitsInTick / time->timeScale;
+                double cpbEarliestAT = rce->hrdTiming->cpbRemovalTime - (double)m_bufPeriodSEI.m_initialCpbRemovalDelay / 90000;
+                if (!pic->m_lowres.bKeyframe)
+                    cpbEarliestAT -= (double)m_bufPeriodSEI.m_initialCpbRemovalDelayOffset / 90000;
+
+                rce->hrdTiming->cpbInitialAT = hrd->cbrFlag ? m_prevCpbFinalAT : X265_MAX(m_prevCpbFinalAT, cpbEarliestAT);
+            }
+
+            uint32_t cpbsizeUnscale = hrd->cpbSizeValue << (hrd->cpbSizeScale + CPB_SHIFT);
+            rce->hrdTiming->cpbFinalAT = m_prevCpbFinalAT = rce->hrdTiming->cpbInitialAT + actualBits / cpbsizeUnscale;
+            rce->hrdTiming->dpbOutputTime = (double)rce->picTimingSEI->m_picDpbOutputDelay * time->numUnitsInTick / time->timeScale + rce->hrdTiming->cpbRemovalTime;
         }
-        updateVbv(bits, rce);
-        rce->isActive = false;
     }
+    // Allow rateControlStart of next frame only when rateControlEnd of previous frame is over
+    m_startEndOrder.incr();
+    rce->isActive = false;
     return 0;
+
+writeFailure:
+    x265_log(m_param, X265_LOG_ERROR, "RatecontrolEnd: stats file write failure\n");
+    return 1;
 }
+
+#if defined(_MSC_VER)
+#pragma warning(disable: 4996) // POSIX function names are just fine, thank you
+#endif
+
+/* called when the encoder is flushing, and thus the final frame count is
+ * unambiguously known */
+void RateControl::setFinalFrameCount(int count)
+{
+    m_finalFrameCount = count;
+    /* unblock waiting threads */
+    m_startEndOrder.set(m_startEndOrder.get());
+}
+
+/* called when the encoder is closing, and no more frames will be output.
+ * all blocked functions must finish so the frame encoder threads can be
+ * closed */
+void RateControl::terminate()
+{
+    m_bTerminated = true;
+    /* unblock waiting threads */
+    m_startEndOrder.set(m_startEndOrder.get());
+}
+
+void RateControl::destroy()
+{
+    const char *fileName = m_param->rc.statFileName;
+    if (!fileName)
+        fileName = s_defaultStatFileName;
+
+    if (m_statFileOut)
+    {
+        fclose(m_statFileOut);
+        char *tmpFileName = strcatFilename(fileName, ".temp");
+        int bError = 1;
+        if (tmpFileName)
+        {
+           unlink(fileName);
+           bError = rename(tmpFileName, fileName);
+        }
+        if (bError)
+        {
+            x265_log(m_param, X265_LOG_ERROR, "failed to rename output stats file to \"%s\"\n",
+                     fileName);
+        }
+        X265_FREE(tmpFileName);
+    }
+
+    if (m_cutreeStatFileOut)
+    {
+        fclose(m_cutreeStatFileOut);
+        char *tmpFileName = strcatFilename(fileName, ".cutree.temp");
+        char *newFileName = strcatFilename(fileName, ".cutree");
+        int bError = 1;
+        if (tmpFileName && newFileName)
+        {
+           unlink(newFileName);
+           bError = rename(tmpFileName, newFileName);
+        }
+        if (bError)
+        {
+            x265_log(m_param, X265_LOG_ERROR, "failed to rename cutree output stats file to \"%s\"\n",
+                     newFileName);
+        }
+        X265_FREE(tmpFileName);
+        X265_FREE(newFileName);
+    }
+
+    if (m_cutreeStatFileIn)
+        fclose(m_cutreeStatFileIn);
+
+    X265_FREE(m_rce2Pass);
+    for (int i = 0; i < 2; i++)
+        X265_FREE(m_cuTreeStats.qpBuffer[i]);
+}
+

x265-hg20140324.tar.bz2/source/encoder/ratecontrol.h -> x265-hg20140928.tar.bz2/source/encoder/ratecontrol.h Changed

@@ -20,21 +20,19 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef X265_RATECONTROL_H
 #define X265_RATECONTROL_H
 
-#include "TLibCommon/CommonDef.h"
-
 namespace x265 {
 // encoder namespace
 
-struct Lookahead;
 class Encoder;
-class TComPic;
-
+class Frame;
+struct SPS;
+class SEIBufferingPeriod;
 #define BASE_FRAME_DURATION 0.04
 
 /* Arbitrary limitations as a sanity check. */
@@ -43,6 +41,23 @@
 
 #define CLIP_DURATION(f) Clip3(MIN_FRAME_DURATION, MAX_FRAME_DURATION, f)
 
+/* Current frame stats for 2 pass */
+struct FrameStats
+{
+    int         mvBits;    /* MV bits (MV+Ref+Block Type) */
+    int         coeffBits; /* Texture bits (DCT coefs) */
+    int         miscBits;
+
+    int         iCuCnt;
+    int         pCuCnt;
+    int         skipCuCnt;
+    
+    /* CU type counts stored as percentage */
+    double      percentIntra;
+    double      percentInter;
+    double      percentSkip;
+};
+
 struct Predictor
 {
     double coeff;
@@ -51,103 +66,197 @@
     double offset;
 };
 
+struct HRDTiming
+{
+    double cpbInitialAT;
+    double cpbFinalAT;
+    double dpbOutputTime;
+    double cpbRemovalTime;
+};
+
 struct RateControlEntry
 {
-    int64_t texBits;  /* Required in 2-pass rate control */
     int64_t lastSatd; /* Contains the picture cost of the previous frame, required for resetAbr and VBV */
-
     int sliceType;
-    int mvBits;
     int bframes;
     int poc;
+    int encodeOrder;
     int64_t leadingNoBSatd;
     bool bLastMiniGopBFrame;
     double blurredComplexity;
     double qpaRc;
+    double qpAq;
     double qRceq;
     double frameSizePlanned;  /* frame Size decided by RateCotrol before encoding the frame */
     double bufferRate;
     double movingAvgSum;
+    double   rowCplxrSum;
+    int64_t  rowTotalBits;  /* update cplxrsum and totalbits at the end of 2 rows */
     double qpNoVbv;
     double bufferFill;
+    double frameDuration;
+    double clippedDuration;
     Predictor rowPreds[3][2];
     Predictor* rowPred[2];
     double frameSizeEstimated;  /* hold frameSize, updated from cu level vbv rc */
     bool isActive;
+    SEIPictureTiming *picTimingSEI;
+    HRDTiming        *hrdTiming;
+    /* Required in 2-pass rate control */
+    double iCuCount;
+    double pCuCount;
+    double skipCuCount;
+    bool keptAsRef;
+    double expectedVbv;
+    double qScale;
+    double newQScale;
+    double newQp;
+    int mvBits;
+    int miscBits;
+    int coeffBits;
+    uint64_t expectedBits; /* total expected bits up to the current frame (current one excluded) */
 };
 
-struct RateControl
+class RateControl
 {
-    TComSlice *curSlice;      /* all info about the current frame */
-    x265_param* param;
-    SliceType sliceType;      /* Current frame type */
-    int ncu;                  /* number of CUs in a frame */
-    int keyFrameInterval;     /* TODO: need to initialize in init */
-    int qp;                   /* updated qp for current frame */
-
-    double frameDuration;     /* current frame duration in seconds */
-    double bitrate;
-    double rateFactorConstant;
-    bool   isAbr;
-    double bufferSize;
-    double bufferFillFinal;  /* real buffer as of the last finished frame */
-    double bufferFill;       /* planned buffer, if all in-progress frames hit their bit budget */
-    double bufferRate;       /* # of bits added to buffer_fill after each frame */
-    double vbvMaxRate;       /* in kbps */
-    double vbvMinRate;       /* in kbps */
-    bool singleFrameVbv;
-    double rateFactorMaxIncrement; /* Don't allow RF above (CRF + this value). */
-    bool isVbv;
-    Predictor pred[5];
-    Predictor predBfromP;
-    int bframes;
-    int bframeBits;
-    bool isAbrReset;
-    int lastAbrResetPoc;
-    int64_t currentSatd;
-    int    qpConstant[3];
-    double cplxrSum;          /* sum of bits*qscale/rceq */
-    double wantedBitsWindow;  /* target bitrate * window */
-    double ipOffset;
-    double pbOffset;
-    int lastNonBPictType;
-    int64_t leadingNoBSatd;
-    double accumPQp;          /* for determining I-frame quant */
-    double accumPNorm;
-    double lastQScaleFor[3];  /* last qscale for a specific pict type, used for max_diff & ipb factor stuff */
-    double lstep;
-    double lmin[3];           /* min qscale by frame type */
-    double lmax[3];
-    double shortTermCplxSum;
-    double shortTermCplxCount;
-    int64_t totalBits;        /* totalbits used for already encoded frames */
-    double lastRceq;
-    int framesDone;           /* framesDone keeps track of # of frames passed through RateCotrol already */
-    double qCompress;
-    double qpNoVbv;             /* QP for the current frame if 1-pass VBV was disabled. */
-    RateControl(Encoder * _cfg);
+public:
+
+    x265_param* m_param;
+    Slice*      m_curSlice;      /* all info about the current frame */
+    SliceType   m_sliceType;     /* Current frame type */
+    int         m_ncu;           /* number of CUs in a frame */
+    int         m_qp;            /* updated qp for current frame */
+
+    bool   m_isAbr;
+    bool   m_isVbv;
+    bool   m_isCbr;
+    bool   m_singleFrameVbv;
+
+    bool   m_isAbrReset;
+    int    m_lastAbrResetPoc;
+
+    double m_frameDuration;     /* current frame duration in seconds */
+    double m_bitrate;
+    double m_rateFactorConstant;
+    double m_bufferSize;
+    double m_bufferFillFinal;  /* real buffer as of the last finished frame */
+    double m_bufferFill;       /* planned buffer, if all in-progress frames hit their bit budget */
+    double m_bufferRate;       /* # of bits added to buffer_fill after each frame */
+    double m_vbvMaxRate;       /* in kbps */
+    double m_rateFactorMaxIncrement; /* Don't allow RF above (CRF + this value). */
+    double m_rateFactorMaxDecrement; /* don't allow RF below (this value). */
+
+    Predictor m_pred[5];
+    Predictor m_predBfromP;
+
+    int       m_leadingBframes;
+    int64_t   m_bframeBits;
+    int64_t   m_currentSatd;
+    int       m_qpConstant[3];
+    double    m_ipOffset;
+    double    m_pbOffset;
+
+    int      m_lastNonBPictType;
+    int64_t  m_leadingNoBSatd;
+
+    double   m_cplxrSum;          /* sum of bits*qscale/rceq */
+    double   m_wantedBitsWindow;  /* target bitrate * window */
+    double   m_accumPQp;          /* for determining I-frame quant */
+    double   m_accumPNorm;
+    double   m_lastQScaleFor[3];  /* last qscale for a specific pict type, used for max_diff & ipb factor stuff */
+    double   m_lstep;
+    double   m_shortTermCplxSum;
+    double   m_shortTermCplxCount;
+    double   m_lastRceq;
+    double   m_qCompress;
+    int64_t  m_totalBits;        /* total bits used for already encoded frames */
+    int      m_framesDone;       /* # of frames passed through RateCotrol already */
+    double   m_fps;
+
+    /* a common variable on which rateControlStart, rateControlEnd and rateControUpdateStats waits to
+     * sync the calls to these functions. For example
+     * -F2:
+     * rceStart  10
+     * rceUpdate 10
+     * rceEnd    9
+     * rceStart  11
+     * rceUpdate 11
+     * rceEnd    10
+     * rceStart  12
+     * rceUpdate 12
+     * rceEnd    11 */
+    ThreadSafeInteger m_startEndOrder;
+    int      m_finalFrameCount;   /* set when encoder begins flushing */
+    bool     m_bTerminated;       /* set true when encoder is closing */
+
+    /* hrd stuff */
+    SEIBufferingPeriod m_bufPeriodSEI;
+    double   m_nominalRemovalTime;
+    double   m_prevCpbFinalAT;
+
+    /* 2 pass */
+    bool     m_2pass;
+    FILE*    m_statFileOut;
+    FILE*    m_cutreeStatFileOut;
+    FILE*    m_cutreeStatFileIn;
+    int      m_numEntries;
+    RateControlEntry *m_rce2Pass;
+    double   m_lastAccumPNorm;
+    int64_t  m_predictedBits;
+    double   m_expectedBitsSum;   /* sum of qscale2bits after rceq, ratefactor, and overflow, only includes finished frames */
+    struct
+    {
+        uint16_t *qpBuffer[2]; /* Global buffers for converting MB-tree quantizer data. */
+        int qpBufPos;          /* In order to handle pyramid reordering, QP buffer acts as a stack.
+                                * This value is the current position (0 or 1). */
+    } m_cuTreeStats;
+
+    RateControl(x265_param *p);
+    void setFinalFrameCount(int count);
+    void terminate();          /* un-block all waiting functions so encoder may close */
+    void destroy();
 
     // to be called for each frame to process RateControl and set QP
-    void rateControlStart(TComPic* pic, Lookahead *, RateControlEntry* rce, Encoder* enc);
-    void calcAdaptiveQuantFrame(TComPic *pic);
-    int rateControlEnd(TComPic* pic, int64_t bits, RateControlEntry* rce);
-    int rowDiagonalVbvRateControl(TComPic* pic, uint32_t row, RateControlEntry* rce, double& qpVbv);
+    int rateControlStart(Frame* pic, RateControlEntry* rce, Encoder* enc);
+    void calcAdaptiveQuantFrame(Frame *pic);
+    void rateControlUpdateStats(RateControlEntry* rce);
+    int rateControlEnd(Frame* pic, int64_t bits, RateControlEntry* rce, FrameStats* stats);
+    int rowDiagonalVbvRateControl(Frame* pic, uint32_t row, RateControlEntry* rce, double& qpVbv);
+    void hrdFullness(SEIBufferingPeriod* sei);
+    bool init(const SPS* sps);
+    void initHRD(SPS* sps);
+    int rateControlSliceType(int frameNum);
+    bool cuTreeReadFor2Pass(Frame* frame);
 
 protected:
 
-    void reInit();
+    static const double s_amortizeFraction;
+    static const int    s_amortizeFrames;
+    static const char  *s_defaultStatFileName;
+
+    int m_residualFrames;
+    int m_partialResidualFrames;
+    int m_residualCost;
+    int m_partialResidualCost;
+
     double getQScale(RateControlEntry *rce, double rateFactor);
-    double rateEstimateQscale(TComPic* pic, RateControlEntry *rce); // main logic for calculating QP based on ABR
+    double rateEstimateQscale(Frame* pic, RateControlEntry *rce); // main logic for calculating QP based on ABR
     void accumPQpUpdate();
-    uint32_t acEnergyCu(TComPic* pic, uint32_t block_x, uint32_t block_y);
+    uint32_t acEnergyCu(Frame* pic, uint32_t block_x, uint32_t block_y);
 
     void updateVbv(int64_t bits, RateControlEntry* rce);
     void updatePredictor(Predictor *p, double q, double var, double bits);
-    double clipQscale(TComPic* pic, double q);
+    double clipQscale(Frame* pic, double q);
     void updateVbvPlan(Encoder* enc);
     double predictSize(Predictor *p, double q, double var);
     void checkAndResetABR(RateControlEntry* rce, bool isFrameDone);
-    double predictRowsSizeSum(TComPic* pic, RateControlEntry* rce, double qpm, int32_t& encodedBits);
+    double predictRowsSizeSum(Frame* pic, RateControlEntry* rce, double qpm, int32_t& encodedBits);
+    bool initPass2();
+    double getDiffLimitedQScale(RateControlEntry *rce, double q);
+    double countExpectedBits();
+    bool vbv2Pass(uint64_t allAvailableBits);
+    bool findUnderflow(double *fills, int *t0, int *t1, int over);
+    bool fixUnderflow(int t0, int t1, double adjustment, double qscaleMin, double qscaleMax);
 };
 }
 #endif // ifndef X265_RATECONTROL_H

x265-hg20140928.tar.bz2/source/encoder/rdcost.h Added

@@ -0,0 +1,100 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_RDCOST_H
+#define X265_RDCOST_H
+
+#include "common.h"
+
+namespace x265 {
+// private namespace
+
+class RDCost
+{
+public:
+
+    /* all weights and factors stored as FIX8 */
+    uint64_t  m_lambda2;
+    uint64_t  m_lambda;
+    uint64_t  m_cbDistortionWeight;
+    uint64_t  m_crDistortionWeight;
+    uint32_t  m_psyRd;
+
+    void setPsyRdScale(double scale)                { m_psyRd = (uint32_t)floor(256.0 * scale * 0.33); }
+    void setCbDistortionWeight(uint16_t weightFix8) { m_cbDistortionWeight = weightFix8; }
+    void setCrDistortionWeight(uint16_t weightFix8) { m_crDistortionWeight = weightFix8; }
+
+    void setLambda(double lambda2, double lambda)
+    {
+        m_lambda2 = (uint64_t)floor(256.0 * lambda2);
+        m_lambda = (uint64_t)floor(256.0 * lambda);
+    }
+
+    inline uint64_t calcRdCost(uint32_t distortion, uint32_t bits)
+    {
+        X265_CHECK(bits <= (UINT64_MAX - 128) / m_lambda2,
+                   "calcRdCost wrap detected dist: %d, bits %d, lambda: %d\n", distortion, bits, (int)m_lambda2);
+        return distortion + ((bits * m_lambda2 + 128) >> 8);
+    }
+
+    /* return the difference in energy between the source block and the recon block */
+    inline int psyCost(int size, pixel *source, intptr_t sstride, pixel *recon, intptr_t rstride)
+    {
+        return primitives.psy_cost[size](source, sstride, recon, rstride);
+    }
+
+    /* return the RD cost of this prediction, including the effect of psy-rd */
+    inline uint64_t calcPsyRdCost(uint32_t distortion, uint32_t bits, uint32_t psycost)
+    {
+        return distortion + ((m_lambda * m_psyRd * psycost) >> 16) + ((bits * m_lambda2) >> 8);
+    }
+
+    inline uint64_t calcRdSADCost(uint32_t sadCost, uint32_t bits)
+    {
+        X265_CHECK(bits <= (UINT64_MAX - 128) / m_lambda,
+                   "calcRdSADCost wrap detected dist: %d, bits %d, lambda: "X265_LL"\n", sadCost, bits, m_lambda);
+        return sadCost + ((bits * m_lambda + 128) >> 8);
+    }
+
+    inline uint32_t scaleChromaDistCb(uint32_t dist)
+    {
+        X265_CHECK(dist <= (UINT64_MAX - 128) / m_cbDistortionWeight,
+                   "scaleChromaDistCb wrap detected dist: %d, lambda: "X265_LL"\n", dist, m_cbDistortionWeight);
+        return (uint32_t)(((dist * m_cbDistortionWeight) + 128) >> 8);
+    }
+
+    inline uint32_t scaleChromaDistCr(uint32_t dist)
+    {
+        X265_CHECK(dist <= (UINT64_MAX - 128) / m_crDistortionWeight,
+                   "scaleChromaDistCr wrap detected dist: %d, lambda: "X265_LL"\n", dist, m_crDistortionWeight);
+        return (uint32_t)(((dist * m_crDistortionWeight) + 128) >> 8);
+    }
+
+    inline uint32_t getCost(uint32_t bits)
+    {
+        return (uint32_t)((bits * m_lambda + 128) >> 8);
+    }
+};
+}
+
+#endif // ifndef X265_TCOMRDCOST_H

x265-hg20140324.tar.bz2/source/encoder/reference.cpp -> x265-hg20140928.tar.bz2/source/encoder/reference.cpp Changed

@@ -19,15 +19,15 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TypeDef.h"
-#include "TLibCommon/TComPicYuv.h"
-#include "TLibCommon/TComSlice.h"
+#include "common.h"
 #include "primitives.h"
+#include "slice.h"
 #include "reference.h"
-#include "common.h"
+#include "TLibCommon/TypeDef.h"
+#include "TLibCommon/TComPicYuv.h"
 
 using namespace x265;
 
@@ -36,14 +36,14 @@
     m_weightBuffer = NULL;
 }
 
-int MotionReference::init(TComPicYuv* pic, wpScalingParam *w)
+int MotionReference::init(TComPicYuv* pic, WeightParam *w)
 {
     m_reconPic = pic;
     lumaStride = pic->getStride();
     intptr_t startpad = pic->m_lumaMarginY * lumaStride + pic->m_lumaMarginX;
 
     /* directly reference the pre-extended integer pel plane */
-    fpelPlane = pic->m_picBufY + startpad;
+    fpelPlane = pic->m_picBuf[0] + startpad;
     isWeighted = false;
 
     if (w)

x265-hg20140324.tar.bz2/source/encoder/reference.h -> x265-hg20140928.tar.bz2/source/encoder/reference.h Changed

x265-hg20140928.tar.bz2/source/encoder/sao.cpp Added

@@ -0,0 +1,2531 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *          Min Chen <chenm003@163.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "common.h"
+#include "sao.h"
+
+namespace {
+
+#if HIGH_BIT_DEPTH
+inline double roundIDBI2(double x)
+{
+    return ((x) > 0) ? (int)(((int)(x) + (1 << (X265_DEPTH - 8 - 1))) / (1 << (X265_DEPTH - 8))) :
+                       ((int)(((int)(x) - (1 << (X265_DEPTH - 8 - 1))) / (1 << (X265_DEPTH - 8))));
+}
+#endif
+
+/* rounding with IBDI */
+inline double roundIDBI(double x)
+{
+#if HIGH_BIT_DEPTH
+    return X265_DEPTH > 8 ? roundIDBI2(x) : ((x) >= 0 ? ((int)((x) + 0.5)) : ((int)((x) - 0.5)));
+#else
+    return (x) >= 0 ? ((int)((x) + 0.5)) : ((int)((x) - 0.5));
+#endif
+}
+
+/* get the sign of input variable (TODO: this is a dup, make common) */
+inline int signOf(int x)
+{
+    return (x >> 31) | ((int)((((uint32_t)-x)) >> 31));
+}
+
+int convertLevelRowCol2Idx(int level, int row, int col)
+{
+    if (!level)
+        return 0;
+    else if (level == 1)
+        return 1 + row * 2 + col;
+    else if (level == 2)
+        return 5 + row * 4 + col;
+    else if (level == 3)
+        return 21 + row * 8 + col;
+    else // (level == 4)
+        return 85 + row * 16 + col;
+}
+
+} // end anonymous namespace
+
+
+namespace x265 {
+
+const int SAO::s_numCulPartsLevel[5] =
+{
+    1,   // level 0
+    5,   // level 1
+    21,  // level 2
+    85,  // level 3
+    341, // level 4
+};
+
+const uint32_t SAO::s_eoTable[NUM_EDGETYPE] =
+{
+    1, // 0
+    2, // 1
+    0, // 2
+    3, // 3
+    4  // 4
+};
+
+SAO::SAO()
+{
+    m_count = NULL;
+    m_offset = NULL;
+    m_offsetOrg = NULL;
+    m_countPreDblk = NULL;
+    m_offsetOrgPreDblk = NULL;
+    m_rate = NULL;
+    m_dist = NULL;
+    m_cost = NULL;
+    m_costPartBest = NULL;
+    m_distOrg = NULL;
+    m_typePartBest = NULL;
+    m_refDepth = 0;
+    m_lumaLambda = 0;
+    m_chromaLambda = 0;
+    m_param = NULL;
+    m_numTotalParts = 0;
+    m_clipTable = NULL;
+    m_clipTableBase = NULL;
+    m_offsetBo = NULL;
+    m_tmpU1[0] = NULL;
+    m_tmpU1[1] = NULL;
+    m_tmpU1[2] = NULL;
+    m_tmpU2[0] = NULL;
+    m_tmpU2[1] = NULL;
+    m_tmpU2[2] = NULL;
+    m_tmpL1 = NULL;
+    m_tmpL2 = NULL;
+
+    m_depthSaoRate[0][0] = 0;
+    m_depthSaoRate[0][1] = 0;
+    m_depthSaoRate[0][2] = 0;
+    m_depthSaoRate[0][3] = 0;
+    m_depthSaoRate[1][0] = 0;
+    m_depthSaoRate[1][1] = 0;
+    m_depthSaoRate[1][2] = 0;
+    m_depthSaoRate[1][3] = 0;
+}
+
+bool SAO::create(x265_param *param)
+{
+    m_param = param;
+    m_hChromaShift = CHROMA_H_SHIFT(param->internalCsp);
+    m_vChromaShift = CHROMA_V_SHIFT(param->internalCsp);
+
+    m_numCuInWidth =  (m_param->sourceWidth + g_maxCUSize - 1) / g_maxCUSize;
+    m_numCuInHeight = (m_param->sourceHeight + g_maxCUSize - 1) / g_maxCUSize;
+
+    int maxSplitLevelHeight = (int)(logf((float)m_numCuInHeight) / logf(2.0));
+    int maxSplitLevelWidth  = (int)(logf((float)m_numCuInWidth) / logf(2.0));
+
+    m_maxSplitLevel = maxSplitLevelHeight < maxSplitLevelWidth ? maxSplitLevelHeight : maxSplitLevelWidth;
+    m_maxSplitLevel = X265_MIN(m_maxSplitLevel, SAO_MAX_DEPTH);
+
+    /* various structures are overloaded to store per component data.
+     * m_numTotalParts must allow for sufficient storage in any allocated arrays */
+    m_numTotalParts = X265_MAX(3, s_numCulPartsLevel[m_maxSplitLevel]);
+
+    const pixel maxY = (1 << X265_DEPTH) - 1;
+    const pixel rangeExt = maxY >> 1;
+    int numLcu = m_numCuInWidth * m_numCuInHeight;
+
+    CHECKED_MALLOC(m_clipTableBase,  pixel, maxY + 2 * rangeExt);
+    CHECKED_MALLOC(m_offsetBo,       pixel, maxY + 2 * rangeExt);
+
+    CHECKED_MALLOC(m_tmpL1, pixel, g_maxCUSize + 1);
+    CHECKED_MALLOC(m_tmpL2, pixel, g_maxCUSize + 1);
+
+    for (int i = 0; i < 3; i++)
+    {
+        CHECKED_MALLOC(m_tmpU1[i], pixel, m_param->sourceWidth);
+        CHECKED_MALLOC(m_tmpU2[i], pixel, m_param->sourceWidth);
+    }
+
+    CHECKED_MALLOC(m_distOrg, int64_t, m_numTotalParts);
+    CHECKED_MALLOC(m_costPartBest, double, m_numTotalParts);
+    CHECKED_MALLOC(m_typePartBest, int, m_numTotalParts);
+
+    CHECKED_MALLOC(m_rate, PerType, m_numTotalParts);
+    CHECKED_MALLOC(m_dist, PerType, m_numTotalParts);
+    CHECKED_MALLOC(m_cost, PerTypeD, m_numTotalParts);
+
+    CHECKED_MALLOC(m_count, PerClass, m_numTotalParts);
+    CHECKED_MALLOC(m_offset, PerClass, m_numTotalParts);
+    CHECKED_MALLOC(m_offsetOrg, PerClass, m_numTotalParts);
+
+    CHECKED_MALLOC(m_countPreDblk, PerPlane, numLcu);
+    CHECKED_MALLOC(m_offsetOrgPreDblk, PerPlane, numLcu);
+
+    m_clipTable = &(m_clipTableBase[rangeExt]);
+
+    for (int i = 0; i < rangeExt; i++)
+        m_clipTableBase[i] = 0;
+
+    for (int i = 0; i < maxY; i++)
+        m_clipTable[i] = (pixel)i;
+
+    for (int i = maxY; i < maxY + rangeExt; i++)
+        m_clipTable[i] = maxY;
+
+    return true;
+
+fail:
+    return false;
+}
+
+void SAO::destroy()
+{
+    X265_FREE(m_clipTableBase);
+    X265_FREE(m_offsetBo);
+
+    X265_FREE(m_tmpL1);
+    X265_FREE(m_tmpL2);
+
+    for (int i = 0; i < 3; i++)
+    {
+        X265_FREE(m_tmpU1[i]);
+        X265_FREE(m_tmpU2[i]);
+    }
+
+    X265_FREE(m_distOrg);
+    X265_FREE(m_costPartBest);
+    X265_FREE(m_typePartBest);
+    X265_FREE(m_rate);
+    X265_FREE(m_dist);
+    X265_FREE(m_cost);
+    X265_FREE(m_count);
+    X265_FREE(m_offset);
+    X265_FREE(m_offsetOrg);
+    X265_FREE(m_countPreDblk);
+    X265_FREE(m_offsetOrgPreDblk);
+}
+
+/* allocate memory for SAO parameters */
+void SAO::allocSaoParam(SAOParam *saoParam) const
+{
+    saoParam->maxSplitLevel = m_maxSplitLevel;
+    saoParam->numCuInWidth  = m_numCuInWidth;
+    saoParam->numCuInHeight = m_numCuInHeight;
+
+    saoParam->saoPart[0] = new SAOQTPart[s_numCulPartsLevel[saoParam->maxSplitLevel]];
+    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 0);
+
+    saoParam->saoPart[1] = new SAOQTPart[s_numCulPartsLevel[saoParam->maxSplitLevel]];
+    saoParam->saoPart[2] = new SAOQTPart[s_numCulPartsLevel[saoParam->maxSplitLevel]];
+    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 1);
+    initSAOParam(saoParam, 0, 0, 0, -1, 0, m_numCuInWidth - 1,  0, m_numCuInHeight - 1, 2);
+
+    saoParam->saoLcuParam[0] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
+    saoParam->saoLcuParam[1] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
+    saoParam->saoLcuParam[2] = new SaoLcuParam[m_numCuInHeight * m_numCuInWidth];
+}
+
+/* recursively initialize SAO parameters (only once) */
+void SAO::initSAOParam(SAOParam *saoParam, int partLevel, int partRow, int partCol, int parentPartIdx, int startCUX, int endCUX, int startCUY, int endCUY, int plane) const
+{
+    int partIdx = convertLevelRowCol2Idx(partLevel, partRow, partCol);
+
+    SAOQTPart* saoPart = &(saoParam->saoPart[plane][partIdx]);
+
+    saoPart->partIdx   = partIdx;
+    saoPart->partLevel = partLevel;
+    saoPart->partRow   = partRow;
+    saoPart->partCol   = partCol;
+
+    saoPart->startCUX  = startCUX;
+    saoPart->endCUX    = endCUX;
+    saoPart->startCUY  = startCUY;
+    saoPart->endCUY    = endCUY;
+
+    saoPart->upPartIdx = parentPartIdx;
+    saoPart->bestType  = -1;
+
+    saoPart->subTypeIdx = 0;
+
+    for (int j = 0; j < SAO_NUM_OFFSET; j++)
+        saoPart->offset[j] = 0;
+
+    if (saoPart->partLevel < m_maxSplitLevel)
+    {
+        int downLevel    = (partLevel + 1);
+        int downRowStart = (partRow << 1);
+        int downColStart = (partCol << 1);
+
+        int numCUWidth  = endCUX - startCUX + 1;
+        int numCUHeight = endCUY - startCUY + 1;
+        int numCULeft   = (numCUWidth  >> 1);
+        int numCUTop    = (numCUHeight >> 1);
+
+        int downStartCUX = startCUX;
+        int downEndCUX  = downStartCUX + numCULeft - 1;
+        int downStartCUY = startCUY;
+        int downEndCUY  = downStartCUY + numCUTop  - 1;
+        int downRowIdx = downRowStart + 0;
+        int downColIdx = downColStart + 0;
+
+        saoPart->downPartsIdx[0] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
+
+        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, plane);
+
+        downStartCUX = startCUX + numCULeft;
+        downEndCUX   = endCUX;
+        downStartCUY = startCUY;
+        downEndCUY   = downStartCUY + numCUTop - 1;
+        downRowIdx  = downRowStart + 0;
+        downColIdx  = downColStart + 1;
+
+        saoPart->downPartsIdx[1] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
+
+        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx,  downStartCUX, downEndCUX, downStartCUY, downEndCUY, plane);
+
+        downStartCUX = startCUX;
+        downEndCUX   = downStartCUX + numCULeft - 1;
+        downStartCUY = startCUY + numCUTop;
+        downEndCUY   = endCUY;
+        downRowIdx  = downRowStart + 1;
+        downColIdx  = downColStart + 0;
+
+        saoPart->downPartsIdx[2] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
+
+        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, plane);
+
+        downStartCUX = startCUX + numCULeft;
+        downEndCUX   = endCUX;
+        downStartCUY = startCUY + numCUTop;
+        downEndCUY   = endCUY;
+        downRowIdx  = downRowStart + 1;
+        downColIdx  = downColStart + 1;
+
+        saoPart->downPartsIdx[3] = convertLevelRowCol2Idx(downLevel, downRowIdx, downColIdx);
+
+        initSAOParam(saoParam, downLevel, downRowIdx, downColIdx, partIdx, downStartCUX, downEndCUX, downStartCUY, downEndCUY, plane);
+    }
+    else
+    {
+        saoPart->downPartsIdx[0] = saoPart->downPartsIdx[1] = saoPart->downPartsIdx[2] = saoPart->downPartsIdx[3] = -1;
+    }
+}
+
+/* reset SAO parameters once per frame */
+void SAO::resetSAOParam(SAOParam *saoParam)
+{
+    int numComponet = 3;
+
+    for (int c = 0; c < numComponet; c++)
+    {
+        if (c < 2)
+            saoParam->bSaoFlag[c] = false;
+
+        for (int i = 0; i < s_numCulPartsLevel[m_maxSplitLevel]; i++)
+        {
+            saoParam->saoPart[c][i].bestType     = -1;
+            saoParam->saoPart[c][i].bSplit       = false;
+            saoParam->saoPart[c][i].bProcessed   = false;
+            saoParam->saoPart[c][i].minCost      = MAX_DOUBLE;
+            saoParam->saoPart[c][i].minDist      = MAX_INT;
+            saoParam->saoPart[c][i].minRate      = MAX_INT;
+            saoParam->saoPart[c][i].subTypeIdx   = 0;
+            for (int j = 0; j < SAO_NUM_OFFSET; j++)
+            {
+                saoParam->saoPart[c][i].offset[j] = 0;
+                saoParam->saoPart[c][i].offset[j] = 0;
+                saoParam->saoPart[c][i].offset[j] = 0;
+            }
+        }
+
+        saoParam->oneUnitFlag[0] = 0;
+        saoParam->oneUnitFlag[1] = 0;
+        saoParam->oneUnitFlag[2] = 0;
+        resetLcuPart(saoParam->saoLcuParam[0]);
+        resetLcuPart(saoParam->saoLcuParam[1]);
+        resetLcuPart(saoParam->saoLcuParam[2]);
+    }
+}
+
+void SAO::startSlice(Frame *pic, Entropy& initState, int qp)
+{
+    Slice* slice = pic->m_picSym->m_slice;
+
+    int qpCb = Clip3(0, QP_MAX_MAX, qp + slice->m_pps->chromaCbQpOffset);
+    m_lumaLambda = x265_lambda2_tab[qp];
+    m_chromaLambda = x265_lambda2_tab[qpCb]; // Use Cb QP for SAO chroma
+    m_pic = pic;
+
+    switch (slice->m_sliceType)
+    {
+    case I_SLICE:
+        m_refDepth = 0;
+        break;
+    case P_SLICE:
+        m_refDepth = 1;
+        break;
+    case B_SLICE:
+        m_refDepth = 2 + !IS_REFERENCED(slice);
+        break;
+    }
+
+    resetStats();
+
+    m_entropyCoder.load(initState);
+    m_rdEntropyCoders[0][CI_NEXT_BEST].load(initState);
+    m_rdEntropyCoders[0][CI_CURR_BEST].load(initState);
+
+    SAOParam* saoParam = pic->getPicSym()->m_saoParam;
+    if (!saoParam)
+    {
+        saoParam = new SAOParam;
+        allocSaoParam(saoParam);
+        pic->getPicSym()->m_saoParam = saoParam;
+    }
+
+    resetSAOParam(saoParam);
+    rdoSaoUnitRowInit(saoParam);
+
+    // NOTE: Disable SAO automatic turn-off when frame parallelism is
+    // enabled for output exact independent of frame thread count
+    if (m_param->frameNumThreads > 1)
+    {
+        saoParam->bSaoFlag[0] = true;
+        saoParam->bSaoFlag[1] = true;
+    }
+}
+
+/* sample adaptive offset process for one LCU crossing LCU boundary */
+void SAO::processSaoCu(int addr, int saoType, int plane)
+{
+    int x, y;
+    TComDataCU *tmpCu = m_pic->getCU(addr);
+    pixel* rec;
+    int stride;
+    int lcuWidth;
+    int lcuHeight;
+    int rpelx;
+    int bpely;
+    int picWidthTmp;
+    int picHeightTmp;
+    int startX;
+    int startY;
+    int endX;
+    int endY;
+    pixel* tmpL;
+    pixel* tmpU;
+    uint32_t lpelx = tmpCu->getCUPelX();
+    uint32_t tpely = tmpCu->getCUPelY();
+    bool isLuma = !plane;
+
+    picWidthTmp  = isLuma ? m_param->sourceWidth  : m_param->sourceWidth  >> m_hChromaShift;
+    picHeightTmp = isLuma ? m_param->sourceHeight : m_param->sourceHeight >> m_vChromaShift;
+    lcuWidth     = isLuma ? g_maxCUSize : g_maxCUSize >> m_hChromaShift;
+    lcuHeight    = isLuma ? g_maxCUSize : g_maxCUSize >> m_vChromaShift;
+    lpelx        = isLuma ? lpelx       : lpelx       >> m_hChromaShift;
+    tpely        = isLuma ? tpely       : tpely       >> m_vChromaShift;
+
+    rpelx        = lpelx + lcuWidth;
+    bpely        = tpely + lcuHeight;
+    rpelx        = rpelx > picWidthTmp  ? picWidthTmp  : rpelx;
+    bpely        = bpely > picHeightTmp ? picHeightTmp : bpely;
+    lcuWidth     = rpelx - lpelx;
+    lcuHeight    = bpely - tpely;
+
+    if (!tmpCu->m_pic)
+        return;
+
+    if (plane)
+    {
+        rec    = m_pic->getPicYuvRec()->getChromaAddr(plane, addr);
+        stride = m_pic->getCStride();
+    }
+    else
+    {
+        rec    = m_pic->getPicYuvRec()->getLumaAddr(addr);
+        stride = m_pic->getStride();
+    }
+
+    int32_t _upBuff1[MAX_CU_SIZE + 2], *upBuff1 = _upBuff1 + 1;
+    int32_t _upBufft[MAX_CU_SIZE + 2], *upBufft = _upBufft + 1;
+
+//   if (iSaoType!=SAO_BO_0 || iSaoType!=SAO_BO_1)
+    {
+        int cuHeightTmp = isLuma ? g_maxCUSize : (g_maxCUSize  >> m_vChromaShift);
+        pixel* recR = &rec[isLuma ? (g_maxCUSize - 1) : ((g_maxCUSize >> m_hChromaShift) - 1)];
+        for (int i = 0; i < cuHeightTmp + 1; i++)
+        {
+            m_tmpL2[i] = *recR;
+            recR += stride;
+        }
+
+        tmpL = m_tmpL1;
+        tmpU = &(m_tmpU1[plane][lpelx]);
+    }
+
+    switch (saoType)
+    {
+    case SAO_EO_0: // dir: -
+    {
+        pixel firstPxl = 0, lastPxl = 0;
+        startX = !lpelx;
+        endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+        if (lcuWidth & 15)
+        {
+            for (y = 0; y < lcuHeight; y++)
+            {
+                int signLeft = signOf(rec[startX] - tmpL[y]);
+                for (x = startX; x < endX; x++)
+                {
+                    int signRight = signOf(rec[x] - rec[x + 1]);
+                    int edgeType = signRight + signLeft + 2;
+                    signLeft = -signRight;
+
+                    rec[x] = m_clipTable[rec[x] + m_offsetEo[edgeType]];
+                }
+
+                rec += stride;
+            }
+        }
+        else
+        {
+            for (y = 0; y < lcuHeight; y++)
+            {
+                int signLeft = signOf(rec[startX] - tmpL[y]);
+
+                if (!lpelx)
+                    firstPxl = rec[0];
+
+                if (rpelx == picWidthTmp)
+                    lastPxl = rec[lcuWidth - 1];
+
+                primitives.saoCuOrgE0(rec, m_offsetEo, lcuWidth, (int8_t)signLeft);
+
+                if (!lpelx)
+                    rec[0] = firstPxl;
+
+                if (rpelx == picWidthTmp)
+                    rec[lcuWidth - 1] = lastPxl;
+
+                rec += stride;
+            }
+        }
+        break;
+    }
+    case SAO_EO_1: // dir: |
+    {
+        startY = !tpely;
+        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+        if (!tpely)
+            rec += stride;
+
+        for (x = 0; x < lcuWidth; x++)
+            upBuff1[x] = signOf(rec[x] - tmpU[x]);
+
+        for (y = startY; y < endY; y++)
+        {
+            for (x = 0; x < lcuWidth; x++)
+            {
+                int signDown = signOf(rec[x] - rec[x + stride]);
+                int edgeType = signDown + upBuff1[x] + 2;
+                upBuff1[x] = -signDown;
+
+                rec[x] = m_clipTable[rec[x] + m_offsetEo[edgeType]];
+            }
+
+            rec += stride;
+        }
+
+        break;
+    }
+    case SAO_EO_2: // dir: 135
+    {
+        startX = !lpelx;
+        endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+
+        startY = !tpely;
+        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+
+        if (!tpely)
+            rec += stride;
+
+        for (x = startX; x < endX; x++)
+            upBuff1[x] = signOf(rec[x] - tmpU[x - 1]);
+
+        for (y = startY; y < endY; y++)
+        {
+            int signDown2 = signOf(rec[stride + startX] - tmpL[y]);
+            for (x = startX; x < endX; x++)
+            {
+                int signDown1 = signOf(rec[x] - rec[x + stride + 1]);
+                int edgeType  = signDown1 + upBuff1[x] + 2;
+                upBufft[x + 1] = -signDown1;
+                rec[x] = m_clipTable[rec[x] + m_offsetEo[edgeType]];
+            }
+
+            upBufft[startX] = signDown2;
+
+            std::swap(upBuff1, upBufft);
+
+            rec += stride;
+        }
+
+        break;
+    }
+    case SAO_EO_3: // dir: 45
+    {
+        startX = (lpelx == 0) ? 1 : 0;
+        endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+
+        startY = (tpely == 0) ? 1 : 0;
+        endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+
+        if (startY == 1)
+            rec += stride;
+
+        for (x = startX - 1; x < endX; x++)
+            upBuff1[x] = signOf(rec[x] - tmpU[x + 1]);
+
+        for (y = startY; y < endY; y++)
+        {
+            x = startX;
+            int signDown1 = signOf(rec[x] - tmpL[y + 1]);
+            int edgeType  = signDown1 + upBuff1[x] + 2;
+            upBuff1[x - 1] = -signDown1;
+            rec[x] = m_clipTable[rec[x] + m_offsetEo[edgeType]];
+            for (x = startX + 1; x < endX; x++)
+            {
+                signDown1 = signOf(rec[x] - rec[x + stride - 1]);
+                edgeType  = signDown1 + upBuff1[x] + 2;
+                upBuff1[x - 1] = -signDown1;
+                rec[x] = m_clipTable[rec[x] + m_offsetEo[edgeType]];
+            }
+
+            upBuff1[endX - 1] = signOf(rec[endX - 1 + stride] - rec[endX]);
+
+            rec += stride;
+        }
+
+        break;
+    }
+    case SAO_BO:
+    {
+        const pixel* offsetBo = m_offsetBo;
+
+        for (y = 0; y < lcuHeight; y++)
+        {
+            for (x = 0; x < lcuWidth; x++)
+                rec[x] = offsetBo[rec[x]];
+
+            rec += stride;
+        }
+
+        break;
+    }
+    default: break;
+    }
+
+//   if (iSaoType!=SAO_BO_0 || iSaoType!=SAO_BO_1)
+    std::swap(m_tmpL1, m_tmpL2);
+}
+
+/* Process SAO all units */
+void SAO::processSaoUnitAll(SaoLcuParam* saoLcuParam, bool oneUnitFlag, int plane)
+{
+    pixel *rec;
+    int picWidthTmp;
+
+    if (plane)
+    {
+        rec         = m_pic->getPicYuvRec()->getChromaAddr(plane);
+        picWidthTmp = m_param->sourceWidth >> m_hChromaShift;
+    }
+    else
+    {
+        rec         = m_pic->getPicYuvRec()->getLumaAddr();
+        picWidthTmp = m_param->sourceWidth;
+    }
+
+    memcpy(m_tmpU1[plane], rec, sizeof(pixel) * picWidthTmp);
+
+    int frameWidthInCU = m_pic->getFrameWidthInCU();
+    int frameHeightInCU = m_pic->getFrameHeightInCU();
+    int stride;
+    bool isChroma = !!plane;
+    uint32_t cuHeightTmp = isChroma ? (g_maxCUSize >> m_vChromaShift) : g_maxCUSize;
+
+    const int boShift = X265_DEPTH - SAO_BO_BITS;
+
+    for (int idxY = 0; idxY < frameHeightInCU; idxY++)
+    {
+        int addr = idxY * frameWidthInCU;
+        if (plane == 0)
+        {
+            rec = m_pic->getPicYuvRec()->getLumaAddr(addr);
+            stride = m_pic->getStride();
+            picWidthTmp = m_param->sourceWidth;
+        }
+        else
+        {
+            rec = m_pic->getPicYuvRec()->getChromaAddr(plane, addr);
+            stride = m_pic->getCStride();
+            picWidthTmp = m_param->sourceWidth >> m_hChromaShift;
+        }
+        for (uint32_t i = 0; i < cuHeightTmp + 1; i++)
+        {
+            m_tmpL1[i] = rec[0];
+            rec += stride;
+        }
+
+        rec -= (stride << 1);
+
+        memcpy(m_tmpU2[plane], rec, sizeof(pixel) * picWidthTmp);
+
+        for (int idxX = 0; idxX < frameWidthInCU; idxX++)
+        {
+            addr = idxY * frameWidthInCU + idxX;
+
+            int typeIdx;
+            bool mergeLeftFlag;
+
+            if (oneUnitFlag)
+            {
+                typeIdx = saoLcuParam[0].typeIdx;
+                mergeLeftFlag = (addr == 0) ? 0 : 1;
+            }
+            else
+            {
+                typeIdx = saoLcuParam[addr].typeIdx;
+                mergeLeftFlag = saoLcuParam[addr].mergeLeftFlag;
+            }
+            if (typeIdx >= 0)
+            {
+                if (!mergeLeftFlag)
+                {
+                    if (typeIdx == SAO_BO)
+                    {
+                        pixel* offsetBo = m_offsetBo;
+                        int offset[SAO_NUM_BO_CLASSES];
+                        memset(offset, 0, sizeof(offset));
+
+                        for (int i = 0; i < SAO_NUM_OFFSET; i++)
+                            offset[((saoLcuParam[addr].subTypeIdx + i) & (SAO_NUM_BO_CLASSES - 1))] = saoLcuParam[addr].offset[i] << SAO_BIT_INC;
+
+                        for (int i = 0; i < (1 << X265_DEPTH); i++)
+                            offsetBo[i] = m_clipTable[i + offset[i >> boShift]];
+                    }
+                    else // if (typeIdx == SAO_EO_0 || typeIdx == SAO_EO_1 || typeIdx == SAO_EO_2 || typeIdx == SAO_EO_3)
+                    {
+                        int offset[NUM_EDGETYPE];
+                        offset[0] = 0;
+                        for (int i = 0; i < SAO_NUM_OFFSET; i++)
+                            offset[i + 1] = saoLcuParam[addr].offset[i] << SAO_BIT_INC;
+
+                        for (int edgeType = 0; edgeType < NUM_EDGETYPE; edgeType++)
+                            m_offsetEo[edgeType] = (int8_t)offset[s_eoTable[edgeType]];
+                    }
+                }
+                processSaoCu(addr, typeIdx, plane);
+            }
+            else
+            {
+                if (idxX != (frameWidthInCU - 1))
+                {
+                    if (isChroma)
+                    {
+                        rec = m_pic->getPicYuvRec()->getChromaAddr(plane, addr);
+                        stride = m_pic->getCStride();
+                    }
+                    else
+                    {
+                        rec = m_pic->getPicYuvRec()->getLumaAddr(addr);
+                        stride = m_pic->getStride();
+                    }
+
+                    int widthShift = isChroma ? (g_maxCUSize >> m_hChromaShift) : g_maxCUSize;
+                    for (uint32_t i = 0; i < cuHeightTmp + 1; i++)
+                    {
+                        m_tmpL1[i] = rec[widthShift - 1];
+                        rec += stride;
+                    }
+                }
+            }
+        }
+
+        std::swap(m_tmpU1[plane], m_tmpU2[plane]);
+    }
+}
+
+/* Process SAO all units */
+void SAO::processSaoUnitRow(SaoLcuParam* saoLcuParam, int idxY, int plane)
+{
+    pixel *rec;
+    int picWidthTmp;
+
+    if (plane)
+    {
+        rec         = m_pic->getPicYuvRec()->getChromaAddr(plane);
+        picWidthTmp = m_param->sourceWidth >> m_hChromaShift;
+    }
+    else
+    {
+        rec         = m_pic->getPicYuvRec()->getLumaAddr();
+        picWidthTmp = m_param->sourceWidth;
+    }
+
+    if (!idxY)
+        memcpy(m_tmpU1[plane], rec, sizeof(pixel) * picWidthTmp);
+
+    int frameWidthInCU = m_pic->getFrameWidthInCU();
+    int stride;
+    bool isChroma = !!plane;
+
+    const int boShift = X265_DEPTH - SAO_BO_BITS;
+
+    int addr = idxY * frameWidthInCU;
+    if (isChroma)
+    {
+        rec = m_pic->getPicYuvRec()->getChromaAddr(plane, addr);
+        stride = m_pic->getCStride();
+        picWidthTmp = m_param->sourceWidth >> m_hChromaShift;
+    }
+    else
+    {
+        rec = m_pic->getPicYuvRec()->getLumaAddr(addr);
+        stride = m_pic->getStride();
+        picWidthTmp = m_param->sourceWidth;
+    }
+    int maxCUHeight = isChroma ? (g_maxCUSize >> m_vChromaShift) : g_maxCUSize;
+    for (int i = 0; i < maxCUHeight + 1; i++)
+    {
+        m_tmpL1[i] = rec[0];
+        rec += stride;
+    }
+
+    rec -= (stride << 1);
+
+    memcpy(m_tmpU2[plane], rec, sizeof(pixel) * picWidthTmp);
+
+    for (int idxX = 0; idxX < frameWidthInCU; idxX++)
+    {
+        addr = idxY * frameWidthInCU + idxX;
+
+        int typeIdx = saoLcuParam[addr].typeIdx;
+        bool mergeLeftFlag = saoLcuParam[addr].mergeLeftFlag;
+
+        if (typeIdx >= 0)
+        {
+            if (!mergeLeftFlag)
+            {
+                if (typeIdx == SAO_BO)
+                {
+                    pixel* offsetBo = m_offsetBo;
+                    int offset[SAO_NUM_BO_CLASSES];
+                    memset(offset, 0, sizeof(offset));
+
+                    for (int i = 0; i < SAO_NUM_OFFSET; i++)
+                        offset[((saoLcuParam[addr].subTypeIdx + i) & (SAO_NUM_BO_CLASSES - 1))] = saoLcuParam[addr].offset[i] << SAO_BIT_INC;
+
+                    for (int i = 0; i < (1 << X265_DEPTH); i++)
+                        offsetBo[i] = m_clipTable[i + offset[i >> boShift]];
+                }
+                else // if (typeIdx == SAO_EO_0 || typeIdx == SAO_EO_1 || typeIdx == SAO_EO_2 || typeIdx == SAO_EO_3)
+                {
+                    int offset[NUM_EDGETYPE];
+                    offset[0] = 0;
+                    for (int i = 0; i < SAO_NUM_OFFSET; i++)
+                        offset[i + 1] = saoLcuParam[addr].offset[i] << SAO_BIT_INC;
+
+                    for (int edgeType = 0; edgeType < NUM_EDGETYPE; edgeType++)
+                        m_offsetEo[edgeType] = (int8_t)offset[s_eoTable[edgeType]];
+                }
+            }
+            processSaoCu(addr, typeIdx, plane);
+        }
+        else
+        {
+            if (idxX != (frameWidthInCU - 1))
+            {
+                if (isChroma)
+                {
+                    rec = m_pic->getPicYuvRec()->getChromaAddr(plane, addr);
+                    stride = m_pic->getCStride();
+                }
+                else
+                {
+                    rec = m_pic->getPicYuvRec()->getLumaAddr(addr);
+                    stride = m_pic->getStride();
+                }
+
+                int widthShift = isChroma ? (g_maxCUSize >> m_hChromaShift) : g_maxCUSize;
+                for (int i = 0; i < maxCUHeight + 1; i++)
+                {
+                    m_tmpL1[i] = rec[widthShift - 1];
+                    rec += stride;
+                }
+            }
+        }
+    }
+
+    std::swap(m_tmpU1[plane], m_tmpU2[plane]);
+}
+
+void SAO::resetLcuPart(SaoLcuParam* saoLcuParam)
+{
+    for (int i = 0; i < m_numCuInWidth * m_numCuInHeight; i++)
+    {
+        saoLcuParam[i].mergeUpFlag   =  1;
+        saoLcuParam[i].mergeLeftFlag =  0;
+        saoLcuParam[i].partIdx       =  0;
+        saoLcuParam[i].typeIdx       = -1;
+        saoLcuParam[i].subTypeIdx    =  0;
+        for (int j = 0; j < SAO_NUM_OFFSET; j++)
+            saoLcuParam[i].offset[j] = 0;
+    }
+}
+
+void SAO::resetSaoUnit(SaoLcuParam* saoUnit)
+{
+    saoUnit->mergeUpFlag   = 0;
+    saoUnit->mergeLeftFlag = 0;
+    saoUnit->partIdx       = 0;
+    saoUnit->partIdxTmp    = 0;
+    saoUnit->typeIdx       = -1;
+    saoUnit->subTypeIdx    = 0;
+
+    for (int i = 0; i < SAO_NUM_OFFSET; i++)
+        saoUnit->offset[i] = 0;
+}
+
+void SAO::copySaoUnit(SaoLcuParam* saoUnitDst, SaoLcuParam* saoUnitSrc)
+{
+    saoUnitDst->mergeLeftFlag = saoUnitSrc->mergeLeftFlag;
+    saoUnitDst->mergeUpFlag   = saoUnitSrc->mergeUpFlag;
+    saoUnitDst->typeIdx       = saoUnitSrc->typeIdx;
+
+    saoUnitDst->subTypeIdx  = saoUnitSrc->subTypeIdx;
+    for (int i = 0; i < SAO_NUM_OFFSET; i++)
+        saoUnitDst->offset[i] = saoUnitSrc->offset[i];
+}
+
+/* convert QP part to SAO unit */
+void SAO::convertQT2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int plane)
+{
+    SAOQTPart* saoPart = &(saoParam->saoPart[plane][partIdx]);
+
+    if (!saoPart->bSplit)
+    {
+        convertOnePart2SaoUnit(saoParam, partIdx, plane);
+        return;
+    }
+
+    if (saoPart->partLevel < m_maxSplitLevel)
+    {
+        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[0], plane);
+        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[1], plane);
+        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[2], plane);
+        convertQT2SaoUnit(saoParam, saoPart->downPartsIdx[3], plane);
+    }
+}
+
+/* convert one SAO part to SAO unit */
+void SAO::convertOnePart2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int plane)
+{
+    int frameWidthInCU = m_pic->getFrameWidthInCU();
+    SAOQTPart* saoQTPart = saoParam->saoPart[plane];
+    SaoLcuParam* saoLcuParam = saoParam->saoLcuParam[plane];
+
+    for (int idxY = saoQTPart[partIdx].startCUY; idxY <= saoQTPart[partIdx].endCUY; idxY++)
+    {
+        for (int idxX = saoQTPart[partIdx].startCUX; idxX <= saoQTPart[partIdx].endCUX; idxX++)
+        {
+            int addr = idxY * frameWidthInCU + idxX;
+            saoLcuParam[addr].partIdxTmp = (int)partIdx;
+            saoLcuParam[addr].typeIdx    = saoQTPart[partIdx].bestType;
+            saoLcuParam[addr].subTypeIdx = saoQTPart[partIdx].subTypeIdx;
+            if (saoLcuParam[addr].typeIdx >= 0)
+            {
+                for (int j = 0; j < SAO_NUM_OFFSET; j++)
+                    saoLcuParam[addr].offset[j] = saoQTPart[partIdx].offset[j];
+            }
+            else
+            {
+                saoLcuParam[addr].subTypeIdx = saoQTPart[partIdx].subTypeIdx;
+                for (int j = 0; j < SAO_NUM_OFFSET; j++)
+                    saoLcuParam[addr].offset[j] = 0;
+            }
+        }
+    }
+}
+
+/* process SAO for one partition */
+void SAO::rdoSaoOnePart(SAOQTPart *psQTPart, int partIdx, int plane)
+{
+    SAOQTPart* onePart = &(psQTPart[partIdx]);
+
+    int64_t estDist;
+
+    m_distOrg[partIdx] = 0;
+
+    int    bestClassTableBo = 0;
+    int    currentDistortionTableBo[MAX_NUM_SAO_CLASS];
+    double currentRdCostTableBo[MAX_NUM_SAO_CLASS];
+    double bestRDCostTableBo = MAX_DOUBLE;
+
+    int allowMergeLeft;
+    int allowMergeUp;
+    SaoLcuParam saoLcuParamRdo;
+
+    for (int typeIdx = -1; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
+    {
+        m_entropyCoder.load(m_rdEntropyCoders[onePart->partLevel][CI_CURR_BEST]);
+        m_entropyCoder.resetBits();
+
+        if (typeIdx >= 0)
+        {
+            estDist = estSaoTypeDist(partIdx, typeIdx, 0, m_lumaLambda, currentDistortionTableBo, currentRdCostTableBo);
+            if (typeIdx == SAO_BO)
+            {
+                // Estimate Best Position
+                for (int i = 0; i < SAO_NUM_BO_CLASSES - SAO_BO_LEN + 1; i++)
+                {
+                    double currentRDCost = 0.0;
+                    for (int j = i; j < i + SAO_BO_LEN; j++)
+                        currentRDCost += currentRdCostTableBo[j];
+
+                    if (currentRDCost < bestRDCostTableBo)
+                    {
+                        bestRDCostTableBo = currentRDCost;
+                        bestClassTableBo  = i;
+                    }
+                }
+
+                // Recode all offsets
+                for (int classIdx = bestClassTableBo; classIdx < bestClassTableBo + SAO_BO_LEN; classIdx++)
+                    estDist += currentDistortionTableBo[classIdx];
+            }
+
+            for (int ry = onePart->startCUY; ry <= onePart->endCUY; ry++)
+            {
+                for (int rx = onePart->startCUX; rx <= onePart->endCUX; rx++)
+                {
+                    // get bits for typeIdx = -1
+                    allowMergeLeft = 1;
+                    allowMergeUp   = 1;
+
+                    // reset
+                    resetSaoUnit(&saoLcuParamRdo);
+
+                    // set merge flag
+                    saoLcuParamRdo.mergeUpFlag   = 1;
+                    saoLcuParamRdo.mergeLeftFlag = 1;
+
+                    if (ry == onePart->startCUY)
+                        saoLcuParamRdo.mergeUpFlag = 0;
+
+                    if (rx == onePart->startCUX)
+                        saoLcuParamRdo.mergeLeftFlag = 0;
+
+                    // set type and offsets
+                    saoLcuParamRdo.typeIdx = typeIdx;
+                    saoLcuParamRdo.subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo : 0;
+                    for (int classIdx = 0; classIdx < SAO_NUM_OFFSET; classIdx++)
+                        saoLcuParamRdo.offset[classIdx] = (int)m_offset[partIdx][typeIdx][classIdx + saoLcuParamRdo.subTypeIdx + 1];
+
+                    m_entropyCoder.codeSaoUnitInterleaving(plane, 1, rx, ry, &saoLcuParamRdo, 1, 1, allowMergeLeft, allowMergeUp);
+                }
+            }
+
+            m_dist[partIdx][typeIdx] = estDist;
+            m_rate[partIdx][typeIdx] = m_entropyCoder.getNumberOfWrittenBits();
+
+            m_cost[partIdx][typeIdx] = (double)((double)m_dist[partIdx][typeIdx] + m_lumaLambda * (double)m_rate[partIdx][typeIdx]);
+
+            if (m_cost[partIdx][typeIdx] < m_costPartBest[partIdx])
+            {
+                m_distOrg[partIdx] = 0;
+                m_costPartBest[partIdx] = m_cost[partIdx][typeIdx];
+                m_typePartBest[partIdx] = typeIdx;
+                m_entropyCoder.store(m_rdEntropyCoders[onePart->partLevel][CI_TEMP_BEST]);
+            }
+        }
+        else
+        {
+            for (int ry = onePart->startCUY; ry <= onePart->endCUY; ry++)
+            {
+                for (int rx = onePart->startCUX; rx <= onePart->endCUX; rx++)
+                {
+                    // get bits for iTypeIdx = -1
+                    allowMergeLeft = 1;
+                    allowMergeUp   = 1;
+
+                    // reset
+                    resetSaoUnit(&saoLcuParamRdo);
+
+                    // set merge flag
+                    saoLcuParamRdo.mergeUpFlag   = 1;
+                    saoLcuParamRdo.mergeLeftFlag = 1;
+
+                    if (ry == onePart->startCUY)
+                        saoLcuParamRdo.mergeUpFlag = 0;
+
+                    if (rx == onePart->startCUX)
+                        saoLcuParamRdo.mergeLeftFlag = 0;
+
+                    m_entropyCoder.codeSaoUnitInterleaving(plane, 1, rx, ry,  &saoLcuParamRdo, 1,  1,  allowMergeLeft, allowMergeUp);
+                }
+            }
+            if (m_distOrg[partIdx] < m_costPartBest[partIdx])
+            {
+                m_costPartBest[partIdx] = (double)m_distOrg[partIdx] + m_entropyCoder.getNumberOfWrittenBits() * m_lumaLambda;
+                m_typePartBest[partIdx] = -1;
+                m_entropyCoder.store(m_rdEntropyCoders[onePart->partLevel][CI_TEMP_BEST]);
+            }
+        }
+    }
+
+    onePart->bProcessed = true;
+    onePart->bSplit    = false;
+    onePart->minDist   =       m_typePartBest[partIdx] >= 0 ? m_dist[partIdx][m_typePartBest[partIdx]] : m_distOrg[partIdx];
+    onePart->minRate   = (int)(m_typePartBest[partIdx] >= 0 ? m_rate[partIdx][m_typePartBest[partIdx]] : 0);
+    onePart->minCost   = onePart->minDist + m_lumaLambda * onePart->minRate;
+    onePart->bestType  = m_typePartBest[partIdx];
+
+    if (onePart->bestType != -1)
+    {
+        int minIndex = 0;
+        if (onePart->bestType == SAO_BO)
+        {
+            onePart->subTypeIdx = bestClassTableBo;
+            minIndex = onePart->subTypeIdx;
+        }
+        for (int i = 0; i < SAO_NUM_OFFSET; i++)
+            onePart->offset[i] = (int)m_offset[partIdx][onePart->bestType][minIndex + i + 1];
+    }
+}
+
+/* Run partition tree disable */
+void SAO::disablePartTree(SAOQTPart *psQTPart, int partIdx)
+{
+    SAOQTPart* pOnePart = &(psQTPart[partIdx]);
+
+    pOnePart->bSplit   = false;
+    pOnePart->bestType = -1;
+
+    if (pOnePart->partLevel < (int)m_maxSplitLevel)
+    {
+        for (int i = 0; i < SAOQTPart::NUM_DOWN_PART; i++)
+            disablePartTree(psQTPart, pOnePart->downPartsIdx[i]);
+    }
+}
+
+/* Run quadtree decision function */
+void SAO::runQuadTreeDecision(SAOQTPart *qtPart, int partIdx, double &costFinal, int maxLevel, int plane)
+{
+    SAOQTPart* onePart = &(qtPart[partIdx]);
+
+    uint32_t nextDepth = onePart->partLevel + 1;
+
+    if (!partIdx)
+        costFinal = 0;
+
+    // SAO for this part
+    if (!onePart->bProcessed)
+        rdoSaoOnePart(qtPart, partIdx, plane);
+
+    // SAO for sub 4 parts
+    if (onePart->partLevel < maxLevel)
+    {
+        double costNotSplit = m_lumaLambda + onePart->minCost;
+        double costSplit    = m_lumaLambda;
+
+        for (int i = 0; i < SAOQTPart::NUM_DOWN_PART; i++)
+        {
+            if (i) //initialize RD with previous depth buffer
+                m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+            else
+                m_rdEntropyCoders[nextDepth][CI_CURR_BEST].load(m_rdEntropyCoders[onePart->partLevel][CI_CURR_BEST]);
+
+            runQuadTreeDecision(qtPart, onePart->downPartsIdx[i], costFinal, maxLevel, plane);
+            costSplit += costFinal;
+            m_rdEntropyCoders[nextDepth][CI_NEXT_BEST].load(m_rdEntropyCoders[nextDepth][CI_TEMP_BEST]);
+        }
+
+        if (costSplit < costNotSplit)
+        {
+            costFinal = costSplit;
+            onePart->bSplit   = true;
+            onePart->bestType = -1;
+            m_rdEntropyCoders[onePart->partLevel][CI_NEXT_BEST].load(m_rdEntropyCoders[nextDepth][CI_NEXT_BEST]);
+        }
+        else
+        {
+            costFinal = costNotSplit;
+            onePart->bSplit = false;
+            for (int i = 0; i < SAOQTPart::NUM_DOWN_PART; i++)
+                disablePartTree(qtPart, onePart->downPartsIdx[i]);
+
+            m_rdEntropyCoders[onePart->partLevel][CI_NEXT_BEST].load(m_rdEntropyCoders[onePart->partLevel][CI_TEMP_BEST]);
+        }
+    }
+    else
+        costFinal = onePart->minCost;
+}
+
+/* Calculate SAO statistics for current LCU without non-crossing slice */
+void SAO::calcSaoStatsCu(int addr, int partIdx, int plane)
+{
+    int x, y;
+    TComDataCU *cu = m_pic->getCU(addr);
+
+    pixel* fenc;
+    pixel* recon;
+    int stride;
+    int lcuHeight;
+    int lcuWidth;
+    uint32_t rpelx;
+    uint32_t bpely;
+    uint32_t picWidthTmp;
+    uint32_t picHeightTmp;
+    int64_t* stats;
+    int64_t* counts;
+    int startX;
+    int startY;
+    int endX;
+    int endY;
+    uint32_t lpelx = cu->getCUPelX();
+    uint32_t tpely = cu->getCUPelY();
+
+    int isLuma = !plane;
+    int isChroma = !!plane;
+    int numSkipLine = isChroma ? 4 - (2 * m_vChromaShift) : 4;
+
+    if (!m_param->saoLcuBasedOptimization)
+        numSkipLine = 0;
+
+    int numSkipLineRight = isChroma ? 5 - (2 * m_hChromaShift) : 5;
+
+    if (!m_param->saoLcuBasedOptimization)
+        numSkipLineRight = 0;
+
+    picWidthTmp  = isLuma ? m_param->sourceWidth  : m_param->sourceWidth  >> m_hChromaShift;
+    picHeightTmp = isLuma ? m_param->sourceHeight : m_param->sourceHeight >> m_vChromaShift;
+    lcuWidth     = isLuma ? g_maxCUSize : g_maxCUSize >> m_hChromaShift;
+    lcuHeight    = isLuma ? g_maxCUSize : g_maxCUSize >> m_vChromaShift;
+    lpelx        = isLuma ? lpelx       : lpelx       >> m_hChromaShift;
+    tpely        = isLuma ? tpely       : tpely       >> m_vChromaShift;
+
+    rpelx     = lpelx + lcuWidth;
+    bpely     = tpely + lcuHeight;
+    rpelx     = rpelx > picWidthTmp  ? picWidthTmp  : rpelx;
+    bpely     = bpely > picHeightTmp ? picHeightTmp : bpely;
+    lcuWidth  = rpelx - lpelx;
+    lcuHeight = bpely - tpely;
+    stride    =  (plane == 0) ? m_pic->getStride() : m_pic->getCStride();
+
+    //if(iSaoType == BO_0 || iSaoType == BO_1)
+    {
+        const int boShift = X265_DEPTH - SAO_BO_BITS;
+
+        if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+        {
+            numSkipLine      = isChroma ? 3 - (2 * m_vChromaShift) : 3;
+            numSkipLineRight = isChroma ? 4 - (2 * m_hChromaShift) : 4;
+        }
+        stats = m_offsetOrg[partIdx][SAO_BO];
+        counts = m_count[partIdx][SAO_BO];
+
+        fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+        recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+        endX = (rpelx == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
+        endY = (bpely == picHeightTmp) ? lcuHeight : lcuHeight - numSkipLine;
+        for (y = 0; y < endY; y++)
+        {
+            for (x = 0; x < endX; x++)
+            {
+                int classIdx = 1 + (recon[x] >> boShift);
+                stats[classIdx] += (fenc[x] - recon[x]);
+                counts[classIdx]++;
+            }
+
+            fenc += stride;
+            recon += stride;
+        }
+    }
+
+    int32_t _upBuff1[MAX_CU_SIZE + 2], *upBuff1 = _upBuff1 + 1;
+    int32_t _upBufft[MAX_CU_SIZE + 2], *upBufft = _upBufft + 1;
+
+    //if (iSaoType == EO_0  || iSaoType == EO_1 || iSaoType == EO_2 || iSaoType == EO_3)
+    {
+        //if (iSaoType == EO_0)
+        {
+            if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+            {
+                numSkipLine      = isChroma ? 3 - (2 * m_vChromaShift) : 3;
+                numSkipLineRight = isChroma ? 5 - (2 * m_hChromaShift) : 5;
+            }
+            stats = m_offsetOrg[partIdx][SAO_EO_0];
+            counts = m_count[partIdx][SAO_EO_0];
+
+            fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+            recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+            startX = (lpelx == 0) ? 1 : 0;
+            endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+            for (y = 0; y < lcuHeight - numSkipLine; y++)
+            {
+                int signLeft = signOf(recon[startX] - recon[startX - 1]);
+                for (x = startX; x < endX; x++)
+                {
+                    int signRight = signOf(recon[x] - recon[x + 1]);
+                    int edgeType = signRight + signLeft + 2;
+                    signLeft = -signRight;
+
+                    stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                    counts[s_eoTable[edgeType]]++;
+                }
+
+                fenc += stride;
+                recon += stride;
+            }
+        }
+
+        //if (iSaoType == EO_1)
+        {
+            if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+            {
+                numSkipLine      = isChroma ? 4 - (2 * m_vChromaShift) : 4;
+                numSkipLineRight = isChroma ? 4 - (2 * m_hChromaShift) : 4;
+            }
+            stats = m_offsetOrg[partIdx][SAO_EO_1];
+            counts = m_count[partIdx][SAO_EO_1];
+
+            fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+            recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+            startY = (tpely == 0) ? 1 : 0;
+            endX   = (rpelx == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
+            endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+            if (tpely == 0)
+            {
+                fenc += stride;
+                recon += stride;
+            }
+
+            for (x = 0; x < lcuWidth; x++)
+                upBuff1[x] = signOf(recon[x] - recon[x - stride]);
+
+            for (y = startY; y < endY; y++)
+            {
+                for (x = 0; x < endX; x++)
+                {
+                    int signDown = signOf(recon[x] - recon[x + stride]);
+                    int edgeType = signDown + upBuff1[x] + 2;
+                    upBuff1[x] = -signDown;
+
+                    stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                    counts[s_eoTable[edgeType]]++;
+                }
+
+                fenc += stride;
+                recon += stride;
+            }
+        }
+        //if (iSaoType == EO_2)
+        {
+            if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+            {
+                numSkipLine      = isChroma ? 4 - (2 * m_vChromaShift) : 4;
+                numSkipLineRight = isChroma ? 5 - (2 * m_hChromaShift) : 5;
+            }
+            stats = m_offsetOrg[partIdx][SAO_EO_2];
+            counts = m_count[partIdx][SAO_EO_2];
+
+            fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+            recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+            startX = (lpelx == 0) ? 1 : 0;
+            endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+
+            startY = (tpely == 0) ? 1 : 0;
+            endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+            if (tpely == 0)
+            {
+                fenc += stride;
+                recon += stride;
+            }
+
+            for (x = startX; x < endX; x++)
+                upBuff1[x] = signOf(recon[x] - recon[x - stride - 1]);
+
+            for (y = startY; y < endY; y++)
+            {
+                int signDown2 = signOf(recon[stride + startX] - recon[startX - 1]);
+                for (x = startX; x < endX; x++)
+                {
+                    int signDown1 = signOf(recon[x] - recon[x + stride + 1]);
+                    int edgeType  = signDown1 + upBuff1[x] + 2;
+                    upBufft[x + 1] = -signDown1;
+                    stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                    counts[s_eoTable[edgeType]]++;
+                }
+
+                upBufft[startX] = signDown2;
+                std::swap(upBuff1, upBufft);
+
+                recon += stride;
+                fenc += stride;
+            }
+        }
+        //if (iSaoType == EO_3)
+        {
+            if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+            {
+                numSkipLine      = isChroma ? 4 - (2 * m_vChromaShift) : 4;
+                numSkipLineRight = isChroma ? 5 - (2 * m_hChromaShift) : 5;
+            }
+            stats = m_offsetOrg[partIdx][SAO_EO_3];
+            counts = m_count[partIdx][SAO_EO_3];
+
+            fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+            recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+            startX = (lpelx == 0) ? 1 : 0;
+            endX   = (rpelx == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+
+            startY = (tpely == 0) ? 1 : 0;
+            endY   = (bpely == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+            if (startY == 1)
+            {
+                fenc += stride;
+                recon += stride;
+            }
+
+            for (x = startX - 1; x < endX; x++)
+                upBuff1[x] = signOf(recon[x] - recon[x - stride + 1]);
+
+            for (y = startY; y < endY; y++)
+            {
+                for (x = startX; x < endX; x++)
+                {
+                    int signDown1 = signOf(recon[x] - recon[x + stride - 1]);
+                    int edgeType  = signDown1 + upBuff1[x] + 2;
+                    upBuff1[x - 1] = -signDown1;
+                    stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                    counts[s_eoTable[edgeType]]++;
+                }
+
+                upBuff1[endX - 1] = signOf(recon[endX - 1 + stride] - recon[endX]);
+
+                recon += stride;
+                fenc += stride;
+            }
+        }
+    }
+}
+
+void SAO::calcSaoStatsCu_BeforeDblk(Frame* pic, int idxX, int idxY)
+{
+    int x, y;
+
+    pixel* fenc;
+    pixel* recon;
+    int stride;
+    uint32_t rPelX;
+    uint32_t bPelY;
+    int64_t* stats;
+    int64_t* count;
+    int startX;
+    int startY;
+    int endX;
+    int endY;
+    int firstX, firstY;
+
+    int frameWidthInCU  = m_numCuInWidth;
+
+    int isChroma;
+    int numSkipLine, numSkipLineRight;
+
+    uint32_t lPelX, tPelY;
+    TComDataCU *cu;
+    int32_t _upBuff1[MAX_CU_SIZE + 2], *upBuff1 = _upBuff1 + 1;
+    int32_t _upBufft[MAX_CU_SIZE + 2], *upBufft = _upBufft + 1;
+
+    const int boShift = X265_DEPTH - SAO_BO_BITS;
+
+    // NOTE: Row
+    {
+        // NOTE: Col
+        {
+            int addr    = idxX + frameWidthInCU * idxY;
+            cu      = pic->getCU(addr);
+
+            uint32_t picWidthTmp  = m_param->sourceWidth;
+            uint32_t picHeightTmp = m_param->sourceHeight;
+            int lcuWidth  = g_maxCUSize;
+            int lcuHeight = g_maxCUSize;
+            lPelX   = cu->getCUPelX();
+            tPelY   = cu->getCUPelY();
+            rPelX     = lPelX + lcuWidth;
+            bPelY     = tPelY + lcuHeight;
+            rPelX     = rPelX > picWidthTmp  ? picWidthTmp  : rPelX;
+            bPelY     = bPelY > picHeightTmp ? picHeightTmp : bPelY;
+            lcuWidth  = rPelX - lPelX;
+            lcuHeight = bPelY - tPelY;
+
+            memset(m_countPreDblk[addr], 0, sizeof(PerPlane));
+            memset(m_offsetOrgPreDblk[addr], 0, sizeof(PerPlane));
+
+            for (int plane = 0; plane < 3; plane++)
+            {
+                isChroma = !!plane;
+                if (plane == 1)
+                {
+                    picWidthTmp  >>= m_hChromaShift;
+                    picHeightTmp >>= m_vChromaShift;
+                    lcuWidth     >>= m_hChromaShift;
+                    lcuHeight    >>= m_vChromaShift;
+                    lPelX        >>= m_hChromaShift;
+                    tPelY        >>= m_vChromaShift;
+                    rPelX     = lPelX + lcuWidth;
+                    bPelY     = tPelY + lcuHeight;
+                }
+
+                stride   = (plane == 0) ? pic->getStride() : pic->getCStride();
+
+                //if(iSaoType == BO)
+
+                numSkipLine = isChroma ? 1 : 3;
+                numSkipLineRight = isChroma ? 2 : 4;
+
+                stats = m_offsetOrgPreDblk[addr][plane][SAO_BO];
+                count = m_countPreDblk[addr][plane][SAO_BO];
+
+                fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+                recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+                startX = (rPelX == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
+                startY = (bPelY == picHeightTmp) ? lcuHeight : lcuHeight - numSkipLine;
+
+                for (y = 0; y < lcuHeight; y++)
+                {
+                    for (x = 0; x < lcuWidth; x++)
+                    {
+                        if (x < startX && y < startY)
+                            continue;
+
+                        int classIdx = 1 + (recon[x] >> boShift);
+                        stats[classIdx] += (fenc[x] - recon[x]);
+                        count[classIdx]++;
+                    }
+
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                //if (iSaoType == EO_0)
+
+                numSkipLine = isChroma ? 1 : 3;
+                numSkipLineRight = isChroma ? 3 : 5;
+
+                stats = m_offsetOrgPreDblk[addr][plane][SAO_EO_0];
+                count = m_countPreDblk[addr][plane][SAO_EO_0];
+
+                fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+                recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+                startX = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+                startY = (bPelY == picHeightTmp) ? lcuHeight : lcuHeight - numSkipLine;
+                firstX = (lPelX == 0) ? 1 : 0;
+                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+
+                for (y = 0; y < lcuHeight; y++)
+                {
+                    int signLeft = signOf(recon[firstX] - recon[firstX - 1]);
+                    for (x = firstX; x < endX; x++)
+                    {
+                        int signRight =  signOf(recon[x] - recon[x + 1]);
+                        int edgeType =  signRight + signLeft + 2;
+                        signLeft  = -signRight;
+
+                        if (x < startX && y < startY)
+                            continue;
+
+                        stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                        count[s_eoTable[edgeType]]++;
+                    }
+
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                //if (iSaoType == EO_1)
+
+                numSkipLine = isChroma ? 2 : 4;
+                numSkipLineRight = isChroma ? 2 : 4;
+
+                stats = m_offsetOrgPreDblk[addr][plane][SAO_EO_1];
+                count = m_countPreDblk[addr][plane][SAO_EO_1];
+
+                fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+                recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+                startX = (rPelX == picWidthTmp) ? lcuWidth : lcuWidth - numSkipLineRight;
+                startY = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+                firstY = (tPelY == 0) ? 1 : 0;
+                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+                if (firstY == 1)
+                {
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                for (x = 0; x < lcuWidth; x++)
+                    upBuff1[x] = signOf(recon[x] - recon[x - stride]);
+
+                for (y = firstY; y < endY; y++)
+                {
+                    for (x = 0; x < lcuWidth; x++)
+                    {
+                        int signDown = signOf(recon[x] - recon[x + stride]);
+                        int edgeType = signDown + upBuff1[x] + 2;
+                        upBuff1[x] = -signDown;
+
+                        if (x < startX && y < startY)
+                            continue;
+
+                        stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                        count[s_eoTable[edgeType]]++;
+                    }
+
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                //if (iSaoType == EO_2)
+
+                numSkipLine = isChroma ? 2 : 4;
+                numSkipLineRight = isChroma ? 3 : 5;
+
+                stats = m_offsetOrgPreDblk[addr][plane][SAO_EO_2];
+                count = m_countPreDblk[addr][plane][SAO_EO_2];
+
+                fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+                recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+                startX = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+                startY = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+                firstX = (lPelX == 0) ? 1 : 0;
+                firstY = (tPelY == 0) ? 1 : 0;
+                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+                if (firstY == 1)
+                {
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                for (x = firstX; x < endX; x++)
+                    upBuff1[x] = signOf(recon[x] - recon[x - stride - 1]);
+
+                for (y = firstY; y < endY; y++)
+                {
+                    int signDown2 = signOf(recon[stride + startX] - recon[startX - 1]);
+                    for (x = firstX; x < endX; x++)
+                    {
+                        int signDown1 = signOf(recon[x] - recon[x + stride + 1]);
+                        int edgeType = signDown1 + upBuff1[x] + 2;
+                        upBufft[x + 1] = -signDown1;
+
+                        if (x < startX && y < startY)
+                            continue;
+
+                        stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                        count[s_eoTable[edgeType]]++;
+                    }
+
+                    upBufft[firstX] = signDown2;
+                    std::swap(upBuff1, upBufft);
+
+                    recon += stride;
+                    fenc += stride;
+                }
+
+                //if (iSaoType == EO_3)
+
+                numSkipLine = isChroma ? 2 : 4;
+                numSkipLineRight = isChroma ? 3 : 5;
+
+                stats = m_offsetOrgPreDblk[addr][plane][SAO_EO_3];
+                count = m_countPreDblk[addr][plane][SAO_EO_3];
+
+                fenc = m_pic->getPicYuvOrg()->getPlaneAddr(plane, addr);
+                recon = m_pic->getPicYuvRec()->getPlaneAddr(plane, addr);
+
+                startX = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth - numSkipLineRight;
+                startY = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight - numSkipLine;
+                firstX = (lPelX == 0) ? 1 : 0;
+                firstY = (tPelY == 0) ? 1 : 0;
+                endX   = (rPelX == picWidthTmp) ? lcuWidth - 1 : lcuWidth;
+                endY   = (bPelY == picHeightTmp) ? lcuHeight - 1 : lcuHeight;
+                if (firstY == 1)
+                {
+                    fenc += stride;
+                    recon += stride;
+                }
+
+                for (x = firstX - 1; x < endX; x++)
+                    upBuff1[x] = signOf(recon[x] - recon[x - stride + 1]);
+
+                for (y = firstY; y < endY; y++)
+                {
+                    for (x = firstX; x < endX; x++)
+                    {
+                        int signDown1 = signOf(recon[x] - recon[x + stride - 1]);
+                        int edgeType  = signDown1 + upBuff1[x] + 2;
+                        upBuff1[x - 1] = -signDown1;
+
+                        if (x < startX && y < startY)
+                            continue;
+
+                        stats[s_eoTable[edgeType]] += (fenc[x] - recon[x]);
+                        count[s_eoTable[edgeType]]++;
+                    }
+
+                    upBuff1[endX - 1] = signOf(recon[endX - 1 + stride] - recon[endX]);
+
+                    recon += stride;
+                    fenc += stride;
+                }
+            }
+        }
+    }
+}
+
+void SAO::getSaoStats(SAOQTPart *psQTPart, int plane)
+{
+    int levelIdx, partIdx;
+    int i;
+    int lcuIdx;
+    int lcuIdy;
+    int frameWidthInCU = m_pic->getFrameWidthInCU();
+    int downPartIdx;
+    int partStart;
+    int partEnd;
+    SAOQTPart* onePart;
+
+    if (!m_maxSplitLevel)
+    {
+        partIdx = 0;
+        onePart = &(psQTPart[partIdx]);
+        for (lcuIdy = onePart->startCUY; lcuIdy <= onePart->endCUY; lcuIdy++)
+        {
+            for (lcuIdx = onePart->startCUX; lcuIdx <= onePart->endCUX; lcuIdx++)
+            {
+                int addr = lcuIdy * frameWidthInCU + lcuIdx;
+                calcSaoStatsCu(addr, partIdx, plane);
+            }
+        }
+    }
+    else
+    {
+        for (partIdx = s_numCulPartsLevel[m_maxSplitLevel - 1]; partIdx < s_numCulPartsLevel[m_maxSplitLevel]; partIdx++)
+        {
+            onePart = &(psQTPart[partIdx]);
+            for (lcuIdy = onePart->startCUY; lcuIdy <= onePart->endCUY; lcuIdy++)
+            {
+                for (lcuIdx = onePart->startCUX; lcuIdx <= onePart->endCUX; lcuIdx++)
+                {
+                    int addr = lcuIdy * frameWidthInCU + lcuIdx;
+                    calcSaoStatsCu(addr, partIdx, plane);
+                }
+            }
+        }
+
+        for (levelIdx = m_maxSplitLevel - 1; levelIdx >= 0; levelIdx--)
+        {
+            partStart = (levelIdx > 0) ? s_numCulPartsLevel[levelIdx - 1] : 0;
+            partEnd   = s_numCulPartsLevel[levelIdx];
+
+            for (partIdx = partStart; partIdx < partEnd; partIdx++)
+            {
+                onePart = &(psQTPart[partIdx]);
+                for (i = 0; i < SAOQTPart::NUM_DOWN_PART; i++)
+                {
+                    downPartIdx = onePart->downPartsIdx[i];
+                    for (int typeIdx = 0; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
+                    {
+                        for (int classIdx = 0; classIdx < (typeIdx < SAO_BO ? SAO_EO_LEN : SAO_NUM_BO_CLASSES) + 1; classIdx++)
+                        {
+                            m_offsetOrg[partIdx][typeIdx][classIdx] += m_offsetOrg[downPartIdx][typeIdx][classIdx];
+                            m_count[partIdx][typeIdx][classIdx]    += m_count[downPartIdx][typeIdx][classIdx];
+                        }
+                    }
+                }
+            }
+        }
+    }
+}
+
+/* reset offset statistics */
+void SAO::resetStats()
+{
+    for (int i = 0; i < m_numTotalParts; i++)
+    {
+        m_costPartBest[i] = MAX_DOUBLE;
+        m_typePartBest[i] = -1;
+        m_distOrg[i] = 0;
+        for (int j = 0; j < MAX_NUM_SAO_TYPE; j++)
+        {
+            m_dist[i][j] = 0;
+            m_rate[i][j] = 0;
+            m_cost[i][j] = 0;
+            for (int k = 0; k < MAX_NUM_SAO_CLASS; k++)
+            {
+                m_count[i][j][k] = 0;
+                m_offset[i][j][k] = 0;
+                m_offsetOrg[i][j][k] = 0;
+            }
+        }
+    }
+}
+
+/* Sample adaptive offset process */
+void SAO::SAOProcess(SAOParam *saoParam)
+{
+    X265_CHECK(!m_param->saoLcuBasedOptimization, "SAO LCU mode failure\n"); 
+    double costFinal = 0;
+    saoParam->bSaoFlag[0] = true;
+    saoParam->bSaoFlag[1] = false;
+
+    getSaoStats(saoParam->saoPart[0], 0);
+    runQuadTreeDecision(saoParam->saoPart[0], 0, costFinal, m_maxSplitLevel, 0);
+    saoParam->bSaoFlag[0] = costFinal < 0;
+
+    if (saoParam->bSaoFlag[0])
+    {
+        convertQT2SaoUnit(saoParam, 0, 0);
+        assignSaoUnitSyntax(saoParam->saoLcuParam[0], saoParam->saoPart[0], saoParam->oneUnitFlag[0]);
+        processSaoUnitAll(saoParam->saoLcuParam[0], saoParam->oneUnitFlag[0], 0);
+    }
+    if (saoParam->bSaoFlag[1])
+    {
+        processSaoUnitAll(saoParam->saoLcuParam[1], saoParam->oneUnitFlag[1], 1);
+        processSaoUnitAll(saoParam->saoLcuParam[2], saoParam->oneUnitFlag[2], 2);
+    }
+}
+
+/* Check merge SAO unit */
+void SAO::checkMerge(SaoLcuParam * saoUnitCurr, SaoLcuParam * saoUnitCheck, int dir)
+{
+    int countDiff = 0;
+
+    if (saoUnitCurr->partIdx != saoUnitCheck->partIdx)
+    {
+        if (saoUnitCurr->typeIdx >= 0)
+        {
+            if (saoUnitCurr->typeIdx == saoUnitCheck->typeIdx)
+            {
+                for (int i = 0; i < SAO_NUM_OFFSET; i++)
+                    countDiff += (saoUnitCurr->offset[i] != saoUnitCheck->offset[i]);
+
+                countDiff += (saoUnitCurr->subTypeIdx != saoUnitCheck->subTypeIdx);
+                if (countDiff == 0)
+                {
+                    saoUnitCurr->partIdx = saoUnitCheck->partIdx;
+                    if (dir == 1)
+                    {
+                        saoUnitCurr->mergeUpFlag = 1;
+                        saoUnitCurr->mergeLeftFlag = 0;
+                    }
+                    else
+                    {
+                        saoUnitCurr->mergeUpFlag = 0;
+                        saoUnitCurr->mergeLeftFlag = 1;
+                    }
+                }
+            }
+        }
+        else
+        {
+            if (saoUnitCurr->typeIdx == saoUnitCheck->typeIdx)
+            {
+                saoUnitCurr->partIdx = saoUnitCheck->partIdx;
+                if (dir == 1)
+                {
+                    saoUnitCurr->mergeUpFlag = 1;
+                    saoUnitCurr->mergeLeftFlag = 0;
+                }
+                else
+                {
+                    saoUnitCurr->mergeUpFlag = 0;
+                    saoUnitCurr->mergeLeftFlag = 1;
+                }
+            }
+        }
+    }
+}
+
+/** Assign SAO unit syntax from picture-based algorithm */
+void SAO::assignSaoUnitSyntax(SaoLcuParam* saoLcuParam,  SAOQTPart* saoPart, bool &oneUnitFlag)
+{
+    if (saoPart->bSplit == 0)
+        oneUnitFlag = 1;
+    else
+    {
+        oneUnitFlag = 0;
+
+        int idxCount = -1;
+        saoLcuParam[0].mergeUpFlag = 0;
+        saoLcuParam[0].mergeLeftFlag = 0;
+
+        for (int j = 0; j < m_numCuInHeight; j++)
+        {
+            for (int i = 0; i < m_numCuInWidth; i++)
+            {
+                int addr     = i + j * m_numCuInWidth;
+                int addrUp   = (j == 0) ? -1 : addr - m_numCuInWidth;
+                int addrLeft = (i == 0) ? -1 : addr - 1;
+                int idx      = saoLcuParam[addr].partIdxTmp;
+                int idxLeft  = (addrLeft == -1) ? -1 : saoLcuParam[addrLeft].partIdxTmp;
+                int idxUp    = (addrUp == -1)   ? -1 : saoLcuParam[addrUp].partIdxTmp;
+
+                if (idx != idxLeft && idx != idxUp)
+                {
+                    saoLcuParam[addr].mergeUpFlag   = 0;
+                    idxCount++;
+                    saoLcuParam[addr].mergeLeftFlag = 0;
+                    saoLcuParam[addr].partIdx = idxCount;
+                }
+                else if (idx == idxLeft)
+                {
+                    saoLcuParam[addr].mergeUpFlag   = 1;
+                    saoLcuParam[addr].mergeLeftFlag = 1;
+                    saoLcuParam[addr].partIdx = saoLcuParam[addrLeft].partIdx;
+                }
+                else if (idx == idxUp)
+                {
+                    saoLcuParam[addr].mergeUpFlag   = 1;
+                    saoLcuParam[addr].mergeLeftFlag = 0;
+                    saoLcuParam[addr].partIdx = saoLcuParam[addrUp].partIdx;
+                }
+                if (addrUp != -1)
+                    checkMerge(&saoLcuParam[addr], &saoLcuParam[addrUp], 1);
+                if (addrLeft != -1)
+                    checkMerge(&saoLcuParam[addr], &saoLcuParam[addrLeft], 0);
+            }
+        }
+    }
+}
+
+void SAO::rdoSaoUnitRowInit(SAOParam *saoParam)
+{
+    saoParam->bSaoFlag[0] = true;
+    saoParam->bSaoFlag[1] = true;
+    saoParam->oneUnitFlag[0] = false;
+    saoParam->oneUnitFlag[1] = false;
+    saoParam->oneUnitFlag[2] = false;
+
+    m_numNoSao[0] = 0; // Luma
+    m_numNoSao[1] = 0; // Chroma
+    if (m_refDepth > 0 && m_depthSaoRate[0][m_refDepth - 1] > SAO_ENCODING_RATE)
+        saoParam->bSaoFlag[0] = false;
+    if (m_refDepth > 0 && m_depthSaoRate[1][m_refDepth - 1] > SAO_ENCODING_RATE_CHROMA)
+        saoParam->bSaoFlag[1] = false;
+}
+
+void SAO::rdoSaoUnitRowEnd(SAOParam *saoParam, int numlcus)
+{
+    if (!saoParam->bSaoFlag[0])
+        m_depthSaoRate[0][m_refDepth] = 1.0;
+    else
+        m_depthSaoRate[0][m_refDepth] = m_numNoSao[0] / ((double)numlcus);
+
+    if (!saoParam->bSaoFlag[1])
+        m_depthSaoRate[1][m_refDepth] = 1.0;
+    else
+        m_depthSaoRate[1][m_refDepth] = m_numNoSao[1] / ((double)numlcus * 2);
+}
+
+void SAO::rdoSaoUnitRow(SAOParam *saoParam, int idxY)
+{
+    int frameWidthInCU  = saoParam->numCuInWidth;
+    int j, k;
+    int compIdx = 0;
+    SaoLcuParam mergeSaoParam[3][2];
+    double compDistortion[3];
+
+    for (int idxX = 0; idxX < frameWidthInCU; idxX++)
+    {
+        int addr     = idxX + idxY * frameWidthInCU;
+        int addrUp   = idxY == 0 ? -1 : addr - frameWidthInCU;
+        int addrLeft = idxX == 0 ? -1 : addr - 1;
+        int allowMergeLeft = 1;
+        int allowMergeUp   = 1;
+        uint32_t rate;
+        double bestCost, mergeCost;
+        if (idxX == 0)
+            allowMergeLeft = 0;
+        if (idxY == 0)
+            allowMergeUp = 0;
+
+        compDistortion[0] = 0;
+        compDistortion[1] = 0;
+        compDistortion[2] = 0;
+        m_entropyCoder.load(m_rdEntropyCoders[0][CI_CURR_BEST]);
+        if (allowMergeLeft)
+            m_entropyCoder.codeSaoMerge(0);
+        if (allowMergeUp)
+            m_entropyCoder.codeSaoMerge(0);
+        m_entropyCoder.store(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+        // reset stats Y, Cb, Cr
+        for (compIdx = 0; compIdx < 3; compIdx++)
+        {
+            for (j = 0; j < MAX_NUM_SAO_TYPE; j++)
+            {
+                for (k = 0; k < MAX_NUM_SAO_CLASS; k++)
+                {
+                    m_offset[compIdx][j][k] = 0;
+                    if (m_param->saoLcuBasedOptimization && m_param->saoLcuBoundary)
+                    {
+                        m_count[compIdx][j][k] = m_countPreDblk[addr][compIdx][j][k];
+                        m_offsetOrg[compIdx][j][k] = m_offsetOrgPreDblk[addr][compIdx][j][k];
+                    }
+                    else
+                    {
+                        m_count[compIdx][j][k] = 0;
+                        m_offsetOrg[compIdx][j][k] = 0;
+                    }
+                }
+            }
+
+            saoParam->saoLcuParam[compIdx][addr].typeIdx       = -1;
+            saoParam->saoLcuParam[compIdx][addr].mergeUpFlag   = 0;
+            saoParam->saoLcuParam[compIdx][addr].mergeLeftFlag = 0;
+            saoParam->saoLcuParam[compIdx][addr].subTypeIdx    = 0;
+            if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
+                calcSaoStatsCu(addr, compIdx,  compIdx);
+        }
+
+        saoComponentParamDist(allowMergeLeft, allowMergeUp, saoParam, addr, addrUp, addrLeft, 0, 
+                              &mergeSaoParam[0][0], &compDistortion[0]);
+
+        sao2ChromaParamDist(allowMergeLeft, allowMergeUp, saoParam, addr, addrUp, addrLeft,
+                            &mergeSaoParam[1][0], &mergeSaoParam[2][0], &compDistortion[0]);
+
+        if (saoParam->bSaoFlag[0] || saoParam->bSaoFlag[1])
+        {
+            // Cost of new SAO_params
+            m_entropyCoder.load(m_rdEntropyCoders[0][CI_CURR_BEST]);
+            m_entropyCoder.resetBits();
+            if (allowMergeLeft)
+                m_entropyCoder.codeSaoMerge(0);
+            if (allowMergeUp)
+                m_entropyCoder.codeSaoMerge(0);
+            for (compIdx = 0; compIdx < 3; compIdx++)
+            {
+                if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
+                    m_entropyCoder.codeSaoOffset(&saoParam->saoLcuParam[compIdx][addr], compIdx);
+            }
+
+            rate = m_entropyCoder.getNumberOfWrittenBits();
+            bestCost = compDistortion[0] + (double)rate;
+            m_entropyCoder.store(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+
+            // Cost of Merge
+            for (int mergeUp = 0; mergeUp < 2; ++mergeUp)
+            {
+                if ((allowMergeLeft && !mergeUp) || (allowMergeUp && mergeUp))
+                {
+                    m_entropyCoder.load(m_rdEntropyCoders[0][CI_CURR_BEST]);
+                    m_entropyCoder.resetBits();
+                    if (allowMergeLeft)
+                        m_entropyCoder.codeSaoMerge(1 - mergeUp);
+                    if (allowMergeUp && (mergeUp == 1))
+                        m_entropyCoder.codeSaoMerge(1);
+
+                    rate = m_entropyCoder.getNumberOfWrittenBits();
+                    mergeCost = compDistortion[mergeUp + 1] + (double)rate;
+                    if (mergeCost < bestCost)
+                    {
+                        bestCost = mergeCost;
+                        m_entropyCoder.store(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+                        for (compIdx = 0; compIdx < 3; compIdx++)
+                        {
+                            mergeSaoParam[compIdx][mergeUp].mergeLeftFlag = !mergeUp;
+                            mergeSaoParam[compIdx][mergeUp].mergeUpFlag = !!mergeUp;
+                            if ((compIdx == 0 && saoParam->bSaoFlag[0]) || (compIdx > 0 && saoParam->bSaoFlag[1]))
+                                copySaoUnit(&saoParam->saoLcuParam[compIdx][addr], &mergeSaoParam[compIdx][mergeUp]);
+                        }
+                    }
+                }
+            }
+
+            if (saoParam->saoLcuParam[0][addr].typeIdx < 0)
+                m_numNoSao[0]++;
+            if (saoParam->saoLcuParam[1][addr].typeIdx < 0)
+                m_numNoSao[1] += 2;
+            m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+            m_entropyCoder.store(m_rdEntropyCoders[0][CI_CURR_BEST]);
+        }
+    }
+}
+
+/** rate distortion optimization of SAO unit */
+inline int64_t SAO::estSaoTypeDist(int compIdx, int typeIdx, int shift, double lambda, int32_t *currentDistortionTableBo, double *currentRdCostTableBo)
+{
+    int64_t estDist = 0;
+
+    for (int classIdx = 1; classIdx < ((typeIdx < SAO_BO) ?  SAO_EO_LEN + 1 : SAO_NUM_BO_CLASSES + 1); classIdx++)
+    {
+        if (typeIdx == SAO_BO)
+        {
+            currentDistortionTableBo[classIdx - 1] = 0;
+            currentRdCostTableBo[classIdx - 1] = lambda;
+        }
+        if (m_count[compIdx][typeIdx][classIdx])
+        {
+            m_offset[compIdx][typeIdx][classIdx] = (int64_t)roundIDBI((double)(m_offsetOrg[compIdx][typeIdx][classIdx] << (X265_DEPTH - 8)) / (double)(m_count[compIdx][typeIdx][classIdx] << SAO_BIT_INC));
+            m_offset[compIdx][typeIdx][classIdx] = Clip3(-OFFSET_THRESH + 1, OFFSET_THRESH - 1, (int)m_offset[compIdx][typeIdx][classIdx]);
+            if (typeIdx < SAO_BO)
+            {
+                if (m_offset[compIdx][typeIdx][classIdx] < 0 && classIdx < 3)
+                    m_offset[compIdx][typeIdx][classIdx] = 0;
+                if (m_offset[compIdx][typeIdx][classIdx] > 0 && classIdx >= 3)
+                    m_offset[compIdx][typeIdx][classIdx] = 0;
+            }
+            m_offset[compIdx][typeIdx][classIdx] = estIterOffset(typeIdx, classIdx, lambda, m_offset[compIdx][typeIdx][classIdx], m_count[compIdx][typeIdx][classIdx], m_offsetOrg[compIdx][typeIdx][classIdx], shift, SAO_BIT_INC, currentDistortionTableBo, currentRdCostTableBo, OFFSET_THRESH);
+        }
+        else
+        {
+            m_offsetOrg[compIdx][typeIdx][classIdx] = 0;
+            m_offset[compIdx][typeIdx][classIdx] = 0;
+        }
+        if (typeIdx != SAO_BO)
+            estDist += estSaoDist(m_count[compIdx][typeIdx][classIdx], m_offset[compIdx][typeIdx][classIdx] << SAO_BIT_INC, m_offsetOrg[compIdx][typeIdx][classIdx], shift);
+    }
+
+    return estDist;
+}
+
+inline int64_t SAO::estSaoDist(int64_t count, int64_t offset, int64_t offsetOrg, int shift)
+{
+    return (count * offset * offset - offsetOrg * offset * 2) >> shift;
+}
+
+inline int64_t SAO::estIterOffset(int typeIdx, int classIdx, double lambda, int64_t offsetInput, int64_t count, int64_t offsetOrg, int shift, int bitIncrease, int32_t *currentDistortionTableBo, double *currentRdCostTableBo, int offsetTh)
+{
+    //Clean up, best_q_offset.
+    int64_t iterOffset, tempOffset;
+    int64_t tempDist, tempRate;
+    int64_t offsetOutput = 0;
+
+    iterOffset = offsetInput;
+    // Assuming sending quantized value 0 results in zero offset and sending the value zero needs 1 bit. entropy coder can be used to measure the exact rate here.
+    double tempMinCost = lambda;
+    while (iterOffset != 0)
+    {
+        // Calculate the bits required for signalling the offset
+        tempRate = (typeIdx == SAO_BO) ? (abs((int)iterOffset) + 2) : (abs((int)iterOffset) + 1);
+        if (abs((int)iterOffset) == offsetTh - 1)
+            tempRate--;
+
+        // Do the dequntization before distorion calculation
+        tempOffset = iterOffset << bitIncrease;
+        tempDist   = estSaoDist(count, tempOffset, offsetOrg, shift);
+        double tempCost   = ((double)tempDist + lambda * (double)tempRate);
+        if (tempCost < tempMinCost)
+        {
+            tempMinCost = tempCost;
+            offsetOutput = iterOffset;
+            if (typeIdx == SAO_BO)
+            {
+                currentDistortionTableBo[classIdx - 1] = (int)tempDist;
+                currentRdCostTableBo[classIdx - 1] = tempCost;
+            }
+        }
+        iterOffset = (iterOffset > 0) ? (iterOffset - 1) : (iterOffset + 1);
+    }
+
+    return offsetOutput;
+}
+
+void SAO::saoComponentParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, int plane,
+                                SaoLcuParam *compSaoParam, double *compDistortion)
+{
+    int64_t estDist;
+    int64_t bestDist;
+
+    SaoLcuParam* saoLcuParam = &(saoParam->saoLcuParam[plane][addr]);
+    SaoLcuParam* saoLcuParamNeighbor = NULL;
+
+    resetSaoUnit(saoLcuParam);
+    resetSaoUnit(&compSaoParam[0]);
+    resetSaoUnit(&compSaoParam[1]);
+
+    double dCostPartBest = MAX_DOUBLE;
+    double bestRDCostTableBo = MAX_DOUBLE;
+    int    bestClassTableBo  = 0;
+    int    currentDistortionTableBo[MAX_NUM_SAO_CLASS];
+    double currentRdCostTableBo[MAX_NUM_SAO_CLASS];
+
+    SaoLcuParam saoLcuParamRdo;
+
+    resetSaoUnit(&saoLcuParamRdo);
+
+    m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+    m_entropyCoder.resetBits();
+    m_entropyCoder.codeSaoOffset(&saoLcuParamRdo, plane);
+    dCostPartBest = m_entropyCoder.getNumberOfWrittenBits() * m_lumaLambda;
+    copySaoUnit(saoLcuParam, &saoLcuParamRdo);
+    bestDist = 0;
+
+    for (int typeIdx = 0; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
+    {
+        estDist = estSaoTypeDist(plane, typeIdx, 0, m_lumaLambda, currentDistortionTableBo, currentRdCostTableBo);
+
+        if (typeIdx == SAO_BO)
+        {
+            // Estimate Best Position
+            for (int i = 0; i < SAO_NUM_BO_CLASSES - SAO_BO_LEN + 1; i++)
+            {
+                double currentRDCost = 0.0;
+                for (int j = i; j < i + SAO_BO_LEN; j++)
+                    currentRDCost += currentRdCostTableBo[j];
+
+                if (currentRDCost < bestRDCostTableBo)
+                {
+                    bestRDCostTableBo = currentRDCost;
+                    bestClassTableBo  = i;
+                }
+            }
+
+            // Re code all Offsets
+            // Code Center
+            estDist = 0;
+            for (int classIdx = bestClassTableBo; classIdx < bestClassTableBo + SAO_BO_LEN; classIdx++)
+                estDist += currentDistortionTableBo[classIdx];
+        }
+        resetSaoUnit(&saoLcuParamRdo);
+        saoLcuParamRdo.typeIdx = typeIdx;
+        saoLcuParamRdo.mergeLeftFlag = 0;
+        saoLcuParamRdo.mergeUpFlag   = 0;
+        saoLcuParamRdo.subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo : 0;
+        for (int classIdx = 0; classIdx < SAO_NUM_OFFSET; classIdx++)
+            saoLcuParamRdo.offset[classIdx] = (int)m_offset[plane][typeIdx][classIdx + saoLcuParamRdo.subTypeIdx + 1];
+
+        m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+        m_entropyCoder.resetBits();
+        m_entropyCoder.codeSaoOffset(&saoLcuParamRdo, plane);
+
+        uint32_t estRate = m_entropyCoder.getNumberOfWrittenBits();
+        m_cost[plane][typeIdx] = (double)((double)estDist + m_lumaLambda * (double)estRate);
+
+        if (m_cost[plane][typeIdx] < dCostPartBest)
+        {
+            dCostPartBest = m_cost[plane][typeIdx];
+            copySaoUnit(saoLcuParam, &saoLcuParamRdo);
+            bestDist = estDist;
+        }
+    }
+
+    compDistortion[0] += ((double)bestDist / m_lumaLambda);
+    m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+    m_entropyCoder.codeSaoOffset(saoLcuParam, plane);
+    m_entropyCoder.store(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+
+    // merge left or merge up
+
+    for (int idxNeighbor = 0; idxNeighbor < 2; idxNeighbor++)
+    {
+        saoLcuParamNeighbor = NULL;
+        if (allowMergeLeft && addrLeft >= 0 && idxNeighbor == 0)
+            saoLcuParamNeighbor = &(saoParam->saoLcuParam[plane][addrLeft]);
+        else if (allowMergeUp && addrUp >= 0 && idxNeighbor == 1)
+            saoLcuParamNeighbor = &(saoParam->saoLcuParam[plane][addrUp]);
+        if (saoLcuParamNeighbor != NULL)
+        {
+            estDist = 0;
+            int typeIdx = saoLcuParamNeighbor->typeIdx;
+            if (typeIdx >= 0)
+            {
+                int mergeBandPosition = (typeIdx == SAO_BO) ? saoLcuParamNeighbor->subTypeIdx : 0;
+                int mergeOffset;
+                for (int classIdx = 0; classIdx < SAO_NUM_OFFSET; classIdx++)
+                {
+                    mergeOffset = saoLcuParamNeighbor->offset[classIdx];
+                    estDist += estSaoDist(m_count[plane][typeIdx][classIdx + mergeBandPosition + 1], mergeOffset, m_offsetOrg[plane][typeIdx][classIdx + mergeBandPosition + 1],  0);
+                }
+            }
+            else
+                estDist = 0;
+
+            copySaoUnit(&compSaoParam[idxNeighbor], saoLcuParamNeighbor);
+            compSaoParam[idxNeighbor].mergeUpFlag   = !!idxNeighbor;
+            compSaoParam[idxNeighbor].mergeLeftFlag = !idxNeighbor;
+
+            compDistortion[idxNeighbor + 1] += ((double)estDist / m_lumaLambda);
+        }
+    }
+}
+
+void SAO::sao2ChromaParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft,
+                              SaoLcuParam *crSaoParam, SaoLcuParam *cbSaoParam, double *distortion)
+{
+    int64_t estDist[2];
+    int64_t bestDist = 0;
+
+    SaoLcuParam* saoLcuParam[2] = { &(saoParam->saoLcuParam[1][addr]), &(saoParam->saoLcuParam[2][addr]) };
+    SaoLcuParam* saoLcuParamNeighbor[2] = { NULL, NULL };
+    SaoLcuParam* saoMergeParam[2][2];
+
+    saoMergeParam[0][0] = &crSaoParam[0];
+    saoMergeParam[0][1] = &crSaoParam[1];
+    saoMergeParam[1][0] = &cbSaoParam[0];
+    saoMergeParam[1][1] = &cbSaoParam[1];
+
+    resetSaoUnit(saoLcuParam[0]);
+    resetSaoUnit(saoLcuParam[1]);
+    resetSaoUnit(saoMergeParam[0][0]);
+    resetSaoUnit(saoMergeParam[0][1]);
+    resetSaoUnit(saoMergeParam[1][0]);
+    resetSaoUnit(saoMergeParam[1][1]);
+
+    double costPartBest = MAX_DOUBLE;
+    double bestRDCostTableBo;
+    double currentRdCostTableBo[MAX_NUM_SAO_CLASS];
+    int    bestClassTableBo[2] = { 0, 0 };
+    int    currentDistortionTableBo[MAX_NUM_SAO_CLASS];
+
+    SaoLcuParam saoLcuParamRdo[2];
+
+    resetSaoUnit(&saoLcuParamRdo[0]);
+    resetSaoUnit(&saoLcuParamRdo[1]);
+
+    m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+    m_entropyCoder.resetBits();
+    m_entropyCoder.codeSaoOffset(&saoLcuParamRdo[0], 1);
+    m_entropyCoder.codeSaoOffset(&saoLcuParamRdo[1], 2);
+
+    costPartBest = m_entropyCoder.getNumberOfWrittenBits() * m_chromaLambda;
+    copySaoUnit(saoLcuParam[0], &saoLcuParamRdo[0]);
+    copySaoUnit(saoLcuParam[1], &saoLcuParamRdo[1]);
+
+    for (int typeIdx = 0; typeIdx < MAX_NUM_SAO_TYPE; typeIdx++)
+    {
+        if (typeIdx == SAO_BO)
+        {
+            // Estimate Best Position
+            for (int compIdx = 0; compIdx < 2; compIdx++)
+            {
+                bestRDCostTableBo = MAX_DOUBLE;
+                estDist[compIdx] = estSaoTypeDist(compIdx + 1, typeIdx, 0, m_chromaLambda, currentDistortionTableBo, currentRdCostTableBo);
+                for (int i = 0; i < SAO_NUM_BO_CLASSES - SAO_BO_LEN + 1; i++)
+                {
+                    double currentRDCost = 0.0;
+                    for (int j = i; j < i + SAO_BO_LEN; j++)
+                        currentRDCost += currentRdCostTableBo[j];
+
+                    if (currentRDCost < bestRDCostTableBo)
+                    {
+                        bestRDCostTableBo = currentRDCost;
+                        bestClassTableBo[compIdx]  = i;
+                    }
+                }
+
+                // Re code all Offsets
+                // Code Center
+                estDist[compIdx] = 0;
+                for (int classIdx = bestClassTableBo[compIdx]; classIdx < bestClassTableBo[compIdx] + SAO_BO_LEN; classIdx++)
+                    estDist[compIdx] += currentDistortionTableBo[classIdx];
+            }
+        }
+        else
+        {
+            estDist[0] = estSaoTypeDist(1, typeIdx, 0, m_chromaLambda, currentDistortionTableBo, currentRdCostTableBo);
+            estDist[1] = estSaoTypeDist(2, typeIdx, 0, m_chromaLambda, currentDistortionTableBo, currentRdCostTableBo);
+        }
+
+        m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+        m_entropyCoder.resetBits();
+
+        for (int compIdx = 0; compIdx < 2; compIdx++)
+        {
+            resetSaoUnit(&saoLcuParamRdo[compIdx]);
+            saoLcuParamRdo[compIdx].typeIdx = typeIdx;
+            saoLcuParamRdo[compIdx].mergeLeftFlag = 0;
+            saoLcuParamRdo[compIdx].mergeUpFlag   = 0;
+            saoLcuParamRdo[compIdx].subTypeIdx = (typeIdx == SAO_BO) ? bestClassTableBo[compIdx] : 0;
+            for (int classIdx = 0; classIdx < SAO_NUM_OFFSET; classIdx++)
+                saoLcuParamRdo[compIdx].offset[classIdx] = (int)m_offset[compIdx + 1][typeIdx][classIdx + saoLcuParamRdo[compIdx].subTypeIdx + 1];
+
+            m_entropyCoder.codeSaoOffset(&saoLcuParamRdo[compIdx], compIdx + 1);
+        }
+
+        uint32_t estRate = m_entropyCoder.getNumberOfWrittenBits();
+        m_cost[1][typeIdx] = (double)((double)(estDist[0] + estDist[1]) + m_chromaLambda * (double)estRate);
+
+        if (m_cost[1][typeIdx] < costPartBest)
+        {
+            costPartBest = m_cost[1][typeIdx];
+            copySaoUnit(saoLcuParam[0], &saoLcuParamRdo[0]);
+            copySaoUnit(saoLcuParam[1], &saoLcuParamRdo[1]);
+            bestDist = (estDist[0] + estDist[1]);
+        }
+    }
+
+    distortion[0] += ((double)bestDist / m_chromaLambda);
+    m_entropyCoder.load(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+    m_entropyCoder.codeSaoOffset(saoLcuParam[0], 1);
+    m_entropyCoder.codeSaoOffset(saoLcuParam[1], 2);
+    m_entropyCoder.store(m_rdEntropyCoders[0][CI_TEMP_BEST]);
+
+    // merge left or merge up
+
+    for (int idxNeighbor = 0; idxNeighbor < 2; idxNeighbor++)
+    {
+        for (int compIdx = 0; compIdx < 2; compIdx++)
+        {
+            saoLcuParamNeighbor[compIdx] = NULL;
+            if (allowMergeLeft && addrLeft >= 0 && idxNeighbor == 0)
+                saoLcuParamNeighbor[compIdx] = &(saoParam->saoLcuParam[compIdx + 1][addrLeft]);
+            else if (allowMergeUp && addrUp >= 0 && idxNeighbor == 1)
+                saoLcuParamNeighbor[compIdx] = &(saoParam->saoLcuParam[compIdx + 1][addrUp]);
+            if (saoLcuParamNeighbor[compIdx] != NULL)
+            {
+                estDist[compIdx] = 0;
+                int typeIdx = saoLcuParamNeighbor[compIdx]->typeIdx;
+                if (typeIdx >= 0)
+                {
+                    int mergeBandPosition = (typeIdx == SAO_BO) ? saoLcuParamNeighbor[compIdx]->subTypeIdx : 0;
+                    for (int classIdx = 0; classIdx < SAO_NUM_OFFSET; classIdx++)
+                    {
+                        int mergeOffset = saoLcuParamNeighbor[compIdx]->offset[classIdx];
+                        estDist[compIdx] += estSaoDist(m_count[compIdx + 1][typeIdx][classIdx + mergeBandPosition + 1], mergeOffset, m_offsetOrg[compIdx + 1][typeIdx][classIdx + mergeBandPosition + 1],  0);
+                    }
+                }
+                else
+                    estDist[compIdx] = 0;
+
+                copySaoUnit(saoMergeParam[compIdx][idxNeighbor], saoLcuParamNeighbor[compIdx]);
+                saoMergeParam[compIdx][idxNeighbor]->mergeUpFlag   = !!idxNeighbor;
+                saoMergeParam[compIdx][idxNeighbor]->mergeLeftFlag = !idxNeighbor;
+                distortion[idxNeighbor + 1] += ((double)estDist[compIdx] / m_chromaLambda);
+            }
+        }
+    }
+}
+
+static void restoreOrigLosslessYuv(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth);
+
+/* Original Lossless YUV LF disable process */
+void restoreLFDisabledOrigYuv(Frame* pic)
+{
+    if (pic->m_picSym->m_slice->m_pps->bTransquantBypassEnabled)
+    {
+        for (uint32_t cuAddr = 0; cuAddr < pic->getNumCUsInFrame(); cuAddr++)
+        {
+            TComDataCU* cu = pic->getCU(cuAddr);
+
+            origCUSampleRestoration(cu, 0, 0);
+        }
+    }
+}
+
+/* Original YUV restoration for CU in lossless coding */
+void origCUSampleRestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth)
+{
+    // go to sub-CU
+    if (cu->getDepth(absZOrderIdx) > depth)
+    {
+        uint32_t curNumParts = NUM_CU_PARTITIONS >> (depth << 1);
+        uint32_t qNumParts   = curNumParts >> 2;
+        uint32_t xmax = cu->m_slice->m_sps->picWidthInLumaSamples  - cu->getCUPelX();
+        uint32_t ymax = cu->m_slice->m_sps->picHeightInLumaSamples - cu->getCUPelY();
+        for (uint32_t partIdx = 0; partIdx < 4; partIdx++, absZOrderIdx += qNumParts)
+        {
+            if (g_zscanToPelX[absZOrderIdx] < xmax && g_zscanToPelY[absZOrderIdx] < ymax)
+                origCUSampleRestoration(cu, absZOrderIdx, depth + 1);
+        }
+
+        return;
+    }
+
+    // restore original YUV samples
+    if (cu->isLosslessCoded(absZOrderIdx))
+        restoreOrigLosslessYuv(cu, absZOrderIdx, depth);
+}
+
+/* Original Lossless YUV sample restoration */
+static void restoreOrigLosslessYuv(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth)
+{
+    TComPicYuv* pcPicYuvRec = cu->m_pic->getPicYuvRec();
+    int hChromaShift = cu->getHorzChromaShift();
+    int vChromaShift = cu->getVertChromaShift();
+    uint32_t lumaOffset   = absZOrderIdx << (LOG2_UNIT_SIZE * 2);
+    uint32_t chromaOffset = lumaOffset >> (hChromaShift + vChromaShift);
+
+    pixel* dst = pcPicYuvRec->getLumaAddr(cu->getAddr(), absZOrderIdx);
+    pixel* src = cu->getLumaOrigYuv() + lumaOffset;
+    uint32_t stride = pcPicYuvRec->getStride();
+    uint32_t width  = (g_maxCUSize >> depth);
+    uint32_t height = (g_maxCUSize >> depth);
+
+    //TODO Optimized Primitives
+    for (uint32_t y = 0; y < height; y++)
+    {
+        for (uint32_t x = 0; x < width; x++)
+        {
+            dst[x] = src[x];
+        }
+
+        src += width;
+        dst += stride;
+    }
+
+    pixel* dstCb = pcPicYuvRec->getChromaAddr(1, cu->getAddr(), absZOrderIdx);
+    pixel* srcCb = cu->getChromaOrigYuv(1) + chromaOffset;
+
+    pixel* dstCr = pcPicYuvRec->getChromaAddr(2, cu->getAddr(), absZOrderIdx);
+    pixel* srcCr = cu->getChromaOrigYuv(2) + chromaOffset;
+
+    stride = pcPicYuvRec->getCStride();
+    width  = ((g_maxCUSize >> depth) >> hChromaShift);
+    height = ((g_maxCUSize >> depth) >> vChromaShift);
+
+    //TODO Optimized Primitives
+    for (uint32_t y = 0; y < height; y++)
+    {
+        for (uint32_t x = 0; x < width; x++)
+        {
+            dstCb[x] = srcCb[x];
+            dstCr[x] = srcCr[x];
+        }
+
+        srcCb += width;
+        dstCb += stride;
+        srcCr += width;
+        dstCr += stride;
+    }
+}
+
+}

