Request 5292 (accepted)

Overview

Request 5292 (accepted)

- Update to version 26.1.1:
* win-dshow: Fix dshowcapture not linking audio of certain devices
* linux-jack: fix deadlock when closing the client
* linux-jack: mark ports as JackPortIsTerminal
* linux-jack: fix timestamp calculation
* obs-browser: Initialize CEF early to fix macOS crash
* libobs: Update version to 26.1.1
* rtmp-services: Add Loola.tv service
* rtmp-services: Fix json formatting
* libobs: Avoid unnecessary mallocs in audio processing
* UI: Fix padding on Acri context bar buttons
* image-source: Fix slideshow transition bug when randomized
* docs/sphinx: Add missing obs_frontend_open_projector
* libobs: Update to SIMDe 0.7.1
* libobs: Set lock state when duplicating scene item
* libobs: Add definitions in ARCH_SIMD_DEFINES
* cmake: Add ARCH_SIMD_DEFINES variable
* coreaudio-encoder: Fix cmake for mingw
* Revert "UI: Only apply new scaling behavior on newer installs"
* UI: Only apply new scaling behavior on newer installs
* UI: Support fractional scaling for Canvas/Base size
* mac-virtualcam: Remove unnecessary logging
* mac-virtualcam: Mark parameters as unused
* image-source: Add .webp to "All formats" option
* image-source: Add webp to file filter
* CI: Remove jack, speex and fdk-aac from default builds for macOS
* libobs, obs-ffmpeg: Use correct value for EINVAL error check
* UI/updater: Increase number of download workers
* UI/updater: Enable HTTP2 and TLS 1.3
* UI: Fix name of kab-KAB locale

Created by boombatower almost 4 years ago
In state accepted
Package maintainers: boombatower and frispete

Submit obs-studio

Submit package home:boombatow...es:Multimedia / obs-studio to package Multimedia / obs-studio

obs-studio.changes Changed

@@ -1,4 +1,50 @@
 -------------------------------------------------------------------
+Wed Jan 06 18:27:38 UTC 2021 - jimmy@boombatower.com
+
+- Update to version 26.1.1:
+ * win-dshow: Fix dshowcapture not linking audio of certain devices
+ * linux-jack: fix deadlock when closing the client
+ * linux-jack: mark ports as JackPortIsTerminal
+ * linux-jack: fix timestamp calculation
+ * obs-browser: Initialize CEF early to fix macOS crash
+ * libobs: Update version to 26.1.1
+ * rtmp-services: Add Loola.tv service
+ * rtmp-services: Fix json formatting
+ * libobs: Avoid unnecessary mallocs in audio processing
+ * UI: Fix padding on Acri context bar buttons
+ * image-source: Fix slideshow transition bug when randomized
+ * docs/sphinx: Add missing obs_frontend_open_projector
+ * libobs: Update to SIMDe 0.7.1
+ * libobs: Set lock state when duplicating scene item
+ * libobs: Add definitions in ARCH_SIMD_DEFINES
+ * cmake: Add ARCH_SIMD_DEFINES variable
+ * coreaudio-encoder: Fix cmake for mingw
+ * Revert "UI: Only apply new scaling behavior on newer installs"
+ * UI: Only apply new scaling behavior on newer installs
+ * UI: Support fractional scaling for Canvas/Base size
+ * mac-virtualcam: Remove unnecessary logging
+ * mac-virtualcam: Mark parameters as unused
+ * image-source: Add .webp to "All formats" option
+ * image-source: Add webp to file filter
+ * CI: Remove jack, speex and fdk-aac from default builds for macOS
+ * libobs, obs-ffmpeg: Use correct value for EINVAL error check
+ * UI/updater: Increase number of download workers
+ * UI/updater: Enable HTTP2 and TLS 1.3
+ * UI: Fix name of kab-KAB locale
+ * decklink: Fix automatic pixel format detection
+ * CI: Fix macOS 10.13 crashes due to unsupported library symbols
+ * UI/installer: Add additional VS2019 DLL check
+ * mac-virtualcam: Fix file mode
+ * CI: Run make with -j$(nproc)
+ * CI: Remove obsolete and unused files
+ * libobs: Add texture sharing support for macOS/OpenGL
+ * CI: Add necessary changes for CEF 4183
+ * UI/updater: Move in-use files away before writing
+ * UI/updater: Always clean up temporary files
+ * UI: Remove Smashcast from AutoConfig
+ * rtmp-services: Remove Smashcast
+
+-------------------------------------------------------------------
 Tue Dec 15 23:25:38 UTC 2020 - Jimmy Berry <jimmy@boombatower.com>
 
 - Add modinfo-use-full-path.patch for new v4l2lookback support.

obs-studio.spec Changed

_service Changed

_servicedata Changed

obs-studio-26.1.0.tar.xz/CI/before-deploy-osx.sh Deleted

@@ -1,114 +0,0 @@
-hr() {
-  echo "───────────────────────────────────────────────────"
-  echo $1
-  echo "───────────────────────────────────────────────────"
-}
-
-# Exit if something fails
-set -e
-
-# Generate file name variables
-export GIT_TAG=$(git describe --abbrev=0)
-export GIT_HASH=$(git rev-parse --short HEAD)
-export FILE_DATE=$(date +%Y-%m-%d.%H-%M-%S)
-export FILENAME=$FILE_DATE-$GIT_HASH-$TRAVIS_BRANCH-osx.dmg
-
-echo "git tag: $GIT_TAG"
-
-cd ./build
-
-# Move obslua
-hr "Moving OBS LUA"
-mv ./rundir/RelWithDebInfo/data/obs-scripting/obslua.so ./rundir/RelWithDebInfo/bin/
-
-# Move obspython
-hr "Moving OBS Python"
-# mv ./rundir/RelWithDebInfo/data/obs-scripting/_obspython.so ./rundir/RelWithDebInfo/bin/
-# mv ./rundir/RelWithDebInfo/data/obs-scripting/obspython.py ./rundir/RelWithDebInfo/bin/
-
-# Package everything into a nice .app
-hr "Packaging .app"
-STABLE=false
-if [ -n "${TRAVIS_TAG}" ]; then
-  STABLE=true
-fi
-
-#sudo python ../CI/install/osx/build_app.py --public-key ../CI/install/osx/OBSPublicDSAKey.pem --sparkle-framework ../../sparkle/Sparkle.framework --stable=$STABLE
-
-../CI/install/osx/packageApp.sh
-
-# fix obs outputs plugin it doesn't play nicely with dylibBundler at the moment
-if [ -f /usr/local/opt/mbedtls/lib/libmbedtls.12.dylib ]; then
-    cp /usr/local/opt/mbedtls/lib/libmbedtls.12.dylib ./OBS.app/Contents/Frameworks/
-    cp /usr/local/opt/mbedtls/lib/libmbedcrypto.3.dylib ./OBS.app/Contents/Frameworks/
-    cp /usr/local/opt/mbedtls/lib/libmbedx509.0.dylib ./OBS.app/Contents/Frameworks/
-    chmod +w ./OBS.app/Contents/Frameworks/*.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedtls.12.dylib ./OBS.app/Contents/Frameworks/libmbedtls.12.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedcrypto.3.dylib ./OBS.app/Contents/Frameworks/libmbedcrypto.3.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedx509.0.dylib ./OBS.app/Contents/Frameworks/libmbedx509.0.dylib
-    install_name_tool -change libmbedtls.12.dylib @executable_path/../Frameworks/libmbedtls.12.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-    install_name_tool -change libmbedcrypto.3.dylib @executable_path/../Frameworks/libmbedcrypto.3.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-    install_name_tool -change libmbedx509.0.dylib @executable_path/../Frameworks/libmbedx509.0.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-elif [ -f /usr/local/opt/mbedtls/lib/libmbedtls.13.dylib ]; then
-    cp /usr/local/opt/mbedtls/lib/libmbedtls.13.dylib ./OBS.app/Contents/Frameworks/
-    cp /usr/local/opt/mbedtls/lib/libmbedcrypto.5.dylib ./OBS.app/Contents/Frameworks/
-    cp /usr/local/opt/mbedtls/lib/libmbedx509.1.dylib ./OBS.app/Contents/Frameworks/
-    chmod +w ./OBS.app/Contents/Frameworks/*.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedtls.13.dylib ./OBS.app/Contents/Frameworks/libmbedtls.13.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedcrypto.5.dylib ./OBS.app/Contents/Frameworks/libmbedcrypto.5.dylib
-    install_name_tool -id @executable_path/../Frameworks/libmbedx509.1.dylib ./OBS.app/Contents/Frameworks/libmbedx509.1.dylib
-    install_name_tool -change libmbedtls.13.dylib @executable_path/../Frameworks/libmbedtls.13.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-    install_name_tool -change libmbedcrypto.5.dylib @executable_path/../Frameworks/libmbedcrypto.5.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-    install_name_tool -change libmbedx509.1.dylib @executable_path/../Frameworks/libmbedx509.1.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-fi
-
-install_name_tool -change /usr/local/opt/curl/lib/libcurl.4.dylib @executable_path/../Frameworks/libcurl.4.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-install_name_tool -change @rpath/libobs.0.dylib @executable_path/../Frameworks/libobs.0.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-install_name_tool -change /tmp/obsdeps/bin/libjansson.4.dylib @executable_path/../Frameworks/libjansson.4.dylib ./OBS.app/Contents/Plugins/obs-outputs.so
-
-# copy sparkle into the app
-hr "Copying Sparkle.framework"
-cp -R ../../sparkle/Sparkle.framework ./OBS.app/Contents/Frameworks/
-install_name_tool -change @rpath/Sparkle.framework/Versions/A/Sparkle @executable_path/../Frameworks/Sparkle.framework/Versions/A/Sparkle ./OBS.app/Contents/MacOS/obs
-
-# Copy Chromium embedded framework to app Frameworks directory
-hr "Copying Chromium Embedded Framework.framework"
-sudo mkdir -p OBS.app/Contents/Frameworks
-sudo cp -R ../../cef_binary_${CEF_BUILD_VERSION}_macosx64/Release/Chromium\ Embedded\ Framework.framework OBS.app/Contents/Frameworks/
-
-install_name_tool -change /usr/local/opt/qt/lib/QtGui.framework/Versions/5/QtGui @executable_path/../Frameworks/QtGui.framework/Versions/5/QtGui ./OBS.app/Contents/Plugins/obs-browser.so
-install_name_tool -change /usr/local/opt/qt/lib/QtCore.framework/Versions/5/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore ./OBS.app/Contents/Plugins/obs-browser.so
-install_name_tool -change /usr/local/opt/qt/lib/QtWidgets.framework/Versions/5/QtWidgets @executable_path/../Frameworks/QtWidgets.framework/Versions/5/QtWidgets ./OBS.app/Contents/Plugins/obs-browser.so
-
-cp ../CI/install/osx/OBSPublicDSAKey.pem OBS.app/Contents/Resources
-
-# edit plist
-plutil -insert CFBundleVersion -string $GIT_TAG ./OBS.app/Contents/Info.plist
-plutil -insert CFBundleShortVersionString -string $GIT_TAG ./OBS.app/Contents/Info.plist
-plutil -insert OBSFeedsURL -string https://obsproject.com/osx_update/feeds.xml ./OBS.app/Contents/Info.plist
-plutil -insert SUFeedURL -string https://obsproject.com/osx_update/stable/updates.xml ./OBS.app/Contents/Info.plist
-plutil -insert SUPublicDSAKeyFile -string OBSPublicDSAKey.pem ./OBS.app/Contents/Info.plist
-
-dmgbuild -s ../CI/install/osx/settings.json "OBS" obs.dmg
-
-if [ -v "$TRAVIS" ]; then
-	# Signing stuff
-	hr "Decrypting Cert"
-	openssl aes-256-cbc -K $encrypted_dd3c7f5e9db9_key -iv $encrypted_dd3c7f5e9db9_iv -in ../CI/osxcert/Certificates.p12.enc -out Certificates.p12 -d
-	hr "Creating Keychain"
-	security create-keychain -p mysecretpassword build.keychain
-	security default-keychain -s build.keychain
-	security unlock-keychain -p mysecretpassword build.keychain
-	security set-keychain-settings -t 3600 -u build.keychain
-	hr "Importing certs into keychain"
-	security import ./Certificates.p12 -k build.keychain -T /usr/bin/productsign -P ""
-	# macOS 10.12+
-	security set-key-partition-list -S apple-tool:,apple: -s -k mysecretpassword build.keychain
-fi
-
-cp ./OBS.dmg ./$FILENAME
-
-# Move to the folder that travis uses to upload artifacts from
-hr "Moving package to nightly folder for distribution"
-mkdir ../nightly
-sudo mv ./$FILENAME ../nightly

obs-studio-26.1.0.tar.xz/CI/before-script-osx.sh Deleted

obs-studio-26.1.0.tar.xz/CI/install Deleted

obs-studio-26.1.0.tar.xz/CI/install-dependencies-osx.sh Deleted

@@ -1,81 +0,0 @@
-hr() {
-  echo "───────────────────────────────────────────────────"
-  echo $1
-  echo "───────────────────────────────────────────────────"
-}
-
-# Exit if something fails
-set -e
-
-# Echo all commands before executing
-set -v
-
-if [[ $TRAVIS ]]; then
-  git fetch --unshallow
-fi
-
-git fetch origin --tags
-
-# Leave obs-studio folder
-cd ../
-
-# Install Packages app so we can build a package later
-# http://s.sudre.free.fr/Software/Packages/about.html
-hr "Downloading Packages app"
-wget --quiet --retry-connrefused --waitretry=1 https://s3-us-west-2.amazonaws.com/obs-nightly/Packages.pkg
-sudo installer -pkg ./Packages.pkg -target /
-
-brew update
-
-#Base OBS Deps and ccache
-for DEPENDENCY in jack speexdsp ccache mbedtls freetype fdk-aac cmocka; do
-    if [ ! -d "$(brew --cellar)/${DEPENDENCY}" ]; then
-        brew install $DEPENDENCY
-    else
-        brew upgrade $DEPENDENCY
-    fi
-done
-
-brew install https://gist.githubusercontent.com/DDRBoxman/9c7a2b08933166f4b61ed9a44b242609/raw/ef4de6c587c6bd7f50210eccd5bd51ff08e6de13/qt.rb
-if [ -d "$(brew --cellar)/swig" ]; then
-    brew unlink swig
-fi
-brew install https://gist.githubusercontent.com/DDRBoxman/4cada55c51803a2f963fa40ce55c9d3e/raw/572c67e908bfbc1bcb8c476ea77ea3935133f5b5/swig.rb
-
-pip install dmgbuild
-
-export PATH=/usr/local/opt/ccache/libexec:$PATH
-ccache -s || echo "CCache is not available."
-
-# Fetch and untar prebuilt OBS deps that are compatible with older versions of OSX
-hr "Downloading OBS deps"
-wget --quiet --retry-connrefused --waitretry=1 https://github.com/obsproject/obs-deps/releases/download/2020-04-24/osx-deps-2020-04-24.tar.gz
-tar -xf ./osx-deps-2020-04-24.tar.gz -C /tmp
-
-# Fetch vlc codebase
-hr "Downloading VLC repo"
-wget --quiet --retry-connrefused --waitretry=1 https://downloads.videolan.org/vlc/3.0.8/vlc-3.0.8.tar.xz
-tar -xf vlc-3.0.8.tar.xz
-
-# Get sparkle
-hr "Downloading Sparkle framework"
-wget --quiet --retry-connrefused --waitretry=1 -O sparkle.tar.bz2 https://github.com/sparkle-project/Sparkle/releases/download/1.23.0/Sparkle-1.23.0.tar.bz2
-mkdir ./sparkle
-tar -xf ./sparkle.tar.bz2 -C ./sparkle
-sudo cp -R ./sparkle/Sparkle.framework /Library/Frameworks/Sparkle.framework
-
-# CEF Stuff
-hr "Downloading CEF"
-wget --quiet --retry-connrefused --waitretry=1 https://obs-nightly.s3-us-west-2.amazonaws.com/cef_binary_${CEF_BUILD_VERSION}_macosx64.tar.bz2
-tar -xf ./cef_binary_${CEF_BUILD_VERSION}_macosx64.tar.bz2
-cd ./cef_binary_${CEF_BUILD_VERSION}_macosx64
-# remove a broken test
-sed -i '.orig' '/add_subdirectory(tests\/ceftests)/d' ./CMakeLists.txt
-# target 10.11
-sed -i '.orig' s/\"10.9\"/\"10.11\"/ ./cmake/cef_variables.cmake
-mkdir build
-cd ./build
-cmake -DCMAKE_CXX_FLAGS="-std=c++11 -stdlib=libc++" -DCMAKE_EXE_LINKER_FLAGS="-std=c++11 -stdlib=libc++" -DCMAKE_OSX_DEPLOYMENT_TARGET=10.11 ..
-make -j4
-mkdir libcef_dll
-cd ../../

obs-studio-26.1.0.tar.xz/CI/install/osx Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/CMakeLists.pkgproj Deleted