x265-hg20140928.tar.bz2/source/encoder/sao.h Added

@@ -0,0 +1,176 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Steve Borho <steve@borho.org>
+ *          Min Chen <chenm003@163.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_SAO_H
+#define X265_SAO_H
+
+#include "common.h"
+#include "frame.h"
+#include "entropy.h"
+
+namespace x265 {
+// private namespace
+
+enum SAOTypeLen
+{
+    SAO_EO_LEN = 4,
+    SAO_BO_LEN = 4,
+    SAO_NUM_BO_CLASSES = 32
+};
+
+enum SAOType
+{
+    SAO_EO_0 = 0,
+    SAO_EO_1,
+    SAO_EO_2,
+    SAO_EO_3,
+    SAO_BO,
+    MAX_NUM_SAO_TYPE
+};
+
+class SAO
+{
+protected:
+
+    enum { SAO_MAX_DEPTH = 4 };
+    enum { SAO_BO_BITS  = 5 };
+    enum { MAX_NUM_SAO_CLASS = 33 };
+    enum { SAO_BIT_INC = X265_MAX(X265_DEPTH - 10, 0) };
+    enum { OFFSET_THRESH = 1 << X265_MIN(X265_DEPTH - 5, 5) };
+    enum { NUM_EDGETYPE = 5 };
+
+    static const int      s_numCulPartsLevel[5];
+    static const uint32_t s_eoTable[NUM_EDGETYPE];
+
+    typedef int64_t (PerClass[MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS]);
+    typedef int64_t (PerType[MAX_NUM_SAO_TYPE]);
+    typedef double  (PerTypeD[MAX_NUM_SAO_TYPE]);
+    typedef int64_t (PerPlane[3][MAX_NUM_SAO_TYPE][MAX_NUM_SAO_CLASS]);
+
+    /* allocated per part */
+    PerClass*   m_count;
+    PerClass*   m_offset;
+    PerClass*   m_offsetOrg;
+    PerType*    m_rate;
+    PerType*    m_dist;
+    PerTypeD*   m_cost;
+    double*     m_costPartBest;
+    int64_t*    m_distOrg;
+    int*        m_typePartBest;
+
+    /* allocated per LCU */
+    PerPlane*   m_countPreDblk;
+    PerPlane*   m_offsetOrgPreDblk;
+
+    double      m_depthSaoRate[2][4];
+    pixel*      m_offsetBo;
+    int8_t      m_offsetEo[NUM_EDGETYPE];
+
+    int         m_maxSplitLevel;
+
+    int         m_numCuInWidth;
+    int         m_numCuInHeight;
+    int         m_numTotalParts;
+    int         m_hChromaShift;
+    int         m_vChromaShift;
+
+    pixel*      m_clipTable;
+    pixel*      m_clipTableBase;
+
+    pixel*      m_tmpU1[3];
+    pixel*      m_tmpU2[3];
+    pixel*      m_tmpL1;
+    pixel*      m_tmpL2;
+
+public:
+
+    Frame*      m_pic;
+    Entropy     m_rdEntropyCoders[5][CI_NUM_SAO];
+    Entropy     m_entropyCoder;
+
+    x265_param* m_param;
+    int         m_refDepth;
+    int         m_numNoSao[2];
+    
+    double      m_lumaLambda;
+    double      m_chromaLambda;
+    /* TODO: No doubles for distortion */
+
+    SAO();
+
+    bool create(x265_param *param);
+    void destroy();
+
+    void initSAOParam(SAOParam* saoParam, int partLevel, int partRow, int partCol, int parentPartIdx, int startCUX, int endCUX, int startCUY, int endCUY, int plane) const;
+    void allocSaoParam(SAOParam* saoParam) const;
+
+    void startSlice(Frame *pic, Entropy& initState, int qp);
+    void resetSAOParam(SAOParam* saoParam);
+    void resetStats();
+    void resetSaoUnit(SaoLcuParam* saoUnit);
+
+    void SAOProcess(SAOParam* saoParam);
+
+    // LCU-basd SAO process without slice granularity
+    void processSaoCu(int addr, int partIdx, int plane);
+
+    void resetLcuPart(SaoLcuParam* saoLcuParam);
+    void convertQT2SaoUnit(SAOParam* saoParam, uint32_t partIdx, int plane);
+    void convertOnePart2SaoUnit(SAOParam *saoParam, uint32_t partIdx, int plane);
+    void processSaoUnitAll(SaoLcuParam* saoLcuParam, bool oneUnitFlag, int plane);
+    void processSaoUnitRow(SaoLcuParam* saoLcuParam, int idxY, int plane);
+
+    void copySaoUnit(SaoLcuParam* saoUnitDst, SaoLcuParam* saoUnitSrc);
+
+    void runQuadTreeDecision(SAOQTPart *psQTPart, int partIdx, double &costFinal, int maxLevel, int plane);
+    void rdoSaoOnePart(SAOQTPart *psQTPart, int partIdx, int plane);
+
+    void disablePartTree(SAOQTPart *psQTPart, int partIdx);
+    void getSaoStats(SAOQTPart *psQTPart, int plane);
+    void calcSaoStatsCu(int addr, int partIdx, int plane);
+    void calcSaoStatsCu_BeforeDblk(Frame* pic, int idxX, int idxY);
+    void assignSaoUnitSyntax(SaoLcuParam* saoLcuParam,  SAOQTPart* saoPart, bool &oneUnitFlag);
+    void checkMerge(SaoLcuParam* lcuParamCurr, SaoLcuParam * lcuParamCheck, int dir);
+
+    void saoComponentParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft, int plane,
+                               SaoLcuParam *compSaoParam, double *distortion);
+    void sao2ChromaParamDist(int allowMergeLeft, int allowMergeUp, SAOParam *saoParam, int addr, int addrUp, int addrLeft,
+                            SaoLcuParam *crSaoParam, SaoLcuParam *cbSaoParam, double *distortion);
+
+    inline int64_t estSaoDist(int64_t count, int64_t offset, int64_t offsetOrg, int shift);
+    inline int64_t estIterOffset(int typeIdx, int classIdx, double lambda, int64_t offsetInput, int64_t count, int64_t offsetOrg, int shift,
+                                 int bitIncrease, int32_t *currentDistortionTableBo, double *currentRdCostTableBo, int offsetTh);
+    inline int64_t estSaoTypeDist(int compIdx, int typeIdx, int shift, double lambda, int32_t *currentDistortionTableBo, double *currentRdCostTableBo);
+
+    void rdoSaoUnitRowInit(SAOParam *saoParam);
+    void rdoSaoUnitRowEnd(SAOParam *saoParam, int numlcus);
+    void rdoSaoUnitRow(SAOParam *saoParam, int idxY);
+};
+
+void restoreLFDisabledOrigYuv(Frame* pic);
+void origCUSampleRestoration(TComDataCU* cu, uint32_t absZOrderIdx, uint32_t depth);
+
+}
+
+#endif // ifndef X265_SAO_H