@@ -1,1026 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
-<plist version="1.0">
-<dict>
-	<key>PACKAGES</key>
-	<array>
-		<dict>
-			<key>PACKAGE_FILES</key>
-			<dict>
-				<key>DEFAULT_INSTALL_LOCATION</key>
-				<string>/</string>
-				<key>HIERARCHY</key>
-				<dict>
-					<key>CHILDREN</key>
-					<array>
-						<dict>
-							<key>CHILDREN</key>
-							<array>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>80</integer>
-									<key>PATH</key>
-									<string>../../../build/OBS.app</string>
-									<key>PATH_TYPE</key>
-									<integer>3</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>3</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>80</integer>
-									<key>PATH</key>
-									<string>Utilities</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-							</array>
-							<key>GID</key>
-							<integer>80</integer>
-							<key>PATH</key>
-							<string>Applications</string>
-							<key>PATH_TYPE</key>
-							<integer>0</integer>
-							<key>PERMISSIONS</key>
-							<integer>509</integer>
-							<key>TYPE</key>
-							<integer>1</integer>
-							<key>UID</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>CHILDREN</key>
-							<array>
-								<dict>
-									<key>CHILDREN</key>
-									<array>
-										<dict>
-											<key>CHILDREN</key>
-											<array>
-												<dict>
-													<key>CHILDREN</key>
-													<array>
-														<dict>
-															<key>CHILDREN</key>
-															<array>
-																<dict>
-																	<key>CHILDREN</key>
-																	<array>
-																		<dict>
-																			<key>CHILDREN</key>
-																			<array/>
-																			<key>GID</key>
-																			<integer>80</integer>
-																			<key>PATH</key>
-																			<string>../../../build/plugins/obs-browser/obs-browser-page</string>
-																			<key>PATH_TYPE</key>
-																			<integer>3</integer>
-																			<key>PERMISSIONS</key>
-																			<integer>493</integer>
-																			<key>TYPE</key>
-																			<integer>3</integer>
-																			<key>UID</key>
-																			<integer>0</integer>
-																		</dict>
-																		<dict>
-																			<key>CHILDREN</key>
-																			<array/>
-																			<key>GID</key>
-																			<integer>80</integer>
-																			<key>PATH</key>
-																			<string>../../../build/plugins/obs-browser/obs-browser.so</string>
-																			<key>PATH_TYPE</key>
-																			<integer>3</integer>
-																			<key>PERMISSIONS</key>
-																			<integer>493</integer>
-																			<key>TYPE</key>
-																			<integer>3</integer>
-																			<key>UID</key>
-																			<integer>0</integer>
-																		</dict>
-																	</array>
-																	<key>GID</key>
-																	<integer>80</integer>
-																	<key>PATH</key>
-																	<string>bin</string>
-																	<key>PATH_TYPE</key>
-																	<integer>0</integer>
-																	<key>PERMISSIONS</key>
-																	<integer>493</integer>
-																	<key>TYPE</key>
-																	<integer>2</integer>
-																	<key>UID</key>
-																	<integer>0</integer>
-																</dict>
-															</array>
-															<key>GID</key>
-															<integer>80</integer>
-															<key>PATH</key>
-															<string>obs-browser</string>
-															<key>PATH_TYPE</key>
-															<integer>0</integer>
-															<key>PERMISSIONS</key>
-															<integer>493</integer>
-															<key>TYPE</key>
-															<integer>2</integer>
-															<key>UID</key>
-															<integer>0</integer>
-														</dict>
-													</array>
-													<key>GID</key>
-													<integer>80</integer>
-													<key>PATH</key>
-													<string>plugins</string>
-													<key>PATH_TYPE</key>
-													<integer>0</integer>
-													<key>PERMISSIONS</key>
-													<integer>493</integer>
-													<key>TYPE</key>
-													<integer>2</integer>
-													<key>UID</key>
-													<integer>0</integer>
-												</dict>
-											</array>
-											<key>GID</key>
-											<integer>80</integer>
-											<key>PATH</key>
-											<string>obs-studio</string>
-											<key>PATH_TYPE</key>
-											<integer>0</integer>
-											<key>PERMISSIONS</key>
-											<integer>493</integer>
-											<key>TYPE</key>
-											<integer>2</integer>
-											<key>UID</key>
-											<integer>0</integer>
-										</dict>
-									</array>
-									<key>GID</key>
-									<integer>80</integer>
-									<key>PATH</key>
-									<string>Application Support</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Documentation</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Filesystems</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Frameworks</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Input Methods</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Internet Plug-Ins</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>LaunchAgents</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>LaunchDaemons</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>PreferencePanes</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Preferences</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>80</integer>
-									<key>PATH</key>
-									<string>Printers</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>PrivilegedHelperTools</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>QuickLook</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>QuickTime</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Screen Savers</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Scripts</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Services</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Widgets</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-							</array>
-							<key>GID</key>
-							<integer>0</integer>
-							<key>PATH</key>
-							<string>Library</string>
-							<key>PATH_TYPE</key>
-							<integer>0</integer>
-							<key>PERMISSIONS</key>
-							<integer>493</integer>
-							<key>TYPE</key>
-							<integer>1</integer>
-							<key>UID</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>CHILDREN</key>
-							<array>
-								<dict>
-									<key>CHILDREN</key>
-									<array>
-										<dict>
-											<key>CHILDREN</key>
-											<array/>
-											<key>GID</key>
-											<integer>0</integer>
-											<key>PATH</key>
-											<string>Extensions</string>
-											<key>PATH_TYPE</key>
-											<integer>0</integer>
-											<key>PERMISSIONS</key>
-											<integer>493</integer>
-											<key>TYPE</key>
-											<integer>1</integer>
-											<key>UID</key>
-											<integer>0</integer>
-										</dict>
-									</array>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Library</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>493</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-							</array>
-							<key>GID</key>
-							<integer>0</integer>
-							<key>PATH</key>
-							<string>System</string>
-							<key>PATH_TYPE</key>
-							<integer>0</integer>
-							<key>PERMISSIONS</key>
-							<integer>493</integer>
-							<key>TYPE</key>
-							<integer>1</integer>
-							<key>UID</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>CHILDREN</key>
-							<array>
-								<dict>
-									<key>CHILDREN</key>
-									<array/>
-									<key>GID</key>
-									<integer>0</integer>
-									<key>PATH</key>
-									<string>Shared</string>
-									<key>PATH_TYPE</key>
-									<integer>0</integer>
-									<key>PERMISSIONS</key>
-									<integer>1023</integer>
-									<key>TYPE</key>
-									<integer>1</integer>
-									<key>UID</key>
-									<integer>0</integer>
-								</dict>
-							</array>
-							<key>GID</key>
-							<integer>80</integer>
-							<key>PATH</key>
-							<string>Users</string>
-							<key>PATH_TYPE</key>
-							<integer>0</integer>
-							<key>PERMISSIONS</key>
-							<integer>493</integer>
-							<key>TYPE</key>
-							<integer>1</integer>
-							<key>UID</key>
-							<integer>0</integer>
-						</dict>
-					</array>
-					<key>GID</key>
-					<integer>0</integer>
-					<key>PATH</key>
-					<string>/</string>
-					<key>PATH_TYPE</key>
-					<integer>0</integer>
-					<key>PERMISSIONS</key>
-					<integer>493</integer>
-					<key>TYPE</key>
-					<integer>1</integer>
-					<key>UID</key>
-					<integer>0</integer>
-				</dict>
-				<key>PAYLOAD_TYPE</key>
-				<integer>0</integer>
-				<key>VERSION</key>
-				<integer>2</integer>
-			</dict>
-			<key>PACKAGE_SCRIPTS</key>
-			<dict>
-				<key>POSTINSTALL_PATH</key>
-				<dict>
-					<key>PATH</key>
-					<string>post-install.sh</string>
-					<key>PATH_TYPE</key>
-					<integer>3</integer>
-				</dict>
-				<key>PREINSTALL_PATH</key>
-				<dict/>
-				<key>RESOURCES</key>
-				<array/>
-			</dict>
-			<key>PACKAGE_SETTINGS</key>
-			<dict>
-				<key>AUTHENTICATION</key>
-				<integer>1</integer>
-				<key>CONCLUSION_ACTION</key>
-				<integer>0</integer>
-				<key>IDENTIFIER</key>
-				<string>org.obsproject.pkg.obs-studio</string>
-				<key>NAME</key>
-				<string>OBS</string>
-				<key>OVERWRITE_PERMISSIONS</key>
-				<false/>
-				<key>VERSION</key>
-				<string>1.0</string>
-			</dict>
-			<key>UUID</key>
-			<string>19CCE3F2-8911-4364-B673-8B5BC3ABD4DA</string>
-		</dict>
-		<dict>
-			<key>PACKAGE_SETTINGS</key>
-			<dict>
-				<key>LOCATION</key>
-				<integer>0</integer>
-				<key>NAME</key>
-				<string>SyphonInject</string>
-			</dict>
-			<key>PATH</key>
-			<dict>
-				<key>PATH</key>
-				<string>SyphonInject.pkg</string>
-				<key>PATH_TYPE</key>
-				<integer>1</integer>
-			</dict>
-			<key>TYPE</key>
-			<integer>1</integer>
-			<key>UUID</key>
-			<string>0CC9C67E-4D14-4794-9930-019925513B1C</string>
-		</dict>
-	</array>
-	<key>PROJECT</key>
-	<dict>
-		<key>PROJECT_COMMENTS</key>
-		<dict>
-			<key>NOTES</key>
-			<data>
-			PCFET0NUWVBFIGh0bWwgUFVCTElDICItLy9XM0MvL0RURCBIVE1M
-			IDQuMDEvL0VOIiAiaHR0cDovL3d3dy53My5vcmcvVFIvaHRtbDQv
-			c3RyaWN0LmR0ZCI+CjxodG1sPgo8aGVhZD4KPG1ldGEgaHR0cC1l
-			cXVpdj0iQ29udGVudC1UeXBlIiBjb250ZW50PSJ0ZXh0L2h0bWw7
-			IGNoYXJzZXQ9VVRGLTgiPgo8bWV0YSBodHRwLWVxdWl2PSJDb250
-			ZW50LVN0eWxlLVR5cGUiIGNvbnRlbnQ9InRleHQvY3NzIj4KPHRp
-			dGxlPjwvdGl0bGU+CjxtZXRhIG5hbWU9IkdlbmVyYXRvciIgY29u
-			dGVudD0iQ29jb2EgSFRNTCBXcml0ZXIiPgo8bWV0YSBuYW1lPSJD
-			b2NvYVZlcnNpb24iIGNvbnRlbnQ9IjE1MDQuODEiPgo8c3R5bGUg
-			dHlwZT0idGV4dC9jc3MiPgo8L3N0eWxlPgo8L2hlYWQ+Cjxib2R5
-			Pgo8L2JvZHk+CjwvaHRtbD4K
-			</data>
-		</dict>
-		<key>PROJECT_PRESENTATION</key>
-		<dict>
-			<key>BACKGROUND</key>
-			<dict>
-				<key>ALIGNMENT</key>
-				<integer>4</integer>
-				<key>BACKGROUND_PATH</key>
-				<dict>
-					<key>PATH</key>
-					<string>obs.png</string>
-					<key>PATH_TYPE</key>
-					<integer>1</integer>
-				</dict>
-				<key>CUSTOM</key>
-				<integer>1</integer>
-				<key>SCALING</key>
-				<integer>0</integer>
-			</dict>
-			<key>INSTALLATION TYPE</key>
-			<dict>
-				<key>HIERARCHIES</key>
-				<dict>
-					<key>INSTALLER</key>
-					<dict>
-						<key>LIST</key>
-						<array>
-							<dict>
-								<key>DESCRIPTION</key>
-								<array/>
-								<key>OPTIONS</key>
-								<dict>
-									<key>HIDDEN</key>
-									<false/>
-									<key>STATE</key>
-									<integer>0</integer>
-								</dict>
-								<key>PACKAGE_UUID</key>
-								<string>19CCE3F2-8911-4364-B673-8B5BC3ABD4DA</string>
-								<key>REQUIREMENTS</key>
-								<array/>
-								<key>TITLE</key>
-								<array/>
-								<key>TOOLTIP</key>
-								<array/>
-								<key>TYPE</key>
-								<integer>0</integer>
-								<key>UUID</key>
-								<string>7C540711-59F4-479C-9CFD-8C4D6594992E</string>
-							</dict>
-							<dict>
-								<key>DESCRIPTION</key>
-								<array/>
-								<key>OPTIONS</key>
-								<dict>
-									<key>HIDDEN</key>
-									<false/>
-									<key>STATE</key>
-									<integer>1</integer>
-								</dict>
-								<key>PACKAGE_UUID</key>
-								<string>0CC9C67E-4D14-4794-9930-019925513B1C</string>
-								<key>REQUIREMENTS</key>
-								<array/>
-								<key>TITLE</key>
-								<array/>
-								<key>TOOLTIP</key>
-								<array/>
-								<key>TYPE</key>
-								<integer>0</integer>
-								<key>UUID</key>
-								<string>BBDE08F6-D7EE-47CB-881F-7F208B3A604B</string>
-							</dict>
-						</array>
-						<key>REMOVED</key>
-						<dict/>
-					</dict>
-				</dict>
-				<key>INSTALLATION TYPE</key>
-				<integer>0</integer>
-				<key>MODE</key>
-				<integer>0</integer>
-			</dict>
-			<key>INSTALLATION_STEPS</key>
-			<array>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewIntroductionController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>Introduction</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewReadMeController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>ReadMe</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewLicenseController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>License</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewDestinationSelectController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>TargetSelect</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewInstallationTypeController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>PackageSelection</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewInstallationController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>Install</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-				<dict>
-					<key>ICPRESENTATION_CHAPTER_VIEW_CONTROLLER_CLASS</key>
-					<string>ICPresentationViewSummaryController</string>
-					<key>INSTALLER_PLUGIN</key>
-					<string>Summary</string>
-					<key>LIST_TITLE_KEY</key>
-					<string>InstallerSectionTitle</string>
-				</dict>
-			</array>
-			<key>INTRODUCTION</key>
-			<dict>
-				<key>LOCALIZATIONS</key>
-				<array/>
-			</dict>
-			<key>LICENSE</key>
-			<dict>
-				<key>KEYWORDS</key>
-				<dict/>
-				<key>LOCALIZATIONS</key>
-				<array/>
-				<key>MODE</key>
-				<integer>0</integer>
-			</dict>
-			<key>README</key>
-			<dict>
-				<key>LOCALIZATIONS</key>
-				<array/>
-			</dict>
-			<key>SUMMARY</key>
-			<dict>
-				<key>LOCALIZATIONS</key>
-				<array/>
-			</dict>
-			<key>TITLE</key>
-			<dict>
-				<key>LOCALIZATIONS</key>
-				<array>
-					<dict>
-						<key>LANGUAGE</key>
-						<string>English</string>
-						<key>VALUE</key>
-						<string>OBS</string>
-					</dict>
-				</array>
-			</dict>
-		</dict>
-		<key>PROJECT_REQUIREMENTS</key>
-		<dict>
-			<key>LIST</key>
-			<array/>
-			<key>POSTINSTALL_PATH</key>
-			<dict/>
-			<key>PREINSTALL_PATH</key>
-			<dict/>
-			<key>RESOURCES</key>
-			<array/>
-			<key>ROOT_VOLUME_ONLY</key>
-			<false/>
-		</dict>
-		<key>PROJECT_SETTINGS</key>
-		<dict>
-			<key>ADVANCED_OPTIONS</key>
-			<dict/>
-			<key>BUILD_FORMAT</key>
-			<integer>0</integer>
-			<key>BUILD_PATH</key>
-			<dict>
-				<key>PATH</key>
-				<string>../../../build</string>
-				<key>PATH_TYPE</key>
-				<integer>3</integer>
-			</dict>
-			<key>EXCLUDED_FILES</key>
-			<array>
-				<dict>
-					<key>PATTERNS_ARRAY</key>
-					<array>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.DS_Store</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-					</array>
-					<key>PROTECTED</key>
-					<true/>
-					<key>PROXY_NAME</key>
-					<string>Remove .DS_Store files</string>
-					<key>PROXY_TOOLTIP</key>
-					<string>Remove ".DS_Store" files created by the Finder.</string>
-					<key>STATE</key>
-					<true/>
-				</dict>
-				<dict>
-					<key>PATTERNS_ARRAY</key>
-					<array>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.pbdevelopment</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-					</array>
-					<key>PROTECTED</key>
-					<true/>
-					<key>PROXY_NAME</key>
-					<string>Remove .pbdevelopment files</string>
-					<key>PROXY_TOOLTIP</key>
-					<string>Remove ".pbdevelopment" files created by ProjectBuilder or Xcode.</string>
-					<key>STATE</key>
-					<true/>
-				</dict>
-				<dict>
-					<key>PATTERNS_ARRAY</key>
-					<array>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>CVS</string>
-							<key>TYPE</key>
-							<integer>1</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.cvsignore</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.cvspass</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.svn</string>
-							<key>TYPE</key>
-							<integer>1</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.git</string>
-							<key>TYPE</key>
-							<integer>1</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>.gitignore</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-					</array>
-					<key>PROTECTED</key>
-					<true/>
-					<key>PROXY_NAME</key>
-					<string>Remove SCM metadata</string>
-					<key>PROXY_TOOLTIP</key>
-					<string>Remove helper files and folders used by the CVS, SVN or Git Source Code Management systems.</string>
-					<key>STATE</key>
-					<true/>
-				</dict>
-				<dict>
-					<key>PATTERNS_ARRAY</key>
-					<array>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>classes.nib</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>designable.db</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>info.nib</string>
-							<key>TYPE</key>
-							<integer>0</integer>
-						</dict>
-					</array>
-					<key>PROTECTED</key>
-					<true/>
-					<key>PROXY_NAME</key>
-					<string>Optimize nib files</string>
-					<key>PROXY_TOOLTIP</key>
-					<string>Remove "classes.nib", "info.nib" and "designable.nib" files within .nib bundles.</string>
-					<key>STATE</key>
-					<true/>
-				</dict>
-				<dict>
-					<key>PATTERNS_ARRAY</key>
-					<array>
-						<dict>
-							<key>REGULAR_EXPRESSION</key>
-							<false/>
-							<key>STRING</key>
-							<string>Resources Disabled</string>
-							<key>TYPE</key>
-							<integer>1</integer>
-						</dict>
-					</array>
-					<key>PROTECTED</key>
-					<true/>
-					<key>PROXY_NAME</key>
-					<string>Remove Resources Disabled folders</string>
-					<key>PROXY_TOOLTIP</key>
-					<string>Remove "Resources Disabled" folders.</string>
-					<key>STATE</key>
-					<true/>
-				</dict>
-				<dict>
-					<key>SEPARATOR</key>
-					<true/>
-				</dict>
-			</array>
-			<key>NAME</key>
-			<string>OBS</string>
-		</dict>
-	</dict>
-	<key>TYPE</key>
-	<integer>0</integer>
-	<key>VERSION</key>
-	<integer>2</integer>
-</dict>
-</plist>

obs-studio-26.1.0.tar.xz/CI/install/osx/Info.plist Deleted

@@ -1,28 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
-<plist version="1.0">
-<dict>
-	<key>CFBundleIconFile</key>
-	<string>obs.icns</string>
-	<key>CFBundleName</key>
-	<string>OBS</string>
-	<key>CFBundleGetInfoString</key>
-	<string>OBS - Free and Open Source Streaming/Recording Software</string>
-	<key>CFBundleExecutable</key>
-	<string>obs</string>
-	<key>CFBundleIdentifier</key>
-	<string>com.obsproject.obs-studio</string>
-	<key>CFBundlePackageType</key>
-	<string>APPL</string>
-	<key>LSMinimumSystemVersion</key>
-	<string>10.8.5</string>
-	<key>NSHighResolutionCapable</key>
-	<true/>
-	<key>LSAppNapIsDisabled</key>
-	<true/>
-	<key>NSCameraUsageDescription</key>
- 	<string>OBS needs to access the camera to enable camera sources to work.</string>
- 	<key>NSMicrophoneUsageDescription</key>
- 	<string>OBS needs to access the microphone to enable audio input.</string>
-</dict>
-</plist>

obs-studio-26.1.0.tar.xz/CI/install/osx/OBSPublicDSAKey.pem Deleted

@@ -1,36 +0,0 @@
------BEGIN PUBLIC KEY-----
-MIIGPDCCBC4GByqGSM44BAEwggQhAoICAQCZZZ2y7H2GJmMfP4KQihJTJOoiGNUw
-mue6sqMbH+utRykRnSKBZux6R665eRFMpNgrgFO1TLLGbdD2U31KiGtCvFJOmOl3
-+QP055BuXjEG36NU7AWEFLAlbDlr/2D3oumq3Ib3iMnnr9RrVztJ2VFOvVio1eWr
-ZxboVwKPK8D6BqsWiv15vbYlJnTC4Fls6ySmdjVBxwoPlTaMu1ysi5DfbIZ93s5u
-aQt1FvXuWtPBWjyVUORcNbcWf49E5R2pV0OSBK95Hw2/wXz4vmj+w92dTePGnVaW
-Me4CoF5PIeZILwp6DCLStX4eW2WG1NChJTC8zeQ/3bMMoGyKM/MadyvrDqMywsKY
-caxkIwHrDKOEdXXGo80dIwZMMLipPA8DKhx5ojphfkeXjIhKSx+49knXT3ED5okE
-Wai7tGUXj/8D8sGh+7b+AVsdujvr4v8WQaZiKUOZ2IIHOg3VLz9T9v0zet1Yt987
-KNymFcp2CHeJ6KnDP/ZGQ6Nl0HsPxUgscsXV+R2FEc8Q1j0Ukkuxnopa0E4/huUu
-gjyRzpXD734qFMDf7LcXca6qNjBor6gVj5sRyRKCpZ+KQfMUlr8jp506ztYSyeJu
-dxJV30tQgztwkbrs02CqOt4Z3Peo6sdht7hWKSPVwmja3tq8/TfUSSoo6wKYN9/w
-Mf3dVeRF8hCzJQIVAJnzuzmzQhCKPiQnl3jh5qGII2XfAoICAQCCVATAff89ceHj
-ROHEbHTQFpVxJ/kRZPfxnU46DSw79Tih7tthV68oakPSOTP3cx/Tga0GwogarZ9N
-F2VVan5w9OQSSewXsr5UDT5bnmJF+h+JB7TMy+sXZBYobUqjlUd5VtKc8RsN86P4
-s7xbK0mA+hfe+27r18JT81/eH3xUfh7UOUGSdMN2Ch9f7RFSMZIgUAZUzu2K3ODp
-hPgtc2QJ8QVAp7GLvQgw8ZUME/ChZslyBIyJvYgUIxfxlgRWYro5pQT7/ngkgdXo
-wlghHKkldwMuY3zaFdhPnFNuEUEtc18ILsbz0+AnagCUd6n+3safskCRqLIHMOY6
-iLBSZPX9hJQhVCqSqz1VNDDww8FNa/fojJ1Lr/TI0I+0Ib2pCiY2LChXUqGY5SLZ
-2KNs5qFsyZP+I0L8YsGwqvUYyFwk7Ok224n0NtaOwqpLCrtXd/i6DaDNiaoJuwJC
-1ELCfaZivorgkC5rhBt2H7qWUAR+EtrFE/gb0k/G5EIhjYql7onGbX+G2re38vQA
-fg1pzguhig2dafP/BxMLZrn1Gg61xzmEYPuS9gclktaf675srv8GVb46VkOxXL+D
-YvTmpJPP7UUOVlmAMCo4j4y09MW3jq9TDp42VTLeZVubyjslGnavlnq1O+ZyXUye
-1FMeby65sIbSHHHwoFnRv3hLSEXI5gOCAgYAAoICAQCUkYnZkPfHfOJZI403xUYP
-CE/bLpkza074Xo6EXElsWRnpQgNTx+JFOvItgj3v0OkIqDin9UredKOwfkiftslV
-jxUVKA6I5kwnGvCpvTpQMLyLjq+VQr+J2D6eId6tV/iajhdu5r4JThU8KllT7Ywb
-NAur34ftLNCVAMRUaDNeEoHfePgderW384e+lbvpmtifmBluammGSxxRtUsdjvJZ
-BFkhaJu86CKxcU7D1lbPVOtV/jaxz6d16VdGcfBdi2LzXZzZtYpT9XGPX3NF+xii
-spAURWsoe11LTRXF+eJhgCm5iIDN3kh1HEQKYKAVpmrcM0aFzk/NpS+tFyU72vaq
-IRSSJw/aa1oELOAakG5oPldc4RcYWl32sbnVwXHO7TZvgTrBSC10o65MAC5CHP/s
-b07heDYAIt7re7szvOYq+c/9zAMAlu3pcO8MqaXYMmybdHBXHQ2b+DdJWHmIUWcX
-CbUzr09vzGkJAvqsXqbmJPr8aixrO75DhT0iDTILLWe/GWK51nf+Tg0pNxVgGyAl
-BqvRqqo7SSDu9FMkwQesFFHhuoHLyEHwVPJ+sMQTNwQcm9c6YuW8EYDRSkeKLWYk
-3fkjG+Pe9uVE8a1taDg3FjSY0UqjUT6XMw+i0Lajyus2L6wFBwrrGM6E4xa6x1CC
-MGjmuSOlPA1umQsToIcO4g==
------END PUBLIC KEY-----

obs-studio-26.1.0.tar.xz/CI/install/osx/SyphonInject.pkg Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.png Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/QuickLook Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/QuickLook/Icon.tiff Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/QuickLook/Preview.tiff Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/QuickLook/Thumbnail.tiff Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/data Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/data/556CF265-5721-4F18-BE83-8CF39483B4C2 Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/data/8CA689C3-ED2A-459E-952C-E08026CFCD07 Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.pxd/metadata.info Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background.tiff Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/background@2x.png Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/buildDMG Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/build_app.py Deleted

@@ -1,238 +0,0 @@
-#!/usr/bin/env python
-
-candidate_paths = "bin obs-plugins data".split()
-
-plist_path = "../cmake/osxbundle/Info.plist"
-icon_path = "../cmake/osxbundle/obs.icns"
-run_path = "../cmake/osxbundle/obslaunch.sh"
-
-#not copied
-blacklist = """/usr /System""".split()
-
-#copied
-whitelist = """/usr/local""".split()
-
-#
-#
-#
-
-
-from sys import argv
-from glob import glob
-from subprocess import check_output, call
-from collections import namedtuple
-from shutil import copy, copytree, rmtree
-from os import makedirs, rename, walk, path as ospath
-import plistlib
-
-import argparse
-
-def _str_to_bool(s):
-    """Convert string to bool (in argparse context)."""
-    if s.lower() not in ['true', 'false']:
-        raise ValueError('Need bool; got %r' % s)
-    return {'true': True, 'false': False}[s.lower()]
-
-def add_boolean_argument(parser, name, default=False):
-    """Add a boolean argument to an ArgumentParser instance."""
-    group = parser.add_mutually_exclusive_group()
-    group.add_argument(
-        '--' + name, nargs='?', default=default, const=True, type=_str_to_bool)
-    group.add_argument('--no' + name, dest=name, action='store_false')
-
-parser = argparse.ArgumentParser(description='obs-studio package util')
-parser.add_argument('-d', '--base-dir', dest='dir', default='rundir/RelWithDebInfo')
-parser.add_argument('-n', '--build-number', dest='build_number', default='0')
-parser.add_argument('-k', '--public-key', dest='public_key', default='OBSPublicDSAKey.pem')
-parser.add_argument('-f', '--sparkle-framework', dest='sparkle', default=None)
-parser.add_argument('-b', '--base-url', dest='base_url', default='https://obsproject.com/osx_update')
-parser.add_argument('-u', '--user', dest='user', default='jp9000')
-parser.add_argument('-c', '--channel', dest='channel', default='master')
-add_boolean_argument(parser, 'stable', default=False)
-parser.add_argument('-p', '--prefix', dest='prefix', default='')
-args = parser.parse_args()
-
-def cmd(cmd):
-    import subprocess
-    import shlex
-    return subprocess.check_output(shlex.split(cmd)).rstrip('\r\n')
-
-LibTarget = namedtuple("LibTarget", ("path", "external", "copy_as"))
-
-inspect = list()
-
-inspected = set()
-
-build_path = args.dir
-build_path = build_path.replace("\\ ", " ")
-
-def add(name, external=False, copy_as=None):
-	if external and copy_as is None:
-		copy_as = name.split("/")[-1]
-	if name[0] != "/":
-		name = build_path+"/"+name
-	t = LibTarget(name, external, copy_as)
-	if t in inspected:
-		return
-	inspect.append(t)
-	inspected.add(t)
-
-
-for i in candidate_paths:
-	print("Checking " + i)
-	for root, dirs, files in walk(build_path+"/"+i):
-		for file_ in files:
-			if ".ini" in file_:
-				continue
-			if ".png" in file_:
-				continue
-			if ".effect" in file_:
-				continue
-			if ".py" in file_:
-				continue
-			if ".json" in file_:
-				continue
-			path = root + "/" + file_
-			try:
-				out = check_output("{0}otool -L '{1}'".format(args.prefix, path), shell=True,
-						universal_newlines=True)
-				if "is not an object file" in out:
-					continue
-			except:
-				continue
-			rel_path = path[len(build_path)+1:]
-			print(repr(path), repr(rel_path))
-			add(rel_path)
-
-def add_plugins(path, replace):
-	for img in glob(path.replace(
-		"lib/QtCore.framework/Versions/5/QtCore",
-		"plugins/%s/*"%replace).replace(
-			"Library/Frameworks/QtCore.framework/Versions/5/QtCore",
-			"share/qt5/plugins/%s/*"%replace)):
-		if "_debug" in img:
-			continue
-		add(img, True, img.split("plugins/")[-1])
-
-actual_sparkle_path = '@loader_path/Frameworks/Sparkle.framework/Versions/A/Sparkle'
-
-while inspect:
-	target = inspect.pop()
-	print("inspecting", repr(target))
-	path = target.path
-	if path[0] == "@":
-		continue
-	out = check_output("{0}otool -L '{1}'".format(args.prefix, path), shell=True,
-			universal_newlines=True)
-
-	if "QtCore" in path:
-		add_plugins(path, "platforms")
-		add_plugins(path, "imageformats")
-		add_plugins(path, "accessible")
-		add_plugins(path, "styles")
-
-
-	for line in out.split("\n")[1:]:
-		new = line.strip().split(" (")[0]
-		if '@' in new and "sparkle.framework" in new.lower():
-			actual_sparkle_path = new
-			print "Using sparkle path:", repr(actual_sparkle_path)
-		if not new or new[0] == "@" or new.endswith(path.split("/")[-1]):
-			continue
-		whitelisted = False
-		for i in whitelist:
-			if new.startswith(i):
-				whitelisted = True
-		if not whitelisted:
-			blacklisted = False
-			for i in blacklist:
-				if new.startswith(i):
-					blacklisted = True
-					break
-			if blacklisted:
-				continue
-		add(new, True)
-
-changes = list()
-for path, external, copy_as in inspected:
-	if not external:
-		continue #built with install_rpath hopefully
-	changes.append("-change '%s' '@rpath/%s'"%(path, copy_as))
-changes = " ".join(changes)
-
-info = plistlib.readPlist(plist_path)
-
-latest_tag = cmd('git describe --tags --abbrev=0')
-log = cmd('git log --pretty=oneline {0}...HEAD'.format(latest_tag))
-
-from os import path
-# set version
-if args.stable:
-    info["CFBundleVersion"] = latest_tag
-    info["CFBundleShortVersionString"] = latest_tag
-    info["SUFeedURL"] = '{0}/stable/updates.xml'.format(args.base_url)
-else:
-    info["CFBundleVersion"] = args.build_number
-    info["CFBundleShortVersionString"] = '{0}.{1}'.format(latest_tag, args.build_number)
-    info["SUFeedURL"] = '{0}/{1}/{2}/updates.xml'.format(args.base_url, args.user, args.channel)
-
-info["SUPublicDSAKeyFile"] = path.basename(args.public_key)
-info["OBSFeedsURL"] = '{0}/feeds.xml'.format(args.base_url)
-
-app_name = info["CFBundleName"]+".app"
-icon_file = "tmp/Contents/Resources/%s"%info["CFBundleIconFile"]
-
-copytree(build_path, "tmp/Contents/Resources/", symlinks=True)
-copy(icon_path, icon_file)
-plistlib.writePlist(info, "tmp/Contents/Info.plist")
-makedirs("tmp/Contents/MacOS")
-copy(run_path, "tmp/Contents/MacOS/%s"%info["CFBundleExecutable"])
-try:
-	copy(args.public_key, "tmp/Contents/Resources")
-except:
-	pass
-
-if args.sparkle is not None:
-    copytree(args.sparkle, "tmp/Contents/Frameworks/Sparkle.framework", symlinks=True)
-
-prefix = "tmp/Contents/Resources/"
-sparkle_path = '@loader_path/{0}/Frameworks/Sparkle.framework/Versions/A/Sparkle'
-
-cmd('{0}install_name_tool -change {1} {2} {3}/bin/obs'.format(
-    args.prefix, actual_sparkle_path, sparkle_path.format('../..'), prefix))
-
-
-
-for path, external, copy_as in inspected:
-	id_ = ""
-	filename = path
-	rpath = ""
-	if external:
-		if copy_as == "Python":
-			continue
-		id_ = "-id '@rpath/%s'"%copy_as
-		filename = prefix + "bin/" +copy_as
-		rpath = "-add_rpath @loader_path/ -add_rpath @executable_path/"
-		if "/" in copy_as:
-			try:
-				dirs = copy_as.rsplit("/", 1)[0]
-				makedirs(prefix + "bin/" + dirs)
-			except:
-				pass
-		copy(path, filename)
-	else:
-		filename = path[len(build_path)+1:]
-		id_ = "-id '@rpath/../%s'"%filename
-		if not filename.startswith("bin"):
-			print(filename)
-			rpath = "-add_rpath '@loader_path/{}/'".format(ospath.relpath("bin/", ospath.dirname(filename)))
-		filename = prefix + filename
-
-	cmd = "{0}install_name_tool {1} {2} {3} '{4}'".format(args.prefix, changes, id_, rpath, filename)
-	call(cmd, shell=True)
-
-try:
-	rename("tmp", app_name)
-except:
-	print("App already exists")
-	rmtree("tmp")

obs-studio-26.1.0.tar.xz/CI/install/osx/dylibBundler Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/makeRetinaBG Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/obs.icns Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/obs.png Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/packageApp.sh Deleted

@@ -1,77 +0,0 @@
-# Exit if something fails
-set -e
-
-rm -rf ./OBS.app
-
-mkdir OBS.app
-mkdir OBS.app/Contents
-mkdir OBS.app/Contents/MacOS
-mkdir OBS.app/Contents/PlugIns
-mkdir OBS.app/Contents/Resources
-
-cp -R rundir/RelWithDebInfo/bin/ ./OBS.app/Contents/MacOS
-cp -R rundir/RelWithDebInfo/data ./OBS.app/Contents/Resources
-cp ../CI/install/osx/obs.icns ./OBS.app/Contents/Resources
-cp -R rundir/RelWithDebInfo/obs-plugins/ ./OBS.app/Contents/PlugIns
-cp ../CI/install/osx/Info.plist ./OBS.app/Contents
-
-../CI/install/osx/dylibBundler -b -cd -d ./OBS.app/Contents/Frameworks -p @executable_path/../Frameworks/ \
--s ./OBS.app/Contents/MacOS \
--s /usr/local/opt/mbedtls/lib/ \
--x ./OBS.app/Contents/PlugIns/coreaudio-encoder.so \
--x ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so \
--x ./OBS.app/Contents/PlugIns/frontend-tools.so \
--x ./OBS.app/Contents/PlugIns/image-source.so \
--x ./OBS.app/Contents/PlugIns/linux-jack.so \
--x ./OBS.app/Contents/PlugIns/mac-avcapture.so \
--x ./OBS.app/Contents/PlugIns/mac-capture.so \
--x ./OBS.app/Contents/PlugIns/mac-decklink.so \
--x ./OBS.app/Contents/PlugIns/mac-syphon.so \
--x ./OBS.app/Contents/PlugIns/mac-vth264.so \
--x ./OBS.app/Contents/PlugIns/obs-browser.so \
--x ./OBS.app/Contents/PlugIns/obs-browser-page \
--x ./OBS.app/Contents/PlugIns/obs-ffmpeg.so \
--x ./OBS.app/Contents/PlugIns/obs-filters.so \
--x ./OBS.app/Contents/PlugIns/obs-transitions.so \
--x ./OBS.app/Contents/PlugIns/obs-vst.so \
--x ./OBS.app/Contents/PlugIns/rtmp-services.so \
--x ./OBS.app/Contents/MacOS/obs \
--x ./OBS.app/Contents/MacOS/obs-ffmpeg-mux \
--x ./OBS.app/Contents/MacOS/obslua.so \
--x ./OBS.app/Contents/PlugIns/obs-x264.so \
--x ./OBS.app/Contents/PlugIns/text-freetype2.so \
--x ./OBS.app/Contents/PlugIns/obs-libfdk.so
-# -x ./OBS.app/Contents/MacOS/_obspython.so \
-# -x ./OBS.app/Contents/PlugIns/obs-outputs.so \
-
-/usr/local/Cellar/qt/5.14.1/bin/macdeployqt ./OBS.app
-
-mv ./OBS.app/Contents/MacOS/libobs-opengl.so ./OBS.app/Contents/Frameworks
-
-rm -f -r ./OBS.app/Contents/Frameworks/QtNetwork.framework
-
-# put qt network in here becasuse streamdeck uses it
-cp -R /usr/local/opt/qt/lib/QtNetwork.framework ./OBS.app/Contents/Frameworks
-chmod -R +w ./OBS.app/Contents/Frameworks/QtNetwork.framework
-rm -r ./OBS.app/Contents/Frameworks/QtNetwork.framework/Headers
-rm -r ./OBS.app/Contents/Frameworks/QtNetwork.framework/Versions/5/Headers/
-chmod 644 ./OBS.app/Contents/Frameworks/QtNetwork.framework/Versions/5/Resources/Info.plist
-install_name_tool -id @executable_path/../Frameworks/QtNetwork.framework/Versions/5/QtNetwork ./OBS.app/Contents/Frameworks/QtNetwork.framework/Versions/5/QtNetwork
-install_name_tool -change /usr/local/Cellar/qt/5.14.1/lib/QtCore.framework/Versions/5/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore ./OBS.app/Contents/Frameworks/QtNetwork.framework/Versions/5/QtNetwork
-
-
-# decklink ui qt
-install_name_tool -change /usr/local/opt/qt/lib/QtGui.framework/Versions/5/QtGui @executable_path/../Frameworks/QtGui.framework/Versions/5/QtGui ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so
-install_name_tool -change /usr/local/opt/qt/lib/QtCore.framework/Versions/5/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so
-install_name_tool -change /usr/local/opt/qt/lib/QtWidgets.framework/Versions/5/QtWidgets @executable_path/../Frameworks/QtWidgets.framework/Versions/5/QtWidgets ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so
-
-# frontend tools qt
-install_name_tool -change /usr/local/opt/qt/lib/QtGui.framework/Versions/5/QtGui @executable_path/../Frameworks/QtGui.framework/Versions/5/QtGui ./OBS.app/Contents/PlugIns/frontend-tools.so
-install_name_tool -change /usr/local/opt/qt/lib/QtCore.framework/Versions/5/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore ./OBS.app/Contents/PlugIns/frontend-tools.so
-install_name_tool -change /usr/local/opt/qt/lib/QtWidgets.framework/Versions/5/QtWidgets @executable_path/../Frameworks/QtWidgets.framework/Versions/5/QtWidgets ./OBS.app/Contents/PlugIns/frontend-tools.so
-
-# vst qt
-install_name_tool -change /usr/local/opt/qt/lib/QtGui.framework/Versions/5/QtGui @executable_path/../Frameworks/QtGui.framework/Versions/5/QtGui ./OBS.app/Contents/PlugIns/obs-vst.so
-install_name_tool -change /usr/local/opt/qt/lib/QtCore.framework/Versions/5/QtCore @executable_path/../Frameworks/QtCore.framework/Versions/5/QtCore ./OBS.app/Contents/PlugIns/obs-vst.so
-install_name_tool -change /usr/local/opt/qt/lib/QtWidgets.framework/Versions/5/QtWidgets @executable_path/../Frameworks/QtWidgets.framework/Versions/5/QtWidgets ./OBS.app/Contents/PlugIns/obs-vst.so
-install_name_tool -change /usr/local/opt/qt/lib/QtMacExtras.framework/Versions/5/QtMacExtras @executable_path/../Frameworks/QtMacExtras.framework/Versions/5/QtMacExtras ./OBS.app/Contents/PlugIns/obs-vst.so

obs-studio-26.1.0.tar.xz/CI/install/osx/package_util.py Deleted

@@ -1,94 +0,0 @@
-def cmd(cmd):
- import subprocess
- import shlex
- return subprocess.check_output(shlex.split(cmd)).rstrip('\r\n')
-
-def get_tag_info(tag):
- rev = cmd('git rev-parse {0}'.format(latest_tag))
- anno = cmd('git cat-file -p {0}'.format(rev))
- tag_info = []
- for i, v in enumerate(anno.splitlines()):
- if i <= 4:
- continue
- tag_info.append(v.lstrip())
-
- return tag_info
-
-def gen_html(github_user, latest_tag):
-
- url = 'https://github.com/{0}/obs-studio/commit/%H'.format(github_user)
-
- with open('readme.html', 'w') as f:
- f.write("<html><body>")
- log_cmd = """git log {0}...HEAD --pretty=format:'<li>&bull; <a href="{1}">(view)</a> %s</li>'"""
- log_res = cmd(log_cmd.format(latest_tag, url))
- if len(log_res.splitlines()):
- f.write('Changes since {0}: (Newest to oldest)'.format(latest_tag))
- f.write(log_res)
-
- ul = False
- f.write('')
- import re
-
- for l in get_tag_info(latest_tag):
- if not len(l):
- continue
- if l.startswith('*'):
- ul = True
- if not ul:
- f.write('<ul>')
- f.write('<li>&bull; {0}</li>'.format(re.sub(r'^(\s*)?[*](\s*)?', '', l)))
- else:
- if ul:
- f.write('</ul>')
- ul = False
- f.write('{0}'.format(l))
- if ul:
- f.write('</ul>')
- f.write('</body></html>')
-
- cmd('textutil -convert rtf readme.html -output readme.rtf')
- cmd("""sed -i '' 's/Times-Roman/Verdana/g' readme.rtf""")
-
-def save_manifest(latest_tag, user, jenkins_build, branch, stable):
- log = cmd('git log --pretty=oneline {0}...HEAD'.format(latest_tag))
- manifest = {}
- manifest['commits'] = []
- for v in log.splitlines():
- manifest['commits'].append(v)
- manifest['tag'] = {
- 'name': latest_tag,
- 'description': get_tag_info(latest_tag)
- }
-
- manifest['version'] = cmd('git rev-list HEAD --count')
- manifest['sha1'] = cmd('git rev-parse HEAD')
- manifest['jenkins_build'] = jenkins_build
-
- manifest['user'] = user
- manifest['branch'] = branch
- manifest['stable'] = stable
-
- import cPickle
- with open('manifest', 'w') as f:
- cPickle.dump(manifest, f)
-
-def prepare_pkg(project, package_id):
- cmd('packagesutil --file "{0}" set package-1 identifier {1}'.format(project, package_id))
- cmd('packagesutil --file "{0}" set package-1 version {1}'.format(project, '1.0'))
-
-
-import argparse
-parser = argparse.ArgumentParser(description='obs-studio package util')
-parser.add_argument('-u', '--user', dest='user', default='jp9000')
-parser.add_argument('-p', '--package-id', dest='package_id', default='org.obsproject.pkg.obs-studio')
-parser.add_argument('-f', '--project-file', dest='project', default='OBS.pkgproj')
-parser.add_argument('-j', '--jenkins-build', dest='jenkins_build', default='0')
-parser.add_argument('-b', '--branch', dest='branch', default='master')
-parser.add_argument('-s', '--stable', dest='stable', required=False, action='store_true', default=False)
-args = parser.parse_args()
-
-latest_tag = cmd('git describe --tags --abbrev=0')
-gen_html(args.user, latest_tag)
-prepare_pkg(args.project, args.package_id)
-save_manifest(latest_tag, args.user, args.jenkins_build, args.branch, args.stable)

obs-studio-26.1.0.tar.xz/CI/install/osx/post-install.sh Deleted

obs-studio-26.1.0.tar.xz/CI/install/osx/settings.json Deleted

obs-studio-26.1.0.tar.xz/CI/osxcert Deleted

obs-studio-26.1.0.tar.xz/CI/osxcert/Certificates.p12.enc Deleted

obs-studio-26.1.0.tar.xz/CI/util Deleted

obs-studio-26.1.0.tar.xz/CI/util/build-package-deps-osx.sh Deleted

@@ -1,161 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-# This script builds a tar file that contains a bunch of deps that OBS needs for
-# advanced functionality on OSX. Currently this tar file is pulled down off of s3
-# and used in the CI build process on travis.
-# Mostly this sets build flags to compile with older SDKS and make sure that 
-# the libs are portable.
-
-exists()
-{
-  command -v "$1" >/dev/null 2>&1
-}
-
-if ! exists nasm; then
-    echo "nasm not found. Try brew install nasm"
-    exit
-fi
-
-CURDIR=$(pwd)
-
-# the temp directory
-WORK_DIR=`mktemp -d`
-
-# deletes the temp directory
-function cleanup {
-  #rm -rf "$WORK_DIR"
-  echo "Deleted temp working directory $WORK_DIR"
-}
-
-# register the cleanup function to be called on the EXIT signal
-trap cleanup EXIT
-
-cd $WORK_DIR
-
-DEPS_DEST=$WORK_DIR/obsdeps
-
-# make dest dirs
-mkdir $DEPS_DEST
-mkdir $DEPS_DEST/bin
-mkdir $DEPS_DEST/include
-mkdir $DEPS_DEST/lib
-
-# OSX COMPAT
-export MACOSX_DEPLOYMENT_TARGET=10.11
-
-# If you need an olders SDK and Xcode won't give it to you
-# https://github.com/phracker/MacOSX-SDKs
-
-# libopus
-curl -L -O https://ftp.osuosl.org/pub/xiph/releases/opus/opus-1.2.1.tar.gz
-tar -xf opus-1.2.1.tar.gz
-cd ./opus-1.2.1
-mkdir build
-cd ./build
-../configure --disable-shared --enable-static --prefix="/tmp/obsdeps"
-make -j 12
-make install
-
-cd $WORK_DIR
-
-# libogg
-curl -L -O https://ftp.osuosl.org/pub/xiph/releases/ogg/libogg-1.3.3.tar.gz
-tar -xf libogg-1.3.3.tar.gz
-cd ./libogg-1.3.3
-mkdir build
-cd ./build
-../configure --disable-shared --enable-static --prefix="/tmp/obsdeps"
-make -j 12
-make install
-
-cd $WORK_DIR
-
-# libvorbis
-curl -L -O https://ftp.osuosl.org/pub/xiph/releases/vorbis/libvorbis-1.3.6.tar.gz
-tar -xf libvorbis-1.3.6.tar.gz
-cd ./libvorbis-1.3.6
-mkdir build
-cd ./build
-../configure --disable-shared --enable-static --prefix="/tmp/obsdeps"
-make -j 12
-make install
-
-cd $WORK_DIR
-
-# libvpx
-curl -L -O https://chromium.googlesource.com/webm/libvpx/+archive/v1.7.0.tar.gz
-mkdir -p ./libvpx-v1.7.0
-tar -xf v1.7.0.tar.gz -C $PWD/libvpx-v1.7.0
-cd ./libvpx-v1.7.0
-mkdir -p build
-cd ./build
-../configure --disable-shared --prefix="/tmp/obsdeps" --libdir="/tmp/obsdeps/lib"
-make -j 12
-make install
-
-cd $WORK_DIR
-
-# x264
-git clone git://git.videolan.org/x264.git
-cd ./x264
-git checkout origin/stable
-mkdir build
-cd ./build
-../configure --extra-ldflags="-mmacosx-version-min=10.11" --enable-static --prefix="/tmp/obsdeps"
-make -j 12
-make install
-../configure --extra-ldflags="-mmacosx-version-min=10.11" --enable-shared --libdir="/tmp/obsdeps/bin" --prefix="/tmp/obsdeps"
-make -j 12
-ln -f -s libx264.*.dylib libx264.dylib
-find . -name \*.dylib -exec cp \{\} $DEPS_DEST/bin/ \;
-rsync -avh --include="*/" --include="*.h" --exclude="*" ../* $DEPS_DEST/include/
-rsync -avh --include="*/" --include="*.h" --exclude="*" ./* $DEPS_DEST/include/
-
-cd $WORK_DIR
-
-# janson
-curl -L -O http://www.digip.org/jansson/releases/jansson-2.11.tar.gz
-tar -xf jansson-2.11.tar.gz
-cd jansson-2.11
-mkdir build
-cd ./build
-../configure --libdir="/tmp/obsdeps/bin" --enable-shared --disable-static
-make -j 12
-find . -name \*.dylib -exec cp \{\} $DEPS_DEST/bin/ \;
-rsync -avh --include="*/" --include="*.h" --exclude="*" ../* $DEPS_DEST/include/
-rsync -avh --include="*/" --include="*.h" --exclude="*" ./* $DEPS_DEST/include/
-
-cd $WORK_DIR
-
-export LDFLAGS="-L/tmp/obsdeps/lib"
-export CFLAGS="-I/tmp/obsdeps/include"
-
-# FFMPEG
-curl -L -O https://github.com/FFmpeg/FFmpeg/archive/n4.0.2.zip
-unzip ./n4.0.2.zip
-cd ./FFmpeg-n4.0.2
-mkdir build
-cd ./build
-../configure --pkg-config-flags="--static" --extra-ldflags="-mmacosx-version-min=10.11" --enable-shared --disable-static --shlibdir="/tmp/obsdeps/bin" --enable-gpl --disable-doc --enable-libx264 --enable-libopus --enable-libvorbis --enable-libvpx --disable-outdev=sdl
-make -j 12
-find . -name \*.dylib -exec cp \{\} $DEPS_DEST/bin/ \;
-rsync -avh --include="*/" --include="*.h" --exclude="*" ../* $DEPS_DEST/include/
-rsync -avh --include="*/" --include="*.h" --exclude="*" ./* $DEPS_DEST/include/
-
-#luajit
-curl -L -O https://luajit.org/download/LuaJIT-2.0.5.tar.gz
-tar -xf LuaJIT-2.0.5.tar.gz
-cd LuaJIT-2.0.5
-make PREFIX=/tmp/obsdeps
-make PREFIX=/tmp/obsdeps install
-find /tmp/obsdeps/lib -name libluajit\*.dylib -exec cp \{\} $DEPS_DEST/lib/ \;
-rsync -avh --include="*/" --include="*.h" --exclude="*" src/* $DEPS_DEST/include/
-make PREFIX=/tmp/obsdeps uninstall
-
-cd $WORK_DIR
-
-tar -czf osx-deps.tar.gz obsdeps
-
-cp ./osx-deps.tar.gz $CURDIR