x265-hg20140928.tar.bz2/source/encoder/search.cpp Added

@@ -0,0 +1,3497 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "TLibCommon/TComRom.h"
+#include "TLibCommon/TComMotionInfo.h"
+
+#include "common.h"
+#include "primitives.h"
+
+#include "search.h"
+#include "entropy.h"
+#include "rdcost.h"
+
+using namespace x265;
+
+ALIGN_VAR_32(const pixel, Search::zeroPel[MAX_CU_SIZE]) = { 0 };
+
+Search::Search()
+{
+    memset(m_qtTempCoeff, 0, sizeof(m_qtTempCoeff));
+    m_qtTempShortYuv = NULL;
+    for (int i = 0; i < 3; i++)
+    {
+        m_qtTempTransformSkipFlag[i] = NULL;
+        m_qtTempCbf[i] = NULL;
+    }
+
+    m_numLayers = 0;
+    m_param = NULL;
+    m_entropyCoder = NULL;
+    m_rdEntropyCoders = NULL;
+}
+
+Search::~Search()
+{
+    for (int i = 0; i < m_numLayers; ++i)
+    {
+        X265_FREE(m_qtTempCoeff[0][i]);
+        m_qtTempShortYuv[i].destroy();
+    }
+
+    X265_FREE(m_qtTempCbf[0]);
+    X265_FREE(m_qtTempTransformSkipFlag[0]);
+    m_predTempYuv.destroy();
+
+    delete[] m_qtTempShortYuv;
+}
+
+bool Search::initSearch(x265_param *param, ScalingList& scalingList)
+{
+    m_param = param;
+    m_bEnableRDOQ = param->rdLevel >= 4;
+    m_bFrameParallel = param->frameNumThreads > 1;
+    m_numLayers = g_log2Size[param->maxCUSize] - 2;
+
+    bool ok = m_quant.init(m_bEnableRDOQ, param->psyRdoq, scalingList);
+
+    m_rdCost.setPsyRdScale(param->psyRd);
+
+    ok &= Predict::allocBuffers(param->internalCsp);
+    ok &= m_predTempYuv.create(MAX_CU_SIZE, MAX_CU_SIZE, param->internalCsp);
+    m_me.setSearchMethod(param->searchMethod);
+    m_me.setSubpelRefine(param->subpelRefine);
+
+    /* When frame parallelism is active, only 'refLagPixels' of reference frames will be guaranteed
+     * available for motion reference.  See refLagRows in FrameEncoder::compressCTURows() */
+    m_refLagPixels = m_bFrameParallel ? param->searchRange : param->sourceHeight;
+
+    m_qtTempShortYuv = new ShortYuv[m_numLayers];
+    uint32_t sizeL = 1 << (g_maxLog2CUSize * 2);
+    uint32_t sizeC = sizeL >> (CHROMA_H_SHIFT(m_csp) + CHROMA_V_SHIFT(m_csp));
+    for (int i = 0; i < m_numLayers; ++i)
+    {
+        m_qtTempCoeff[0][i] = X265_MALLOC(coeff_t, sizeL + sizeC * 2);
+        m_qtTempCoeff[1][i] = m_qtTempCoeff[0][i] + sizeL;
+        m_qtTempCoeff[2][i] = m_qtTempCoeff[0][i] + sizeL + sizeC;
+        ok &= m_qtTempShortYuv[i].create(MAX_CU_SIZE, MAX_CU_SIZE, param->internalCsp);
+    }
+
+    const uint32_t numPartitions = 1 << (g_maxFullDepth * 2);
+    CHECKED_MALLOC(m_qtTempCbf[0], uint8_t, numPartitions * 3);
+    m_qtTempCbf[1] = m_qtTempCbf[0] + numPartitions;
+    m_qtTempCbf[2] = m_qtTempCbf[0] + numPartitions * 2;
+    CHECKED_MALLOC(m_qtTempTransformSkipFlag[0], uint8_t, numPartitions * 3);
+    m_qtTempTransformSkipFlag[1] = m_qtTempTransformSkipFlag[0] + numPartitions;
+    m_qtTempTransformSkipFlag[2] = m_qtTempTransformSkipFlag[0] + numPartitions * 2;
+
+    return ok;
+
+fail:
+    return false;
+}
+
+void Search::xEncSubdivCbfQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t absPartIdxStep, uint32_t width, uint32_t height)
+{
+    uint32_t fullDepth  = cu->getDepth(0) + trDepth;
+    uint32_t trMode     = cu->getTransformIdx(absPartIdx);
+    uint32_t subdiv     = (trMode > trDepth ? 1 : 0);
+    uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    if ((log2TrSize > 2) && !(m_csp == X265_CSP_I444))
+    {
+        if (trDepth == 0 || cu->getCbf(absPartIdx, TEXT_CHROMA_U, trDepth - 1))
+            m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, (width >> hChromaShift), (height >> vChromaShift), TEXT_CHROMA_U, trDepth, (subdiv == 0));
+
+        if (trDepth == 0 || cu->getCbf(absPartIdx, TEXT_CHROMA_V, trDepth - 1))
+            m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, (width >> hChromaShift), (height >> vChromaShift), TEXT_CHROMA_V, trDepth, (subdiv == 0));
+    }
+
+    if (subdiv)
+    {
+        absPartIdxStep >>= 2;
+        width  >>= 1;
+        height >>= 1;
+
+        uint32_t qtPartNum = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        for (uint32_t part = 0; part < 4; part++)
+            xEncSubdivCbfQTChroma(cu, trDepth + 1, absPartIdx + part * qtPartNum, absPartIdxStep, width, height);
+    }
+}
+
+void Search::xEncCoeffQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TextType ttype)
+{
+    if (!cu->getCbf(absPartIdx, ttype, trDepth))
+        return;
+
+    uint32_t fullDepth = cu->getDepth(0) + trDepth;
+    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+
+    if (trMode > trDepth)
+    {
+        uint32_t qtPartNum = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        for (uint32_t part = 0; part < 4; part++)
+            xEncCoeffQTChroma(cu, trDepth + 1, absPartIdx + part * qtPartNum, ttype);
+
+        return;
+    }
+
+    uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+
+    uint32_t trDepthC = trDepth;
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+    
+    if ((log2TrSize == 2) && !(m_csp == X265_CSP_I444))
+    {
+        X265_CHECK(trDepth > 0, "transform size too small\n");
+        trDepthC--;
+        log2TrSizeC++;
+        uint32_t qpdiv = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepthC) << 1);
+        bool bFirstQ = ((absPartIdx & (qpdiv - 1)) == 0);
+        if (!bFirstQ)
+            return;
+    }
+
+    uint32_t qtLayer    = log2TrSize - 2;
+
+    if (m_csp != X265_CSP_I422)
+    {
+        uint32_t shift = (m_csp == X265_CSP_I420) ? 2 : 0;
+        uint32_t coeffOffset = absPartIdx << (LOG2_UNIT_SIZE * 2 - shift);
+        coeff_t* coeff = m_qtTempCoeff[ttype][qtLayer] + coeffOffset;
+        m_entropyCoder->codeCoeffNxN(cu, coeff, absPartIdx, log2TrSizeC, ttype);
+    }
+    else
+    {
+        uint32_t coeffOffset = absPartIdx << (LOG2_UNIT_SIZE * 2 - 1);
+        coeff_t* coeff = m_qtTempCoeff[ttype][qtLayer] + coeffOffset;
+        uint32_t subTUSize = 1 << (log2TrSizeC * 2);
+        uint32_t partIdxesPerSubTU  = NUM_CU_PARTITIONS >> (((cu->getDepth(absPartIdx) + trDepthC) << 1) + 1);
+        if (cu->getCbf(absPartIdx, ttype, trDepth + 1))
+            m_entropyCoder->codeCoeffNxN(cu, coeff, absPartIdx, log2TrSizeC, ttype);
+        if (cu->getCbf(absPartIdx + partIdxesPerSubTU, ttype, trDepth + 1))
+            m_entropyCoder->codeCoeffNxN(cu, coeff + subTUSize, absPartIdx + partIdxesPerSubTU, log2TrSizeC, ttype);
+    }
+}
+
+void Search::xEncIntraHeaderLuma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx)
+{
+    // CU header
+    if (!absPartIdx)
+    {
+        if (!cu->m_slice->isIntra())
+        {
+            if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+                m_entropyCoder->codeCUTransquantBypassFlag(cu->getCUTransquantBypass(0));
+            m_entropyCoder->codeSkipFlag(cu, 0);
+            m_entropyCoder->codePredMode(cu->getPredictionMode(0));
+        }
+
+        m_entropyCoder->codePartSize(cu, 0, cu->getDepth(0));
+    }
+    // luma prediction mode
+    if (cu->getPartitionSize(0) == SIZE_2Nx2N)
+    {
+        if (!absPartIdx)
+            m_entropyCoder->codeIntraDirLumaAng(cu, 0, false);
+    }
+    else
+    {
+        uint32_t qtNumParts = cu->getTotalNumPart() >> 2;
+        if (!trDepth)
+        {
+            X265_CHECK(absPartIdx == 0, "unexpected absPartIdx %d\n", absPartIdx);
+            for (uint32_t part = 0; part < 4; part++)
+                m_entropyCoder->codeIntraDirLumaAng(cu, part * qtNumParts, false);
+        }
+        else if (!(absPartIdx & (qtNumParts - 1)))
+            m_entropyCoder->codeIntraDirLumaAng(cu, absPartIdx, false);
+    }
+}
+
+void Search::xEncIntraHeaderChroma(TComDataCU* cu, uint32_t absPartIdx)
+{
+    // chroma prediction mode
+    if (cu->getPartitionSize(0) == SIZE_2Nx2N || cu->getChromaFormat() != X265_CSP_I444)
+    {
+        if (!absPartIdx)
+            m_entropyCoder->codeIntraDirChroma(cu, absPartIdx);
+    }
+    else
+    {
+        uint32_t qtNumParts = cu->getTotalNumPart() >> 2;
+        if (!(absPartIdx & (qtNumParts - 1)))
+            m_entropyCoder->codeIntraDirChroma(cu, absPartIdx);
+    }
+}
+
+uint32_t Search::xGetIntraBitsQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t absPartIdxStep)
+{
+    int cuSize = 1 << cu->getLog2CUSize(absPartIdx);
+    m_entropyCoder->resetBits();
+    xEncIntraHeaderChroma(cu, absPartIdx);
+    xEncSubdivCbfQTChroma(cu, trDepth, absPartIdx, absPartIdxStep, cuSize, cuSize);
+    xEncCoeffQTChroma(cu, trDepth, absPartIdx, TEXT_CHROMA_U);
+    xEncCoeffQTChroma(cu, trDepth, absPartIdx, TEXT_CHROMA_V);
+    return m_entropyCoder->getNumberOfWrittenBits();
+}
+
+uint32_t Search::xGetIntraBitsLuma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t log2TrSize, coeff_t* coeff, uint32_t depthRange[2])
+{
+    m_entropyCoder->resetBits();
+    xEncIntraHeaderLuma(cu, trDepth, absPartIdx);
+
+    // Transform subdiv flag
+    if (log2TrSize != *depthRange)
+        m_entropyCoder->codeTransformSubdivFlag(0, 5 - log2TrSize);
+
+    // Cbfs
+    uint32_t trMode = cu->getTransformIdx(absPartIdx);
+    m_entropyCoder->codeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
+
+    if (cu->getCbf(absPartIdx, TEXT_LUMA, trDepth))
+        m_entropyCoder->codeCoeffNxN(cu, coeff, absPartIdx, log2TrSize, TEXT_LUMA);
+
+    return m_entropyCoder->getNumberOfWrittenBits();
+}
+
+uint32_t Search::xGetIntraBitsChroma(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSizeC, uint32_t chromaId, coeff_t* coeff)
+{
+    m_entropyCoder->resetBits();
+    m_entropyCoder->codeCoeffNxN(cu, coeff, absPartIdx, log2TrSizeC, (TextType)chromaId);
+    return m_entropyCoder->getNumberOfWrittenBits();
+}
+
+/* returns distortion */
+uint32_t Search::xIntraCodingLumaBlk(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSize, TComYuv* fencYuv, TComYuv* predYuv,
+                                     ShortYuv* resiYuv, int16_t* reconQt, uint32_t reconQtStride, coeff_t* coeff, uint32_t& cbf)
+{
+    uint32_t stride       = fencYuv->getStride();
+    pixel*   fenc         = fencYuv->getLumaAddr(absPartIdx);
+    pixel*   pred         = predYuv->getLumaAddr(absPartIdx);
+    int16_t* residual     = resiYuv->getLumaAddr(absPartIdx);
+
+    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
+    pixel*   reconIPred       = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+    uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getStride();
+    bool     useTransformSkip = !!cu->getTransformSkip(absPartIdx, TEXT_LUMA);
+    int      part = partitionFromLog2Size(log2TrSize);
+    int      sizeIdx = log2TrSize - 2;
+
+#if CHECKED_BUILD || _DEBUG
+    uint32_t tuSize       = 1 << log2TrSize;
+    X265_CHECK(!((intptr_t)fenc & (tuSize - 1)), "fenc alignment check fail\n");
+    X265_CHECK(!((intptr_t)pred & (tuSize - 1)), "pred alignment check fail\n");
+    X265_CHECK(!((intptr_t)residual & (tuSize - 1)), "residual alignment check fail\n");
+#endif
+    // get residual signal
+    primitives.calcresidual[sizeIdx](fenc, pred, residual, stride);
+
+    // transform and quantization
+    if (m_bEnableRDOQ)
+        m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSize, true);
+
+    uint32_t numSig = m_quant.transformNxN(cu, fenc, stride, residual, stride, coeff, log2TrSize, TEXT_LUMA, absPartIdx, useTransformSkip);
+    if (numSig)
+    {
+        X265_CHECK(log2TrSize <= 5, "log2TrSize is too large %d\n", log2TrSize);
+        m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), residual, stride, coeff, log2TrSize, TEXT_LUMA, true, useTransformSkip, numSig);
+        primitives.calcrecon[sizeIdx](pred, residual, reconQt, reconIPred, stride, reconQtStride, reconIPredStride);
+        cbf = 1;
+        return primitives.sse_sp[part](reconQt, reconQtStride, fenc, stride);
+    }
+    else
+    {
+#if CHECKED_BUILD || _DEBUG
+        memset(coeff, 0, sizeof(coeff_t) << (log2TrSize * 2));
+#endif
+        // reconstruction
+        primitives.square_copy_ps[sizeIdx](reconQt,    reconQtStride,    pred, stride);
+        primitives.square_copy_pp[sizeIdx](reconIPred, reconIPredStride, pred, stride);
+        cbf = 0;
+        return primitives.sse_pp[part](pred, stride, fenc, stride);
+    }
+}
+
+uint32_t Search::xIntraCodingChromaBlk(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, int16_t* reconQt,
+                                       uint32_t reconQtStride, coeff_t* coeff, uint32_t& cbf, uint32_t chromaId, uint32_t log2TrSizeC)
+{
+    TextType ttype        = (TextType)chromaId;
+    uint32_t stride       = fencYuv->getCStride();
+    pixel*   fenc         = fencYuv->getChromaAddr(chromaId, absPartIdx);
+    pixel*   pred         = predYuv->getChromaAddr(chromaId, absPartIdx);
+    int16_t* residual     = resiYuv->getChromaAddr(chromaId, absPartIdx);
+
+    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
+    pixel*   reconIPred       = cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), zorder);
+    uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+    bool     useTransformSkipC = !!cu->getTransformSkip(absPartIdx, ttype);
+    int      part = partitionFromLog2Size(log2TrSizeC);
+    int      sizeIdxC = log2TrSizeC - 2;
+    uint32_t dist;
+
+#if CHECKED_BUILD || _DEBUG
+    uint32_t tuSize       = 1 << log2TrSizeC;
+    X265_CHECK(!((intptr_t)fenc & (tuSize - 1)), "fenc alignment check fail\n");
+    X265_CHECK(!((intptr_t)pred & (tuSize - 1)), "pred alignment check fail\n");
+    X265_CHECK(!((intptr_t)residual & (tuSize - 1)), "residual alignment check fail\n");
+#endif
+    primitives.calcresidual[sizeIdxC](fenc, pred, residual, stride);
+
+    // init rate estimation arrays for RDOQ
+    if (m_bEnableRDOQ)
+        m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSizeC, false);
+
+    uint32_t numSig = m_quant.transformNxN(cu, fenc, stride, residual, stride, coeff, log2TrSizeC, ttype, absPartIdx, useTransformSkipC);
+
+    if (numSig)
+    {
+        X265_CHECK(log2TrSizeC <= 5, "log2TrSizeC is too large %d\n", log2TrSizeC);
+        m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), residual, stride, coeff, log2TrSizeC, ttype, true, useTransformSkipC, numSig);
+        primitives.calcrecon[sizeIdxC](pred, residual, reconQt, reconIPred, stride, reconQtStride, reconIPredStride);
+        cbf = 1;
+        dist = primitives.sse_sp[part](reconQt, reconQtStride, fenc, stride);
+    }
+    else
+    {
+#if CHECKED_BUILD || _DEBUG
+        memset(coeff, 0, sizeof(coeff_t) << (log2TrSizeC * 2));
+#endif
+        // reconstruction
+        primitives.square_copy_ps[sizeIdxC](reconQt,    reconQtStride,    pred, stride);
+        primitives.square_copy_pp[sizeIdxC](reconIPred, reconIPredStride, pred, stride);
+        cbf = 0;
+        dist = primitives.sse_pp[part](pred, stride, fenc, stride);
+    }
+
+    X265_CHECK(ttype == TEXT_CHROMA_U || ttype == TEXT_CHROMA_V, "invalid ttype\n");
+    return (ttype == TEXT_CHROMA_U) ? m_rdCost.scaleChromaDistCb(dist) : m_rdCost.scaleChromaDistCr(dist);
+}
+
+/* returns distortion. TODO reorder params */
+uint32_t Search::xRecurIntraCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, 
+                                     bool bAllowRQTSplit, uint64_t& rdCost, uint32_t& rdBits, uint32_t& psyEnergy, uint32_t depthRange[2])
+{
+    uint32_t fullDepth   = cu->getDepth(0) + trDepth;
+    uint32_t log2TrSize  = g_maxLog2CUSize - fullDepth;
+    uint32_t outDist     = 0;
+    bool bCheckSplit = log2TrSize > *depthRange;
+    bool bCheckFull  = log2TrSize <= *(depthRange + 1);
+
+    int isIntraSlice = (cu->m_slice->m_sliceType == I_SLICE);
+
+    // don't check split if TU size is less or equal to max TU size
+    bool noSplitIntraMaxTuSize = bCheckFull;
+
+    if (m_param->rdPenalty && !isIntraSlice)
+    {
+        int maxTuSize = cu->m_slice->m_sps->quadtreeTULog2MaxSize;
+        // in addition don't check split if TU size is less or equal to 16x16 TU size for non-intra slice
+        noSplitIntraMaxTuSize = (log2TrSize <= (uint32_t)X265_MIN(maxTuSize, 4));
+
+        // TODO: this check for tuQTMaxIntraDepth depth is a hack, it needs a better fix
+        if (m_param->rdPenalty == 2 && m_param->tuQTMaxIntraDepth > fullDepth)
+            // if maximum RD-penalty don't check TU size 32x32
+            bCheckFull = (log2TrSize <= (uint32_t)X265_MIN(maxTuSize, 4));
+    }
+    if (!bAllowRQTSplit && noSplitIntraMaxTuSize)
+        bCheckSplit = false;
+
+    uint64_t singleCost  = MAX_INT64;
+    uint32_t singleDistY = 0;
+    uint32_t singleBits  = 0;
+    uint32_t singlePsyEnergyY = 0;
+    uint32_t singleCbfY   = 0;
+    int      bestModeId   = 0;
+    bool     bestTQbypass = 0;
+
+    if (bCheckFull)
+    {
+        uint32_t tuSize = 1 << log2TrSize;
+
+        bool checkTransformSkip = (cu->m_slice->m_pps->bTransformSkipEnabled &&
+                                   log2TrSize <= MAX_LOG2_TS_SIZE &&
+                                   !cu->getCUTransquantBypass(0));
+        if (checkTransformSkip)
+        {
+            checkTransformSkip &= !((cu->getQP(0) == 0));
+            if (m_param->bEnableTSkipFast)
+                checkTransformSkip &= (cu->getPartitionSize(absPartIdx) == SIZE_NxN);
+        }
+
+        bool checkTQbypass = cu->m_slice->m_pps->bTransquantBypassEnabled && !m_param->bLossless;
+
+        // NOTE: transform_quant_bypass just at cu level
+        if ((cu->m_slice->m_pps->bTransquantBypassEnabled) && cu->getCUTransquantBypass(0) != checkTQbypass)
+            checkTQbypass = cu->getCUTransquantBypass(0) && !m_param->bLossless;
+
+        uint32_t stride = fencYuv->getStride();
+        pixel*   pred   = predYuv->getLumaAddr(absPartIdx);
+
+        // init availability pattern
+        uint32_t lumaPredMode = cu->getLumaIntraDir(absPartIdx);
+        TComPattern::initAdiPattern(cu, absPartIdx, trDepth, m_predBuf, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt, lumaPredMode);
+
+        // get prediction signal
+        predIntraLumaAng(lumaPredMode, pred, stride, log2TrSize);
+
+        cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
+
+        uint32_t qtLayer        = log2TrSize - 2;
+        uint32_t coeffOffsetY   = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        coeff_t* coeffY         = m_qtTempCoeff[0][qtLayer] + coeffOffsetY;
+        int16_t* reconQt        = m_qtTempShortYuv[qtLayer].getLumaAddr(absPartIdx);
+        X265_CHECK(m_qtTempShortYuv[qtLayer].m_width == MAX_CU_SIZE, "width is not max CU size\n");
+        const uint32_t reconQtStride = MAX_CU_SIZE;
+
+        if (checkTransformSkip || checkTQbypass)
+        {
+            // store original entropy coding status
+            m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+
+            uint32_t  singleDistYTmp = 0;
+            uint32_t  singlePsyEnergyYTmp = 0;
+            uint32_t  singleCbfYTmp  = 0;
+            uint64_t  singleCostTmp  = 0;
+            bool      singleTQbypass = 0;
+            const int firstCheckId   = 0;
+
+            ALIGN_VAR_32(coeff_t, tsCoeffY[32 * 32]);
+            ALIGN_VAR_32(int16_t, tsReconY[32 * 32]);
+
+            for (int modeId = firstCheckId; modeId < 2; modeId++)
+            {
+                coeff_t* coeff = (modeId ? tsCoeffY : coeffY);
+                int16_t* recon = (modeId ? tsReconY : reconQt);
+                uint32_t reconStride = (modeId ? tuSize : reconQtStride);
+
+                cu->setTransformSkipSubParts(checkTransformSkip ? modeId : 0, TEXT_LUMA, absPartIdx, fullDepth);
+
+                bool bIsLossLess = modeId != firstCheckId;
+                if ((cu->m_slice->m_pps->bTransquantBypassEnabled))
+                    cu->setCUTransquantBypassSubParts(bIsLossLess, absPartIdx, fullDepth);
+
+                // code luma block with given intra prediction mode and store Cbf
+                singleDistYTmp = xIntraCodingLumaBlk(cu, absPartIdx, log2TrSize, fencYuv, predYuv, resiYuv, recon, reconStride, coeff, singleCbfYTmp);
+                singlePsyEnergyYTmp = 0;
+                if (m_rdCost.m_psyRd)
+                {
+                    uint32_t zorder = cu->getZorderIdxInCU() + absPartIdx;
+                    singlePsyEnergyYTmp = m_rdCost.psyCost(log2TrSize - 2, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(),
+                        cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getStride());
+                }
+                cu->setCbfSubParts(singleCbfYTmp << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
+                singleTQbypass = cu->getCUTransquantBypass(absPartIdx);
+
+                if ((modeId == 1) && !singleCbfYTmp && checkTransformSkip)
+                    // In order not to code TS flag when cbf is zero, the case for TS with cbf being zero is forbidden.
+                    break;
+                else
+                {
+                    singleBits = xGetIntraBitsLuma(cu, trDepth, absPartIdx, log2TrSize, coeff, depthRange);
+                    if (m_rdCost.m_psyRd)
+                        singleCostTmp = m_rdCost.calcPsyRdCost(singleDistYTmp, singleBits, singlePsyEnergyYTmp);
+                    else
+                        singleCostTmp = m_rdCost.calcRdCost(singleDistYTmp, singleBits);
+                }
+
+                if (singleCostTmp < singleCost)
+                {
+                    singleCost   = singleCostTmp;
+                    singleDistY  = singleDistYTmp;
+                    singlePsyEnergyY = singlePsyEnergyYTmp;
+                    singleCbfY   = singleCbfYTmp;
+                    bestTQbypass = singleTQbypass;
+                    bestModeId   = modeId;
+                    if (bestModeId == firstCheckId)
+                        m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_TEMP_BEST]);
+                }
+                if (modeId == firstCheckId)
+                    m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+            }
+
+            cu->setTransformSkipSubParts(checkTransformSkip ? bestModeId : 0, TEXT_LUMA, absPartIdx, fullDepth);
+            if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+                cu->setCUTransquantBypassSubParts(bestTQbypass, absPartIdx, fullDepth);
+
+            if (bestModeId == firstCheckId)
+            {
+                xLoadIntraResultQT(cu, absPartIdx, log2TrSize, reconQt, reconQtStride);
+                cu->setCbfSubParts(singleCbfY << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
+                m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_TEMP_BEST]);
+            }
+            else
+            {
+                ::memcpy(coeffY, tsCoeffY, sizeof(coeff_t) << (log2TrSize * 2));
+                int sizeIdx = log2TrSize - 2;
+                primitives.square_copy_ss[sizeIdx](reconQt, reconQtStride, tsReconY, tuSize);
+            }
+        }
+        else
+        {
+            m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+
+            // code luma block with given intra prediction mode and store Cbf
+            cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, fullDepth);
+            singleDistY = xIntraCodingLumaBlk(cu, absPartIdx, log2TrSize, fencYuv, predYuv, resiYuv, reconQt, reconQtStride, coeffY, singleCbfY);
+            if (m_rdCost.m_psyRd)
+            {
+                uint32_t zorder = cu->getZorderIdxInCU() + absPartIdx;
+                singlePsyEnergyY = m_rdCost.psyCost(log2TrSize - 2, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(),
+                    cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getStride());
+            }
+            cu->setCbfSubParts(singleCbfY << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
+
+            singleBits = xGetIntraBitsLuma(cu, trDepth, absPartIdx, log2TrSize, coeffY, depthRange);
+            if (m_param->rdPenalty && (log2TrSize == 5) && !isIntraSlice)
+                singleBits *= 4;
+
+            if (m_rdCost.m_psyRd)
+                singleCost = m_rdCost.calcPsyRdCost(singleDistY, singleBits, singlePsyEnergyY);
+            else
+                singleCost = m_rdCost.calcRdCost(singleDistY, singleBits);
+        }
+    }
+
+    if (bCheckSplit)
+    {
+        // store full entropy coding status, load original entropy coding status
+        if (bCheckFull)
+        {
+            m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_TEST]);
+            m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+        }
+        else
+            m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+
+        // code splitted block
+        uint64_t splitCost     = 0;
+        uint32_t splitDistY    = 0;
+        uint32_t splitPsyEnergyY = 0;
+        uint32_t qPartsDiv     = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        uint32_t absPartIdxSub = absPartIdx;
+        uint32_t splitCbfY     = 0;
+        uint32_t splitBits     = 0;
+
+        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
+        {
+            splitDistY += xRecurIntraCodingQT(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, bAllowRQTSplit, splitCost, splitBits, 
+                                              splitPsyEnergyY, depthRange);
+            splitCbfY |= cu->getCbf(absPartIdxSub, TEXT_LUMA, trDepth + 1);
+        }
+
+        if (bCheckFull && log2TrSize != depthRange[0])
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeTransformSubdivFlag(1, 5 - log2TrSize);
+            splitBits += m_entropyCoder->getNumberOfWrittenBits();
+        }
+
+        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
+            cu->getCbf(TEXT_LUMA)[absPartIdx + offs] |= (splitCbfY << trDepth);
+
+        if (m_rdCost.m_psyRd)
+            splitCost = m_rdCost.calcPsyRdCost(splitDistY, splitBits, splitPsyEnergyY);
+        else
+            splitCost = m_rdCost.calcRdCost(splitDistY, splitBits);
+
+        if (splitCost < singleCost)
+        {
+            outDist   += splitDistY;
+            rdCost    += splitCost;
+            rdBits    += splitBits;
+            psyEnergy += splitPsyEnergyY;
+            return outDist;
+        }
+
+        // set entropy coding status
+        m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_TEST]);
+
+        // set transform index and Cbf values
+        cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
+        cu->setCbfSubParts(singleCbfY << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
+        cu->setTransformSkipSubParts(bestModeId, TEXT_LUMA, absPartIdx, fullDepth);
+
+        // set reconstruction for next intra prediction blocks
+        uint32_t qtLayer   = log2TrSize - 2;
+        uint32_t zorder    = cu->getZorderIdxInCU() + absPartIdx;
+        int16_t* reconQt   = m_qtTempShortYuv[qtLayer].getLumaAddr(absPartIdx);
+        X265_CHECK(m_qtTempShortYuv[qtLayer].m_width == MAX_CU_SIZE, "width is not max CU size\n");
+        const uint32_t reconQtStride = MAX_CU_SIZE;
+
+        pixel*   dst       = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+        uint32_t dststride = cu->m_pic->getPicYuvRec()->getStride();
+        int sizeIdx = log2TrSize - 2;
+        primitives.square_copy_sp[sizeIdx](dst, dststride, reconQt, reconQtStride);
+    }
+
+    rdCost    += singleCost;
+    rdBits    += singleBits;
+    psyEnergy += singlePsyEnergyY;
+    return outDist + singleDistY;
+}
+
+void Search::residualTransformQuantIntra(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv,
+                                         ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2])
+{
+    uint32_t fullDepth   = cu->getDepth(0) +  trDepth;
+    uint32_t log2TrSize  = g_maxLog2CUSize - fullDepth;
+    bool     bCheckFull  = log2TrSize <= depthRange[1];
+    bool     bCheckSplit = log2TrSize > depthRange[0];
+
+    int isIntraSlice = (cu->m_slice->m_sliceType == I_SLICE);
+
+    if (m_param->rdPenalty == 2 && !isIntraSlice)
+    {
+        int maxTuSize = cu->m_slice->m_sps->quadtreeTULog2MaxSize;
+        // if maximum RD-penalty don't check TU size 32x32
+        bCheckFull = (log2TrSize <= (uint32_t)X265_MIN(maxTuSize, 4));
+    }
+
+    if (bCheckFull)
+    {
+        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, fullDepth);
+
+        // code luma block with given intra prediction mode and store Cbf
+        uint32_t lumaPredMode = cu->getLumaIntraDir(absPartIdx);
+        uint32_t stride       = fencYuv->getStride();
+        pixel*   fenc         = fencYuv->getLumaAddr(absPartIdx);
+        pixel*   pred         = predYuv->getLumaAddr(absPartIdx);
+        int16_t* residual     = resiYuv->getLumaAddr(absPartIdx);
+        pixel*   recon        = reconYuv->getLumaAddr(absPartIdx);
+        uint32_t coeffOffsetY = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        coeff_t* coeff        = cu->getCoeffY() + coeffOffsetY;
+
+        uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
+        pixel*   reconIPred       = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+        uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getStride();
+
+        bool     useTransformSkip = !!cu->getTransformSkip(absPartIdx, TEXT_LUMA);
+
+        // init availability pattern
+        TComPattern::initAdiPattern(cu, absPartIdx, trDepth, m_predBuf, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt, lumaPredMode);
+        // get prediction signal
+        predIntraLumaAng(lumaPredMode, pred, stride, log2TrSize);
+
+        cu->setTrIdxSubParts(trDepth, absPartIdx, fullDepth);
+
+#if CHECKED_BUILD || _DEBUG
+        uint32_t tuSize       = 1 << log2TrSize;
+        X265_CHECK(!((intptr_t)fenc & (tuSize - 1)), "fenc alignment failure\n");
+        X265_CHECK(!((intptr_t)pred & (tuSize - 1)), "pred alignment failure\n");
+        X265_CHECK(!((intptr_t)residual & (tuSize - 1)), "residual alignment failure\n");
+#endif
+        int sizeIdx = log2TrSize - 2;
+        primitives.calcresidual[sizeIdx](fenc, pred, residual, stride);
+        uint32_t numSig = m_quant.transformNxN(cu, fenc, stride, residual, stride, coeff, log2TrSize, TEXT_LUMA, absPartIdx, useTransformSkip);
+
+        // set coded block flag
+        cu->setCbfSubParts((!!numSig) << trDepth, TEXT_LUMA, absPartIdx, fullDepth);
+
+        if (numSig)
+        {
+            m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), residual, stride, coeff, log2TrSize, TEXT_LUMA, true, useTransformSkip, numSig);
+
+            // Generate Recon
+            primitives.luma_add_ps[sizeIdx](recon, stride, pred, residual, stride, stride);
+            primitives.square_copy_pp[sizeIdx](reconIPred, reconIPredStride, recon, stride);
+        }
+        else
+        {
+#if CHECKED_BUILD || _DEBUG
+            memset(coeff, 0, sizeof(coeff_t) << (log2TrSize * 2));
+#endif
+
+            // Generate Recon
+            primitives.square_copy_pp[sizeIdx](recon,      stride,           pred, stride);
+            primitives.square_copy_pp[sizeIdx](reconIPred, reconIPredStride, pred, stride);
+        }
+    }
+
+    if (bCheckSplit && !bCheckFull)
+    {
+        // code splitted block
+
+        uint32_t qPartsDiv     = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        uint32_t absPartIdxSub = absPartIdx;
+        uint32_t splitCbfY     = 0;
+
+        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
+        {
+            residualTransformQuantIntra(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, reconYuv, depthRange);
+            splitCbfY |= cu->getCbf(absPartIdxSub, TEXT_LUMA, trDepth + 1);
+        }
+
+        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
+            cu->getCbf(TEXT_LUMA)[absPartIdx + offs] |= (splitCbfY << trDepth);
+    }
+}
+
+void Search::xSetIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv)
+{
+    uint32_t fullDepth = cu->getDepth(0) + trDepth;
+    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+
+    if (trMode == trDepth)
+    {
+        uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+        uint32_t qtLayer    = log2TrSize - 2;
+
+        // copy transform coefficients
+        uint32_t coeffOffsetY = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        coeff_t* coeffSrcY    = m_qtTempCoeff[0][qtLayer] + coeffOffsetY;
+        coeff_t* coeffDestY   = cu->getCoeffY()           + coeffOffsetY;
+        ::memcpy(coeffDestY, coeffSrcY, sizeof(coeff_t) << (log2TrSize * 2));
+
+        // copy reconstruction
+        m_qtTempShortYuv[qtLayer].copyPartToPartLuma(reconYuv, absPartIdx, log2TrSize);
+    }
+    else
+    {
+        uint32_t numQPart = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        for (uint32_t part = 0; part < 4; part++)
+            xSetIntraResultQT(cu, trDepth + 1, absPartIdx + part * numQPart, reconYuv);
+    }
+}
+
+void Search::xLoadIntraResultQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSize, int16_t* reconQt, uint32_t reconQtStride)
+{
+    // copy reconstruction
+    int sizeIdx = log2TrSize - 2;
+    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
+    pixel*   reconIPred       = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+    uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getStride();
+    primitives.square_copy_sp[sizeIdx](reconIPred, reconIPredStride, reconQt, reconQtStride);
+}
+
+void Search::xLoadIntraResultChromaQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSizeC, uint32_t chromaId,
+                                      int16_t* reconQt, uint32_t reconQtStride)
+{
+    X265_CHECK(chromaId == 1 || chromaId == 2, "invalid chroma id");
+
+    // copy reconstruction
+    int sizeIdxC = log2TrSizeC - 2;
+    uint32_t zorder           = cu->getZorderIdxInCU() + absPartIdx;
+    pixel*   reconIPred       = cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), zorder);
+    uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+    primitives.square_copy_sp[sizeIdxC](reconIPred, reconIPredStride, reconQt, reconQtStride);
+}
+
+void Search::offsetSubTUCBFs(TComDataCU* cu, TextType ttype, uint32_t trDepth, uint32_t absPartIdx)
+{
+    uint32_t depth = cu->getDepth(0);
+    uint32_t fullDepth = depth + trDepth;
+    uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+
+    uint32_t trDepthC = trDepth;
+    if (log2TrSize == 2 && cu->getChromaFormat() != X265_CSP_I444)
+    {
+        X265_CHECK(trDepthC > 0, "trDepthC invalid\n");
+        trDepthC--;
+    }
+
+    uint32_t partIdxesPerSubTU = (NUM_CU_PARTITIONS >> ((depth + trDepthC) << 1)) >> 1;
+
+    // move the CBFs down a level and set the parent CBF
+    uint8_t subTUCBF[2];
+    uint8_t combinedSubTUCBF = 0;
+
+    for (uint32_t subTU = 0; subTU < 2; subTU++)
+    {
+        const uint32_t subTUAbsPartIdx = absPartIdx + (subTU * partIdxesPerSubTU);
+
+        subTUCBF[subTU]   = cu->getCbf(subTUAbsPartIdx, ttype, trDepth);
+        combinedSubTUCBF |= subTUCBF[subTU];
+    }
+
+    for (uint32_t subTU = 0; subTU < 2; subTU++)
+    {
+        const uint32_t subTUAbsPartIdx = absPartIdx + (subTU * partIdxesPerSubTU);
+        const uint8_t compositeCBF = (subTUCBF[subTU] << 1) | combinedSubTUCBF;
+
+        cu->setCbfPartRange((compositeCBF << trDepth), ttype, subTUAbsPartIdx, partIdxesPerSubTU);
+    }
+}
+
+/* returns distortion */
+uint32_t Search::xRecurIntraChromaCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
+                                           uint32_t& psyEnergy)
+{
+    uint32_t fullDepth = cu->getDepth(0) + trDepth;
+    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+    uint32_t outDist = 0;
+
+    if (trMode == trDepth)
+    {        
+        uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+        uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+
+        uint32_t trDepthC = trDepth;
+        if (log2TrSize == 2 && m_csp != X265_CSP_I444)
+        {
+            X265_CHECK(trDepth > 0, "invalid trDepth\n");
+            trDepthC--;
+            log2TrSizeC++;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepthC) << 1);
+            bool bFirstQ = ((absPartIdx & (qpdiv - 1)) == 0);
+            if (!bFirstQ)
+                return outDist;
+        }
+
+        uint32_t qtLayer = log2TrSize - 2;
+        uint32_t tuSize = 1 << log2TrSizeC;
+        uint32_t stride = fencYuv->getCStride();
+        const bool splitIntoSubTUs = (m_csp == X265_CSP_I422);
+
+        bool checkTransformSkip = (cu->m_slice->m_pps->bTransformSkipEnabled &&
+                                   log2TrSizeC <= MAX_LOG2_TS_SIZE &&
+                                   !cu->getCUTransquantBypass(0));
+
+        if (m_param->bEnableTSkipFast)
+        {
+            checkTransformSkip &= (log2TrSize <= MAX_LOG2_TS_SIZE);
+            if (checkTransformSkip)
+            {
+                int nbLumaSkip = 0;
+                for (uint32_t absPartIdxSub = absPartIdx; absPartIdxSub < absPartIdx + 4; absPartIdxSub++)
+                    nbLumaSkip += cu->getTransformSkip(absPartIdxSub, TEXT_LUMA);
+
+                checkTransformSkip &= (nbLumaSkip > 0);
+            }
+        }
+        uint32_t singlePsyEnergy = 0;
+        for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+        {
+            uint32_t curPartNum = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepthC) << 1);
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, curPartNum, absPartIdx);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+                pixel*   pred        = predYuv->getChromaAddr(chromaId, absPartIdxC);
+
+                // init availability pattern
+                TComPattern::initAdiPatternChroma(cu, absPartIdxC, trDepthC, m_predBuf, chromaId);
+                pixel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, tuSize, m_predBuf);
+
+                uint32_t chromaPredMode = cu->getChromaIntraDir(absPartIdxC);
+                // update chroma mode
+                if (chromaPredMode == DM_CHROMA_IDX)
+                    chromaPredMode = cu->getLumaIntraDir((m_csp == X265_CSP_I444) ? absPartIdxC : 0);
+                chromaPredMode = (m_csp == X265_CSP_I422) ? g_chroma422IntraAngleMappingTable[chromaPredMode] : chromaPredMode;
+                // get prediction signal
+                predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, log2TrSizeC, m_csp);
+
+                uint32_t singleCbfC     = 0;
+                uint32_t singlePsyEnergyTmp = 0;
+
+                int16_t* reconQt        = m_qtTempShortYuv[qtLayer].getChromaAddr(chromaId, absPartIdxC);
+                uint32_t reconQtStride  = m_qtTempShortYuv[qtLayer].m_cwidth;
+                uint32_t coeffOffsetC   = absPartIdxC << (LOG2_UNIT_SIZE * 2 - (hChromaShift + vChromaShift));
+                coeff_t* coeffC         = m_qtTempCoeff[chromaId][qtLayer] + coeffOffsetC;
+
+                if (checkTransformSkip)
+                {
+                    // use RDO to decide whether Cr/Cb takes TS
+                    m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+
+                    uint64_t singleCost     = MAX_INT64;
+                    int      bestModeId     = 0;
+                    uint32_t singleDistC    = 0;
+                    uint32_t singleDistCTmp = 0;
+                    uint64_t singleCostTmp  = 0;
+                    uint32_t singleCbfCTmp  = 0;
+
+                    const int firstCheckId  = 0;
+
+                    ALIGN_VAR_32(coeff_t, tsCoeffC[MAX_TS_SIZE * MAX_TS_SIZE]);
+                    ALIGN_VAR_32(int16_t, tsReconC[MAX_TS_SIZE * MAX_TS_SIZE]);
+
+                    for (int chromaModeId = firstCheckId; chromaModeId < 2; chromaModeId++)
+                    {
+                        coeff_t* coeff = (chromaModeId ? tsCoeffC : coeffC);
+                        int16_t* recon = (chromaModeId ? tsReconC : reconQt);
+                        uint32_t reconStride = (chromaModeId ? tuSize : reconQtStride);
+
+                        cu->setTransformSkipPartRange(chromaModeId, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+
+                        singleDistCTmp = xIntraCodingChromaBlk(cu, absPartIdxC, fencYuv, predYuv, resiYuv, recon, reconStride, coeff, singleCbfCTmp, chromaId, log2TrSizeC);
+                        cu->setCbfPartRange(singleCbfCTmp << trDepth, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+
+                        if (chromaModeId == 1 && !singleCbfCTmp)
+                            //In order not to code TS flag when cbf is zero, the case for TS with cbf being zero is forbidden.
+                            break;
+                        else
+                        {
+                            uint32_t bitsTmp = singleCbfCTmp ? xGetIntraBitsChroma(cu, absPartIdxC, log2TrSizeC, chromaId, coeff) : 0;
+                            if (m_rdCost.m_psyRd)
+                            {
+                                uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                                singlePsyEnergyTmp = m_rdCost.psyCost(log2TrSizeC - 2, fencYuv->getChromaAddr(chromaId, absPartIdxC), fencYuv->getCStride(),
+                                    cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getCStride());
+                                singleCostTmp = m_rdCost.calcPsyRdCost(singleDistCTmp, bitsTmp, singlePsyEnergyTmp);
+                            }
+                            else
+                                singleCostTmp = m_rdCost.calcRdCost(singleDistCTmp, bitsTmp);
+                        }
+
+                        if (singleCostTmp < singleCost)
+                        {
+                            singleCost  = singleCostTmp;
+                            singleDistC = singleDistCTmp;
+                            bestModeId  = chromaModeId;
+                            singleCbfC  = singleCbfCTmp;
+                            singlePsyEnergy = singlePsyEnergyTmp;
+                            if (bestModeId == firstCheckId)
+                                m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_TEMP_BEST]);
+                        }
+                        if (chromaModeId == firstCheckId)
+                            m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+                    }
+
+                    if (bestModeId == firstCheckId)
+                    {
+                        xLoadIntraResultChromaQT(cu, absPartIdxC, log2TrSizeC, chromaId, reconQt, reconQtStride);
+                        cu->setCbfPartRange(singleCbfC << trDepth, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+                        m_entropyCoder->load(m_rdEntropyCoders[fullDepth][CI_TEMP_BEST]);
+                    }
+                    else
+                    {
+                        ::memcpy(coeffC, tsCoeffC, sizeof(coeff_t) << (log2TrSizeC * 2));
+                        int sizeIdxC = log2TrSizeC - 2;
+                        primitives.square_copy_ss[sizeIdxC](reconQt, reconQtStride, tsReconC, tuSize);
+                    }
+
+                    cu->setTransformSkipPartRange(bestModeId, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+
+                    outDist += singleDistC;
+
+                    if (chromaId == 1)
+                        m_entropyCoder->store(m_rdEntropyCoders[fullDepth][CI_QT_TRAFO_ROOT]);
+                }
+                else
+                {
+                    cu->setTransformSkipPartRange(0, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+                    outDist += xIntraCodingChromaBlk(cu, absPartIdxC, fencYuv, predYuv, resiYuv, reconQt, reconQtStride, coeffC, singleCbfC, chromaId, log2TrSizeC);
+                    if (m_rdCost.m_psyRd)
+                    {
+                        uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                        singlePsyEnergyTmp = m_rdCost.psyCost(log2TrSizeC - 2, fencYuv->getChromaAddr(chromaId, absPartIdxC), fencYuv->getCStride(),
+                            cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getCStride());
+                    }
+                    cu->setCbfPartRange(singleCbfC << trDepth, (TextType)chromaId, absPartIdxC, tuIterator.absPartIdxStep);
+                }
+                singlePsyEnergy += singlePsyEnergyTmp;
+            }
+            while (tuIterator.isNextSection());
+
+            if (splitIntoSubTUs)
+                offsetSubTUCBFs(cu, (TextType)chromaId, trDepth, absPartIdx);
+        }
+        psyEnergy = singlePsyEnergy;
+    }
+    else
+    {
+        uint32_t splitCbfU     = 0;
+        uint32_t splitCbfV     = 0;
+        uint32_t splitPsyEnergy = 0;
+        uint32_t qPartsDiv     = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        uint32_t absPartIdxSub = absPartIdx;
+        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
+        {
+            uint32_t psyEnergyTemp = 0;
+            outDist += xRecurIntraChromaCodingQT(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, psyEnergyTemp);
+            splitPsyEnergy += psyEnergyTemp;
+            splitCbfU |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_U, trDepth + 1);
+            splitCbfV |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_V, trDepth + 1);
+        }
+
+        psyEnergy = splitPsyEnergy;
+        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
+        {
+            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + offs] |= (splitCbfU << trDepth);
+            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + offs] |= (splitCbfV << trDepth);
+        }
+    }
+
+    return outDist;
+}
+
+void Search::xSetIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv)
+{
+    uint32_t fullDepth = cu->getDepth(0) + trDepth;
+    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    if (trMode == trDepth)
+    {
+        uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+        uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+
+        if (log2TrSize == 2 && m_csp != X265_CSP_I444)
+        {
+            X265_CHECK(trDepth > 0, "invalid trDepth\n");
+            trDepth--;
+            log2TrSizeC++;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepth) << 1);
+            if (absPartIdx & (qpdiv - 1))
+                return;
+        }
+
+        // copy transform coefficients
+
+        uint32_t numCoeffC = 1 << (log2TrSizeC * 2 + (m_csp == X265_CSP_I422));
+        uint32_t coeffOffsetC = absPartIdx << (LOG2_UNIT_SIZE * 2 - (hChromaShift + vChromaShift));
+
+        uint32_t qtLayer   = log2TrSize - 2;
+        coeff_t* coeffSrcU = m_qtTempCoeff[1][qtLayer] + coeffOffsetC;
+        coeff_t* coeffSrcV = m_qtTempCoeff[2][qtLayer] + coeffOffsetC;
+        coeff_t* coeffDstU = cu->getCoeffCb()          + coeffOffsetC;
+        coeff_t* coeffDstV = cu->getCoeffCr()          + coeffOffsetC;
+        ::memcpy(coeffDstU, coeffSrcU, sizeof(coeff_t) * numCoeffC);
+        ::memcpy(coeffDstV, coeffSrcV, sizeof(coeff_t) * numCoeffC);
+
+        // copy reconstruction
+        m_qtTempShortYuv[qtLayer].copyPartToPartChroma(reconYuv, absPartIdx, log2TrSizeC + hChromaShift);
+    }
+    else
+    {
+        uint32_t numQPart = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        for (uint32_t part = 0; part < 4; part++)
+            xSetIntraResultChromaQT(cu, trDepth + 1, absPartIdx + part * numQPart, reconYuv);
+    }
+}
+
+void Search::residualQTIntraChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx,
+                                   TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv)
+{
+    uint32_t fullDepth = cu->getDepth(0) + trDepth;
+    uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+    
+    if (trMode == trDepth)
+    {
+        uint32_t log2TrSize = g_maxLog2CUSize - fullDepth;
+        uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+        uint32_t trDepthC = trDepth;
+        if (log2TrSize == 2 && m_csp != X265_CSP_I444)
+        {
+            X265_CHECK(trDepth > 0, "invalid trDepth\n");
+            trDepthC--;
+            log2TrSizeC++;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepthC) << 1);
+            bool bFirstQ = ((absPartIdx & (qpdiv - 1)) == 0);
+            if (!bFirstQ)
+                return;
+        }
+
+        uint32_t tuSize = 1 << log2TrSizeC;
+        uint32_t stride = fencYuv->getCStride();
+        const bool splitIntoSubTUs = (m_csp == X265_CSP_I422);
+        const int sizeIdxC = log2TrSizeC - 2;
+
+        for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+        {
+            uint32_t curPartNum = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trDepthC) << 1);
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, curPartNum, absPartIdx);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+
+                TextType ttype          = (TextType)chromaId;
+                pixel*   fenc           = fencYuv->getChromaAddr(chromaId, absPartIdxC);
+                pixel*   pred           = predYuv->getChromaAddr(chromaId, absPartIdxC);
+                int16_t* residual       = resiYuv->getChromaAddr(chromaId, absPartIdxC);
+                pixel*   recon          = reconYuv->getChromaAddr(chromaId, absPartIdxC);
+                uint32_t coeffOffsetC   = absPartIdxC << (LOG2_UNIT_SIZE * 2 - (hChromaShift + vChromaShift));
+                coeff_t* coeff          = cu->getCoeff(ttype) + coeffOffsetC;
+                uint32_t zorder         = cu->getZorderIdxInCU() + absPartIdxC;
+                pixel*   reconIPred     = cu->m_pic->getPicYuvRec()->getChromaAddr(chromaId, cu->getAddr(), zorder);
+                uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+
+                const bool useTransformSkipC = !!cu->getTransformSkip(absPartIdxC, ttype);
+                cu->setTransformSkipPartRange(0, ttype, absPartIdxC, tuIterator.absPartIdxStep);
+
+                uint32_t chromaPredMode = cu->getChromaIntraDir(absPartIdxC);
+
+                // update chroma mode
+                if (chromaPredMode == DM_CHROMA_IDX)
+                    chromaPredMode = cu->getLumaIntraDir((m_csp == X265_CSP_I444) ? absPartIdxC : 0);
+                chromaPredMode = (m_csp == X265_CSP_I422) ? g_chroma422IntraAngleMappingTable[chromaPredMode] : chromaPredMode;
+                // init availability pattern
+                TComPattern::initAdiPatternChroma(cu, absPartIdxC, trDepthC, m_predBuf, chromaId);
+                pixel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, tuSize, m_predBuf);
+
+                // get prediction signal
+                predIntraChromaAng(chromaPred, chromaPredMode, pred, stride, log2TrSizeC, m_csp);
+
+                // get residual signal
+                X265_CHECK(!((intptr_t)fenc & (tuSize - 1)), "fenc alignment failure\n");
+                X265_CHECK(!((intptr_t)pred & (tuSize - 1)), "pred alignment failure\n");
+                X265_CHECK(!((intptr_t)residual & (tuSize - 1)), "residual alignment failure\n");
+                primitives.calcresidual[sizeIdxC](fenc, pred, residual, stride);
+
+                uint32_t numSig = m_quant.transformNxN(cu, fenc, stride, residual, stride, coeff, log2TrSizeC, ttype, absPartIdxC, useTransformSkipC);
+
+                cu->setCbfPartRange((!!numSig) << trDepth, ttype, absPartIdxC, tuIterator.absPartIdxStep);
+
+                if (numSig)
+                {
+                    // inverse transform
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), residual, stride, coeff, log2TrSizeC, ttype, true, useTransformSkipC, numSig);
+
+                    // reconstruction
+                    primitives.chroma[X265_CSP_I444].add_ps[sizeIdxC](recon, stride, pred, residual, stride, stride);
+                    primitives.square_copy_pp[sizeIdxC](reconIPred, reconIPredStride, recon, stride);
+                }
+                else
+                {
+#if CHECKED_BUILD || _DEBUG
+                    memset(coeff, 0, sizeof(coeff_t) * tuSize * tuSize);
+#endif
+                    primitives.square_copy_pp[sizeIdxC](recon,      stride,           pred, stride);
+                    primitives.square_copy_pp[sizeIdxC](reconIPred, reconIPredStride, pred, stride);
+                }
+            }
+            while (tuIterator.isNextSection());
+
+            if (splitIntoSubTUs)
+                offsetSubTUCBFs(cu, (TextType)chromaId, trDepth, absPartIdx);
+        }
+    }
+    else
+    {
+        uint32_t splitCbfU     = 0;
+        uint32_t splitCbfV     = 0;
+        uint32_t qPartsDiv     = NUM_CU_PARTITIONS >> ((fullDepth + 1) << 1);
+        uint32_t absPartIdxSub = absPartIdx;
+        for (uint32_t part = 0; part < 4; part++, absPartIdxSub += qPartsDiv)
+        {
+            residualQTIntraChroma(cu, trDepth + 1, absPartIdxSub, fencYuv, predYuv, resiYuv, reconYuv);
+            splitCbfU |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_U, trDepth + 1);
+            splitCbfV |= cu->getCbf(absPartIdxSub, TEXT_CHROMA_V, trDepth + 1);
+        }
+
+        for (uint32_t offs = 0; offs < 4 * qPartsDiv; offs++)
+        {
+            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + offs] |= (splitCbfU << trDepth);
+            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + offs] |= (splitCbfV << trDepth);
+        }
+    }
+}
+
+void Search::estIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2])
+{
+    uint32_t depth        = cu->getDepth(0);
+    uint32_t initTrDepth  = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
+    uint32_t numPU        = 1 << (2 * initTrDepth);
+    uint32_t log2TrSize   = cu->getLog2CUSize(0) - initTrDepth;
+    uint32_t tuSize       = 1 << log2TrSize;
+    uint32_t qNumParts    = cu->getTotalNumPart() >> 2;
+    uint32_t sizeIdx      = log2TrSize - 2;
+    uint32_t partOffset   = 0;
+    uint32_t srcstride   = reconYuv->getStride();
+    uint32_t dststride   = cu->m_pic->getPicYuvRec()->getStride();
+
+    // loop over partitions
+    for (uint32_t pu = 0; pu < numPU; pu++, partOffset += qNumParts)
+    {
+        // Reference sample smoothing
+        TComPattern::initAdiPattern(cu, partOffset, initTrDepth, m_predBuf, m_refAbove, m_refLeft, m_refAboveFlt, m_refLeftFlt, ALL_IDX);
+
+        // determine set of modes to be tested (using prediction signal only)
+        pixel*   fenc   = fencYuv->getLumaAddr(partOffset);
+        uint32_t stride = predYuv->getStride();
+
+        pixel *above         = m_refAbove    + tuSize - 1;
+        pixel *aboveFiltered = m_refAboveFlt + tuSize - 1;
+        pixel *left          = m_refLeft     + tuSize - 1;
+        pixel *leftFiltered  = m_refLeftFlt  + tuSize - 1;
+
+        // 33 Angle modes once
+        ALIGN_VAR_32(pixel, buf_trans[32 * 32]);
+        ALIGN_VAR_32(pixel, tmp[33 * 32 * 32]);
+        ALIGN_VAR_32(pixel, bufScale[32 * 32]);
+        pixel _above[4 * 32 + 1];
+        pixel _left[4 * 32 + 1];
+        int scaleTuSize = tuSize;
+        int scaleStride = stride;
+        int costShift = 0;
+
+        if (tuSize > 32)
+        {
+            pixel *aboveScale  = _above + 2 * 32;
+            pixel *leftScale   = _left + 2 * 32;
+
+            // origin is 64x64, we scale to 32x32 and setup required parameters
+            primitives.scale2D_64to32(bufScale, fenc, stride);
+            fenc = bufScale;
+
+            // reserve space in case primitives need to store data in above
+            // or left buffers
+            aboveScale[0] = leftScale[0] = above[0];
+            primitives.scale1D_128to64(aboveScale + 1, above + 1, 0);
+            primitives.scale1D_128to64(leftScale + 1, left + 1, 0);
+
+            scaleTuSize = 32;
+            scaleStride = 32;
+            costShift = 2;
+            sizeIdx = 5 - 2; // log2(scaleTuSize) - 2
+
+            // Filtered and Unfiltered refAbove and refLeft pointing to above and left.
+            above         = aboveScale;
+            left          = leftScale;
+            aboveFiltered = aboveScale;
+            leftFiltered  = leftScale;
+        }
+
+        uint32_t preds[3];
+        cu->getIntraDirLumaPredictor(partOffset, preds);
+        
+        uint64_t mpms;
+        uint32_t rbits = getIntraRemModeBits(cu, partOffset, depth, preds, mpms);
+
+        pixelcmp_t sa8d = primitives.sa8d[sizeIdx];
+        uint64_t modeCosts[35];
+        uint64_t bcost;
+
+        // DC
+        primitives.intra_pred[DC_IDX][sizeIdx](tmp, scaleStride, left, above, 0, (scaleTuSize <= 16));
+        uint32_t bits = (mpms & ((uint64_t)1 << DC_IDX)) ? getIntraModeBits(cu, DC_IDX, partOffset, depth) : rbits;
+        uint32_t sad  = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
+        modeCosts[DC_IDX] = bcost = m_rdCost.calcRdSADCost(sad, bits);
+
+        // PLANAR
+        pixel *abovePlanar = above;
+        pixel *leftPlanar  = left;
+        if (tuSize >= 8 && tuSize <= 32)
+        {
+            abovePlanar = aboveFiltered;
+            leftPlanar  = leftFiltered;
+        }
+        primitives.intra_pred[PLANAR_IDX][sizeIdx](tmp, scaleStride, leftPlanar, abovePlanar, 0, 0);
+        bits = (mpms & ((uint64_t)1 << PLANAR_IDX)) ? getIntraModeBits(cu, PLANAR_IDX, partOffset, depth) : rbits;
+        sad  = sa8d(fenc, scaleStride, tmp, scaleStride) << costShift;
+        modeCosts[PLANAR_IDX] = m_rdCost.calcRdSADCost(sad, bits);
+        COPY1_IF_LT(bcost, modeCosts[PLANAR_IDX]);
+
+        // angular predictions
+        primitives.intra_pred_allangs[sizeIdx](tmp, above, left, aboveFiltered, leftFiltered, (scaleTuSize <= 16));
+
+        primitives.transpose[sizeIdx](buf_trans, fenc, scaleStride);
+        for (int mode = 2; mode < 35; mode++)
+        {
+            bool modeHor = (mode < 18);
+            pixel *cmp = (modeHor ? buf_trans : fenc);
+            intptr_t srcStride = (modeHor ? scaleTuSize : scaleStride);
+            bits = (mpms & ((uint64_t)1 << mode)) ? getIntraModeBits(cu, mode, partOffset, depth) : rbits;
+            sad = sa8d(cmp, srcStride, &tmp[(mode - 2) * (scaleTuSize * scaleTuSize)], scaleTuSize) << costShift;
+            modeCosts[mode] = m_rdCost.calcRdSADCost(sad, bits);
+            COPY1_IF_LT(bcost, modeCosts[mode]);
+        }
+
+        /* Find the top maxCandCount candidate modes with cost within 25% of best
+         * or among the most probable modes. maxCandCount is derived from the
+         * rdLevel and depth. In general we want to try more modes at slower RD
+         * levels and at higher depths */
+        uint64_t candCostList[MAX_RD_INTRA_MODES];
+        uint32_t rdModeList[MAX_RD_INTRA_MODES];
+        int maxCandCount = 2 + m_param->rdLevel + ((depth + initTrDepth) >> 1);
+        for (int i = 0; i < maxCandCount; i++)
+            candCostList[i] = MAX_INT64;
+
+        uint64_t paddedBcost = bcost + (bcost >> 3); // 1.12%
+        for (int mode = 0; mode < 35; mode++)
+            if (modeCosts[mode] < paddedBcost || (mpms & ((uint64_t)1 << mode)))
+                updateCandList(mode, modeCosts[mode], maxCandCount, rdModeList, candCostList);
+
+        /* measure best candidates using simple RDO (no TU splits) */
+        uint32_t bmode = 0;
+        uint64_t cost;
+        bcost = MAX_INT64;
+        for (int i = 0; i < maxCandCount; i++)
+        {
+            if (candCostList[i] == MAX_INT64)
+                break;
+            m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+            cu->setLumaIntraDirSubParts(rdModeList[i], partOffset, depth + initTrDepth);
+            cost = bits = 0;
+            uint32_t psyEnergy = 0;
+            xRecurIntraCodingQT(cu, initTrDepth, partOffset, fencYuv, predYuv, resiYuv, false, cost, bits, psyEnergy, depthRange);
+            COPY2_IF_LT(bcost, cost, bmode, rdModeList[i]);
+        }
+
+        /* remeasure best mode, allowing TU splits */
+        cu->setLumaIntraDirSubParts(bmode, partOffset, depth + initTrDepth);
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+        uint32_t psyEnergy = 0;
+        // update distortion (rate and r-d costs are determined later)
+        cu->m_totalDistortion += xRecurIntraCodingQT(cu, initTrDepth, partOffset, fencYuv, predYuv, resiYuv, true, cost, bits, psyEnergy, depthRange);
+
+        xSetIntraResultQT(cu, initTrDepth, partOffset, reconYuv);
+
+        // set reconstruction for next intra prediction blocks
+        if (pu != numPU - 1)
+        {
+            uint32_t zorder      = cu->getZorderIdxInCU() + partOffset;
+            pixel*   dst         = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+            pixel*   src         = reconYuv->getLumaAddr(partOffset);
+            primitives.square_copy_pp[log2TrSize - 2](dst, dststride, src, srcstride);
+        }
+    }
+
+    if (numPU > 1)
+    {
+        uint32_t combCbfY = 0;
+        uint32_t partIdx  = 0;
+        for (uint32_t part = 0; part < 4; part++, partIdx += qNumParts)
+            combCbfY |= cu->getCbf(partIdx, TEXT_LUMA, 1);
+
+        for (uint32_t offs = 0; offs < 4 * qNumParts; offs++)
+            cu->getCbf(TEXT_LUMA)[offs] |= combCbfY;
+    }
+
+    // reset context models
+    m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+    x265_emms();
+}
+
+void Search::sharedEstIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2], uint8_t* sharedModes)
+{
+    uint32_t depth       = cu->getDepth(0);
+    uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
+    uint32_t numPU       = 1 << (2 * initTrDepth);
+    uint32_t log2TrSize  = cu->getLog2CUSize(0) - initTrDepth;
+    uint32_t qNumParts   = cu->getTotalNumPart() >> 2;
+
+    // loop over partitions
+    uint32_t partOffset  = 0;
+    uint64_t puCost      = 0;
+    uint32_t bits        = 0;
+    uint32_t dststride   = cu->m_pic->getPicYuvRec()->getStride();
+    uint32_t srcstride   = reconYuv->getStride();
+
+    for (uint32_t pu = 0; pu < numPU; pu++, partOffset += qNumParts)
+    {
+        cu->setLumaIntraDirSubParts(sharedModes[pu], partOffset, depth + initTrDepth);
+
+        // set context models
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+        uint32_t psyEnergy = 0;
+        // update overall distortion (rate and r-d costs are determined later)
+        cu->m_totalDistortion += xRecurIntraCodingQT(cu, initTrDepth, partOffset, fencYuv, predYuv, resiYuv, true, puCost, bits, psyEnergy, depthRange);
+        xSetIntraResultQT(cu, initTrDepth, partOffset, reconYuv);
+
+        if (pu != numPU - 1)
+        {
+            uint32_t zorder      = cu->getZorderIdxInCU() + partOffset;
+            pixel*   dst         = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+            pixel*   src         = reconYuv->getLumaAddr(partOffset);
+            primitives.luma_copy_pp[log2TrSize - 2](dst, dststride, src, srcstride);
+        }
+
+        // update PU data
+        cu->setLumaIntraDirSubParts(sharedModes[pu], partOffset, depth + initTrDepth);
+        cu->copyToPic((uint8_t)depth, pu, initTrDepth);
+    }
+
+    if (numPU > 1)
+    {
+        // set Cbf for all blocks
+        uint32_t combCbfY = 0;
+        uint32_t partIdx  = 0;
+        for (uint32_t part = 0; part < 4; part++, partIdx += qNumParts)
+            combCbfY |= cu->getCbf(partIdx, TEXT_LUMA, 1);
+
+        for (uint32_t offs = 0; offs < 4 * qNumParts; offs++)
+            cu->getCbf(TEXT_LUMA)[offs] |= combCbfY;
+    }
+
+    // reset context models
+    m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+}
+
+void Search::getBestIntraModeChroma(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv)
+{
+    uint32_t bestMode  = 0;
+    uint64_t bestCost  = MAX_INT64;
+    uint32_t modeList[NUM_CHROMA_MODE];
+
+    uint32_t log2TrSizeC = cu->getLog2CUSize(0) - CHROMA_H_SHIFT(m_csp);
+    uint32_t tuSize = 1 << log2TrSizeC;
+    int32_t scaleTuSize = tuSize;
+    int32_t costShift = 0;
+
+    if (tuSize > 32)
+    {
+        scaleTuSize = 32;
+        costShift = 2;
+        log2TrSizeC = 5;
+    }
+    int32_t sizeIdx = log2TrSizeC - 2;
+    pixelcmp_t sa8d = primitives.sa8d[sizeIdx];
+
+    TComPattern::initAdiPatternChroma(cu, 0, 0, m_predBuf, 1);
+    TComPattern::initAdiPatternChroma(cu, 0, 0, m_predBuf, 2);
+    cu->getAllowedChromaDir(0, modeList);
+
+    // check chroma modes
+    for (uint32_t mode = 0; mode < NUM_CHROMA_MODE; mode++)
+    {
+        uint32_t chromaPredMode = modeList[mode];
+        if (chromaPredMode == DM_CHROMA_IDX)
+            chromaPredMode = cu->getLumaIntraDir(0);
+        chromaPredMode = (m_csp == X265_CSP_I422) ? g_chroma422IntraAngleMappingTable[chromaPredMode] : chromaPredMode;
+        uint64_t cost = 0;
+        for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+        {
+            pixel* fenc = fencYuv->getChromaAddr(chromaId, 0);
+            pixel* pred = predYuv->getChromaAddr(chromaId, 0);
+            pixel* chromaPred = TComPattern::getAdiChromaBuf(chromaId, scaleTuSize, m_predBuf);
+
+            // get prediction signal
+            predIntraChromaAng(chromaPred, chromaPredMode, pred, fencYuv->getCStride(), log2TrSizeC, m_csp);
+            cost += sa8d(fenc, fencYuv->getCStride(), pred, fencYuv->getCStride()) << costShift;
+        }
+
+        if (cost < bestCost)
+        {
+            bestCost = cost;
+            bestMode = modeList[mode];
+        }
+    }
+
+    cu->setChromIntraDirSubParts(bestMode, 0, cu->getDepth(0));
+}
+
+void Search::estIntraPredChromaQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv)
+{
+    uint32_t depth       = cu->getDepth(0);
+    uint32_t initTrDepth = (cu->getPartitionSize(0) != SIZE_2Nx2N) && (cu->getChromaFormat() == X265_CSP_I444 ? 1 : 0);
+    uint32_t log2TrSize  = cu->getLog2CUSize(0) - initTrDepth;
+    uint32_t absPartIdx  = (NUM_CU_PARTITIONS >> (depth << 1));
+
+    int part = partitionFromLog2Size(log2TrSize);
+
+    TURecurse tuIterator((initTrDepth == 0) ? DONT_SPLIT : QUAD_SPLIT, absPartIdx, 0);
+
+    do
+    {
+        uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+
+        uint32_t bestMode = 0;
+        uint32_t bestDist = 0;
+        uint64_t bestCost = MAX_INT64;
+
+        // init mode list
+        uint32_t minMode = 0;
+        uint32_t maxMode = NUM_CHROMA_MODE;
+        uint32_t modeList[NUM_CHROMA_MODE];
+
+        cu->getAllowedChromaDir(absPartIdxC, modeList);
+
+        // check chroma modes
+        for (uint32_t mode = minMode; mode < maxMode; mode++)
+        {
+            // restore context models
+            m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+            // chroma coding
+            cu->setChromIntraDirSubParts(modeList[mode], absPartIdxC, depth + initTrDepth);
+
+            uint32_t psyEnergy = 0;
+            uint32_t dist = xRecurIntraChromaCodingQT(cu, initTrDepth, absPartIdxC, fencYuv, predYuv, resiYuv, psyEnergy);
+
+            if (cu->m_slice->m_pps->bTransformSkipEnabled)
+                m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+            uint32_t bits = xGetIntraBitsQTChroma(cu, initTrDepth, absPartIdxC, tuIterator.absPartIdxStep);
+            uint64_t cost = 0; 
+            if (m_rdCost.m_psyRd)
+                cost = m_rdCost.calcPsyRdCost(dist, bits, psyEnergy);
+            else
+                cost = m_rdCost.calcRdCost(dist, bits);
+
+            if (cost < bestCost)
+            {
+                bestCost = cost;
+                bestDist = dist;
+                bestMode = modeList[mode];
+                xSetIntraResultChromaQT(cu, initTrDepth, absPartIdxC, reconYuv);
+                ::memcpy(m_qtTempCbf[1], cu->getCbf(TEXT_CHROMA_U) + absPartIdxC, tuIterator.absPartIdxStep * sizeof(uint8_t));
+                ::memcpy(m_qtTempCbf[2], cu->getCbf(TEXT_CHROMA_V) + absPartIdxC, tuIterator.absPartIdxStep * sizeof(uint8_t));
+                ::memcpy(m_qtTempTransformSkipFlag[1], cu->getTransformSkip(TEXT_CHROMA_U) + absPartIdxC, tuIterator.absPartIdxStep * sizeof(uint8_t));
+                ::memcpy(m_qtTempTransformSkipFlag[2], cu->getTransformSkip(TEXT_CHROMA_V) + absPartIdxC, tuIterator.absPartIdxStep * sizeof(uint8_t));
+            }
+        }
+
+        if (!tuIterator.isLastSection())
+        {
+            uint32_t zorder      = cu->getZorderIdxInCU() + absPartIdxC;
+            uint32_t dststride   = cu->m_pic->getPicYuvRec()->getCStride();
+            uint32_t srcstride   = reconYuv->getCStride();
+            pixel *src, *dst;
+
+            dst = cu->m_pic->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder);
+            src = reconYuv->getCbAddr(absPartIdxC);
+            primitives.chroma[m_csp].copy_pp[part](dst, dststride, src, srcstride);
+
+            dst = cu->m_pic->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder);
+            src = reconYuv->getCrAddr(absPartIdxC);
+            primitives.chroma[m_csp].copy_pp[part](dst, dststride, src, srcstride);
+        }
+
+        ::memcpy(cu->getCbf(TEXT_CHROMA_U) + absPartIdxC, m_qtTempCbf[1], tuIterator.absPartIdxStep * sizeof(uint8_t));
+        ::memcpy(cu->getCbf(TEXT_CHROMA_V) + absPartIdxC, m_qtTempCbf[2], tuIterator.absPartIdxStep * sizeof(uint8_t));
+        ::memcpy(cu->getTransformSkip(TEXT_CHROMA_U) + absPartIdxC, m_qtTempTransformSkipFlag[1], tuIterator.absPartIdxStep * sizeof(uint8_t));
+        ::memcpy(cu->getTransformSkip(TEXT_CHROMA_V) + absPartIdxC, m_qtTempTransformSkipFlag[2], tuIterator.absPartIdxStep * sizeof(uint8_t));
+        cu->setChromIntraDirSubParts(bestMode, absPartIdxC, depth + initTrDepth);
+        cu->m_totalDistortion += bestDist;
+    }
+    while (tuIterator.isNextSection());
+
+    // restore context models
+    if (initTrDepth != 0)
+    {   // set Cbf for all blocks
+        uint32_t combCbfU = 0;
+        uint32_t combCbfV = 0;
+        uint32_t partIdx  = 0;
+        for (uint32_t p = 0; p < 4; p++, partIdx += tuIterator.absPartIdxStep)
+        {
+            combCbfU |= cu->getCbf(partIdx, TEXT_CHROMA_U, 1);
+            combCbfV |= cu->getCbf(partIdx, TEXT_CHROMA_V, 1);
+        }
+
+        for (uint32_t offs = 0; offs < 4 * tuIterator.absPartIdxStep; offs++)
+        {
+            cu->getCbf(TEXT_CHROMA_U)[offs] |= combCbfU;
+            cu->getCbf(TEXT_CHROMA_V)[offs] |= combCbfV;
+        }
+    }
+
+    m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+}
+
+/* estimation of best merge coding */
+uint32_t Search::mergeEstimation(TComDataCU* cu, int puIdx, MergeData& m)
+{
+    X265_CHECK(cu->getPartitionSize(0) != SIZE_2Nx2N, "merge tested on non-2Nx2N partition\n");
+
+    cu->getInterMergeCandidates(m.absPartIdx, puIdx, m.mvFieldNeighbours, m.interDirNeighbours, m.maxNumMergeCand);
+
+    /* convert bidir merge candidates into unidir
+     * TODO: why did the HM do this?, why use MV pairs below? */
+    if (cu->isBipredRestriction())
+    {
+        for (uint32_t mergeCand = 0; mergeCand < m.maxNumMergeCand; ++mergeCand)
+        {
+            if (m.interDirNeighbours[mergeCand] == 3)
+            {
+                m.interDirNeighbours[mergeCand] = 1;
+                m.mvFieldNeighbours[mergeCand][1].refIdx = NOT_VALID;
+            }
+        }
+    }
+
+    uint32_t outCost = MAX_UINT;
+    for (uint32_t mergeCand = 0; mergeCand < m.maxNumMergeCand; ++mergeCand)
+    {
+        /* Prevent TMVP candidates from using unavailable reference pixels */
+        if (m_bFrameParallel &&
+            (m.mvFieldNeighbours[mergeCand][0].mv.y >= (m_param->searchRange + 1) * 4 ||
+             m.mvFieldNeighbours[mergeCand][1].mv.y >= (m_param->searchRange + 1) * 4))
+            continue;
+
+        cu->getCUMvField(REF_PIC_LIST_0)->m_mv[m.absPartIdx] = m.mvFieldNeighbours[mergeCand][0].mv;
+        cu->getCUMvField(REF_PIC_LIST_0)->m_refIdx[m.absPartIdx] = (char)m.mvFieldNeighbours[mergeCand][0].refIdx;
+        cu->getCUMvField(REF_PIC_LIST_1)->m_mv[m.absPartIdx] = m.mvFieldNeighbours[mergeCand][1].mv;
+        cu->getCUMvField(REF_PIC_LIST_1)->m_refIdx[m.absPartIdx] = (char)m.mvFieldNeighbours[mergeCand][1].refIdx;
+
+        prepMotionCompensation(cu, puIdx);
+        motionCompensation(&m_predTempYuv, true, false);
+        uint32_t costCand = m_me.bufSATD(m_predTempYuv.getLumaAddr(m.absPartIdx), m_predTempYuv.getStride());
+        uint32_t bitsCand = getTUBits(mergeCand, m.maxNumMergeCand);
+        costCand = costCand + m_rdCost.getCost(bitsCand);
+        if (costCand < outCost)
+        {
+            outCost = costCand;
+            m.bits = bitsCand;
+            m.index = mergeCand;
+        }
+    }
+
+    m.mvField[0] = m.mvFieldNeighbours[m.index][0];
+    m.mvField[1] = m.mvFieldNeighbours[m.index][1];
+    m.interDir = m.interDirNeighbours[m.index];
+
+    return outCost;
+}
+
+/* search of the best candidate for inter prediction
+ * returns true if predYuv was filled with a motion compensated prediction */
+bool Search::predInterSearch(TComDataCU* cu, TComYuv* predYuv, bool bMergeOnly, bool bChroma)
+{
+    MV amvpCand[2][MAX_NUM_REF][AMVP_NUM_CANDS];
+    MV mvc[(MD_ABOVE_LEFT + 1) * 2 + 1];
+
+    Slice *slice        = cu->m_slice;
+    TComPicYuv *fenc    = slice->m_pic->getPicYuvOrg();
+    PartSize partSize   = cu->getPartitionSize(0);
+    int      numPart    = cu->getNumPartInter();
+    int      numPredDir = slice->isInterP() ? 1 : 2;
+    uint32_t lastMode = 0;
+    int      totalmebits = 0;
+    TComYuv   m_predYuv[2];
+
+    const int* numRefIdx = slice->m_numRefIdx;
+
+    MergeData merge;
+
+    memset(&merge, 0, sizeof(merge));
+
+    m_predYuv[0].create(MAX_CU_SIZE, MAX_CU_SIZE, m_param->internalCsp);
+    m_predYuv[1].create(MAX_CU_SIZE, MAX_CU_SIZE, m_param->internalCsp);
+
+    for (int partIdx = 0; partIdx < numPart; partIdx++)
+    {
+        uint32_t partAddr;
+        int      roiWidth, roiHeight;
+        cu->getPartIndexAndSize(partIdx, partAddr, roiWidth, roiHeight);
+
+        pixel* pu = fenc->getLumaAddr(cu->getAddr(), cu->getZorderIdxInCU() + partAddr);
+        m_me.setSourcePU(pu - fenc->getLumaAddr(), roiWidth, roiHeight);
+
+        uint32_t mrgCost = MAX_UINT;
+
+        /* find best cost merge candidate */
+        if (cu->getPartitionSize(partAddr) != SIZE_2Nx2N)
+        {
+            merge.absPartIdx = partAddr;
+            merge.width = roiWidth;
+            merge.height = roiHeight;
+            mrgCost = mergeEstimation(cu, partIdx, merge);
+
+            if (bMergeOnly && cu->getLog2CUSize(0) > 3)
+            {
+                if (mrgCost == MAX_UINT)
+                {
+                    /* No valid merge modes were found, there is no possible way to
+                     * perform a valid motion compensation prediction, so early-exit */
+                    return false;
+                }
+                // set merge result
+                cu->setMergeFlag(partAddr, true);
+                cu->setMergeIndex(partAddr, merge.index);
+                cu->setInterDirSubParts(merge.interDir, partAddr, partIdx, cu->getDepth(partAddr));
+                cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(merge.mvField[0], partSize, partAddr, 0, partIdx);
+                cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(merge.mvField[1], partSize, partAddr, 0, partIdx);
+                totalmebits += merge.bits;
+
+                prepMotionCompensation(cu, partIdx);
+                motionCompensation(predYuv, true, bChroma);
+                continue;
+            }
+        }
+
+        MotionData list[2];
+        MotionData bidir[2];
+        uint32_t listSelBits[3]; // cost in bits of selecting a particular ref list
+        uint32_t bidirCost = MAX_UINT;
+        int bidirBits = 0;
+
+        list[0].cost = MAX_UINT;
+        list[1].cost = MAX_UINT;
+
+        getBlkBits(partSize, slice->isInterP(), partIdx, lastMode, listSelBits);
+
+        // Uni-directional prediction
+        for (int l = 0; l < numPredDir; l++)
+        {
+            for (int ref = 0; ref < numRefIdx[l]; ref++)
+            {
+                uint32_t bits = listSelBits[l] + MVP_IDX_BITS;
+                bits += getTUBits(ref, numRefIdx[l]);
+
+                int numMvc = cu->fillMvpCand(partIdx, partAddr, l, ref, amvpCand[l][ref], mvc);
+
+                // Pick the best possible MVP from AMVP candidates based on least residual
+                uint32_t bestCost = MAX_INT;
+                int mvpIdx = 0;
+                int merange = m_param->searchRange;
+
+                for (int i = 0; i < AMVP_NUM_CANDS; i++)
+                {
+                    MV mvCand = amvpCand[l][ref][i];
+
+                    // NOTE: skip mvCand if Y is > merange and -FN>1
+                    if (m_bFrameParallel && (mvCand.y >= (merange + 1) * 4))
+                        continue;
+
+                    cu->clipMv(mvCand);
+
+                    prepMotionCompensation(cu, partIdx);
+                    predInterLumaBlk(slice->m_refPicList[l][ref]->getPicYuvRec(), &m_predTempYuv, &mvCand);
+                    uint32_t cost = m_me.bufSAD(m_predTempYuv.getLumaAddr(partAddr), m_predTempYuv.getStride());
+                    cost = (uint32_t)m_rdCost.calcRdSADCost(cost, MVP_IDX_BITS);
+
+                    if (bestCost > cost)
+                    {
+                        bestCost = cost;
+                        mvpIdx  = i;
+                    }
+                }
+
+                MV mvmin, mvmax, outmv, mvp = amvpCand[l][ref][mvpIdx];
+
+                setSearchRange(cu, mvp, merange, mvmin, mvmax);
+                int satdCost = m_me.motionEstimate(&slice->m_mref[l][ref], mvmin, mvmax, mvp, numMvc, mvc, merange, outmv);
+
+                /* Get total cost of partition, but only include MV bit cost once */
+                bits += m_me.bitcost(outmv);
+                uint32_t cost = (satdCost - m_me.mvcost(outmv)) + m_rdCost.getCost(bits);
+
+                /* Refine MVP selection, updates: mvp, mvpIdx, bits, cost */
+                checkBestMVP(amvpCand[l][ref], outmv, mvp, mvpIdx, bits, cost);
+
+                if (cost < list[l].cost)
+                {
+                    list[l].mv = outmv;
+                    list[l].mvp = mvp;
+                    list[l].mvpIdx = mvpIdx;
+                    list[l].ref = ref;
+                    list[l].cost = cost;
+                    list[l].bits = bits;
+                }
+            }
+        }
+
+        // Bi-directional prediction
+        if (slice->isInterB() && !cu->isBipredRestriction() && list[0].cost != MAX_UINT && list[1].cost != MAX_UINT)
+        {
+            ALIGN_VAR_32(pixel, avg[MAX_CU_SIZE * MAX_CU_SIZE]);
+
+            bidir[0] = list[0];
+            bidir[1] = list[1];
+
+            // Generate reference subpels
+            TComPicYuv *refPic0 = slice->m_refPicList[0][list[0].ref]->getPicYuvRec();
+            TComPicYuv *refPic1 = slice->m_refPicList[1][list[1].ref]->getPicYuvRec();
+            
+            prepMotionCompensation(cu, partIdx);
+            predInterLumaBlk(refPic0, &m_predYuv[0], &list[0].mv);
+            predInterLumaBlk(refPic1, &m_predYuv[1], &list[1].mv);
+
+            pixel *pred0 = m_predYuv[0].getLumaAddr(partAddr);
+            pixel *pred1 = m_predYuv[1].getLumaAddr(partAddr);
+
+            int partEnum = partitionFromSizes(roiWidth, roiHeight);
+            primitives.pixelavg_pp[partEnum](avg, roiWidth, pred0, m_predYuv[0].getStride(), pred1, m_predYuv[1].getStride(), 32);
+            int satdCost = m_me.bufSATD(avg, roiWidth);
+
+            bidirBits = list[0].bits + list[1].bits + listSelBits[2] - (listSelBits[0] + listSelBits[1]);
+            bidirCost = satdCost + m_rdCost.getCost(bidirBits);
+
+            MV mvzero(0, 0);
+            bool bTryZero = list[0].mv.notZero() || list[1].mv.notZero();
+            if (bTryZero)
+            {
+                /* Do not try zero MV if unidir motion predictors are beyond
+                 * valid search area */
+                MV mvmin, mvmax;
+                int merange = X265_MAX(m_param->sourceWidth, m_param->sourceHeight);
+                setSearchRange(cu, mvzero, merange, mvmin, mvmax);
+                mvmax.y += 2; // there is some pad for subpel refine
+                mvmin <<= 2;
+                mvmax <<= 2;
+
+                bTryZero &= list[0].mvp.checkRange(mvmin, mvmax);
+                bTryZero &= list[1].mvp.checkRange(mvmin, mvmax);
+            }
+            if (bTryZero)
+            {
+                // coincident blocks of the two reference pictures
+                pixel *ref0 = slice->m_mref[0][list[0].ref].fpelPlane + (pu - fenc->getLumaAddr());
+                pixel *ref1 = slice->m_mref[1][list[1].ref].fpelPlane + (pu - fenc->getLumaAddr());
+                intptr_t refStride = slice->m_mref[0][0].lumaStride;
+
+                primitives.pixelavg_pp[partEnum](avg, roiWidth, ref0, refStride, ref1, refStride, 32);
+                satdCost = m_me.bufSATD(avg, roiWidth);
+
+                MV mvp0 = list[0].mvp;
+                int mvpIdx0 = list[0].mvpIdx;
+                uint32_t bits0 = list[0].bits - m_me.bitcost(list[0].mv, mvp0) + m_me.bitcost(mvzero, mvp0);
+
+                MV mvp1 = list[1].mvp;
+                int mvpIdx1 = list[1].mvpIdx;
+                uint32_t bits1 = list[1].bits - m_me.bitcost(list[1].mv, mvp1) + m_me.bitcost(mvzero, mvp1);
+
+                uint32_t cost = satdCost + m_rdCost.getCost(bits0) + m_rdCost.getCost(bits1);
+
+                /* refine MVP selection for zero mv, updates: mvp, mvpidx, bits, cost */
+                checkBestMVP(amvpCand[0][list[0].ref], mvzero, mvp0, mvpIdx0, bits0, cost);
+                checkBestMVP(amvpCand[1][list[1].ref], mvzero, mvp1, mvpIdx1, bits1, cost);
+
+                if (cost < bidirCost)
+                {
+                    bidir[0].mv = mvzero;
+                    bidir[1].mv = mvzero;
+                    bidir[0].mvp = mvp0;
+                    bidir[1].mvp = mvp1;
+                    bidir[0].mvpIdx = mvpIdx0;
+                    bidir[1].mvpIdx = mvpIdx1;
+                    bidirCost = cost;
+                    bidirBits = bits0 + bits1 + listSelBits[2] - (listSelBits[0] + listSelBits[1]);
+                }
+            }
+        }
+
+        /* select best option and store into CU */
+        cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(TComMvField(), partSize, partAddr, 0, partIdx);
+        cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(TComMvField(), partSize, partAddr, 0, partIdx);
+
+        if (mrgCost < bidirCost && mrgCost < list[0].cost && mrgCost < list[1].cost)
+        {
+            cu->setMergeFlag(partAddr, true);
+            cu->setMergeIndex(partAddr, merge.index);
+            cu->setInterDirSubParts(merge.interDir, partAddr, partIdx, cu->getDepth(partAddr));
+            cu->getCUMvField(REF_PIC_LIST_0)->setAllMvField(merge.mvField[0], partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_1)->setAllMvField(merge.mvField[1], partSize, partAddr, 0, partIdx);
+
+            totalmebits += merge.bits;
+        }
+        else if (bidirCost < list[0].cost && bidirCost < list[1].cost)
+        {
+            lastMode = 2;
+
+            cu->setMergeFlag(partAddr, false);
+            cu->setInterDirSubParts(3, partAddr, partIdx, cu->getDepth(0));
+            cu->getCUMvField(REF_PIC_LIST_0)->setAllMv(bidir[0].mv, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_0)->setAllRefIdx(list[0].ref, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_0)->setMvd(partAddr, bidir[0].mv - bidir[0].mvp);
+            cu->setMVPIdx(REF_PIC_LIST_0, partAddr, bidir[0].mvpIdx);
+
+            cu->getCUMvField(REF_PIC_LIST_1)->setAllMv(bidir[1].mv, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_1)->setAllRefIdx(list[1].ref, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_1)->setMvd(partAddr, bidir[1].mv - bidir[1].mvp);
+            cu->setMVPIdx(REF_PIC_LIST_1, partAddr, bidir[1].mvpIdx);
+
+            totalmebits += bidirBits;
+        }
+        else if (list[0].cost <= list[1].cost)
+        {
+            lastMode = 0;
+
+            cu->setMergeFlag(partAddr, false);
+            cu->setInterDirSubParts(1, partAddr, partIdx, cu->getDepth(0));
+            cu->getCUMvField(REF_PIC_LIST_0)->setAllMv(list[0].mv, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_0)->setAllRefIdx(list[0].ref, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_0)->setMvd(partAddr, list[0].mv - list[0].mvp);
+            cu->setMVPIdx(REF_PIC_LIST_0, partAddr, list[0].mvpIdx);
+
+            totalmebits += list[0].bits;
+        }
+        else
+        {
+            lastMode = 1;
+
+            cu->setMergeFlag(partAddr, false);
+            cu->setInterDirSubParts(2, partAddr, partIdx, cu->getDepth(0));
+            cu->getCUMvField(REF_PIC_LIST_1)->setAllMv(list[1].mv, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_1)->setAllRefIdx(list[1].ref, partSize, partAddr, 0, partIdx);
+            cu->getCUMvField(REF_PIC_LIST_1)->setMvd(partAddr, list[1].mv - list[1].mvp);
+            cu->setMVPIdx(REF_PIC_LIST_1, partAddr, list[1].mvpIdx);
+
+            totalmebits += list[1].bits;
+        }
+        prepMotionCompensation(cu, partIdx);
+        motionCompensation(predYuv, true, bChroma);
+    }
+
+    m_predYuv[0].destroy();
+    m_predYuv[1].destroy();
+
+    x265_emms();
+    cu->m_totalBits = totalmebits;
+    return true;
+}
+
+void Search::getBlkBits(PartSize cuMode, bool bPSlice, int partIdx, uint32_t lastMode, uint32_t blockBit[3])
+{
+    if (cuMode == SIZE_2Nx2N)
+    {
+        blockBit[0] = (!bPSlice) ? 3 : 1;
+        blockBit[1] = 3;
+        blockBit[2] = 5;
+    }
+    else if ((cuMode == SIZE_2NxN || cuMode == SIZE_2NxnU) || cuMode == SIZE_2NxnD)
+    {
+        static const uint32_t listBits[2][3][3] =
+        {
+            { { 0, 0, 3 }, { 0, 0, 0 }, { 0, 0, 0 } },
+            { { 5, 7, 7 }, { 7, 5, 7 }, { 9 - 3, 9 - 3, 9 - 3 } }
+        };
+        if (bPSlice)
+        {
+            blockBit[0] = 3;
+            blockBit[1] = 0;
+            blockBit[2] = 0;
+        }
+        else
+            ::memcpy(blockBit, listBits[partIdx][lastMode], 3 * sizeof(uint32_t));
+    }
+    else if ((cuMode == SIZE_Nx2N || cuMode == SIZE_nLx2N) || cuMode == SIZE_nRx2N)
+    {
+        static const uint32_t listBits[2][3][3] =
+        {
+            { { 0, 2, 3 }, { 0, 0, 0 }, { 0, 0, 0 } },
+            { { 5, 7, 7 }, { 7 - 2, 7 - 2, 9 - 2 }, { 9 - 3, 9 - 3, 9 - 3 } }
+        };
+        if (bPSlice)
+        {
+            blockBit[0] = 3;
+            blockBit[1] = 0;
+            blockBit[2] = 0;
+        }
+        else
+            ::memcpy(blockBit, listBits[partIdx][lastMode], 3 * sizeof(uint32_t));
+    }
+    else if (cuMode == SIZE_NxN)
+    {
+        blockBit[0] = (!bPSlice) ? 3 : 1;
+        blockBit[1] = 3;
+        blockBit[2] = 5;
+    }
+    else
+    {
+        X265_CHECK(0, "getBlkBits: unknown cuMode\n");
+    }
+}
+
+/* Check if using an alternative MVP would result in a smaller MVD + signal bits */
+void Search::checkBestMVP(MV* amvpCand, MV mv, MV& mvPred, int& outMvpIdx, uint32_t& outBits, uint32_t& outCost)
+{
+    X265_CHECK(amvpCand[outMvpIdx] == mvPred, "checkBestMVP: unexpected mvPred\n");
+
+    int mvpIdx = !outMvpIdx;
+    MV mvp = amvpCand[mvpIdx];
+    int diffBits = m_me.bitcost(mv, mvp) - m_me.bitcost(mv, mvPred);
+    if (diffBits < 0)
+    {
+        outMvpIdx = mvpIdx;
+        mvPred = mvp;
+        uint32_t origOutBits = outBits;
+        outBits = origOutBits + diffBits;
+        outCost = (outCost - m_rdCost.getCost(origOutBits)) + m_rdCost.getCost(outBits);
+    }
+}
+
+void Search::setSearchRange(TComDataCU* cu, MV mvp, int merange, MV& mvmin, MV& mvmax)
+{
+    cu->clipMv(mvp);
+
+    MV dist((int16_t)merange << 2, (int16_t)merange << 2);
+    mvmin = mvp - dist;
+    mvmax = mvp + dist;
+
+    cu->clipMv(mvmin);
+    cu->clipMv(mvmax);
+
+    /* Clip search range to signaled maximum MV length.
+     * We do not support this VUI field being changed from the default */
+    const int maxMvLen = (1 << 15) - 1;
+    mvmin.x = X265_MAX(mvmin.x, -maxMvLen);
+    mvmin.y = X265_MAX(mvmin.y, -maxMvLen);
+    mvmax.x = X265_MIN(mvmax.x, maxMvLen);
+    mvmax.y = X265_MIN(mvmax.y, maxMvLen);
+
+    mvmin >>= 2;
+    mvmax >>= 2;
+
+    /* conditional clipping for frame parallelism */
+    mvmin.y = X265_MIN(mvmin.y, (int16_t)m_refLagPixels);
+    mvmax.y = X265_MIN(mvmax.y, (int16_t)m_refLagPixels);
+}
+
+void Search::encodeResAndCalcRdSkipCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, TComYuv* outReconYuv)
+{
+    X265_CHECK(!cu->isIntra(0), "intra CU not expected\n");
+
+    uint32_t cuSize = 1 << cu->getLog2CUSize(0);
+    uint32_t depth  = cu->getDepth(0);
+
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    // No residual coding : SKIP mode
+
+    cu->setSkipFlagSubParts(true, 0, depth);
+    cu->setTrIdxSubParts(0, 0, depth);
+    cu->clearCbf(0, depth);
+
+    outReconYuv->copyFromYuv(predYuv);
+    // Luma
+    int part = partitionFromLog2Size(cu->getLog2CUSize(0));
+    cu->m_totalDistortion = primitives.sse_pp[part](fencYuv->getLumaAddr(), fencYuv->getStride(), outReconYuv->getLumaAddr(), outReconYuv->getStride());
+    // Chroma
+    part = partitionFromSizes(cuSize >> hChromaShift, cuSize >> vChromaShift);
+    cu->m_totalDistortion += m_rdCost.scaleChromaDistCb(primitives.sse_pp[part](fencYuv->getCbAddr(), fencYuv->getCStride(), outReconYuv->getCbAddr(), outReconYuv->getCStride()));
+    cu->m_totalDistortion += m_rdCost.scaleChromaDistCr(primitives.sse_pp[part](fencYuv->getCrAddr(), fencYuv->getCStride(), outReconYuv->getCrAddr(), outReconYuv->getCStride()));
+
+    m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+    m_entropyCoder->resetBits();
+    if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+        m_entropyCoder->codeCUTransquantBypassFlag(cu->getCUTransquantBypass(0));
+    m_entropyCoder->codeSkipFlag(cu, 0);
+    m_entropyCoder->codeMergeIndex(cu, 0);
+
+    cu->m_mvBits = m_entropyCoder->getNumberOfWrittenBits();
+    cu->m_coeffBits = 0;
+    cu->m_totalBits = m_entropyCoder->getNumberOfWrittenBits();
+
+    if (m_rdCost.m_psyRd)
+    {
+        int size = cu->getLog2CUSize(0) - 2;
+        cu->m_psyEnergy = m_rdCost.psyCost(size, fencYuv->getLumaAddr(), fencYuv->getStride(),
+                                           outReconYuv->getLumaAddr(), outReconYuv->getStride());
+        cu->m_totalPsyCost = m_rdCost.calcPsyRdCost(cu->m_totalDistortion, cu->m_totalBits, cu->m_psyEnergy);
+    }
+    else
+        cu->m_totalRDCost = m_rdCost.calcRdCost(cu->m_totalDistortion, cu->m_totalBits);
+
+    m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+}
+
+/** encode residual and calculate rate-distortion for a CU block */
+void Search::encodeResAndCalcRdInterCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* outResiYuv,
+                                       ShortYuv* outBestResiYuv, TComYuv* outReconYuv)
+{
+    X265_CHECK(!cu->isIntra(0), "intra CU not expected\n");
+
+    uint32_t bestBits = 0, bestCoeffBits = 0;
+
+    uint32_t log2CUSize = cu->getLog2CUSize(0);
+    uint32_t cuSize = 1 << log2CUSize;
+    uint32_t depth  = cu->getDepth(0);
+
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    m_quant.setQPforQuant(cu);
+
+    outResiYuv->subtract(fencYuv, predYuv, log2CUSize);
+
+    bool bIsTQBypassEnable = cu->m_slice->m_pps->bTransquantBypassEnabled;
+    uint32_t numModes = 1;
+
+    if (bIsTQBypassEnable && !m_param->bLossless)
+    {
+        /* When cu-lossless mode is enabled, and lossless mode is not,
+         * 2 modes need to be checked, normal and lossless mode*/
+        numModes = 2;
+    }
+
+    uint32_t tuDepthRange[2];
+    cu->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+
+    uint64_t bestCost = MAX_INT64;
+    uint32_t bestMode = 0;
+
+    for (uint32_t modeId = 0; modeId < numModes; modeId++)
+    {
+        bool bIsLosslessMode = bIsTQBypassEnable && !modeId;
+
+        cu->setCUTransquantBypassSubParts(bIsLosslessMode, 0, depth);
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+        uint64_t cost = 0;
+        uint32_t zeroDistortion = 0;
+        uint32_t bits = 0;
+        uint32_t distortion = xEstimateResidualQT(cu, 0, fencYuv, predYuv, outResiYuv, depth, cost, bits, &zeroDistortion, tuDepthRange);
+
+        m_entropyCoder->resetBits();
+        m_entropyCoder->codeQtRootCbfZero();
+        uint32_t zeroResiBits = m_entropyCoder->getNumberOfWrittenBits();
+        uint64_t zeroCost = 0;
+        uint32_t zeroPsyEnergyY = 0;
+        if (m_rdCost.m_psyRd)
+        {
+            // need to check whether zero distortion is similar to psyenergy of fenc
+            int size = log2CUSize - 2;
+            zeroPsyEnergyY = m_rdCost.psyCost(size, fencYuv->getLumaAddr(), fencYuv->getStride(), (pixel*)zeroPel, 0);
+            zeroCost = m_rdCost.calcPsyRdCost(zeroDistortion, zeroResiBits, zeroPsyEnergyY);
+        }
+        else
+            zeroCost = m_rdCost.calcRdCost(zeroDistortion, zeroResiBits);
+
+        if (bIsLosslessMode)
+            zeroCost = cost + 1;
+
+        if (zeroCost < cost)
+        {
+            distortion = zeroDistortion;
+            cu->m_psyEnergy = zeroPsyEnergyY;
+
+            const uint32_t qpartnum = NUM_CU_PARTITIONS >> (depth << 1);
+            ::memset(cu->getTransformIdx(), 0, qpartnum * sizeof(uint8_t));
+            ::memset(cu->getCbf(TEXT_LUMA), 0, qpartnum * sizeof(uint8_t));
+            ::memset(cu->getCbf(TEXT_CHROMA_U), 0, qpartnum * sizeof(uint8_t));
+            ::memset(cu->getCbf(TEXT_CHROMA_V), 0, qpartnum * sizeof(uint8_t));
+#if CHECKED_BUILD || _DEBUG
+            ::memset(cu->getCoeffY(), 0, cuSize * cuSize * sizeof(coeff_t));
+            ::memset(cu->getCoeffCb(), 0, cuSize * cuSize * sizeof(coeff_t) >> (hChromaShift + vChromaShift));
+            ::memset(cu->getCoeffCr(), 0, cuSize * cuSize * sizeof(coeff_t) >> (hChromaShift + vChromaShift));
+#endif
+            cu->setTransformSkipSubParts(0, 0, 0, 0, depth);
+        }
+        else
+            xSetResidualQTData(cu, 0, NULL, depth, false);
+
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+        bits = getInterSymbolBits(cu, tuDepthRange);
+
+        if (m_rdCost.m_psyRd)
+            cost = m_rdCost.calcPsyRdCost(distortion, bits, cu->m_psyEnergy);
+        else
+            cost = m_rdCost.calcRdCost(distortion, bits);
+
+        if (cost < bestCost)
+        {
+            if (cu->getQtRootCbf(0))
+                xSetResidualQTData(cu, 0, outBestResiYuv, depth, true);
+
+            bestMode = modeId; // 0 for lossless
+            bestBits = bits;
+            bestCost = cost;
+            bestCoeffBits = cu->m_coeffBits;
+            m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+        }
+    }
+
+    X265_CHECK(bestCost != MAX_INT64, "no best cost\n");
+
+    if (bIsTQBypassEnable && !bestMode)
+    {
+        cu->setCUTransquantBypassSubParts(true, 0, depth);
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+        uint64_t cost = 0;
+        uint32_t bits = 0;
+        xEstimateResidualQT(cu, 0, fencYuv, predYuv, outResiYuv, depth, cost, bits, NULL, tuDepthRange);
+        xSetResidualQTData(cu, 0, NULL, depth, false);
+        m_entropyCoder->store(m_rdEntropyCoders[depth][CI_TEMP_BEST]);
+    }
+
+    if (cu->getQtRootCbf(0))
+        outReconYuv->addClip(predYuv, outBestResiYuv, log2CUSize);
+    else
+        outReconYuv->copyFromYuv(predYuv);
+
+    // update with clipped distortion and cost (qp estimation loop uses unclipped values)
+    int part = partitionFromLog2Size(log2CUSize);
+    uint32_t bestDist = primitives.sse_pp[part](fencYuv->getLumaAddr(), fencYuv->getStride(), outReconYuv->getLumaAddr(), outReconYuv->getStride());
+    part = partitionFromSizes(cuSize >> hChromaShift, cuSize >> vChromaShift);
+    bestDist += m_rdCost.scaleChromaDistCb(primitives.sse_pp[part](fencYuv->getCbAddr(), fencYuv->getCStride(), outReconYuv->getCbAddr(), outReconYuv->getCStride()));
+    bestDist += m_rdCost.scaleChromaDistCr(primitives.sse_pp[part](fencYuv->getCrAddr(), fencYuv->getCStride(), outReconYuv->getCrAddr(), outReconYuv->getCStride()));
+    if (m_rdCost.m_psyRd)
+    {
+        int size = log2CUSize - 2;
+        cu->m_psyEnergy = m_rdCost.psyCost(size, fencYuv->getLumaAddr(), fencYuv->getStride(),
+                                           outReconYuv->getLumaAddr(), outReconYuv->getStride());
+        cu->m_totalPsyCost = m_rdCost.calcPsyRdCost(bestDist, bestBits, cu->m_psyEnergy);
+    }
+    else
+        cu->m_totalRDCost = m_rdCost.calcRdCost(bestDist, bestBits);
+
+    cu->m_totalBits       = bestBits;
+    cu->m_totalDistortion = bestDist;
+    cu->m_coeffBits = bestCoeffBits;
+    cu->m_mvBits = bestBits - bestCoeffBits;
+
+    if (cu->isSkipped(0))
+        cu->clearCbf(0, depth);
+}
+
+void Search::generateCoeffRecon(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv)
+{
+    m_quant.setQPforQuant(cu);
+
+    uint32_t tuDepthRange[2];
+    cu->getQuadtreeTULog2MinSizeInCU(tuDepthRange, 0);
+
+    if (cu->getPredictionMode(0) == MODE_INTER)
+    {
+        residualTransformQuantInter(cu, 0, fencYuv, resiYuv, cu->getDepth(0), tuDepthRange);
+        if (cu->getQtRootCbf(0))
+            reconYuv->addClip(predYuv, resiYuv, cu->getLog2CUSize(0));
+        else
+        {
+            reconYuv->copyFromYuv(predYuv);
+            if (cu->getMergeFlag(0) && cu->getPartitionSize(0) == SIZE_2Nx2N)
+                cu->setSkipFlagSubParts(true, 0, cu->getDepth(0));
+        }
+    }
+    else if (cu->getPredictionMode(0) == MODE_INTRA)
+    {
+        uint32_t initTrDepth = cu->getPartitionSize(0) == SIZE_2Nx2N ? 0 : 1;
+        residualTransformQuantIntra(cu, initTrDepth, 0, fencYuv, predYuv, resiYuv, reconYuv, tuDepthRange);
+        getBestIntraModeChroma(cu, fencYuv, predYuv);
+        residualQTIntraChroma(cu, 0, 0, fencYuv, predYuv, resiYuv, reconYuv);
+    }
+}
+
+void Search::residualTransformQuantInter(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, ShortYuv* resiYuv,
+                                         const uint32_t depth, uint32_t depthRange[2])
+{
+    X265_CHECK(cu->getDepth(0) == cu->getDepth(absPartIdx), "invalid depth\n");
+    const uint32_t trMode = depth - cu->getDepth(0);
+    const uint32_t log2TrSize = g_maxLog2CUSize - depth;
+    const uint32_t setCbf     = 1 << trMode;
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    bool bSplitFlag = ((cu->m_slice->m_sps->quadtreeTUMaxDepthInter == 1) && cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N));
+    bool bCheckFull;
+    if (bSplitFlag && depth == cu->getDepth(absPartIdx) && log2TrSize > depthRange[0])
+        bCheckFull = false;
+    else
+        bCheckFull = log2TrSize <= depthRange[1];
+    const bool bCheckSplit = log2TrSize > depthRange[0];
+    X265_CHECK(bCheckFull || bCheckSplit, "check-full or check-split must be set\n");
+
+    // code full block
+    if (bCheckFull)
+    {
+        uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+        bool bCodeChroma = true;
+        uint32_t trModeC = trMode;
+        if ((log2TrSize == 2) && !(m_csp == X265_CSP_I444))
+        {
+            log2TrSizeC++;
+            trModeC--;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+            bCodeChroma = ((absPartIdx & (qpdiv - 1)) == 0);
+        }
+
+        const bool splitIntoSubTUs = (m_csp == X265_CSP_I422);
+        uint32_t absPartIdxStep = NUM_CU_PARTITIONS >> ((cu->getDepth(0) +  trModeC) << 1);
+
+        uint32_t coeffOffsetY = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        uint32_t coeffOffsetC = coeffOffsetY >> (hChromaShift + vChromaShift);
+        coeff_t *coeffCurY = cu->getCoeffY()  + coeffOffsetY;
+        coeff_t *coeffCurU = cu->getCoeffCb() + coeffOffsetC;
+        coeff_t *coeffCurV = cu->getCoeffCr() + coeffOffsetC;
+
+        uint32_t sizeIdx  = log2TrSize  - 2;
+        uint32_t sizeIdxC = log2TrSizeC - 2;
+        cu->setTrIdxSubParts(depth - cu->getDepth(0), absPartIdx, depth);
+
+        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
+
+        int16_t *curResiY = resiYuv->getLumaAddr(absPartIdx);
+        const uint32_t strideResiY = resiYuv->m_width;
+        const uint32_t strideResiC = resiYuv->m_cwidth;
+
+        uint32_t numSigY = m_quant.transformNxN(cu, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(), curResiY, strideResiY, coeffCurY,
+                                                log2TrSize, TEXT_LUMA, absPartIdx, false);
+
+        cu->setCbfSubParts(numSigY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+
+        if (numSigY)
+            m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), curResiY, strideResiY, coeffCurY, log2TrSize, TEXT_LUMA, false, false, numSigY);
+        else
+            primitives.blockfill_s[sizeIdx](curResiY, strideResiY, 0);
+
+        if (bCodeChroma)
+        {
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, absPartIdxStep, absPartIdx);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+                uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+
+                int16_t *curResiU = resiYuv->getCbAddr(absPartIdxC);
+                int16_t *curResiV = resiYuv->getCrAddr(absPartIdxC);
+
+                cu->setTransformSkipPartRange(0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setTransformSkipPartRange(0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                uint32_t numSigU = m_quant.transformNxN(cu, fencYuv->getCbAddr(absPartIdxC), fencYuv->getCStride(), curResiU, strideResiC, coeffCurU + subTUOffset,
+                                                        log2TrSizeC, TEXT_CHROMA_U, absPartIdxC, false);
+                uint32_t numSigV = m_quant.transformNxN(cu, fencYuv->getCrAddr(absPartIdxC), fencYuv->getCStride(), curResiV, strideResiC, coeffCurV + subTUOffset,
+                                                        log2TrSizeC, TEXT_CHROMA_V, absPartIdxC, false);
+
+                cu->setCbfPartRange(numSigU ? setCbf : 0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setCbfPartRange(numSigV ? setCbf : 0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                if (numSigU)
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), curResiU, strideResiC, coeffCurU + subTUOffset, log2TrSizeC, TEXT_CHROMA_U, false, false, numSigU);
+                else
+                    primitives.blockfill_s[sizeIdxC](curResiU, strideResiC, 0);
+
+                if (numSigV)
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), curResiV, strideResiC, coeffCurV + subTUOffset, log2TrSizeC, TEXT_CHROMA_V, false, false, numSigV);
+                else
+                    primitives.blockfill_s[sizeIdxC](curResiV, strideResiC, 0);
+            }
+            while (tuIterator.isNextSection());
+
+            if (splitIntoSubTUs)
+            {
+                offsetSubTUCBFs(cu, TEXT_CHROMA_U, trMode, absPartIdx);
+                offsetSubTUCBFs(cu, TEXT_CHROMA_V, trMode, absPartIdx);
+            }
+        }
+        return;
+    }
+
+    // code sub-blocks
+    if (bCheckSplit && !bCheckFull)
+    {
+        const uint32_t qPartNumSubdiv = NUM_CU_PARTITIONS >> ((depth + 1) << 1);
+        for (uint32_t i = 0; i < 4; ++i)
+            residualTransformQuantInter(cu, absPartIdx + i * qPartNumSubdiv, fencYuv, resiYuv, depth + 1, depthRange);
+
+        uint32_t ycbf = 0;
+        uint32_t ucbf = 0;
+        uint32_t vcbf = 0;
+        for (uint32_t i = 0; i < 4; ++i)
+        {
+            ycbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_LUMA,     trMode + 1);
+            ucbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_U, trMode + 1);
+            vcbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_V, trMode + 1);
+        }
+
+        for (uint32_t i = 0; i < 4 * qPartNumSubdiv; ++i)
+        {
+            cu->getCbf(TEXT_LUMA)[absPartIdx + i] |= ycbf << trMode;
+            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + i] |= ucbf << trMode;
+            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + i] |= vcbf << trMode;
+        }
+    }
+}
+
+uint32_t Search::xEstimateResidualQT(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
+                                     uint32_t depth, uint64_t& rdCost, uint32_t& outBits, uint32_t* outZeroDist, uint32_t depthRange[2])
+{
+    X265_CHECK(cu->getDepth(0) == cu->getDepth(absPartIdx), "depth not matching\n");
+    const uint32_t trMode = depth - cu->getDepth(0);
+    const uint32_t log2TrSize = g_maxLog2CUSize - depth;
+    const uint32_t subTUDepth = trMode + 1;
+    const uint32_t setCbf     = 1 << trMode;
+    uint32_t outDist = 0;
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    bool bSplitFlag = ((cu->m_slice->m_sps->quadtreeTUMaxDepthInter == 1) && cu->getPredictionMode(absPartIdx) == MODE_INTER && (cu->getPartitionSize(absPartIdx) != SIZE_2Nx2N));
+    bool bCheckFull;
+    if (bSplitFlag && depth == cu->getDepth(absPartIdx) && log2TrSize > depthRange[0])
+        bCheckFull = false;
+    else
+        bCheckFull = log2TrSize <= depthRange[1];
+    const bool bCheckSplit = log2TrSize > depthRange[0];
+
+    X265_CHECK(bCheckFull || bCheckSplit, "check-full or check-split must be set\n");
+
+    uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+    bool bCodeChroma = true;
+    uint32_t trModeC = trMode;
+    if ((log2TrSize == 2) && !(m_csp == X265_CSP_I444))
+    {
+        log2TrSizeC++;
+        trModeC--;
+        uint32_t qpdiv = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+        bCodeChroma = ((absPartIdx & (qpdiv - 1)) == 0);
+    }
+
+    // code full block
+    uint64_t singleCost = MAX_INT64;
+    uint32_t singleBits = 0;
+    uint32_t singleDist = 0;
+    uint32_t singlePsyEnergy = 0;
+    uint32_t singleBitsComp[MAX_NUM_COMPONENT][2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/] = { { 0, 0 }, { 0, 0 }, { 0, 0 } };
+    uint32_t singleDistComp[MAX_NUM_COMPONENT][2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/] = { { 0, 0 }, { 0, 0 }, { 0, 0 } };
+    uint32_t singlePsyEnergyComp[MAX_NUM_COMPONENT][2] = { { 0, 0 }, { 0, 0 }, { 0, 0 } };
+    uint32_t numSigY = 0;
+    uint32_t bestTransformMode[MAX_NUM_COMPONENT][2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/] = { { 0, 0 }, { 0, 0 }, { 0, 0 } };
+    uint64_t minCost[MAX_NUM_COMPONENT][2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/];
+
+    uint32_t bestCBF[MAX_NUM_COMPONENT];
+    uint32_t bestsubTUCBF[MAX_NUM_COMPONENT][2];
+    m_entropyCoder->store(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+
+    uint32_t trSize = 1 << log2TrSize;
+    const bool splitIntoSubTUs = (m_csp == X265_CSP_I422);
+    uint32_t absPartIdxStep = NUM_CU_PARTITIONS >> ((cu->getDepth(0) +  trModeC) << 1);
+
+    // code full block
+    if (bCheckFull)
+    {
+        uint32_t numSigU[2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/] = { 0, 0 };
+        uint32_t numSigV[2 /*0 = top (or whole TU for non-4:2:2) sub-TU, 1 = bottom sub-TU*/] = { 0, 0 };
+        uint32_t trSizeC = 1 << log2TrSizeC;
+        int sizeIdx  = log2TrSize - 2;
+        int sizeIdxC = log2TrSizeC - 2;
+        const uint32_t qtLayer = log2TrSize - 2;
+        uint32_t coeffOffsetY = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        uint32_t coeffOffsetC = coeffOffsetY >> (hChromaShift + vChromaShift);
+        coeff_t* coeffCurY = m_qtTempCoeff[0][qtLayer] + coeffOffsetY;
+        coeff_t* coeffCurU = m_qtTempCoeff[1][qtLayer] + coeffOffsetC;
+        coeff_t* coeffCurV = m_qtTempCoeff[2][qtLayer] + coeffOffsetC;
+
+        cu->setTrIdxSubParts(depth - cu->getDepth(0), absPartIdx, depth);
+        bool checkTransformSkip   = cu->m_slice->m_pps->bTransformSkipEnabled && !cu->getCUTransquantBypass(0);
+        bool checkTransformSkipY  = checkTransformSkip && log2TrSize  <= MAX_LOG2_TS_SIZE;
+        bool checkTransformSkipUV = checkTransformSkip && log2TrSizeC <= MAX_LOG2_TS_SIZE;
+
+        cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
+
+        if (m_bEnableRDOQ)
+            m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSize, true);
+
+        numSigY = m_quant.transformNxN(cu, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(), resiYuv->getLumaAddr(absPartIdx), resiYuv->m_width, coeffCurY,
+                                       log2TrSize, TEXT_LUMA, absPartIdx, false);
+
+        cu->setCbfSubParts(numSigY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+
+        m_entropyCoder->resetBits();
+        m_entropyCoder->codeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
+        if (numSigY)
+            m_entropyCoder->codeCoeffNxN(cu, coeffCurY, absPartIdx, log2TrSize, TEXT_LUMA);
+        singleBitsComp[TEXT_LUMA][0] = m_entropyCoder->getNumberOfWrittenBits();
+
+        uint32_t singleBitsPrev = singleBitsComp[TEXT_LUMA][0];
+
+        if (bCodeChroma)
+        {
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, absPartIdxStep, absPartIdx);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+                uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+
+                cu->setTransformSkipPartRange(0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setTransformSkipPartRange(0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                if (m_bEnableRDOQ)
+                    m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSizeC, false);
+
+                numSigU[tuIterator.section] = m_quant.transformNxN(cu, fencYuv->getCbAddr(absPartIdxC), fencYuv->getCStride(), resiYuv->getCbAddr(absPartIdxC), resiYuv->m_cwidth, coeffCurU + subTUOffset,
+                                                                   log2TrSizeC, TEXT_CHROMA_U, absPartIdxC, false);
+                numSigV[tuIterator.section] = m_quant.transformNxN(cu, fencYuv->getCrAddr(absPartIdxC), fencYuv->getCStride(), resiYuv->getCrAddr(absPartIdxC), resiYuv->m_cwidth, coeffCurV + subTUOffset,
+                                                                   log2TrSizeC, TEXT_CHROMA_V, absPartIdxC, false);
+
+                cu->setCbfPartRange(numSigU[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setCbfPartRange(numSigV[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                m_entropyCoder->codeQtCbf(cu, absPartIdxC, TEXT_CHROMA_U, trMode);
+                if (numSigU[tuIterator.section])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurU + subTUOffset, absPartIdxC, log2TrSizeC, TEXT_CHROMA_U);
+                singleBitsComp[TEXT_CHROMA_U][tuIterator.section] = m_entropyCoder->getNumberOfWrittenBits() - singleBitsPrev;
+
+                m_entropyCoder->codeQtCbf(cu, absPartIdxC, TEXT_CHROMA_V, trMode);
+                if (numSigV[tuIterator.section])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurV + subTUOffset, absPartIdxC, log2TrSizeC, TEXT_CHROMA_V);
+
+                uint32_t newBits = m_entropyCoder->getNumberOfWrittenBits();
+                singleBitsComp[TEXT_CHROMA_V][tuIterator.section] = newBits - (singleBitsPrev + singleBitsComp[TEXT_CHROMA_U][tuIterator.section]);
+
+                singleBitsPrev = newBits;
+            }
+            while (tuIterator.isNextSection());
+        }
+
+        const uint32_t numCoeffY = 1 << (log2TrSize * 2);
+        const uint32_t numCoeffC = 1 << (log2TrSizeC * 2);
+
+        for (uint32_t subTUIndex = 0; subTUIndex < 2; subTUIndex++)
+        {
+            minCost[TEXT_LUMA][subTUIndex]     = MAX_INT64;
+            minCost[TEXT_CHROMA_U][subTUIndex] = MAX_INT64;
+            minCost[TEXT_CHROMA_V][subTUIndex] = MAX_INT64;
+        }
+
+        int partSize = partitionFromLog2Size(log2TrSize);
+        X265_CHECK(log2TrSize <= 5, "log2TrSize is too large\n");
+        uint32_t distY = primitives.ssd_s[log2TrSize - 2](resiYuv->getLumaAddr(absPartIdx), resiYuv->m_width);
+        uint32_t psyEnergyY = 0;
+        if (m_rdCost.m_psyRd)
+        {
+            // need to check whether zero distortion is similar to psyenergy of fenc
+            int size = log2TrSize - 2;
+            psyEnergyY = m_rdCost.psyCost(size, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(), (pixel*)zeroPel, 0);
+        }
+        int16_t *curResiY = m_qtTempShortYuv[qtLayer].getLumaAddr(absPartIdx);
+        X265_CHECK(m_qtTempShortYuv[qtLayer].m_width == MAX_CU_SIZE, "width not full CU\n");
+        const uint32_t strideResiY = MAX_CU_SIZE;
+        const uint32_t strideResiC = m_qtTempShortYuv[qtLayer].m_cwidth;
+
+        if (outZeroDist)
+            *outZeroDist += distY;
+
+        if (numSigY)
+        {
+            m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), curResiY, strideResiY, coeffCurY, log2TrSize, TEXT_LUMA, false, false, numSigY); //this is for inter mode only
+
+            const uint32_t nonZeroDistY = primitives.sse_ss[partSize](resiYuv->getLumaAddr(absPartIdx), resiYuv->m_width, curResiY, strideResiY);
+            uint32_t nonZeroPsyEnergyY = 0;
+            if (m_rdCost.m_psyRd)
+            {
+                pixel*   pred = predYuv->getLumaAddr(absPartIdx);
+                uint32_t zorder = cu->getZorderIdxInCU() + absPartIdx;
+                pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+                uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getStride();
+                uint32_t stride = fencYuv->getStride();
+                //===== reconstruction =====
+                primitives.luma_add_ps[sizeIdx](reconIPred, reconIPredStride, pred, curResiY, stride, strideResiY);
+                int size = log2TrSize - 2;
+                nonZeroPsyEnergyY = m_rdCost.psyCost(size, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(),
+                    cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getStride());
+            }
+            if (cu->getCUTransquantBypass(0))
+            {
+                distY = nonZeroDistY;
+                psyEnergyY = nonZeroPsyEnergyY;
+            }
+            else
+            {
+                uint64_t singleCostY = 0;
+                if (m_rdCost.m_psyRd)
+                    singleCostY = m_rdCost.calcPsyRdCost(nonZeroDistY, singleBitsComp[TEXT_LUMA][0], nonZeroPsyEnergyY);
+                else
+                    singleCostY = m_rdCost.calcRdCost(nonZeroDistY, singleBitsComp[TEXT_LUMA][0]);
+                m_entropyCoder->resetBits();
+                m_entropyCoder->codeQtCbfZero(TEXT_LUMA, trMode);
+                const uint32_t nullBitsY = m_entropyCoder->getNumberOfWrittenBits();
+                uint64_t nullCostY = 0;
+                if (m_rdCost.m_psyRd)
+                    nullCostY = m_rdCost.calcPsyRdCost(distY, nullBitsY, psyEnergyY);
+                else
+                    nullCostY = m_rdCost.calcRdCost(distY, nullBitsY);
+                if (nullCostY < singleCostY)
+                {
+                    numSigY = 0;
+#if CHECKED_BUILD || _DEBUG
+                    ::memset(coeffCurY, 0, sizeof(coeff_t) * numCoeffY);
+#endif
+                    if (checkTransformSkipY)
+                        minCost[TEXT_LUMA][0] = nullCostY;
+                }
+                else
+                {
+                    distY = nonZeroDistY;
+                    psyEnergyY = nonZeroPsyEnergyY;
+                    if (checkTransformSkipY)
+                        minCost[TEXT_LUMA][0] = singleCostY;
+                }
+            }
+        }
+        else if (checkTransformSkipY)
+        {
+            m_entropyCoder->resetBits();
+            m_entropyCoder->codeQtCbfZero(TEXT_LUMA, trMode);
+            const uint32_t nullBitsY = m_entropyCoder->getNumberOfWrittenBits();
+            if (m_rdCost.m_psyRd)
+                minCost[TEXT_LUMA][0] = m_rdCost.calcPsyRdCost(distY, nullBitsY, psyEnergyY);
+            else
+                minCost[TEXT_LUMA][0] = m_rdCost.calcRdCost(distY, nullBitsY);
+        }
+
+        singleDistComp[TEXT_LUMA][0] = distY;
+        singlePsyEnergyComp[TEXT_LUMA][0] = psyEnergyY;
+        if (!numSigY)
+            primitives.blockfill_s[sizeIdx](curResiY, strideResiY, 0);
+        cu->setCbfSubParts(numSigY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+
+        uint32_t distU = 0;
+        uint32_t distV = 0;
+        uint32_t psyEnergyU = 0;
+        uint32_t psyEnergyV = 0;
+        if (bCodeChroma)
+        {
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, absPartIdxStep, absPartIdx);
+
+            int partSizeC = partitionFromLog2Size(log2TrSizeC);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+                uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+
+                int16_t *curResiU = m_qtTempShortYuv[qtLayer].getCbAddr(absPartIdxC);
+                int16_t *curResiV = m_qtTempShortYuv[qtLayer].getCrAddr(absPartIdxC);
+
+                distU = m_rdCost.scaleChromaDistCb(primitives.ssd_s[log2TrSizeC - 2](resiYuv->getCbAddr(absPartIdxC), resiYuv->m_cwidth));
+                if (outZeroDist)
+                    *outZeroDist += distU;
+
+                if (numSigU[tuIterator.section])
+                {
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), curResiU, strideResiC, coeffCurU + subTUOffset,
+                                            log2TrSizeC, TEXT_CHROMA_U, false, false, numSigU[tuIterator.section]);
+                    uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCbAddr(absPartIdxC), resiYuv->m_cwidth,
+                                                                 curResiU, strideResiC);
+                    const uint32_t nonZeroDistU = m_rdCost.scaleChromaDistCb(dist);
+                    uint32_t nonZeroPsyEnergyU = 0;
+
+                    if (m_rdCost.m_psyRd)
+                    {
+                        pixel*   pred = predYuv->getCbAddr(absPartIdxC);
+                        uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                        pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder);
+                        uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+                        uint32_t stride = fencYuv->getCStride();
+                        //===== reconstruction =====
+                        int size = log2TrSizeC - 2;
+                        primitives.luma_add_ps[size](reconIPred, reconIPredStride, pred, curResiU, stride, strideResiC);
+                        nonZeroPsyEnergyU = m_rdCost.psyCost(size, fencYuv->getCbAddr(absPartIdxC), fencYuv->getCStride(),
+                                                             cu->m_pic->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder),
+                                                             cu->m_pic->getPicYuvRec()->getCStride());
+                    }
+                    if (cu->getCUTransquantBypass(0))
+                    {
+                        distU = nonZeroDistU;
+                        psyEnergyU = nonZeroPsyEnergyU;
+                    }
+                    else
+                    {
+                        uint64_t singleCostU = 0;
+                        if (m_rdCost.m_psyRd)
+                            singleCostU = m_rdCost.calcPsyRdCost(nonZeroDistU, singleBitsComp[TEXT_CHROMA_U][tuIterator.section], nonZeroPsyEnergyU);
+                        else
+                            singleCostU = m_rdCost.calcRdCost(nonZeroDistU, singleBitsComp[TEXT_CHROMA_U][tuIterator.section]);
+                        m_entropyCoder->resetBits();
+                        m_entropyCoder->codeQtCbfZero(TEXT_CHROMA_U, trMode);
+                        const uint32_t nullBitsU = m_entropyCoder->getNumberOfWrittenBits();
+                        uint64_t nullCostU = 0;
+                        if (m_rdCost.m_psyRd)
+                            nullCostU = m_rdCost.calcPsyRdCost(distU, nullBitsU, psyEnergyU);
+                        else
+                            nullCostU = m_rdCost.calcRdCost(distU, nullBitsU);
+                        if (nullCostU < singleCostU)
+                        {
+                            numSigU[tuIterator.section] = 0;
+#if CHECKED_BUILD || _DEBUG
+                            ::memset(coeffCurU + subTUOffset, 0, sizeof(coeff_t) * numCoeffC);
+#endif
+                            if (checkTransformSkipUV)
+                                minCost[TEXT_CHROMA_U][tuIterator.section] = nullCostU;
+                        }
+                        else
+                        {
+                            distU = nonZeroDistU;
+                            psyEnergyU = nonZeroPsyEnergyU;
+                            if (checkTransformSkipUV)
+                                minCost[TEXT_CHROMA_U][tuIterator.section] = singleCostU;
+                        }
+                    }
+                }
+                else if (checkTransformSkipUV)
+                {
+                    m_entropyCoder->resetBits();
+                    m_entropyCoder->codeQtCbfZero(TEXT_CHROMA_U, trModeC);
+                    const uint32_t nullBitsU = m_entropyCoder->getNumberOfWrittenBits();
+                    if (m_rdCost.m_psyRd)
+                        minCost[TEXT_CHROMA_U][tuIterator.section] = m_rdCost.calcPsyRdCost(distU, nullBitsU, psyEnergyU);
+                    else
+                        minCost[TEXT_CHROMA_U][tuIterator.section] = m_rdCost.calcRdCost(distU, nullBitsU);
+                }
+
+                singleDistComp[TEXT_CHROMA_U][tuIterator.section] = distU;
+                singlePsyEnergyComp[TEXT_CHROMA_U][tuIterator.section] = psyEnergyU;
+
+                if (!numSigU[tuIterator.section])
+                    primitives.blockfill_s[sizeIdxC](curResiU, strideResiC, 0);
+
+                distV = m_rdCost.scaleChromaDistCr(primitives.ssd_s[log2TrSizeC - 2](resiYuv->getCrAddr(absPartIdxC), resiYuv->m_cwidth));
+                if (outZeroDist)
+                    *outZeroDist += distV;
+
+                if (numSigV[tuIterator.section])
+                {
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), curResiV, strideResiC, coeffCurV + subTUOffset,
+                                            log2TrSizeC, TEXT_CHROMA_V, false, false, numSigV[tuIterator.section]);
+                    uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCrAddr(absPartIdxC), resiYuv->m_cwidth,
+                                                                 curResiV, strideResiC);
+                    const uint32_t nonZeroDistV = m_rdCost.scaleChromaDistCr(dist);
+                    uint32_t nonZeroPsyEnergyV = 0;
+
+                    if (m_rdCost.m_psyRd)
+                    {
+                        pixel*   pred = predYuv->getCrAddr(absPartIdxC);
+                        uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                        pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder);
+                        uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+                        uint32_t stride = fencYuv->getCStride();
+                        //===== reconstruction =====
+                        int size = log2TrSizeC - 2;
+                        primitives.luma_add_ps[size](reconIPred, reconIPredStride, pred, curResiV, stride, strideResiC);
+                        nonZeroPsyEnergyV = m_rdCost.psyCost(size, fencYuv->getCrAddr(absPartIdxC), fencYuv->getCStride(),
+                                                             cu->m_pic->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder),
+                                                             cu->m_pic->getPicYuvRec()->getCStride());
+                    }
+                    if (cu->getCUTransquantBypass(0))
+                    {
+                        distV = nonZeroDistV;
+                        psyEnergyV = nonZeroPsyEnergyV;
+                    }
+                    else
+                    {
+                        uint64_t singleCostV = 0;
+                        if (m_rdCost.m_psyRd)
+                            singleCostV = m_rdCost.calcPsyRdCost(nonZeroDistV, singleBitsComp[TEXT_CHROMA_V][tuIterator.section], nonZeroPsyEnergyV);
+                        else
+                            singleCostV = m_rdCost.calcRdCost(nonZeroDistV, singleBitsComp[TEXT_CHROMA_V][tuIterator.section]);
+                        m_entropyCoder->resetBits();
+                        m_entropyCoder->codeQtCbfZero(TEXT_CHROMA_V, trMode);
+                        const uint32_t nullBitsV = m_entropyCoder->getNumberOfWrittenBits();
+                        uint64_t nullCostV = 0;
+                        if (m_rdCost.m_psyRd)
+                            nullCostV = m_rdCost.calcPsyRdCost(distV, nullBitsV, psyEnergyV);
+                        else
+                            nullCostV = m_rdCost.calcRdCost(distV, nullBitsV);
+                        if (nullCostV < singleCostV)
+                        {
+                            numSigV[tuIterator.section] = 0;
+#if CHECKED_BUILD || _DEBUG
+                            ::memset(coeffCurV + subTUOffset, 0, sizeof(coeff_t) * numCoeffC);
+#endif
+                            if (checkTransformSkipUV)
+                                minCost[TEXT_CHROMA_V][tuIterator.section] = nullCostV;
+                        }
+                        else
+                        {
+                            distV = nonZeroDistV;
+                            psyEnergyV = nonZeroPsyEnergyV;
+                            if (checkTransformSkipUV)
+                                minCost[TEXT_CHROMA_V][tuIterator.section] = singleCostV;
+                        }
+                    }
+                }
+                else if (checkTransformSkipUV)
+                {
+                    m_entropyCoder->resetBits();
+                    m_entropyCoder->codeQtCbfZero(TEXT_CHROMA_V, trModeC);
+                    const uint32_t nullBitsV = m_entropyCoder->getNumberOfWrittenBits();
+                    if (m_rdCost.m_psyRd)
+                        minCost[TEXT_CHROMA_V][tuIterator.section] = m_rdCost.calcPsyRdCost(distV, nullBitsV, psyEnergyV);
+                    else
+                        minCost[TEXT_CHROMA_V][tuIterator.section] = m_rdCost.calcRdCost(distV, nullBitsV);
+                }
+
+                singleDistComp[TEXT_CHROMA_V][tuIterator.section] = distV;
+                singlePsyEnergyComp[TEXT_CHROMA_V][tuIterator.section] = psyEnergyV;
+
+                if (!numSigV[tuIterator.section])
+                    primitives.blockfill_s[sizeIdxC](curResiV, strideResiC, 0);
+
+                cu->setCbfPartRange(numSigU[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setCbfPartRange(numSigV[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+            }
+            while (tuIterator.isNextSection());
+        }
+
+        if (checkTransformSkipY)
+        {
+            uint32_t nonZeroDistY = 0;
+            uint32_t nonZeroPsyEnergyY = 0;
+            uint64_t singleCostY = MAX_INT64;
+
+            ALIGN_VAR_32(coeff_t, tsCoeffY[MAX_TS_SIZE * MAX_TS_SIZE]);
+            ALIGN_VAR_32(int16_t, tsResiY[MAX_TS_SIZE * MAX_TS_SIZE]);
+
+            m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+
+            cu->setTransformSkipSubParts(1, TEXT_LUMA, absPartIdx, depth);
+
+            if (m_bEnableRDOQ)
+                m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSize, true);
+
+            uint32_t numSigTSkipY = m_quant.transformNxN(cu, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(), resiYuv->getLumaAddr(absPartIdx), resiYuv->m_width, tsCoeffY,
+                                                         log2TrSize, TEXT_LUMA, absPartIdx, true);
+            cu->setCbfSubParts(numSigTSkipY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+
+            if (numSigTSkipY)
+            {
+                m_entropyCoder->resetBits();
+                m_entropyCoder->codeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
+                m_entropyCoder->codeCoeffNxN(cu, tsCoeffY, absPartIdx, log2TrSize, TEXT_LUMA);
+                const uint32_t skipSingleBitsY = m_entropyCoder->getNumberOfWrittenBits();
+
+                m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdx), tsResiY, trSize, tsCoeffY, log2TrSize, TEXT_LUMA, false, true, numSigTSkipY);
+
+                nonZeroDistY = primitives.sse_ss[partSize](resiYuv->getLumaAddr(absPartIdx), resiYuv->m_width, tsResiY, trSize);
+
+                if (m_rdCost.m_psyRd)
+                {
+                    pixel*   pred = predYuv->getLumaAddr(absPartIdx);
+                    uint32_t zorder = cu->getZorderIdxInCU() + absPartIdx;
+                    pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder);
+                    uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getStride();
+                    uint32_t stride = fencYuv->getStride();
+                    //===== reconstruction =====
+                    int size = log2TrSize - 2;
+                    primitives.luma_add_ps[size](reconIPred, reconIPredStride, pred, tsResiY, stride, trSize);
+                    nonZeroPsyEnergyY = m_rdCost.psyCost(size, fencYuv->getLumaAddr(absPartIdx), fencYuv->getStride(),
+                                                         cu->m_pic->getPicYuvRec()->getLumaAddr(cu->getAddr(), zorder),
+                                                         cu->m_pic->getPicYuvRec()->getStride());
+                    singleCostY = m_rdCost.calcPsyRdCost(nonZeroDistY, skipSingleBitsY, nonZeroPsyEnergyY);
+                }
+                else
+                    singleCostY = m_rdCost.calcRdCost(nonZeroDistY, skipSingleBitsY);
+            }
+
+            if (!numSigTSkipY || minCost[TEXT_LUMA][0] < singleCostY)
+                cu->setTransformSkipSubParts(0, TEXT_LUMA, absPartIdx, depth);
+            else
+            {
+                singleDistComp[TEXT_LUMA][0] = nonZeroDistY;
+                singlePsyEnergyComp[TEXT_LUMA][0] = nonZeroPsyEnergyY;
+                numSigY = numSigTSkipY;
+                bestTransformMode[TEXT_LUMA][0] = 1;
+                memcpy(coeffCurY, tsCoeffY, sizeof(coeff_t) * numCoeffY);
+                primitives.square_copy_ss[sizeIdx](curResiY, strideResiY, tsResiY, trSize);
+            }
+
+            cu->setCbfSubParts(numSigY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+        }
+
+        if (bCodeChroma && checkTransformSkipUV)
+        {
+            uint32_t nonZeroDistU = 0, nonZeroDistV = 0;
+            uint32_t nonZeroPsyEnergyU = 0, nonZeroPsyEnergyV = 0;
+            uint64_t singleCostU = MAX_INT64;
+            uint64_t singleCostV = MAX_INT64;
+
+            m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+
+            TURecurse tuIterator(splitIntoSubTUs ? VERTICAL_SPLIT : DONT_SPLIT, absPartIdxStep, absPartIdx);
+
+            int partSizeC = partitionFromLog2Size(log2TrSizeC);
+
+            do
+            {
+                uint32_t absPartIdxC = tuIterator.absPartIdxTURelCU;
+                uint32_t subTUOffset = tuIterator.section << (log2TrSizeC * 2);
+
+                int16_t *curResiU = m_qtTempShortYuv[qtLayer].getCbAddr(absPartIdxC);
+                int16_t *curResiV = m_qtTempShortYuv[qtLayer].getCrAddr(absPartIdxC);
+
+                ALIGN_VAR_32(coeff_t, tsCoeffU[MAX_TS_SIZE * MAX_TS_SIZE]);
+                ALIGN_VAR_32(int16_t, tsResiU[MAX_TS_SIZE * MAX_TS_SIZE]);
+                ALIGN_VAR_32(coeff_t, tsCoeffV[MAX_TS_SIZE * MAX_TS_SIZE]);
+                ALIGN_VAR_32(int16_t, tsResiV[MAX_TS_SIZE * MAX_TS_SIZE]);
+
+                cu->setTransformSkipPartRange(1, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setTransformSkipPartRange(1, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                if (m_bEnableRDOQ)
+                    m_entropyCoder->estBit(m_entropyCoder->m_estBitsSbac, log2TrSizeC, false);
+
+                uint32_t numSigTSkipU = m_quant.transformNxN(cu, fencYuv->getCbAddr(absPartIdxC), fencYuv->getCStride(), resiYuv->getCbAddr(absPartIdxC), resiYuv->m_cwidth, tsCoeffU,
+                                                             log2TrSizeC, TEXT_CHROMA_U, absPartIdxC, true);
+                uint32_t numSigTSkipV = m_quant.transformNxN(cu, fencYuv->getCrAddr(absPartIdxC), fencYuv->getCStride(), resiYuv->getCrAddr(absPartIdxC), resiYuv->m_cwidth, tsCoeffV,
+                                                             log2TrSizeC, TEXT_CHROMA_V, absPartIdxC, true);
+
+                cu->setCbfPartRange(numSigTSkipU ? setCbf : 0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setCbfPartRange(numSigTSkipV ? setCbf : 0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+
+                m_entropyCoder->resetBits();
+                singleBitsComp[TEXT_CHROMA_U][tuIterator.section] = 0;
+
+                if (numSigTSkipU)
+                {
+                    m_entropyCoder->codeQtCbf(cu, absPartIdxC, TEXT_CHROMA_U, trMode);
+                    m_entropyCoder->codeCoeffNxN(cu, tsCoeffU, absPartIdxC, log2TrSizeC, TEXT_CHROMA_U);
+                    singleBitsComp[TEXT_CHROMA_U][tuIterator.section] = m_entropyCoder->getNumberOfWrittenBits();
+
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), tsResiU, trSizeC, tsCoeffU,
+                                            log2TrSizeC, TEXT_CHROMA_U, false, true, numSigTSkipU);
+                    uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCbAddr(absPartIdxC), resiYuv->m_cwidth,
+                                                                 tsResiU, trSizeC);
+                    nonZeroDistU = m_rdCost.scaleChromaDistCb(dist);
+                    if (m_rdCost.m_psyRd)
+                    {
+                        pixel*   pred = predYuv->getCbAddr(absPartIdxC);
+                        uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                        pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder);
+                        uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+                        uint32_t stride = fencYuv->getCStride();
+                        //===== reconstruction =====
+                        int size = log2TrSizeC - 2;
+                        primitives.luma_add_ps[size](reconIPred, reconIPredStride, pred, tsResiU, stride, trSizeC);
+                        nonZeroPsyEnergyU = m_rdCost.psyCost(size, fencYuv->getCbAddr(absPartIdxC), fencYuv->getCStride(),
+                            cu->m_pic->getPicYuvRec()->getCbAddr(cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getCStride());
+                        singleCostU = m_rdCost.calcPsyRdCost(nonZeroDistU, singleBitsComp[TEXT_CHROMA_U][tuIterator.section], nonZeroPsyEnergyU);
+                    }
+                    else
+                        singleCostU = m_rdCost.calcRdCost(nonZeroDistU, singleBitsComp[TEXT_CHROMA_U][tuIterator.section]);
+                }
+
+                if (!numSigTSkipU || minCost[TEXT_CHROMA_U][tuIterator.section] < singleCostU)
+                    cu->setTransformSkipPartRange(0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                else
+                {
+                    singleDistComp[TEXT_CHROMA_U][tuIterator.section] = nonZeroDistU;
+                    singlePsyEnergyComp[TEXT_CHROMA_U][tuIterator.section] = nonZeroPsyEnergyU;
+                    numSigU[tuIterator.section] = numSigTSkipU;
+                    bestTransformMode[TEXT_CHROMA_U][tuIterator.section] = 1;
+                    memcpy(coeffCurU + subTUOffset, tsCoeffU, sizeof(coeff_t) * numCoeffC);
+                    primitives.square_copy_ss[sizeIdxC](curResiU, strideResiC, tsResiU, trSizeC);
+                }
+
+                if (numSigTSkipV)
+                {
+                    m_entropyCoder->codeQtCbf(cu, absPartIdxC, TEXT_CHROMA_V, trMode);
+                    m_entropyCoder->codeCoeffNxN(cu, tsCoeffV, absPartIdxC, log2TrSizeC, TEXT_CHROMA_V);
+                    singleBitsComp[TEXT_CHROMA_V][tuIterator.section] = m_entropyCoder->getNumberOfWrittenBits() - singleBitsComp[TEXT_CHROMA_U][tuIterator.section];
+
+                    m_quant.invtransformNxN(cu->getCUTransquantBypass(absPartIdxC), tsResiV, trSizeC, tsCoeffV,
+                                            log2TrSizeC, TEXT_CHROMA_V, false, true, numSigTSkipV);
+                    uint32_t dist = primitives.sse_ss[partSizeC](resiYuv->getCrAddr(absPartIdxC), resiYuv->m_cwidth,
+                                                                 tsResiV, trSizeC);
+                    nonZeroDistV = m_rdCost.scaleChromaDistCr(dist);
+                    if (m_rdCost.m_psyRd)
+                    {
+                        pixel*   pred = predYuv->getCrAddr(absPartIdxC);
+                        uint32_t zorder = cu->getZorderIdxInCU() + absPartIdxC;
+                        pixel*   reconIPred = cu->m_pic->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder);
+                        uint32_t reconIPredStride = cu->m_pic->getPicYuvRec()->getCStride();
+                        uint32_t stride = fencYuv->getCStride();
+                        //===== reconstruction =====
+                        int size = log2TrSizeC - 2;
+                        primitives.luma_add_ps[size](reconIPred, reconIPredStride, pred, tsResiV, stride, trSizeC);
+                        nonZeroPsyEnergyV = m_rdCost.psyCost(size, fencYuv->getCrAddr(absPartIdxC), fencYuv->getCStride(),
+                            cu->m_pic->getPicYuvRec()->getCrAddr(cu->getAddr(), zorder), cu->m_pic->getPicYuvRec()->getCStride());
+                        singleCostV = m_rdCost.calcPsyRdCost(nonZeroDistV, singleBitsComp[TEXT_CHROMA_V][tuIterator.section], nonZeroPsyEnergyV);
+                    }
+                    else
+                        singleCostV = m_rdCost.calcRdCost(nonZeroDistV, singleBitsComp[TEXT_CHROMA_V][tuIterator.section]);
+                }
+
+                if (!numSigTSkipV || minCost[TEXT_CHROMA_V][tuIterator.section] < singleCostV)
+                    cu->setTransformSkipPartRange(0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+                else
+                {
+                    singleDistComp[TEXT_CHROMA_V][tuIterator.section] = nonZeroDistV;
+                    singlePsyEnergyComp[TEXT_CHROMA_V][tuIterator.section] = nonZeroPsyEnergyV;
+                    numSigV[tuIterator.section] = numSigTSkipV;
+                    bestTransformMode[TEXT_CHROMA_V][tuIterator.section] = 1;
+                    memcpy(coeffCurV + subTUOffset, tsCoeffV, sizeof(coeff_t) * numCoeffC);
+                    primitives.square_copy_ss[sizeIdxC](curResiV, strideResiC, tsResiV, trSizeC);
+                }
+
+                cu->setCbfPartRange(numSigU[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_U, absPartIdxC, tuIterator.absPartIdxStep);
+                cu->setCbfPartRange(numSigV[tuIterator.section] ? setCbf : 0, TEXT_CHROMA_V, absPartIdxC, tuIterator.absPartIdxStep);
+            }
+            while (tuIterator.isNextSection());
+        }
+
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+
+        m_entropyCoder->resetBits();
+
+        if (log2TrSize > depthRange[0])
+            m_entropyCoder->codeTransformSubdivFlag(0, 5 - log2TrSize);
+
+        if (bCodeChroma)
+        {
+            if (splitIntoSubTUs)
+            {
+                offsetSubTUCBFs(cu, TEXT_CHROMA_U, trMode, absPartIdx);
+                offsetSubTUCBFs(cu, TEXT_CHROMA_V, trMode, absPartIdx);
+            }
+
+            uint32_t trHeightC = (m_csp == X265_CSP_I422) ? (trSizeC << 1) : trSizeC;
+            m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, trSizeC, trHeightC, TEXT_CHROMA_U, trMode, true);
+            m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, trSizeC, trHeightC, TEXT_CHROMA_V, trMode, true);
+        }
+
+        m_entropyCoder->codeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
+        if (numSigY)
+            m_entropyCoder->codeCoeffNxN(cu, coeffCurY, absPartIdx, log2TrSize, TEXT_LUMA);
+
+        if (bCodeChroma)
+        {
+            if (!splitIntoSubTUs)
+            {
+                if (numSigU[0])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurU, absPartIdx, log2TrSizeC, TEXT_CHROMA_U);
+                if (numSigV[0])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurV, absPartIdx, log2TrSizeC, TEXT_CHROMA_V);
+            }
+            else
+            {
+                uint32_t subTUSize = 1 << (log2TrSizeC * 2);
+                uint32_t partIdxesPerSubTU = absPartIdxStep >> 1;
+
+                if (numSigU[0])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurU, absPartIdx, log2TrSizeC, TEXT_CHROMA_U);
+                if (numSigU[1])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurU + subTUSize, absPartIdx + partIdxesPerSubTU, log2TrSizeC, TEXT_CHROMA_U);
+                if (numSigV[0])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurV, absPartIdx, log2TrSizeC, TEXT_CHROMA_V);
+                if (numSigV[1])
+                    m_entropyCoder->codeCoeffNxN(cu, coeffCurV + subTUSize, absPartIdx + partIdxesPerSubTU, log2TrSizeC, TEXT_CHROMA_V);
+            }
+        }
+
+        singleDist += singleDistComp[TEXT_LUMA][0];
+        singlePsyEnergy += singlePsyEnergyComp[TEXT_LUMA][0];// need to check we need to add chroma also
+        for (uint32_t subTUIndex = 0; subTUIndex < 2; subTUIndex++)
+        {
+            singleDist += singleDistComp[TEXT_CHROMA_U][subTUIndex];
+            singleDist += singleDistComp[TEXT_CHROMA_V][subTUIndex];
+        }
+
+        singleBits = m_entropyCoder->getNumberOfWrittenBits();
+        if (m_rdCost.m_psyRd)
+            singleCost = m_rdCost.calcPsyRdCost(singleDist, singleBits, singlePsyEnergy);
+        else
+            singleCost = m_rdCost.calcRdCost(singleDist, singleBits);
+
+        bestCBF[TEXT_LUMA] = cu->getCbf(absPartIdx, TEXT_LUMA, trMode);
+        if (bCodeChroma)
+        {
+            for (uint32_t chromId = TEXT_CHROMA_U; chromId <= TEXT_CHROMA_V; chromId++)
+            {
+                bestCBF[chromId] = cu->getCbf(absPartIdx, (TextType)chromId, trMode);
+                if (splitIntoSubTUs)
+                {
+                    uint32_t partIdxesPerSubTU = absPartIdxStep >> 1;
+                    for (uint32_t subTU = 0; subTU < 2; subTU++)
+                        bestsubTUCBF[chromId][subTU] = cu->getCbf((absPartIdx + (subTU * partIdxesPerSubTU)), (TextType)chromId, subTUDepth);
+                }
+            }
+        }
+    }
+
+    // code sub-blocks
+    if (bCheckSplit)
+    {
+        if (bCheckFull)
+        {
+            m_entropyCoder->store(m_rdEntropyCoders[depth][CI_QT_TRAFO_TEST]);
+            m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+        }
+        uint32_t subdivDist = 0;
+        uint32_t subdivBits = 0;
+        uint64_t subDivCost = 0;
+        uint32_t subDivPsyEnergy = 0;
+        bestCBF[TEXT_LUMA] = cu->getCbf(absPartIdx, TEXT_LUMA, trMode);
+        if (bCodeChroma)
+        {
+            for (uint32_t chromId = TEXT_CHROMA_U; chromId <= TEXT_CHROMA_V; chromId++)
+            {
+                bestCBF[chromId] = cu->getCbf(absPartIdx, (TextType)chromId, trMode);
+                if (splitIntoSubTUs)
+                {
+                    uint32_t partIdxesPerSubTU     = absPartIdxStep >> 1;
+                    for (uint32_t subTU = 0; subTU < 2; subTU++)
+                        bestsubTUCBF[chromId][subTU] = cu->getCbf((absPartIdx + (subTU * partIdxesPerSubTU)), (TextType)chromId, subTUDepth);
+                }
+            }
+        }
+
+        const uint32_t qPartNumSubdiv = NUM_CU_PARTITIONS >> ((depth + 1) << 1);
+        for (uint32_t i = 0; i < 4; ++i)
+        {
+            cu->m_psyEnergy = 0;
+            subdivDist += xEstimateResidualQT(cu, absPartIdx + i * qPartNumSubdiv, fencYuv, predYuv, resiYuv, depth + 1, subDivCost, subdivBits, bCheckFull ? NULL : outZeroDist, depthRange);
+            subDivPsyEnergy += cu->m_psyEnergy;
+        }
+
+        uint32_t ycbf = 0;
+        uint32_t ucbf = 0;
+        uint32_t vcbf = 0;
+        for (uint32_t i = 0; i < 4; ++i)
+        {
+            ycbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_LUMA,     trMode + 1);
+            ucbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_U, trMode + 1);
+            vcbf |= cu->getCbf(absPartIdx + i * qPartNumSubdiv, TEXT_CHROMA_V, trMode + 1);
+        }
+
+        for (uint32_t i = 0; i < 4 * qPartNumSubdiv; ++i)
+        {
+            cu->getCbf(TEXT_LUMA)[absPartIdx + i]     |= ycbf << trMode;
+            cu->getCbf(TEXT_CHROMA_U)[absPartIdx + i] |= ucbf << trMode;
+            cu->getCbf(TEXT_CHROMA_V)[absPartIdx + i] |= vcbf << trMode;
+        }
+
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_ROOT]);
+        m_entropyCoder->resetBits();
+
+        xEncodeResidualQT(cu, absPartIdx, depth, true,  TEXT_LUMA, depthRange);
+        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_LUMA, depthRange);
+        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_CHROMA_U, depthRange);
+        xEncodeResidualQT(cu, absPartIdx, depth, false, TEXT_CHROMA_V, depthRange);
+
+        subdivBits = m_entropyCoder->getNumberOfWrittenBits();
+
+        if (m_rdCost.m_psyRd)
+            subDivCost = m_rdCost.calcPsyRdCost(subdivDist, subdivBits, subDivPsyEnergy);
+        else
+            subDivCost = m_rdCost.calcRdCost(subdivDist, subdivBits);
+
+        if (ycbf || ucbf || vcbf || !bCheckFull)
+        {
+            if (subDivCost < singleCost)
+            {
+                rdCost += subDivCost;
+                outBits += subdivBits;
+                outDist += subdivDist;
+                cu->m_psyEnergy = subDivPsyEnergy;
+                return outDist;
+            }
+            else
+                cu->m_psyEnergy = singlePsyEnergy;
+        }
+
+        cu->setTransformSkipSubParts(bestTransformMode[TEXT_LUMA][0], TEXT_LUMA, absPartIdx, depth);
+        if (bCodeChroma)
+        {
+            const uint32_t numberOfSections  = splitIntoSubTUs ? 2 : 1;
+
+            uint32_t partIdxesPerSubTU  = absPartIdxStep >> (splitIntoSubTUs ? 1 : 0);
+            for (uint32_t subTUIndex = 0; subTUIndex < numberOfSections; subTUIndex++)
+            {
+                const uint32_t  subTUPartIdx = absPartIdx + (subTUIndex * partIdxesPerSubTU);
+
+                cu->setTransformSkipPartRange(bestTransformMode[TEXT_CHROMA_U][subTUIndex], TEXT_CHROMA_U, subTUPartIdx, partIdxesPerSubTU);
+                cu->setTransformSkipPartRange(bestTransformMode[TEXT_CHROMA_V][subTUIndex], TEXT_CHROMA_V, subTUPartIdx, partIdxesPerSubTU);
+            }
+        }
+        X265_CHECK(bCheckFull, "check-full must be set\n");
+        m_entropyCoder->load(m_rdEntropyCoders[depth][CI_QT_TRAFO_TEST]);
+    }
+
+    rdCost += singleCost;
+    outBits += singleBits;
+    outDist += singleDist;
+    cu->m_psyEnergy = singlePsyEnergy;
+
+    cu->setTrIdxSubParts(trMode, absPartIdx, depth);
+    cu->setCbfSubParts(numSigY ? setCbf : 0, TEXT_LUMA, absPartIdx, depth);
+
+    if (bCodeChroma)
+    {
+        const uint32_t numberOfSections  = splitIntoSubTUs ? 2 : 1;
+        uint32_t partIdxesPerSubTU  = absPartIdxStep >> (splitIntoSubTUs ? 1 : 0);
+
+        for (uint32_t chromaId = TEXT_CHROMA_U; chromaId <= TEXT_CHROMA_V; chromaId++)
+        {
+            for (uint32_t subTUIndex = 0; subTUIndex < numberOfSections; subTUIndex++)
+            {
+                const uint32_t  subTUPartIdx = absPartIdx + (subTUIndex * partIdxesPerSubTU);
+
+                if (splitIntoSubTUs)
+                {
+                    const uint8_t combinedCBF = (uint8_t)((bestsubTUCBF[chromaId][subTUIndex] << subTUDepth) | (bestCBF[chromaId] << trMode));
+                    cu->setCbfPartRange(combinedCBF, (TextType)chromaId, subTUPartIdx, partIdxesPerSubTU);
+                }
+                else
+                    cu->setCbfPartRange((bestCBF[chromaId] << trMode), (TextType)chromaId, subTUPartIdx, partIdxesPerSubTU);
+            }
+        }
+    }
+
+    return outDist;
+}
+
+void Search::xEncodeResidualQT(TComDataCU* cu, uint32_t absPartIdx, const uint32_t depth, bool bSubdivAndCbf, TextType ttype, uint32_t depthRange[2])
+{
+    X265_CHECK(cu->getDepth(0) == cu->getDepth(absPartIdx), "depth not matching\n");
+    const uint32_t curTrMode   = depth - cu->getDepth(0);
+    const uint32_t trMode      = cu->getTransformIdx(absPartIdx);
+    const bool     bSubdiv     = curTrMode != trMode;
+    const uint32_t log2TrSize  = g_maxLog2CUSize - depth;
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+    
+    const bool splitIntoSubTUs = (m_csp == X265_CSP_I422);
+
+    if (bSubdivAndCbf && log2TrSize <= depthRange[1] && log2TrSize > depthRange[0])
+        m_entropyCoder->codeTransformSubdivFlag(bSubdiv, 5 - log2TrSize);
+
+    X265_CHECK(cu->getPredictionMode(absPartIdx) != MODE_INTRA, "xEncodeResidualQT() with intra block\n");
+
+    bool mCodeAll = true;
+    uint32_t trWidthC  = 1 << log2TrSizeC;
+    uint32_t trHeightC = splitIntoSubTUs ? (trWidthC << 1) : trWidthC;
+
+    const uint32_t numPels = trWidthC * trHeightC;
+    if (numPels < (MIN_TU_SIZE * MIN_TU_SIZE))
+        mCodeAll = false;
+
+    if (bSubdivAndCbf)
+    {
+        const bool bFirstCbfOfCU = curTrMode == 0;
+        if (bFirstCbfOfCU || mCodeAll)
+        {
+            uint32_t absPartIdxStep = NUM_CU_PARTITIONS >> ((cu->getDepth(0) +  curTrMode) << 1);
+            if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode - 1))
+                m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, trWidthC, trHeightC, TEXT_CHROMA_U, curTrMode, !bSubdiv);
+            if (bFirstCbfOfCU || cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode - 1))
+                m_entropyCoder->codeQtCbf(cu, absPartIdx, absPartIdxStep, trWidthC, trHeightC, TEXT_CHROMA_V, curTrMode, !bSubdiv);
+        }
+        else
+        {
+            X265_CHECK(cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode) == cu->getCbf(absPartIdx, TEXT_CHROMA_U, curTrMode - 1), "chroma CBF not matching\n");
+            X265_CHECK(cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode) == cu->getCbf(absPartIdx, TEXT_CHROMA_V, curTrMode - 1), "chroma CBF not matching\n");
+        }
+    }
+
+    if (!bSubdiv)
+    {
+        //Luma
+        const uint32_t qtLayer = log2TrSize - 2;
+        uint32_t coeffOffsetY = absPartIdx << (LOG2_UNIT_SIZE * 2);
+        coeff_t* coeffCurY = m_qtTempCoeff[0][qtLayer] + coeffOffsetY;
+
+        //Chroma
+        bool bCodeChroma = true;
+        uint32_t trModeC = trMode;
+        if ((log2TrSize == 2) && !(m_csp == X265_CSP_I444))
+        {
+            log2TrSizeC++;
+            trModeC--;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((depth - 1) << 1);
+            bCodeChroma = ((absPartIdx & (qpdiv - 1)) == 0);
+        }
+
+        if (bSubdivAndCbf)
+            m_entropyCoder->codeQtCbf(cu, absPartIdx, TEXT_LUMA, trMode);
+        else
+        {
+            if (ttype == TEXT_LUMA && cu->getCbf(absPartIdx, TEXT_LUMA, trMode))
+                m_entropyCoder->codeCoeffNxN(cu, coeffCurY, absPartIdx, log2TrSize, TEXT_LUMA);
+
+            if (bCodeChroma)
+            {
+                uint32_t coeffOffsetC = coeffOffsetY >> (hChromaShift + vChromaShift);
+                coeff_t* coeffCurU = m_qtTempCoeff[1][qtLayer] + coeffOffsetC;
+                coeff_t* coeffCurV = m_qtTempCoeff[2][qtLayer] + coeffOffsetC;
+
+                if (!splitIntoSubTUs)
+                {
+                    if (ttype == TEXT_CHROMA_U && cu->getCbf(absPartIdx, TEXT_CHROMA_U, trMode))
+                        m_entropyCoder->codeCoeffNxN(cu, coeffCurU, absPartIdx, log2TrSizeC, TEXT_CHROMA_U);
+                    if (ttype == TEXT_CHROMA_V && cu->getCbf(absPartIdx, TEXT_CHROMA_V, trMode))
+                        m_entropyCoder->codeCoeffNxN(cu, coeffCurV, absPartIdx, log2TrSizeC, TEXT_CHROMA_V);
+                }
+                else
+                {
+                    uint32_t partIdxesPerSubTU  = NUM_CU_PARTITIONS >> (((cu->getDepth(absPartIdx) + trModeC) << 1) + 1);
+                    uint32_t subTUSize = 1 << (log2TrSizeC * 2);
+                    if (ttype == TEXT_CHROMA_U && cu->getCbf(absPartIdx, TEXT_CHROMA_U, trMode))
+                    {
+                        if (cu->getCbf(absPartIdx, ttype, trMode + 1))
+                            m_entropyCoder->codeCoeffNxN(cu, coeffCurU, absPartIdx, log2TrSizeC, TEXT_CHROMA_U);
+                        if (cu->getCbf(absPartIdx + partIdxesPerSubTU, ttype, trMode + 1))
+                            m_entropyCoder->codeCoeffNxN(cu, coeffCurU + subTUSize, absPartIdx + partIdxesPerSubTU, log2TrSizeC, TEXT_CHROMA_U);
+                    }
+                    if (ttype == TEXT_CHROMA_V && cu->getCbf(absPartIdx, TEXT_CHROMA_V, trMode))
+                    {
+                        if (cu->getCbf(absPartIdx, ttype, trMode + 1))
+                            m_entropyCoder->codeCoeffNxN(cu, coeffCurV, absPartIdx, log2TrSizeC, TEXT_CHROMA_V);
+                        if (cu->getCbf(absPartIdx + partIdxesPerSubTU, ttype, trMode + 1))
+                            m_entropyCoder->codeCoeffNxN(cu, coeffCurV + subTUSize, absPartIdx + partIdxesPerSubTU, log2TrSizeC, TEXT_CHROMA_V);
+                    }
+                }
+            }
+        }
+    }
+    else
+    {
+        if (bSubdivAndCbf || cu->getCbf(absPartIdx, ttype, curTrMode))
+        {
+            const uint32_t qpartNumSubdiv = NUM_CU_PARTITIONS >> ((depth + 1) << 1);
+            for (uint32_t i = 0; i < 4; ++i)
+                xEncodeResidualQT(cu, absPartIdx + i * qpartNumSubdiv, depth + 1, bSubdivAndCbf, ttype, depthRange);
+        }
+    }
+}
+
+void Search::xSetResidualQTData(TComDataCU* cu, uint32_t absPartIdx, ShortYuv* resiYuv, uint32_t depth, bool bSpatial)
+{
+    X265_CHECK(cu->getDepth(0) == cu->getDepth(absPartIdx), "depth not matching\n");
+    const uint32_t curTrMode = depth - cu->getDepth(0);
+    const uint32_t trMode    = cu->getTransformIdx(absPartIdx);
+
+    int hChromaShift = CHROMA_H_SHIFT(m_csp);
+    int vChromaShift = CHROMA_V_SHIFT(m_csp);
+
+    if (curTrMode == trMode)
+    {
+        const uint32_t log2TrSize = g_maxLog2CUSize - depth;
+        const uint32_t qtLayer    = log2TrSize - 2;
+
+        uint32_t log2TrSizeC = log2TrSize - hChromaShift;
+        bool bCodeChroma = true;
+        uint32_t trModeC = trMode;
+        if ((log2TrSize == 2) && !(m_csp == X265_CSP_I444))
+        {
+            log2TrSizeC++;
+            trModeC--;
+            uint32_t qpdiv = NUM_CU_PARTITIONS >> ((cu->getDepth(0) + trModeC) << 1);
+            bCodeChroma = ((absPartIdx & (qpdiv - 1)) == 0);
+        }
+
+        if (bSpatial)
+        {
+            m_qtTempShortYuv[qtLayer].copyPartToPartLuma(resiYuv, absPartIdx, log2TrSize);
+            if (bCodeChroma)
+                m_qtTempShortYuv[qtLayer].copyPartToPartChroma(resiYuv, absPartIdx, log2TrSizeC + hChromaShift);
+        }
+        else
+        {
+            uint32_t numCoeffY = 1 << (log2TrSize * 2);
+            uint32_t coeffOffsetY = absPartIdx << LOG2_UNIT_SIZE * 2;
+            coeff_t* coeffSrcY = m_qtTempCoeff[0][qtLayer] + coeffOffsetY;
+            coeff_t* coeffDstY = cu->getCoeffY()           + coeffOffsetY;
+            ::memcpy(coeffDstY, coeffSrcY, sizeof(coeff_t) * numCoeffY);
+            if (bCodeChroma)
+            {
+                uint32_t numCoeffC = 1 << (log2TrSizeC * 2 + (m_csp == X265_CSP_I422));
+                uint32_t coeffOffsetC = coeffOffsetY >> (hChromaShift + vChromaShift);
+
+                coeff_t* coeffSrcU = m_qtTempCoeff[1][qtLayer] + coeffOffsetC;
+                coeff_t* coeffSrcV = m_qtTempCoeff[2][qtLayer] + coeffOffsetC;
+                coeff_t* coeffDstU = cu->getCoeffCb()          + coeffOffsetC;
+                coeff_t* coeffDstV = cu->getCoeffCr()          + coeffOffsetC;
+                ::memcpy(coeffDstU, coeffSrcU, sizeof(coeff_t) * numCoeffC);
+                ::memcpy(coeffDstV, coeffSrcV, sizeof(coeff_t) * numCoeffC);
+            }
+        }
+    }
+    else
+    {
+        const uint32_t qPartNumSubdiv =  NUM_CU_PARTITIONS >> ((depth + 1) << 1);
+        for (uint32_t i = 0; i < 4; ++i)
+            xSetResidualQTData(cu, absPartIdx + i * qPartNumSubdiv, resiYuv, depth + 1, bSpatial);
+    }
+}
+
+uint32_t Search::getIntraModeBits(TComDataCU* cu, uint32_t mode, uint32_t partOffset, uint32_t depth)
+{
+    // Reload only contexts required for coding intra mode information
+    m_entropyCoder->loadIntraDirModeLuma(m_rdEntropyCoders[depth][CI_CURR_BEST]);
+
+    cu->getLumaIntraDir()[partOffset] = (uint8_t)mode;
+
+    m_entropyCoder->resetBits();
+    m_entropyCoder->codeIntraDirLumaAng(cu, partOffset, false);
+
+    return m_entropyCoder->getNumberOfWrittenBits();
+}
+
+/* returns the number of bits required to signal a non-most-probable mode.
+ * on return mpm contains bitmap of most probable modes */
+uint32_t Search::getIntraRemModeBits(TComDataCU* cu, uint32_t partOffset, uint32_t depth, uint32_t preds[3], uint64_t& mpms)
+{
+    mpms = 0;
+    for (int i = 0; i < 3; ++i)
+        mpms |= ((uint64_t)1 << preds[i]);
+
+    uint32_t mode = 34;
+    while (mpms & ((uint64_t)1 << mode))
+        --mode;
+
+    return getIntraModeBits(cu, mode, partOffset, depth);
+}
+
+/* swap the current mode/cost with the mode with the highest cost in the
+ * current candidcate list, if its cost is better (maintain a top N list) */
+void Search::updateCandList(uint32_t mode, uint64_t cost, int maxCandCount, uint32_t* candModeList, uint64_t* candCostList)
+{
+    uint32_t maxIndex = 0;
+    uint64_t maxValue = 0;
+
+    for (int i = 0; i < maxCandCount; i++)
+    {
+        if (maxValue < candCostList[i])
+        {
+            maxValue = candCostList[i];
+            maxIndex = i;
+        }
+    }
+
+    if (cost < maxValue)
+    {
+        candCostList[maxIndex] = cost;
+        candModeList[maxIndex] = mode;
+    }
+}
+
+/* add inter-prediction syntax elements for a CU block */
+uint32_t Search::getInterSymbolBits(TComDataCU* cu, uint32_t depthRange[2])
+{
+    if (cu->getMergeFlag(0) && cu->getPartitionSize(0) == SIZE_2Nx2N && !cu->getQtRootCbf(0))
+    {
+        cu->setSkipFlagSubParts(true, 0, cu->getDepth(0));
+
+        m_entropyCoder->resetBits();
+        if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+            m_entropyCoder->codeCUTransquantBypassFlag(cu->getCUTransquantBypass(0));
+        if (!cu->m_slice->isIntra())
+            m_entropyCoder->codeSkipFlag(cu, 0);
+        m_entropyCoder->codeMergeIndex(cu, 0);
+        cu->m_mvBits = m_entropyCoder->getNumberOfWrittenBits();
+        cu->m_coeffBits = 0;
+        return m_entropyCoder->getNumberOfWrittenBits();
+    }
+    else
+    {
+        m_entropyCoder->resetBits();
+        if (cu->m_slice->m_pps->bTransquantBypassEnabled)
+            m_entropyCoder->codeCUTransquantBypassFlag(cu->getCUTransquantBypass(0));
+        if (!cu->m_slice->isIntra())
+        {
+            m_entropyCoder->codeSkipFlag(cu, 0);
+            m_entropyCoder->codePredMode(cu->getPredictionMode(0));
+        }
+        m_entropyCoder->codePartSize(cu, 0, cu->getDepth(0));
+        m_entropyCoder->codePredInfo(cu, 0);
+        bool bDummy = false;
+        cu->m_mvBits = m_entropyCoder->getNumberOfWrittenBits();
+        m_entropyCoder->codeCoeff(cu, 0, cu->getDepth(0), bDummy, depthRange);
+        int totalBits = m_entropyCoder->getNumberOfWrittenBits();
+        cu->m_coeffBits = totalBits - cu->m_mvBits;
+        return totalBits;
+    }
+}