obs-studio-26.1.0.tar.xz/CI/util/win32.sh Deleted

@@ -1,69 +0,0 @@
-#/bin/bash
-
-cd x264
-make clean
-LDFLAGS="-static-libgcc" ./configure --enable-shared --enable-win32thread --disable-avs --disable-ffms --disable-gpac --disable-interlaced --disable-lavf --cross-prefix=i686-w64-mingw32- --host=i686-pc-mingw32 --prefix="/home/jim/packages/win32"
-make -j6 fprofiled VIDS="CITY_704x576_60_orig_01.yuv"
-make install
-i686-w64-mingw32-dlltool -z /home/jim/packages/win32/bin/x264.orig.def --export-all-symbols /home/jim/packages/win32/bin/libx264-148.dll
-grep "EXPORTS\|x264" /home/jim/packages/win32/bin/x264.orig.def > /home/jim/packages/win32/bin/x264.def
-rm -f /home/jim/packages/win32/bin/x264.org.def
-sed -i -e "/\\t.*DATA/d" -e "/\\t\".*/d" -e "s/\s@.*//" /home/jim/packages/win32/bin/x264.def
-i686-w64-mingw32-dlltool -m i386 -d /home/jim/packages/win32/bin/x264.def -l /home/jim/packages/win32/bin/x264.lib -D /home/jim/win32/packages/bin/libx264-148.dll
-cd ..
-
-cd opus
-make clean
-LDFLAGS="-static-libgcc" ./configure -host=i686-w64-mingw32 --prefix="/home/jim/packages/win32" --enable-shared
-make -j6
-make install
-cd ..
-
-cd zlib/build32
-make clean
-cmake .. -DCMAKE_SYSTEM_NAME=Windows -DCMAKE_C_COMPILER=i686-w64-mingw32-gcc -DCMAKE_INSTALL_PREFIX=/home/jim/packages/win32 -DINSTALL_PKGCONFIG_DIR=/home/jim/packages/win32/lib/pkgconfig -DCMAKE_RC_COMPILER=i686-w64-mingw32-windres -DCMAKE_SHARED_LINKER_FLAGS="-static-libgcc"
-make -j6
-make install
-mv ../../win32/lib/libzlib.dll.a ../../win32/lib/libz.dll.a
-mv ../../win32/lib/libzlibstatic.a ../../win32/lib/libz.a
-cp ../win32/zlib.def /home/jim/packages/win32/bin
-i686-w64-mingw32-dlltool -m i386 -d ../win32/zlib.def -l /home/jim/packages/win32/bin/zlib.lib -D /home/jim/win32/packages/bin/zlib.dll
-cd ../..
-
-cd libpng
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win32/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win32/lib -static-libgcc" CPPFLAGS="-I/home/jim/packages/win32/include" ./configure -host=i686-w64-mingw32 --prefix="/home/jim/packages/win32" --enable-shared
-make -j6
-make install
-cd ..
-
-cd libogg
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win32/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win32/lib -static-libgcc" CPPFLAGS="-I/home/jim/packages/win32/include" ./configure -host=i686-w64-mingw32 --prefix="/home/jim/packages/win32" --enable-shared
-make -j6
-make install
-cd ..
-
-cd libvorbis
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win32/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win32/lib -static-libgcc" CPPFLAGS="-I/home/jim/packages/win32/include" ./configure -host=i686-w64-mingw32 --prefix="/home/jim/packages/win32" --enable-shared --with-ogg="/home/jim/packages/win32"
-make -j6
-make install
-cd ..
-
-cd libvpxbuild
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win32/lib/pkgconfig" CROSS=i686-w64-mingw32- LDFLAGS="-static-libgcc" ../libvpx/configure --prefix=/home/jim/packages/win32 --enable-vp8 --enable-vp9 --disable-docs --disable-examples --enable-shared --disable-static --enable-runtime-cpu-detect --enable-realtime-only --disable-install-bins --disable-install-docs --disable-unit-tests --target=x86-win32-gcc
-make -j6
-make install
-i686-w64-mingw32-dlltool -m i386 -d libvpx.def -l /home/jim/packages/win32/bin/vpx.lib -D /home/jim/win32/packages/bin/libvpx-1.dll
-cd ..
-
-cd ffmpeg
-make clean
-cp /media/sf_linux/nvEncodeAPI.h /home/jim/packages/win32/include
-PKG_CONFIG_PATH="/home/jim/packages/win32/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win32/lib -static-libgcc" CFLAGS="-I/home/jim/packages/win32/include" ./configure --enable-memalign-hack --enable-gpl --disable-programs --disable-doc --arch=x86 --enable-shared --enable-nvenc --enable-libx264 --enable-libopus --enable-libvorbis --enable-libvpx --disable-debug --cross-prefix=i686-w64-mingw32- --target-os=mingw32 --pkg-config=pkg-config --prefix="/home/jim/packages/win32" --disable-postproc
-read -n1 -r -p "Press any key to continue building FFmpeg..." key
-make -j6
-make install
-cd ..

obs-studio-26.1.0.tar.xz/CI/util/win64.sh Deleted

@@ -1,69 +0,0 @@
-#/bin/bash
-
-cd x264
-make clean
-LDFLAGS="-static-libgcc" ./configure --enable-shared --enable-win32thread --disable-avs --disable-ffms --disable-gpac --disable-interlaced --disable-lavf --cross-prefix=x86_64-w64-mingw32- --host=x86_64-pc-mingw32 --prefix="/home/jim/packages/win64"
-make -j6 fprofiled VIDS="CITY_704x576_60_orig_01.yuv"
-make install
-x86_64-w64-mingw32-dlltool -z /home/jim/packages/win64/bin/x264.orig.def --export-all-symbols /home/jim/packages/win64/bin/libx264-148.dll
-grep "EXPORTS\|x264" /home/jim/packages/win64/bin/x264.orig.def > /home/jim/packages/win64/bin/x264.def
-rm -f /home/jim/packages/win64/bin/x264.org.def
-sed -i -e "/\\t.*DATA/d" -e "/\\t\".*/d" -e "s/\s@.*//" /home/jim/packages/win64/bin/x264.def
-x86_64-w64-mingw32-dlltool -m i386:x86-64 -d /home/jim/packages/win64/bin/x264.def -l /home/jim/packages/win64/bin/x264.lib -D /home/jim/win64/packages/bin/libx264-148.dll
-cd ..
-
-cd opus
-make clean
-LDFLAGS="-static-libgcc" ./configure -host=x86_64-w64-mingw32 --prefix="/home/jim/packages/win64" --enable-shared
-make -j6
-make install
-cd ..
-
-cd zlib/build64
-make clean
-cmake .. -DCMAKE_SYSTEM_NAME=Windows -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_INSTALL_PREFIX=/home/jim/packages/win64 -DCMAKE_RC_COMPILER=x86_64-w64-mingw32-windres -DCMAKE_SHARED_LINKER_FLAGS="-static-libgcc"
-make -j6
-make install
-mv ../../win64/lib/libzlib.dll.a ../../win64/lib/libz.dll.a
-mv ../../win64/lib/libzlibstatic.a ../../win64/lib/libz.a
-cp ../win64/zlib.def /home/jim/packages/win64/bin
-x86_64-w64-mingw32-dlltool -m i386:x86-64 -d ../win32/zlib.def -l /home/jim/packages/win64/bin/zlib.lib -D /home/jim/win64/packages/bin/zlib.dll
-cd ../..
-
-cd libpng
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win64/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win64/lib" CPPFLAGS="-I/home/jim/packages/win64/include" ./configure -host=x86_64-w64-mingw32 --prefix="/home/jim/packages/win64" --enable-shared
-make -j6
-make install
-cd ..
-
-cd libogg
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win64/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win64/lib -static-libgcc" CPPFLAGS="-I/home/jim/packages/win64/include" ./configure -host=x86_64-w64-mingw32 --prefix="/home/jim/packages/win64" --enable-shared
-make -j6
-make install
-cd ..
-
-cd libvorbis
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win64/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win64/lib -static-libgcc" CPPFLAGS="-I/home/jim/packages/win64/include" ./configure -host=x86_64-w64-mingw32 --prefix="/home/jim/packages/win64" --enable-shared --with-ogg="/home/jim/packages/win64"
-make -j6
-make install
-cd ..
-
-cd libvpxbuild
-make clean
-PKG_CONFIG_PATH="/home/jim/packages/win64/lib/pkgconfig" CROSS=x86_64-w64-mingw32- LDFLAGS="-static-libgcc" ../libvpx/configure --prefix=/home/jim/packages/win64 --enable-vp8 --enable-vp9 --disable-docs --disable-examples --enable-shared --disable-static --enable-runtime-cpu-detect --enable-realtime-only --disable-install-bins --disable-install-docs --disable-unit-tests --target=x86_64-win64-gcc
-make -j6
-make install
-x86_64-w64-mingw32-dlltool -m i386:x86-64 -d libvpx.def -l /home/jim/packages/win64/bin/vpx.lib -D /home/jim/win64/packages/bin/libvpx-1.dll
-cd ..
-
-cd ffmpeg
-make clean
-cp /media/sf_linux/nvEncodeAPI.h /home/jim/packages/win64/include
-PKG_CONFIG_PATH="/home/jim/packages/win64/lib/pkgconfig" LDFLAGS="-L/home/jim/packages/win64/lib" CPPFLAGS="-I/home/jim/packages/win64/include" ./configure --enable-memalign-hack --enable-gpl --disable-doc --arch=x86_64 --enable-shared --enable-nvenc --enable-libx264 --enable-libopus --enable-libvorbis --enable-libvpx --disable-debug --cross-prefix=x86_64-w64-mingw32- --target-os=mingw32 --pkg-config=pkg-config --prefix="/home/jim/packages/win64" --disable-postproc
-read -n1 -r -p "Press any key to continue building FFmpeg..." key
-make -j6
-make install
-cd ..