x265-hg20140928.tar.bz2/source/encoder/search.h Added

@@ -0,0 +1,184 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_SEARCH_H
+#define X265_SEARCH_H
+
+#include "common.h"
+#include "predict.h"
+#include "quant.h"
+#include "bitcost.h"
+
+#include "TLibCommon/TComYuv.h"
+#include "TLibCommon/TComMotionInfo.h"
+#include "TLibCommon/TComPattern.h"
+
+#include "rdcost.h"
+#include "motion.h"
+
+#define MVP_IDX_BITS 1
+#define NUM_LAYERS 4
+
+namespace x265 {
+// private namespace
+
+class Entropy;
+
+inline int getTUBits(int idx, int numIdx)
+{
+    return idx + (idx < numIdx - 1);
+}
+
+class Search : public Predict
+{
+public:
+
+    static const pixel zeroPel[MAX_CU_SIZE];
+
+    MotionEstimate  m_me;
+    Quant           m_quant;
+    RDCost          m_rdCost;
+    x265_param*     m_param;
+
+    Entropy*        m_entropyCoder;
+    Entropy       (*m_rdEntropyCoders)[CI_NUM];
+
+    TComYuv         m_predTempYuv;
+    ShortYuv*       m_qtTempShortYuv;
+    coeff_t*        m_qtTempCoeff[3][NUM_LAYERS];
+    uint8_t*        m_qtTempCbf[3];
+    uint8_t*        m_qtTempTransformSkipFlag[3];
+
+    bool            m_bFrameParallel;
+    bool            m_bEnableRDOQ;
+    int             m_numLayers;
+    int             m_refLagPixels;
+
+    Search();
+    ~Search();
+
+    bool     initSearch(x265_param *param, ScalingList& scalingList);
+
+    void     estIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2]);
+    void     sharedEstIntraPredQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2], uint8_t* sharedModes);
+    void     estIntraPredChromaQT(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
+
+    // estimation inter prediction (non-skip)
+    bool     predInterSearch(TComDataCU* cu, TComYuv* predYuv, bool bMergeOnly, bool bChroma);
+
+    // encode residual and compute rd-cost for inter mode
+    void     encodeResAndCalcRdInterCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, ShortYuv* bestResiYuv, TComYuv* reconYuv);
+    void     encodeResAndCalcRdSkipCU(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, TComYuv* reconYuv);
+
+    void     generateCoeffRecon(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
+    void     residualTransformQuantInter(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, ShortYuv* resiYuv, uint32_t depth, uint32_t depthRange[2]);
+
+    uint32_t getIntraModeBits(TComDataCU* cu, uint32_t mode, uint32_t partOffset, uint32_t depth);
+    uint32_t getIntraRemModeBits(TComDataCU * cu, uint32_t partOffset, uint32_t depth, uint32_t preds[3], uint64_t& mpms);
+
+protected:
+
+    void     xSetResidualQTData(TComDataCU* cu, uint32_t absPartIdx, ShortYuv* resiYuv, uint32_t depth, bool bSpatial);
+    void     xSetIntraResultQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv);
+
+    void     xEncSubdivCbfQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx,  uint32_t absPartIdxStep, uint32_t width, uint32_t height);
+    void     xEncCoeffQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TextType ttype);
+    void     xEncIntraHeaderLuma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx);
+    void     xEncIntraHeaderChroma(TComDataCU* cu, uint32_t absPartIdx);
+
+    uint32_t xGetIntraBitsQTChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t absPartIdxStep);
+    uint32_t xGetIntraBitsLuma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, uint32_t log2TrSize, coeff_t* coeff, uint32_t depthRange[2]);
+    uint32_t xGetIntraBitsChroma(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSizeC, uint32_t chromaId, coeff_t* coeff);
+    uint32_t xIntraCodingLumaBlk(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSize, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
+                                 int16_t* reconQt, uint32_t reconQtStride, coeff_t* coeff, uint32_t& cbf);
+
+    uint32_t xEstimateResidualQT(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv, uint32_t depth,
+                                 uint64_t &rdCost, uint32_t &outBits, uint32_t *zeroDist, uint32_t tuDepthRange[2]);
+
+    uint32_t xRecurIntraCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv,
+                                 ShortYuv* resiYuv, bool bAllowRQTSplit, uint64_t& dRDCost, uint32_t& puBits, uint32_t& psyEnergy, uint32_t depthRange[2]);
+
+    uint32_t xRecurIntraChromaCodingQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
+                                       uint32_t& psyEnergy);
+
+    uint32_t xIntraCodingChromaBlk(TComDataCU* cu, uint32_t absPartIdx, TComYuv* fencYuv, TComYuv* predYuv, ShortYuv* resiYuv,
+                                   int16_t* reconQt, uint32_t reconQtStride, coeff_t* coeff, uint32_t& cbf, uint32_t chromaId, uint32_t log2TrSizeC);
+
+    void     residualTransformQuantIntra(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
+                                         TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv, uint32_t depthRange[2]);
+
+    void     residualQTIntraChroma(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* fencYuv,
+                                   TComYuv* predYuv, ShortYuv* resiYuv, TComYuv* reconYuv);
+
+    void     xEncodeResidualQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t depth, bool bSubdivAndCbf, TextType ttype, uint32_t depthRange[2]);
+    void     xSetIntraResultChromaQT(TComDataCU* cu, uint32_t trDepth, uint32_t absPartIdx, TComYuv* reconYuv);
+
+    void     xLoadIntraResultQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSize, int16_t* reconQt, uint32_t reconQtStride);
+    void     xLoadIntraResultChromaQT(TComDataCU* cu, uint32_t absPartIdx, uint32_t log2TrSizeC, uint32_t chromaId, int16_t* reconQt, uint32_t reconQtStride);
+
+    void     offsetSubTUCBFs(TComDataCU* cu, TextType ttype, uint32_t trDepth, uint32_t absPartIdx);
+
+    struct MergeData
+    {
+        /* merge candidate data, cached between calls to mergeEstimation */
+        TComMvField mvFieldNeighbours[MRG_MAX_NUM_CANDS][2];
+        uint8_t     interDirNeighbours[MRG_MAX_NUM_CANDS];
+        uint32_t    maxNumMergeCand;
+
+        /* data updated for each partition */
+        uint32_t    absPartIdx;
+        int         width;
+        int         height;
+
+        /* outputs */
+        TComMvField mvField[2];
+        uint32_t    interDir;
+        uint32_t    index;
+        uint32_t    bits;
+    };
+
+    struct MotionData
+    {
+        MV  mv;
+        MV  mvp;
+        int mvpIdx;
+        int ref;
+        uint32_t cost;
+        int bits;
+    };
+
+    /* inter/ME helper functions */
+    void     checkBestMVP(MV* amvpCand, MV cMv, MV& mvPred, int& mvpIdx, uint32_t& outBits, uint32_t& outCost);
+    void     getBlkBits(PartSize cuMode, bool bPSlice, int partIdx, uint32_t lastMode, uint32_t blockBit[3]);
+    uint32_t getInterSymbolBits(TComDataCU* cu, uint32_t depthRange[2]);
+    uint32_t mergeEstimation(TComDataCU* cu, int partIdx, MergeData& m);
+    void     setSearchRange(TComDataCU* cu, MV mvp, int merange, MV& mvmin, MV& mvmax);
+
+    /* intra helper functions */
+    enum { MAX_RD_INTRA_MODES = 16 };
+    void     updateCandList(uint32_t mode, uint64_t cost, int maxCandCount, uint32_t* candModeList, uint64_t* candCostList);
+    void     getBestIntraModeChroma(TComDataCU* cu, TComYuv* fencYuv, TComYuv* predYuv);
+};
+}
+
+#endif // ifndef X265_SEARCH_H