obs-studio-26.1.0.tar.xz/libobs/util/simde/mmx.h Deleted

@@ -1,2270 +0,0 @@
-/* SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person
- * obtaining a copy of this software and associated documentation
- * files (the "Software"), to deal in the Software without
- * restriction, including without limitation the rights to use, copy,
- * modify, merge, publish, distribute, sublicense, and/or sell copies
- * of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- *
- * Copyright:
- * 2017-2020 Evan Nemerson <evan@nemerson.com>
- */
-
-#if !defined(SIMDE_X86_MMX_H)
-#define SIMDE_X86_MMX_H
-
-#include "simde-common.h"
-
-#if !defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES
-#endif
-
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
-
-#if defined(SIMDE_X86_MMX_NATIVE)
-#define SIMDE_X86_MMX_USE_NATIVE_TYPE
-#elif defined(SIMDE_X86_SSE_NATIVE)
-#define SIMDE_X86_MMX_USE_NATIVE_TYPE
-#endif
-
-#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
-#include <mmintrin.h>
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#include <arm_neon.h>
-#endif
-
-#include <stdint.h>
-#include <limits.h>
-
-SIMDE_BEGIN_DECLS_
-
-typedef union {
-#if defined(SIMDE_VECTOR_SUBSCRIPT)
-	SIMDE_ALIGN(8) int8_t i8 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) int16_t i16 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) int32_t i32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) int64_t i64 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) uint8_t u8 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) uint16_t u16 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) uint32_t u32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) uint64_t u64 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) simde_float32 f32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) int_fast32_t i32f SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(8) uint_fast32_t u32f SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-#else
-	SIMDE_ALIGN(8) int8_t i8[8];
-	SIMDE_ALIGN(8) int16_t i16[4];
-	SIMDE_ALIGN(8) int32_t i32[2];
-	SIMDE_ALIGN(8) int64_t i64[1];
-	SIMDE_ALIGN(8) uint8_t u8[8];
-	SIMDE_ALIGN(8) uint16_t u16[4];
-	SIMDE_ALIGN(8) uint32_t u32[2];
-	SIMDE_ALIGN(8) uint64_t u64[1];
-	SIMDE_ALIGN(8) simde_float32 f32[2];
-	SIMDE_ALIGN(8) int_fast32_t i32f[8 / sizeof(int_fast32_t)];
-	SIMDE_ALIGN(8) uint_fast32_t u32f[8 / sizeof(uint_fast32_t)];
-#endif
-
-#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
-	__m64 n;
-#endif
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int8x8_t neon_i8;
-	int16x4_t neon_i16;
-	int32x2_t neon_i32;
-	int64x1_t neon_i64;
-	uint8x8_t neon_u8;
-	uint16x4_t neon_u16;
-	uint32x2_t neon_u32;
-	uint64x1_t neon_u64;
-	float32x2_t neon_f32;
-#endif
-} simde__m64_private;
-
-#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
-typedef __m64 simde__m64;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-typedef int32x2_t simde__m64;
-#elif defined(SIMDE_VECTOR_SUBSCRIPT)
-typedef int32_t simde__m64 SIMDE_ALIGN(8) SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
-#else
-typedef simde__m64_private simde__m64;
-#endif
-
-#if !defined(SIMDE_X86_MMX_USE_NATIVE_TYPE) && \
-	defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES
-typedef simde__m64 __m64;
-#endif
-
-HEDLEY_STATIC_ASSERT(8 == sizeof(simde__m64), "__m64 size incorrect");
-HEDLEY_STATIC_ASSERT(8 == sizeof(simde__m64_private), "__m64 size incorrect");
-#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m64) == 8,
-		 "simde__m64 is not 8-byte aligned");
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m64_private) == 8,
-		 "simde__m64_private is not 8-byte aligned");
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde__m64_from_private(simde__m64_private v)
-{
-	simde__m64 r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64_private simde__m64_to_private(simde__m64 v)
-{
-	simde__m64_private r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-#define SIMDE_X86_GENERATE_CONVERSION_FUNCTION(simde_type, source_type, isax, \
-					 fragment) \
-	SIMDE_FUNCTION_ATTRIBUTES \
-	simde__##simde_type simde__##simde_type##_from_##isax##_##fragment( \
-		source_type value) \
-	{ \
-		simde__##simde_type##_private r_; \
-		r_.isax##_##fragment = value; \
-		return simde__##simde_type##_from_private(r_); \
-	} \
- \
-	SIMDE_FUNCTION_ATTRIBUTES \
-	source_type simde__##simde_type##_to_##isax##_##fragment( \
-		simde__##simde_type value) \
-	{ \
-		simde__##simde_type##_private r_ = \
-			simde__##simde_type##_to_private(value); \
-		return r_.isax##_##fragment; \
-	}
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int8x8_t, neon, i8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int16x4_t, neon, i16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int32x2_t, neon, i32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int64x1_t, neon, i64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint8x8_t, neon, u8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint16x4_t, neon, u16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint32x2_t, neon, u32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint64x1_t, neon, u64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, float32x2_t, neon, f32)
-#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_add_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_add_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vadd_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = a_.i8 + b_.i8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = a_.i8[i] + b_.i8[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddb(a, b) simde_mm_add_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_add_pi8(a, b) simde_mm_add_pi8(a, b)
-#define _m_paddb(a, b) simde_m_paddb(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_add_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_add_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vadd_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = a_.i16 + b_.i16;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] + b_.i16[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddw(a, b) simde_mm_add_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_add_pi16(a, b) simde_mm_add_pi16(a, b)
-#define _m_add_paddw(a, b) simde_mm_add_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_add_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_add_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vadd_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = a_.i32 + b_.i32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] + b_.i32[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddd(a, b) simde_mm_add_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_add_pi32(a, b) simde_mm_add_pi32(a, b)
-#define _m_add_paddd(a, b) simde_mm_add_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_adds_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_adds_pi8(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vqadd_s8(a_.neon_i8, b_.neon_i8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		if ((((b_.i8[i]) > 0) &&
-		 ((a_.i8[i]) > (INT8_MAX - (b_.i8[i]))))) {
-			r_.i8[i] = INT8_MAX;
-		} else if ((((b_.i8[i]) < 0) &&
-			 ((a_.i8[i]) < (INT8_MIN - (b_.i8[i]))))) {
-			r_.i8[i] = INT8_MIN;
-		} else {
-			r_.i8[i] = (a_.i8[i]) + (b_.i8[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddsb(a, b) simde_mm_adds_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_pi8(a, b) simde_mm_adds_pi8(a, b)
-#define _m_add_paddsb(a, b) simde_mm_adds_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_adds_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_adds_pu8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vqadd_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		const uint_fast16_t x =
-			HEDLEY_STATIC_CAST(uint_fast16_t, a_.u8[i]) +
-			HEDLEY_STATIC_CAST(uint_fast16_t, b_.u8[i]);
-		if (x > UINT8_MAX)
-			r_.u8[i] = UINT8_MAX;
-		else
-			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddusb(a, b) simde_mm_adds_pu8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_pu8(a, b) simde_mm_adds_pu8(a, b)
-#define _m_paddusb(a, b) simde_mm_adds_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_adds_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_adds_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vqadd_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if ((((b_.i16[i]) > 0) &&
-		 ((a_.i16[i]) > (INT16_MAX - (b_.i16[i]))))) {
-			r_.i16[i] = INT16_MAX;
-		} else if ((((b_.i16[i]) < 0) &&
-			 ((a_.i16[i]) < (SHRT_MIN - (b_.i16[i]))))) {
-			r_.i16[i] = SHRT_MIN;
-		} else {
-			r_.i16[i] = (a_.i16[i]) + (b_.i16[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddsw(a, b) simde_mm_adds_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_pi16(a, b) simde_mm_adds_pi16(a, b)
-#define _m_paddsw(a, b) simde_mm_adds_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_adds_pu16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_adds_pu16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vqadd_u16(a_.neon_u16, b_.neon_u16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		const uint32_t x = a_.u16[i] + b_.u16[i];
-		if (x > UINT16_MAX)
-			r_.u16[i] = UINT16_MAX;
-		else
-			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_paddusw(a, b) simde_mm_adds_pu16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_pu16(a, b) simde_mm_adds_pu16(a, b)
-#define _m_paddusw(a, b) simde_mm_adds_pu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_and_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_and_si64(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vand_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 & b_.i64;
-#else
-	r_.i64[0] = a_.i64[0] & b_.i64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pand(a, b) simde_mm_and_si64(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_and_si64(a, b) simde_mm_and_si64(a, b)
-#define _m_pand(a, b) simde_mm_and_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_andnot_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_andnot_si64(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vbic_s32(b_.neon_i32, a_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = ~a_.i32f & b_.i32f;
-#else
-	r_.u64[0] = (~(a_.u64[0])) & (b_.u64[0]);
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pandn(a, b) simde_mm_andnot_si64(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_andnot_si64(a, b) simde_mm_andnot_si64(a, b)
-#define _m_pandn(a, b) simde_mm_andnot_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpeq_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpeq_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vreinterpret_s8_u8(vceq_s8(a_.neon_i8, b_.neon_i8));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = (a_.i8[i] == b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpeqb(a, b) simde_mm_cmpeq_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_pi8(a, b) simde_mm_cmpeq_pi8(a, b)
-#define _m_pcmpeqb(a, b) simde_mm_cmpeq_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpeq_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpeq_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vreinterpret_s16_u16(vceq_s16(a_.neon_i16, b_.neon_i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] == b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpeqw(a, b) simde_mm_cmpeq_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_pi16(a, b) simde_mm_cmpeq_pi16(a, b)
-#define _m_pcmpeqw(a, b) simde_mm_cmpeq_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpeq_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpeq_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vreinterpret_s32_u32(vceq_s32(a_.neon_i32, b_.neon_i32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = (a_.i32[i] == b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpeqd(a, b) simde_mm_cmpeq_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_pi32(a, b) simde_mm_cmpeq_pi32(a, b)
-#define _m_pcmpeqd(a, b) simde_mm_cmpeq_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpgt_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpgt_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vreinterpret_s8_u8(vcgt_s8(a_.neon_i8, b_.neon_i8));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = (a_.i8[i] > b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpgtb(a, b) simde_mm_cmpgt_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_pi8(a, b) simde_mm_cmpgt_pi8(a, b)
-#define _m_pcmpgtb(a, b) simde_mm_cmpgt_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpgt_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpgt_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vreinterpret_s16_u16(vcgt_s16(a_.neon_i16, b_.neon_i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpgtw(a, b) simde_mm_cmpgt_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_pi16(a, b) simde_mm_cmpgt_pi16(a, b)
-#define _m_pcmpgtw(a, b) simde_mm_cmpgt_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cmpgt_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cmpgt_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vreinterpret_s32_u32(vcgt_s32(a_.neon_i32, b_.neon_i32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = (a_.i32[i] > b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pcmpgtd(a, b) simde_mm_cmpgt_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_pi32(a, b) simde_mm_cmpgt_pi32(a, b)
-#define _m_pcmpgtd(a, b) simde_mm_cmpgt_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvtm64_si64(simde__m64 a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
-	!defined(__PGI)
-	return _mm_cvtm64_si64(a);
-#else
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vget_lane_s64(a_.neon_i64, 0);
-#else
-	return a_.i64[0];
-#endif
-#endif
-}
-#define simde_m_to_int64(a) simde_mm_cvtm64_si64(a)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtm64_si64(a) simde_mm_cvtm64_si64(a)
-#define _m_to_int64(a) simde_mm_cvtm64_si64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtsi32_si64(int32_t a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtsi32_si64(a);
-#else
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int32_t av[sizeof(r_.neon_i32) / sizeof(r_.neon_i32[0])] = {a, 0};
-	r_.neon_i32 = vld1_s32(av);
-#else
-	r_.i32[0] = a;
-	r_.i32[1] = 0;
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_from_int(a) simde_mm_cvtsi32_si64(a)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi32_si64(a) simde_mm_cvtsi32_si64(a)
-#define _m_from_int(a) simde_mm_cvtsi32_si64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtsi64_m64(int64_t a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
-	!defined(__PGI)
-	return _mm_cvtsi64_m64(a);
-#else
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vld1_s64(&a);
-#else
-	r_.i64[0] = a;
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_from_int64(a) simde_mm_cvtsi64_m64(a)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi64_m64(a) simde_mm_cvtsi64_m64(a)
-#define _m_from_int64(a) simde_mm_cvtsi64_m64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvtsi64_si32(simde__m64 a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtsi64_si32(a);
-#else
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vget_lane_s32(a_.neon_i32, 0);
-#else
-	return a_.i32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi64_si32(a) simde_mm_cvtsi64_si32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_empty(void)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	_mm_empty();
-#else
-#endif
-}
-#define simde_m_empty() simde_mm_empty()
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_empty() simde_mm_empty()
-#define _m_empty() simde_mm_empty()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_madd_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_madd_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int32x4_t i1 = vmull_s16(a_.neon_i16, b_.neon_i16);
-	r_.neon_i32 = vpadd_s32(vget_low_s32(i1), vget_high_s32(i1));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i += 2) {
-		r_.i32[i / 2] = (a_.i16[i] * b_.i16[i]) +
-				(a_.i16[i + 1] * b_.i16[i + 1]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmaddwd(a, b) simde_mm_madd_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_madd_pi16(a, b) simde_mm_madd_pi16(a, b)
-#define _m_pmaddwd(a, b) simde_mm_madd_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_mulhi_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_mulhi_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int32x4_t t1 = vmull_s16(a_.neon_i16, b_.neon_i16);
-	const uint32x4_t t2 = vshrq_n_u32(vreinterpretq_u32_s32(t1), 16);
-	const uint16x4_t t3 = vmovn_u32(t2);
-	r_.neon_i16 = vreinterpret_s16_u16(t3);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = HEDLEY_STATIC_CAST(int16_t,
-					 ((a_.i16[i] * b_.i16[i]) >> 16));
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmulhw(a, b) simde_mm_mulhi_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_mulhi_pi16(a, b) simde_mm_mulhi_pi16(a, b)
-#define _m_pmulhw(a, b) simde_mm_mulhi_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_mullo_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_mullo_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int32x4_t t1 = vmull_s16(a_.neon_i16, b_.neon_i16);
-	const uint16x4_t t2 = vmovn_u32(vreinterpretq_u32_s32(t1));
-	r_.neon_i16 = vreinterpret_s16_u16(t2);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = HEDLEY_STATIC_CAST(
-			int16_t, ((a_.i16[i] * b_.i16[i]) & 0xffff));
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmullw(a, b) simde_mm_mullo_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_mullo_pi16(a, b) simde_mm_mullo_pi16(a, b)
-#define _m_pmullw(a, b) simde_mm_mullo_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_or_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_or_si64(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vorr_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 | b_.i64;
-#else
-	r_.i64[0] = a_.i64[0] | b_.i64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_por(a, b) simde_mm_or_si64(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_or_si64(a, b) simde_mm_or_si64(a, b)
-#define _m_por(a, b) simde_mm_or_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_packs_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_packs_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vqmovn_s16(vcombine_s16(a_.neon_i16, b_.neon_i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if (a_.i16[i] < INT8_MIN) {
-			r_.i8[i] = INT8_MIN;
-		} else if (a_.i16[i] > INT8_MAX) {
-			r_.i8[i] = INT8_MAX;
-		} else {
-			r_.i8[i] = HEDLEY_STATIC_CAST(int8_t, a_.i16[i]);
-		}
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if (b_.i16[i] < INT8_MIN) {
-			r_.i8[i + 4] = INT8_MIN;
-		} else if (b_.i16[i] > INT8_MAX) {
-			r_.i8[i + 4] = INT8_MAX;
-		} else {
-			r_.i8[i + 4] = HEDLEY_STATIC_CAST(int8_t, b_.i16[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_packsswb(a, b) simde_mm_packs_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_packs_pi16(a, b) simde_mm_packs_pi16(a, b)
-#define _m_packsswb(a, b) mm_packs_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_packs_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_packs_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vqmovn_s32(vcombine_s32(a_.neon_i32, b_.neon_i32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (8 / sizeof(a_.i32[0])); i++) {
-		if (a_.i32[i] < SHRT_MIN) {
-			r_.i16[i] = SHRT_MIN;
-		} else if (a_.i32[i] > INT16_MAX) {
-			r_.i16[i] = INT16_MAX;
-		} else {
-			r_.i16[i] = HEDLEY_STATIC_CAST(int16_t, a_.i32[i]);
-		}
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (8 / sizeof(b_.i32[0])); i++) {
-		if (b_.i32[i] < SHRT_MIN) {
-			r_.i16[i + 2] = SHRT_MIN;
-		} else if (b_.i32[i] > INT16_MAX) {
-			r_.i16[i + 2] = INT16_MAX;
-		} else {
-			r_.i16[i + 2] = HEDLEY_STATIC_CAST(int16_t, b_.i32[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_packssdw(a, b) simde_mm_packs_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_packs_pi32(a, b) simde_mm_packs_pi32(a, b)
-#define _m_packssdw(a, b) simde_mm_packs_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_packs_pu16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_packs_pu16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	const int16x8_t t1 = vcombine_s16(a_.neon_i16, b_.neon_i16);
-
-	/* Set elements which are < 0 to 0 */
-	const int16x8_t t2 =
-		vandq_s16(t1, vreinterpretq_s16_u16(vcgezq_s16(t1)));
-
-	/* Vector with all s16 elements set to UINT8_MAX */
-	const int16x8_t vmax = vmovq_n_s16((int16_t)UINT8_MAX);
-
-	/* Elements which are within the acceptable range */
-	const int16x8_t le_max =
-		vandq_s16(t2, vreinterpretq_s16_u16(vcleq_s16(t2, vmax)));
-	const int16x8_t gt_max =
-		vandq_s16(vmax, vreinterpretq_s16_u16(vcgtq_s16(t2, vmax)));
-
-	/* Final values as 16-bit integers */
-	const int16x8_t values = vorrq_s16(le_max, gt_max);
-
-	r_.neon_u8 = vmovn_u16(vreinterpretq_u16_s16(values));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if (a_.i16[i] > UINT8_MAX) {
-			r_.u8[i] = UINT8_MAX;
-		} else if (a_.i16[i] < 0) {
-			r_.u8[i] = 0;
-		} else {
-			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, a_.i16[i]);
-		}
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if (b_.i16[i] > UINT8_MAX) {
-			r_.u8[i + 4] = UINT8_MAX;
-		} else if (b_.i16[i] < 0) {
-			r_.u8[i + 4] = 0;
-		} else {
-			r_.u8[i + 4] = HEDLEY_STATIC_CAST(uint8_t, b_.i16[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_packuswb(a, b) simde_mm_packs_pu16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_packs_pu16(a, b) simde_mm_packs_pu16(a, b)
-#define _m_packuswb(a, b) simde_mm_packs_pu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set_pi8(int8_t e7, int8_t e6, int8_t e5, int8_t e4,
-			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0);
-#else
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int8_t v[sizeof(r_.i8) / sizeof(r_.i8[0])] = {e0, e1, e2, e3,
-							 e4, e5, e6, e7};
-	r_.neon_i8 = vld1_s8(v);
-#else
-	r_.i8[0] = e0;
-	r_.i8[1] = e1;
-	r_.i8[2] = e2;
-	r_.i8[3] = e3;
-	r_.i8[4] = e4;
-	r_.i8[5] = e5;
-	r_.i8[6] = e6;
-	r_.i8[7] = e7;
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0) \
-	simde_mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_set_pu8(uint8_t e7, uint8_t e6, uint8_t e5, uint8_t e4,
-			 uint8_t e3, uint8_t e2, uint8_t e1, uint8_t e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_X86_MMX_NATIVE)
-	r_.n = _mm_set_pi8(
-		HEDLEY_STATIC_CAST(int8_t, e7), HEDLEY_STATIC_CAST(int8_t, e6),
-		HEDLEY_STATIC_CAST(int8_t, e5), HEDLEY_STATIC_CAST(int8_t, e4),
-		HEDLEY_STATIC_CAST(int8_t, e3), HEDLEY_STATIC_CAST(int8_t, e2),
-		HEDLEY_STATIC_CAST(int8_t, e1), HEDLEY_STATIC_CAST(int8_t, e0));
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const uint8_t v[sizeof(r_.u8) / sizeof(r_.u8[0])] = {e0, e1, e2, e3,
-							 e4, e5, e6, e7};
-	r_.neon_u8 = vld1_u8(v);
-#else
-	r_.u8[0] = e0;
-	r_.u8[1] = e1;
-	r_.u8[2] = e2;
-	r_.u8[3] = e3;
-	r_.u8[4] = e4;
-	r_.u8[5] = e5;
-	r_.u8[6] = e6;
-	r_.u8[7] = e7;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set_pi16(int16_t e3, int16_t e2, int16_t e1, int16_t e0)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set_pi16(e3, e2, e1, e0);
-#else
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int16_t v[sizeof(r_.i16) / sizeof(r_.i16[0])] = {e0, e1, e2, e3};
-	r_.neon_i16 = vld1_s16(v);
-#else
-	r_.i16[0] = e0;
-	r_.i16[1] = e1;
-	r_.i16[2] = e2;
-	r_.i16[3] = e3;
-#endif
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set_pi16(e3, e2, e1, e0) simde_mm_set_pi16(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_set_pu16(uint16_t e3, uint16_t e2, uint16_t e1,
-			 uint16_t e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_X86_MMX_NATIVE)
-	r_.n = _mm_set_pi16(HEDLEY_STATIC_CAST(int16_t, e3),
-			 HEDLEY_STATIC_CAST(int16_t, e2),
-			 HEDLEY_STATIC_CAST(int16_t, e1),
-			 HEDLEY_STATIC_CAST(int16_t, e0));
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const uint16_t v[sizeof(r_.u16) / sizeof(r_.u16[0])] = {e0, e1, e2, e3};
-	r_.neon_u16 = vld1_u16(v);
-#else
-	r_.u16[0] = e0;