x265-hg20140928.tar.bz2/source/encoder/sei.cpp Added

@@ -0,0 +1,74 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#include "common.h"
+#include "bitstream.h"
+#include "slice.h"
+#include "sei.h"
+
+using namespace x265;
+
+/* x265's identifying GUID */
+const uint8_t SEIuserDataUnregistered::m_uuid_iso_iec_11578[16] = {
+    0x2C, 0xA2, 0xDE, 0x09, 0xB5, 0x17, 0x47, 0xDB,
+    0xBB, 0x55, 0xA4, 0xFE, 0x7F, 0xC2, 0xFC, 0x4E
+};
+
+/* marshal a single SEI message sei, storing the marshalled representation
+ * in bitstream bs */
+void SEI::write(Bitstream& bs, const SPS& sps)
+{
+    BitCounter count;
+    m_bitIf = &count;
+
+    /* virtual writeSEI method, write to bit counter */
+    writeSEI(sps);
+
+    m_bitIf = &bs;
+    uint32_t type = payloadType();
+    for (; type >= 0xff; type -= 0xff)
+        WRITE_CODE(0xff, 8, "payload_type");
+    WRITE_CODE(type, 8, "payload_type");
+
+    X265_CHECK(0 == (count.getNumberOfWrittenBits() & 7), "payload unaligned\n");
+    uint32_t payloadSize = count.getNumberOfWrittenBits() >> 3;
+    for (; payloadSize >= 0xff; payloadSize -= 0xff)
+        WRITE_CODE(0xff, 8, "payload_size");
+    WRITE_CODE(payloadSize, 8, "payload_size");
+
+    /* virtual writeSEI method, write to bs */
+    writeSEI(sps);
+}
+
+void SEI::writeByteAlign()
+{
+    // TODO: expose bs.writeByteAlignment() as virtual function
+    if (m_bitIf->getNumberOfWrittenBits() % 8 != 0)
+    {
+        WRITE_FLAG(1, "bit_equal_to_one");
+        while (m_bitIf->getNumberOfWrittenBits() % 8 != 0)
+        {
+            WRITE_FLAG(0, "bit_equal_to_zero");
+        }
+    }
+}

x265-hg20140928.tar.bz2/source/encoder/sei.h Added

@@ -0,0 +1,281 @@
+/*****************************************************************************
+* Copyright (C) 2013 x265 project
+*
+* Authors: Steve Borho <steve@borho.org>
+*
+* This program is free software; you can redistribute it and/or modify
+* it under the terms of the GNU General Public License as published by
+* the Free Software Foundation; either version 2 of the License, or
+* (at your option) any later version.
+*
+* This program is distributed in the hope that it will be useful,
+* but WITHOUT ANY WARRANTY; without even the implied warranty of
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+* GNU General Public License for more details.
+*
+* You should have received a copy of the GNU General Public License
+* along with this program; if not, write to the Free Software
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+*
+* This program is also available under a commercial proprietary license.
+* For more information, contact us at license @ x265.com.
+*****************************************************************************/
+
+#ifndef X265_SEI_H
+#define X265_SEI_H
+
+#include "common.h"
+#include "bitstream.h"
+#include "slice.h"
+
+namespace x265 {
+// private namespace
+
+class SEI : public SyntaxElementWriter
+{
+public:
+
+    /* SEI users call write() to marshal an SEI to a bitstream. SEI
+     * subclasses may implement write() or accept the default write()
+     * method which calls writeSEI() with a bitcounter to determine
+     * the size, then it encodes the header and calls writeSEI a
+     * second time for the real encode. */
+    virtual void write(Bitstream& bs, const SPS& sps);
+
+    virtual ~SEI() {}
+
+protected:
+
+    enum PayloadType
+    {
+        BUFFERING_PERIOD                     = 0,
+        PICTURE_TIMING                       = 1,
+        PAN_SCAN_RECT                        = 2,
+        FILLER_PAYLOAD                       = 3,
+        USER_DATA_REGISTERED_ITU_T_T35       = 4,
+        USER_DATA_UNREGISTERED               = 5,
+        RECOVERY_POINT                       = 6,
+        SCENE_INFO                           = 9,
+        FULL_FRAME_SNAPSHOT                  = 15,
+        PROGRESSIVE_REFINEMENT_SEGMENT_START = 16,
+        PROGRESSIVE_REFINEMENT_SEGMENT_END   = 17,
+        FILM_GRAIN_CHARACTERISTICS           = 19,
+        POST_FILTER_HINT                     = 22,
+        TONE_MAPPING_INFO                    = 23,
+        FRAME_PACKING                        = 45,
+        DISPLAY_ORIENTATION                  = 47,
+        SOP_DESCRIPTION                      = 128,
+        ACTIVE_PARAMETER_SETS                = 129,
+        DECODING_UNIT_INFO                   = 130,
+        TEMPORAL_LEVEL0_INDEX                = 131,
+        DECODED_PICTURE_HASH                 = 132,
+        SCALABLE_NESTING                     = 133,
+        REGION_REFRESH_INFO                  = 134,
+    };
+
+    virtual PayloadType payloadType() const = 0;
+
+    virtual void writeSEI(const SPS&) { X265_CHECK(0, "empty writeSEI method called\n");  }
+
+    void writeByteAlign();
+};
+
+class SEIuserDataUnregistered : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return USER_DATA_UNREGISTERED; }
+
+    SEIuserDataUnregistered() : m_userData(NULL) {}
+
+    static const uint8_t m_uuid_iso_iec_11578[16];
+    uint32_t m_userDataLength;
+    uint8_t *m_userData;
+
+    void write(Bitstream& bs, const SPS&)
+    {
+        m_bitIf = &bs;
+
+        WRITE_CODE(USER_DATA_UNREGISTERED, 8, "payload_type");
+
+        uint32_t payloadSize = 16 + m_userDataLength;
+        for (; payloadSize >= 0xff; payloadSize -= 0xff)
+            WRITE_CODE(0xff, 8, "payload_size");
+        WRITE_CODE(payloadSize, 8, "payload_size");
+
+        for (uint32_t i = 0; i < 16; i++)
+            WRITE_CODE(m_uuid_iso_iec_11578[i], 8, "sei.uuid_iso_iec_11578[i]");
+
+        for (uint32_t i = 0; i < m_userDataLength; i++)
+            WRITE_CODE(m_userData[i], 8, "user_data");
+    }
+};
+
+class SEIDecodedPictureHash : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return DECODED_PICTURE_HASH; }
+
+    enum Method
+    {
+        MD5,
+        CRC,
+        CHECKSUM,
+    } m_method;
+
+    uint8_t m_digest[3][16];
+
+    void write(Bitstream& bs, const SPS&)
+    {
+        m_bitIf = &bs;
+
+        WRITE_CODE(DECODED_PICTURE_HASH, 8, "payload_type");
+
+        switch (m_method)
+        {
+        case MD5:
+            WRITE_CODE(1 + 16 * 3, 8, "payload_size");
+            WRITE_CODE(MD5, 8, "hash_type");
+            break;
+        case CRC:
+            WRITE_CODE(1 + 2 * 3, 8, "payload_size");
+            WRITE_CODE(CRC, 8, "hash_type");
+            break;
+        case CHECKSUM:
+            WRITE_CODE(1 + 4 * 3, 8, "payload_size");
+            WRITE_CODE(CHECKSUM, 8, "hash_type");
+            break;
+        }
+
+        for (int yuvIdx = 0; yuvIdx < 3; yuvIdx++)
+        {
+            if (m_method == MD5)
+            {
+                for (uint32_t i = 0; i < 16; i++)
+                    WRITE_CODE(m_digest[yuvIdx][i], 8, "picture_md5");
+            }
+            else if (m_method == CRC)
+            {
+                uint32_t val = (m_digest[yuvIdx][0] << 8) + m_digest[yuvIdx][1];
+                WRITE_CODE(val, 16, "picture_crc");
+            }
+            else if (m_method == CHECKSUM)
+            {
+                uint32_t val = (m_digest[yuvIdx][0] << 24) + (m_digest[yuvIdx][1] << 16) + (m_digest[yuvIdx][2] << 8) + m_digest[yuvIdx][3];
+                WRITE_CODE(val, 32, "picture_checksum");
+            }
+        }
+    }
+};
+
+class SEIActiveParameterSets : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return ACTIVE_PARAMETER_SETS; }
+
+    bool m_selfContainedCvsFlag;
+    bool m_noParamSetUpdateFlag;
+
+    void writeSEI(const SPS&)
+    {
+        WRITE_CODE(0, 4, "active_vps_id");
+        WRITE_FLAG(m_selfContainedCvsFlag, "self_contained_cvs_flag");
+        WRITE_FLAG(m_noParamSetUpdateFlag, "no_param_set_update_flag");
+        WRITE_UVLC(0, "num_sps_ids_minus1");
+        WRITE_UVLC(0, "active_seq_param_set_id");
+        writeByteAlign();
+    }
+};
+
+class SEIBufferingPeriod : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return BUFFERING_PERIOD; }
+
+    SEIBufferingPeriod()
+        : m_cpbDelayOffset(0)
+        , m_dpbDelayOffset(0)
+        , m_auCpbRemovalDelayDelta(1)
+    {
+    }
+
+    bool     m_cpbDelayOffset;
+    bool     m_dpbDelayOffset;
+    uint32_t m_initialCpbRemovalDelay;
+    uint32_t m_initialCpbRemovalDelayOffset;
+    uint32_t m_auCpbRemovalDelayDelta;
+
+    void writeSEI(const SPS& sps)
+    {
+        const HRDInfo& hrd = sps.vuiParameters.hrdParameters;
+
+        WRITE_UVLC(0, "bp_seq_parameter_set_id");
+        WRITE_FLAG(0, "rap_cpb_params_present_flag");
+        WRITE_FLAG(0, "concatenation_flag");
+        WRITE_CODE(m_auCpbRemovalDelayDelta - 1,   hrd.cpbRemovalDelayLength,       "au_cpb_removal_delay_delta_minus1");
+        WRITE_CODE(m_initialCpbRemovalDelay,       hrd.initialCpbRemovalDelayLength,        "initial_cpb_removal_delay");
+        WRITE_CODE(m_initialCpbRemovalDelayOffset, hrd.initialCpbRemovalDelayLength, "initial_cpb_removal_delay_offset");
+
+        writeByteAlign();
+    }
+};
+
+class SEIPictureTiming : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return PICTURE_TIMING; }
+
+    uint32_t  m_picStruct;
+    uint32_t  m_sourceScanType;
+    bool      m_duplicateFlag;
+
+    uint32_t  m_auCpbRemovalDelay;
+    uint32_t  m_picDpbOutputDelay;
+
+    void writeSEI(const SPS& sps)
+    {
+        const VUI *vui = &sps.vuiParameters;
+        const HRDInfo *hrd = &vui->hrdParameters;
+
+        if (vui->frameFieldInfoPresentFlag)
+        {
+            WRITE_CODE(m_picStruct, 4,          "pic_struct");
+            WRITE_CODE(m_sourceScanType, 2,     "source_scan_type");
+            WRITE_FLAG(m_duplicateFlag,         "duplicate_flag");
+        }
+
+        if (vui->hrdParametersPresentFlag)
+        {
+            WRITE_CODE(m_auCpbRemovalDelay - 1, hrd->cpbRemovalDelayLength, "au_cpb_removal_delay_minus1");
+            WRITE_CODE(m_picDpbOutputDelay, hrd->dpbOutputDelayLength, "pic_dpb_output_delay");
+            /* Removed sub-pic signaling June 2014 */
+        }
+        writeByteAlign();
+    }
+};
+
+class SEIRecoveryPoint : public SEI
+{
+public:
+
+    PayloadType payloadType() const { return RECOVERY_POINT; }
+
+    int  m_recoveryPocCnt;
+    bool m_exactMatchingFlag;
+    bool m_brokenLinkFlag;
+
+    void writeSEI(const SPS&)
+    {
+        WRITE_SVLC(m_recoveryPocCnt,    "recovery_poc_cnt");
+        WRITE_FLAG(m_exactMatchingFlag, "exact_matching_flag");
+        WRITE_FLAG(m_brokenLinkFlag,    "broken_link_flag");
+        writeByteAlign();
+    }
+};
+}
+
+#endif // ifndef X265_SEI_H

x265-hg20140324.tar.bz2/source/encoder/slicetype.cpp -> x265-hg20140928.tar.bz2/source/encoder/slicetype.cpp Changed

@@ -19,23 +19,22 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComRom.h"
-#include "TLibCommon/TComPic.h"
+#include "common.h"
+#include "frame.h"
 #include "primitives.h"
 #include "lowres.h"
+#include "TLibCommon/TComRom.h"
+#include "mv.h"
 
 #include "encoder.h"
 #include "slicetype.h"
 #include "motion.h"
-#include "mv.h"
 #include "ratecontrol.h"
 
-#define LOWRES_COST_MASK  ((1 << 14) - 1)
-#define LOWRES_COST_SHIFT 14
-#define NUM_CUS (widthInCU > 2 && heightInCU > 2 ? (widthInCU - 2) * (heightInCU - 2) : widthInCU * heightInCU)
+#define NUM_CUS (m_widthInCU > 2 && m_heightInCU > 2 ? (m_widthInCU - 2) * (m_heightInCU - 2) : m_widthInCU * m_heightInCU)
 
 using namespace x265;
 
@@ -56,140 +55,212 @@
     dst.y = median(a.y, b.y, c.y);
 }
 
-Lookahead::Lookahead(Encoder *_cfg, ThreadPool* pool)
-    : est(pool)
+Lookahead::Lookahead(x265_param *param, ThreadPool* pool, Encoder* enc)
+    : JobProvider(pool)
+    , m_est(pool)
 {
-    param = _cfg->param;
-    lastKeyframe = -param->keyframeMax;
-    lastNonB = NULL;
-    widthInCU = ((param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
-    heightInCU = ((param->sourceHeight / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
-    scratch = (int*)x265_malloc(widthInCU * sizeof(int));
-    memset(histogram, 0, sizeof(histogram));
+    m_bReady = 0;
+    m_param = param;
+    m_top = enc;
+    m_lastKeyframe = -m_param->keyframeMax;
+    m_lastNonB = NULL;
+    m_bFilling = true;
+    m_bFlushed = false;
+    m_widthInCU = ((m_param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    m_heightInCU = ((m_param->sourceHeight / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    m_scratch = (int*)x265_malloc(m_widthInCU * sizeof(int));
+    memset(m_histogram, 0, sizeof(m_histogram));
 }
 
 Lookahead::~Lookahead() { }
 
-void Lookahead::init() { }
+void Lookahead::init()
+{
+    if (m_pool && m_pool->getThreadCount() >= 4 &&
+        ((m_param->bFrameAdaptive && m_param->bframes) ||
+         m_param->rc.cuTree || m_param->scenecutThreshold ||
+         (m_param->lookaheadDepth && m_param->rc.vbvBufferSize)))
+        m_pool = m_pool; /* allow use of worker thread */
+    else
+        m_pool = NULL; /* disable use of worker thread */
+}
 
 void Lookahead::destroy()
 {
+    if (m_pool)
+        // flush will dequeue, if it is necessary
+        JobProvider::flush();
+
     // these two queues will be empty unless the encode was aborted
-    while (!inputQueue.empty())
+    while (!m_inputQueue.empty())
     {
-        TComPic* pic = inputQueue.popFront();
+        Frame* pic = m_inputQueue.popFront();
         pic->destroy();
         delete pic;
     }
 
-    while (!outputQueue.empty())
+    while (!m_outputQueue.empty())
     {
-        TComPic* pic = outputQueue.popFront();
+        Frame* pic = m_outputQueue.popFront();
         pic->destroy();
         delete pic;
     }
 
-    x265_free(scratch);
+    x265_free(m_scratch);
 }
 
-void Lookahead::addPicture(TComPic *pic, int sliceType)
+/* Called by API thread */
+void Lookahead::addPicture(Frame *pic, int sliceType)
 {
     TComPicYuv *orig = pic->getPicYuvOrg();
 
-    pic->m_lowres.init(orig, pic->getSlice()->getPOC(), sliceType);
-    inputQueue.pushBack(*pic);
+    pic->m_lowres.init(orig, pic->getPOC(), sliceType);
 
-    if (inputQueue.size() >= param->lookaheadDepth)
-        slicetypeDecide();
+    m_inputQueueLock.acquire();
+    m_inputQueue.pushBack(*pic);
+
+    if (m_inputQueue.size() >= m_param->lookaheadDepth)
+    {
+        /* when queue fills the first time, run slicetypeDecide synchronously,
+         * since the encoder will always be blocked here */
+        if (m_pool && !m_bFilling)
+        {
+            m_inputQueueLock.release();
+            m_bReady = 1;
+            m_pool->pokeIdleThread();
+        }
+        else
+            slicetypeDecide();
+
+        if (m_bFilling && m_pool)
+            JobProvider::enqueue();
+        m_bFilling = false;
+    }
+    else
+        m_inputQueueLock.release();
 }
 
+/* Called by API thread */
 void Lookahead::flush()
 {
-    if (!inputQueue.empty())
+    /* just in case the input queue is never allowed to fill */
+    m_bFilling = false;
+
+    /* flush synchronously */
+    m_inputQueueLock.acquire();
+    if (!m_inputQueue.empty())
+    {
         slicetypeDecide();
+    }
+    else
+        m_inputQueueLock.release();
+
+    m_inputQueueLock.acquire();
+
+    /* bFlushed indicates that an empty output queue actually means all frames
+     * have been decided (no more inputs for the encoder) */
+    if (m_inputQueue.empty())
+        m_bFlushed = true;
+    m_inputQueueLock.release();
 }
 
-// Called by RateControl to get the estimated SATD cost for a given picture.
-// It assumes dpb->prepareEncode() has already been called for the picture and
-// all the references are established
-int64_t Lookahead::getEstimatedPictureCost(TComPic *pic)
+/* Called by API thread. If the lookahead queue has not yet been filled the
+ * first time, it immediately returns NULL.  Else the function blocks until
+ * outputs are available and then pops the first frame from the output queue. If
+ * flush() has been called and the output queue is empty, NULL is returned. */
+Frame* Lookahead::getDecidedPicture()
+{
+    m_outputQueueLock.acquire();
+
+    if (m_bFilling)
+    {
+        m_outputQueueLock.release();
+        return NULL;
+    }
+
+    while (m_outputQueue.empty() && !m_bFlushed)
+    {
+        m_outputQueueLock.release();
+        m_outputAvailable.wait();
+        m_outputQueueLock.acquire();
+    }
+
+    Frame *fenc = m_outputQueue.popFront();
+    m_outputQueueLock.release();
+    return fenc;
+}
+
+/* Called by pool worker threads */
+bool Lookahead::findJob(int)
+{
+    if (m_bReady && ATOMIC_CAS32(&m_bReady, 1, 0) == 1)
+    {
+        m_inputQueueLock.acquire();
+        slicetypeDecide();
+        return true;
+    }
+    else
+        return false;
+}
+
+/* Called by rate-control to calculate the estimated SATD cost for a given
+ * picture.  It assumes dpb->prepareEncode() has already been called for the
+ * picture and all the references are established */
+void Lookahead::getEstimatedPictureCost(Frame *pic)
 {
     Lowres *frames[X265_LOOKAHEAD_MAX];
 
     // POC distances to each reference
-    int d0, d1, p0, p1, b;
-    int poc = pic->getSlice()->getPOC();
-    int l0poc = pic->getSlice()->getRefPOC(REF_PIC_LIST_0, 0);
-    int l1poc = pic->getSlice()->getRefPOC(REF_PIC_LIST_1, 0);
+    Slice *slice = pic->m_picSym->m_slice;
+    int p0 = 0, p1, b;
+    int poc = slice->m_poc;
+    int l0poc = slice->m_refPOCList[0][0];
+    int l1poc = slice->m_refPOCList[1][0];
 
-    switch (pic->getSlice()->getSliceType())
+    switch (slice->m_sliceType)
     {
     case I_SLICE:
-        frames[0] = &pic->m_lowres;
-        p0 = p1 = b = 0;
+        frames[p0] = &pic->m_lowres;
+        b = p1 = 0;
         break;
 
     case P_SLICE:
-        d0 = poc - l0poc;
-        frames[0] = &pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0)->m_lowres;
-        frames[d0] = &pic->m_lowres;
-        p0 = 0;
-        p1 = d0;
-        b = d0;
+        b = p1 = poc - l0poc;
+        frames[p0] = &slice->m_refPicList[0][0]->m_lowres;
+        frames[b] = &pic->m_lowres;
         break;
 
     case B_SLICE:
-        d0 = poc - l0poc;
-        if (l1poc > poc)
-        {
-            // L1 reference is truly in the future
-            d1 = l1poc - poc;
-            frames[0] = &pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0)->m_lowres;
-            frames[d0] = &pic->m_lowres;
-            frames[d0 + d1] = &pic->getSlice()->getRefPic(REF_PIC_LIST_1, 0)->m_lowres;
-            p0 = 0;
-            p1 = d0 + d1;
-            b = d0;
-        }
-        else
-        {
-            frames[0] = &pic->getSlice()->getRefPic(REF_PIC_LIST_0, 0)->m_lowres;
-            frames[d0] = &pic->m_lowres;
-            p0 = 0;
-            p1 = d0;
-            b = d0;
-        }
+        b = poc - l0poc;
+        p1 = b + l1poc - poc;
+        frames[p0] = &slice->m_refPicList[0][0]->m_lowres;
+        frames[b] = &pic->m_lowres;
+        frames[p1] = &slice->m_refPicList[1][0]->m_lowres;
         break;
 
     default:
-        return 0;
+        return;
     }
 
-    if (pic->m_lowres.costEst[b - p0][p1 - b] < 0)
-    {
-        CostEstimate cost(ThreadPool::getThreadPool());
-        cost.init(param, pic);
-        cost.estimateFrameCost(frames, p0, p1, b, false);
-        cost.flush();
-    }
-    if (param->rc.cuTree)
-    {
+    if (m_param->rc.cuTree && !m_param->rc.bStatRead)
+        /* update row satds based on cutree offsets */
         pic->m_lowres.satdCost = frameCostRecalculate(frames, p0, p1, b);
-        if (b && param->rc.vbvBufferSize)
-            frameCostRecalculate(frames, b, b, b);
-    }
-
-    else if (param->rc.aqMode)
+    else if (m_param->rc.aqMode)
         pic->m_lowres.satdCost = pic->m_lowres.costEstAq[b - p0][p1 - b];
     else
         pic->m_lowres.satdCost = pic->m_lowres.costEst[b - p0][p1 - b];
 
-    if (param->rc.vbvBufferSize > 0 && param->rc.vbvMaxBitrate > 0)
+    if (m_param->rc.vbvBufferSize && m_param->rc.vbvMaxBitrate)
     {
+        /* aggregate lowres row satds to CTU resolution */
         pic->m_lowres.lowresCostForRc = pic->m_lowres.lowresCosts[b - p0][p1 - b];
         uint32_t lowresRow = 0, lowresCol = 0, lowresCuIdx = 0, sum = 0;
-        uint32_t scale = param->maxCUSize / (2 * X265_LOWRES_CU_SIZE);
-        uint32_t widthInLowresCu = (uint32_t)widthInCU, heightInLowresCu = (uint32_t)heightInCU;
+        uint32_t scale = m_param->maxCUSize / (2 * X265_LOWRES_CU_SIZE);
+        uint32_t widthInLowresCu = (uint32_t)m_widthInCU, heightInLowresCu = (uint32_t)m_heightInCU;
+        double *qp_offset = 0;
+        /* Factor in qpoffsets based on Aq/Cutree in CU costs */
+        if (m_param->rc.aqMode)
+            qp_offset = (frames[b]->sliceType == X265_TYPE_B || !m_param->rc.cuTree) ? frames[b]->qpAqOffset : frames[b]->qpCuTreeOffset;
 
         for (uint32_t row = 0; row < pic->getFrameHeightInCU(); row++)
         {
@@ -200,84 +271,117 @@
                 lowresCuIdx = lowresRow * widthInLowresCu;
                 for (lowresCol = 0; lowresCol < widthInLowresCu; lowresCol++, lowresCuIdx++)
                 {
-                    sum +=  pic->m_lowres.lowresCostForRc[lowresCuIdx];
+                    uint16_t lowresCuCost = pic->m_lowres.lowresCostForRc[lowresCuIdx] & LOWRES_COST_MASK;
+                    if (qp_offset)
+                    {
+                        lowresCuCost = (uint16_t)((lowresCuCost * x265_exp2fix8(qp_offset[lowresCuIdx]) + 128) >> 8);
+                        int32_t intraCuCost = pic->m_lowres.intraCost[lowresCuIdx]; 
+                        pic->m_lowres.intraCost[lowresCuIdx] = (intraCuCost * x265_exp2fix8(qp_offset[lowresCuIdx]) + 128) >> 8;
+                    }
+                    pic->m_lowres.lowresCostForRc[lowresCuIdx] = lowresCuCost;
+                    sum += lowresCuCost;
                 }
-
                 pic->m_rowSatdForVbv[row] += sum;
             }
         }
     }
-    return pic->m_lowres.satdCost;
 }
 
+/* called by API thread or worker thread with inputQueueLock acquired */
 void Lookahead::slicetypeDecide()
 {
-    Lowres *frames[X265_LOOKAHEAD_MAX];
-    TComPic *list[X265_LOOKAHEAD_MAX];
-    TComPic *ipic = inputQueue.first();
-    bool isKeyFrameAnalyse = (param->rc.cuTree || (param->rc.vbvBufferSize && param->lookaheadDepth));
+    ScopedLock lock(m_decideLock);
 
-    if (!est.rows && ipic)
-        est.init(param, ipic);
+    Lowres *frames[X265_LOOKAHEAD_MAX];
+    Frame *list[X265_LOOKAHEAD_MAX];
+    int maxSearch = X265_MIN(m_param->lookaheadDepth, X265_LOOKAHEAD_MAX);
 
-    if ((param->bFrameAdaptive && param->bframes) ||
-        param->rc.cuTree || param->scenecutThreshold ||
-        (param->lookaheadDepth && param->rc.vbvBufferSize))
+    memset(frames, 0, sizeof(frames));
+    memset(list, 0, sizeof(list));
+    int numFrames = 0;
     {
-        slicetypeAnalyse(frames, false);
+        Frame *pic = m_inputQueue.first();
+        int j;
+        for (j = 0; j < m_param->bframes + 2; j++)
+        {
+            if (!pic) break;
+            list[j] = pic;
+            pic = pic->m_next;
+        }
+        numFrames = j;
+
+        pic = m_inputQueue.first();
+        frames[0] = m_lastNonB;
+        for (j = 0; j < maxSearch; j++)
+        {
+            if (!pic) break;
+            frames[j + 1] = &pic->m_lowres;
+            pic = pic->m_next;
+        }
+
+        maxSearch = j;
     }
-    else
-        frames[0] = lastNonB;
 
-    int j;
-    for (j = 0; ipic && j < param->bframes + 2; ipic = ipic->m_next)
+    m_inputQueueLock.release();
+
+    if (!m_est.m_rows && list[0])
+        m_est.init(m_param, list[0]);
+
+    if (m_param->rc.bStatRead)
     {
-        list[j++] = ipic;
+        /* Use the frame types from the first pass */
+        for (int i = 0; i < numFrames; i++)
+            list[i]->m_lowres.sliceType = m_top->m_rateControl->rateControlSliceType(list[i]->getPOC());
+    }
+    else if (m_lastNonB &&
+        ((m_param->bFrameAdaptive && m_param->bframes) ||
+         m_param->rc.cuTree || m_param->scenecutThreshold ||
+         (m_param->lookaheadDepth && m_param->rc.vbvBufferSize)))
+    {
+        slicetypeAnalyse(frames, false);
     }
-
-    list[j] = NULL;
 
     int bframes, brefs;
     for (bframes = 0, brefs = 0;; bframes++)
     {
         Lowres& frm = list[bframes]->m_lowres;
 
-        if (frm.sliceType == X265_TYPE_BREF && !param->bBPyramid && brefs == param->bBPyramid)
+        if (frm.sliceType == X265_TYPE_BREF && !m_param->bBPyramid && brefs == m_param->bBPyramid)
         {
             frm.sliceType = X265_TYPE_B;
-            x265_log(param, X265_LOG_WARNING, "B-ref at frame %d incompatible with B-pyramid\n",
+            x265_log(m_param, X265_LOG_WARNING, "B-ref at frame %d incompatible with B-pyramid\n",
                      frm.frameNum);
         }
 
         /* pyramid with multiple B-refs needs a big enough dpb that the preceding P-frame stays available.
            smaller dpb could be supported by smart enough use of mmco, but it's easier just to forbid it.*/
-        else if (frm.sliceType == X265_TYPE_BREF && param->bBPyramid && brefs &&
-                 param->maxNumReferences <= (brefs + 3))
+        else if (frm.sliceType == X265_TYPE_BREF && m_param->bBPyramid && brefs &&
+                 m_param->maxNumReferences <= (brefs + 3))
         {
             frm.sliceType = X265_TYPE_B;
-            x265_log(param, X265_LOG_WARNING, "B-ref at frame %d incompatible with B-pyramid and %d reference frames\n",
-                     frm.sliceType, param->maxNumReferences);
+            x265_log(m_param, X265_LOG_WARNING, "B-ref at frame %d incompatible with B-pyramid and %d reference frames\n",
+                     frm.sliceType, m_param->maxNumReferences);
         }
 
-        if ( /*(!param->intraRefresh || frm.frameNum == 0) && */ frm.frameNum - lastKeyframe >= param->keyframeMax)
+        if ( /*(!param->intraRefresh || frm.frameNum == 0) && */ frm.frameNum - m_lastKeyframe >= m_param->keyframeMax)
         {
             if (frm.sliceType == X265_TYPE_AUTO || frm.sliceType == X265_TYPE_I)
-                frm.sliceType = param->bOpenGOP && lastKeyframe >= 0 ? X265_TYPE_I : X265_TYPE_IDR;
+                frm.sliceType = m_param->bOpenGOP && m_lastKeyframe >= 0 ? X265_TYPE_I : X265_TYPE_IDR;
             bool warn = frm.sliceType != X265_TYPE_IDR;
-            if (warn && param->bOpenGOP)
+            if (warn && m_param->bOpenGOP)
                 warn &= frm.sliceType != X265_TYPE_I;
             if (warn)
             {
-                x265_log(param, X265_LOG_WARNING, "specified frame type (%d) at %d is not compatible with keyframe interval\n",
+                x265_log(m_param, X265_LOG_WARNING, "specified frame type (%d) at %d is not compatible with keyframe interval\n",
                          frm.sliceType, frm.frameNum);
-                frm.sliceType = param->bOpenGOP && lastKeyframe >= 0 ? X265_TYPE_I : X265_TYPE_IDR;
+                frm.sliceType = m_param->bOpenGOP && m_lastKeyframe >= 0 ? X265_TYPE_I : X265_TYPE_IDR;
             }
         }
-        if (frm.sliceType == X265_TYPE_I && frm.frameNum - lastKeyframe >= param->keyframeMin)
+        if (frm.sliceType == X265_TYPE_I && frm.frameNum - m_lastKeyframe >= m_param->keyframeMin)
         {
-            if (param->bOpenGOP)
+            if (m_param->bOpenGOP)
             {
-                lastKeyframe = frm.frameNum;
+                m_lastKeyframe = frm.frameNum;
                 frm.bKeyframe = true;
             }
             else
@@ -286,18 +390,18 @@
         if (frm.sliceType == X265_TYPE_IDR)
         {
             /* Closed GOP */
-            lastKeyframe = frm.frameNum;
+            m_lastKeyframe = frm.frameNum;
             frm.bKeyframe = true;
             if (bframes > 0)
             {
-                frames[bframes]->sliceType = X265_TYPE_P;
+                list[bframes - 1]->m_lowres.sliceType = X265_TYPE_P;
                 bframes--;
             }
         }
-        if (bframes == param->bframes || !list[bframes + 1])
+        if (bframes == m_param->bframes || !list[bframes + 1])
         {
             if (IS_X265_TYPE_B(frm.sliceType))
-                x265_log(param, X265_LOG_WARNING, "specified frame type is not compatible with max B-frames\n");
+                x265_log(m_param, X265_LOG_WARNING, "specified frame type is not compatible with max B-frames\n");
             if (frm.sliceType == X265_TYPE_AUTO || IS_X265_TYPE_B(frm.sliceType))
                 frm.sliceType = X265_TYPE_P;
         }
@@ -312,83 +416,82 @@
     if (bframes)
         list[bframes - 1]->m_lowres.bLastMiniGopBFrame = true;
     list[bframes]->m_lowres.leadingBframes = bframes;
-    lastNonB = &list[bframes]->m_lowres;
-    histogram[bframes]++;
+    m_lastNonB = &list[bframes]->m_lowres;
+    m_histogram[bframes]++;
 
     /* insert a bref into the sequence */
-    if (param->bBPyramid && bframes > 1 && !brefs)
+    if (m_param->bBPyramid && bframes > 1 && !brefs)
     {
         list[bframes / 2]->m_lowres.sliceType = X265_TYPE_BREF;
         brefs++;
     }
 
-    /* calculate the frame costs ahead of time for x264_rc_analyse_slice while we still have lowres */
-    if (param->rc.rateControlMode != X265_RC_CQP)
+    /* calculate the frame costs ahead of time for estimateFrameCost while we still have lowres */
+    if (m_param->rc.rateControlMode != X265_RC_CQP)
     {
         int p0, p1, b;
-        p1 = b = bframes + 1;
-
-        for (int i = 0; i <= bframes; i++)
+        /* For zero latency tuning, calculate frame cost to be used later in RC */
+        if (!maxSearch)
         {
-            frames[i + 1] = &list[i]->m_lowres;
+            for (int i = 0; i <= bframes; i++)
+               frames[i + 1] = &list[i]->m_lowres;
         }
 
-        if (IS_X265_TYPE_I(frames[bframes + 1]->sliceType))
-            p0 = bframes + 1;
-        else // P
-            p0 = 0;
-
-        est.estimateFrameCost(frames, p0, p1, b, 0);
+        /* estimate new non-B cost */
+        p1 = b = bframes + 1;
+        p0 = (IS_X265_TYPE_I(frames[bframes + 1]->sliceType)) ? b : 0;
+        m_est.estimateFrameCost(frames, p0, p1, b, 0);
 
-        if ((p0 != p1 || bframes) && param->rc.vbvBufferSize)
+        if (bframes)
         {
-            // We need the intra costs for row SATDs
-            est.estimateFrameCost(frames, b, b, b, 0);
-
-            // We need B-frame costs for row SATDs
-            p0 = 0;
+            p0 = 0; // last nonb
             for (b = 1; b <= bframes; b++)
             {
                 if (frames[b]->sliceType == X265_TYPE_B)
-                    for (p1 = b; frames[p1]->sliceType == X265_TYPE_B; )
-                    {
-                        p1++;
-                    }
-
+                    for (p1 = b; frames[p1]->sliceType == X265_TYPE_B; p1++)
+                        ; // find new nonb or bref
                 else
                     p1 = bframes + 1;
-                est.estimateFrameCost(frames, p0, p1, b, 0);
+
+                m_est.estimateFrameCost(frames, p0, p1, b, 0);
+
                 if (frames[b]->sliceType == X265_TYPE_BREF)
                     p0 = b;
             }
         }
     }
 
+    m_inputQueueLock.acquire();
+
     /* dequeue all frames from inputQueue that are about to be enqueued
      * in the output queue. The order is important because TComPic can
      * only be in one list at a time */
     int64_t pts[X265_BFRAME_MAX + 1];
     for (int i = 0; i <= bframes; i++)
     {
-        TComPic *pic;
-        pic = inputQueue.popFront();
+        Frame *pic;
+        pic = m_inputQueue.popFront();
         pts[i] = pic->m_pts;
+        maxSearch--;
     }
 
+    m_inputQueueLock.release();
+
+    m_outputQueueLock.acquire();
     /* add non-B to output queue */
     int idx = 0;
     list[bframes]->m_reorderedPts = pts[idx++];
-    outputQueue.pushBack(*list[bframes]);
+    m_outputQueue.pushBack(*list[bframes]);
 
     /* Add B-ref frame next to P frame in output queue, the B-ref encode before non B-ref frame */
-    if (bframes > 1 && param->bBPyramid)
+    if (bframes > 1 && m_param->bBPyramid)
     {
         for (int i = 0; i < bframes; i++)
         {
             if (list[i]->m_lowres.sliceType == X265_TYPE_BREF)
             {
                 list[i]->m_reorderedPts = pts[idx++];
-                outputQueue.pushBack(*list[i]);
+                m_outputQueue.pushBack(*list[i]);
             }
         }
     }
@@ -400,26 +503,42 @@
         if (list[i]->m_lowres.sliceType != X265_TYPE_BREF)
         {
             list[i]->m_reorderedPts = pts[idx++];
-            outputQueue.pushBack(*list[i]);
+            m_outputQueue.pushBack(*list[i]);
         }
     }
 
-    if (isKeyFrameAnalyse && IS_X265_TYPE_I(lastNonB->sliceType))
+    bool isKeyFrameAnalyse = (m_param->rc.cuTree || (m_param->rc.vbvBufferSize && m_param->lookaheadDepth)) && !m_param->rc.bStatRead;
+    if (isKeyFrameAnalyse && IS_X265_TYPE_I(m_lastNonB->sliceType))
     {
+        m_inputQueueLock.acquire();
+        Frame *pic = m_inputQueue.first();
+        frames[0] = m_lastNonB;
+        int j;
+        for (j = 0; j < maxSearch; j++)
+        {
+            frames[j + 1] = &pic->m_lowres;
+            pic = pic->m_next;
+        }
+
+        frames[j + 1] = NULL;
+        m_inputQueueLock.release();
         slicetypeAnalyse(frames, true);
     }
+
+    m_outputQueueLock.release();
+    m_outputAvailable.trigger();
 }
 
 void Lookahead::vbvLookahead(Lowres **frames, int numFrames, int keyframe)
 {
     int prevNonB = 0, curNonB = 1, idx = 0;
+    bool isNextNonB = false;
 
     while (curNonB < numFrames && frames[curNonB]->sliceType == X265_TYPE_B)
-    {
         curNonB++;
-    }
 
     int nextNonB = keyframe ? prevNonB : curNonB;
+    int nextB = keyframe ? prevNonB + 1 : curNonB + 1;
 
     while (curNonB < numFrames + !keyframe)
     {
@@ -438,12 +557,33 @@
             frames[nextNonB]->plannedType[idx] = X265_TYPE_B;
         }
 
+        for (int i = nextB; i <= curNonB; i++)
+        {
+            for (int j = frames[i]->indB + i + 1; j <= curNonB; j++, frames[i]->indB++)
+            {
+                if (j == curNonB)
+                {
+                    if (isNextNonB)
+                    {
+                        int p0 = IS_X265_TYPE_I(frames[curNonB]->sliceType) ? curNonB : prevNonB;
+                        frames[i]->plannedSatd[frames[i]->indB] = vbvFrameCost(frames, p0, curNonB, curNonB);
+                        frames[i]->plannedType[frames[i]->indB] = frames[curNonB]->sliceType;
+                    }
+                }
+                else
+                {
+                    frames[i]->plannedSatd[frames[i]->indB] = vbvFrameCost(frames, prevNonB, curNonB, j);
+                    frames[i]->plannedType[frames[i]->indB] = X265_TYPE_B;
+                }
+            }
+            if (i == curNonB && !isNextNonB)
+                isNextNonB = true;
+        }
+
         prevNonB = curNonB;
         curNonB++;
         while (curNonB <= numFrames && frames[curNonB]->sliceType == X265_TYPE_B)
-        {
             curNonB++;
-        }
     }
 
     frames[nextNonB]->plannedType[idx] = X265_TYPE_AUTO;
@@ -451,11 +591,11 @@
 
 int64_t Lookahead::vbvFrameCost(Lowres **frames, int p0, int p1, int b)
 {
-    int64_t cost = est.estimateFrameCost(frames, p0, p1, b, 0);
+    int64_t cost = m_est.estimateFrameCost(frames, p0, p1, b, 0);
 
-    if (param->rc.aqMode)
+    if (m_param->rc.aqMode)
     {
-        if (param->rc.cuTree)
+        if (m_param->rc.cuTree)
             return frameCostRecalculate(frames, p0, p1, b);
         else
             return frames[b]->costEstAq[b - p0][p1 - b];
@@ -466,37 +606,34 @@
 void Lookahead::slicetypeAnalyse(Lowres **frames, bool bKeyframe)
 {
     int numFrames, origNumFrames, keyintLimit, framecnt;
-    int maxSearch = X265_MIN(param->lookaheadDepth, X265_LOOKAHEAD_MAX);
+    int maxSearch = X265_MIN(m_param->lookaheadDepth, X265_LOOKAHEAD_MAX);
     int cuCount = NUM_CUS;
     int resetStart;
-    bool bIsVbvLookahead = param->rc.vbvBufferSize && param->lookaheadDepth;
-
-    if (!lastNonB)
-        return;
+    bool bIsVbvLookahead = m_param->rc.vbvBufferSize && m_param->lookaheadDepth;
 
-    frames[0] = lastNonB;
-    TComPic* pic = inputQueue.first();
-    for (framecnt = 0; (framecnt < maxSearch) && pic && pic->m_lowres.sliceType == X265_TYPE_AUTO; framecnt++)
+    /* count undecided frames */
+    for (framecnt = 0; framecnt < maxSearch; framecnt++)
     {
-        frames[framecnt + 1] = &pic->m_lowres;
-        pic = pic->m_next;
+        Lowres *fenc = frames[framecnt + 1];
+        if (!fenc || fenc->sliceType != X265_TYPE_AUTO)
+            break;
     }
 
     if (!framecnt)
     {
-        if (param->rc.cuTree)
+        if (m_param->rc.cuTree)
             cuTree(frames, 0, bKeyframe);
         return;
     }
 
     frames[framecnt + 1] = NULL;
 
-    keyintLimit = param->keyframeMax - frames[0]->frameNum + lastKeyframe - 1;
+    keyintLimit = m_param->keyframeMax - frames[0]->frameNum + m_lastKeyframe - 1;
     origNumFrames = numFrames = X265_MIN(framecnt, keyintLimit);
 
     if (bIsVbvLookahead)
         numFrames = framecnt;
-    else if (param->bOpenGOP && numFrames < framecnt)
+    else if (m_param->bOpenGOP && numFrames < framecnt)
         numFrames++;
     else if (numFrames == 0)
     {
@@ -506,15 +643,15 @@
 
     int numBFrames = 0;
     int numAnalyzed = numFrames;
-    if (param->scenecutThreshold && scenecut(frames, 0, 1, true, origNumFrames, maxSearch))
+    if (m_param->scenecutThreshold && scenecut(frames, 0, 1, true, origNumFrames, maxSearch))
     {
         frames[1]->sliceType = X265_TYPE_I;
         return;
     }
 
-    if (param->bframes)
+    if (m_param->bframes)
     {
-        if (param->bFrameAdaptive == X265_B_ADAPT_TRELLIS)
+        if (m_param->bFrameAdaptive == X265_B_ADAPT_TRELLIS)
         {
             if (numFrames > 1)
             {
@@ -537,13 +674,13 @@
             }
             frames[numFrames]->sliceType = X265_TYPE_P;
         }
-        else if (param->bFrameAdaptive == X265_B_ADAPT_FAST)
+        else if (m_param->bFrameAdaptive == X265_B_ADAPT_FAST)
         {
             int64_t cost1p0, cost2p0, cost1b1, cost2p1;
 
             for (int i = 0; i <= numFrames - 2; )
             {
-                cost2p1 = est.estimateFrameCost(frames, i + 0, i + 2, i + 2, 1);
+                cost2p1 = m_est.estimateFrameCost(frames, i + 0, i + 2, i + 2, 1);
                 if (frames[i + 2]->intraMbs[2] > cuCount / 2)
                 {
                     frames[i + 1]->sliceType = X265_TYPE_P;
@@ -552,9 +689,9 @@
                     continue;
                 }
 
-                cost1b1 = est.estimateFrameCost(frames, i + 0, i + 2, i + 1, 0);
-                cost1p0 = est.estimateFrameCost(frames, i + 0, i + 1, i + 1, 0);
-                cost2p0 = est.estimateFrameCost(frames, i + 1, i + 2, i + 2, 0);
+                cost1b1 = m_est.estimateFrameCost(frames, i + 0, i + 2, i + 1, 0);
+                cost1p0 = m_est.estimateFrameCost(frames, i + 0, i + 1, i + 1, 0);
+                cost2p0 = m_est.estimateFrameCost(frames, i + 1, i + 2, i + 2, 0);
 
                 if (cost1p0 + cost2p0 < cost1b1 + cost2p1)
                 {
@@ -565,14 +702,14 @@
 
 // arbitrary and untuned
 #define INTER_THRESH 300
-#define P_SENS_BIAS (50 - param->bFrameBias)
+#define P_SENS_BIAS (50 - m_param->bFrameBias)
                 frames[i + 1]->sliceType = X265_TYPE_B;
 
                 int j;
-                for (j = i + 2; j <= X265_MIN(i + param->bframes, numFrames - 1); j++)
+                for (j = i + 2; j <= X265_MIN(i + m_param->bframes, numFrames - 1); j++)
                 {
                     int64_t pthresh = X265_MAX(INTER_THRESH - P_SENS_BIAS * (j - i - 1), INTER_THRESH / 10);
-                    int64_t pcost = est.estimateFrameCost(frames, i + 0, j + 1, j + 1, 1);
+                    int64_t pcost = m_est.estimateFrameCost(frames, i + 0, j + 1, j + 1, 1);
                     if (pcost > pthresh * cuCount || frames[j + 1]->intraMbs[j - i + 1] > cuCount / 3)
                         break;
                     frames[j]->sliceType = X265_TYPE_B;
@@ -590,7 +727,7 @@
         }
         else
         {
-            numBFrames = X265_MIN(numFrames - 1, param->bframes);
+            numBFrames = X265_MIN(numFrames - 1, m_param->bframes);
             for (int j = 1; j < numFrames; j++)
             {
                 frames[j]->sliceType = (j % (numBFrames + 1)) ? X265_TYPE_B : X265_TYPE_P;
@@ -601,7 +738,7 @@
         /* Check scenecut on the first minigop. */
         for (int j = 1; j < numBFrames + 1; j++)
         {
-            if (param->scenecutThreshold && scenecut(frames, j, j + 1, false, origNumFrames, maxSearch))
+            if (m_param->scenecutThreshold && scenecut(frames, j, j + 1, false, origNumFrames, maxSearch))
             {
                 frames[j]->sliceType = X265_TYPE_P;
                 numAnalyzed = j;
@@ -621,11 +758,11 @@
         resetStart = bKeyframe ? 1 : 2;
     }
 
-    if (param->rc.cuTree)
-        cuTree(frames, X265_MIN(numFrames, param->keyframeMax), bKeyframe);
+    if (m_param->rc.cuTree)
+        cuTree(frames, X265_MIN(numFrames, m_param->keyframeMax), bKeyframe);
 
     // if (!param->bIntraRefresh)
-    for (int j = keyintLimit + 1; j <= numFrames; j += param->keyframeMax)
+    for (int j = keyintLimit + 1; j <= numFrames; j += m_param->keyframeMax)
     {
         frames[j]->sliceType = X265_TYPE_I;
         resetStart = X265_MIN(resetStart, j + 1);
@@ -644,13 +781,13 @@
 bool Lookahead::scenecut(Lowres **frames, int p0, int p1, bool bRealScenecut, int numFrames, int maxSearch)
 {
     /* Only do analysis during a normal scenecut check. */
-    if (bRealScenecut && param->bframes)
+    if (bRealScenecut && m_param->bframes)
     {
         int origmaxp1 = p0 + 1;
         /* Look ahead to avoid coding short flashes as scenecuts. */
-        if (param->bFrameAdaptive == X265_B_ADAPT_TRELLIS)
+        if (m_param->bFrameAdaptive == X265_B_ADAPT_TRELLIS)
             /* Don't analyse any more frames than the trellis would have covered. */
-            origmaxp1 += param->bframes;
+            origmaxp1 += m_param->bframes;
         else
             origmaxp1++;
         int maxp1 = X265_MIN(origmaxp1, numFrames);
@@ -691,29 +828,29 @@
 {
     Lowres *frame = frames[p1];
 
-    est.estimateFrameCost(frames, p0, p1, p1, 0);
+    m_est.estimateFrameCost(frames, p0, p1, p1, 0);
 
     int64_t icost = frame->costEst[0][0];
     int64_t pcost = frame->costEst[p1 - p0][0];
-    int gopSize = frame->frameNum - lastKeyframe;
-    float threshMax = (float)(param->scenecutThreshold / 100.0);
+    int gopSize = frame->frameNum - m_lastKeyframe;
+    float threshMax = (float)(m_param->scenecutThreshold / 100.0);
 
     /* magic numbers pulled out of thin air */
     float threshMin = (float)(threshMax * 0.25);
     float bias;
 
-    if (param->keyframeMin == param->keyframeMax)
+    if (m_param->keyframeMin == m_param->keyframeMax)
         threshMin = threshMax;
-    if (gopSize <= param->keyframeMin / 4)
+    if (gopSize <= m_param->keyframeMin / 4)
         bias = threshMin / 4;
-    else if (gopSize <= param->keyframeMin)
-        bias = threshMin * gopSize / param->keyframeMin;
+    else if (gopSize <= m_param->keyframeMin)
+        bias = threshMin * gopSize / m_param->keyframeMin;
     else
     {
         bias = threshMin
             + (threshMax - threshMin)
-            * (gopSize - param->keyframeMin)
-            / (param->keyframeMax - param->keyframeMin);
+            * (gopSize - m_param->keyframeMin)
+            / (m_param->keyframeMax - m_param->keyframeMin);
     }
 
     bool res = pcost >= (1.0 - bias) * icost;
@@ -721,7 +858,7 @@
     {
         int imb = frame->intraMbs[p1 - p0];
         int pmb = NUM_CUS - imb;
-        x265_log(param, X265_LOG_DEBUG, "scene cut at %d Icost:%d Pcost:%d ratio:%.4f bias:%.4f gop:%d (imb:%d pmb:%d)\n",
+        x265_log(m_param, X265_LOG_DEBUG, "scene cut at %d Icost:%d Pcost:%d ratio:%.4f bias:%.4f gop:%d (imb:%d pmb:%d)\n",
                  frame->frameNum, icost, pcost, 1. - (double)pcost / icost, bias, gopSize, imb, pmb);
     }
     return res;
@@ -730,7 +867,7 @@
 void Lookahead::slicetypePath(Lowres **frames, int length, char(*best_paths)[X265_LOOKAHEAD_MAX + 1])
 {
     char paths[2][X265_LOOKAHEAD_MAX + 1];
-    int num_paths = X265_MIN(param->bframes + 1, length);
+    int num_paths = X265_MIN(m_param->bframes + 1, length);
     int64_t best_cost = 1LL << 62;
     int idx = 0;
 
@@ -773,30 +910,30 @@
         }
 
         /* Add the cost of the P-frame found above */
-        cost += est.estimateFrameCost(frames, cur_p, next_p, next_p, 0);
+        cost += m_est.estimateFrameCost(frames, cur_p, next_p, next_p, 0);
         /* Early terminate if the cost we have found is larger than the best path cost so far */
         if (cost > threshold)
             break;
 
-        if (param->bBPyramid && next_p - cur_p > 2)
+        if (m_param->bBPyramid && next_p - cur_p > 2)
         {
             int middle = cur_p + (next_p - cur_p) / 2;
-            cost += est.estimateFrameCost(frames, cur_p, next_p, middle, 0);
+            cost += m_est.estimateFrameCost(frames, cur_p, next_p, middle, 0);
             for (int next_b = loc; next_b < middle && cost < threshold; next_b++)
             {
-                cost += est.estimateFrameCost(frames, cur_p, middle, next_b, 0);
+                cost += m_est.estimateFrameCost(frames, cur_p, middle, next_b, 0);
             }
 
             for (int next_b = middle + 1; next_b < next_p && cost < threshold; next_b++)
             {
-                cost += est.estimateFrameCost(frames, middle, next_p, next_b, 0);
+                cost += m_est.estimateFrameCost(frames, middle, next_p, next_b, 0);
             }
         }
         else
         {
             for (int next_b = loc; next_b < next_p && cost < threshold; next_b++)
             {
-                cost += est.estimateFrameCost(frames, cur_p, next_p, next_b, 0);
+                cost += m_est.estimateFrameCost(frames, cur_p, next_p, next_b, 0);
             }
         }
 
@@ -817,16 +954,16 @@
     double totalDuration = 0.0;
     for (int j = 0; j <= numframes; j++)
     {
-        totalDuration += (double)param->fpsDenom / param->fpsNum;
+        totalDuration += (double)m_param->fpsDenom / m_param->fpsNum;
     }
 
     double averageDuration = totalDuration / (numframes + 1);
 
     int i = numframes;
-    int cuCount = widthInCU * heightInCU;
+    int cuCount = m_widthInCU * m_heightInCU;
 
     if (bIntra)
-        est.estimateFrameCost(frames, 0, 0, 0, 0);
+        m_est.estimateFrameCost(frames, 0, 0, 0, 0);
 
     while (i > 0 && frames[i]->sliceType == X265_TYPE_B)
     {
@@ -838,12 +975,12 @@
     /* Lookaheadless MB-tree is not a theoretically distinct case; the same extrapolation could
      * be applied to the end of a lookahead buffer of any size.  However, it's most needed when
      * lookahead=0, so that's what's currently implemented. */
-    if (!param->lookaheadDepth)
+    if (!m_param->lookaheadDepth)
     {
         if (bIntra)
         {
             memset(frames[0]->propagateCost, 0, cuCount * sizeof(uint16_t));
-            memcpy(frames[0]->qpOffset, frames[0]->qpAqOffset, cuCount * sizeof(double));
+            memcpy(frames[0]->qpCuTreeOffset, frames[0]->qpAqOffset, cuCount * sizeof(double));
             return;
         }
         std::swap(frames[lastnonb]->propagateCost, frames[0]->propagateCost);
@@ -867,13 +1004,13 @@
         if (curnonb < idx)
             break;
 
-        est.estimateFrameCost(frames, curnonb, lastnonb, lastnonb, 0);
+        m_est.estimateFrameCost(frames, curnonb, lastnonb, lastnonb, 0);
         memset(frames[curnonb]->propagateCost, 0, cuCount * sizeof(uint16_t));
         bframes = lastnonb - curnonb - 1;
-        if (param->bBPyramid && bframes > 1)
+        if (m_param->bBPyramid && bframes > 1)
         {
             int middle = (bframes + 1) / 2 + curnonb;
-            est.estimateFrameCost(frames, curnonb, lastnonb, middle, 0);
+            m_est.estimateFrameCost(frames, curnonb, lastnonb, middle, 0);
             memset(frames[middle]->propagateCost, 0, cuCount * sizeof(uint16_t));
             while (i > curnonb)
             {
@@ -881,7 +1018,7 @@
                 int p1 = i < middle ? middle : lastnonb;
                 if (i != middle)
                 {
-                    est.estimateFrameCost(frames, p0, p1, i, 0);
+                    m_est.estimateFrameCost(frames, p0, p1, i, 0);
                     estimateCUPropagate(frames, averageDuration, p0, p1, i, 0);
                 }
                 i--;
@@ -893,7 +1030,7 @@
         {
             while (i > curnonb)
             {
-                est.estimateFrameCost(frames, curnonb, lastnonb, i, 0);
+                m_est.estimateFrameCost(frames, curnonb, lastnonb, i, 0);
                 estimateCUPropagate(frames, averageDuration, curnonb, lastnonb, i, 0);
                 i--;
             }
@@ -902,15 +1039,15 @@
         lastnonb = curnonb;
     }
 
-    if (!param->lookaheadDepth)
+    if (!m_param->lookaheadDepth)
     {
-        est.estimateFrameCost(frames, 0, lastnonb, lastnonb, 0);
+        m_est.estimateFrameCost(frames, 0, lastnonb, lastnonb, 0);
         estimateCUPropagate(frames, averageDuration, 0, lastnonb, lastnonb, 1);
         std::swap(frames[lastnonb]->propagateCost, frames[0]->propagateCost);
     }
 
     cuTreeFinish(frames[lastnonb], averageDuration, lastnonb);
-    if (param->bBPyramid && bframes > 1 && !param->rc.vbvBufferSize)
+    if (m_param->bBPyramid && bframes > 1 && !m_param->rc.vbvBufferSize)
         cuTreeFinish(frames[lastnonb + (bframes + 1) / 2], averageDuration, 0);
 }
 
@@ -918,35 +1055,34 @@
 {
     uint16_t *refCosts[2] = { frames[p0]->propagateCost, frames[p1]->propagateCost };
     int32_t distScaleFactor = (((b - p0) << 8) + ((p1 - p0) >> 1)) / (p1 - p0);
-    int32_t bipredWeight = param->bEnableWeightedBiPred ? 64 - (distScaleFactor >> 2) : 32;
+    int32_t bipredWeight = m_param->bEnableWeightedBiPred ? 64 - (distScaleFactor >> 2) : 32;
     MV *mvs[2] = { frames[b]->lowresMvs[0][b - p0 - 1], frames[b]->lowresMvs[1][p1 - b - 1] };
     int32_t bipredWeights[2] = { bipredWeight, 64 - bipredWeight };
 
-    memset(scratch, 0, widthInCU * sizeof(int));
+    memset(m_scratch, 0, m_widthInCU * sizeof(int));
 
     uint16_t *propagateCost = frames[b]->propagateCost;
 
     x265_emms();
-    double fpsFactor = CLIP_DURATION((double)param->fpsDenom / param->fpsNum) / CLIP_DURATION(averageDuration);
+    double fpsFactor = CLIP_DURATION((double)m_param->fpsDenom / m_param->fpsNum) / CLIP_DURATION(averageDuration);
 
     /* For non-refferd frames the source costs are always zero, so just memset one row and re-use it. */
     if (!referenced)
-        memset(frames[b]->propagateCost, 0, widthInCU * sizeof(uint16_t));
+        memset(frames[b]->propagateCost, 0, m_widthInCU * sizeof(uint16_t));
 
-    int32_t StrideInCU = widthInCU;
-    for (uint16_t blocky = 0; blocky < heightInCU; blocky++)
+    int32_t StrideInCU = m_widthInCU;
+    for (uint16_t blocky = 0; blocky < m_heightInCU; blocky++)
     {
         int cuIndex = blocky * StrideInCU;
-        /* TODO This function go into ASM */
-        estimateCUPropagateCost(scratch, propagateCost,
-                                frames[b]->intraCost + cuIndex, frames[b]->lowresCosts[b - p0][p1 - b] + cuIndex,
-                                frames[b]->invQscaleFactor + cuIndex, &fpsFactor, widthInCU);
+        primitives.propagateCost(m_scratch, propagateCost,
+                                 frames[b]->intraCost + cuIndex, frames[b]->lowresCosts[b - p0][p1 - b] + cuIndex,
+                                 frames[b]->invQscaleFactor + cuIndex, &fpsFactor, m_widthInCU);
 
         if (referenced)
-            propagateCost += widthInCU;
-        for (uint16_t blockx = 0; blockx < widthInCU; blockx++, cuIndex++)
+            propagateCost += m_widthInCU;
+        for (uint16_t blockx = 0; blockx < m_widthInCU; blockx++, cuIndex++)
         {
-            int32_t propagate_amount = scratch[blockx];
+            int32_t propagate_amount = m_scratch[blockx];
             /* Don't propagate for an intra block. */
             if (propagate_amount > 0)
             {
@@ -987,7 +1123,7 @@
 
                         /* We could just clip the MVs, but pixels that lie outside the frame probably shouldn't
                          * be counted. */
-                        if (cux < widthInCU - 1 && cuy < heightInCU - 1 && cux >= 0 && cuy >= 0)
+                        if (cux < m_widthInCU - 1 && cuy < m_heightInCU - 1 && cux >= 0 && cuy >= 0)
                         {
                             CLIP_ADD(refCosts[list][idx0], (listamount * idx0weight + 512) >> 10);
                             CLIP_ADD(refCosts[list][idx1], (listamount * idx1weight + 512) >> 10);
@@ -996,13 +1132,13 @@
                         }
                         else /* Check offsets individually */
                         {
-                            if (cux < widthInCU && cuy < heightInCU && cux >= 0 && cuy >= 0)
+                            if (cux < m_widthInCU && cuy < m_heightInCU && cux >= 0 && cuy >= 0)
                                 CLIP_ADD(refCosts[list][idx0], (listamount * idx0weight + 512) >> 10);
-                            if (cux + 1 < widthInCU && cuy < heightInCU && cux + 1 >= 0 && cuy >= 0)
+                            if (cux + 1 < m_widthInCU && cuy < m_heightInCU && cux + 1 >= 0 && cuy >= 0)
                                 CLIP_ADD(refCosts[list][idx1], (listamount * idx1weight + 512) >> 10);
-                            if (cux < widthInCU && cuy + 1 < heightInCU && cux >= 0 && cuy + 1 >= 0)
+                            if (cux < m_widthInCU && cuy + 1 < m_heightInCU && cux >= 0 && cuy + 1 >= 0)
                                 CLIP_ADD(refCosts[list][idx2], (listamount * idx2weight + 512) >> 10);
-                            if (cux + 1 < widthInCU && cuy + 1 < heightInCU && cux + 1 >= 0 && cuy + 1 >= 0)
+                            if (cux + 1 < m_widthInCU && cuy + 1 < m_heightInCU && cux + 1 >= 0 && cuy + 1 >= 0)
                                 CLIP_ADD(refCosts[list][idx3], (listamount * idx3weight + 512) >> 10);
                         }
                     }
@@ -1011,13 +1147,13 @@
         }
     }
 
-    if (param->rc.vbvBufferSize && param->logLevel && referenced)
+    if (m_param->rc.vbvBufferSize && m_param->lookaheadDepth && referenced)
         cuTreeFinish(frames[b], averageDuration, b == p1 ? b - p0 : 0);
 }
 
 void Lookahead::cuTreeFinish(Lowres *frame, double averageDuration, int ref0Distance)
 {
-    int fpsFactor = (int)(CLIP_DURATION(averageDuration) / CLIP_DURATION((double)param->fpsDenom / param->fpsNum) * 256);
+    int fpsFactor = (int)(CLIP_DURATION(averageDuration) / CLIP_DURATION((double)m_param->fpsDenom / m_param->fpsNum) * 256);
     double weightdelta = 0.0;
 
     if (ref0Distance && frame->weightedCostDelta[ref0Distance - 1] > 0)
@@ -1026,8 +1162,8 @@
     /* Allow the strength to be adjusted via qcompress, since the two
      * concepts are very similar. */
 