-	r_.u16[1] = e1;
-	r_.u16[2] = e2;
-	r_.u16[3] = e3;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_set_pu32(uint32_t e1, uint32_t e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_X86_MMX_NATIVE)
-	r_.n = _mm_set_pi32(HEDLEY_STATIC_CAST(int32_t, e1),
-			 HEDLEY_STATIC_CAST(int32_t, e0));
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const uint32_t v[sizeof(r_.u32) / sizeof(r_.u32[0])] = {e0, e1};
-	r_.neon_u32 = vld1_u32(v);
-#else
-	r_.u32[0] = e0;
-	r_.u32[1] = e1;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set_pi32(int32_t e1, int32_t e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_X86_MMX_NATIVE)
-	r_.n = _mm_set_pi32(e1, e0);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int32_t v[sizeof(r_.i32) / sizeof(r_.i32[0])] = {e0, e1};
-	r_.neon_i32 = vld1_s32(v);
-#else
-	r_.i32[0] = e0;
-	r_.i32[1] = e1;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set_pi32(e1, e0) simde_mm_set_pi32(e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_set_pi64(int64_t e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const int64_t v[sizeof(r_.i64) / sizeof(r_.i64[0])] = {e0};
-	r_.neon_i64 = vld1_s64(v);
-#else
-	r_.i64[0] = e0;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_set_f32x2(simde_float32 e1, simde_float32 e0)
-{
-	simde__m64_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	const simde_float32 v[sizeof(r_.f32) / sizeof(r_.f32[0])] = {e0, e1};
-	r_.neon_f32 = vld1_f32(v);
-#else
-	r_.f32[0] = e0;
-	r_.f32[1] = e1;
-#endif
-
-	return simde__m64_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set1_pi8(int8_t a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set1_pi8(a);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	simde__m64_private r_;
-	r_.neon_i8 = vmov_n_s8(a);
-	return simde__m64_from_private(r_);
-#else
-	return simde_mm_set_pi8(a, a, a, a, a, a, a, a);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_pi8(a) simde_mm_set1_pi8(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set1_pi16(int16_t a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set1_pi16(a);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	simde__m64_private r_;
-	r_.neon_i16 = vmov_n_s16(a);
-	return simde__m64_from_private(r_);
-#else
-	return simde_mm_set_pi16(a, a, a, a);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_pi16(a) simde_mm_set1_pi16(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_set1_pi32(int32_t a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set1_pi32(a);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	simde__m64_private r_;
-	r_.neon_i32 = vmov_n_s32(a);
-	return simde__m64_from_private(r_);
-#else
-	return simde_mm_set_pi32(a, a);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_pi32(a) simde_mm_set1_pi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_setr_pi8(int8_t e7, int8_t e6, int8_t e5, int8_t e4,
-			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0);
-#else
-	return simde_mm_set_pi8(e0, e1, e2, e3, e4, e5, e6, e7);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0) \
-	simde_mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_setr_pi16(int16_t e3, int16_t e2, int16_t e1, int16_t e0)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_setr_pi16(e3, e2, e1, e0);
-#else
-	return simde_mm_set_pi16(e0, e1, e2, e3);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_pi16(e3, e2, e1, e0) simde_mm_setr_pi16(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_setr_pi32(int32_t e1, int32_t e0)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_setr_pi32(e1, e0);
-#else
-	return simde_mm_set_pi32(e0, e1);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_pi32(e1, e0) simde_mm_setr_pi32(e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_setzero_si64(void)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_setzero_si64();
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	simde__m64_private r_;
-	r_.neon_u32 = vmov_n_u32(0);
-	return simde__m64_from_private(r_);
-#else
-	return simde_mm_set_pi32(0, 0);
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_setzero_si64() simde_mm_setzero_si64()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_x_mm_setone_si64(void)
-{
-	return simde_mm_set1_pi32(~INT32_C(0));
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sll_pi16(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sll_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vshl_s16(a_.neon_i16, vmov_n_s16((int16_t)vget_lane_u64(
-						 count_.neon_u64, 0)));
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i16 = a_.i16 << count_.u64[0];
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 15)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t,
-					 a_.u16[i] << count_.u64[0]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psllw(a, count) simde_mm_sll_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_pi16(a, count) simde_mm_sll_pi16(a, count)
-#define _m_psllw(a, count) simde_mm_sll_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sll_pi32(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sll_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vshl_s32(a_.neon_i32, vmov_n_s32((int32_t)vget_lane_u64(
-						 count_.neon_u64, 0)));
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i32 = a_.i32 << count_.u64[0];
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 31)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] << count_.u64[0];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pslld(a, count) simde_mm_sll_pi32(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_pi32(a, count) simde_mm_sll_pi32(a, count)
-#define _m_pslld(a, count) simde_mm_sll_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_slli_pi16(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_slli_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i16 = a_.i16 << count;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vshl_s16(a_.neon_i16, vmov_n_s16((int16_t)count));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, a_.u16[i] << count);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psllwi(a, count) simde_mm_slli_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_pi16(a, count) simde_mm_slli_pi16(a, count)
-#define _m_psllwi(a, count) simde_mm_slli_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_slli_pi32(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_slli_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i32 = a_.i32 << count;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vshl_s32(a_.neon_i32, vmov_n_s32((int32_t)count));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] << count;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pslldi(a, b) simde_mm_slli_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_pi32(a, count) simde_mm_slli_pi32(a, count)
-#define _m_pslldi(a, count) simde_mm_slli_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_slli_si64(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_slli_si64(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i64 = a_.i64 << count;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vshl_s64(a_.neon_i64, vmov_n_s64((int64_t)count));
-#else
-	r_.u64[0] = a_.u64[0] << count;
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psllqi(a, count) simde_mm_slli_si64(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_si64(a, count) simde_mm_slli_si64(a, count)
-#define _m_psllqi(a, count) simde_mm_slli_si64(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sll_si64(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sll_si64(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vshl_s64(a_.neon_i64, count_.neon_i64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 << count_.i64;
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 63)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	r_.u64[0] = a_.u64[0] << count_.u64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psllq(a, count) simde_mm_sll_si64(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_si64(a, count) simde_mm_sll_si64(a, count)
-#define _m_psllq(a, count) simde_mm_sll_si64(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srl_pi16(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_srl_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u16 = a_.u16 >> count_.u64[0];
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vshl_u16(
-		a_.neon_u16,
-		vmov_n_s16(-((int16_t)vget_lane_u64(count_.neon_u64, 0))));
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 15)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < sizeof(r_.u16) / sizeof(r_.u16[0]); i++) {
-		r_.u16[i] = a_.u16[i] >> count_.u64[0];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrlw(a, count) simde_mm_srl_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_pi16(a, count) simde_mm_srl_pi16(a, count)
-#define _m_psrlw(a, count) simde_mm_srl_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srl_pi32(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_srl_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u32 = a_.u32 >> count_.u64[0];
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vshl_u32(
-		a_.neon_u32,
-		vmov_n_s32(-((int32_t)vget_lane_u64(count_.neon_u64, 0))));
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 31)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < sizeof(r_.u32) / sizeof(r_.u32[0]); i++) {
-		r_.u32[i] = a_.u32[i] >> count_.u64[0];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrld(a, count) simde_mm_srl_pi32(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_pi32(a, count) simde_mm_srl_pi32(a, count)
-#define _m_psrld(a, count) simde_mm_srl_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srli_pi16(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_srli_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u16 = a_.u16 >> count;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vshl_u16(a_.neon_u16, vmov_n_s16(-((int16_t)count)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = a_.u16[i] >> count;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrlwi(a, count) simde_mm_srli_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_pi16(a, count) simde_mm_srli_pi16(a, count)
-#define _m_psrlwi(a, count) simde_mm_srli_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srli_pi32(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_srli_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u32 = a_.u32 >> count;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vshl_u32(a_.neon_u32, vmov_n_s32(-((int32_t)count)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] >> count;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrldi(a, count) simde_mm_srli_pi32(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_pi32(a, count) simde_mm_srli_pi32(a, count)
-#define _m_psrldi(a, count) simde_mm_srli_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srli_si64(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_srli_si64(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u64 = vshl_u64(a_.neon_u64, vmov_n_s64(-count));
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u64 = a_.u64 >> count;
-#else
-	r_.u64[0] = a_.u64[0] >> count;
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrlqi(a, count) simde_mm_srli_si64(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_si64(a, count) simde_mm_srli_si64(a, count)
-#define _m_psrlqi(a, count) simde_mm_srli_si64(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srl_si64(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_srl_si64(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_u64 = vshl_u64(a_.neon_u64, vneg_s64(count_.neon_i64));
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.u64 = a_.u64 >> count_.u64;
-#else
-	if (HEDLEY_UNLIKELY(count_.u64[0] > 63)) {
-		simde_memset(&r_, 0, sizeof(r_));
-		return simde__m64_from_private(r_);
-	}
-
-	r_.u64[0] = a_.u64[0] >> count_.u64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrlq(a, count) simde_mm_srl_si64(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_si64(a, count) simde_mm_srl_si64(a, count)
-#define _m_psrlq(a, count) simde_mm_srl_si64(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srai_pi16(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_srai_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i16 = a_.i16 >> (count & 0xff);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
- r_.neon_i16 = vshl_s16(a_.neon_i16, vmov_n_s16(-HEDLEY_STATIC_CAST(int16_t, count));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] >> (count & 0xff);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrawi(a, count) simde_mm_srai_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srai_pi16(a, count) simde_mm_srai_pi16(a, count)
-#define _m_psrawi(a, count) simde_mm_srai_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_srai_pi32(simde__m64 a, int count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
-	return _mm_srai_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i32 = a_.i32 >> (count & 0xff);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vshl_s32(a_.neon_i32,
-			 vmov_n_s32(-HEDLEY_STATIC_CAST(int32_t, count)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] >> (count & 0xff);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psradi(a, count) simde_mm_srai_pi32(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_srai_pi32(a, count) simde_mm_srai_pi32(a, count)
-#define _m_srai_pi32(a, count) simde_mm_srai_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sra_pi16(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sra_pi16(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-	const int cnt = HEDLEY_STATIC_CAST(
-		int, (count_.i64[0] > 15 ? 15 : count_.i64[0]));
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i16 = a_.i16 >> cnt;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 =
-		vshl_s16(a_.neon_i16,
-			 vmov_n_s16(-HEDLEY_STATIC_CAST(
-				 int16_t, vget_lane_u64(count_.neon_u64, 0))));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] >> cnt;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psraw(a, count) simde_mm_sra_pi16(a, count)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sra_pi16(a, count) simde_mm_sra_pi16(a, count)
-#define _m_psraw(a, count) simde_mm_sra_pi16(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sra_pi32(simde__m64 a, simde__m64 count)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sra_pi32(a, count);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private count_ = simde__m64_to_private(count);
-	const int32_t cnt =
-		(count_.u64[0] > 31)
-			? 31
-			: HEDLEY_STATIC_CAST(int32_t, count_.u64[0]);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i32 = a_.i32 >> cnt;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 =
-		vshl_s32(a_.neon_i32,
-			 vmov_n_s32(-HEDLEY_STATIC_CAST(
-				 int32_t, vget_lane_u64(count_.neon_u64, 0))));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] >> cnt;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psrad(a, b) simde_mm_sra_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sra_pi32(a, count) simde_mm_sra_pi32(a, count)
-#define _m_psrad(a, count) simde_mm_sra_pi32(a, count)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sub_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sub_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vsub_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = a_.i8 - b_.i8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = a_.i8[i] - b_.i8[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubb(a, b) simde_mm_sub_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_pi8(a, b) simde_mm_sub_pi8(a, b)
-#define _m_psubb(a, b) simde_mm_sub_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sub_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sub_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vsub_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = a_.i16 - b_.i16;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] - b_.i16[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubw(a, b) simde_mm_sub_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_pi16(a, b) simde_mm_sub_pi16(a, b)
-#define _m_psubw(a, b) simde_mm_sub_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sub_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sub_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vsub_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = a_.i32 - b_.i32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] - b_.i32[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubd(a, b) simde_mm_sub_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_pi32(a, b) simde_mm_sub_pi32(a, b)
-#define _m_psubd(a, b) simde_mm_sub_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_subs_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_subs_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vqsub_s8(a_.neon_i8, b_.neon_i8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		if (((b_.i8[i]) > 0 && (a_.i8[i]) < INT8_MIN + (b_.i8[i]))) {
-			r_.i8[i] = INT8_MIN;
-		} else if ((b_.i8[i]) < 0 &&
-			 (a_.i8[i]) > INT8_MAX + (b_.i8[i])) {
-			r_.i8[i] = INT8_MAX;
-		} else {
-			r_.i8[i] = (a_.i8[i]) - (b_.i8[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubsb(a, b) simde_mm_subs_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_pi8(a, b) simde_mm_subs_pi8(a, b)
-#define _m_psubsb(a, b) simde_mm_subs_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_subs_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_subs_pu8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vqsub_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		const int32_t x = a_.u8[i] - b_.u8[i];
-		if (x < 0) {
-			r_.u8[i] = 0;
-		} else if (x > UINT8_MAX) {
-			r_.u8[i] = UINT8_MAX;
-		} else {
-			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubusb(a, b) simde_mm_subs_pu8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_pu8(a, b) simde_mm_subs_pu8(a, b)
-#define _m_psubusb(a, b) simde_mm_subs_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_subs_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_subs_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vqsub_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		if (((b_.i16[i]) > 0 && (a_.i16[i]) < SHRT_MIN + (b_.i16[i]))) {
-			r_.i16[i] = SHRT_MIN;
-		} else if ((b_.i16[i]) < 0 &&
-			 (a_.i16[i]) > INT16_MAX + (b_.i16[i])) {
-			r_.i16[i] = INT16_MAX;
-		} else {
-			r_.i16[i] = (a_.i16[i]) - (b_.i16[i]);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubsw(a, b) simde_mm_subs_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_pi16(a, b) simde_mm_subs_pi16(a, b)
-#define _m_psubsw(a, b) simde_mm_subs_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_subs_pu16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_subs_pu16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vqsub_u16(a_.neon_u16, b_.neon_u16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		const int x = a_.u16[i] - b_.u16[i];
-		if (x < 0) {
-			r_.u16[i] = 0;
-		} else if (x > UINT16_MAX) {
-			r_.u16[i] = UINT16_MAX;
-		} else {
-			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
-		}
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psubusw(a, b) simde_mm_subs_pu16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_pu16(a, b) simde_mm_subs_pu16(a, b)
-#define _m_psubusw(a, b) simde_mm_subs_pu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpackhi_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpackhi_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i8 = vzip2_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 8, a_.i8, b_.i8, 4, 12, 5, 13, 6, 14,
-				 7, 15);
-#else
-	r_.i8[0] = a_.i8[4];
-	r_.i8[1] = b_.i8[4];
-	r_.i8[2] = a_.i8[5];
-	r_.i8[3] = b_.i8[5];
-	r_.i8[4] = a_.i8[6];
-	r_.i8[5] = b_.i8[6];
-	r_.i8[6] = a_.i8[7];
-	r_.i8[7] = b_.i8[7];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpckhbw(a, b) simde_mm_unpackhi_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_pi8(a, b) simde_mm_unpackhi_pi8(a, b)
-#define _m_punpckhbw(a, b) simde_mm_unpackhi_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpackhi_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpackhi_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i16 = vzip2_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 8, a_.i16, b_.i16, 2, 6, 3, 7);
-#else
-	r_.i16[0] = a_.i16[2];
-	r_.i16[1] = b_.i16[2];
-	r_.i16[2] = a_.i16[3];
-	r_.i16[3] = b_.i16[3];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpckhwd(a, b) simde_mm_unpackhi_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_pi16(a, b) simde_mm_unpackhi_pi16(a, b)
-#define _m_punpckhwd(a, b) simde_mm_unpackhi_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpackhi_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpackhi_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i32 = vzip2_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 8, a_.i32, b_.i32, 1, 3);
-#else
-	r_.i32[0] = a_.i32[1];
-	r_.i32[1] = b_.i32[1];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpckhdq(a, b) simde_mm_unpackhi_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_pi32(a, b) simde_mm_unpackhi_pi32(a, b)
-#define _m_punpckhdq(a, b) simde_mm_unpackhi_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpacklo_pi8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpacklo_pi8(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i8 = vzip1_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 8, a_.i8, b_.i8, 0, 8, 1, 9, 2, 10, 3,
-				 11);
-#else
-	r_.i8[0] = a_.i8[0];
-	r_.i8[1] = b_.i8[0];
-	r_.i8[2] = a_.i8[1];
-	r_.i8[3] = b_.i8[1];
-	r_.i8[4] = a_.i8[2];
-	r_.i8[5] = b_.i8[2];
-	r_.i8[6] = a_.i8[3];
-	r_.i8[7] = b_.i8[3];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpcklbw(a, b) simde_mm_unpacklo_pi8(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_pi8(a, b) simde_mm_unpacklo_pi8(a, b)
-#define _m_punpcklbw(a, b) simde_mm_unpacklo_pi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpacklo_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpacklo_pi16(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i16 = vzip1_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 8, a_.i16, b_.i16, 0, 4, 1, 5);
-#else
-	r_.i16[0] = a_.i16[0];
-	r_.i16[1] = b_.i16[0];
-	r_.i16[2] = a_.i16[1];
-	r_.i16[3] = b_.i16[1];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpcklwd(a, b) simde_mm_unpacklo_pi16(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_pi16(a, b) simde_mm_unpacklo_pi16(a, b)
-#define _m_punpcklwd(a, b) simde_mm_unpacklo_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_unpacklo_pi32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_unpacklo_pi32(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i32 = vzip1_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 8, a_.i32, b_.i32, 0, 2);
-#else
-	r_.i32[0] = a_.i32[0];
-	r_.i32[1] = b_.i32[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_punpckldq(a, b) simde_mm_unpacklo_pi32(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_pi32(a, b) simde_mm_unpacklo_pi32(a, b)
-#define _m_punpckldq(a, b) simde_mm_unpacklo_pi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_xor_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_xor_si64(a, b);
-#else
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = veor_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f ^ b_.i32f;
-#else
-	r_.u64[0] = a_.u64[0] ^ b_.u64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pxor(a, b) simde_mm_xor_si64(a, b)
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _mm_xor_si64(a, b) simde_mm_xor_si64(a, b)
-#define _m_pxor(a, b) simde_mm_xor_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_m_to_int(simde__m64 a)
-{
-#if defined(SIMDE_X86_MMX_NATIVE)
-	return _m_to_int(a);
-#else
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vget_lane_s32(a_.neon_i32, 0);
-#else
-	return a_.i32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
-#define _m_to_int(a) simde_m_to_int(a)
-#endif
-
-SIMDE_END_DECLS_
-
-HEDLEY_DIAGNOSTIC_POP
-
-#endif /* !defined(SIMDE_X86_MMX_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/sse.h Deleted

@@ -1,3718 +0,0 @@
-/* SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person
- * obtaining a copy of this software and associated documentation
- * files (the "Software"), to deal in the Software without
- * restriction, including without limitation the rights to use, copy,
- * modify, merge, publish, distribute, sublicense, and/or sell copies
- * of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- *
- * Copyright:
- * 2017-2020 Evan Nemerson <evan@nemerson.com>
- * 2015-2017 John W. Ratcliff <jratcliffscarab@gmail.com>
- * 2015 Brandon Rowlett <browlett@nvidia.com>
- * 2015 Ken Fast <kfast@gdeb.com>
- */
-
-#if !defined(SIMDE_X86_SSE_H)
-#define SIMDE_X86_SSE_H
-
-#include "mmx.h"
-
-#if !defined(SIMDE_X86_AVX512F_NATIVE) && defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES
-#endif
-
-#if defined(_WIN32)
-#include <windows.h>
-#endif
-
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
-SIMDE_BEGIN_DECLS_
-
-typedef union {
-#if defined(SIMDE_VECTOR_SUBSCRIPT)
-	SIMDE_ALIGN(16) int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#if defined(SIMDE_HAVE_INT128_)
-	SIMDE_ALIGN(16) simde_int128 i128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) simde_uint128 u128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#endif
-	SIMDE_ALIGN(16) simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-	SIMDE_ALIGN(16) int8_t i8[16];
-	SIMDE_ALIGN(16) int16_t i16[8];
-	SIMDE_ALIGN(16) int32_t i32[4];
-	SIMDE_ALIGN(16) int64_t i64[2];
-	SIMDE_ALIGN(16) uint8_t u8[16];
-	SIMDE_ALIGN(16) uint16_t u16[8];
-	SIMDE_ALIGN(16) uint32_t u32[4];
-	SIMDE_ALIGN(16) uint64_t u64[2];
-#if defined(SIMDE_HAVE_INT128_)
-	SIMDE_ALIGN(16) simde_int128 i128[1];
-	SIMDE_ALIGN(16) simde_uint128 u128[1];
-#endif
-	SIMDE_ALIGN(16) simde_float32 f32[4];
-	SIMDE_ALIGN(16) int_fast32_t i32f[16 / sizeof(int_fast32_t)];
-	SIMDE_ALIGN(16) uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
-#endif
-
-	SIMDE_ALIGN(16) simde__m64_private m64_private[2];
-	SIMDE_ALIGN(16) simde__m64 m64[2];
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	SIMDE_ALIGN(16) __m128 n;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN(16) int8x16_t neon_i8;
-	SIMDE_ALIGN(16) int16x8_t neon_i16;
-	SIMDE_ALIGN(16) int32x4_t neon_i32;
-	SIMDE_ALIGN(16) int64x2_t neon_i64;
-	SIMDE_ALIGN(16) uint8x16_t neon_u8;
-	SIMDE_ALIGN(16) uint16x8_t neon_u16;
-	SIMDE_ALIGN(16) uint32x4_t neon_u32;
-	SIMDE_ALIGN(16) uint64x2_t neon_u64;
-	SIMDE_ALIGN(16) float32x4_t neon_f32;
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	SIMDE_ALIGN(16) float64x2_t neon_f64;
-#endif
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	SIMDE_ALIGN(16) v128_t wasm_v128;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
-#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
-#endif
-#endif
-} simde__m128_private;
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-typedef __m128 simde__m128;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-typedef float32x4_t simde__m128;
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-typedef v128_t simde__m128;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128;
-#elif defined(SIMDE_VECTOR_SUBSCRIPT)
-typedef simde_float32 simde__m128 SIMDE_ALIGN(16)
-	SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-typedef simde__m128_private simde__m128;
-#endif
-
-#if !defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES
-typedef simde__m128 __m128;
-#endif
-
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128), "simde__m128 size incorrect");
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128_private),
-		 "simde__m128_private size incorrect");
-#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128) == 16,
-		 "simde__m128 is not 16-byte aligned");
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128_private) == 16,
-		 "simde__m128_private is not 16-byte aligned");
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde__m128_from_private(simde__m128_private v)
-{
-	simde__m128 r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128_private simde__m128_to_private(simde__m128 v)
-{
-	simde__m128_private r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int8x16_t, neon, i8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int16x8_t, neon, i16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int32x4_t, neon, i32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int64x2_t, neon, i64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint8x16_t, neon, u8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint16x8_t, neon, u16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint32x4_t, neon, u32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint64x2_t, neon, u64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float32x4_t, neon, f32)
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float64x2_t, neon, f64)
-#endif
-#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_POP
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_set_ps(simde_float32 e3, simde_float32 e2,
-			 simde_float32 e1, simde_float32 e0)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_set_ps(e3, e2, e1, e0);
-#else
-	simde__m128_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN(16) simde_float32 data[4] = {e0, e1, e2, e3};
-	r_.neon_f32 = vld1q_f32(data);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_make(e0, e1, e2, e3);
-#else
-	r_.f32[0] = e0;
-	r_.f32[1] = e1;
-	r_.f32[2] = e2;
-	r_.f32[3] = e3;
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_set_ps(e3, e2, e1, e0) simde_mm_set_ps(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_set_ps1(simde_float32 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_set_ps1(a);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vdupq_n_f32(a);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	(void)a;
-	return vec_splats(a);
-#else
-	return simde_mm_set_ps(a, a, a, a);
-#endif
-}
-#define simde_mm_set1_ps(a) simde_mm_set_ps1(a)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_set_ps1(a) simde_mm_set_ps1(a)
-#define _mm_set1_ps(a) simde_mm_set1_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_move_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_move_ss(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 =
-		vsetq_lane_f32(vgetq_lane_f32(b_.neon_f32, 0), a_.neon_f32, 0);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
-	m = {16, 17, 18, 19, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
-	r_.altivec_f32 = vec_perm(a_.altivec_f32, b_.altivec_f32, m);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_v8x16_shuffle(b_.wasm_v128, a_.wasm_v128, 0, 1, 2,
-					 3, 20, 21, 22, 23, 24, 25, 26, 27, 28,
-					 29, 30, 31);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 4, 1, 2, 3);
-#else
-	r_.f32[0] = b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_move_ss(a, b) simde_mm_move_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_add_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_add_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vaddq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_add(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_add(a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f32 = a_.f32 + b_.f32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = a_.f32[i] + b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_add_ps(a, b) simde_mm_add_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_add_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_add_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_add_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = a_.f32[0] + b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_add_ss(a, b) simde_mm_add_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_and_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_and_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vandq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = a_.i32 & b_.i32;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_and(a_.altivec_f32, b_.altivec_f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] & b_.i32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_and_ps(a, b) simde_mm_and_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_andnot_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_andnot_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_andc(b_.altivec_f32, a_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = ~a_.i32 & b_.i32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = ~(a_.i32[i]) & b_.i32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_andnot_ps(a, b) simde_mm_andnot_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_avg_pu16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_avg_pu16(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vrhadd_u16(b_.neon_u16, a_.neon_u16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
-	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
-	defined(SIMDE_CONVERT_VECTOR_)
-	uint32_t wa SIMDE_VECTOR(16);
-	uint32_t wb SIMDE_VECTOR(16);
-	uint32_t wr SIMDE_VECTOR(16);
-	SIMDE_CONVERT_VECTOR_(wa, a_.u16);
-	SIMDE_CONVERT_VECTOR_(wb, b_.u16);
-	wr = (wa + wb + 1) >> 1;
-	SIMDE_CONVERT_VECTOR_(r_.u16, wr);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = (a_.u16[i] + b_.u16[i] + 1) >> 1;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pavgw(a, b) simde_mm_avg_pu16(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_avg_pu16(a, b) simde_mm_avg_pu16(a, b)
-#define _m_pavgw(a, b) simde_mm_avg_pu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_avg_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_avg_pu8(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vrhadd_u8(b_.neon_u8, a_.neon_u8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
-	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
-	defined(SIMDE_CONVERT_VECTOR_)
-	uint16_t wa SIMDE_VECTOR(16);
-	uint16_t wb SIMDE_VECTOR(16);
-	uint16_t wr SIMDE_VECTOR(16);
-	SIMDE_CONVERT_VECTOR_(wa, a_.u8);
-	SIMDE_CONVERT_VECTOR_(wb, b_.u8);
-	wr = (wa + wb + 1) >> 1;
-	SIMDE_CONVERT_VECTOR_(r_.u8, wr);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] + b_.u8[i] + 1) >> 1;
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pavgb(a, b) simde_mm_avg_pu8(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_avg_pu8(a, b) simde_mm_avg_pu8(a, b)
-#define _m_pavgb(a, b) simde_mm_avg_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpeq_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpeq_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vceqq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_eq(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmpeq(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), a_.f32 == b_.f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] == b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_ps(a, b) simde_mm_cmpeq_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpeq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpeq_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpeq_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] == b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_ss(a, b) simde_mm_cmpeq_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpge_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpge_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vcgeq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_ge(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmpge(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 >= b_.f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] >= b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpge_ps(a, b) simde_mm_cmpge_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpge_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
-	return _mm_cmpge_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpge_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] >= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpge_ss(a, b) simde_mm_cmpge_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpgt_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpgt_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vcgtq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_gt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmpgt(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 > b_.f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] > b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_ps(a, b) simde_mm_cmpgt_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpgt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
-	return _mm_cmpgt_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpgt_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] > b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_ss(a, b) simde_mm_cmpgt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmple_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmple_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vcleq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_le(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmple(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 <= b_.f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] <= b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmple_ps(a, b) simde_mm_cmple_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmple_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmple_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmple_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] <= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmple_ss(a, b) simde_mm_cmple_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmplt_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmplt_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vcltq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_lt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmplt(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 < b_.f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] < b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_ps(a, b) simde_mm_cmplt_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmplt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmplt_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmplt_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] < b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_ss(a, b) simde_mm_cmplt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpneq_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpneq_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_ne(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P9_NATIVE) && SIMDE_ARCH_POWER_CHECK(900) && \
-	!defined(HEDLEY_IBM_VERSION)
-	/* vec_cmpne(vector float, vector float) is missing from XL C/C++ v16.1.1,
- though the documentation (table 89 on page 432 of the IBM XL C/C++ for
- Linux Compiler Reference, Version 16.1.1) shows that it should be
- present. Both GCC and clang support it. */
-	r_.altivec_f32 = (SIMDE_POWER_ALTIVEC_VECTOR(float))vec_cmpne(
-		a_.altivec_f32, b_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 != b_.f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (a_.f32[i] != b_.f32[i]) ? ~UINT32_C(0)
-						 : UINT32_C(0);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpneq_ps(a, b) simde_mm_cmpneq_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpneq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpneq_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpneq_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.u32[0] = (a_.f32[0] != b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpneq_ss(a, b) simde_mm_cmpneq_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnge_ps(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmplt_ps(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnge_ps(a, b) simde_mm_cmpnge_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnge_ss(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmplt_ss(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnge_ss(a, b) simde_mm_cmpnge_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpngt_ps(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmple_ps(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpngt_ps(a, b) simde_mm_cmpngt_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpngt_ss(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmple_ss(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpngt_ss(a, b) simde_mm_cmpngt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnle_ps(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmpgt_ps(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnle_ps(a, b) simde_mm_cmpnle_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnle_ss(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmpgt_ss(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnle_ss(a, b) simde_mm_cmpnle_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnlt_ps(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmpge_ps(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnlt_ps(a, b) simde_mm_cmpnlt_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpnlt_ss(simde__m128 a, simde__m128 b)
-{
-	return simde_mm_cmpge_ss(a, b);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnlt_ss(a, b) simde_mm_cmpnlt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpord_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpord_ps(a, b);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	return wasm_v128_and(wasm_f32x4_eq(a, a), wasm_f32x4_eq(b, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	/* Note: NEON does not have ordered compare builtin
- Need to compare a eq a and b eq b to check for NaN
- Do AND of results to get final */
-	uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	r_.neon_u32 = vandq_u32(ceqaa, ceqbb);
-#elif defined(simde_math_isnanf)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (simde_math_isnanf(a_.f32[i]) ||
-			 simde_math_isnanf(b_.f32[i]))
-				 ? UINT32_C(0)
-				 : ~UINT32_C(0);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpord_ps(a, b) simde_mm_cmpord_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpunord_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpunord_ps(a, b);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	return wasm_v128_or(wasm_f32x4_ne(a, a), wasm_f32x4_ne(b, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	r_.neon_u32 = vmvnq_u32(vandq_u32(ceqaa, ceqbb));
-#elif defined(simde_math_isnanf)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = (simde_math_isnanf(a_.f32[i]) ||
-			 simde_math_isnanf(b_.f32[i]))
-				 ? ~UINT32_C(0)
-				 : UINT32_C(0);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpunord_ps(a, b) simde_mm_cmpunord_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpunord_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
-	return _mm_cmpunord_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpunord_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(simde_math_isnanf)
-	r_.u32[0] =
-		(simde_math_isnanf(a_.f32[0]) || simde_math_isnanf(b_.f32[0]))
-			? ~UINT32_C(0)
-			: UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpunord_ss(a, b) simde_mm_cmpunord_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comieq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comieq_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
-	uint32x4_t a_eq_b = vceqq_f32(a_.neon_f32, b_.neon_f32);
-	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_eq_b), 0) != 0);
-#else
-	return a_.f32[0] == b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comieq_ss(a, b) simde_mm_comieq_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comige_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comige_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_ge_b = vcgeq_f32(a_.neon_f32, b_.neon_f32);
-	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0);
-#else
-	return a_.f32[0] >= b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comige_ss(a, b) simde_mm_comige_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comigt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comigt_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_gt_b = vcgtq_f32(a_.neon_f32, b_.neon_f32);
-	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0);
-#else
-	return a_.f32[0] > b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comigt_ss(a, b) simde_mm_comigt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comile_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comile_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
-	uint32x4_t a_le_b = vcleq_f32(a_.neon_f32, b_.neon_f32);
-	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_le_b), 0) != 0);
-#else
-	return a_.f32[0] <= b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comile_ss(a, b) simde_mm_comile_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comilt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comilt_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
-	uint32x4_t a_lt_b = vcltq_f32(a_.neon_f32, b_.neon_f32);
-	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_lt_b), 0) != 0);
-#else
-	return a_.f32[0] < b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comilt_ss(a, b) simde_mm_comilt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comineq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_comineq_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
-	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
-	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_neq_b), 0) != 0);
-#else
-	return a_.f32[0] != b_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_comineq_ss(a, b) simde_mm_comineq_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvt_pi2ps(simde__m128 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvt_pi2ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32),
-				 vget_high_f32(a_.neon_f32));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
-	r_.m64_private[1] = a_.m64_private[1];
-
-#else
-	r_.f32[0] = (simde_float32)b_.i32[0];
-	r_.f32[1] = (simde_float32)b_.i32[1];
-	r_.i32[2] = a_.i32[2];
-	r_.i32[3] = a_.i32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvt_pi2ps(a, b) simde_mm_cvt_pi2ps((a), b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvt_ps2pi(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvt_ps2pi(a);
-#else
-	simde__m64_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
-#elif defined(SIMDE_CONVERT_VECTOR_) && !defined(__clang__)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.m64_private[0].f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = HEDLEY_STATIC_CAST(int32_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvt_ps2pi(a) simde_mm_cvt_ps2pi((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvt_si2ss(simde__m128 a, int32_t b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cvt_si2ss(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vsetq_lane_f32((float)b, a_.neon_f32, 0);
-#else
-	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
-	r_.i32[1] = a_.i32[1];
-	r_.i32[2] = a_.i32[2];
-	r_.i32[3] = a_.i32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvt_si2ss(a, b) simde_mm_cvt_si2ss((a), b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvt_ss2si(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cvt_ss2si(a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
-	return vgetq_lane_s32(vcvtnq_s32_f32(a_.neon_f32), 0);
-#elif defined(simde_math_nearbyintf)
-	return SIMDE_CONVERT_FTOI(int32_t, simde_math_nearbyintf(a_.f32[0]));
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvt_ss2si(a) simde_mm_cvt_ss2si((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpi16_ps(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpi16_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && 0 /* TODO */
-	r_.neon_f32 = vmovl_s16(
-		vget_low_s16(vuzp1q_s16(a_.neon_i16, vmovq_n_s16(0))));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f32, a_.i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		simde_float32 v = a_.i16[i];
-		r_.f32[i] = v;
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpi16_ps(a) simde_mm_cvtpi16_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpi32_ps(simde__m128 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpi32_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-	simde__m64_private b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32),
-				 vget_high_f32(a_.neon_f32));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
-	r_.m64_private[1] = a_.m64_private[1];
-#else
-	r_.f32[0] = (simde_float32)b_.i32[0];
-	r_.f32[1] = (simde_float32)b_.i32[1];
-	r_.i32[2] = a_.i32[2];
-	r_.i32[3] = a_.i32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpi32_ps(a, b) simde_mm_cvtpi32_ps((a), b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpi32x2_ps(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpi32x2_ps(a, b);
-#else
-	simde__m128_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcvtq_f32_s32(vcombine_s32(a_.neon_i32, b_.neon_i32));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, a_.i32);
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[1].f32, b_.i32);
-#else
-	r_.f32[0] = (simde_float32)a_.i32[0];
-	r_.f32[1] = (simde_float32)a_.i32[1];
-	r_.f32[2] = (simde_float32)b_.i32[0];
-	r_.f32[3] = (simde_float32)b_.i32[1];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpi32x2_ps(a, b) simde_mm_cvtpi32x2_ps(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpi8_ps(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpi8_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 =
-		vcvtq_f32_s32(vmovl_s16(vget_low_s16(vmovl_s8(a_.neon_i8))));
-#else
-	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[0]);
-	r_.f32[1] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[1]);
-	r_.f32[2] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[2]);
-	r_.f32[3] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[3]);
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpi8_ps(a) simde_mm_cvtpi8_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtps_pi16(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtps_pi16(a);
-#else
-	simde__m64_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i16, a_.f32);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vmovn_s32(vcvtq_s32_f32(a_.neon_f32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = SIMDE_CONVERT_FTOI(int16_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtps_pi16(a) simde_mm_cvtps_pi16((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtps_pi32(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtps_pi32(a);
-#else
-	simde__m64_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.m64_private[0].f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtps_pi32(a) simde_mm_cvtps_pi32((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtps_pi8(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtps_pi8(a);
-#else
-	simde__m64_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int16x4_t b = vmovn_s32(vcvtq_s32_f32(a_.neon_f32));
-	int16x8_t c = vcombine_s16(b, vmov_n_s16(0));
-	r_.neon_i8 = vmovn_s16(c);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(a_.f32) / sizeof(a_.f32[0])); i++) {
-		r_.i8[i] = SIMDE_CONVERT_FTOI(int8_t, a_.f32[i]);
-	}
-	/* Note: the upper half is undefined */
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtps_pi8(a) simde_mm_cvtps_pi8((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpu16_ps(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpu16_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcvtq_f32_u32(vmovl_u16(a_.neon_u16));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f32, a_.u16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = (simde_float32)a_.u16[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpu16_ps(a) simde_mm_cvtpu16_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpu8_ps(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpu8_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 =
-		vcvtq_f32_u32(vmovl_u16(vget_low_u16(vmovl_u8(a_.neon_u8))));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = HEDLEY_STATIC_CAST(simde_float32, a_.u8[i]);
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpu8_ps(a) simde_mm_cvtpu8_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtsi32_ss(simde__m128 a, int32_t b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cvtsi32_ss(a, b);
-#else
-	simde__m128_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vsetq_lane_f32((simde_float32)b, a_.neon_f32, 0);
-#else
-	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi32_ss(a, b) simde_mm_cvtsi32_ss((a), b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtsi64_ss(simde__m128 a, int64_t b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if !defined(__PGI)
-	return _mm_cvtsi64_ss(a, b);
-#else
-	return _mm_cvtsi64x_ss(a, b);
-#endif
-#else
-	simde__m128_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vsetq_lane_f32((simde_float32)b, a_.neon_f32, 0);
-#else
-	r_ = a_;
-	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi64_ss(a, b) simde_mm_cvtsi64_ss((a), b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde_float32 simde_mm_cvtss_f32(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cvtss_f32(a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vgetq_lane_f32(a_.neon_f32, 0);
-#else
-	return a_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtss_f32(a) simde_mm_cvtss_f32((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvtss_si32(simde__m128 a)
-{
-	return simde_mm_cvt_ss2si(a);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtss_si32(a) simde_mm_cvtss_si32((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvtss_si64(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if !defined(__PGI)
-	return _mm_cvtss_si64(a);
-#else
-	return _mm_cvtss_si64x(a);
-#endif
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return SIMDE_CONVERT_FTOI(int64_t, vgetq_lane_f32(a_.neon_f32, 0));
-#else
-	return SIMDE_CONVERT_FTOI(int64_t, a_.f32[0]);
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtss_si64(a) simde_mm_cvtss_si64((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtt_ps2pi(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtt_ps2pi(a);
-#else
-	simde__m64_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.m64_private[0].f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_mm_cvttps_pi32(a) simde_mm_cvtt_ps2pi(a)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtt_ps2pi(a) simde_mm_cvtt_ps2pi((a))
-#define _mm_cvttps_pi32(a) simde_mm_cvttps_pi32((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvtt_ss2si(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cvtt_ss2si(a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return SIMDE_CONVERT_FTOI(int32_t, vgetq_lane_f32(a_.neon_f32, 0));
-#else
-	return SIMDE_CONVERT_FTOI(int32_t, a_.f32[0]);
-#endif
-#endif
-}
-#define simde_mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtt_ss2si(a) simde_mm_cvtt_ss2si((a))
-#define _mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvttss_si64(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
-	!defined(_MSC_VER)
-#if defined(__PGI)
-	return _mm_cvttss_si64x(a);
-#else
-	return _mm_cvttss_si64(a);
-#endif
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return SIMDE_CONVERT_FTOI(int64_t, vgetq_lane_f32(a_.neon_f32, 0));
-#else
-	return SIMDE_CONVERT_FTOI(int64_t, a_.f32[0]);
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttss_si64(a) simde_mm_cvttss_si64((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cmpord_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_cmpord_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_cmpord_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(simde_math_isnanf)
-	r_.u32[0] = (simde_math_isnanf(simde_mm_cvtss_f32(a)) ||
-		 simde_math_isnanf(simde_mm_cvtss_f32(b)))
-			 ? UINT32_C(0)
-			 : ~UINT32_C(0);
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.u32[i] = a_.u32[i];
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpord_ss(a, b) simde_mm_cmpord_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_div_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_div_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f32 = vdivq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	float32x4_t recip0 = vrecpeq_f32(b_.neon_f32);
-	float32x4_t recip1 =
-		vmulq_f32(recip0, vrecpsq_f32(recip0, b_.neon_f32));
-	r_.neon_f32 = vmulq_f32(a_.neon_f32, recip1);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_div(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f32 = a_.f32 / b_.f32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = a_.f32[i] / b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_div_ps(a, b) simde_mm_div_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_div_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_div_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_div_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = a_.f32[0] / b_.f32[0];
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = a_.f32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_div_ss(a, b) simde_mm_div_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int16_t simde_mm_extract_pi16(simde__m64 a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 3)
-{
-	simde__m64_private a_ = simde__m64_to_private(a);
-	return a_.i16[imm8];
-}
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
-	!defined(HEDLEY_PGI_VERSION)
-#if HEDLEY_HAS_WARNING("-Wvector-conversion")
-/* https://bugs.llvm.org/show_bug.cgi?id=44589 */
-#define simde_mm_extract_pi16(a, imm8) \
-	(HEDLEY_DIAGNOSTIC_PUSH _Pragma( \
-		"clang diagnostic ignored \"-Wvector-conversion\"") \
-		 HEDLEY_STATIC_CAST(int16_t, _mm_extract_pi16((a), (imm8))) \
-			 HEDLEY_DIAGNOSTIC_POP)
-#else
-#define simde_mm_extract_pi16(a, imm8) \
-	HEDLEY_STATIC_CAST(int16_t, _mm_extract_pi16(a, imm8))
-#endif
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_extract_pi16(a, imm8) \
-	vget_lane_s16(simde__m64_to_private(a).neon_i16, imm8)
-#endif
-#define simde_m_pextrw(a, imm8) simde_mm_extract_pi16(a, imm8)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_extract_pi16(a, imm8) simde_mm_extract_pi16((a), (imm8))
-#endif
-
-enum {
-#if defined(SIMDE_X86_SSE_NATIVE)
-	SIMDE_MM_ROUND_NEAREST = _MM_ROUND_NEAREST,
-	SIMDE_MM_ROUND_DOWN = _MM_ROUND_DOWN,
-	SIMDE_MM_ROUND_UP = _MM_ROUND_UP,
-	SIMDE_MM_ROUND_TOWARD_ZERO = _MM_ROUND_TOWARD_ZERO
-#else
-	SIMDE_MM_ROUND_NEAREST
-#if defined(FE_TONEAREST)
-	= FE_TONEAREST
-#endif
-	,
-
-	SIMDE_MM_ROUND_DOWN
-#if defined(FE_DOWNWARD)
-	= FE_DOWNWARD
-#endif
-	,
-
-	SIMDE_MM_ROUND_UP
-#if defined(FE_UPWARD)
-	= FE_UPWARD
-#endif
-	,
-
-	SIMDE_MM_ROUND_TOWARD_ZERO
-#if defined(FE_TOWARDZERO)
-	= FE_TOWARDZERO
-#endif
-#endif
-};
-
-SIMDE_FUNCTION_ATTRIBUTES
-unsigned int SIMDE_MM_GET_ROUNDING_MODE(void)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _MM_GET_ROUNDING_MODE();
-#elif defined(SIMDE_HAVE_FENV_H)
-	return HEDLEY_STATIC_CAST(unsigned int, fegetround());
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _MM_GET_ROUNDING_MODE() SIMDE_MM_GET_ROUNDING_MODE()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void SIMDE_MM_SET_ROUNDING_MODE(unsigned int a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_MM_SET_ROUNDING_MODE(a);
-#elif defined(SIMDE_HAVE_FENV_H)
-	fesetround(HEDLEY_STATIC_CAST(int, a));
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _MM_SET_ROUNDING_MODE(a) SIMDE_MM_SET_ROUNDING_MODE(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_insert_pi16(simde__m64 a, int16_t i, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 3)