-    int cuCount = widthInCU * heightInCU;
-    double strength = 5.0 * (1.0 - param->rc.qCompress);
+    int cuCount = m_widthInCU * m_heightInCU;
+    double strength = 5.0 * (1.0 - m_param->rc.qCompress);
 
     for (int cuIndex = 0; cuIndex < cuCount; cuIndex++)
     {
@@ -1036,50 +1172,33 @@
         {
             int propagateCost = (frame->propagateCost[cuIndex] * fpsFactor + 128) >> 8;
             double log2_ratio = X265_LOG2(intracost + propagateCost) - X265_LOG2(intracost) + weightdelta;
-            frame->qpOffset[cuIndex] = frame->qpAqOffset[cuIndex] - strength * log2_ratio;
+            frame->qpCuTreeOffset[cuIndex] = frame->qpAqOffset[cuIndex] - strength * log2_ratio;
         }
     }
 }
 
-/* Estimate the total amount of influence on future quality that could be had if we
- * were to improve the reference samples used to inter predict any given macroblock. */
-void Lookahead::estimateCUPropagateCost(int *dst, uint16_t *propagateIn, int32_t *intraCosts, uint16_t *interCosts,
-                                        int32_t *invQscales, double *fpsFactor, int len)
-{
-    double fps = *fpsFactor / 256;
-
-    for (int i = 0; i < len; i++)
-    {
-        double intraCost       = intraCosts[i] * invQscales[i];
-        double propagateAmount = (double)propagateIn[i] + intraCost * fps;
-        double propagateNum    = (double)intraCosts[i] - (interCosts[i] & LOWRES_COST_MASK);
-        double propagateDenom  = (double)intraCosts[i];
-        dst[i] = (int)(propagateAmount * propagateNum / propagateDenom + 0.5);
-    }
-}
-
 /* If MB-tree changes the quantizers, we need to recalculate the frame cost without
  * re-running lookahead. */
 int64_t Lookahead::frameCostRecalculate(Lowres** frames, int p0, int p1, int b)
 {
     int64_t score = 0;
     int *rowSatd = frames[b]->rowSatds[b - p0][p1 - b];
-    double *qp_offset = IS_X265_TYPE_B(frames[b]->sliceType) ? frames[b]->qpAqOffset : frames[b]->qpOffset;
+    double *qp_offset = (frames[b]->sliceType == X265_TYPE_B) ? frames[b]->qpAqOffset : frames[b]->qpCuTreeOffset;
 
     x265_emms();
-    for (int cuy = heightInCU - 1; cuy >= 0; cuy--)
+    for (int cuy = m_heightInCU - 1; cuy >= 0; cuy--)
     {
         rowSatd[cuy] = 0;
-        for (int cux = widthInCU - 1; cux >= 0; cux--)
+        for (int cux = m_widthInCU - 1; cux >= 0; cux--)
         {
-            int cuxy = cux + cuy * widthInCU;
+            int cuxy = cux + cuy * m_widthInCU;
             int cuCost = frames[b]->lowresCosts[b - p0][p1 - b][cuxy] & LOWRES_COST_MASK;
             double qp_adj = qp_offset[cuxy];
             cuCost = (cuCost * x265_exp2fix8(qp_adj) + 128) >> 8;
             rowSatd[cuy] += cuCost;
-            if ((cuy > 0 && cuy < heightInCU - 1 &&
-                 cux > 0 && cux < widthInCU - 1) ||
-                widthInCU <= 2 || heightInCU <= 2)
+            if ((cuy > 0 && cuy < m_heightInCU - 1 &&
+                 cux > 0 && cux < m_widthInCU - 1) ||
+                m_widthInCU <= 2 || m_heightInCU <= 2)
             {
                 score += cuCost;
             }
@@ -1092,40 +1211,41 @@
 CostEstimate::CostEstimate(ThreadPool *p)
     : WaveFront(p)
 {
-    param = NULL;
-    curframes = NULL;
-    wbuffer[0] = wbuffer[1] = wbuffer[2] = wbuffer[3] = 0;
-    rows = NULL;
-    paddedLines = widthInCU = heightInCU = 0;
-    bDoSearch[0] = bDoSearch[1] = false;
-    curb = curp0 = curp1 = 0;
-    rowsCompleted = 0;
+    m_param = NULL;
+    m_curframes = NULL;
+    m_wbuffer[0] = m_wbuffer[1] = m_wbuffer[2] = m_wbuffer[3] = 0;
+    m_rows = NULL;
+    m_paddedLines = m_widthInCU = m_heightInCU = 0;
+    m_bDoSearch[0] = m_bDoSearch[1] = false;
+    m_curb = m_curp0 = m_curp1 = 0;
+    m_bFrameCompleted = false;
 }
 
 CostEstimate::~CostEstimate()
 {
     for (int i = 0; i < 4; i++)
     {
-        x265_free(wbuffer[i]);
+        x265_free(m_wbuffer[i]);
     }
 
-    delete[] rows;
+    delete[] m_rows;
 }
 
-void CostEstimate::init(x265_param *_param, TComPic *pic)
+void CostEstimate::init(x265_param *_param, Frame *pic)
 {
-    param = _param;
-    widthInCU = ((param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
-    heightInCU = ((param->sourceHeight / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    m_param = _param;
+    m_widthInCU = ((m_param->sourceWidth / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
+    m_heightInCU = ((m_param->sourceHeight / 2) + X265_LOWRES_CU_SIZE - 1) >> X265_LOWRES_CU_BITS;
 
-    rows = new EstimateRow[heightInCU];
-    for (int i = 0; i < heightInCU; i++)
+    m_rows = new EstimateRow[m_heightInCU];
+    for (int i = 0; i < m_heightInCU; i++)
     {
-        rows[i].widthInCU = widthInCU;
-        rows[i].heightInCU = heightInCU;
+        m_rows[i].m_widthInCU = m_widthInCU;
+        m_rows[i].m_heightInCU = m_heightInCU;
+        m_rows[i].m_param = m_param;
     }
 
-    if (!WaveFront::init(heightInCU))
+    if (!WaveFront::init(m_heightInCU))
     {
         m_pool = NULL;
     }
@@ -1134,23 +1254,23 @@
         WaveFront::enableAllRows();
     }
 
-    if (param->bEnableWeightedPred)
+    if (m_param->bEnableWeightedPred)
     {
         TComPicYuv *orig = pic->getPicYuvOrg();
-        paddedLines = pic->m_lowres.lines + 2 * orig->getLumaMarginY();
+        m_paddedLines = pic->m_lowres.lines + 2 * orig->getLumaMarginY();
         int padoffset = pic->m_lowres.lumaStride * orig->getLumaMarginY() + orig->getLumaMarginX();
 
         /* allocate weighted lowres buffers */
         for (int i = 0; i < 4; i++)
         {
-            wbuffer[i] = (pixel*)x265_malloc(sizeof(pixel) * (pic->m_lowres.lumaStride * paddedLines));
-            weightedRef.lowresPlane[i] = wbuffer[i] + padoffset;
+            m_wbuffer[i] = (pixel*)x265_malloc(sizeof(pixel) * (pic->m_lowres.lumaStride * m_paddedLines));
+            m_weightedRef.lowresPlane[i] = m_wbuffer[i] + padoffset;
         }
 
-        weightedRef.fpelPlane = weightedRef.lowresPlane[0];
-        weightedRef.lumaStride = pic->m_lowres.lumaStride;
-        weightedRef.isLowres = true;
-        weightedRef.isWeighted = false;
+        m_weightedRef.fpelPlane = m_weightedRef.lowresPlane[0];
+        m_weightedRef.lumaStride = pic->m_lowres.lumaStride;
+        m_weightedRef.isLowres = true;
+        m_weightedRef.isWeighted = false;
     }
 }
 
@@ -1163,8 +1283,8 @@
         score = fenc->costEst[b - p0][p1 - b];
     else
     {
-        weightedRef.isWeighted = false;
-        if (param->bEnableWeightedPred && b == p1 && b != p0 && fenc->lowresMvs[0][b - p0 - 1][0].x == 0x7FFF)
+        m_weightedRef.isWeighted = false;
+        if (m_param->bEnableWeightedPred && b == p1 && b != p0 && fenc->lowresMvs[0][b - p0 - 1][0].x == 0x7FFF)
         {
             if (!fenc->bIntraCalculated)
                 estimateFrameCost(frames, b, b, b, 0);
@@ -1172,26 +1292,29 @@
         }
 
         /* For each list, check to see whether we have lowres motion-searched this reference */
-        bDoSearch[0] = b != p0 && fenc->lowresMvs[0][b - p0 - 1][0].x == 0x7FFF;
-        bDoSearch[1] = b != p1 && fenc->lowresMvs[1][p1 - b - 1][0].x == 0x7FFF;
+        m_bDoSearch[0] = b != p0 && fenc->lowresMvs[0][b - p0 - 1][0].x == 0x7FFF;
+        m_bDoSearch[1] = b != p1 && fenc->lowresMvs[1][p1 - b - 1][0].x == 0x7FFF;
 
-        if (bDoSearch[0]) fenc->lowresMvs[0][b - p0 - 1][0].x = 0;
-        if (bDoSearch[1]) fenc->lowresMvs[1][p1 - b - 1][0].x = 0;
+        if (m_bDoSearch[0]) fenc->lowresMvs[0][b - p0 - 1][0].x = 0;
+        if (m_bDoSearch[1]) fenc->lowresMvs[1][p1 - b - 1][0].x = 0;
 
-        curb = b;
-        curp0 = p0;
-        curp1 = p1;
-        curframes = frames;
+        m_curb = b;
+        m_curp0 = p0;
+        m_curp1 = p1;
+        m_curframes = frames;
         fenc->costEst[b - p0][p1 - b] = 0;
         fenc->costEstAq[b - p0][p1 - b] = 0;
 
-        for (int i = 0; i < heightInCU; i++)
+        for (int i = 0; i < m_heightInCU; i++)
         {
-            rows[i].init();
-            rows[i].me.setSourcePlane(fenc->lowresPlane[0], fenc->lumaStride);
+            m_rows[i].init();
+            m_rows[i].m_me.setSourcePlane(fenc->lowresPlane[0], fenc->lumaStride);
+            if (!fenc->bIntraCalculated)
+                fenc->rowSatds[0][0][i] = 0;
+            fenc->rowSatds[b - p0][p1 - b][i] = 0;
         }
 
-        rowsCompleted = false;
+        m_bFrameCompleted = false;
 
         if (m_pool)
         {
@@ -1199,40 +1322,40 @@
 
             // enableAllRows must be already called
             enqueueRow(0);
-            while (!rowsCompleted)
+            while (!m_bFrameCompleted)
             {
-                WaveFront::findJob();
+                WaveFront::findJob(-1);
             }
 
             WaveFront::dequeue();
         }
         else
         {
-            for (int row = 0; row < heightInCU; row++)
+            for (int row = 0; row < m_heightInCU; row++)
             {
-                processRow(row);
+                processRow(row, -1);
             }
 
             x265_emms();
         }
 
         // Accumulate cost from each row
-        for (int row = 0; row < heightInCU; row++)
+        for (int row = 0; row < m_heightInCU; row++)
         {
-            score += rows[row].costEst;
-            fenc->costEst[0][0] += rows[row].costIntra;
-            if (param->rc.aqMode)
+            score += m_rows[row].m_costEst;
+            fenc->costEst[0][0] += m_rows[row].m_costIntra;
+            if (m_param->rc.aqMode)
             {
-                fenc->costEstAq[0][0] += rows[row].costIntraAq;
-                fenc->costEstAq[b - p0][p1 - b] += rows[row].costEstAq;
+                fenc->costEstAq[0][0] += m_rows[row].m_costIntraAq;
+                fenc->costEstAq[b - p0][p1 - b] += m_rows[row].m_costEstAq;
             }
-            fenc->intraMbs[b - p0] += rows[row].intraMbs;
+            fenc->intraMbs[b - p0] += m_rows[row].m_intraMbs;
         }
 
         fenc->bIntraCalculated = true;
 
         if (b != p1)
-            score = (uint64_t)score * 100 / (130 + param->bFrameBias);
+            score = (uint64_t)score * 100 / (130 + m_param->bFrameBias);
         if (b != p0 || b != p1) //Not Intra cost
             fenc->costEst[b - p0][p1 - b] = score;
     }
@@ -1246,7 +1369,7 @@
     return score;
 }
 
-uint32_t CostEstimate::weightCostLuma(Lowres **frames, int b, int p0, wpScalingParam *wp)
+uint32_t CostEstimate::weightCostLuma(Lowres **frames, int b, int p0, WeightParam *wp)
 {
     Lowres *fenc = frames[b];
     Lowres *ref  = frames[p0];
@@ -1261,9 +1384,9 @@
         int round = denom ? 1 << (denom - 1) : 0;
         int correction = IF_INTERNAL_PREC - X265_DEPTH; // intermediate interpolation depth
 
-        primitives.weight_pp(ref->buffer[0], wbuffer[0], stride, stride, stride, paddedLines,
+        primitives.weight_pp(ref->buffer[0], m_wbuffer[0], stride, stride, stride, m_paddedLines,
                              scale, round << correction, denom + correction, offset);
-        src = weightedRef.fpelPlane;
+        src = m_weightedRef.fpelPlane;
     }
 
     uint32_t cost = 0;
@@ -1310,9 +1433,9 @@
     unsigned int minscore = 0, origscore = 1;
     int found = 0;
 
-    w.setFromWeightAndOffset((int)(guessScale * 128 + 0.5f), 0);
-    mindenom = w.log2WeightDenom;
-    minscale = w.inputWeight;
+    m_w.setFromWeightAndOffset((int)(guessScale * 128 + 0.5f), 0, 7, true);
+    mindenom = m_w.log2WeightDenom;
+    minscale = m_w.inputWeight;
 
     origscore = minscore = weightCostLuma(frames, b, p0, NULL);
 
@@ -1331,8 +1454,8 @@
         curScale = (int)((1 << mindenom) * (fencMean - curOffset) / refMean + 0.5f);
         curScale = Clip3(0, 127, curScale);
     }
-    SET_WEIGHT(w, 1, curScale, mindenom, curOffset);
-    s = weightCostLuma(frames, b, p0, &w);
+    SET_WEIGHT(m_w, 1, curScale, mindenom, curOffset);
+    s = weightCostLuma(frames, b, p0, &m_w);
     COPY4_IF_LT(minscore, s, minscale, curScale, minoff, curOffset, found, 1);
 
     /* Use a smaller denominator if possible */
@@ -1346,83 +1469,79 @@
         return;
     else
     {
-        SET_WEIGHT(w, 1, minscale, mindenom, minoff);
+        SET_WEIGHT(m_w, 1, minscale, mindenom, minoff);
         // set weighted delta cost
         fenc->weightedCostDelta[deltaIndex] = minscore / origscore;
 
-        int offset = w.inputOffset << (X265_DEPTH - 8);
-        int scale = w.inputWeight;
-        int denom = w.log2WeightDenom;
+        int offset = m_w.inputOffset << (X265_DEPTH - 8);
+        int scale = m_w.inputWeight;
+        int denom = m_w.log2WeightDenom;
         int round = denom ? 1 << (denom - 1) : 0;
         int correction = IF_INTERNAL_PREC - X265_DEPTH; // intermediate interpolation depth
         int stride = ref->lumaStride;
 
         for (int i = 0; i < 4; i++)
         {
-            primitives.weight_pp(ref->buffer[i], wbuffer[i], stride, stride, stride, paddedLines,
+            primitives.weight_pp(ref->buffer[i], m_wbuffer[i], stride, stride, stride, m_paddedLines,
                                  scale, round << correction, denom + correction, offset);
         }
 
-        weightedRef.isWeighted = true;
+        m_weightedRef.isWeighted = true;
     }
 }
 
-void CostEstimate::processRow(int row)
+void CostEstimate::processRow(int row, int /*threadId*/)
 {
-    int realrow = heightInCU - 1 - row;
-    Lowres **frames = curframes;
-    Lowres *fenc = frames[curb];
-    ReferencePlanes *wfref0 = weightedRef.isWeighted ? &weightedRef : frames[curp0];
-
-    if (!fenc->bIntraCalculated)
-        fenc->rowSatds[0][0][realrow] = 0;
-    fenc->rowSatds[curb - curp0][curp1 - curb][realrow] = 0;
+    int realrow = m_heightInCU - 1 - row;
+    Lowres **frames = m_curframes;
+    ReferencePlanes *wfref0 = m_weightedRef.isWeighted ? &m_weightedRef : frames[m_curp0];
 
     /* Lowres lookahead goes backwards because the MVs are used as
      * predictors in the main encode.  This considerably improves MV
      * prediction overall. */
-    for (int i = widthInCU - 1 - rows[row].completed; i >= 0; i--)
+    for (int i = m_widthInCU - 1 - m_rows[row].m_completed; i >= 0; i--)
     {
         // TODO: use lowres MVs as motion candidates in full-res search
-        rows[row].estimateCUCost(frames, wfref0, i, realrow, curp0, curp1, curb, bDoSearch);
-        rows[row].completed++;
+        m_rows[row].estimateCUCost(frames, wfref0, i, realrow, m_curp0, m_curp1, m_curb, m_bDoSearch);
+        m_rows[row].m_completed++;
 
-        if (rows[row].completed >= 2 && row < heightInCU - 1)
+        if (m_rows[row].m_completed >= 2 && row < m_heightInCU - 1)
         {
-            ScopedLock below(rows[row + 1].lock);
-            if (rows[row + 1].active == false &&
-                rows[row + 1].completed + 2 <= rows[row].completed)
+            ScopedLock below(m_rows[row + 1].m_lock);
+            if (m_rows[row + 1].m_active == false &&
+                m_rows[row + 1].m_completed + 2 <= m_rows[row].m_completed)
             {
-                rows[row + 1].active = true;
+                m_rows[row + 1].m_active = true;
                 enqueueRow(row + 1);
             }
         }
 
-        ScopedLock self(rows[row].lock);
-        if (row > 0 && (int32_t)rows[row].completed < widthInCU - 1 && rows[row - 1].completed < rows[row].completed + 2)
+        ScopedLock self(m_rows[row].m_lock);
+        if (row > 0 && (int32_t)m_rows[row].m_completed < m_widthInCU - 1 &&
+            m_rows[row - 1].m_completed < m_rows[row].m_completed + 2)
         {
-            rows[row].active = false;
+            m_rows[row].m_active = false;
             x265_emms();
             return;
         }
     }
 
-    if (row == heightInCU - 1)
+    if (row == m_heightInCU - 1)
     {
-        rowsCompleted = true;
+        m_bFrameCompleted = true;
     }
     x265_emms();
 }
 
 void EstimateRow::init()
 {
-    costEst = 0;
-    costEstAq = 0;
-    costIntra = 0;
-    costIntraAq = 0;
-    intraMbs = 0;
-    active = false;
-    completed = 0;
+    m_costEst = 0;
+    m_costEstAq = 0;
+    m_costIntra = 0;
+    m_costIntraAq = 0;
+    m_intraMbs = 0;
+    m_active = false;
+    m_completed = 0;
 }
 
 void EstimateRow::estimateCUCost(Lowres **frames, ReferencePlanes *wfref0, int cux, int cuy, int p0, int p1, int b, bool bDoSearch[2])
@@ -1431,15 +1550,15 @@
     Lowres *fenc  = frames[b];
 
     const int bBidir = (b < p1);
-    const int cuXY = cux + cuy * widthInCU;
+    const int cuXY = cux + cuy * m_widthInCU;
     const int cuSize = X265_LOWRES_CU_SIZE;
     const int pelOffset = cuSize * cux + cuSize * cuy * fenc->lumaStride;
 
     // should this CU's cost contribute to the frame cost?
-    const bool bFrameScoreCU = (cux > 0 && cux < widthInCU - 1 &&
-                                cuy > 0 && cuy < heightInCU - 1) || widthInCU <= 2 || heightInCU <= 2;
+    const bool bFrameScoreCU = (cux > 0 && cux < m_widthInCU - 1 &&
+                                cuy > 0 && cuy < m_heightInCU - 1) || m_widthInCU <= 2 || m_heightInCU <= 2;
 
-    me.setSourcePU(pelOffset, cuSize, cuSize);
+    m_me.setSourcePU(pelOffset, cuSize, cuSize);
 
     /* A small, arbitrary bias to avoid VBV problems caused by zero-residual lookahead blocks. */
     int lowresPenalty = 4;
@@ -1450,14 +1569,14 @@
                             &fenc->lowresMvCosts[1][p1 - b - 1][cuXY] };
 
     MV mvmin, mvmax;
-    int bcost = me.COST_MAX;
+    int bcost = m_me.COST_MAX;
     int listused = 0;
 
     // establish search bounds that don't cross extended frame boundaries
     mvmin.x = (int16_t)(-cux * cuSize - 8);
     mvmin.y = (int16_t)(-cuy * cuSize - 8);
-    mvmax.x = (int16_t)((widthInCU - cux - 1) * cuSize + 8);
-    mvmax.y = (int16_t)((heightInCU - cuy - 1) * cuSize + 8);
+    mvmax.x = (int16_t)((m_widthInCU - cux - 1) * cuSize + 8);
+    mvmax.y = (int16_t)((m_heightInCU - cuy - 1) * cuSize + 8);
 
     if (p0 != p1)
     {
@@ -1477,15 +1596,15 @@
             mvc[0] = 0;
             mvc[2] = 0;
 #define MVC(mv) mvc[numc++] = mv;
-            if (cux < widthInCU - 1)
+            if (cux < m_widthInCU - 1)
                 MVC(fenc_mv[1]);
-            if (cuy < heightInCU - 1)
+            if (cuy < m_heightInCU - 1)
             {
-                MVC(fenc_mv[widthInCU]);
+                MVC(fenc_mv[m_widthInCU]);
                 if (cux > 0)
-                    MVC(fenc_mv[widthInCU - 1]);
-                if (cux < widthInCU - 1)
-                    MVC(fenc_mv[widthInCU + 1]);
+                    MVC(fenc_mv[m_widthInCU - 1]);
+                if (cux < m_widthInCU - 1)
+                    MVC(fenc_mv[m_widthInCU + 1]);
             }
 #undef MVC
             if (numc <= 1)
@@ -1495,7 +1614,7 @@
                 median_mv(mvp, mvc[0], mvc[1], mvc[2]);
             }
 
-            *fenc_costs[i] = me.motionEstimate(i ? fref1 : wfref0, mvmin, mvmax, mvp, numc, mvc, merange, *fenc_mvs[i]);
+            *fenc_costs[i] = m_me.motionEstimate(i ? fref1 : wfref0, mvmin, mvmax, mvp, numc, mvc, m_merange, *fenc_mvs[i]);
             COPY2_IF_LT(bcost, *fenc_costs[i], listused, i + 1);
         }
         if (bBidir)
@@ -1520,7 +1639,7 @@
     }
     if (!fenc->bIntraCalculated)
     {
-        int nLog2SizeMinus2 = g_convertToBit[cuSize]; // partition size
+        const int sizeIdx = X265_LOWRES_CU_BITS - 2; // partition size
 
         pixel _above0[X265_LOWRES_CU_SIZE * 4 + 1], *const above0 = _above0 + 2 * X265_LOWRES_CU_SIZE;
         pixel _above1[X265_LOWRES_CU_SIZE * 4 + 1], *const above1 = _above1 + 2 * X265_LOWRES_CU_SIZE;
@@ -1558,45 +1677,75 @@
 
         int predsize = cuSize * cuSize;
 
-        // generate 35 intra predictions into tmp
-        primitives.intra_pred[nLog2SizeMinus2][DC_IDX](predictions, cuSize, left0, above0, 0, (cuSize <= 16));
+        // generate 35 intra predictions into m_predictions
+        pixelcmp_t satd = primitives.satd[partitionFromLog2Size(X265_LOWRES_CU_BITS)];
+        int icost = m_me.COST_MAX, cost;
+        primitives.intra_pred[DC_IDX][sizeIdx](m_predictions, cuSize, left0, above0, 0, (cuSize <= 16));
+        cost = satd(m_me.fenc, FENC_STRIDE, m_predictions, cuSize);
+        if (cost < icost)
+            icost = cost;
         pixel *above = (cuSize >= 8) ? above1 : above0;
         pixel *left  = (cuSize >= 8) ? left1 : left0;
-        primitives.intra_pred[nLog2SizeMinus2][PLANAR_IDX](predictions + predsize, cuSize, left, above, 0, 0);
-        primitives.intra_pred_allangs[nLog2SizeMinus2](predictions + 2 * predsize, above0, left0, above1, left1, (cuSize <= 16));
+        primitives.intra_pred[PLANAR_IDX][sizeIdx](m_predictions, cuSize, left, above, 0, 0);
+        cost = satd(m_me.fenc, FENC_STRIDE, m_predictions, cuSize);
+        if (cost < icost)
+            icost = cost;
+        primitives.intra_pred_allangs[sizeIdx](m_predictions + 2 * predsize, above0, left0, above1, left1, (cuSize <= 16));
 
-        // calculate 35 satd costs, keep least cost
+        // calculate satd costs, keep least cost
         ALIGN_VAR_32(pixel, buf_trans[32 * 32]);
-        primitives.transpose[nLog2SizeMinus2](buf_trans, me.fenc, FENC_STRIDE);
-        pixelcmp_t satd = primitives.satd[partitionFromSizes(cuSize, cuSize)];
-        int icost = me.COST_MAX, cost;
-        for (uint32_t mode = 0; mode < 35; mode++)
+        primitives.transpose[sizeIdx](buf_trans, m_me.fenc, FENC_STRIDE);
+
+        int acost = m_me.COST_MAX;
+        uint32_t mode, lowmode = 4;
+        for (mode = 5; mode < 35; mode += 5)
         {
-            if ((mode >= 2) && (mode < 18))
-                cost = satd(buf_trans, cuSize, &predictions[mode * predsize], cuSize);
+            if (mode < 18)
+                cost = satd(buf_trans, cuSize, &m_predictions[mode * predsize], cuSize);
             else
-                cost = satd(me.fenc, FENC_STRIDE, &predictions[mode * predsize], cuSize);
-            if (cost < icost)
-                icost = cost;
+                cost = satd(m_me.fenc, FENC_STRIDE, &m_predictions[mode * predsize], cuSize);
+            COPY2_IF_LT(acost, cost, lowmode, mode);
         }
+        for (uint32_t dist = 2; dist >= 1; dist--)
+        {
+            mode = lowmode - dist;
+            if (mode < 18)
+                cost = satd(buf_trans, cuSize, &m_predictions[mode * predsize], cuSize);
+            else
+                cost = satd(m_me.fenc, FENC_STRIDE, &m_predictions[mode * predsize], cuSize);
+            COPY2_IF_LT(acost, cost, lowmode, mode);
 
-        const int intraPenalty = 5 * lookAheadLambda;
-        icost += intraPenalty + lowresPenalty;
+            mode = lowmode + dist;
+            if (mode < 18)
+                cost = satd(buf_trans, cuSize, &m_predictions[mode * predsize], cuSize);
+            else
+                cost = satd(m_me.fenc, FENC_STRIDE, &m_predictions[mode * predsize], cuSize);
+            COPY2_IF_LT(acost, cost, lowmode, mode);
+        }
+        if (acost < icost)
+            icost = acost;
+
+        const int intraPenalty = 5 * m_lookAheadLambda;
+        icost += intraPenalty + lowresPenalty; /* estimate intra signal cost */
         fenc->intraCost[cuXY] = icost;
-        fenc->rowSatds[0][0][cuy] += icost;
+        int icostAq = icost;
         if (bFrameScoreCU)
         {
-            costIntra += icost;
+            m_costIntra += icost;
             if (fenc->invQscaleFactor)
-                costIntraAq += (icost * fenc->invQscaleFactor[cuXY] + 128) >> 8;
+            {
+                icostAq = (icost * fenc->invQscaleFactor[cuXY] + 128) >> 8;
+                m_costIntraAq += icostAq;
+            }
         }
+        fenc->rowSatds[0][0][cuy] += icostAq;
     }
     bcost += lowresPenalty;
     if (!bBidir)
     {
         if (fenc->intraCost[cuXY] < bcost)
         {
-            if (bFrameScoreCU) intraMbs++;
+            if (bFrameScoreCU) m_intraMbs++;
             bcost = fenc->intraCost[cuXY];
             listused = 0;
         }
@@ -1605,13 +1754,17 @@
     /* For I frames these costs were accumulated earlier */
     if (p0 != p1)
     {
-        fenc->rowSatds[b - p0][p1 - b][cuy] += bcost;
+        int bcostAq = bcost;
         if (bFrameScoreCU)
         {
-            costEst += bcost;
+            m_costEst += bcost;
             if (fenc->invQscaleFactor)
-                costEstAq += (bcost * fenc->invQscaleFactor[cuXY] + 128) >> 8;
+            {
+                bcostAq = (bcost * fenc->invQscaleFactor[cuXY] + 128) >> 8;
+                m_costEstAq += bcostAq;
+            }
         }
+        fenc->rowSatds[b - p0][p1 - b][cuy] += bcostAq;
     }
     fenc->lowresCosts[b - p0][p1 - b][cuXY] = (uint16_t)(X265_MIN(bcost, LOWRES_COST_MASK) | (listused << LOWRES_COST_SHIFT));
 }

x265-hg20140324.tar.bz2/source/encoder/slicetype.h -> x265-hg20140928.tar.bz2/source/encoder/slicetype.h Changed

x265-hg20140324.tar.bz2/source/encoder/weightPrediction.cpp -> x265-hg20140928.tar.bz2/source/encoder/weightPrediction.cpp Changed

@@ -20,18 +20,18 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
-#include "TLibCommon/TComPic.h"
+#include "common.h"
+#include "frame.h"
 #include "lowres.h"
 #include "mv.h"
 #include "slicetype.h"
 #include "bitstream.h"
 
 using namespace x265;
-namespace weightp {
-
+namespace {
 struct Cache
 {
     const int * intraCost;
@@ -43,7 +43,7 @@
     int         lowresHeightInCU;
 };
 
-int sliceHeaderCost(wpScalingParam *w, int lambda, int bChroma)
+int sliceHeaderCost(WeightParam *w, int lambda, int bChroma)
 {
     /* 4 times higher, because chroma is analyzed at full resolution. */
     if (bChroma)
@@ -61,6 +61,7 @@
     MV mvmin, mvmax;
 
     int cu = 0;
+
     for (int y = 0; y < ref.lines; y += cuSize)
     {
         int pixoff = y * stride;
@@ -171,7 +172,7 @@
                     const Cache &   cache,
                     int             width,
                     int             height,
-                    wpScalingParam *w,
+                    WeightParam *w,
                     bool            bLuma)
 {
     if (w)
@@ -216,13 +217,17 @@
     x265_emms();
     return cost;
 }
+}
 
-void analyzeWeights(TComSlice& slice, x265_param& param, wpScalingParam wp[2][MAX_NUM_REF][3])
+namespace x265 {
+void weightAnalyse(Slice& slice, x265_param& param)
 {
-    TComPicYuv *fencYuv = slice.getPic()->getPicYuvOrg();
-    Lowres& fenc        = slice.getPic()->m_lowres;
+    WeightParam wp[2][MAX_NUM_REF][3];
+    TComPicYuv *fencYuv = slice.m_pic->getPicYuvOrg();
+    Lowres& fenc        = slice.m_pic->m_lowres;
+
+    Cache cache;
 
-    weightp::Cache cache;
     memset(&cache, 0, sizeof(cache));
     cache.intraCost = fenc.intraCost;
     cache.numPredDir = slice.isInterP() ? 1 : 2;
@@ -232,37 +237,31 @@
     cache.hshift = CHROMA_H_SHIFT(cache.csp);
     cache.vshift = CHROMA_V_SHIFT(cache.csp);
 
-    /* reset weight states */
-    for (int list = 0; list < cache.numPredDir; list++)
-    {
-        for (int ref = 0; ref < slice.getNumRefIdx(list); ref++)
-        {
-            SET_WEIGHT(wp[list][ref][0], false, 1, 0, 0);
-            SET_WEIGHT(wp[list][ref][1], false, 1, 0, 0);
-            SET_WEIGHT(wp[list][ref][2], false, 1, 0, 0);
-        }
-    }
-
     /* Use single allocation for motion compensated ref and weight buffers */
     pixel *mcbuf = X265_MALLOC(pixel, 2 * fencYuv->getStride() * fencYuv->getHeight());
     if (!mcbuf)
+    {
+        slice.disableWeights();
         return;
+    }
     pixel *weightTemp = mcbuf + fencYuv->getStride() * fencYuv->getHeight();
 
-    int lambda = (int)x265_lambda2_non_I[X265_LOOKAHEAD_QP];
-    int curPoc = slice.getPOC();
+    int lambda = (int)x265_lambda_tab[X265_LOOKAHEAD_QP];
+    int curPoc = slice.m_poc;
     const float epsilon = 1.f / 128.f;
 
+    int chromaDenom, lumaDenom, denom;
+    chromaDenom = lumaDenom = 7;
     int numpixels[3];
-    int w = ((fencYuv->getWidth()  + 15) >> 4) << 4;
-    int h = ((fencYuv->getHeight() + 15) >> 4) << 4;
-    numpixels[0] = w * h;
+    int w16 = ((fencYuv->getWidth()  + 15) >> 4) << 4;
+    int h16 = ((fencYuv->getHeight() + 15) >> 4) << 4;
+    numpixels[0] = w16 * h16;
     numpixels[1] = numpixels[2] = numpixels[0] >> (cache.hshift + cache.vshift);
 
     for (int list = 0; list < cache.numPredDir; list++)
     {
-        wpScalingParam *weights = wp[list][0];
-        TComPic *refPic = slice.getRefPic(list, 0);
+        WeightParam *weights = wp[list][0];
+        Frame *refPic = slice.m_refPicList[list][0];
         Lowres& refLowres = refPic->m_lowres;
         int diffPoc = abs(curPoc - refPic->getPOC());
 
@@ -270,6 +269,7 @@
         float guessScale[3], fencMean[3], refMean[3];
         for (int plane = 0; plane < 3; plane++)
         {
+            SET_WEIGHT(weights[plane], false, 1, 0, 0);
             uint64_t fencVar = fenc.wp_ssd[plane] + !refLowres.wp_ssd[plane];
             uint64_t refVar  = refLowres.wp_ssd[plane] + !refLowres.wp_ssd[plane];
             guessScale[plane] = sqrt((float)fencVar / refVar);
@@ -278,14 +278,14 @@
         }
 
         /* make sure both our scale factors fit */
-        int chromaDenom = 7;
-        while (chromaDenom > 0)
+        while (!list && chromaDenom > 0)
         {
             float thresh = 127.f / (1 << chromaDenom);
             if (guessScale[1] < thresh && guessScale[2] < thresh)
                 break;
             chromaDenom--;
         }
+
         SET_WEIGHT(weights[1], false, 1 << chromaDenom, chromaDenom, 0);
         SET_WEIGHT(weights[2], false, 1 << chromaDenom, chromaDenom, 0);
 
@@ -293,6 +293,7 @@
 
         for (int plane = 0; plane < 3; plane++)
         {
+            denom = plane ? chromaDenom : lumaDenom;
             if (plane && !weights[0].bPresentFlag)
                 break;
 
@@ -300,20 +301,20 @@
             x265_emms();
             if (fabsf(refMean[plane] - fencMean[plane]) < 0.5f && fabsf(1.f - guessScale[plane]) < epsilon)
             {
-                SET_WEIGHT(weights[plane], 0, 1, 0, 0);
+                SET_WEIGHT(weights[plane], 0, 1 << denom, denom, 0);
                 continue;
             }
 
             if (plane)
             {
-                int scale = Clip3(0, 255, (int)(guessScale[plane] * (1 << chromaDenom) + 0.5f));
+                int scale = Clip3(0, 255, (int)(guessScale[plane] * (1 << denom) + 0.5f));
                 if (scale > 127)
                     continue;
                 weights[plane].inputWeight = scale;
             }
             else
             {
-                weights[plane].setFromWeightAndOffset((int)(guessScale[plane] * 128 + 0.5f), 0, 7);
+                weights[plane].setFromWeightAndOffset((int)(guessScale[plane] * (1 << denom) + 0.5f), 0, denom, !list);
             }
 
             int mindenom = weights[plane].log2WeightDenom;
@@ -400,13 +401,17 @@
                 break;
 
             default:
+                slice.disableWeights();
                 X265_FREE(mcbuf);
                 return;
             }
 
             uint32_t origscore = weightCost(orig, fref, weightTemp, stride, cache, width, height, NULL, !plane);
             if (!origscore)
+            {
+                SET_WEIGHT(weights[plane], 0, 1 << denom, denom, 0);
                 continue;
+            }
 
             uint32_t minscore = origscore;
             bool bFound = false;
@@ -439,10 +444,10 @@
                 int endOffset   = Clip3(-128, 127, curOffset + offsetDist);
                 for (int off = startOffset; off <= endOffset; off++)
                 {
-                    wpScalingParam wsp;
+                    WeightParam wsp;
                     SET_WEIGHT(wsp, true, curScale, mindenom, off);
                     uint32_t s = weightCost(orig, fref, weightTemp, stride, cache, width, height, &wsp, !plane) +
-                                 sliceHeaderCost(&wsp, lambda, !!plane);
+                        sliceHeaderCost(&wsp, lambda, !!plane);
                     COPY4_IF_LT(minscore, s, minscale, curScale, minoff, off, bFound, true);
 
                     /* Don't check any more offsets if the previous one had a lower cost than the current one */
@@ -452,7 +457,7 @@
             }
 
             /* Use a smaller luma denominator if possible */
-            if (!plane)
+            if (!(plane || list))
             {
                 while (mindenom > 0 && !(minscale & 1))
                 {
@@ -463,7 +468,7 @@
 
             if (!bFound || (minscale == (1 << mindenom) && minoff == 0) || (float)minscore / origscore > 0.998f)
             {
-                SET_WEIGHT(weights[plane], false, 1, 0, 0);
+                SET_WEIGHT(weights[plane], false, 1 << denom, denom, 0);
             }
             else
             {
@@ -482,20 +487,22 @@
                     weights[1] = weights[2];
             }
         }
-    }
 
-    X265_FREE(mcbuf);
-}
-}
+        lumaDenom = weights[0].log2WeightDenom;
+        chromaDenom = weights[1].log2WeightDenom;
 
-namespace x265 {
-void weightAnalyse(TComSlice& slice, x265_param& param)
-{
-    wpScalingParam wp[2][MAX_NUM_REF][3];
+        /* reset weight states */
+        for (int ref = 1; ref < slice.m_numRefIdx[list]; ref++)
+        {
+            SET_WEIGHT(wp[list][ref][0], false, 1 << lumaDenom, lumaDenom, 0);
+            SET_WEIGHT(wp[list][ref][1], false, 1 << chromaDenom, chromaDenom, 0);
+            SET_WEIGHT(wp[list][ref][2], false, 1 << chromaDenom, chromaDenom, 0);
+        }
+    }
 
-    weightp::analyzeWeights(slice, param, wp);
+    X265_FREE(mcbuf);
 
-    slice.setWpScaling(wp);
+    memcpy(slice.m_weightPredTable, wp, sizeof(WeightParam) * 2 * MAX_NUM_REF * 3);
 
     if (param.logLevel >= X265_LOG_FULL)
     {
@@ -503,11 +510,11 @@
         int p = 0;
         bool bWeighted = false;
 
-        p = sprintf(buf, "poc: %d weights:", slice.getPOC());
+        p = sprintf(buf, "poc: %d weights:", slice.m_poc);
         int numPredDir = slice.isInterP() ? 1 : 2;
         for (int list = 0; list < numPredDir; list++)
         {
-            wpScalingParam* w = &wp[list][0][0];
+            WeightParam* w = &wp[list][0][0];
             if (w[0].bPresentFlag || w[1].bPresentFlag || w[2].bPresentFlag)
             {
                 bWeighted = true;

x265-hg20140928.tar.bz2/source/filters Added

x265-hg20140928.tar.bz2/source/filters/filters.cpp Added

@@ -0,0 +1,79 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Selvakumar Nithiyaruban <selvakumar@multicorewareinc.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#include "filters.h"
+#include "common.h"
+
+/* The dithering algorithm is based on Sierra-2-4A error diffusion. */
+void ditherPlane(pixel *dst, int dstStride, uint16_t *src, int srcStride,
+                 int width, int height, int16_t *errors, int bitDepth)
+{
+    const int lShift = 16 - bitDepth;
+    const int rShift = 16 - bitDepth + 2;
+    const int half = (1 << (16 - bitDepth + 1));
+    const int pixelMax = (1 << bitDepth) - 1;
+
+    memset(errors, 0, (width + 1) * sizeof(int16_t));
+    int pitch = 1;
+    for (int y = 0; y < height; y++, src += srcStride, dst += dstStride)
+    {
+        int16_t err = 0;
+        for (int x = 0; x < width; x++)
+        {
+            err = err * 2 + errors[x] + errors[x + 1];
+            dst[x * pitch] = (pixel)Clip3(0, pixelMax, ((src[x * 1] << 2) + err + half) >> rShift);
+            errors[x] = err = src[x * pitch] - (dst[x * pitch] << lShift);
+        }
+    }
+}
+
+void ditherImage(x265_picture& picIn, int picWidth, int picHeight, int16_t *errorBuf, int bitDepth)
+{
+    /* This portion of code is from readFrame in x264. */
+    for (int i = 0; i < x265_cli_csps[picIn.colorSpace].planes; i++)
+    {
+        if ((picIn.bitDepth & 7) && (picIn.bitDepth != 16))
+        {
+            /* upconvert non 16bit high depth planes to 16bit */
+            uint16_t *plane = (uint16_t*)picIn.planes[i];
+            uint32_t pixelCount = x265_picturePlaneSize(picIn.colorSpace, picWidth, picHeight, i);
+            int lShift = 16 - picIn.bitDepth;
+
+            /* This loop assumes width is equal to stride which
+               happens to be true for file reader outputs */
+            for (uint32_t j = 0; j < pixelCount; j++)
+            {
+                plane[j] = plane[j] << lShift;
+            }
+        }
+    }
+
+    for (int i = 0; i < x265_cli_csps[picIn.colorSpace].planes; i++)
+    {
+        int height = (int)(picHeight >> x265_cli_csps[picIn.colorSpace].height[i]);
+        int width = (int)(picWidth >> x265_cli_csps[picIn.colorSpace].width[i]);
+
+        ditherPlane(((pixel*)picIn.planes[i]), picIn.stride[i] / sizeof(pixel), ((uint16_t*)picIn.planes[i]),
+                    picIn.stride[i] / 2, width, height, errorBuf, bitDepth);
+    }
+}

x265-hg20140928.tar.bz2/source/filters/filters.h Added

@@ -0,0 +1,31 @@
+/*****************************************************************************
+ * Copyright (C) 2013 x265 project
+ *
+ * Authors: Selvakumar Nithiyaruban <selvakumar@multicorewareinc.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at license @ x265.com.
+ *****************************************************************************/
+
+#ifndef X265_FILTERS_H
+#define X265_FILTERS_H
+
+#include "x265.h"
+
+void ditherImage(x265_picture&, int picWidth, int picHeight, int16_t *errorBuf, int bitDepth);
+
+#endif //X265_FILTERS_H

x265-hg20140324.tar.bz2/source/input/input.cpp -> x265-hg20140928.tar.bz2/source/input/input.cpp Changed

x265-hg20140324.tar.bz2/source/input/input.h -> x265-hg20140928.tar.bz2/source/input/input.h Changed

x265-hg20140324.tar.bz2/source/input/y4m.cpp -> x265-hg20140928.tar.bz2/source/input/y4m.cpp Changed

@@ -18,11 +18,10 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "y4m.h"
-#include "PPA/ppa.h"
 #include "common.h"
 
 #include <iostream>
@@ -44,15 +43,13 @@
 
 Y4MInput::Y4MInput(InputFileInfo& info)
 {
-    for (uint32_t i = 0; i < QUEUE_SIZE; i++)
-    {
-        plane[i][2] = plane[i][1] = plane[i][0] = NULL;
-        frameStat[i] = false;
-    }
+    for (int i = 0; i < QUEUE_SIZE; i++)
+        buf[i] = NULL;
 
-    head.set(0);
-    tail.set(0);
+    readCount.set(0);
+    writeCount.set(0);
 
+    threadActive = false;
     colorSpace = info.csp;
     sarWidth = info.sarWidth;
     sarHeight = info.sarHeight;
@@ -60,6 +57,8 @@
     height = info.height;
     rateNum = info.fpsNum;
     rateDenom = info.fpsDenom;
+    depth = info.depth;
+    framesize = 0;
 
     ifs = NULL;
     if (!strcmp(info.filename, "-"))
@@ -72,13 +71,25 @@
     else
         ifs = new ifstream(info.filename, ios::binary | ios::in);
 
-    threadActive = false;
     if (ifs && ifs->good() && parseHeader())
     {
+        int pixelbytes = depth > 8 ? 2 : 1;
+        for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
+        {
+            int stride = (width >> x265_cli_csps[colorSpace].width[i]) * pixelbytes;
+            framesize += (stride * (height >> x265_cli_csps[colorSpace].height[i]));
+        }
+
         threadActive = true;
-        for (uint32_t i = 0; i < QUEUE_SIZE; i++)
+        for (int q = 0; q < QUEUE_SIZE; q++)
         {
-            pictureAlloc(i);
+            buf[q] = X265_MALLOC(char, framesize);
+            if (!buf[q])
+            {
+                x265_log(NULL, X265_LOG_ERROR, "y4m: buffer allocation failure, aborting");
+                threadActive = false;
+                break;
+            }
         }
     }
     if (!threadActive)
@@ -96,15 +107,10 @@
     info.fpsNum = rateNum;
     info.fpsDenom = rateDenom;
     info.csp = colorSpace;
-    info.depth = 8;
+    info.depth = depth;
     info.frameCount = -1;
 
-    size_t frameSize = strlen(header) + 1;
-    for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
-    {
-        frameSize += (size_t)((width  >> x265_cli_csps[colorSpace].width[i]) *
-                              (height >> x265_cli_csps[colorSpace].height[i]));
-    }
+    size_t estFrameSize = framesize + strlen(header) + 1; /* assume basic FRAME\n headers */
 
     /* try to estimate frame count, if this is not stdin */
     if (ifs != &cin)
@@ -113,34 +119,36 @@
 
 #if defined(_MSC_VER) && _MSC_VER < 1700
         /* Older MSVC versions cannot handle 64bit file sizes properly, so go native */
-        HANDLE hFile = CreateFileA(info.filename, GENERIC_READ, 
-            FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING, 
-            FILE_ATTRIBUTE_NORMAL, NULL);
+        HANDLE hFile = CreateFileA(info.filename, GENERIC_READ,
+                                   FILE_SHARE_READ | FILE_SHARE_WRITE, NULL, OPEN_EXISTING,
+                                   FILE_ATTRIBUTE_NORMAL, NULL);
         if (hFile != INVALID_HANDLE_VALUE)
         {
             LARGE_INTEGER size;
             if (GetFileSizeEx(hFile, &size))
-                info.frameCount = (int)((size.QuadPart - (int64_t)cur) / frameSize);
+                info.frameCount = (int)((size.QuadPart - (int64_t)cur) / estFrameSize);
             CloseHandle(hFile);
         }
-#else
+#else // if defined(_MSC_VER) && _MSC_VER < 1700
         if (cur >= 0)
         {
             ifs->seekg(0, ios::end);
             istream::pos_type size = ifs->tellg();
             ifs->seekg(cur, ios::beg);
             if (size > 0)
-                info.frameCount = (int)((size - cur) / frameSize);
+                info.frameCount = (int)((size - cur) / estFrameSize);
         }
-#endif
+#endif // if defined(_MSC_VER) && _MSC_VER < 1700
     }
 
     if (info.skipFrames)
     {
+#if X86_64
+        ifs->seekg((uint64_t)estFrameSize * info.skipFrames, ios::cur);
+#else
         for (int i = 0; i < info.skipFrames; i++)
-        {
-            ifs->ignore(frameSize);
-        }
+            ifs->ignore(estFrameSize);
+#endif
     }
 }
 
@@ -149,29 +157,16 @@
     if (ifs && ifs != &cin)
         delete ifs;
 
-    for (uint32_t i = 0; i < QUEUE_SIZE; i++)
-    {
-        for (int j = 0; j < x265_cli_csps[colorSpace].planes; j++)
-        {
-            delete[] plane[i][j];
-        }
-    }
+    for (int i = 0; i < QUEUE_SIZE; i++)
+        X265_FREE(buf[i]);
 }
 
-void Y4MInput::pictureAlloc(int queueindex)
+void Y4MInput::release()
 {
-    for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
-    {
-        plane_stride[i] = (uint32_t)(width >> x265_cli_csps[colorSpace].width[i]);
-        plane_size[i] =   (uint32_t)(plane_stride[i] * (height >> x265_cli_csps[colorSpace].height[i]));
-        plane[queueindex][i] = new char[plane_size[i]];
-        if (!plane[queueindex][i])
-        {
-            x265_log(NULL, X265_LOG_ERROR, "y4m: buffer allocation failure, aborting");
-            threadActive = false;
-            return;
-        }
-    }
+    threadActive = false;
+    readCount.set(readCount.get()); // unblock file reader
+    stop();
+    delete this;
 }
 
 bool Y4MInput::parseHeader()
@@ -180,6 +175,7 @@
         return false;
 
     int csp = 0;
+    int d = 0;
 
     while (!ifs->eof())
     {
@@ -309,20 +305,35 @@
 
             case 'C':
                 csp = 0;
+                d = 0;
                 while (!ifs->eof())
                 {
                     c = ifs->get();
 
-                    if (c == ' ' || c == '\n')
+                    if (c <= '9' && c >= '0')
                     {
-                        break;
+                        csp = csp * 10 + (c - '0');
                     }
-                    else
+                    else if (c == 'p')
                     {
-                        csp = csp * 10 + (c - '0');
+                        // example: C420p16
+                        while (!ifs->eof())
+                        {
+                            c = ifs->get();
+
+                            if (c <= '9' && c >= '0')
+                                d = d * 10 + (c - '0');
+                            else
+                                break;
+                        }
+                        break;
                     }
+                    else
+                        break;
                 }
 
+                if (d >= 8 && d <= 16)
+                    depth = d;
                 colorSpace = (csp == 444) ? X265_CSP_I444 : (csp == 422) ? X265_CSP_I422 : X265_CSP_I420;
                 break;
 
@@ -371,98 +382,85 @@
     }
     while (threadActive);
 
-    /* "open the throttle" at the end, allow reader to consume
-     * remaining valid queue entries */
     threadActive = false;
-    frameStat[tail.get()] = false;
-    tail.set(QUEUE_SIZE);
+    writeCount.set(writeCount.get()); // unblock readPicture
 }
 
-bool Y4MInput::readPicture(x265_picture& pic)
+bool Y4MInput::populateFrameQueue()
 {
-    PPAStartCpuEventFunc(read_yuv);
-    int curHead = head.get();
-    int curTail = tail.get();
-
-#if ENABLE_THREADING
-
-    while (curHead == curTail)
-    {
-        curTail = tail.waitForChange(curTail);
-        if (!threadActive)
-            return false;
-    }
-
-#else
-
-    populateFrameQueue();
-
-#endif
-
-    if (!frameStat[curHead])
+    if (!ifs || ifs->fail())
         return false;
 
-    pic.bitDepth = 8;
-    pic.colorSpace = colorSpace;
-    for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
-    {
-        pic.planes[i] = plane[curHead][i];
-        pic.stride[i] = plane_stride[i];
-    }
-
-    head.set((curHead + 1) % QUEUE_SIZE);
-
-    PPAStopCpuEventFunc(read_yuv);
-    return true;
-}
-
-bool Y4MInput::populateFrameQueue()
-{
     /* strip off the FRAME header */
     char hbuf[sizeof(header)];
 
-    if (!ifs)
-        return false;
-
     ifs->read(hbuf, strlen(header));
-    if (!ifs->good())
+    if (ifs->eof())
         return false;
 
-    if (memcmp(hbuf, header, strlen(header)))
+    if (!ifs->good() || memcmp(hbuf, header, strlen(header)))
     {
         x265_log(NULL, X265_LOG_ERROR, "y4m: frame header missing\n");
         return false;
     }
+
     /* consume bytes up to line feed */
     int c = ifs->get();
     while (c != '\n' && ifs->good())
-    {
         c = ifs->get();
-    }
 
-    int curTail = tail.get();
-    int curHead = head.get();
-    while ((curTail + 1) % QUEUE_SIZE == curHead)
+    /* wait for room in the ring buffer */
+    int written = writeCount.get();
+    int read = readCount.get();
+    while (written - read > QUEUE_SIZE - 2)
     {
-        curHead = head.waitForChange(curHead);
+        read = readCount.waitForChange(read);
         if (!threadActive)
             return false;
     }
 
-    for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
+    ifs->read(buf[written % QUEUE_SIZE], framesize);
+    if (ifs->good())
     {
-        ifs->read(plane[curTail][i], plane_size[i]);
+        writeCount.incr();
+        return true;
     }
-
-    frameStat[curTail] = !ifs->fail();
-    tail.set((curTail + 1) % QUEUE_SIZE);
-    return !ifs->fail();
+    else
+        return false;
 }
 
-void Y4MInput::release()
+bool Y4MInput::readPicture(x265_picture& pic)
 {
-    threadActive = false;
-    head.set(QUEUE_SIZE); // unblock file reader
-    stop();
-    delete this;
+    int read = readCount.get();
+    int written = writeCount.get();
+
+#if ENABLE_THREADING
+
+    /* only wait if the read thread is still active */
+    while (threadActive && read == written)
+        written = writeCount.waitForChange(written);
+
+#else
+
+    populateFrameQueue();
+
+#endif // if ENABLE_THREADING
+
+    if (read < written)
+    {
+        int pixelbytes = depth > 8 ? 2 : 1;
+        pic.bitDepth = depth;
+        pic.colorSpace = colorSpace;
+        pic.stride[0] = width * pixelbytes;
+        pic.stride[1] = pic.stride[0] >> x265_cli_csps[colorSpace].width[1];
+        pic.stride[2] = pic.stride[0] >> x265_cli_csps[colorSpace].width[2];
+        pic.planes[0] = buf[read % QUEUE_SIZE];
+        pic.planes[1] = (char*)pic.planes[0] + pic.stride[0] * height;
+        pic.planes[2] = (char*)pic.planes[1] + pic.stride[1] * (height >> x265_cli_csps[colorSpace].height[1]);
+        readCount.incr();
+        return true;
+    }
+    else
+        return false;
 }
+

x265-hg20140324.tar.bz2/source/input/y4m.h -> x265-hg20140928.tar.bz2/source/input/y4m.h Changed

x265-hg20140324.tar.bz2/source/input/yuv.cpp -> x265-hg20140928.tar.bz2/source/input/yuv.cpp Changed

x265-hg20140324.tar.bz2/source/input/yuv.h -> x265-hg20140928.tar.bz2/source/input/yuv.h Changed

x265-hg20140324.tar.bz2/source/output/output.cpp -> x265-hg20140928.tar.bz2/source/output/output.cpp Changed

x265-hg20140324.tar.bz2/source/output/output.h -> x265-hg20140928.tar.bz2/source/output/output.h Changed

x265-hg20140324.tar.bz2/source/output/y4m.cpp -> x265-hg20140928.tar.bz2/source/output/y4m.cpp Changed

@@ -18,12 +18,10 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "common.h"
-#include "PPA/ppa.h"
-#include "common.h"
 #include "output.h"
 #include "y4m.h"
 
@@ -61,7 +59,6 @@
 
 bool Y4MOutput::writePicture(const x265_picture& pic)
 {
-    PPAStartCpuEventFunc(write_yuv);
     std::ofstream::pos_type outPicPos = header;
     outPicPos += (uint64_t)pic.poc * (6 + frameSize);
     ofs.seekp(outPicPos);
@@ -78,12 +75,13 @@
         x265_log(NULL, X265_LOG_WARNING, "y4m: forcing reconstructed pixels to 8 bits\n");
     }
 #endif
-    assert(pic.colorSpace == colorSpace);
+
+    X265_CHECK(pic.colorSpace == colorSpace, "invalid color space\n");
 
 #if HIGH_BIT_DEPTH
 
     // encoder gave us short pixels, downshift, then write
-    assert(pic.bitDepth > 8);
+    X265_CHECK(pic.bitDepth > 8, "invalid bit depth\n");
     int shift = pic.bitDepth - 8;
     for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
     {
@@ -102,7 +100,7 @@
 
 #else // if HIGH_BIT_DEPTH
 
-    assert(pic.bitDepth == 8);
+    X265_CHECK(pic.bitDepth == 8, "invalid bit depth\n");
     for (int i = 0; i < x265_cli_csps[colorSpace].planes; i++)
     {
         char *src = (char*)pic.planes[i];
@@ -115,6 +113,5 @@
 
 #endif // if HIGH_BIT_DEPTH
 
-    PPAStopCpuEventFunc(write_yuv);
     return true;
 }

x265-hg20140324.tar.bz2/source/output/y4m.h -> x265-hg20140928.tar.bz2/source/output/y4m.h Changed

x265-hg20140324.tar.bz2/source/output/yuv.cpp -> x265-hg20140928.tar.bz2/source/output/yuv.cpp Changed

x265-hg20140324.tar.bz2/source/output/yuv.h -> x265-hg20140928.tar.bz2/source/output/yuv.h Changed

x265-hg20140324.tar.bz2/source/test/CMakeLists.txt -> x265-hg20140928.tar.bz2/source/test/CMakeLists.txt Changed

x265-hg20140928.tar.bz2/source/test/checkasm-a.asm Added

@@ -0,0 +1,221 @@
+;*****************************************************************************
+;* checkasm-a.asm: assembly check tool
+;*****************************************************************************
+;* Copyright (C) 2008-2014 x264 project
+;*
+;* Authors: Loren Merritt <lorenm@u.washington.edu>
+;*          Henrik Gramner <henrik@gramner.com>
+;*
+;* This program is free software; you can redistribute it and/or modify
+;* it under the terms of the GNU General Public License as published by
+;* the Free Software Foundation; either version 2 of the License, or
+;* (at your option) any later version.
+;*
+;* This program is distributed in the hope that it will be useful,
+;* but WITHOUT ANY WARRANTY; without even the implied warranty of
+;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+;* GNU General Public License for more details.
+;*
+;* You should have received a copy of the GNU General Public License
+;* along with this program; if not, write to the Free Software
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+;*
+;* This program is also available under a commercial proprietary license.
+;* For more information, contact us at license @ x265.com.
+;*****************************************************************************
+
+%include "../common/x86/x86inc.asm"
+
+SECTION_RODATA
+
+error_message: db "failed to preserve register", 0
+
+%if ARCH_X86_64
+; just random numbers to reduce the chance of incidental match
+ALIGN 16
+x6:  ddq 0x79445c159ce790641a1b2550a612b48c
+x7:  ddq 0x86b2536fcd8cf6362eed899d5a28ddcd
+x8:  ddq 0x3f2bf84fc0fcca4eb0856806085e7943
+x9:  ddq 0xd229e1f5b281303facbd382dcf5b8de2
+x10: ddq 0xab63e2e11fa38ed971aeaff20b095fd9
+x11: ddq 0x77d410d5c42c882d89b0c0765892729a
+x12: ddq 0x24b3c1d2a024048bc45ea11a955d8dd5
+x13: ddq 0xdd7b8919edd427862e8ec680de14b47c
+x14: ddq 0x11e53e2b2ac655ef135ce6888fa02cbf
+x15: ddq 0x6de8f4c914c334d5011ff554472a7a10
+n7:   dq 0x21f86d66c8ca00ce
+n8:   dq 0x75b6ba21077c48ad
+n9:   dq 0xed56bb2dcb3c7736
+n10:  dq 0x8bda43d3fd1a7e06
+n11:  dq 0xb64a9c9e5d318408
+n12:  dq 0xdf9a54b303f1d3a3
+n13:  dq 0x4a75479abd64e097
+n14:  dq 0x249214109d5d1c88
+%endif
+
+SECTION .text
+
+cextern_naked puts
+
+; max number of args used by any x265 asm function.
+; (max_args % 4) must equal 3 for stack alignment
+%define max_args 15
+
+%if ARCH_X86_64
+
+;-----------------------------------------------------------------------------
+; void x265_checkasm_stack_clobber( uint64_t clobber, ... )
+;-----------------------------------------------------------------------------
+cglobal checkasm_stack_clobber, 1,2
+    ; Clobber the stack with junk below the stack pointer
+    %define size (max_args+6)*8
+    SUB  rsp, size
+    mov   r1, size-8
+.loop:
+    mov [rsp+r1], r0
+    sub   r1, 8
+    jge .loop
+    ADD  rsp, size
+    RET
+
+%if WIN64
+    %assign free_regs 7
+%else
+    %assign free_regs 9
+%endif
+
+;-----------------------------------------------------------------------------
+; intptr_t x265_checkasm_call( intptr_t (*func)(), int *ok, ... )
+;-----------------------------------------------------------------------------
+cglobal checkasm_call_float
+INIT_XMM
+cglobal checkasm_call, 2,15,16,max_args*8+8
+    mov  r6, r0
+    mov  [rsp+max_args*8], r1
+
+    ; All arguments have been pushed on the stack instead of registers in order to
+    ; test for incorrect assumptions that 32-bit ints are zero-extended to 64-bit.
+    mov  r0, r6mp
+    mov  r1, r7mp
+    mov  r2, r8mp
+    mov  r3, r9mp
+%if UNIX64
+    mov  r4, r10mp
+    mov  r5, r11mp
+    %assign i 6
+    %rep max_args-6
+        mov  r9, [rsp+stack_offset+(i+1)*8]
+        mov  [rsp+(i-6)*8], r9
+        %assign i i+1
+    %endrep
+%else
+    %assign i 4
+    %rep max_args-4
+        mov  r9, [rsp+stack_offset+(i+7)*8]
+        mov  [rsp+i*8], r9
+        %assign i i+1
+    %endrep
+%endif
+
+%if WIN64
+    %assign i 6
+    %rep 16-6
+        mova m %+ i, [x %+ i]
+        %assign i i+1
+    %endrep
+%endif
+
+%assign i 14
+%rep 15-free_regs
+    mov  r %+ i, [n %+ i]
+    %assign i i-1
+%endrep
+    call r6
+%assign i 14
+%rep 15-free_regs
+    xor  r %+ i, [n %+ i]
+    or  r14, r %+ i
+    %assign i i-1
+%endrep
+
+%if WIN64
+    %assign i 6
+    %rep 16-6
+        pxor m %+ i, [x %+ i]
+        por  m6, m %+ i
+        %assign i i+1
+    %endrep
+    packsswb m6, m6
+    movq r5, m6
+    or  r14, r5
+%endif
+
+    jz .ok
+    mov  r9, rax
+    lea  r0, [error_message]
+    call puts
+    mov  r1, [rsp+max_args*8]
+    mov  dword [r1], 0
+    mov  rax, r9
+.ok:
+    RET
+
+%else
+
+; just random numbers to reduce the chance of incidental match
+%define n3 dword 0x6549315c
+%define n4 dword 0xe02f3e23
+%define n5 dword 0xb78d0d1d
+%define n6 dword 0x33627ba7
+
+;-----------------------------------------------------------------------------
+; intptr_t x265_checkasm_call( intptr_t (*func)(), int *ok, ... )
+;-----------------------------------------------------------------------------
+cglobal checkasm_call_float
+cglobal checkasm_call, 1,7
+    mov  r3, n3
+    mov  r4, n4
+    mov  r5, n5
+    mov  r6, n6
+%rep max_args
+    push dword [esp+24+max_args*4]
+%endrep
+    call r0
+    add  esp, max_args*4
+    xor  r3, n3
+    xor  r4, n4
+    xor  r5, n5
+    xor  r6, n6
+    or   r3, r4
+    or   r5, r6
+    or   r3, r5
+    jz .ok
+    mov  r3, eax
+    lea  r1, [error_message]
+    push r1
+    call puts
+    add  esp, 4
+    mov  r1, r1m
+    mov  dword [r1], 0
+    mov  eax, r3
+.ok:
+    REP_RET
+
+%endif ; ARCH_X86_64
+
+;-----------------------------------------------------------------------------
+; int x265_stack_pagealign( int (*func)(), int align )
+;-----------------------------------------------------------------------------
+cglobal stack_pagealign, 2,2
+    movsxdifnidn r1, r1d
+    push rbp
+    mov  rbp, rsp
+%if WIN64
+    sub  rsp, 32 ; shadow space
+%endif
+    and  rsp, ~0xfff
+    sub  rsp, r1
+    call r0
+    leave
+    RET
+

x265-hg20140324.tar.bz2/source/test/intrapredharness.cpp -> x265-hg20140928.tar.bz2/source/test/intrapredharness.cpp Changed

x265-hg20140324.tar.bz2/source/test/intrapredharness.h -> x265-hg20140928.tar.bz2/source/test/intrapredharness.h Changed

x265-hg20140324.tar.bz2/source/test/ipfilterharness.cpp -> x265-hg20140928.tar.bz2/source/test/ipfilterharness.cpp Changed

@@ -21,48 +21,21 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "TLibCommon/TComRom.h"
 #include "common.h"
 #include "ipfilterharness.h"
 
-#define ITERS  100
-#define TEST_CASES 3
-#define SMAX (1 << 12)
-#define SMIN (-1 << 12)
-
 using namespace x265;
 
 IPFilterHarness::IPFilterHarness()
 {
-    // Assuming max_height = max_width = max_srcStride = max_dstStride = 100
-    ipf_t_size = 200 * 200;
-
-    CHECKED_MALLOC(pixel_buff, pixel, ipf_t_size);
-    CHECKED_MALLOC(short_buff, int16_t, ipf_t_size);
-    CHECKED_MALLOC(IPF_vec_output_s, int16_t, ipf_t_size);
-    CHECKED_MALLOC(IPF_C_output_s, int16_t, ipf_t_size);
-    CHECKED_MALLOC(IPF_vec_output_p, pixel, ipf_t_size);
-    CHECKED_MALLOC(IPF_C_output_p, pixel, ipf_t_size);
-
-    /* Array of pixel buffers */
-    CHECKED_MALLOC(pixel_test_buff, pixel*, TEST_CASES);
-
-    /* Array of short buffers */
-    CHECKED_MALLOC(short_test_buff, int16_t*, TEST_CASES);
-
-    for (int i = 0; i < TEST_CASES; i++)
-    {
-        CHECKED_MALLOC(pixel_test_buff[i], pixel, ipf_t_size);
-        CHECKED_MALLOC(short_test_buff[i], int16_t, ipf_t_size);
-    }
-
     /* [0] --- Random values
      * [1] --- Minimum
      * [2] --- Maximum */
-    for (int i = 0; i < ipf_t_size; i++)
+    for (int i = 0; i < TEST_BUF_SIZE; i++)
     {
         pixel_test_buff[0][i] = rand() & PIXEL_MAX;
         short_test_buff[0][i] = (rand() % (2 * SMAX)) - SMAX;
@@ -74,40 +47,19 @@
         short_test_buff[2][i] = SMAX;
     }
 
-    memset(IPF_C_output_p, 0xCD, ipf_t_size);
-    memset(IPF_vec_output_p, 0xCD, ipf_t_size);
-    memset(IPF_C_output_s, 0xCD, ipf_t_size * sizeof(int16_t));
-    memset(IPF_vec_output_s, 0xCD, ipf_t_size * sizeof(int16_t));
-
-    for (int i = 0; i < ipf_t_size; i++)                         // Initialize input buffer
-    {
-        int isPositive = rand() & 1;                             // To randomly generate Positive and Negative values
-        isPositive = (isPositive) ? 1 : -1;
-        pixel_buff[i] = (pixel)(rand() &  ((1 << 8) - 1));
-        short_buff[i] = (int16_t)(isPositive) * (rand() &  SHRT_MAX);
-    }
-    return;
-
-fail:
-    exit(1);
-}
+    memset(IPF_C_output_p, 0xCD, TEST_BUF_SIZE * sizeof(pixel));
+    memset(IPF_vec_output_p, 0xCD, TEST_BUF_SIZE * sizeof(pixel));
+    memset(IPF_C_output_s, 0xCD, TEST_BUF_SIZE * sizeof(int16_t));
+    memset(IPF_vec_output_s, 0xCD, TEST_BUF_SIZE * sizeof(int16_t));
 
-IPFilterHarness::~IPFilterHarness()
-{
-    X265_FREE(IPF_vec_output_s);
-    X265_FREE(IPF_C_output_s);
-    X265_FREE(IPF_vec_output_p);
-    X265_FREE(IPF_C_output_p);
-    X265_FREE(short_buff);
-    X265_FREE(pixel_buff);
-    for (int i = 0; i < TEST_CASES; i++)
+    int pixelMax = (1 << X265_DEPTH) - 1;
+    int shortMax = (1 << 15) - 1;
+    for (int i = 0; i < TEST_BUF_SIZE; i++)
     {
-        X265_FREE(pixel_test_buff[i]);
-        X265_FREE(short_test_buff[i]);
+        pixel_buff[i] = (pixel)(rand() & pixelMax);
+        int isPositive = (rand() & 1) ? 1 : -1;
+        short_buff[i] = (int16_t)(isPositive * (rand() & shortMax));
     }
-
-    X265_FREE(pixel_test_buff);
-    X265_FREE(short_test_buff);
 }
 
 bool IPFilterHarness::check_IPFilter_primitive(filter_p2s_t ref, filter_p2s_t opt, int isChroma, int csp)
@@ -125,10 +77,10 @@
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        int rand_height = (int16_t)rand() % 100;                 // Randomly generated Height
-        int rand_width = (int16_t)rand() % 100;                  // Randomly generated Width
+        int rand_height = (int16_t)rand() % 100;
+        int rand_width = (int16_t)rand() % 100;
 
-        rand_srcStride = rand_width + rand() % 100;              // Randomly generated srcStride
+        rand_srcStride = rand_width + rand() % 100;
         if (rand_srcStride < rand_width)
             rand_srcStride = rand_width;
 
@@ -144,14 +96,16 @@
             rand_width,
             rand_height);
 
-        opt(pixel_test_buff[index],
-            rand_srcStride,
-            IPF_vec_output_s,
-            rand_width,
-            rand_height);
+        checked(opt, pixel_test_buff[index],
+                rand_srcStride,
+                IPF_vec_output_s,
+                rand_width,
+                rand_height);
 
-        if (memcmp(IPF_vec_output_s, IPF_C_output_s, ipf_t_size))
+        if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
             return false;
+
+        reportfail();
     }
 
     return true;
@@ -159,7 +113,7 @@
 
 bool IPFilterHarness::check_IPFilterChroma_primitive(filter_pp_t ref, filter_pp_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -167,14 +121,14 @@
 
         for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++)
         {
-            rand_srcStride = rand() % 100 + 2;          // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 32;         // Randomly generated dstStride
+            rand_srcStride = rand() % 100 + 2;
+            rand_dstStride = rand() % 100 + 64;
 
-            opt(pixel_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_p,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, pixel_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_p,
+                    rand_dstStride,
+                    coeffIdx);
 
             ref(pixel_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -182,8 +136,10 @@
                 rand_dstStride,
                 coeffIdx);
 
-            if (memcmp(IPF_vec_output_p, IPF_C_output_p, ipf_t_size))
+            if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -192,7 +148,7 @@
 
 bool IPFilterHarness::check_IPFilterChroma_ps_primitive(filter_ps_t ref, filter_ps_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -200,8 +156,8 @@
 
         for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;              // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 32;         // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(pixel_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -209,14 +165,16 @@
                 rand_dstStride,
                 coeffIdx);
 
-            opt(pixel_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_s,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, pixel_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_s,
+                    rand_dstStride,
+                    coeffIdx);
 
-            if (memcmp(IPF_vec_output_s, IPF_C_output_s, ipf_t_size * sizeof(int16_t)))
+            if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -225,7 +183,7 @@
 
 bool IPFilterHarness::check_IPFilterChroma_hps_primitive(filter_hps_t ref, filter_hps_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -233,10 +191,11 @@
 
         for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++)
         {
-            for (int isRowExt = 0; isRowExt < 2; isRowExt++)  // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7)
+            // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7)
+            for (int isRowExt = 0; isRowExt < 2; isRowExt++)
             {
-                rand_srcStride = rand() % 100 + 2;            // Randomly generated srcStride
-                rand_dstStride = rand() % 100;                // Randomly generated dstStride
+                rand_srcStride = rand() % 100 + 2;
+                rand_dstStride = rand() % 100 + 64;
 
                 ref(pixel_test_buff[index] + 3 * rand_srcStride,
                     rand_srcStride,
@@ -245,15 +204,17 @@
                     coeffIdx,
                     isRowExt);
 
-                opt(pixel_test_buff[index] + 3 * rand_srcStride,
-                    rand_srcStride,
-                    IPF_vec_output_s,
-                    rand_dstStride,
-                    coeffIdx,
-                    isRowExt);
+                checked(opt, pixel_test_buff[index] + 3 * rand_srcStride,
+                        rand_srcStride,
+                        IPF_vec_output_s,
+                        rand_dstStride,
+                        coeffIdx,
+                        isRowExt);
 
-                if (memcmp(IPF_vec_output_s, IPF_C_output_s, ipf_t_size * sizeof(int16_t)))
+                if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                     return false;
+
+                reportfail();
             }
         }
     }
@@ -263,7 +224,7 @@
 
 bool IPFilterHarness::check_IPFilterChroma_sp_primitive(filter_sp_t ref, filter_sp_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -271,8 +232,8 @@
 
         for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;              // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 32;         // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(short_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -280,14 +241,16 @@
                 rand_dstStride,
                 coeffIdx);
 
-            opt(short_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_p,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, short_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_p,
+                    rand_dstStride,
+                    coeffIdx);
 
-            if (memcmp(IPF_vec_output_p, IPF_C_output_p, ipf_t_size))
+            if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -296,7 +259,7 @@
 
 bool IPFilterHarness::check_IPFilterChroma_ss_primitive(filter_ss_t ref, filter_ss_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -304,8 +267,8 @@
 
         for (int coeffIdx = 0; coeffIdx < 8; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;              // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 32;         // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(short_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -313,14 +276,16 @@
                 rand_dstStride,
                 coeffIdx);
 
-            opt(short_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_s,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, short_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_s,
+                    rand_dstStride,
+                    coeffIdx);
 
-            if (memcmp(IPF_C_output_s, IPF_vec_output_s, ipf_t_size * sizeof(int16_t)))
+            if (memcmp(IPF_C_output_s, IPF_vec_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -329,7 +294,7 @@
 
 bool IPFilterHarness::check_IPFilterLuma_primitive(filter_pp_t ref, filter_pp_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -337,22 +302,25 @@
 
         for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;             // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 64;        // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
+
+            checked(opt, pixel_test_buff[index] + 3 * rand_srcStride + 6,
+                    rand_srcStride,
+                    IPF_vec_output_p,
+                    rand_dstStride,
+                    coeffIdx);
 
-            opt(pixel_test_buff[index] + 3 * rand_srcStride + 6,
-                rand_srcStride,
-                IPF_vec_output_p,
-                rand_dstStride,
-                coeffIdx);
             ref(pixel_test_buff[index] + 3 * rand_srcStride + 6,
                 rand_srcStride,
                 IPF_C_output_p,
                 rand_dstStride,
                 coeffIdx);
 
-            if (memcmp(IPF_vec_output_p, IPF_C_output_p, ipf_t_size))
+            if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -361,7 +329,7 @@
 
 bool IPFilterHarness::check_IPFilterLuma_ps_primitive(filter_ps_t ref, filter_ps_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -369,22 +337,25 @@
 
         for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;             // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 64;        // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(pixel_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
                 IPF_C_output_s,
                 rand_dstStride,
                 coeffIdx);
-            opt(pixel_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_s,
-                rand_dstStride,
-                coeffIdx);
 
-            if (memcmp(IPF_vec_output_s, IPF_C_output_s, ipf_t_size * sizeof(int16_t)))
+            checked(opt, pixel_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_s,
+                    rand_dstStride,
+                    coeffIdx);
+
+            if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -393,7 +364,7 @@
 
 bool IPFilterHarness::check_IPFilterLuma_hps_primitive(filter_hps_t ref, filter_hps_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -401,10 +372,11 @@
 
         for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++)
         {
-            for (int isRowExt = 0; isRowExt < 2; isRowExt++)  // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7)
+            // 0 : Interpolate W x H, 1 : Interpolate W x (H + 7)
+            for (int isRowExt = 0; isRowExt < 2; isRowExt++)
             {
-                rand_srcStride = rand() % 100;                // Randomly generated srcStride
-                rand_dstStride = rand() % 100 + 64;           // Randomly generated dstStride
+                rand_srcStride = rand() % 100;
+                rand_dstStride = rand() % 100 + 64;
 
                 ref(pixel_test_buff[index] + 3 * rand_srcStride + 6,
                     rand_srcStride,
@@ -412,15 +384,18 @@
                     rand_dstStride,
                     coeffIdx,
                     isRowExt);
-                opt(pixel_test_buff[index] + 3 * rand_srcStride + 6,
-                    rand_srcStride,
-                    IPF_vec_output_s,
-                    rand_dstStride,
-                    coeffIdx,
-                    isRowExt);
 
-                if (memcmp(IPF_vec_output_s, IPF_C_output_s, ipf_t_size * sizeof(int16_t)))
+                checked(opt, pixel_test_buff[index] + 3 * rand_srcStride + 6,
+                        rand_srcStride,
+                        IPF_vec_output_s,
+                        rand_dstStride,
+                        coeffIdx,
+                        isRowExt);
+
+                if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                     return false;
+
+                reportfail();
             }
         }
     }
@@ -430,7 +405,7 @@
 
 bool IPFilterHarness::check_IPFilterLuma_sp_primitive(filter_sp_t ref, filter_sp_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -438,8 +413,8 @@
 
         for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;             // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 64;        // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(short_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -447,14 +422,16 @@
                 rand_dstStride,
                 coeffIdx);
 
-            opt(short_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_p,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, short_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_p,
+                    rand_dstStride,
+                    coeffIdx);
 
-            if (memcmp(IPF_vec_output_p, IPF_C_output_p, ipf_t_size))
+            if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -463,7 +440,7 @@
 
 bool IPFilterHarness::check_IPFilterLuma_ss_primitive(filter_ss_t ref, filter_ss_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -471,8 +448,8 @@
 
         for (int coeffIdx = 0; coeffIdx < 4; coeffIdx++)
         {
-            rand_srcStride = rand() % 100;             // Randomly generated srcStride
-            rand_dstStride = rand() % 100 + 64;        // Randomly generated dstStride
+            rand_srcStride = rand() % 100;
+            rand_dstStride = rand() % 100 + 64;
 
             ref(short_test_buff[index] + 3 * rand_srcStride,
                 rand_srcStride,
@@ -480,14 +457,16 @@
                 rand_dstStride,
                 coeffIdx);
 
-            opt(short_test_buff[index] + 3 * rand_srcStride,
-                rand_srcStride,
-                IPF_vec_output_s,
-                rand_dstStride,
-                coeffIdx);
+            checked(opt, short_test_buff[index] + 3 * rand_srcStride,
+                    rand_srcStride,
+                    IPF_vec_output_s,
+                    rand_dstStride,
+                    coeffIdx);
 
-            if (memcmp(IPF_C_output_s, IPF_vec_output_s, ipf_t_size * sizeof(int16_t)))
+            if (memcmp(IPF_C_output_s, IPF_vec_output_s, TEST_BUF_SIZE * sizeof(int16_t)))
                 return false;
+
+            reportfail();
         }
     }
 
@@ -496,18 +475,18 @@
 
 bool IPFilterHarness::check_IPFilterLumaHV_primitive(filter_hv_pp_t ref, filter_hv_pp_t opt)
 {
-    int rand_srcStride, rand_dstStride;
+    intptr_t rand_srcStride, rand_dstStride;
 
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
 
-        for (int coeffIdxX = 0; coeffIdxX < 4; coeffIdxX++)    //coeffIdxX in the filter
+        for (int coeffIdxX = 0; coeffIdxX < 4; coeffIdxX++)
         {
-            for (int coeffIdxY = 0; coeffIdxY < 4; coeffIdxY++)    //coeffIdxY in the filter
+            for (int coeffIdxY = 0; coeffIdxY < 4; coeffIdxY++)
             {
-                rand_srcStride = rand() % 100;             // Randomly generated srcStride
-                rand_dstStride = rand() % 100;             // Randomly generated dstStride
+                rand_srcStride = rand() % 100;
+                rand_dstStride = rand() % 100 + 64;
 
                 ref(pixel_test_buff[index] + 3 * rand_srcStride,
                     rand_srcStride,
@@ -515,15 +494,18 @@
                     rand_dstStride,
                     coeffIdxX,
                     coeffIdxY);
-                opt(pixel_test_buff[index] + 3 * rand_srcStride,
-                    rand_srcStride,
-                    IPF_vec_output_p,
-                    rand_dstStride,
-                    coeffIdxX,
-                    coeffIdxY);
 
-                if (memcmp(IPF_vec_output_p, IPF_C_output_p, ipf_t_size))
+                checked(opt, pixel_test_buff[index] + 3 * rand_srcStride,
+                        rand_srcStride,
+                        IPF_vec_output_p,
+                        rand_dstStride,
+                        coeffIdxX,
+                        coeffIdxY);
+
+                if (memcmp(IPF_vec_output_p, IPF_C_output_p, TEST_BUF_SIZE * sizeof(pixel)))
                     return false;
+
+                reportfail();
             }
         }
     }
@@ -535,7 +517,8 @@
 {
     if (opt.luma_p2s)
     {
-        if (!check_IPFilter_primitive(ref.luma_p2s, opt.luma_p2s, 0, 1))   // last parameter does not matter in case of luma
+        // last parameter does not matter in case of luma
+        if (!check_IPFilter_primitive(ref.luma_p2s, opt.luma_p2s, 0, 1))
         {
             printf("luma_p2s failed\n");
             return false;
@@ -618,7 +601,7 @@
             {
                 if (!check_IPFilterChroma_primitive(ref.chroma[csp].filter_hpp[value], opt.chroma[csp].filter_hpp[value]))
                 {
-                    printf("chroma_hpp[%s]", chromaPartStr[value]);
+                    printf("chroma_hpp[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -626,7 +609,7 @@
             {
                 if (!check_IPFilterChroma_hps_primitive(ref.chroma[csp].filter_hps[value], opt.chroma[csp].filter_hps[value]))
                 {
-                    printf("chroma_hps[%s]", chromaPartStr[value]);
+                    printf("chroma_hps[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -634,7 +617,7 @@
             {
                 if (!check_IPFilterChroma_primitive(ref.chroma[csp].filter_vpp[value], opt.chroma[csp].filter_vpp[value]))
                 {
-                    printf("chroma_vpp[%s]", chromaPartStr[value]);
+                    printf("chroma_vpp[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -642,7 +625,7 @@
             {
                 if (!check_IPFilterChroma_ps_primitive(ref.chroma[csp].filter_vps[value], opt.chroma[csp].filter_vps[value]))
                 {
-                    printf("chroma_vps[%s]", chromaPartStr[value]);
+                    printf("chroma_vps[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -650,7 +633,7 @@
             {
                 if (!check_IPFilterChroma_sp_primitive(ref.chroma[csp].filter_vsp[value], opt.chroma[csp].filter_vsp[value]))
                 {
-                    printf("chroma_vsp[%s]", chromaPartStr[value]);
+                    printf("chroma_vsp[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -658,7 +641,7 @@
             {
                 if (!check_IPFilterChroma_ss_primitive(ref.chroma[csp].filter_vss[value], opt.chroma[csp].filter_vss[value]))
                 {
-                    printf("chroma_vss[%s]", chromaPartStr[value]);
+                    printf("chroma_vss[%s]", chromaPartStr[csp][value]);
                     return false;
                 }
             }
@@ -753,40 +736,40 @@
         {
             if (opt.chroma[csp].filter_hpp[value])
             {
-                printf("chroma_hpp[%s]", chromaPartStr[value]);
+                printf("chroma_hpp[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_hpp[value], ref.chroma[csp].filter_hpp[value],
                                pixel_buff + srcStride, srcStride, IPF_vec_output_p, dstStride, 1);
             }
             if (opt.chroma[csp].filter_hps[value])
             {
-                printf("chroma_hps[%s]", chromaPartStr[value]);
+                printf("chroma_hps[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_hps[value], ref.chroma[csp].filter_hps[value],
                                pixel_buff + srcStride, srcStride, IPF_vec_output_s, dstStride, 1, 1);
             }
             if (opt.chroma[csp].filter_vpp[value])
             {
-                printf("chroma_vpp[%s]", chromaPartStr[value]);
+                printf("chroma_vpp[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_vpp[value], ref.chroma[csp].filter_vpp[value],
                                pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride,
                                IPF_vec_output_p, dstStride, 1);
             }
             if (opt.chroma[csp].filter_vps[value])
             {
-                printf("chroma_vps[%s]", chromaPartStr[value]);
+                printf("chroma_vps[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_vps[value], ref.chroma[csp].filter_vps[value],
                                pixel_buff + maxVerticalfilterHalfDistance * srcStride, srcStride,
                                IPF_vec_output_s, dstStride, 1);
             }
             if (opt.chroma[csp].filter_vsp[value])
             {
-                printf("chroma_vsp[%s]", chromaPartStr[value]);
+                printf("chroma_vsp[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_vsp[value], ref.chroma[csp].filter_vsp[value],
                                short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride,
                                IPF_vec_output_p, dstStride, 1);
             }
             if (opt.chroma[csp].filter_vss[value])
             {
-                printf("chroma_vss[%s]", chromaPartStr[value]);
+                printf("chroma_vss[%s]", chromaPartStr[csp][value]);
                 REPORT_SPEEDUP(opt.chroma[csp].filter_vss[value], ref.chroma[csp].filter_vss[value],
                                short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride,
                                IPF_vec_output_s, dstStride, 1);

x265-hg20140324.tar.bz2/source/test/ipfilterharness.h -> x265-hg20140928.tar.bz2/source/test/ipfilterharness.h Changed

x265-hg20140324.tar.bz2/source/test/mbdstharness.cpp -> x265-hg20140928.tar.bz2/source/test/mbdstharness.cpp Changed

@@ -21,23 +21,22 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "mbdstharness.h"
 #include "TLibCommon/TComRom.h"
 #include "common.h"
 
-#define ITERS  100
-#define TEST_CASES 3
 using namespace x265;
-struct DctConf_t
+
+struct DctConf
 {
     const char *name;
     int width;
 };
 
-const DctConf_t DctConf_infos[] =
+const DctConf dctInfo[] =
 {
     { "dst4x4\t",    4 },
     { "dct4x4\t",    4 },
@@ -46,7 +45,7 @@
     { "dct32x32",   32 },
 };
 
-const DctConf_t IDctConf_infos[] =
+const DctConf idctInfo[] =
 {
     { "idst4x4\t",    4 },
     { "idct4x4\t",    4 },
@@ -58,165 +57,82 @@
 MBDstHarness::MBDstHarness()
 {
     const int idct_max = (1 << (BIT_DEPTH + 4)) - 1;
-    CHECKED_MALLOC(mbuf1, int16_t, mb_t_size);
-    CHECKED_MALLOC(mbufdct, int16_t, mb_t_size);
-    CHECKED_MALLOC(mbufidct, int, mb_t_size);
-
-    CHECKED_MALLOC(mbuf2, int16_t, mem_cmp_size);
-    CHECKED_MALLOC(mbuf3, int16_t, mem_cmp_size);
-    CHECKED_MALLOC(mbuf4, int16_t, mem_cmp_size);
-
-    CHECKED_MALLOC(mintbuf1, int, mb_t_size);
-    CHECKED_MALLOC(mintbuf2, int, mb_t_size);
-    CHECKED_MALLOC(mintbuf3, int, mem_cmp_size);
-    CHECKED_MALLOC(mintbuf4, int, mem_cmp_size);
-    CHECKED_MALLOC(mintbuf5, int, mem_cmp_size);
-    CHECKED_MALLOC(mintbuf6, int, mem_cmp_size);
-    CHECKED_MALLOC(mintbuf7, int, mem_cmp_size);
-    CHECKED_MALLOC(mintbuf8, int, mem_cmp_size);
-
-    CHECKED_MALLOC(short_test_buff, int16_t*, TEST_CASES);
-    CHECKED_MALLOC(int_test_buff, int*, TEST_CASES);
-    CHECKED_MALLOC(int_idct_test_buff, int*, TEST_CASES);
-
-    for (int i = 0; i < TEST_CASES; i++)
-    {
-        CHECKED_MALLOC(short_test_buff[i], int16_t, mb_t_size);
-        CHECKED_MALLOC(int_test_buff[i], int, mb_t_size);
-        CHECKED_MALLOC(int_idct_test_buff[i], int, mb_t_size);
-    }
 
     /* [0] --- Random values
      * [1] --- Minimum
      * [2] --- Maximum */
-    for (int i = 0; i < mb_t_size; i++)
+    for (int i = 0; i < TEST_BUF_SIZE; i++)
     {
         short_test_buff[0][i]    = (rand() & PIXEL_MAX) - (rand() & PIXEL_MAX);
         int_test_buff[0][i]      = rand() % PIXEL_MAX;
         int_idct_test_buff[0][i] = (rand() % (SHORT_MAX - SHORT_MIN)) - SHORT_MAX;
+        int_denoise_test_buff1[0][i] = int_denoise_test_buff2[0][i] = (rand() & UNSIGNED_SHORT_MAX) - (rand() & UNSIGNED_SHORT_MAX);
+
         short_test_buff[1][i]    = -PIXEL_MAX;
         int_test_buff[1][i]      = -PIXEL_MAX;
         int_idct_test_buff[1][i] = SHORT_MIN;
+        int_denoise_test_buff1[1][i] = int_denoise_test_buff2[1][i] = -UNSIGNED_SHORT_MAX;
+
         short_test_buff[2][i]    = PIXEL_MAX;
         int_test_buff[2][i]      = PIXEL_MAX;
         int_idct_test_buff[2][i] = SHORT_MAX;
-    }
+        int_denoise_test_buff1[2][i] = int_denoise_test_buff2[2][i] = UNSIGNED_SHORT_MAX;
 
-    for (int i = 0; i < mb_t_size; i++)
-    {
         mbuf1[i] = rand() & PIXEL_MAX;
         mbufdct[i] = (rand() & PIXEL_MAX) - (rand() & PIXEL_MAX);
         mbufidct[i] = (rand() & idct_max);
     }
 
-    for (int i = 0; i < mb_t_size; i++)
-    {
-        mintbuf1[i] = rand() & PIXEL_MAX;
-        mintbuf2[i] = rand() & PIXEL_MAX;
-    }
-
 #if _DEBUG
-    memset(mbuf2, 0, mem_cmp_size);
-    memset(mbuf3, 0, mem_cmp_size);
-    memset(mbuf4, 0, mem_cmp_size);
-    memset(mbufidct, 0, mb_t_size);
-
-    memset(mintbuf3, 0, mem_cmp_size);
-    memset(mintbuf4, 0, mem_cmp_size);
-    memset(mintbuf5, 0, mem_cmp_size);
-    memset(mintbuf6, 0, mem_cmp_size);
-    memset(mintbuf7, 0, mem_cmp_size);
-    memset(mintbuf8, 0, mem_cmp_size);
-#endif // if _DEBUG
-    return;
-
-fail:
-    exit(1);
-}
-
-MBDstHarness::~MBDstHarness()
-{
-    X265_FREE(mbuf1);
-    X265_FREE(mbuf2);
-    X265_FREE(mbuf3);
-    X265_FREE(mbuf4);
-    X265_FREE(mbufdct);
-    X265_FREE(mbufidct);
-
-    X265_FREE(mintbuf1);
-    X265_FREE(mintbuf2);
-    X265_FREE(mintbuf3);
-    X265_FREE(mintbuf4);
-    X265_FREE(mintbuf5);
-    X265_FREE(mintbuf6);
-    X265_FREE(mintbuf7);
-    X265_FREE(mintbuf8);
-    for (int i = 0; i < TEST_CASES; i++)
-    {
-        X265_FREE(short_test_buff[i]);
-        X265_FREE(int_test_buff[i]);
-        X265_FREE(int_idct_test_buff[i]);
-    }
+    memset(mshortbuf2, 0, MAX_TU_SIZE * sizeof(int16_t));
+    memset(mshortbuf3, 0, MAX_TU_SIZE * sizeof(int16_t));
 
-    X265_FREE(short_test_buff);
-    X265_FREE(int_test_buff);
-    X265_FREE(int_idct_test_buff);
+    memset(mintbuf1, 0, MAX_TU_SIZE * sizeof(int));
+    memset(mintbuf2, 0, MAX_TU_SIZE * sizeof(int));
+    memset(mintbuf3, 0, MAX_TU_SIZE * sizeof(int));
+    memset(mintbuf4, 0, MAX_TU_SIZE * sizeof(int));
+#endif // if _DEBUG
 }
 
-bool MBDstHarness::check_dct_primitive(dct_t ref, dct_t opt, int width)
+bool MBDstHarness::check_dct_primitive(dct_t ref, dct_t opt, intptr_t width)
 {
     int j = 0;
-    int cmp_size = sizeof(int) * width * width;
-    for (int i = 0; i <= 100; i++)
+    intptr_t cmp_size = sizeof(int) * width * width;
+
+    for (int i = 0; i < ITERS; i++)
     {
         int index = rand() % TEST_CASES;
+
         ref(short_test_buff[index] + j, mintbuf3, width);
-        opt(short_test_buff[index] + j, mintbuf4, width);
+        checked(opt, short_test_buff[index] + j, mintbuf4, width);
 
         if (memcmp(mintbuf3, mintbuf4, cmp_size))
-        {
-#if _DEBUG
-            // redo for debug
-            ref(short_test_buff[index] + j, mintbuf3, width);
-            opt(short_test_buff[index] + j, mintbuf4, width);
-#endif
-
             return false;
-        }
-        j += 16;
-#if _DEBUG
-        memset(mbuf2, 0xCD, mem_cmp_size);
-        memset(mbuf3, 0xCD, mem_cmp_size);
-#endif
+
+        reportfail();
+        j += INCR;
     }
 
     return true;
 }
 
-bool MBDstHarness::check_idct_primitive(idct_t ref, idct_t opt, int width)
+bool MBDstHarness::check_idct_primitive(idct_t ref, idct_t opt, intptr_t width)
 {
     int j = 0;
-    int cmp_size = sizeof(int16_t) * width * width;
-    for (int i = 0; i <= 100; i++)
+    intptr_t cmp_size = sizeof(int16_t) * width * width;
+
+    for (int i = 0; i < ITERS; i++)
     {
         int index = rand() % TEST_CASES;
-        ref(int_idct_test_buff[index] + j, mbuf2, width);
-        opt(int_idct_test_buff[index] + j, mbuf3, width);
-        if (memcmp(mbuf2, mbuf3, cmp_size))
-        {
-#if _DEBUG
-            // redo for debug
-            ref(int_idct_test_buff[index] + j, mbuf2, width);
-            opt(int_idct_test_buff[index] + j, mbuf3, width);
-#endif
 
+        ref(int_idct_test_buff[index] + j, mshortbuf2, width);
+        checked(opt, int_idct_test_buff[index] + j, mshortbuf3, width);
+
+        if (memcmp(mshortbuf2, mshortbuf3, cmp_size))
             return false;
-        }
-        j += 16;
-#if _DEBUG
-        memset(mbuf2, 0xCD, mem_cmp_size);
-        memset(mbuf3, 0xCD, mem_cmp_size);
-#endif
+
+        reportfail();
+        j += INCR;
     }
 
     return true;
@@ -226,17 +142,14 @@
 {
     int j = 0;
 
-    for (int i = 0; i <= 5; i++)
+    for (int i = 0; i < ITERS; i++)
     {
+        int index = rand() % TEST_CASES;
         int log2TrSize = (rand() % 4) + 2;
 
         int width = (1 << log2TrSize);
         int height = width;
-#if HIGH_BIT_DEPTH
-        int qp = rand() % 64;
-#else
-        int qp = rand() % 51;
-#endif
+        int qp = rand() % (QP_MAX_SPEC + QP_BD_OFFSET + 1);
         int per = qp / 6;
         int rem = qp % 6;
         static const int invQuantScales[6] = { 40, 45, 51, 57, 64, 72 };
@@ -244,20 +157,14 @@
         int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
         int shift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - transformShift;
 
-        int cmp_size = sizeof(int) * height * width;
-        int index = rand() % TEST_CASES;
-
-        ref(int_test_buff[index] + j, mintbuf3, width * height, scale, shift);
-        opt(int_test_buff[index] + j, mintbuf4, width * height, scale, shift);
+        ref(short_test_buff[index] + j, mintbuf3, width * height, scale, shift);
+        checked(opt, short_test_buff[index] + j, mintbuf4, width * height, scale, shift);
 
-        if (memcmp(mintbuf3, mintbuf4, cmp_size))
+        if (memcmp(mintbuf3, mintbuf4, sizeof(int) * height * width))
             return false;
 
-        j += 16;
-#if _DEBUG
-        memset(mintbuf3, 0xCD, mem_cmp_size);
-        memset(mintbuf4, 0xCD, mem_cmp_size);
-#endif
+        reportfail();
+        j += INCR;
     }
 
     return true;
@@ -267,33 +174,29 @@
 {
     int j = 0;
 
-    for (int i = 0; i <= 5; i++)
+    for (int i = 0; i < ITERS; i++)
     {
         int log2TrSize = (rand() % 4) + 2;
 
         int width = (1 << log2TrSize);
         int height = width;
 
-        int qp = rand() % 52;
+        int qp = rand() % (QP_MAX_SPEC + QP_BD_OFFSET + 1);
         int per = qp / 6;
         int transformShift = MAX_TR_DYNAMIC_RANGE - X265_DEPTH - log2TrSize;
         int shift = QUANT_IQUANT_SHIFT - QUANT_SHIFT - transformShift;
 
         int cmp_size = sizeof(int) * height * width;
         int index1 = rand() % TEST_CASES;
-        int index2 = rand() % TEST_CASES;
 
-        ref(int_test_buff[index1] + j, mintbuf3, int_test_buff[index2] + j, width * height, per, shift);
-        opt(int_test_buff[index1] + j, mintbuf4, int_test_buff[index2] + j, width * height, per, shift);
+        ref(short_test_buff[index1] + j, mintbuf3, mintbuf1, width * height, per, shift);
+        checked(opt, short_test_buff[index1] + j, mintbuf4, mintbuf2, width * height, per, shift);
 
-        if (memcmp(mintbuf3, mintbuf4, cmp_size))
+        if (memcmp(mintbuf1, mintbuf2, cmp_size))
             return false;
 
-        j += 16;
-#if _DEBUG
-        memset(mintbuf3, 0xCD, mem_cmp_size);
-        memset(mintbuf4, 0xCD, mem_cmp_size);
-#endif
+        reportfail();
+        j += INCR;
     }
 
     return true;
@@ -303,51 +206,73 @@
 {
     int j = 0;
 
-    for (int i = 0; i <= ITERS; i++)
+    for (int i = 0; i < ITERS; i++)
     {
         int width = (rand() % 4 + 1) * 4;
-
-        if (width == 12)
-        {
-            width = 32;
-        }
         int height = width;
 
         uint32_t optReturnValue = 0;
         uint32_t refReturnValue = 0;
 
-        int bits = rand() % 32;
-        int valueToAdd = rand() % (32 * 1024);
+        int bits = (rand() % 24) + 8;
+        int valueToAdd = rand() % (1 << bits);
         int cmp_size = sizeof(int) * height * width;
+        int cmp_size1 = sizeof(short) * height * width;
         int numCoeff = height * width;
-        int optLastPos = -1, refLastPos = -1;
 
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
 
-        refReturnValue = ref(int_test_buff[index1] + j, int_test_buff[index2] + j, mintbuf5, mintbuf6, bits, valueToAdd, numCoeff, &refLastPos);
-        optReturnValue = opt(int_test_buff[index1] + j, int_test_buff[index2] + j, mintbuf3, mintbuf4, bits, valueToAdd, numCoeff, &optLastPos);
+        refReturnValue = ref(int_test_buff[index1] + j, int_test_buff[index2] + j, mintbuf1, mshortbuf2, bits, valueToAdd, numCoeff);
+        optReturnValue = (uint32_t)checked(opt, int_test_buff[index1] + j, int_test_buff[index2] + j, mintbuf3, mshortbuf3, bits, valueToAdd, numCoeff);
 
-        if (memcmp(mintbuf3, mintbuf5, cmp_size))
+        if (memcmp(mintbuf1, mintbuf3, cmp_size))
             return false;
 
-        if (memcmp(mintbuf4, mintbuf6, cmp_size))
+        if (memcmp(mshortbuf2, mshortbuf3, cmp_size1))
             return false;
 
         if (optReturnValue != refReturnValue)
             return false;
 
-        if (optLastPos != refLastPos)
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool MBDstHarness::check_nquant_primitive(nquant_t ref, nquant_t opt)
+{
+    int j = 0;
+
+    for (int i = 0; i < ITERS; i++)
+    {
+        int width = (rand() % 4 + 1) * 4;
+        int height = width;
+
+        uint32_t optReturnValue = 0;
+        uint32_t refReturnValue = 0;
+
+        int bits = rand() % 32;
+        int valueToAdd = rand() % (1 << bits);
+        int cmp_size = sizeof(short) * height * width;
+        int numCoeff = height * width;
+
+        int index1 = rand() % TEST_CASES;
+        int index2 = rand() % TEST_CASES;
+
+        refReturnValue = ref(int_test_buff[index1] + j, int_test_buff[index2] + j, mshortbuf2, bits, valueToAdd, numCoeff);
+        optReturnValue = (uint32_t)checked(opt, int_test_buff[index1] + j, int_test_buff[index2] + j, mshortbuf3, bits, valueToAdd, numCoeff);
+
+        if (memcmp(mshortbuf2, mshortbuf3, cmp_size))
             return false;
 
-        j += 16;
+        if (optReturnValue != refReturnValue)
+            return false;
 
-#if _DEBUG
-        memset(mintbuf3, 0xCD, mem_cmp_size);
-        memset(mintbuf4, 0xCD, mem_cmp_size);
-        memset(mintbuf5, 0xCD, mem_cmp_size);
-        memset(mintbuf6, 0xCD, mem_cmp_size);
-#endif
+        reportfail();
+        j += INCR;
     }
 
     return true;
@@ -355,7 +280,7 @@
 
 bool MBDstHarness::check_count_nonzero_primitive(count_nonzero_t ref, count_nonzero_t opt)
 {
-    ALIGN_VAR_32(int32_t, qcoeff[32 * 32]);
+    ALIGN_VAR_32(int16_t, qcoeff[32 * 32]);
 
     for (int i = 0; i < 4; i++)
     {
@@ -365,7 +290,7 @@
 
         for (int n = 0; n <= num; n++)
         {
-            memset(qcoeff, 0, num * sizeof(int32_t));
+            memset(qcoeff, 0, num * sizeof(int16_t));
 
             for (int j = 0; j < n; j++)
             {
@@ -375,29 +300,71 @@
                     k = (k + 11) & mask;
                 }
 
-                qcoeff[k] = rand() - RAND_MAX / 2;
+                qcoeff[k] = (int16_t)rand() - RAND_MAX / 2;
             }
 
             int refval = ref(qcoeff, num);
-            int optval = opt(qcoeff, num);
+            int optval = (int)checked(opt, qcoeff, num);
 
             if (refval != optval)
                 return false;
+
+            reportfail();
+        }
+    }
+
+    return true;
+}
+
+bool MBDstHarness::check_denoise_dct_primitive(denoiseDct_t ref, denoiseDct_t opt)
+{
+    int j = 0;
+
+    for (int s = 0; s < 4; s++)
+    {
+        int log2TrSize = s + 2;
+        int num = 1 << (log2TrSize * 2);
+        int cmp_size = sizeof(int) * num;
+
+        for (int i = 0; i < ITERS; i++)
+        {
+            memset(mubuf1, 0, num * sizeof(uint32_t));
+            memset(mubuf2, 0, num * sizeof(uint32_t));
+            memset(mushortbuf1, 0,  num * sizeof(uint16_t));
+
+            for (int k = 0; k < num; k++)
+                mushortbuf1[k] = rand() % UNSIGNED_SHORT_MAX;
+
+            int index = rand() % TEST_CASES;
+
+            ref(int_denoise_test_buff1[index] + j, mubuf1, mushortbuf1, num);
+            checked(opt, int_denoise_test_buff2[index] + j, mubuf2, mushortbuf1, num);
+
+            if (memcmp(int_denoise_test_buff1[index] + j, int_denoise_test_buff2[index] + j, cmp_size))
+                return false;
+
+            if (memcmp(mubuf1, mubuf2, cmp_size))
+                return false;
+
+            reportfail();
+            j += INCR;
         }
+        j = 0;
     }
 
     return true;
 }
 
+
 bool MBDstHarness::testCorrectness(const EncoderPrimitives& ref, const EncoderPrimitives& opt)
 {
     for (int i = 0; i < NUM_DCTS; i++)
     {
         if (opt.dct[i])
         {
-            if (!check_dct_primitive(ref.dct[i], opt.dct[i], DctConf_infos[i].width))
+            if (!check_dct_primitive(ref.dct[i], opt.dct[i], dctInfo[i].width))
             {
-                printf("\n%s failed\n", DctConf_infos[i].name);
+                printf("\n%s failed\n", dctInfo[i].name);
                 return false;
             }
         }
@@ -407,9 +374,9 @@
     {
         if (opt.idct[i])
         {
-            if (!check_idct_primitive(ref.idct[i], opt.idct[i], IDctConf_infos[i].width))
+            if (!check_idct_primitive(ref.idct[i], opt.idct[i], idctInfo[i].width))
             {
-                printf("%s failed\n", IDctConf_infos[i].name);
+                printf("%s failed\n", idctInfo[i].name);
                 return false;
             }
         }
@@ -424,6 +391,15 @@
         }
     }
 
+    if (opt.dequant_scaling)
+    {
+        if (!check_dequant_primitive(ref.dequant_scaling, opt.dequant_scaling))
+        {
+            printf("dequant_scaling: Failed!\n");
+            return false;
+        }
+    }
+
     if (opt.quant)
     {
         if (!check_quant_primitive(ref.quant, opt.quant))
@@ -433,6 +409,15 @@
         }
     }
 
+    if (opt.nquant)
+    {
+        if (!check_nquant_primitive(ref.nquant, opt.nquant))
+        {
+            printf("nquant: Failed!\n");
+            return false;
+        }
+    }
+
     if (opt.count_nonzero)
     {
         if (!check_count_nonzero_primitive(ref.count_nonzero, opt.count_nonzero))
@@ -442,6 +427,24 @@
         }
     }
 
+    if (opt.dequant_scaling)
+    {
+        if (!check_dequant_primitive(ref.dequant_scaling, opt.dequant_scaling))
+        {
+            printf("dequant_scaling: Failed!\n");
+            return false;
+        }
+    }
+
+    if (opt.denoiseDct)
+    {
+        if (!check_denoise_dct_primitive(ref.denoiseDct, opt.denoiseDct))
+        {
+            printf("denoiseDct: Failed!\n");
+            return false;
+        }
+    }
+
     return true;
 }
 
@@ -451,8 +454,8 @@
     {
         if (opt.dct[value])
         {
-            printf("%s\t", DctConf_infos[value].name);
-            REPORT_SPEEDUP(opt.dct[value], ref.dct[value], mbuf1, mintbuf3, DctConf_infos[value].width);
+            printf("%s\t", dctInfo[value].name);
+            REPORT_SPEEDUP(opt.dct[value], ref.dct[value], mbuf1, mintbuf3, dctInfo[value].width);
         }
     }
 
@@ -460,28 +463,33 @@
     {
         if (opt.idct[value])
         {
-            printf("%s\t", IDctConf_infos[value].name);
-            REPORT_SPEEDUP(opt.idct[value], ref.idct[value], mbufidct, mbuf2, IDctConf_infos[value].width);
+            printf("%s\t", idctInfo[value].name);
+            REPORT_SPEEDUP(opt.idct[value], ref.idct[value], mbufidct, mshortbuf2, idctInfo[value].width);
         }
     }
 
     if (opt.dequant_normal)
     {
         printf("dequant_normal\t");
-        REPORT_SPEEDUP(opt.dequant_normal, ref.dequant_normal, mintbuf1, mintbuf3, 32 * 32, 70, 1);
+        REPORT_SPEEDUP(opt.dequant_normal, ref.dequant_normal, short_test_buff[0], mintbuf3, 32 * 32, 70, 1);
     }
 
     if (opt.dequant_scaling)
     {
         printf("dequant_scaling\t");
-        REPORT_SPEEDUP(opt.dequant_scaling, ref.dequant_scaling, mintbuf1, mintbuf3, mintbuf2, 32 * 32, 5, 1);
+        REPORT_SPEEDUP(opt.dequant_scaling, ref.dequant_scaling, short_test_buff[0], mintbuf3, mintbuf4, 32 * 32, 5, 1);
     }
 
     if (opt.quant)
     {
         printf("quant\t\t");
-        int dummy = -1;
-        REPORT_SPEEDUP(opt.quant, ref.quant, mintbuf1, mintbuf2, mintbuf3, mintbuf4, 23, 23785, 32 * 32, &dummy);
+        REPORT_SPEEDUP(opt.quant, ref.quant, int_test_buff[0], int_test_buff[1], mintbuf3, mshortbuf2, 23, 23785, 32 * 32);
+    }
+
+    if (opt.nquant)
+    {
+        printf("nquant\t\t");
+        REPORT_SPEEDUP(opt.nquant, ref.nquant, int_test_buff[0], int_test_buff[1], mshortbuf2, 23, 23785, 32 * 32);
     }
 
     if (opt.count_nonzero)
@@ -489,7 +497,14 @@
         for (int i = 4; i <= 32; i <<= 1)
         {
             printf("count_nonzero[%dx%d]", i, i);
-            REPORT_SPEEDUP(opt.count_nonzero, ref.count_nonzero, mbufidct, i * i)
+            REPORT_SPEEDUP(opt.count_nonzero, ref.count_nonzero, mbuf1, i * i)
         }
     }
+
+    if (opt.denoiseDct)
+    {
+        printf("denoiseDct\t");
+        REPORT_SPEEDUP(opt.denoiseDct, ref.denoiseDct, int_denoise_test_buff1[0], mubuf1, mushortbuf1, 32 * 32);
+    }
+
 }

x265-hg20140324.tar.bz2/source/test/mbdstharness.h -> x265-hg20140928.tar.bz2/source/test/mbdstharness.h Changed

x265-hg20140324.tar.bz2/source/test/pixelharness.cpp -> x265-hg20140928.tar.bz2/source/test/pixelharness.cpp Changed

@@ -18,7 +18,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "pixelharness.h"
@@ -26,49 +26,8 @@
 
 using namespace x265;
 
-#define INCR   32
-#define STRIDE 64
-#define ITERS  100
-#define MAX_HEIGHT 64
-#define PAD_ROWS   64
-#define BUFFSIZE STRIDE * (MAX_HEIGHT + PAD_ROWS) + INCR * ITERS
-#define TEST_CASES 3
-#define SMAX (1 << 12)
-#define SMIN (-1 << 12)
-
 PixelHarness::PixelHarness()
 {
-    int bufsize = STRIDE * (MAX_HEIGHT + PAD_ROWS) + INCR * ITERS;
-
-    /* 64 pixels wide, 2k deep */
-    CHECKED_MALLOC(pbuf1, pixel, bufsize);
-    CHECKED_MALLOC(pbuf2, pixel, bufsize);
-    CHECKED_MALLOC(pbuf3, pixel, bufsize);
-    CHECKED_MALLOC(pbuf4, pixel, bufsize);
-
-    CHECKED_MALLOC(ibuf1, int, bufsize);
-    CHECKED_MALLOC(psbuf1, int8_t, bufsize);
-
-    CHECKED_MALLOC(sbuf1, int16_t, bufsize);
-    CHECKED_MALLOC(sbuf2, int16_t, bufsize);
-    CHECKED_MALLOC(sbuf3, int16_t, bufsize);
-
-    /* Test Case buffer array */
-    CHECKED_MALLOC(pixel_test_buff, pixel*, TEST_CASES);
-    CHECKED_MALLOC(short_test_buff, int16_t*, TEST_CASES);
-    CHECKED_MALLOC(short_test_buff1, int16_t*, TEST_CASES);
-    CHECKED_MALLOC(short_test_buff2, int16_t*, TEST_CASES);
-    CHECKED_MALLOC(int_test_buff, int*, TEST_CASES);
-
-    for (int i = 0; i < TEST_CASES; i++)
-    {
-        CHECKED_MALLOC(pixel_test_buff[i], pixel, BUFFSIZE);
-        CHECKED_MALLOC(short_test_buff[i], int16_t, BUFFSIZE);
-        CHECKED_MALLOC(short_test_buff1[i], int16_t, BUFFSIZE);
-        CHECKED_MALLOC(short_test_buff2[i], int16_t, BUFFSIZE);
-        CHECKED_MALLOC(int_test_buff[i], int, BUFFSIZE);
-    }
-
     /* [0] --- Random values
      * [1] --- Minimum
      * [2] --- Maximum */
@@ -79,71 +38,53 @@
         short_test_buff1[0][i]  = rand() & PIXEL_MAX;                   // For block copy only
         short_test_buff2[0][i]  = rand() % 16383;                       // for addAvg
         int_test_buff[0][i]     = rand() % SHORT_MAX;
+        ushort_test_buff[0][i]  = rand() % ((1 << 16) - 1);
+        uchar_test_buff[0][i]   = rand() % ((1 << 8) - 1);
+
         pixel_test_buff[1][i]   = PIXEL_MIN;
         short_test_buff[1][i]   = SMIN;
         short_test_buff1[1][i]  = PIXEL_MIN;
         short_test_buff2[1][i]  = -16384;
         int_test_buff[1][i]     = SHORT_MIN;
+        ushort_test_buff[1][i]  = PIXEL_MIN;
+        uchar_test_buff[1][i]   = PIXEL_MIN;
+
         pixel_test_buff[2][i]   = PIXEL_MAX;
         short_test_buff[2][i]   = SMAX;
         short_test_buff1[2][i]  = PIXEL_MAX;
         short_test_buff2[2][i]  = 16383;
         int_test_buff[2][i]     = SHORT_MAX;
-    }
-    for (int i = 0; i < bufsize; i++)
-    {
+        ushort_test_buff[2][i]  = ((1 << 16) - 1);
+        uchar_test_buff[2][i]   = 255;
+
         pbuf1[i] = rand() & PIXEL_MAX;
         pbuf2[i] = rand() & PIXEL_MAX;
         pbuf3[i] = rand() & PIXEL_MAX;
         pbuf4[i] = rand() & PIXEL_MAX;
+
         sbuf1[i] = (rand() % (2 * SMAX + 1)) - SMAX - 1; //max(SHORT_MIN, min(rand(), SMAX));
         sbuf2[i] = (rand() % (2 * SMAX + 1)) - SMAX - 1; //max(SHORT_MIN, min(rand(), SMAX));
         ibuf1[i] = (rand() % (2 * SMAX + 1)) - SMAX - 1;
-        psbuf1[i] = (rand() %65) - 32;                   // range is between -32 to 32
+        psbuf1[i] = (rand() % 65) - 32;                   // range is between -32 to 32
         sbuf3[i] = rand() % PIXEL_MAX; // for blockcopy only
     }
-    return;
-
-fail:
-    exit(1);
-}
-
-PixelHarness::~PixelHarness()
-{
-    X265_FREE(pbuf1);
-    X265_FREE(pbuf2);
-    X265_FREE(pbuf3);
-    X265_FREE(pbuf4);
-    X265_FREE(sbuf1);
-    X265_FREE(sbuf2);
-    X265_FREE(sbuf3);
-    for (int i = 0; i < TEST_CASES; i++)
-    {
-        X265_FREE(pixel_test_buff[i]);
-        X265_FREE(short_test_buff[i]);
-        X265_FREE(short_test_buff1[i]);
-        X265_FREE(int_test_buff[i]);
-    }
-
-    X265_FREE(pixel_test_buff);
-    X265_FREE(short_test_buff);
-    X265_FREE(short_test_buff1);
-    X265_FREE(int_test_buff);
 }
 
 bool PixelHarness::check_pixelcmp(pixelcmp_t ref, pixelcmp_t opt)
 {
     int j = 0;
+    intptr_t stride = STRIDE;
 
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        int vres = opt(pixel_test_buff[index1], STRIDE, pixel_test_buff[index2] + j, STRIDE);
-        int cres = ref(pixel_test_buff[index1], STRIDE, pixel_test_buff[index2] + j, STRIDE);
+        int vres = (int)checked(opt, pixel_test_buff[index1], stride, pixel_test_buff[index2] + j, stride);
+        int cres = ref(pixel_test_buff[index1], stride, pixel_test_buff[index2] + j, stride);
         if (vres != cres)
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -153,16 +94,18 @@
 bool PixelHarness::check_pixelcmp_sp(pixelcmp_sp_t ref, pixelcmp_sp_t opt)
 {
     int j = 0;
+    intptr_t stride = STRIDE;
 
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        int vres = opt(short_test_buff[index1], STRIDE, pixel_test_buff[index2] + j, STRIDE);
-        int cres = ref(short_test_buff[index1], STRIDE, pixel_test_buff[index2] + j, STRIDE);
+        int vres = (int)checked(opt, short_test_buff[index1], stride, pixel_test_buff[index2] + j, stride);
+        int cres = ref(short_test_buff[index1], stride, pixel_test_buff[index2] + j, stride);
         if (vres != cres)
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -172,16 +115,18 @@
 bool PixelHarness::check_pixelcmp_ss(pixelcmp_ss_t ref, pixelcmp_ss_t opt)
 {
     int j = 0;
+    intptr_t stride = STRIDE;
 
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        int vres = opt(short_test_buff[index1], STRIDE, short_test_buff[index2] + j, STRIDE);
-        int cres = ref(short_test_buff[index1], STRIDE, short_test_buff[index2] + j, STRIDE);
+        int vres = (int)checked(opt, short_test_buff[index1], stride, short_test_buff[index2] + j, stride);
+        int cres = ref(short_test_buff[index1], stride, short_test_buff[index2] + j, stride);
         if (vres != cres)
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -193,21 +138,23 @@
     ALIGN_VAR_16(int, cres[16]);
     ALIGN_VAR_16(int, vres[16]);
     int j = 0;
+    intptr_t stride = FENC_STRIDE - 5;
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        opt(pixel_test_buff[index1],
-            pixel_test_buff[index2] + j,
-            pixel_test_buff[index2] + j + 1,
-            pixel_test_buff[index2] + j + 2, FENC_STRIDE - 5, &vres[0]);
+        checked(opt, pixel_test_buff[index1],
+                pixel_test_buff[index2] + j,
+                pixel_test_buff[index2] + j + 1,
+                pixel_test_buff[index2] + j + 2, stride, &vres[0]);
         ref(pixel_test_buff[index1],
             pixel_test_buff[index2] + j,
             pixel_test_buff[index2] + j + 1,
-            pixel_test_buff[index2] + j + 2, FENC_STRIDE - 5, &cres[0]);
+            pixel_test_buff[index2] + j + 2, stride, &cres[0]);
         if ((vres[0] != cres[0]) || ((vres[1] != cres[1])) || ((vres[2] != cres[2])))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -219,78 +166,32 @@
     ALIGN_VAR_16(int, cres[16]);
     ALIGN_VAR_16(int, vres[16]);
     int j = 0;
+    intptr_t stride = FENC_STRIDE - 5;
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        opt(pixel_test_buff[index1],
-            pixel_test_buff[index2] + j,
-            pixel_test_buff[index2] + j + 1,
-            pixel_test_buff[index2] + j + 2,
-            pixel_test_buff[index2] + j + 3, FENC_STRIDE - 5, &vres[0]);
+        checked(opt, pixel_test_buff[index1],
+                pixel_test_buff[index2] + j,
+                pixel_test_buff[index2] + j + 1,
+                pixel_test_buff[index2] + j + 2,
+                pixel_test_buff[index2] + j + 3, stride, &vres[0]);
         ref(pixel_test_buff[index1],
             pixel_test_buff[index2] + j,
             pixel_test_buff[index2] + j + 1,
             pixel_test_buff[index2] + j + 2,
-            pixel_test_buff[index2] + j + 3, FENC_STRIDE - 5, &cres[0]);
+            pixel_test_buff[index2] + j + 3, stride, &cres[0]);
 
         if ((vres[0] != cres[0]) || ((vres[1] != cres[1])) || ((vres[2] != cres[2])) || ((vres[3] != cres[3])))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
     return true;
 }
 
-bool PixelHarness::check_blockcopy_pp(blockcpy_pp_t ref, blockcpy_pp_t opt)
-{
-    ALIGN_VAR_16(pixel, ref_dest[64 * 64]);
-    ALIGN_VAR_16(pixel, opt_dest[64 * 64]);
-    int bx = 64;
-    int by = 64;
-    int j = 0;
-    for (int i = 0; i < ITERS; i++)
-    {
-        int index = i % TEST_CASES;
-        opt(bx, by, opt_dest, 64, pixel_test_buff[index] + j, 128);
-        ref(bx, by, ref_dest, 64, pixel_test_buff[index] + j, 128);
-
-        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
-            return false;
-
-        j += 4;
-        bx = 4 * ((rand() & 15) + 1);
-        by = 4 * ((rand() & 15) + 1);
-    }
-
-    return true;
-}
-
-bool PixelHarness::check_blockcopy_ps(blockcpy_ps_t ref, blockcpy_ps_t opt)
-{
-    ALIGN_VAR_16(pixel, ref_dest[64 * 64]);
-    ALIGN_VAR_16(pixel, opt_dest[64 * 64]);
-    int bx = 64;
-    int by = 64;
-    int j = 0;
-    for (int i = 0; i < ITERS; i++)
-    {
-        int index = i % TEST_CASES;
-        opt(bx, by, opt_dest, 64, short_test_buff1[index] + j, STRIDE);
-        ref(bx, by, ref_dest, 64, short_test_buff1[index] + j, STRIDE);
-
-        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
-            return false;
-
-        j += 4;
-        bx = 4 * ((rand() & 15) + 1);
-        by = 4 * ((rand() & 15) + 1);
-    }
-
-    return true;
-}
-
 bool PixelHarness::check_calresidual(calcresidual_t ref, calcresidual_t opt)
 {
     ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
@@ -299,15 +200,17 @@
     memset(opt_dest, 0, 64 * 64 * sizeof(int16_t));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(pbuf1 + j, pixel_test_buff[index] + j, opt_dest, STRIDE);
-        ref(pbuf1 + j, pixel_test_buff[index] + j, ref_dest, STRIDE);
+        checked(opt, pbuf1 + j, pixel_test_buff[index] + j, opt_dest, stride);
+        ref(pbuf1 + j, pixel_test_buff[index] + j, ref_dest, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -316,48 +219,66 @@
 
 bool PixelHarness::check_calcrecon(calcrecon_t ref, calcrecon_t opt)
 {
-    ALIGN_VAR_16(int16_t, ref_recq[64 * 64]);
-    ALIGN_VAR_16(int16_t, opt_recq[64 * 64]);
+    ALIGN_VAR_16(int16_t, ref_recq[64 * 64 * 2]);
+    ALIGN_VAR_16(int16_t, opt_recq[64 * 64 * 2]);
 
-    ALIGN_VAR_16(pixel, ref_reco[64 * 64]);
-    ALIGN_VAR_16(pixel, opt_reco[64 * 64]);
+    ALIGN_VAR_16(pixel, ref_reco[64 * 64 * 2]);
+    ALIGN_VAR_16(pixel, opt_reco[64 * 64 * 2]);
 
-    ALIGN_VAR_16(pixel, ref_pred[64 * 64]);
-    ALIGN_VAR_16(pixel, opt_pred[64 * 64]);
+    ALIGN_VAR_16(pixel, ref_pred[64 * 64 * 2]);
+    ALIGN_VAR_16(pixel, opt_pred[64 * 64 * 2]);
 
-    memset(ref_recq, 0, 64 * 64 * sizeof(int16_t));
-    memset(opt_recq, 0, 64 * 64 * sizeof(int16_t));
-    memset(ref_reco, 0, 64 * 64 * sizeof(pixel));
-    memset(opt_reco, 0, 64 * 64 * sizeof(pixel));
-    memset(ref_pred, 0, 64 * 64 * sizeof(pixel));
-    memset(opt_pred, 0, 64 * 64 * sizeof(pixel));
+    memset(ref_recq, 0xCD, 64 * 64 * 2 * sizeof(int16_t));
+    memset(opt_recq, 0xCD, 64 * 64 * 2 * sizeof(int16_t));
+    memset(ref_reco, 0xCD, 64 * 64 * 2 * sizeof(pixel));
+    memset(opt_reco, 0xCD, 64 * 64 * 2 * sizeof(pixel));
+    memset(ref_pred, 0xCD, 64 * 64 * 2 * sizeof(pixel));
+    memset(opt_pred, 0xCD, 64 * 64 * 2 * sizeof(pixel));
 
     int j = 0;
     for (int i = 0; i < ITERS; i++)
     {
-        int stride = STRIDE;
+        // NOTE: stride must be multiple of 16, because minimum block is 4x4
+        int stride0 = (STRIDE + (rand() % STRIDE)) & ~15;
+        int stride1 = (STRIDE + (rand() % STRIDE)) & ~15;
+        int stride2 = (STRIDE + (rand() % STRIDE)) & ~15;
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        ref(pixel_test_buff[index1] + j, short_test_buff[index2] + j,
-            ref_reco, ref_recq, ref_pred, stride, stride, stride);
-        opt(pixel_test_buff[index1] + j, short_test_buff[index2] + j,
-            opt_reco, opt_recq, opt_pred, stride, stride, stride);
+        ref(pixel_test_buff[index1] + j, short_test_buff[index2] + j, ref_recq, ref_pred, stride0, stride1, stride2);
+        checked(opt, pixel_test_buff[index1] + j, short_test_buff[index2] + j, opt_recq, opt_pred, stride0, stride1, stride2);
 
-        if (memcmp(ref_recq, opt_recq, 64 * 64 * sizeof(int16_t)))
+        if (memcmp(ref_recq, opt_recq, 64 * stride0 * sizeof(int16_t)))
         {
             return false;
         }
-#if !NEW_CALCRECON
-        if (memcmp(ref_reco, opt_reco, 64 * 64 * sizeof(pixel)))
+        if (memcmp(ref_pred, opt_pred, 64 * stride2 * sizeof(pixel)))
         {
             return false;
         }
-#endif
-        if (memcmp(ref_pred, opt_pred, 64 * 64 * sizeof(pixel)))
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_ssd_s(pixel_ssd_s_t ref, pixel_ssd_s_t opt)
+{
+    int j = 0;
+    for (int i = 0; i < ITERS; i++)
+    {
+        // NOTE: stride must be multiple of 16, because minimum block is 4x4
+        int stride = (STRIDE + (rand() % STRIDE)) & ~15;
+        int cres = ref(sbuf1 + j, stride);
+        int vres = (int)checked(opt, sbuf1 + j, (intptr_t)stride);
+
+        if (cres != vres)
         {
             return false;
         }
 
+        reportfail();
         j += INCR;
     }
 
@@ -378,15 +299,17 @@
     int shift = rand() % 15;
     int round = shift ? (1 << (shift - 1)) : 0;
     int offset = (rand() % 256) - 128;
+    intptr_t stride = 64;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(short_test_buff[index] + j, opt_dest, 64, 64, width, height, w0, round, shift, offset);
-        ref(short_test_buff[index] + j, ref_dest, 64, 64, width, height, w0, round, shift, offset);
+        checked(opt, short_test_buff[index] + j, opt_dest, stride, stride, width, height, w0, round, shift, offset);
+        ref(short_test_buff[index] + j, ref_dest, stride, stride, width, height, w0, round, shift, offset);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -407,48 +330,23 @@
     int shift = rand() % 15;
     int round = shift ? (1 << (shift - 1)) : 0;
     int offset = (rand() % 256) - 128;
+    intptr_t stride = 64;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(pixel_test_buff[index] + j, opt_dest, 64, 64, width, height, w0, round, shift, offset);
-        ref(pixel_test_buff[index] + j, ref_dest, 64, 64, width, height, w0, round, shift, offset);
+        checked(opt, pixel_test_buff[index] + j, opt_dest, stride, stride, width, height, w0, round, shift, offset);
+        ref(pixel_test_buff[index] + j, ref_dest, stride, stride, width, height, w0, round, shift, offset);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
     return true;
 }
 
-bool PixelHarness::check_pixeladd_ss(pixeladd_ss_t ref, pixeladd_ss_t opt)
-{
-    ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
-    ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
-    int bx = 64;
-    int by = 64;
-    int j = 0;
-    for (int i = 0; i < ITERS; i++)
-    {
-        int index1 = rand() % TEST_CASES;
-        int index2 = rand() % TEST_CASES;
-        opt(bx, by, opt_dest, STRIDE, short_test_buff[index1] + j,
-            short_test_buff[index2] + j, STRIDE, STRIDE);
-        ref(bx, by, ref_dest, STRIDE, short_test_buff[index1] + j,
-            short_test_buff[index2] + j, STRIDE, STRIDE);
-
-        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
-            return false;
-
-        j += INCR;
-        bx = 4 * ((rand() & 15) + 1);
-        by = 4 * ((rand() & 15) + 1);
-    }
-
-    return true;
-}
-
 bool PixelHarness::check_downscale_t(downscale_t ref, downscale_t opt)
 {
     ALIGN_VAR_16(pixel, ref_destf[32 * 32]);
@@ -473,8 +371,8 @@
         int index = i % TEST_CASES;
         ref(pixel_test_buff[index] + j, ref_destf, ref_desth, ref_destv,
             ref_destc, src_stride, dst_stride, bx, by);
-        opt(pixel_test_buff[index] + j, opt_destf, opt_desth, opt_destv,
-            opt_destc, src_stride, dst_stride, bx, by);
+        checked(opt, pixel_test_buff[index] + j, opt_destf, opt_desth, opt_destv,
+                opt_destc, src_stride, dst_stride, bx, by);
 
         if (memcmp(ref_destf, opt_destf, 32 * 32 * sizeof(pixel)))
             return false;
@@ -485,6 +383,7 @@
         if (memcmp(ref_destc, opt_destc, 32 * 32 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -496,18 +395,23 @@
     ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
     ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
 
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int shift = (rand() % 7 + 1);
 
         int index = i % TEST_CASES;
-        opt(opt_dest, int_test_buff[index] + j, STRIDE, shift, STRIDE);
-        ref(ref_dest, int_test_buff[index] + j, STRIDE, shift, STRIDE);
+        checked(opt, opt_dest, int_test_buff[index] + j, stride, shift, (int)STRIDE);
+        ref(ref_dest, int_test_buff[index] + j, stride, shift, (int)STRIDE);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -520,17 +424,157 @@
     ALIGN_VAR_16(int32_t, opt_dest[64 * 64]);
 
     int j = 0;
+    intptr_t stride = STRIDE;
+    for (int i = 0; i < ITERS; i++)
+    {
+        int shift = (rand() % 7 + 1);
+
+        int index = i % TEST_CASES;
+        checked(opt, opt_dest, short_test_buff[index] + j, stride, shift, (int)stride);
+        ref(ref_dest, short_test_buff[index] + j, stride, shift, (int)stride);
+
+        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int32_t)))
+            return false;
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_cvt16to32_shr_t(cvt16to32_shr_t ref, cvt16to32_shr_t opt)
+{
+    ALIGN_VAR_16(int32_t, ref_dest[64 * 64]);
+    ALIGN_VAR_16(int32_t, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int shift = (rand() % 7 + 1);
 
         int index = i % TEST_CASES;
-        opt(opt_dest, short_test_buff[index] + j, STRIDE, shift, STRIDE);
-        ref(ref_dest, short_test_buff[index] + j, STRIDE, shift, STRIDE);
+        checked(opt, opt_dest, short_test_buff[index] + j, stride, shift, (int)stride);
+        ref(ref_dest, short_test_buff[index] + j, stride, shift, (int)stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int32_t)))
             return false;
 
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_cvt32to16_shl_t(cvt32to16_shl_t ref, cvt32to16_shl_t opt)
+{
+    ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
+    ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int j = 0;
+    intptr_t stride = STRIDE;
+    for (int i = 0; i < ITERS; i++)
+    {
+        int shift = (rand() % 7 + 1);
+
+        int index = i % TEST_CASES;
+        checked(opt, opt_dest, int_test_buff[index] + j, stride, shift);
+        ref(ref_dest, int_test_buff[index] + j, stride, shift);
+
+        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
+            return false;
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_copy_cnt_t(copy_cnt_t ref, copy_cnt_t opt)
+{
+    ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
+    ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int j = 0;
+    intptr_t stride = STRIDE;
+    for (int i = 0; i < ITERS; i++)
+    {
+        int index = i % TEST_CASES;
+        int opt_cnt = (int)checked(opt, opt_dest, short_test_buff1[index] + j, stride);
+        int ref_cnt = ref(ref_dest, short_test_buff1[index] + j, stride);
+
+        if ((ref_cnt != opt_cnt) || memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
+            return false;
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_copy_shr_t(copy_shr_t ref, copy_shr_t opt)
+{
+    ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
+    ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int j = 0;
+    intptr_t stride = STRIDE;
+    for (int i = 0; i < ITERS; i++)
+    {
+        int shift = (rand() % 7 + 1);
+
+        int index = i % TEST_CASES;
+        checked(opt, opt_dest, short_test_buff[index] + j, stride, shift, (int)STRIDE);
+        ref(ref_dest, short_test_buff[index] + j, stride, shift, (int)STRIDE);
+
+        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
+            return false;
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_copy_shl_t(copy_shl_t ref, copy_shl_t opt)
+{
+    ALIGN_VAR_16(int16_t, ref_dest[64 * 64]);
+    ALIGN_VAR_16(int16_t, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int j = 0;
+    intptr_t stride = STRIDE;
+    for (int i = 0; i < ITERS; i++)
+    {
+        int shift = (rand() % 7 + 1);
+
+        int index = i % TEST_CASES;
+        checked(opt, opt_dest, short_test_buff[index] + j, stride, shift);
+        ref(ref_dest, short_test_buff[index] + j, stride, shift);
+
+        if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
+            return false;
+
+        reportfail();
         j += INCR;
     }
 
@@ -547,18 +591,20 @@
     memset(ref_dest, 0xCD, sizeof(ref_dest));
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        ref(ref_dest, STRIDE, pixel_test_buff[index1] + j,
-            STRIDE, pixel_test_buff[index2] + j, STRIDE, 32);
-        opt(opt_dest, STRIDE, pixel_test_buff[index1] + j,
-            STRIDE, pixel_test_buff[index2] + j, STRIDE, 32);
+        checked(ref, ref_dest, stride, pixel_test_buff[index1] + j,
+                stride, pixel_test_buff[index2] + j, stride, 32);
+        opt(opt_dest, stride, pixel_test_buff[index1] + j,
+            stride, pixel_test_buff[index2] + j, stride, 32);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -576,15 +622,17 @@
     memset(opt_dest, 0, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, STRIDE, pixel_test_buff[index] + j, STRIDE);
-        ref(ref_dest, STRIDE, pixel_test_buff[index] + j, STRIDE);
+        checked(opt, opt_dest, stride, pixel_test_buff[index] + j, stride);
+        ref(ref_dest, stride, pixel_test_buff[index] + j, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -602,15 +650,17 @@
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride1 = 64, stride2 = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, 64, short_test_buff1[index] + j, STRIDE);
-        ref(ref_dest, 64, short_test_buff1[index] + j, STRIDE);
+        checked(opt, opt_dest, stride1, short_test_buff1[index] + j, stride2);
+        ref(ref_dest, stride1, short_test_buff1[index] + j, stride2);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -628,15 +678,17 @@
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, STRIDE, pixel_test_buff[index] + j, STRIDE);
-        ref(ref_dest, STRIDE, pixel_test_buff[index] + j, STRIDE);
+        checked(opt, opt_dest, stride, pixel_test_buff[index] + j, stride);
+        ref(ref_dest, stride, pixel_test_buff[index] + j, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -654,15 +706,17 @@
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, STRIDE, short_test_buff1[index] + j, STRIDE);
-        ref(ref_dest, STRIDE, short_test_buff1[index] + j, STRIDE);
+        checked(opt, opt_dest, stride, short_test_buff1[index] + j, stride);
+        ref(ref_dest, stride, short_test_buff1[index] + j, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -677,15 +731,18 @@
     memset(ref_dest, 0xCD, sizeof(ref_dest));
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
+    intptr_t stride = 64;
     for (int i = 0; i < ITERS; i++)
     {
         int16_t value = (rand() % SHORT_MAX) + 1;
 
-        opt(opt_dest, 64, value);
-        ref(ref_dest, 64, value);
+        checked(opt, opt_dest, stride, value);
+        ref(ref_dest, stride, value);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
+
+        reportfail();
     }
 
     return true;
@@ -700,18 +757,20 @@
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride2 = 64, stride = STRIDE;
     for (int i = 0; i < 1; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        opt(opt_dest, 64, pixel_test_buff[index1] + j,
-            pixel_test_buff[index2] + j, STRIDE, STRIDE);
-        ref(ref_dest, 64, pixel_test_buff[index1] + j,
-            pixel_test_buff[index2] + j, STRIDE, STRIDE);
+        checked(opt, opt_dest, stride2, pixel_test_buff[index1] + j,
+                pixel_test_buff[index2] + j, stride, stride);
+        ref(ref_dest, stride2, pixel_test_buff[index1] + j,
+            pixel_test_buff[index2] + j, stride, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(int16_t)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -727,15 +786,17 @@
     memset(opt_dest, 0, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, pixel_test_buff[index] + j, STRIDE);
-        ref(ref_dest, pixel_test_buff[index] + j, STRIDE);
+        checked(opt, opt_dest, pixel_test_buff[index] + j, stride);
+        ref(ref_dest, pixel_test_buff[index] + j, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -751,15 +812,17 @@
     memset(opt_dest, 0, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        opt(opt_dest, pixel_test_buff[index] + j, STRIDE);
-        ref(ref_dest, pixel_test_buff[index] + j, STRIDE);
+        checked(opt, opt_dest, pixel_test_buff[index] + j, stride);
+        ref(ref_dest, pixel_test_buff[index] + j, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -775,17 +838,18 @@
     memset(opt_dest, 0xCD, sizeof(opt_dest));
 
     int j = 0;
+    intptr_t stride2 = 64, stride = STRIDE;
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        opt(opt_dest, 64, pixel_test_buff[index1] + j, short_test_buff[index2] + j, STRIDE, STRIDE);
-        ref(ref_dest, 64, pixel_test_buff[index1] + j, short_test_buff[index2] + j, STRIDE, STRIDE);
+        checked(opt, opt_dest, stride2, pixel_test_buff[index1] + j, short_test_buff[index2] + j, stride, stride);
+        ref(ref_dest, stride2, pixel_test_buff[index1] + j, short_test_buff[index2] + j, stride, stride);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
-        {
             return false;
-        }
+
+        reportfail();
         j += INCR;
     }
 
@@ -796,14 +860,17 @@
 {
     int j = 0;
 
+    intptr_t stride = STRIDE;
+
     for (int i = 0; i < ITERS; i++)
     {
         int index = i % TEST_CASES;
-        uint64_t vres = opt(pixel_test_buff[index], STRIDE);
-        uint64_t cres = ref(pixel_test_buff[index], STRIDE);
+        uint64_t vres = checked(opt, pixel_test_buff[index], stride);
+        uint64_t cres = ref(pixel_test_buff[index], stride);
         if (vres != cres)
             return false;
 
+        reportfail();
         j += INCR;
     }
 
@@ -817,24 +884,26 @@
 
     for (int i = 0; i < ITERS; i++)
     {
-        int stride = rand() % 64;
+        intptr_t stride = rand() % 64;
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
         ref(pixel_test_buff[index1] + i, stride, pixel_test_buff[index2] + i, stride, sum0);
-        opt(pixel_test_buff[index1] + i, stride, pixel_test_buff[index2] + i, stride, sum1);
+        checked(opt, pixel_test_buff[index1] + i, stride, pixel_test_buff[index2] + i, stride, sum1);
 
         if (memcmp(sum0, sum1, sizeof(sum0)))
             return false;
+
+        reportfail();
     }
 
     return true;
 }
 
+/* TODO: This function causes crashes when checked. Is this a real bug? */
 bool PixelHarness::check_ssim_end(ssim_end4_t ref, ssim_end4_t opt)
 {
     ALIGN_VAR_32(int, sum0[5][4]);
     ALIGN_VAR_32(int, sum1[5][4]);
-    int width;
 
     for (int i = 0; i < ITERS; i++)
     {
@@ -847,14 +916,13 @@
             }
         }
 
-        width = (rand() % 4) + 1;   // range[1-4]
-
+        int width = (rand() % 4) + 1; // range[1-4]
         float cres = ref(sum0, sum1, width);
-        float vres = opt(sum0, sum1, width);
+        float vres = checked_float(opt, sum0, sum1, width);
         if (fabs(vres - cres) > 0.00001)
-        {
             return false;
-        }
+
+        reportfail();
     }
 
     return true;
@@ -869,18 +937,18 @@
 
     memset(ref_dest, 0xCD, sizeof(ref_dest));
     memset(opt_dest, 0xCD, sizeof(opt_dest));
+    intptr_t stride = STRIDE;
 
     for (int i = 0; i < ITERS; i++)
     {
         int index1 = rand() % TEST_CASES;
         int index2 = rand() % TEST_CASES;
-        ref(short_test_buff2[index1] + j, short_test_buff2[index2] + j, ref_dest, STRIDE, STRIDE, STRIDE);
-        opt(short_test_buff2[index1] + j, short_test_buff2[index2] + j, opt_dest, STRIDE, STRIDE, STRIDE);
+        ref(short_test_buff2[index1] + j, short_test_buff2[index2] + j, ref_dest, stride, stride, stride);
+        checked(opt, short_test_buff2[index1] + j, short_test_buff2[index2] + j, opt_dest, stride, stride, stride);
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
-        {
             return false;
-        }
 
+        reportfail();
         j += INCR;
     }
 
@@ -900,18 +968,79 @@
     for (int i = 0; i < ITERS; i++)
     {
         int width = 16 * (rand() % 4 + 1);
-        int8_t sign = rand () % 3;
+        int8_t sign = rand() % 3;
         if (sign == 2)
         {
             sign = -1;
         }
 
         ref(ref_dest, psbuf1 + j, width, sign);
-        opt(opt_dest, psbuf1 + j, width, sign);
+        checked(opt, opt_dest, psbuf1 + j, width, sign);
 
         if (memcmp(ref_dest, opt_dest, 64 * 64 * sizeof(pixel)))
             return false;
 
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_planecopy_sp(planecopy_sp_t ref, planecopy_sp_t opt)
+{
+    ALIGN_VAR_16(pixel, ref_dest[64 * 64]);
+    ALIGN_VAR_16(pixel, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int width = 16 + rand() % 48;
+    int height = 16 + rand() % 48;
+    intptr_t srcStride = 64;
+    intptr_t dstStride = width;
+    int j = 0;
+
+    for (int i = 0; i < ITERS; i++)
+    {
+        int index = i % TEST_CASES;
+        checked(opt, ushort_test_buff[index] + j, srcStride, opt_dest, dstStride, width, height, (int)8, (uint16_t)255);
+        ref(ushort_test_buff[index] + j, srcStride, ref_dest, dstStride, width, height, (int)8, (uint16_t)255);
+
+        if (memcmp(ref_dest, opt_dest, width * height * sizeof(pixel)))
+            return false;
+
+        reportfail();
+        j += INCR;
+    }
+
+    return true;
+}
+
+bool PixelHarness::check_planecopy_cp(planecopy_cp_t ref, planecopy_cp_t opt)
+{
+    ALIGN_VAR_16(pixel, ref_dest[64 * 64]);
+    ALIGN_VAR_16(pixel, opt_dest[64 * 64]);
+
+    memset(ref_dest, 0xCD, sizeof(ref_dest));
+    memset(opt_dest, 0xCD, sizeof(opt_dest));
+
+    int width = 16 + rand() % 48;
+    int height = 16 + rand() % 48;
+    intptr_t srcStride = 64;
+    intptr_t dstStride = width;
+    int j = 0;
+
+    for (int i = 0; i < ITERS; i++)
+    {
+        int index = i % TEST_CASES;
+        checked(opt, uchar_test_buff[index] + j, srcStride, opt_dest, dstStride, width, height, (int)2);
+        ref(uchar_test_buff[index] + j, srcStride, ref_dest, dstStride, width, height, (int)2);
+
+        if (memcmp(ref_dest, opt_dest, width * height * sizeof(pixel)))
+            return false;
+
+        reportfail();
         j += INCR;
     }
 
@@ -1037,30 +1166,33 @@
         }
     }
 
-    if (opt.luma_sub_ps[part])
+    if (opt.luma_addAvg[part])
     {
-        if (!check_pixel_sub_ps(ref.luma_sub_ps[part], opt.luma_sub_ps[part]))
+        if (!check_addAvg(ref.luma_addAvg[part], opt.luma_addAvg[part]))
         {
-            printf("luma_sub_ps[%s] failed\n", lumaPartStr[part]);
+            printf("luma_addAvg[%s] failed\n", lumaPartStr[part]);
             return false;
         }
     }
 
-    if (opt.luma_add_ps[part])
+    if (part < NUM_SQUARE_BLOCKS)
     {
-        if (!check_pixel_add_ps(ref.luma_add_ps[part], opt.luma_add_ps[part]))
+        if (opt.luma_sub_ps[part])
         {
-            printf("luma_add_ps[%s] failed\n", lumaPartStr[part]);
-            return false;
+            if (!check_pixel_sub_ps(ref.luma_sub_ps[part], opt.luma_sub_ps[part]))
+            {
+                printf("luma_sub_ps[%s] failed\n", lumaPartStr[part]);
+                return false;
+            }
         }
-    }
 
-    if (opt.luma_addAvg[part])
-    {
-        if (!check_addAvg(ref.luma_addAvg[part], opt.luma_addAvg[part]))
+        if (opt.luma_add_ps[part])
         {
-            printf("luma_addAvg[%s] failed\n", lumaPartStr[part]);
-            return false;
+            if (!check_pixel_add_ps(ref.luma_add_ps[part], opt.luma_add_ps[part]))
+            {
+                printf("luma_add_ps[%s] failed\n", lumaPartStr[part]);
+                return false;
+            }
         }
     }
 
@@ -1070,7 +1202,7 @@
         {
             if (!check_copy_pp(ref.chroma[i].copy_pp[part], opt.chroma[i].copy_pp[part]))
             {
-                printf("chroma_copy_pp[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
+                printf("chroma_copy_pp[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
                 return false;
             }
         }
@@ -1078,7 +1210,7 @@
         {
             if (!check_copy_sp(ref.chroma[i].copy_sp[part], opt.chroma[i].copy_sp[part]))
             {
-                printf("chroma_copy_sp[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
+                printf("chroma_copy_sp[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
                 return false;
             }
         }
@@ -1086,7 +1218,7 @@
         {
             if (!check_copy_ps(ref.chroma[i].copy_ps[part], opt.chroma[i].copy_ps[part]))
             {
-                printf("chroma_copy_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
+                printf("chroma_copy_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
                 return false;
             }
         }
@@ -1094,32 +1226,35 @@
         {
             if (!check_copy_ss(ref.chroma[i].copy_ss[part], opt.chroma[i].copy_ss[part]))
             {
-                printf("chroma_copy_ss[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
+                printf("chroma_copy_ss[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
                 return false;
             }
         }
-        if (opt.chroma[i].sub_ps[part])
+        if (opt.chroma[i].addAvg[part])
         {
-            if (!check_pixel_sub_ps(ref.chroma[i].sub_ps[part], opt.chroma[i].sub_ps[part]))
+            if (!check_addAvg(ref.chroma[i].addAvg[part], opt.chroma[i].addAvg[part]))
             {
-                printf("chroma_sub_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
+                printf("chroma_addAvg[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
                 return false;
             }
         }
-        if (opt.chroma[i].add_ps[part])
+        if (part < NUM_SQUARE_BLOCKS)
         {
-            if (!check_pixel_add_ps(ref.chroma[i].add_ps[part], opt.chroma[i].add_ps[part]))
+            if (opt.chroma[i].sub_ps[part])
             {
-                printf("chroma_add_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
-                return false;
+                if (!check_pixel_sub_ps(ref.chroma[i].sub_ps[part], opt.chroma[i].sub_ps[part]))
+                {
+                    printf("chroma_sub_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
+                    return false;
+                }
             }
-        }
-        if (opt.chroma[i].addAvg[part])
-        {
-            if (!check_addAvg(ref.chroma[i].addAvg[part], opt.chroma[i].addAvg[part]))
+            if (opt.chroma[i].add_ps[part])
             {
-                printf("chroma_addAvg[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[part]);
-                return false;
+                if (!check_pixel_add_ps(ref.chroma[i].add_ps[part], opt.chroma[i].add_ps[part]))
+                {
+                    printf("chroma_add_ps[%s][%s] failed\n", x265_source_csp_names[i], chromaPartStr[i][part]);
+                    return false;
+                }
             }
         }
     }
@@ -1183,6 +1318,15 @@
             }
         }
 
+        if ((i <= BLOCK_32x32) && opt.ssd_s[i])
+        {
+            if (!check_ssd_s(ref.ssd_s[i], opt.ssd_s[i]))
+            {
+                printf("ssd_s[%dx%d]: failed!\n", 4 << i, 4 << i);
+                return false;
+            }
+        }
+
         if (opt.blockfill_s[i])
         {
             if (!check_blockfill_s(ref.blockfill_s[i], opt.blockfill_s[i]))
@@ -1208,40 +1352,59 @@
                 return false;
             }
         }
-    }
 
-    if (opt.cvt32to16_shr)
-    {
-        if (!check_cvt32to16_shr_t(ref.cvt32to16_shr, opt.cvt32to16_shr))
+        if ((i < BLOCK_64x64) && opt.copy_cnt[i])
         {
-            printf("cvt32to16 failed!\n");
-            return false;
+            if (!check_copy_cnt_t(ref.copy_cnt[i], opt.copy_cnt[i]))
+            {
+                printf("copy_cnt[%dx%d] failed!\n", 4 << i, 4 << i);
+                return false;
+            }
         }
-    }
 
-    if (opt.cvt16to32_shl)
-    {
-        if (!check_cvt16to32_shl_t(ref.cvt16to32_shl, opt.cvt16to32_shl))
+        if ((i < BLOCK_64x64) && opt.cvt16to32_shr[i])
         {
-            printf("cvt16to32 failed!\n");
-            return false;
+            if (!check_cvt16to32_shr_t(ref.cvt16to32_shr[i], opt.cvt16to32_shr[i]))
+            {
+                printf("cvt16to32_shr failed!\n");
+                return false;
+            }
         }
+
+        if ((i < BLOCK_64x64) && opt.cvt32to16_shl[i])
+        {
+            if (!check_cvt32to16_shl_t(ref.cvt32to16_shl[i], opt.cvt32to16_shl[i]))
+            {
+                printf("cvt32to16_shl failed!\n");
+                return false;
+            }
+        }
+
+        if ((i < BLOCK_64x64) && opt.copy_shl[i])
+        {
+            if (!check_copy_shl_t(ref.copy_shl[i], opt.copy_shl[i]))
+            {
+                printf("copy_shl[%dx%d] failed!\n", 4 << i, 4 << i);
+                return false;
+            }
+        }
+
     }
 
-    if (opt.blockcpy_pp)
+    if (opt.cvt32to16_shr)
     {
-        if (!check_blockcopy_pp(ref.blockcpy_pp, opt.blockcpy_pp))
+        if (!check_cvt32to16_shr_t(ref.cvt32to16_shr, opt.cvt32to16_shr))
         {
-            printf("block copy failed!\n");
+            printf("cvt32to16 failed!\n");
             return false;
         }
     }
 
-    if (opt.blockcpy_ps)
+    if (opt.cvt16to32_shl)
     {
-        if (!check_blockcopy_ps(ref.blockcpy_ps, opt.blockcpy_ps))
+        if (!check_cvt16to32_shl_t(ref.cvt16to32_shl, opt.cvt16to32_shl))
         {
-            printf("block copy pixel_short failed!\n");
+            printf("cvt16to32_shl failed!\n");
             return false;
         }
     }
@@ -1264,15 +1427,6 @@
         }
     }
 
-    if (opt.pixeladd_ss)
-    {
-        if (!check_pixeladd_ss(ref.pixeladd_ss, opt.pixeladd_ss))
-        {
-            printf("pixel add clip failed!\n");
-            return false;
-        }
-    }
-
     if (opt.frame_init_lowres_core)
     {
         if (!check_downscale_t(ref.frame_init_lowres_core, opt.frame_init_lowres_core))
@@ -1327,6 +1481,33 @@
         }
     }
 
+    if (opt.planecopy_sp)
+    {
+        if (!check_planecopy_sp(ref.planecopy_sp, opt.planecopy_sp))
+        {
+            printf("planecopy_sp failed\n");
+            return false;
+        }
+    }
+
+    if (opt.planecopy_cp)
+    {
+        if (!check_planecopy_cp(ref.planecopy_cp, opt.planecopy_cp))
+        {
+            printf("planecopy_cp failed\n");
+            return false;
+        }
+    }
+
+    if (opt.copy_shr)
+    {
+        if (!check_copy_shr_t(ref.copy_shr, opt.copy_shr))
+        {
+            printf("copy_shr failed!\n");
+            return false;
+        }
+    }
+
     return true;
 }
 
@@ -1413,61 +1594,65 @@
         HEADER("luma_copy_ss[%s]", lumaPartStr[part]);
         REPORT_SPEEDUP(opt.luma_copy_ss[part], ref.luma_copy_ss[part], sbuf1, 64, sbuf2, 128);
     }
-    if (opt.luma_sub_ps[part])
-    {
-        HEADER("luma_sub_ps[%s]", lumaPartStr[part]);
-        REPORT_SPEEDUP(opt.luma_sub_ps[part], ref.luma_sub_ps[part], (int16_t*)pbuf1, FENC_STRIDE, pbuf2, pbuf1, STRIDE, STRIDE);
-    }
-
-    if (opt.luma_add_ps[part])
-    {
-        HEADER("luma_add_ps[%s]", lumaPartStr[part]);
-        REPORT_SPEEDUP(opt.luma_add_ps[part], ref.luma_add_ps[part], pbuf1, FENC_STRIDE, pbuf2, sbuf1, STRIDE, STRIDE);
-    }
-
     if (opt.luma_addAvg[part])
     {
         HEADER("luma_addAvg[%s]", lumaPartStr[part]);
         REPORT_SPEEDUP(opt.luma_addAvg[part], ref.luma_addAvg[part], sbuf1, sbuf2, pbuf1, STRIDE, STRIDE, STRIDE);
     }
+    if (part < NUM_SQUARE_BLOCKS)
+    {
+        if (opt.luma_sub_ps[part])
+        {
+            HEADER("luma_sub_ps[%s]", lumaPartStr[part]);
+            REPORT_SPEEDUP(opt.luma_sub_ps[part], ref.luma_sub_ps[part], (int16_t*)pbuf1, FENC_STRIDE, pbuf2, pbuf1, STRIDE, STRIDE);
+        }
+        if (opt.luma_add_ps[part])
+        {
+            HEADER("luma_add_ps[%s]", lumaPartStr[part]);
+            REPORT_SPEEDUP(opt.luma_add_ps[part], ref.luma_add_ps[part], pbuf1, FENC_STRIDE, pbuf2, sbuf1, STRIDE, STRIDE);
+        }
+    }
 
     for (int i = 0; i < X265_CSP_COUNT; i++)
     {
         if (opt.chroma[i].copy_pp[part])
         {
-            HEADER("[%s] copy_pp[%s]", x265_source_csp_names[i], chromaPartStr[part]);
+            HEADER("[%s] copy_pp[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
             REPORT_SPEEDUP(opt.chroma[i].copy_pp[part], ref.chroma[i].copy_pp[part], pbuf1, 64, pbuf2, 128);
         }
         if (opt.chroma[i].copy_sp[part])
         {
-            HEADER("[%s] copy_sp[%s]", x265_source_csp_names[i], chromaPartStr[part]);
+            HEADER("[%s] copy_sp[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
             REPORT_SPEEDUP(opt.chroma[i].copy_sp[part], ref.chroma[i].copy_sp[part], pbuf1, 64, sbuf3, 128);
         }
         if (opt.chroma[i].copy_ps[part])
         {
-            HEADER("[%s] copy_ps[%s]", x265_source_csp_names[i], chromaPartStr[part]);
+            HEADER("[%s] copy_ps[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
             REPORT_SPEEDUP(opt.chroma[i].copy_ps[part], ref.chroma[i].copy_ps[part], sbuf1, 64, pbuf1, 128);
         }
         if (opt.chroma[i].copy_ss[part])
         {
-            HEADER("[%s] copy_ss[%s]", x265_source_csp_names[i], chromaPartStr[part]);
+            HEADER("[%s] copy_ss[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
             REPORT_SPEEDUP(opt.chroma[i].copy_ss[part], ref.chroma[i].copy_ss[part], sbuf1, 64, sbuf2, 128);
         }
-        if (opt.chroma[i].sub_ps[part])
-        {
-            HEADER("[%s]  sub_ps[%s]", x265_source_csp_names[i], chromaPartStr[part]);
-            REPORT_SPEEDUP(opt.chroma[i].sub_ps[part], ref.chroma[i].sub_ps[part], (int16_t*)pbuf1, FENC_STRIDE, pbuf2, pbuf1, STRIDE, STRIDE);
-        }
-        if (opt.chroma[i].add_ps[part])
-        {
-            HEADER("[%s]  add_ps[%s]", x265_source_csp_names[i], chromaPartStr[part]);
-            REPORT_SPEEDUP(opt.chroma[i].add_ps[part], ref.chroma[i].add_ps[part], pbuf1, FENC_STRIDE, pbuf2, sbuf1, STRIDE, STRIDE);
-        }
         if (opt.chroma[i].addAvg[part])
         {
-            HEADER("[%s]  addAvg[%s]", x265_source_csp_names[i], chromaPartStr[part]);
+            HEADER("[%s]  addAvg[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
             REPORT_SPEEDUP(opt.chroma[i].addAvg[part], ref.chroma[i].addAvg[part], sbuf1, sbuf2, pbuf1, STRIDE, STRIDE, STRIDE);
         }
+        if (part < NUM_SQUARE_BLOCKS)
+        {
+            if (opt.chroma[i].sub_ps[part])
+            {
+                HEADER("[%s]  sub_ps[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
+                REPORT_SPEEDUP(opt.chroma[i].sub_ps[part], ref.chroma[i].sub_ps[part], (int16_t*)pbuf1, FENC_STRIDE, pbuf2, pbuf1, STRIDE, STRIDE);
+            }
+            if (opt.chroma[i].add_ps[part])
+            {
+                HEADER("[%s]  add_ps[%s]", x265_source_csp_names[i], chromaPartStr[i][part]);
+                REPORT_SPEEDUP(opt.chroma[i].add_ps[part], ref.chroma[i].add_ps[part], pbuf1, FENC_STRIDE, pbuf2, sbuf1, STRIDE, STRIDE);
+            }
+        }
     }
 
 #undef HEADER
@@ -1509,6 +1694,11 @@
 
     for (int i = 0; i < NUM_SQUARE_BLOCKS; i++)
     {
+        if ((i <= BLOCK_32x32) && opt.ssd_s[i])
+        {
+            HEADER("ssd_s[%dx%d]", 4 << i, 4 << i);
+            REPORT_SPEEDUP(opt.ssd_s[i], ref.ssd_s[i], sbuf1, STRIDE);
+        }
         if (opt.sa8d[i])
         {
             HEADER("sa8d[%dx%d]", 4 << i, 4 << i);
@@ -1523,7 +1713,7 @@
         if (opt.calcrecon[i])
         {
             HEADER("recon[%dx%d]", 4 << i, 4 << i);
-            REPORT_SPEEDUP(opt.calcrecon[i], ref.calcrecon[i], pbuf1, sbuf1, pbuf2, sbuf1, pbuf1, 64, 64, 64);
+            REPORT_SPEEDUP(opt.calcrecon[i], ref.calcrecon[i], pbuf1, sbuf1, sbuf1, pbuf1, 64, 64, 64);
         }
 
         if (opt.blockfill_s[i])
@@ -1543,6 +1733,31 @@
             HEADER("var[%dx%d]", 4 << i, 4 << i);
             REPORT_SPEEDUP(opt.var[i], ref.var[i], pbuf1, STRIDE);
         }
+
+        if ((i < BLOCK_64x64) && opt.cvt16to32_shr[i])
+        {
+            HEADER("cvt16to32_shr[%dx%d]", 4 << i, 4 << i);
+            REPORT_SPEEDUP(opt.cvt16to32_shr[i], ref.cvt16to32_shr[i], ibuf1, sbuf2, STRIDE, 3, 4);
+        }
+
+        if ((i < BLOCK_64x64) && opt.cvt32to16_shl[i])
+        {
+            HEADER("cvt32to16_shl[%dx%d]", 4 << i, 4 << i);
+            REPORT_SPEEDUP(opt.cvt32to16_shl[i], ref.cvt32to16_shl[i], sbuf2, ibuf1, STRIDE, 3);
+        }
+
+        if ((i < BLOCK_64x64) && opt.copy_cnt[i])
+        {
+            HEADER("copy_cnt[%dx%d]", 4 << i, 4 << i);
+            REPORT_SPEEDUP(opt.copy_cnt[i], ref.copy_cnt[i], sbuf1, sbuf2, STRIDE);
+        }
+
+        if ((i < BLOCK_64x64) && opt.copy_shl[i])
+        {
+            HEADER("copy_shl[%dx%d]", 4 << i, 4 << i);
+            REPORT_SPEEDUP(opt.copy_shl[i], ref.copy_shl[i], sbuf1, sbuf2, STRIDE, 64);
+        }
+
     }
 
     if (opt.cvt32to16_shr)
@@ -1557,18 +1772,6 @@
         REPORT_SPEEDUP(opt.cvt16to32_shl, ref.cvt16to32_shl, ibuf1, sbuf1, 64, 5, 64);
     }
 
-    if (opt.blockcpy_pp)
-    {
-        HEADER0("blockcpy_pp");
-        REPORT_SPEEDUP(opt.blockcpy_pp, ref.blockcpy_pp, 64, 64, pbuf1, FENC_STRIDE, pbuf2, STRIDE);
-    }
-
-    if (opt.blockcpy_ps)
-    {
-        HEADER0("blockcpy_ps");
-        REPORT_SPEEDUP(opt.blockcpy_ps, ref.blockcpy_ps, 64, 64, pbuf1, FENC_STRIDE, (int16_t*)sbuf3, STRIDE);
-    }
-
     if (opt.weight_pp)
     {
         HEADER0("weight_pp");
@@ -1581,12 +1784,6 @@
         REPORT_SPEEDUP(opt.weight_sp, ref.weight_sp, (int16_t*)sbuf1, pbuf1, 64, 64, 32, 32, 128, 1 << 9, 10, 100);
     }
 
-    if (opt.pixeladd_ss)
-    {
-        HEADER0("pixeladd_ss");
-        REPORT_SPEEDUP(opt.pixeladd_ss, ref.pixeladd_ss, 64, 64, (int16_t*)pbuf1, FENC_STRIDE, (int16_t*)pbuf2, (int16_t*)pbuf1, STRIDE, STRIDE);
-    }
-
     if (opt.frame_init_lowres_core)
     {
         HEADER0("downscale");
@@ -1619,7 +1816,26 @@
 
     if (opt.saoCuOrgE0)
     {
-        printf("SAO_EO_0");
+        HEADER0("SAO_EO_0");
         REPORT_SPEEDUP(opt.saoCuOrgE0, ref.saoCuOrgE0, pbuf1, psbuf1, 64, 1);
     }
+
+    if (opt.planecopy_sp)
+    {
+        HEADER0("planecopy_sp");
+        REPORT_SPEEDUP(opt.planecopy_sp, ref.planecopy_sp, ushort_test_buff[0], 64, pbuf1, 64, 64, 64, 8, 255);
+    }
+
+    if (opt.planecopy_cp)
+    {
+        HEADER0("planecopy_cp");
+        REPORT_SPEEDUP(opt.planecopy_cp, ref.planecopy_cp, uchar_test_buff[0], 64, pbuf1, 64, 64, 64, 2);
+    }
+
+    if (opt.copy_shr)
+    {
+        HEADER0("copy_shr");
+        REPORT_SPEEDUP(opt.copy_shr, ref.copy_shr, sbuf1, sbuf2, 64, 5, 64);
+    }
+
 }

x265-hg20140324.tar.bz2/source/test/pixelharness.h -> x265-hg20140928.tar.bz2/source/test/pixelharness.h Changed

x265-hg20140324.tar.bz2/source/test/testbench.cpp -> x265-hg20140928.tar.bz2/source/test/testbench.cpp Changed

@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #include "common.h"
@@ -36,20 +36,46 @@
 
 const char* lumaPartStr[NUM_LUMA_PARTITIONS] =
 {
-    "  4x4",
-    "  8x8", "  8x4", "  4x8",
-    "16x16", " 16x8", " 8x16", "16x12", "12x16", " 16x4", " 4x16",
-    "32x32", "32x16", "16x32", "32x24", "24x32", " 32x8", " 8x32",
-    "64x64", "64x32", "32x64", "64x48", "48x64", "64x16", "16x64",
+    "  4x4", "  8x8", "16x16", "32x32", "64x64",
+    "  8x4", "  4x8",
+    " 16x8", " 8x16",
+    "32x16", "16x32",
+    "64x32", "32x64",
+    "16x12", "12x16", " 16x4", " 4x16",
+    "32x24", "24x32", " 32x8", " 8x32",
+    "64x48", "48x64", "64x16", "16x64",
 };
 
-const char* chromaPartStr[NUM_CHROMA_PARTITIONS] =
+const char* chromaPartStr420[NUM_CHROMA_PARTITIONS] =
 {
-    "  2x2", // never used by HEVC
-    "  4x4", "  4x2", "  2x4",
-    "  8x8", "  8x4", "  4x8", "  8x6", "  6x8", "  8x2", "  2x8",
-    "16x16", " 16x8", " 8x16", "16x12", "12x16", " 16x4", " 4x16",
-    "32x32", "32x16", "16x32", "32x24", "24x32", " 32x8", " 8x32",
+    "  2x2", "  4x4", "  8x8", "16x16", "32x32",
+    "  4x2", "  2x4",
+    "  8x4", "  4x8",
+    " 16x8", " 8x16",
+    "32x16", "16x32",
+    "  8x6", "  6x8", "  8x2", "  2x8",
+    "16x12", "12x16", " 16x4", " 4x16",
+    "32x24", "24x32", " 32x8", " 8x32",
+};
+
+const char* chromaPartStr422[NUM_CHROMA_PARTITIONS] =
+{
+    "  2x4", "  4x8", " 8x16", "16x32", "32x64",
+    "  4x4", "  2x8",
+    "  8x8", " 4x16",
+    "16x16", " 8x32",
+    "32x32", "16x64",
+    " 8x12", " 6x16", "  8x4", " 2x16",
+    "16x24", "12x32", " 16x8", " 4x32",
+    "32x48", "24x64", "32x16", " 8x64",
+};
+
+const char* const* chromaPartStr[X265_CSP_COUNT] =
+{
+    lumaPartStr,
+    chromaPartStr420,
+    chromaPartStr422,
+    lumaPartStr
 };
 
 void do_help()
@@ -62,6 +88,11 @@
     printf("Options and testbench name may be truncated.\n");
 }
 
+PixelHarness  HPixel;
+MBDstHarness  HMBDist;
+IPFilterHarness HIPFilter;
+IntraPredHarness HIPred;
+
 int main(int argc, char *argv[])
 {
     int cpuid = x265::cpu_detect();
@@ -110,11 +141,6 @@
     printf("Using random seed %X %s\n", seed, bpp[HIGH_BIT_DEPTH]);
     srand(seed);
 
-    PixelHarness  HPixel;
-    MBDstHarness  HMBDist;
-    IPFilterHarness HIPFilter;
-    IntraPredHarness HIPred;
-
     // To disable classes of tests, simply comment them out in this list
     TestHarness *harness[] =
     {
@@ -127,6 +153,7 @@
     EncoderPrimitives cprim;
     memset(&cprim, 0, sizeof(EncoderPrimitives));
     Setup_C_Primitives(cprim);
+    Setup_Alias_Primitives(cprim);
 
     struct test_arch_t
     {
@@ -168,6 +195,7 @@
         EncoderPrimitives asmprim;
         memset(&asmprim, 0, sizeof(asmprim));
         Setup_Assembly_Primitives(asmprim, test_arch[i].flag);
+        memcpy(&primitives, &asmprim, sizeof(EncoderPrimitives));
         for (size_t h = 0; h < sizeof(harness) / sizeof(TestHarness*); h++)
         {
             if (testname && strncmp(testname, harness[h]->getName(), strlen(testname)))
@@ -186,6 +214,12 @@
     memset(&optprim, 0, sizeof(optprim));
     Setup_Instrinsic_Primitives(optprim, cpuid);
     Setup_Assembly_Primitives(optprim, cpuid);
+    Setup_Alias_Primitives(optprim);
+
+    /* some hybrid primitives may rely on other primitives in the
+     * global primitive table, so set up those pointers. This is a
+     * bit ugly, but I don't see a better solution */
+    memcpy(&primitives, &optprim, sizeof(EncoderPrimitives));
 
     printf("\nTest performance improvement with full optimizations\n");

x265-hg20140324.tar.bz2/source/test/testharness.h -> x265-hg20140928.tar.bz2/source/test/testharness.h Changed

@@ -18,16 +18,19 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #ifndef _TESTHARNESS_H_
 #define _TESTHARNESS_H_ 1
 
-#include "TLibCommon/CommonDef.h"
 #include "common.h"
 #include "primitives.h"
 
+#if _MSC_VER
+#pragma warning(disable: 4324) // structure was padded due to __declspec(align())
+#endif
+
 #if HIGH_BIT_DEPTH
 #define BIT_DEPTH 10
 #else
@@ -37,11 +40,12 @@
 #define PIXEL_MIN 0
 #define SHORT_MAX  32767
 #define SHORT_MIN -32767
+#define UNSIGNED_SHORT_MAX 65535
 
 using namespace x265;
 
 extern const char* lumaPartStr[NUM_LUMA_PARTITIONS];
-extern const char* chromaPartStr[NUM_CHROMA_PARTITIONS];
+extern const char* const* chromaPartStr[X265_CSP_COUNT];
 
 class TestHarness
 {
@@ -56,6 +60,13 @@
     virtual void measureSpeed(const EncoderPrimitives& ref, const EncoderPrimitives& opt) = 0;
 
     virtual const char *getName() const = 0;
+
+protected:
+
+    /* Temporary variables for stack checks */
+    int      m_ok;
+
+    uint64_t m_rand;
 };
 
 #ifdef _MSC_VER
@@ -107,4 +118,53 @@
         printf("\t %-8.2lf \t %-8.2lf\n", optperf, refperf); \
     }
 
+extern "C" {
+#if X265_ARCH_X86
+int x265_stack_pagealign(int (*func)(), int align);
+
+/* detect when callee-saved regs aren't saved
+ * needs an explicit asm check because it only sometimes crashes in normal use. */
+intptr_t x265_checkasm_call(intptr_t (*func)(), int *ok, ...);
+float x265_checkasm_call_float(float (*func)(), int *ok, ...);
+#else
+#define x265_stack_pagealign(func, align) func()
+#endif
+
+#if X86_64
+
+/* Evil hack: detect incorrect assumptions that 32-bit ints are zero-extended to 64-bit.
+ * This is done by clobbering the stack with junk around the stack pointer and calling the
+ * assembly function through x265_checkasm_call with added dummy arguments which forces all
+ * real arguments to be passed on the stack and not in registers. For 32-bit argument the
+ * upper half of the 64-bit register location on the stack will now contain junk. Note that
+ * this is dependent on compiler behavior and that interrupts etc. at the wrong time may
+ * overwrite the junk written to the stack so there's no guarantee that it will always
+ * detect all functions that assumes zero-extension.
+ */
+void x265_checkasm_stack_clobber(uint64_t clobber, ...);
+#define checked(func, ...) ( \
+        m_ok = 1, m_rand = (rand() & 0xffff) * 0x0001000100010001ULL, \
+        x265_checkasm_stack_clobber(m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, \
+                                    m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, \
+                                    m_rand, m_rand, m_rand, m_rand, m_rand), /* max_args+6 */ \
+        x265_checkasm_call((intptr_t(*)())func, &m_ok, 0, 0, 0, 0, __VA_ARGS__))
+
+#define checked_float(func, ...) ( \
+        m_ok = 1, m_rand = (rand() & 0xffff) * 0x0001000100010001ULL, \
+        x265_checkasm_stack_clobber(m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, \
+                                    m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, m_rand, \
+                                    m_rand, m_rand, m_rand, m_rand, m_rand), /* max_args+6 */ \
+        x265_checkasm_call_float((float(*)())func, &m_ok, 0, 0, 0, 0, __VA_ARGS__))
+#define reportfail() if (!m_ok) { fprintf(stderr, "stack clobber check failed at %s:%d", __FILE__, __LINE__); abort(); }
+#elif ARCH_X86
+#define checked(func, ...) x265_checkasm_call((intptr_t(*)())func, &m_ok, __VA_ARGS__);
+#define checked_float(func, ...) x265_checkasm_call_float((float(*)())func, &m_ok, __VA_ARGS__);
+
+#else // if X86_64
+#define checked(func, ...) func(__VA_ARGS__)
+#define checked_float(func, ...) func(__VA_ARGS__)
+#define reportfail()
+#endif // if X86_64
+}
+
 #endif // ifndef _TESTHARNESS_H_

x265-hg20140324.tar.bz2/source/test/testpool.cpp -> x265-hg20140928.tar.bz2/source/test/testpool.cpp Changed

x265-hg20140324.tar.bz2/source/x265.cpp -> x265-hg20140928.tar.bz2/source/x265.cpp Changed

@@ -18,7 +18,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  *
  * This program is also available under a commercial proprietary license.
- * For more information, contact us at licensing@multicorewareinc.com.
+ * For more information, contact us at license @ x265.com.
  *****************************************************************************/
 
 #if _MSC_VER
@@ -27,6 +27,7 @@
 
 #include "input/input.h"
 #include "output/output.h"
+#include "filters/filters.h"
 #include "common.h"
 #include "param.h"
 #include "cpu.h"
@@ -68,7 +69,13 @@
     { "tune",           required_argument, NULL, 't' },
     { "frame-threads",  required_argument, NULL, 'F' },
     { "log-level",      required_argument, NULL, 0 },
+    { "profile",        required_argument, NULL, 0 },
+    { "level-idc",      required_argument, NULL, 0 },
+    { "high-tier",            no_argument, NULL, 0 },
+    { "no-high-tier",         no_argument, NULL, 0 },
     { "csv",            required_argument, NULL, 0 },
+    { "no-cu-stats",          no_argument, NULL, 0 },
+    { "cu-stats",             no_argument, NULL, 0 },
     { "y4m",                  no_argument, NULL, 0 },
     { "no-progress",          no_argument, NULL, 0 },
     { "output",         required_argument, NULL, 'o' },
@@ -106,8 +113,12 @@
     { "tskip",                no_argument, NULL, 0 },
     { "no-tskip-fast",        no_argument, NULL, 0 },
     { "tskip-fast",           no_argument, NULL, 0 },
+    { "cu-lossless",          no_argument, NULL, 0 },
+    { "no-cu-lossless",       no_argument, NULL, 0 },
     { "no-constrained-intra", no_argument, NULL, 0 },
     { "constrained-intra",    no_argument, NULL, 0 },
+    { "fast-intra",           no_argument, NULL, 0 },
+    { "no-fast-intra",        no_argument, NULL, 0 },
     { "no-open-gop",          no_argument, NULL, 0 },
     { "open-gop",             no_argument, NULL, 0 },
     { "keyint",         required_argument, NULL, 'I' },
@@ -124,8 +135,11 @@
     { "ref",            required_argument, NULL, 0 },
     { "no-weightp",           no_argument, NULL, 0 },
     { "weightp",              no_argument, NULL, 'w' },
+    { "no-weightb",           no_argument, NULL, 0 },
+    { "weightb",              no_argument, NULL, 0 },
     { "crf",            required_argument, NULL, 0 },
     { "crf-max",        required_argument, NULL, 0 },
+    { "crf-min",        required_argument, NULL, 0 },
     { "vbv-maxrate",    required_argument, NULL, 0 },
     { "vbv-bufsize",    required_argument, NULL, 0 },
     { "vbv-init",       required_argument, NULL, 0 },
@@ -133,9 +147,16 @@
     { "qp",             required_argument, NULL, 'q' },
     { "aq-mode",        required_argument, NULL, 0 },
     { "aq-strength",    required_argument, NULL, 0 },
+    { "ipratio",        required_argument, NULL, 0 },
+    { "pbratio",        required_argument, NULL, 0 },
     { "cbqpoffs",       required_argument, NULL, 0 },
     { "crqpoffs",       required_argument, NULL, 0 },
     { "rd",             required_argument, NULL, 0 },
+    { "psy-rd",         required_argument, NULL, 0 },
+    { "psy-rdoq",       required_argument, NULL, 0 },
+    { "scaling-list",   required_argument, NULL, 0 },
+    { "lossless",             no_argument, NULL, 0 },
+    { "no-lossless",          no_argument, NULL, 0 },
     { "no-signhide",          no_argument, NULL, 0 },
     { "signhide",             no_argument, NULL, 0 },
     { "no-lft",               no_argument, NULL, 0 },
@@ -153,8 +174,8 @@
     { "strong-intra-smoothing",    no_argument, NULL, 0 },
     { "no-cutree",                 no_argument, NULL, 0 },
     { "cutree",                    no_argument, NULL, 0 },
-    { "vui",                  no_argument, NULL, 0 },
-    { "no-vui",               no_argument, NULL, 0 },
+    { "no-hrd",               no_argument, NULL, 0 },
+    { "hrd",                  no_argument, NULL, 0 },
     { "sar",            required_argument, NULL, 0 },
     { "overscan",       required_argument, NULL, 0 },
     { "videoformat",    required_argument, NULL, 0 },
@@ -163,20 +184,26 @@
     { "transfer",       required_argument, NULL, 0 },
     { "colormatrix",    required_argument, NULL, 0 },
     { "chromaloc",      required_argument, NULL, 0 },
-    { "fieldseq",             no_argument, NULL, 0 },
-    { "no-fieldseq",          no_argument, NULL, 0 },
-    { "framefieldinfo",       no_argument, NULL, 0 },
-    { "no-framefieldinfo",    no_argument, NULL, 0 },
     { "crop-rect",      required_argument, NULL, 0 },
-    { "timinginfo",           no_argument, NULL, 0 },
-    { "no-timinginfo",        no_argument, NULL, 0 },
-    { "hrd",                  no_argument, NULL, 0 },
-    { "no-hrd",               no_argument, NULL, 0 },
-    { "nal-hrd",        required_argument, NULL, 0 },
-    { "bitstreamrestriction", no_argument, NULL, 0 },
-    { "no-bitstreamrestriction", no_argument, NULL, 0 },
-    { "subpichrd",            no_argument, NULL, 0 },
-    { "no-subpichrd",         no_argument, NULL, 0 },
+    { "no-dither",            no_argument, NULL, 0 },
+    { "dither",               no_argument, NULL, 0 },
+    { "no-repeat-headers",    no_argument, NULL, 0 },
+    { "repeat-headers",       no_argument, NULL, 0 },
+    { "aud",                  no_argument, NULL, 0 },
+    { "no-aud",               no_argument, NULL, 0 },
+    { "info",                 no_argument, NULL, 0 },
+    { "no-info",              no_argument, NULL, 0 },
+    { "qpfile",         required_argument, NULL, 0 },
+    { "lambda-file",    required_argument, NULL, 0 },
+    { "b-intra",              no_argument, NULL, 0 },
+    { "no-b-intra",           no_argument, NULL, 0 },
+    { "nr",             required_argument, NULL, 0 },
+    { "stats",          required_argument, NULL, 0 },
+    { "pass",           required_argument, NULL, 0 },
+    { "slow-firstpass",       no_argument, NULL, 0 },
+    { "no-slow-firstpass",    no_argument, NULL, 0 },
+    { "analysis-mode",  required_argument, NULL, 0 },
+    { "analysis-file",  required_argument, NULL, 0 },
     { 0, 0, 0, 0 }
 };
 
@@ -194,14 +221,19 @@
     std::fstream bitstreamFile;
     bool bProgress;
     bool bForceY4m;
+    bool bDither;
 
     uint32_t seek;              // number of frames to skip from the beginning
     uint32_t framesToBeEncoded; // number of frames to encode
     uint64_t totalbytes;
+    size_t   analysisRecordSize; // number of bytes read from or dumped into file
+    int      analysisHeaderSize;
 
     int64_t startTime;
     int64_t prevUpdateTime;
     float   frameRate;
+    FILE*   qpfile;
+    FILE*   analysisFile;
 
     /* in microseconds */
     static const int UPDATE_INTERVAL = 250000;
@@ -216,6 +248,11 @@
         bForceY4m = false;
         startTime = x265_mdate();
         prevUpdateTime = 0;
+        bDither = false;
+        qpfile = NULL;
+        analysisFile = NULL;
+        analysisRecordSize = 0;
+        analysisHeaderSize = 0;
     }
 
     void destroy();
@@ -224,6 +261,10 @@
     void printVersion(x265_param *param);
     void showHelp(x265_param *param);
     bool parse(int argc, char **argv, x265_param* param);
+    bool parseQPFile(x265_picture &pic_org);
+    void readAnalysisFile(x265_picture* pic, x265_param*);
+    void writeAnalysisFile(x265_picture* pic, x265_param*);
+    bool validateFanout(x265_param*);
 };
 
 void CLIOptions::destroy()
@@ -234,6 +275,12 @@
     if (recon)
         recon->release();
     recon = NULL;
+    if (qpfile)
+        fclose(qpfile);
+    qpfile = NULL;
+    if (analysisFile)
+        fclose(analysisFile);
+    analysisFile = NULL;
 }
 
 void CLIOptions::writeNALs(const x265_nal* nal, uint32_t nalcount)
@@ -256,9 +303,9 @@
         return;
     int64_t elapsed = time - startTime;
     double fps = elapsed > 0 ? frameNum * 1000000. / elapsed : 0;
+    float bitrate = 0.008f * totalbytes * (param->fpsNum / param->fpsDenom) / ((float)frameNum);
     if (framesToBeEncoded)
     {
-        float bitrate = 0.008f * totalbytes / ((float)frameNum / ((float)param->fpsNum / param->fpsDenom));
         int eta = (int)(elapsed * (framesToBeEncoded - frameNum) / ((int64_t)frameNum * 1000000));
         sprintf(buf, "x265 [%.1f%%] %d/%d frames, %.2f fps, %.2f kb/s, eta %d:%02d:%02d",
                 100. * frameNum / framesToBeEncoded, frameNum, framesToBeEncoded, fps, bitrate,
@@ -266,7 +313,6 @@
     }
     else
     {
-        float bitrate = (float)totalbytes * 8 / ((float)1000 * param->fpsNum / param->fpsDenom);
         sprintf(buf, "x265 %d frames: %.2f fps, %.2f kb/s", frameNum, fps, bitrate);
     }
     fprintf(stderr, "%s  \r", buf + 5);
@@ -292,53 +338,73 @@
     H0("    infile can be YUV or Y4M\n");
     H0("    outfile is raw HEVC bitstream\n");
     H0("\nExecutable Options:\n");
-    H0("-h/--h                           Show this help text and exit\n");
+    H0("-h/--help                        Show this help text and exit\n");
     H0("-V/--version                     Show version info and exit\n");
-    H0("   --[no-]asm <bool|int|string>  Override CPU detection. Default: auto\n");
-    H0("   --threads <integer>           Number of threads for thread pool (0: detect CPU core count, default)\n");
-    H0("-F/--frame-threads <integer>     Number of concurrently encoded frames. 0: auto-determined by core count\n");
-    H0("   --log-level <string>          Logging level: none error warning info debug full. Default %s\n", logLevelNames[param->logLevel+1]);
-    H0("   --csv <filename>              Comma separated log file, log level >= 3 frame log, else one line per run\n");
-    H0("   --no-progress                 Disable CLI progress reports\n");
+    H0("\nOutput Options:\n");
     H0("-o/--output <filename>           Bitstream output file name\n");
+    H0("   --log-level <string>          Logging level: none error warning info debug full. Default %s\n", logLevelNames[param->logLevel + 1]);
+    H0("   --no-progress                 Disable CLI progress reports\n");
+    H0("   --[no-]cu-stats               Enable logging stats about distribution of cu across all modes. Default %s\n",OPT(param->bLogCuStats));
+    H0("   --csv <filename>              Comma separated log file, log level >= 3 frame log, else one line per run\n");
     H0("\nInput Options:\n");
     H0("   --input <filename>            Raw YUV or Y4M input file name. `-` for stdin\n");
     H0("   --y4m                         Force parsing of input stream as YUV4MPEG2 regardless of file extension\n");
-    H0("   --input-depth <integer>       Bit-depth of input file. Default 8\n");
-    H0("   --input-res WxH               Source picture size [w x h], auto-detected if Y4M\n");
-    H0("   --input-csp <string>          Source color space: i420 or i444, auto-detected if Y4M. Default: i420\n");
     H0("   --fps <float|rational>        Source frame rate (float or num/denom), auto-detected if Y4M\n");
-    H0("   --[no-]interlace <bff|tff>    Indicate input pictures are interlace fields in temporal order. Default progressive\n");
-    H0("   --seek <integer>              First frame to encode\n");
-    H0("\nPresets:\n");
+    H0("   --input-res WxH               Source picture size [w x h], auto-detected if Y4M\n");
+    H0("   --input-depth <integer>       Bit-depth of input file. Default 8\n");
+    H0("   --input-csp <string>          Source color space: i420, i444 or i422, auto-detected if Y4M. Default: i420\n");
     H0("-f/--frames <integer>            Maximum number of frames to encode. Default all\n");
-    H0("-p/--preset <string>             Trade off performance for compression efficiency. Default medium\n");
-    H0("                                     ultrafast, veryfast, faster, fast, medium, slow, slower, veryslow, or placebo\n");
-    H0("-t/--tune <string>               Tune the settings for a particular type of source or situation:\n");
-    H0("                                     psnr, ssim, zerolatency, or fastdecode\n");
+    H0("   --seek <integer>              First frame to encode\n");
+    H0("   --[no-]interlace <bff|tff>    Indicate input pictures are interlace fields in temporal order. Default progressive\n");
+    H0("   --dither                      Enable dither if downscaling to 8 bit pixels. Default disabled\n");
     H0("\nQuality reporting metrics:\n");
     H0("   --[no-]ssim                   Enable reporting SSIM metric scores. Default %s\n", OPT(param->bEnableSsim));
     H0("   --[no-]psnr                   Enable reporting PSNR metric scores. Default %s\n", OPT(param->bEnablePsnr));
-    H0("\nQuad-Tree analysis:\n");
+    H0("\nProfile, Level, Tier:\n");
+    H0("   --profile <string>            Enforce an encode profile: main, main10, mainstillpicture\n");
+    H0("   --level-idc <integer|float>   Force a minumum required decoder level (as '5.0' or '50')\n");
+    H0("   --[no-]high-tier              If a decoder level is specified, this modifier selects High tier of that level\n");
+    H0("\nThreading, performance:\n");
+    H0("   --threads <integer>           Number of threads for thread pool (0: detect CPU core count, default)\n");
+    H0("-F/--frame-threads <integer>     Number of concurrently encoded frames. 0: auto-determined by core count\n");
     H0("   --[no-]wpp                    Enable Wavefront Parallel Processing. Default %s\n", OPT(param->bEnableWavefront));
+    H0("   --[no-]asm <bool|int|string>  Override CPU detection. Default: auto\n");
+    H0("\nPresets:\n");
+    H0("-p/--preset <string>             Trade off performance for compression efficiency. Default medium\n");
+    H0("                                 ultrafast, superfast, veryfast, faster, fast, medium, slow, slower, veryslow, or placebo\n");
+    H0("-t/--tune <string>               Tune the settings for a particular type of source or situation:\n");
+    H0("                                 psnr, ssim, zerolatency, or fastdecode\n");
+    H0("\nQuad-Tree size and depth:\n");
     H0("-s/--ctu <64|32|16>              Maximum CU size (default: 64x64). Default %d\n", param->maxCUSize);
     H0("   --tu-intra-depth <integer>    Max TU recursive depth for intra CUs. Default %d\n", param->tuQTMaxIntraDepth);
     H0("   --tu-inter-depth <integer>    Max TU recursive depth for inter CUs. Default %d\n", param->tuQTMaxInterDepth);
+    H0("   --[no-]rect                   Enable rectangular motion partitions Nx2N and 2NxN. Default %s\n", OPT(param->bEnableRectInter));
+    H0("   --[no-]amp                    Enable asymmetric motion partitions, requires --rect. Default %s\n", OPT(param->bEnableAMP));
+    H0("\nAnalysis:\n");
+    H0("   --rd <0..6>                   Level of RD in mode decision 0:least....6:full RDO. Default %d\n", param->rdLevel);
+    H0("   --psy-rd <0..2.0>             Strength of psycho-visual rate distortion optimization, 0 to disable. Default %f\n", param->psyRd);
+    H0("   --psy-rdoq <0..10.0>          Strength of psycho-visual optimization in quantization, 0 to disable. Default %f\n", param->psyRdoq);
+    H0("   --nr <integer>                An integer value in range of 100 to 1000, which denotes strength of noise reduction. Default disabled\n");
+    H0("   --[no-]tskip-fast             Enable fast intra transform skipping. Default %s\n", OPT(param->bEnableTSkipFast));
+    H0("   --[no-]early-skip             Enable early SKIP detection. Default %s\n", OPT(param->bEnableEarlySkip));
+    H0("   --[no-]fast-cbf               Enable early outs based on whether residual is coded. Default %s\n", OPT(param->bEnableCbfFastMode));
+    H0("\nCoding tools:\n");
+    H0("-w/--[no-]weightp                Enable weighted prediction in P slices. Default %s\n", OPT(param->bEnableWeightedPred));
+    H0("   --[no-]weightb                Enable weighted prediction in B slices. Default %s\n", OPT(param->bEnableWeightedBiPred));
+    H0("   --[no-]cu-lossless            Consider lossless mode in CU RDO decisions. Default %s\n", OPT(param->bCULossless));
+    H0("   --[no-]signhide               Hide sign bit of one coeff per TU (rdo). Default %s\n", OPT(param->bEnableSignHiding));
+    H0("   --[no-]tskip                  Enable intra 4x4 transform skipping. Default %s\n", OPT(param->bEnableTransformSkip));
     H0("\nTemporal / motion search options:\n");
     H0("   --me <string>                 Motion search method dia hex umh star full. Default %d\n", param->searchMethod);
     H0("-m/--subme <integer>             Amount of subpel refinement to perform (0:least .. 7:most). Default %d \n", param->subpelRefine);
     H0("   --merange <integer>           Motion search range. Default %d\n", param->searchRange);
-    H0("   --[no-]rect                   Enable rectangular motion partitions Nx2N and 2NxN. Default %s\n", OPT(param->bEnableRectInter));
-    H0("   --[no-]amp                    Enable asymmetric motion partitions, requires --rect. Default %s\n", OPT(param->bEnableAMP));
     H0("   --max-merge <1..5>            Maximum number of merge candidates. Default %d\n", param->maxNumMergeCand);
-    H0("   --[no-]early-skip             Enable early SKIP detection. Default %s\n", OPT(param->bEnableEarlySkip));
-    H0("   --[no-]fast-cbf               Enable Cbf fast mode. Default %s\n", OPT(param->bEnableCbfFastMode));
     H0("\nSpatial / intra options:\n");
-    H0("   --rdpenalty <0..2>            penalty for 32x32 intra TU in non-I slices. 0:disabled 1:RD-penalty 2:maximum. Default %d\n", param->rdPenalty);
-    H0("   --[no-]tskip                  Enable intra transform skipping. Default %s\n", OPT(param->bEnableTransformSkip));
-    H0("   --[no-]tskip-fast             Enable fast intra transform skipping. Default %s\n", OPT(param->bEnableTSkipFast));
     H0("   --[no-]strong-intra-smoothing Enable strong intra smoothing for 32x32 blocks. Default %s\n", OPT(param->bEnableStrongIntraSmoothing));
     H0("   --[no-]constrained-intra      Constrained intra prediction (use only intra coded reference pixels) Default %s\n", OPT(param->bEnableConstrainedIntra));
+    H0("   --[no-]b-intra                Enable intra in B frames in veryslow presets. Default %s\n", OPT(param->bIntraInBFrames));
+    H0("   --[no-]fast-intra             Enable faster search method for angular intra predictions. Default %s\n", OPT(param->bEnableFastIntra));
+    H0("   --rdpenalty <0..2>            penalty for 32x32 intra TU in non-I slices. 0:disabled 1:RD-penalty 2:maximum. Default %d\n", param->rdPenalty);
     H0("\nSlice decision options:\n");
     H0("   --[no-]open-gop               Enable open-GOP, allows I slices to be non-IDR. Default %s\n", OPT(param->bOpenGOP));
     H0("-I/--keyint <integer>            Max IDR period in frames. -1 for infinite-gop. Default %d\n", param->keyframeMax);
@@ -351,30 +417,48 @@
     H0("   --b-adapt <0..2>              0 - none, 1 - fast, 2 - full (trellis) adaptive B frame scheduling. Default %d\n", param->bFrameAdaptive);
     H0("   --[no-]b-pyramid              Use B-frames as references. Default %s\n", OPT(param->bBPyramid));
     H0("   --ref <integer>               max number of L0 references to be allowed (1 .. 16) Default %d\n", param->maxNumReferences);
-    H0("-w/--[no-]weightp                Enable weighted prediction in P slices. Default %s\n", OPT(param->bEnableWeightedPred));
-    H0("\nRate control and rate distortion options:\n");
-    H0("   --bitrate <integer>           Target bitrate (kbps), implies ABR. Default %d\n", param->rc.bitrate);
+    H0("   --qpfile <string>             Force frametypes and QPs for some or all frames\n");
+    H0("                                 Format of each line: framenumber frametype QP\n");
+    H0("                                 QP is optional (none lets x265 choose). Frametypes: I,i,P,B,b.\n");
+    H0("                                 QPs are restricted by qpmin/qpmax.\n");
+    H0("\nRate control, Quantization:\n");
+    H0("   --bitrate <integer>           Target bitrate (kbps) for ABR (implied). Default %d\n", param->rc.bitrate);
+    H0("-q/--qp <integer>                QP for P slices in CQP mode (implied). --ipratio and --pbration determine other slice QPs\n");
     H0("   --crf <float>                 Quality-based VBR (0-51). Default %f\n", param->rc.rfConstant);
-    H0("   --crf-max <float>             With CRF+VBV, limit RF to this value. 0 for no limit (default)\n");
-    H0("                                  May cause VBV underflows!\n");
+    H0("   --[no-]lossless               Enable lossless: bypass transform, quant and loop filters globally. Default %s\n", OPT(param->bLossless));
+    H0("   --crf-max <float>             With CRF+VBV, limit RF to this value. Default %f\n", param->rc.rfConstantMax);
+    H0("                                 May cause VBV underflows!\n");
+    H0("   --crf-min <float>             With CRF+VBV, limit RF to this value. Default %f\n", param->rc.rfConstantMin);
+    H0("                                 this specifies a minimum rate factor value for encode!\n");
     H0("   --vbv-maxrate <integer>       Max local bitrate (kbit/s). Default %d\n", param->rc.vbvMaxBitrate);
     H0("   --vbv-bufsize <integer>       Set size of the VBV buffer (kbit). Default %d\n", param->rc.vbvBufferSize);
     H0("   --vbv-init <float>            Initial VBV buffer occupancy (fraction of bufsize or in kbits). Default %f\n", param->rc.vbvBufferInit);
-    H0("-q/--qp <integer>                Base QP for CQP mode. Default %d\n", param->rc.qp);
     H0("   --aq-mode <integer>           Mode for Adaptive Quantization - 0:none 1:uniform AQ 2:auto variance. Default %d\n", param->rc.aqMode);
     H0("   --aq-strength <float>         Reduces blocking and blurring in flat and textured areas.(0 to 3.0). Default %f\n", param->rc.aqStrength);
     H0("   --[no-]cutree                 Enable cutree for Adaptive Quantization. Default %s\n", OPT(param->rc.cuTree));
+    H0("   --ipratio <float>             QP factor between I and P. Default %f\n", param->rc.ipFactor);
+    H0("   --pbratio <float>             QP factor between P and B. Default %f\n", param->rc.pbFactor);
     H0("   --cbqpoffs <integer>          Chroma Cb QP Offset. Default %d\n", param->cbQpOffset);
     H0("   --crqpoffs <integer>          Chroma Cr QP Offset. Default %d\n", param->crQpOffset);
-    H0("   --rd <0..6>                   Level of RD in mode decision 0:least....6:full RDO. Default %d\n", param->rdLevel);
-    H0("   --[no-]signhide               Hide sign bit of one coeff per TU (rdo). Default %s\n", OPT(param->bEnableSignHiding));
+    H0("   --stats                       Filename for stats file in multipass pass rate control. Default x265_2pass.log\n");
+    H0("   --pass                        Multi pass rate control.\n"
+       "                                   - 1 : First pass, creates stats file\n"
+       "                                   - 2 : Last pass, does not overwrite stats file\n"
+       "                                   - 3 : Nth pass, overwrites stats file\n");
+    H0("   --[no-]slow-firstpass         Enable a slow first pass in a multipass rate control mode. Default %s\n", OPT(param->rc.bEnableSlowFirstPass));
+    H0("   --analysis-mode <string|int>  save - Dump analysis info into file, load - Load analysis buffers from the file. Default %d\n", param->analysisMode);
+    H0("   --analysis-file <filename>    Specify file name used for either dumping or reading analysis data.\n");
+    H0("   --scaling-list <string>       Specify a file containing HM style quant scaling lists or 'default' or 'off'. Default: off\n");
+    H0("   --lambda-file <string>        Specify a file containing replacement values for the lambda tables\n");
+    H0("                                 MAX_MAX_QP+1 floats for lambda table, then again for lambda2 table\n");
+    H0("                                 Blank lines and lines starting with hash(#) are ignored\n");
+    H0("                                 Comma is considered to be white-space\n");
     H0("\nLoop filters (deblock and SAO):\n");
     H0("   --[no-]lft                    Enable Deblocking Loop Filter. Default %s\n", OPT(param->bEnableLoopFilter));
     H0("   --[no-]sao                    Enable Sample Adaptive Offset. Default %s\n", OPT(param->bEnableSAO));
     H0("   --sao-lcu-bounds <integer>    0: right/bottom boundary areas skipped  1: non-deblocked pixels are used. Default %d\n", param->saoLcuBoundary);
     H0("   --sao-lcu-opt <integer>       0: SAO picture-based optimization, 1: SAO LCU-based optimization. Default %d\n", param->saoLcuBasedOptimization);
     H0("\nVUI options:\n");
-    H0("   --[no-]vui                    Add Video Useability Information with all fields to the SPS. Default %s\n", OPT(param->vui.bEnableVuiParametersPresentFlag));
     H0("   --sar <width:height|int>      Sample Aspect Ratio, the ratio of width to height of an individual pixel.\n");
     H0("                                 Choose from 0=undef, 1=1:1(\"square\"), 2=12:11, 3=10:11, 4=16:11,\n");
     H0("                                 5=40:33, 6=24:11, 7=20:11, 8=32:11, 9=80:33, 10=18:11, 11=15:11,\n");
@@ -391,18 +475,18 @@
     H0("   --colormatrix <string>        Specify color matrix setting from undef, bt709, fcc, bt470bg, smpte170m,\n");
     H0("                                 smpte240m, GBR, YCgCo, bt2020nc, bt2020c. Default undef\n");
     H0("   --chromaloc <integer>         Specify chroma sample location (0 to 5). Default of %d\n", param->vui.chromaSampleLocTypeTopField);
-    H0("   --[no-]timinginfo             Add timing information to the VUI. Defaut %s\n", OPT(param->vui.bEnableVuiTimingInfoPresentFlag));
-    H0("   --[no-]hrd                    Signal HRD information. Default %s\n", OPT(param->vui.bEnableVuiHrdParametersPresentFlag));
-    H0("   --nal-hrd <string>            Signal NAL HRD information (requires vbv-buffer size) Choose from none, vbr or cbr.Default none\n");
-    H0("   --[no-]bitstreamrestriction   Add bit stream restriction fields to the VUI. Default %s\n", OPT(param->vui.bEnableBitstreamRestrictionFlag));
-    H0("   --[no-]subpichrd              Add sub picture HRD parameters to the HRD. Default %s\n", OPT(param->vui.bEnableSubPicHrdParamsPresentFlag));
+    H0("\nBitstream options:\n");
+    H0("   --[no-]info                   Emit SEI identifying encoder and parameters. Default %s\n", OPT(param->bEmitInfoSEI));
+    H0("   --[no-]aud                    Emit access unit delimiters at the start of each access unit. Default %s\n", OPT(param->bEnableAccessUnitDelimiters));
+    H0("   --[no-]hrd                    Enable HRD parameters signalling. Default %s\n", OPT(param->bEmitHRDSEI));
+    H0("   --[no-]repeat-headers         Emit SPS and PPS headers at each keyframe. Default %s\n", OPT(param->bRepeatHeaders));
+    H0("   --hash <integer>              Decoded Picture Hash SEI 0: disabled, 1: MD5, 2: CRC, 3: Checksum. Default %d\n", param->decodedPictureHashSEI);
     H0("\nReconstructed video options (debugging):\n");
     H0("-r/--recon <filename>            Reconstructed raw image YUV or Y4M output file name\n");
     H0("   --recon-depth <integer>       Bit-depth of reconstructed raw image file. Defaults to input bit depth, or 8 if Y4M\n");
-    H0("\nSEI options:\n");
-    H0("   --hash <integer>              Decoded Picture Hash SEI 0: disabled, 1: MD5, 2: CRC, 3: Checksum. Default %d\n", param->decodedPictureHashSEI);
 #undef OPT
 #undef H0
+    printf("\n\nFull documentation may be found at http://x265.readthedocs.org/en/default/cli.html\n");
     exit(0);
 }
 
@@ -415,8 +499,10 @@
     const char *inputfn = NULL;
     const char *reconfn = NULL;
     const char *bitstreamfn = NULL;
-    const char *preset = "medium";
-    const char *tune = "ssim";
+    const char *preset = NULL;
+    const char *tune = NULL;
+    const char *profile = NULL;
+    const char *analysisfn = "x265_analysis.dat";
 
     if (argc <= 1)
     {
@@ -502,12 +588,23 @@
             OPT("output") bitstreamfn = optarg;
             OPT("input") inputfn = optarg;
             OPT("recon") reconfn = optarg;
-            OPT("input-res") bError |= sscanf(optarg, "%dx%d", &param->sourceWidth, &param->sourceHeight) != 2;
             OPT("input-depth") inputBitDepth = (uint32_t)x265_atoi(optarg, bError);
+            OPT("dither") this->bDither = true;
             OPT("recon-depth") reconFileBitDepth = (uint32_t)x265_atoi(optarg, bError);
-            OPT("y4m") bForceY4m = true;
+            OPT("y4m") this->bForceY4m = true;
+            OPT("profile") profile = optarg; /* handled last */
             OPT("preset") /* handled above */;
             OPT("tune")   /* handled above */;
+            OPT("analysis-file") analysisfn = optarg;
+            OPT("qpfile")
+            {
+                this->qpfile = fopen(optarg, "rb");
+                if (!this->qpfile)
+                {
+                    x265_log(param, X265_LOG_ERROR, "%s qpfile not found or error in opening qp file \n", optarg);
+                    return false;
+                }
+            }
             else
                 bError |= !!x265_param_parse(param, long_options[long_options_index].name, optarg);
 
@@ -568,17 +665,13 @@
     info.frameCount = 0;
     getParamAspectRatio(param, info.sarWidth, info.sarHeight);
 
-    this->input = Input::open(info, bForceY4m);
+    this->input = Input::open(info, this->bForceY4m);
     if (!this->input || this->input->isFail())
     {
         x265_log(param, X265_LOG_ERROR, "unable to open input file <%s>\n", inputfn);
         return true;
     }
-    if (info.csp != X265_CSP_I420 && info.csp != X265_CSP_I444)
-    {
-        x265_log(param, X265_LOG_ERROR, "Only i420 and i444 color spaces are supported in this build\n");
-        return true;
-    }
+
     if (info.depth < 8 || info.depth > 16)
     {
         x265_log(param, X265_LOG_ERROR, "Input bit depth (%d) must be between 8 and 16\n", inputBitDepth);
@@ -596,16 +689,20 @@
         param->fpsDenom = info.fpsDenom;
         param->fpsNum = info.fpsNum;
     }
-    if (!param->vui.bEnableAspectRatioIdc && info.sarWidth && info.sarHeight)
+    if (!param->vui.aspectRatioIdc && info.sarWidth && info.sarHeight)
         setParamAspectRatio(param, info.sarWidth, info.sarHeight);
-
     if (this->framesToBeEncoded == 0 && info.frameCount > (int)seek)
         this->framesToBeEncoded = info.frameCount - seek;
+    param->totalFrames = this->framesToBeEncoded;
+
+    if (x265_param_apply_profile(param, profile))
+        return true;
+
     if (param->logLevel >= X265_LOG_INFO)
     {
         char buf[128];
-        int p = sprintf(buf, "%dx%d fps %d/%d %s", param->sourceWidth, param->sourceHeight,
-                        param->fpsNum, param->fpsDenom, x265_source_csp_names[param->internalCsp]);
+        int p = sprintf(buf, "%dx%d fps %d/%d %sp%d", param->sourceWidth, param->sourceHeight,
+                        param->fpsNum, param->fpsDenom, x265_source_csp_names[param->internalCsp], info.depth);
 
         int width, height;
         getParamAspectRatio(param, width, height);
@@ -647,10 +744,196 @@
         return true;
     }
 
-    printVersion(param);
+    if (param->analysisMode)
+    {
+        const char *mode = param->analysisMode == X265_ANALYSIS_SAVE ? "wb" : "rb";
+        this->analysisFile = fopen(analysisfn, mode);
+        if (!this->analysisFile)
+        {
+            x265_log(NULL, X265_LOG_ERROR, "failed to open analysis file %s\n", analysisfn);
+            return true;
+        }
+    }
+
     return false;
 }
 
+bool CLIOptions::validateFanout(x265_param *param)
+{
+#define CMP_OPT_FANOUT(opt, param_val)\
+    {\
+        bErr = 0;\
+        p = strstr(paramBuf, opt "=");\
+        char* q = strstr(paramBuf, "no-"opt);\
+        if (p && sscanf(p, opt "=%d" , &i) && param_val != i)\
+            bErr = 1;\
+        else if (!param_val && !q)\
+            bErr = 1;\
+        else if (param_val && (q || !strstr(paramBuf, opt)))\
+            bErr = 1;\
+        if (bErr)\
+        {\
+            x265_log(param, X265_LOG_ERROR, "different " opt " setting than given in analysis file (%d vs %d)\n", param_val, i);\
+            X265_FREE(paramBuf);\
+            return false;\
+        }\
+    }
+
+    char *p = NULL, *paramBuf;
+    int i, j;
+    uint32_t k , l;
+    bool bErr = false;
+
+    paramBuf = X265_MALLOC(char, MAXPARAMSIZE);
+    if (!paramBuf)
+        return false;
+
+    fread(paramBuf, 1, MAXPARAMSIZE, this->analysisFile);
+
+    /* check whether fanout options are compatible */
+    if (strncmp(paramBuf, "#options:", 9))
+    {
+        x265_log(param, X265_LOG_ERROR, "options list in analysis file is not valid\n");
+        X265_FREE(paramBuf);
+        return false;
+    }
+
+    char* buf = strchr(paramBuf, '\n');
+    if (!buf)
+    {
+        x265_log(param, X265_LOG_ERROR, "Malformed analysis file\n");
+        X265_FREE(paramBuf);
+        return false;
+    }
+    *buf = '\0';
+    fseek(this->analysisFile, (int)strlen(paramBuf) + 1, SEEK_SET);
+
+    if (sscanf(paramBuf, "#options: %dx%d", &i, &j) != 2)
+    {
+        x265_log(param, X265_LOG_ERROR, "Resolution specified in analysis file is not valid\n");
+        X265_FREE(paramBuf);
+        return false;
+    }
+    if ((p = strstr(paramBuf, " fps=")) == 0 || sscanf(p, " fps=%u/%u", &k, &l) != 2)
+    {
+        x265_log(param, X265_LOG_ERROR, "fps specified in analysis file is not valid\n");
+        X265_FREE(paramBuf);
+        return false;
+    }
+    if (k != param->fpsNum || l != param->fpsDenom)
+    {
+        x265_log(param, X265_LOG_ERROR, "fps mismatch than given in analysis file (%u/%u vs %u/%u)\n",
+            param->fpsNum, param->fpsDenom, k, l);
+        X265_FREE(paramBuf);
+        return false;
+    }
+
+    CMP_OPT_FANOUT("bitdepth", param->internalBitDepth);
+    CMP_OPT_FANOUT("weightp", param->bEnableWeightedPred);
+    CMP_OPT_FANOUT("bframes", param->bframes);
+    CMP_OPT_FANOUT("b-pyramid", param->bBPyramid);
+    CMP_OPT_FANOUT("b-adapt", param->bFrameAdaptive);
+    CMP_OPT_FANOUT("open-gop", param->bOpenGOP);
+    CMP_OPT_FANOUT("keyint", param->keyframeMax);
+    CMP_OPT_FANOUT("min-keyint", param->keyframeMin);
+    CMP_OPT_FANOUT("scenecut", param->scenecutThreshold);
+    CMP_OPT_FANOUT("ctu", (int)param->maxCUSize);
+    CMP_OPT_FANOUT("ref", param->maxNumReferences);
+    CMP_OPT_FANOUT("rc-lookahead", param->lookaheadDepth);
+
+#undef CMP_OPT_FANOUT
+
+    X265_FREE(paramBuf);
+    return true;
+}
+
+void CLIOptions::readAnalysisFile(x265_picture* pic, x265_param* p)
+{
+    int poc, width, height;
+    uint32_t numPart, numCU;
+    fread(&width, sizeof(int), 1, this->analysisFile);
+    fread(&height, sizeof(int), 1, this->analysisFile);
+    fread(&poc, sizeof(int), 1, this->analysisFile);
+    fread(&pic->sliceType, sizeof(int), 1, this->analysisFile);
+    fread(&numCU, sizeof(int), 1, this->analysisFile);
+    fread(&numPart, sizeof(int), 1, this->analysisFile);
+
+    if (poc != pic->poc || width != p->sourceWidth || height != p->sourceHeight)
+    {
+        x265_log(NULL, X265_LOG_WARNING, "Error in reading intra-inter data.\n");
+        x265_free_analysis_data(pic);
+        return;
+    }
+
+    fread(pic->analysisData.intraData->depth,
+        sizeof(uint8_t), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fread(pic->analysisData.intraData->modes,
+        sizeof(uint8_t), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fread(pic->analysisData.intraData->partSizes,
+        sizeof(char), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fread(pic->analysisData.intraData->poc,
+        sizeof(int), pic->analysisData.numCUsInFrame, this->analysisFile);
+    fread(pic->analysisData.intraData->cuAddr,
+        sizeof(uint32_t), pic->analysisData.numCUsInFrame, this->analysisFile);
+    fread(pic->analysisData.interData, sizeof(x265_inter_data), pic->analysisData.numCUsInFrame * 85, this->analysisFile);
+}
+
+void CLIOptions::writeAnalysisFile(x265_picture* pic, x265_param *p)
+{
+    uint64_t seekTo = pic->poc * this->analysisRecordSize + this->analysisHeaderSize;
+    fseeko(this->analysisFile, seekTo, SEEK_SET);
+    fwrite(&p->sourceWidth, sizeof(int), 1, this->analysisFile);
+    fwrite(&p->sourceHeight, sizeof(int), 1, this->analysisFile);
+    fwrite(&pic->poc, sizeof(int), 1, this->analysisFile);
+    fwrite(&pic->sliceType, sizeof(int), 1, this->analysisFile);
+    fwrite(&pic->analysisData.numCUsInFrame, sizeof(int), 1, this->analysisFile);
+    fwrite(&pic->analysisData.numPartitions, sizeof(int), 1, this->analysisFile);
+
+    fwrite(pic->analysisData.intraData->depth,
+        sizeof(uint8_t), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fwrite(pic->analysisData.intraData->modes,
+        sizeof(uint8_t), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fwrite(pic->analysisData.intraData->partSizes,
+        sizeof(char), pic->analysisData.numPartitions * pic->analysisData.numCUsInFrame, this->analysisFile);
+    fwrite(pic->analysisData.intraData->poc, sizeof(int), pic->analysisData.numCUsInFrame, this->analysisFile);
+    fwrite(pic->analysisData.intraData->cuAddr, sizeof(uint32_t), pic->analysisData.numCUsInFrame, this->analysisFile);
+    fwrite(pic->analysisData.interData, sizeof(x265_inter_data), pic->analysisData.numCUsInFrame * 85, this->analysisFile);
+}
+
+bool CLIOptions::parseQPFile(x265_picture &pic_org)
+{
+    int32_t num = -1, qp, ret;
+    char type;
+    uint32_t filePos;
+    pic_org.forceqp = 0;
+    pic_org.sliceType = X265_TYPE_AUTO;
+    while (num < pic_org.poc)
+    {
+        filePos = ftell(qpfile);
+        qp = -1;
+        ret = fscanf(qpfile, "%d %c%*[ \t]%d\n", &num, &type, &qp);
+
+        if (num > pic_org.poc || ret == EOF)
+        {
+            fseek(qpfile, filePos, SEEK_SET);
+            break;
+        }
+        if (num < pic_org.poc && ret >= 2)
+            continue;
+        if (ret == 3 && qp >= 0)
+            pic_org.forceqp = qp + 1;
+        if (type == 'I') pic_org.sliceType = X265_TYPE_IDR;
+        else if (type == 'i') pic_org.sliceType = X265_TYPE_I;
+        else if (type == 'P') pic_org.sliceType = X265_TYPE_P;
+        else if (type == 'B') pic_org.sliceType = X265_TYPE_BREF;
+        else if (type == 'b') pic_org.sliceType = X265_TYPE_B;
+        else ret = 0;
+        if (ret < 2 || qp < -1 || qp > 51)
+            return 0;
+    }
+    return 1;
+}
+
 int main(int argc, char **argv)
 {
 #if HAVE_VLD
@@ -665,6 +948,7 @@
     if (cliopt.parse(argc, argv, param))
     {
         cliopt.destroy();
+        x265_param_free(param);
         exit(1);
     }
 
@@ -673,10 +957,14 @@
     {
         x265_log(param, X265_LOG_ERROR, "failed to open encoder\n");
         cliopt.destroy();
+        x265_param_free(param);
         x265_cleanup();
         exit(1);
     }
 
+    /* get the encoder parameters post-initialization */
+    x265_encoder_parameters(encoder, param);
+
     /* Control-C handler */
     if (signal(SIGINT, sigint_handler) == SIG_ERR)
         x265_log(param, X265_LOG_ERROR, "Unable to register CTRL+C handler: %s\n", strerror(errno));
@@ -684,23 +972,80 @@
     x265_picture pic_orig, pic_out;
     x265_picture *pic_in = &pic_orig;
     x265_picture *pic_recon = cliopt.recon ? &pic_out : NULL;
+    uint32_t inFrameCount = 0;
+    uint32_t outFrameCount = 0;
     x265_nal *p_nal;
     x265_stats stats;
     uint32_t nal;
+    int16_t *errorBuf = NULL;
 
-    if (!x265_encoder_headers(encoder, &p_nal, &nal))
+    if (!param->bRepeatHeaders)
     {
-        cliopt.writeNALs(p_nal, nal);
+        if (x265_encoder_headers(encoder, &p_nal, &nal) < 0)
+        {
+            x265_log(param, X265_LOG_ERROR, "Failure generating stream headers\n");
+            goto fail;
+        }
+        else
+            cliopt.writeNALs(p_nal, nal);
     }
 
     x265_picture_init(param, pic_in);
 
+    if (param->analysisMode && !pic_recon)
+    {
+        x265_log(NULL, X265_LOG_ERROR, "Must specify recon with analysis-mode option.\n");
+        goto fail;
+    }
+    if (param->analysisMode)
+    {
+        if (param->analysisMode == X265_ANALYSIS_SAVE)
+        {
+            char *p = x265_param2string(param);
+            if (!p)
+            {
+                x265_log(NULL, X265_LOG_ERROR, "analysis: buffer allocation failure, aborting");
+                goto fail;
+            }
+            uint32_t numCU = pic_in->analysisData.numCUsInFrame;
+            uint32_t numPart = pic_in->analysisData.numPartitions;
+
+            cliopt.analysisRecordSize = ((sizeof(int) * 4 + sizeof(uint32_t) * 2) + sizeof(x265_inter_data) * numCU * 85 +
+                    sizeof(uint8_t) * 2 * numPart * numCU + sizeof(char) * numPart * numCU + sizeof(int) * numCU + sizeof(uint32_t) * numCU);
+
+            fprintf(cliopt.analysisFile, "#options: %s\n", p);
+            cliopt.analysisHeaderSize = ftell(cliopt.analysisFile);
+            X265_FREE(p);
+        }
+        else
+        {
+            if (!cliopt.validateFanout(param))
+                goto fail;
+        }
+    }
+
+    if (cliopt.bDither)
+    {
+        errorBuf = X265_MALLOC(int16_t, param->sourceWidth + 1);
+        if (errorBuf)
+            memset(errorBuf, 0, (param->sourceWidth + 1) * sizeof(int16_t));
+        else
+            cliopt.bDither = false;
+    }
+
     // main encoder loop
-    uint32_t inFrameCount = 0;
-    uint32_t outFrameCount = 0;
     while (pic_in && !b_ctrl_c)
     {
         pic_orig.poc = inFrameCount;
+        if (cliopt.qpfile && !param->rc.bStatRead)
+        {
+            if (!cliopt.parseQPFile(pic_orig))
+            {
+                x265_log(NULL, X265_LOG_ERROR, "can't parse qpfile for frame %d\n", pic_in->poc);
+                fclose(cliopt.qpfile);
+                cliopt.qpfile = NULL;
+            }
+        }
 
         if (cliopt.framesToBeEncoded && inFrameCount >= cliopt.framesToBeEncoded)
             pic_in = NULL;
@@ -709,6 +1054,22 @@
         else
             pic_in = NULL;
 
+        if (pic_in)
+        {
+            if (pic_in->bitDepth > X265_DEPTH && cliopt.bDither)
+            {
+                ditherImage(*pic_in, param->sourceWidth, param->sourceHeight, errorBuf, X265_DEPTH);
+                pic_in->bitDepth = X265_DEPTH;
+            }
+            if (param->analysisMode)
+            {
+                x265_alloc_analysis_data(pic_in);
+
+                if (param->analysisMode == X265_ANALYSIS_LOAD)
+                    cliopt.readAnalysisFile(pic_in, param);
+            }
+        }
+
         int numEncoded = x265_encoder_encode(encoder, &p_nal, &nal, pic_in, pic_recon);
         if (numEncoded < 0)
         {
@@ -719,6 +1080,10 @@
         if (numEncoded && pic_recon)
         {
             cliopt.recon->writePicture(pic_out);
+            if (param->analysisMode == X265_ANALYSIS_SAVE)
+                cliopt.writeAnalysisFile(pic_recon, param);
+            if (param->analysisMode)
+                x265_free_analysis_data(pic_recon);
         }
 
         if (nal)
@@ -736,6 +1101,10 @@
         if (numEncoded && pic_recon)
         {
             cliopt.recon->writePicture(pic_out);
+            if (param->analysisMode == X265_ANALYSIS_SAVE)
+                cliopt.writeAnalysisFile(pic_recon, param);
+            if (param->analysisMode)
+                x265_free_analysis_data(pic_recon);
         }
 
         if (nal)
@@ -751,6 +1120,7 @@
     if (cliopt.bProgress)
         fprintf(stderr, "%*s\r", 80, " ");
 
+fail:
     x265_encoder_get_stats(encoder, &stats, sizeof(stats));
     if (param->csvfn && !b_ctrl_c)
         x265_encoder_log(encoder, argc, argv);
@@ -785,6 +1155,8 @@
 
     x265_param_free(param);
 
+    X265_FREE(errorBuf);
+
 #if HAVE_VLD
     assert(VLDReportLeaks() == 0);
 #endif

x265-hg20140324.tar.bz2/source/x265.def.in -> x265-hg20140928.tar.bz2/source/x265.def.in Changed

x265-hg20140324.tar.bz2/source/x265.h -> x265-hg20140928.tar.bz2/source/x265.h Changed

x265-hg20140324.tar.bz2/source/x265.rc.in -> x265-hg20140928.tar.bz2/source/x265.rc.in Changed

x265-hg20140324.tar.bz2/source/x265_config.h.in -> x265-hg20140928.tar.bz2/source/x265_config.h.in Changed