-{
-	simde__m64_private r_, a_ = simde__m64_to_private(a);
-
-	r_.i64[0] = a_.i64[0];
-	r_.i16[imm8] = i;
-
-	return simde__m64_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
-	!defined(__PGI)
-#if HEDLEY_HAS_WARNING("-Wvector-conversion")
-/* https://bugs.llvm.org/show_bug.cgi?id=44589 */
-#define ssimde_mm_insert_pi16(a, i, imm8) \
-	(HEDLEY_DIAGNOSTIC_PUSH _Pragma( \
-		"clang diagnostic ignored \"-Wvector-conversion\"")( \
-		_mm_insert_pi16((a), (i), (imm8))) HEDLEY_DIAGNOSTIC_POP)
-#else
-#define simde_mm_insert_pi16(a, i, imm8) _mm_insert_pi16(a, i, imm8)
-#endif
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_insert_pi16(a, i, imm8) \
-	simde__m64_from_neon_i16( \
-		vset_lane_s16((i), simde__m64_to_neon_i16(a), (imm8)))
-#endif
-#define simde_m_pinsrw(a, i, imm8) (simde_mm_insert_pi16(a, i, imm8))
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_insert_pi16(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
-#define _m_pinsrw(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128
-simde_mm_load_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_load_ps(mem_addr);
-#else
-	simde__m128_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vld1q_f32(mem_addr);
-#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
-	r_.altivec_f32 = vec_vsx_ld(0, mem_addr);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_ld(0, mem_addr);
-#else
-	r_ = *SIMDE_ALIGN_CAST(simde__m128_private const *, mem_addr);
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_load_ps(mem_addr) simde_mm_load_ps(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_load_ps1(simde_float32 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_load_ps1(mem_addr);
-#else
-	simde__m128_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vld1q_dup_f32(mem_addr);
-#else
-	r_ = simde__m128_to_private(simde_mm_set1_ps(*mem_addr));
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#define simde_mm_load1_ps(mem_addr) simde_mm_load_ps1(mem_addr)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_load_ps1(mem_addr) simde_mm_load_ps1(mem_addr)
-#define _mm_load1_ps(mem_addr) simde_mm_load_ps1(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_load_ss(simde_float32 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_load_ss(mem_addr);
-#else
-	simde__m128_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vsetq_lane_f32(*mem_addr, vdupq_n_f32(0), 0);
-#else
-	r_.f32[0] = *mem_addr;
-	r_.i32[1] = 0;
-	r_.i32[2] = 0;
-	r_.i32[3] = 0;
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_load_ss(mem_addr) simde_mm_load_ss(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_loadh_pi(simde__m128 a, simde__m64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_loadh_pi(a,
-			 HEDLEY_REINTERPRET_CAST(__m64 const *, mem_addr));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcombine_f32(
-		vget_low_f32(a_.neon_f32),
-		vld1_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr)));
-#else
-	simde__m64_private b_ =
-		*HEDLEY_REINTERPRET_CAST(simde__m64_private const *, mem_addr);
-	r_.f32[0] = a_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = b_.f32[0];
-	r_.f32[3] = b_.f32[1];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_loadh_pi(a, mem_addr) \
-	simde_mm_loadh_pi((a), (simde__m64 const *)(mem_addr))
-#endif
-
-/* The SSE documentation says that there are no alignment requirements
- for mem_addr. Unfortunately they used the __m64 type for the argument
- which is supposed to be 8-byte aligned, so some compilers (like clang
- with -Wcast-align) will generate a warning if you try to cast, say,
- a simde_float32* to a simde__m64* for this function.
-
- I think the choice of argument type is unfortunate, but I do think we
- need to stick to it here. If there is demand I can always add something
- like simde_x_mm_loadl_f32(simde__m128, simde_float32 mem_addr[2]) */
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_loadl_pi(simde__m128 a, simde__m64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_loadl_pi(a,
-			 HEDLEY_REINTERPRET_CAST(__m64 const *, mem_addr));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcombine_f32(
-		vld1_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr)),
-		vget_high_f32(a_.neon_f32));
-#else
-	simde__m64_private b_;
-	simde_memcpy(&b_, mem_addr, sizeof(b_));
-	r_.i32[0] = b_.i32[0];
-	r_.i32[1] = b_.i32[1];
-	r_.i32[2] = a_.i32[2];
-	r_.i32[3] = a_.i32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_loadl_pi(a, mem_addr) \
-	simde_mm_loadl_pi((a), (simde__m64 const *)(mem_addr))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128
-simde_mm_loadr_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_loadr_ps(mem_addr);
-#else
-	simde__m128_private r_,
-		v_ = simde__m128_to_private(simde_mm_load_ps(mem_addr));
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vrev64q_f32(v_.neon_f32);
-	r_.neon_f32 = vextq_f32(r_.neon_f32, r_.neon_f32, 2);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, v_.f32, v_.f32, 3, 2, 1, 0);
-#else
-	r_.f32[0] = v_.f32[3];
-	r_.f32[1] = v_.f32[2];
-	r_.f32[2] = v_.f32[1];
-	r_.f32[3] = v_.f32[0];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_loadr_ps(mem_addr) simde_mm_loadr_ps(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128
-simde_mm_loadu_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_loadu_ps(mem_addr);
-#else
-	simde__m128_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 =
-		vld1q_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr));
-#else
-	r_.f32[0] = mem_addr[0];
-	r_.f32[1] = mem_addr[1];
-	r_.f32[2] = mem_addr[2];
-	r_.f32[3] = mem_addr[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_loadu_ps(mem_addr) simde_mm_loadu_ps(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_maskmove_si64(simde__m64 a, simde__m64 mask, int8_t *mem_addr)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	_mm_maskmove_si64(a, mask, HEDLEY_REINTERPRET_CAST(char *, mem_addr));
-#else
-	simde__m64_private a_ = simde__m64_to_private(a),
-			 mask_ = simde__m64_to_private(mask);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(a_.i8) / sizeof(a_.i8[0])); i++)
-		if (mask_.i8[i] < 0)
-			mem_addr[i] = a_.i8[i];
-#endif
-}
-#define simde_m_maskmovq(a, mask, mem_addr) \
-	simde_mm_maskmove_si64(a, mask, mem_addr)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_maskmove_si64(a, mask, mem_addr) \
-	simde_mm_maskmove_si64( \
-		(a), (mask), \
-		SIMDE_CHECKED_REINTERPRET_CAST(int8_t *, char *, (mem_addr)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_max_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_max_pi16(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vmax_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? a_.i16[i] : b_.i16[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_max_pi16(a, b) simde_mm_max_pi16(a, b)
-#define _m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_max_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_max_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vmaxq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_max(a_.altivec_f32, b_.altivec_f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = (a_.f32[i] > b_.f32[i]) ? a_.f32[i] : b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_max_ps(a, b) simde_mm_max_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_max_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_max_pu8(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vmax_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] > b_.u8[i]) ? a_.u8[i] : b_.u8[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmaxub(a, b) simde_mm_max_pu8(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_max_pu8(a, b) simde_mm_max_pu8(a, b)
-#define _m_pmaxub(a, b) simde_mm_max_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_max_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_max_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_max_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = (a_.f32[0] > b_.f32[0]) ? a_.f32[0] : b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_max_ss(a, b) simde_mm_max_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_min_pi16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_min_pi16(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vmin_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? a_.i16[i] : b_.i16[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pminsw(a, b) simde_mm_min_pi16(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_min_pi16(a, b) simde_mm_min_pi16(a, b)
-#define _m_pminsw(a, b) simde_mm_min_pi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_min_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_min_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vminq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_min(a_.altivec_f32, b_.altivec_f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = (a_.f32[i] < b_.f32[i]) ? a_.f32[i] : b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_min_ps(a, b) simde_mm_min_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_min_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_min_pu8(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vmin_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] < b_.u8[i]) ? a_.u8[i] : b_.u8[i];
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pminub(a, b) simde_mm_min_pu8(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_min_pu8(a, b) simde_mm_min_pu8(a, b)
-#define _m_pminub(a, b) simde_mm_min_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_min_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_min_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_min_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = (a_.f32[0] < b_.f32[0]) ? a_.f32[0] : b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_min_ss(a, b) simde_mm_min_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_movehl_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_movehl_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 6, 7, 2, 3);
-#else
-	r_.f32[0] = b_.f32[2];
-	r_.f32[1] = b_.f32[3];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_movehl_ps(a, b) simde_mm_movehl_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_movelh_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_movelh_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 1, 4, 5);
-#else
-	r_.f32[0] = a_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = b_.f32[0];
-	r_.f32[3] = b_.f32[1];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_movelh_ps(a, b) simde_mm_movelh_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_movemask_pi8(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_movemask_pi8(a);
-#else
-	simde__m64_private a_ = simde__m64_to_private(a);
-	int r = 0;
-	const size_t nmemb = sizeof(a_.i8) / sizeof(a_.i8[0]);
-
-	SIMDE_VECTORIZE_REDUCTION(| : r)
-	for (size_t i = 0; i < nmemb; i++) {
-		r |= (a_.u8[nmemb - 1 - i] >> 7) << (nmemb - 1 - i);
-	}
-
-	return r;
-#endif
-}
-#define simde_m_pmovmskb(a, b) simde_mm_movemask_pi8(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_movemask_pi8(a) simde_mm_movemask_pi8(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_movemask_ps(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_movemask_ps(a);
-#else
-	int r = 0;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	/* TODO: check to see if NEON version is faster than the portable version */
-	static const uint32x4_t movemask = {1, 2, 4, 8};
-	static const uint32x4_t highbit = {0x80000000, 0x80000000, 0x80000000,
-					 0x80000000};
-	uint32x4_t t0 = a_.neon_u32;
-	uint32x4_t t1 = vtstq_u32(t0, highbit);
-	uint32x4_t t2 = vandq_u32(t1, movemask);
-	uint32x2_t t3 = vorr_u32(vget_low_u32(t2), vget_high_u32(t2));
-	r = vget_lane_u32(t3, 0) | vget_lane_u32(t3, 1);
-#else
-	SIMDE_VECTORIZE_REDUCTION(| : r)
-	for (size_t i = 0; i < sizeof(a_.u32) / sizeof(a_.u32[0]); i++) {
-		r |= (a_.u32[i] >> ((sizeof(a_.u32[i]) * CHAR_BIT) - 1)) << i;
-	}
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_movemask_ps(a) simde_mm_movemask_ps((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_mul_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_mul_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vmulq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_mul(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f32 = a_.f32 * b_.f32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = a_.f32[i] * b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_ps(a, b) simde_mm_mul_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_mul_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_mul_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_mul_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = a_.f32[0] * b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_ss(a, b) simde_mm_mul_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_mulhi_pu16(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_mulhi_pu16(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(
-			uint16_t, ((HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
-				 HEDLEY_STATIC_CAST(uint32_t, b_.u16[i])) >>
-				 UINT32_C(16)));
-	}
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_pmulhuw(a, b) simde_mm_mulhi_pu16(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_mulhi_pu16(a, b) simde_mm_mulhi_pu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_or_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_or_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vorrq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_or(a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f | b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] | b_.u32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_or_ps(a, b) simde_mm_or_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_prefetch(char const *p, int i)
-{
-	(void)p;
-	(void)i;
-}
-#if defined(SIMDE_X86_SSE_NATIVE)
-#define simde_mm_prefetch(p, i) _mm_prefetch(p, i)
-#endif
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_prefetch(p, i) simde_mm_prefetch(p, i)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_rcp_ps(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_rcp_ps(a);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	float32x4_t recip = vrecpeq_f32(a_.neon_f32);
-
-#if SIMDE_ACCURACY_PREFERENCE > 0
-	for (int i = 0; i < SIMDE_ACCURACY_PREFERENCE; ++i) {
-		recip = vmulq_f32(recip, vrecpsq_f32(recip, a_.neon_f32));
-	}
-#endif
-
-	r_.neon_f32 = recip;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_re(a_.altivec_f32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.f32 = 1.0f / a_.f32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = 1.0f / a_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_rcp_ps(a) simde_mm_rcp_ps((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_rcp_ss(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_rcp_ss(a);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_rcp_ps(a));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-	r_.f32[0] = 1.0f / a_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_rcp_ss(a) simde_mm_rcp_ss((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_rsqrt_ps(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_rsqrt_ps(a);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vrsqrteq_f32(a_.neon_f32);
-#elif defined(__STDC_IEC_559__)
-	/* https://basesandframes.files.wordpress.com/2020/04/even_faster_math_functions_green_2020.pdf
- Pages 100 - 103 */
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-#if SIMDE_ACCURACY_PREFERENCE <= 0
-		r_.i32[i] = INT32_C(0x5F37624F) - (a_.i32[i] >> 1);
-#else
-		simde_float32 x = a_.f32[i];
-		simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
-		int32_t ix;
-
-		simde_memcpy(&ix, &x, sizeof(ix));
-
-#if SIMDE_ACCURACY_PREFERENCE == 1
-		ix = INT32_C(0x5F375A82) - (ix >> 1);
-#else
-		ix = INT32_C(0x5F37599E) - (ix >> 1);
-#endif
-
-		simde_memcpy(&x, &ix, sizeof(x));
-
-#if SIMDE_ACCURACY_PREFERENCE >= 2
-		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
-#endif
-		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
-
-		r_.f32[i] = x;
-#endif
-	}
-#elif defined(simde_math_sqrtf)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = 1.0f / simde_math_sqrtf(a_.f32[i]);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_rsqrt_ps(a) simde_mm_rsqrt_ps((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_rsqrt_ss(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_rsqrt_ss(a);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_rsqrt_ps(a));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(__STDC_IEC_559__)
-	{
-#if SIMDE_ACCURACY_PREFERENCE <= 0
-		r_.i32[0] = INT32_C(0x5F37624F) - (a_.i32[0] >> 1);
-#else
-		simde_float32 x = a_.f32[0];
-		simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
-		int32_t ix;
-
-		simde_memcpy(&ix, &x, sizeof(ix));
-
-#if SIMDE_ACCURACY_PREFERENCE == 1
-		ix = INT32_C(0x5F375A82) - (ix >> 1);
-#else
-		ix = INT32_C(0x5F37599E) - (ix >> 1);
-#endif
-
-		simde_memcpy(&x, &ix, sizeof(x));
-
-#if SIMDE_ACCURACY_PREFERENCE >= 2
-		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
-#endif
-		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
-
-		r_.f32[0] = x;
-#endif
-	}
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-#elif defined(simde_math_sqrtf)
-	r_.f32[0] = 1.0f / simde_math_sqrtf(a_.f32[0]);
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_rsqrt_ss(a) simde_mm_rsqrt_ss((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sad_pu8(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sad_pu8(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-	uint16_t sum = 0;
-
-#if defined(SIMDE_HAVE_STDLIB_H)
-	SIMDE_VECTORIZE_REDUCTION(+ : sum)
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		sum += HEDLEY_STATIC_CAST(uint8_t, abs(a_.u8[i] - b_.u8[i]));
-	}
-
-	r_.i16[0] = HEDLEY_STATIC_CAST(int16_t, sum);
-	r_.i16[1] = 0;
-	r_.i16[2] = 0;
-	r_.i16[3] = 0;
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#define simde_m_psadbw(a, b) simde_mm_sad_pu8(a, b)
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sad_pu8(a, b) simde_mm_sad_pu8(a, b)
-#define _m_psadbw(a, b) simde_mm_sad_pu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_set_ss(simde_float32 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_set_ss(a);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vsetq_lane_f32(a, vdupq_n_f32(SIMDE_FLOAT32_C(0.0)), 0);
-#else
-	return simde_mm_set_ps(SIMDE_FLOAT32_C(0.0), SIMDE_FLOAT32_C(0.0),
-			 SIMDE_FLOAT32_C(0.0), a);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_set_ss(a) simde_mm_set_ss(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_setr_ps(simde_float32 e3, simde_float32 e2,
-			 simde_float32 e1, simde_float32 e0)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_setr_ps(e3, e2, e1, e0);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN(16) simde_float32 data[4] = {e3, e2, e1, e0};
-	return vld1q_f32(data);
-#else
-	return simde_mm_set_ps(e0, e1, e2, e3);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_ps(e3, e2, e1, e0) simde_mm_setr_ps(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_setzero_ps(void)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_setzero_ps();
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vdupq_n_f32(SIMDE_FLOAT32_C(0.0));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	return vec_splats((float)0);
-#else
-	simde__m128 r;
-	simde_memset(&r, 0, sizeof(r));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_setzero_ps() simde_mm_setzero_ps()
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_undefined_ps(void)
-{
-	simde__m128_private r_;
-
-#if defined(SIMDE_HAVE_UNDEFINED128)
-	r_.n = _mm_undefined_ps();
-#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-	r_ = simde__m128_to_private(simde_mm_setzero_ps());
-#endif
-
-	return simde__m128_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_undefined_ps() simde_mm_undefined_ps()
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_POP
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_x_mm_setone_ps(void)
-{
-	simde__m128 t = simde_mm_setzero_ps();
-	return simde_mm_cmpeq_ps(t, t);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_sfence(void)
-{
-	/* TODO: Use Hedley. */
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_sfence();
-#elif defined(__GNUC__) && \
-	((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 7))
-	__atomic_thread_fence(__ATOMIC_SEQ_CST);
-#elif !defined(__INTEL_COMPILER) && defined(__STDC_VERSION__) && \
-	(__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_ATOMICS__)
-#if defined(__GNUC__) && (__GNUC__ == 4) && (__GNUC_MINOR__ < 9)
-	__atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
-	atomic_thread_fence(memory_order_seq_cst);
-#endif
-#elif defined(_MSC_VER)
-	MemoryBarrier();
-#elif HEDLEY_HAS_EXTENSION(c_atomic)
-	__c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#elif defined(__GNUC__) && \
-	((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 1))
-	__sync_synchronize();
-#elif defined(_OPENMP)
-#pragma omp critical(simde_mm_sfence_)
-	{
-	}
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sfence() simde_mm_sfence()
-#endif
-
-#define SIMDE_MM_SHUFFLE(z, y, x, w) \
-	(((z) << 6) | ((y) << 4) | ((x) << 2) | (w))
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _MM_SHUFFLE(z, y, x, w) SIMDE_MM_SHUFFLE(z, y, x, w)
-#endif
-
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
-	!defined(__PGI)
-#define simde_mm_shuffle_pi16(a, imm8) _mm_shuffle_pi16(a, imm8)
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shuffle_pi16(a, imm8) \
-	(__extension__({ \
-		const simde__m64_private simde__tmp_a_ = \
-			simde__m64_to_private(a); \
-		simde__m64_from_private((simde__m64_private){ \
-			.i16 = SIMDE_SHUFFLE_VECTOR_( \
-				16, 8, (simde__tmp_a_).i16, \
-				(simde__tmp_a_).i16, (((imm8)) & 3), \
-				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3), \
-				(((imm8) >> 6) & 3))}); \
-	}))
-#else
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_shuffle_pi16(simde__m64 a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m64_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-	for (size_t i = 0; i < sizeof(r_.i16) / sizeof(r_.i16[0]); i++) {
-		r_.i16[i] = a_.i16[(imm8 >> (i * 2)) & 3];
-	}
-
-	HEDLEY_DIAGNOSTIC_PUSH
-#if HEDLEY_HAS_WARNING("-Wconditional-uninitialized")
-#pragma clang diagnostic ignored "-Wconditional-uninitialized"
-#endif
-	return simde__m64_from_private(r_);
-	HEDLEY_DIAGNOSTIC_POP
-}
-#endif
-#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
-#define simde_m_pshufw(a, imm8) _m_pshufw(a, imm8)
-#else
-#define simde_m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
-#endif
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_shuffle_pi16(a, imm8) simde_mm_shuffle_pi16(a, imm8)
-#define _m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
-#endif
-
-#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
-#define simde_mm_shuffle_ps(a, b, imm8) _mm_shuffle_ps(a, b, imm8)
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shuffle_ps(a, b, imm8) \
-	(__extension__({ \
-		simde__m128_from_private((simde__m128_private){ \
-			.f32 = SIMDE_SHUFFLE_VECTOR_( \
-				32, 16, simde__m128_to_private(a).f32, \
-				simde__m128_to_private(b).f32, (((imm8)) & 3), \
-				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3) + 4, \
-				(((imm8) >> 6) & 3) + 4)}); \
-	}))
-#else
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_shuffle_ps(simde__m128 a, simde__m128 b, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = a_.f32[(imm8 >> 0) & 3];
-	r_.f32[1] = a_.f32[(imm8 >> 2) & 3];
-	r_.f32[2] = b_.f32[(imm8 >> 4) & 3];
-	r_.f32[3] = b_.f32[(imm8 >> 6) & 3];
-
-	return simde__m128_from_private(r_);
-}
-#endif
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_shuffle_ps(a, b, imm8) simde_mm_shuffle_ps((a), (b), imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_sqrt_ps(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_sqrt_ps(a);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	float32x4_t recipsq = vrsqrteq_f32(a_.neon_f32);
-	float32x4_t sq = vrecpeq_f32(recipsq);
-	/* ??? use step versions of both sqrt and recip for better accuracy? */
-	r_.neon_f32 = sq;
-#elif defined(simde_math_sqrt)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < sizeof(r_.f32) / sizeof(r_.f32[0]); i++) {
-		r_.f32[i] = simde_math_sqrtf(a_.f32[i]);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sqrt_ps(a) simde_mm_sqrt_ps((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_sqrt_ss(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_sqrt_ss(a);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_sqrt_ps(a));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-
-#if defined(simde_math_sqrtf)
-	r_.f32[0] = simde_math_sqrtf(a_.f32[0]);
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sqrt_ss(a) simde_mm_sqrt_ss((a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_ps(simde_float32 mem_addr[4], simde__m128 a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_store_ps(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_f32(mem_addr, a_.neon_f32);
-#elif defined(SIMDE_POWER_ALTIVE_P7_NATIVE)
-	vec_vsx_st(a_.altivec_32, 0, mem_addr);
-#elif defined(SIMDE_POWER_ALTIVE_P5_NATIVE)
-	vec_st(a_.altivec_32, 0, mem_addr);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	wasm_v128_store(mem_addr, a_.wasm_v128);
-#else
-	SIMDE_VECTORIZE_ALIGNED(mem_addr : 16)
-	for (size_t i = 0; i < sizeof(a_.f32) / sizeof(a_.f32[0]); i++) {
-		mem_addr[i] = a_.f32[i];
-	}
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_store_ps(mem_addr, a) \
-	simde_mm_store_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_ps1(simde_float32 mem_addr[4], simde__m128 a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_store_ps1(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-	SIMDE_VECTORIZE_ALIGNED(mem_addr : 16)
-	for (size_t i = 0; i < sizeof(a_.f32) / sizeof(a_.f32[0]); i++) {
-		mem_addr[i] = a_.f32[0];
-	}
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_store_ps1(mem_addr, a) \
-	simde_mm_store_ps1(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_ss(simde_float32 *mem_addr, simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_store_ss(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_lane_f32(mem_addr, a_.neon_f32, 0);
-#else
-	*mem_addr = a_.f32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_store_ss(mem_addr, a) \
-	simde_mm_store_ss(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store1_ps(simde_float32 mem_addr[4], simde__m128 a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_store1_ps(mem_addr, a);
-#else
-	simde_mm_store_ps1(mem_addr, a);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_store1_ps(mem_addr, a) \
-	simde_mm_store1_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storeh_pi(simde__m64 *mem_addr, simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_storeh_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
-#else
-	simde__m64_private *dest_ =
-		HEDLEY_REINTERPRET_CAST(simde__m64_private *, mem_addr);
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-	dest_->f32[0] = a_.f32[2];
-	dest_->f32[1] = a_.f32[3];
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_storeh_pi(mem_addr, a) simde_mm_storeh_pi(mem_addr, (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storel_pi(simde__m64 *mem_addr, simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_storel_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
-#else
-	simde__m64_private *dest_ =
-		HEDLEY_REINTERPRET_CAST(simde__m64_private *, mem_addr);
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-	dest_->f32[0] = a_.f32[0];
-	dest_->f32[1] = a_.f32[1];
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_storel_pi(mem_addr, a) simde_mm_storel_pi(mem_addr, (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storer_ps(simde_float32 mem_addr[4], simde__m128 a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_storer_ps(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	a_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 3, 2, 1, 0);
-	simde_mm_store_ps(mem_addr, simde__m128_from_private(a_));
-#else
-	SIMDE_VECTORIZE_ALIGNED(mem_addr : 16)
-	for (size_t i = 0; i < sizeof(a_.f32) / sizeof(a_.f32[0]); i++) {
-		mem_addr[i] =
-			a_.f32[((sizeof(a_.f32) / sizeof(a_.f32[0])) - 1) - i];
-	}
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_storer_ps(mem_addr, a) \
-	simde_mm_storer_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storeu_ps(simde_float32 mem_addr[4], simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_storeu_ps(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_f32(mem_addr, a_.neon_f32);
-#else
-	simde_memcpy(mem_addr, &a_, sizeof(a_));
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_storeu_ps(mem_addr, a) \
-	simde_mm_storeu_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_sub_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_sub_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vsubq_f32(a_.neon_f32, b_.neon_f32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f32x4_sub(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f32 = a_.f32 - b_.f32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = a_.f32[i] - b_.f32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_ps(a, b) simde_mm_sub_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_sub_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_sub_ss(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_ss(a, simde_mm_sub_ps(a, b));
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-	r_.f32[0] = a_.f32[0] - b_.f32[0];
-	r_.f32[1] = a_.f32[1];
-	r_.f32[2] = a_.f32[2];
-	r_.f32[3] = a_.f32[3];
-
-	return simde__m128_from_private(r_);
-#endif
-}
-
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_ss(a, b) simde_mm_sub_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomieq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomieq_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] == b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] == b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomieq_ss(a, b) simde_mm_ucomieq_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomige_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomige_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] >= b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] >= b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomige_ss(a, b) simde_mm_ucomige_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomigt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomigt_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] > b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] > b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomigt_ss(a, b) simde_mm_ucomigt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomile_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomile_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] <= b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] <= b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomile_ss(a, b) simde_mm_ucomile_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomilt_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomilt_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] < b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] < b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomilt_ss(a, b) simde_mm_ucomilt_ss((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomineq_ss(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_ucomineq_ss(a, b);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a),
-			 b_ = simde__m128_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f32[0] != b_.f32[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f32[0] != b_.f32[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomineq_ss(a, b) simde_mm_ucomineq_ss((a), (b))
-#endif
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-#if defined(__has_builtin)
-#if __has_builtin(__builtin_ia32_undef128)
-#define SIMDE_HAVE_UNDEFINED128
-#endif
-#elif !defined(__PGI) && !defined(SIMDE_BUG_GCC_REV_208793) && \
-	!defined(_MSC_VER)
-#define SIMDE_HAVE_UNDEFINED128
-#endif
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_unpackhi_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_unpackhi_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	float32x2_t a1 = vget_high_f32(a_.neon_f32);
-	float32x2_t b1 = vget_high_f32(b_.neon_f32);
-	float32x2x2_t result = vzip_f32(a1, b1);
-	r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 2, 6, 3, 7);
-#else
-	r_.f32[0] = a_.f32[2];
-	r_.f32[1] = b_.f32[2];
-	r_.f32[2] = a_.f32[3];
-	r_.f32[3] = b_.f32[3];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_ps(a, b) simde_mm_unpackhi_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_unpacklo_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_unpacklo_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 4, 1, 5);
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	float32x2_t a1 = vget_low_f32(a_.neon_f32);
-	float32x2_t b1 = vget_low_f32(b_.neon_f32);
-	float32x2x2_t result = vzip_f32(a1, b1);
-	r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
-#else
-	r_.f32[0] = a_.f32[0];
-	r_.f32[1] = b_.f32[0];
-	r_.f32[2] = a_.f32[1];
-	r_.f32[3] = b_.f32[1];
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_ps(a, b) simde_mm_unpacklo_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_xor_ps(simde__m128 a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_xor_ps(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a),
-				b_ = simde__m128_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = veorq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_xor(a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f ^ b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] ^ b_.u32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_xor_ps(a, b) simde_mm_xor_ps((a), (b))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_pi(simde__m64 *mem_addr, simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	_mm_stream_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
-#else
-	simde__m64_private *dest = HEDLEY_REINTERPRET_CAST(simde__m64_private *,
-							 mem_addr),
-			 a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	dest->i64[0] = vget_lane_s64(a_.neon_i64, 0);
-#else
-	dest->i64[0] = a_.i64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_pi(mem_addr, a) simde_mm_stream_pi(mem_addr, (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_ps(simde_float32 mem_addr[4], simde__m128 a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_stream_ps(mem_addr, a);
-#else
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_f32(SIMDE_ASSUME_ALIGNED(16, mem_addr), a_.neon_f32);
-#else
-	simde_memcpy(SIMDE_ASSUME_ALIGNED(16, mem_addr), &a_, sizeof(a_));
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_ps(mem_addr, a) \
-	simde_mm_stream_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 float *, simde_float32 *, mem_addr), \
-			 (a))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-uint32_t simde_mm_getcsr(void)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	return _mm_getcsr();
-#else
-	uint32_t r = 0;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	int rounding_mode = fegetround();
-
-	switch (rounding_mode) {
-#if defined(FE_TONEAREST)
-	case FE_TONEAREST:
-		break;
-#endif
-#if defined(FE_UPWARD)
-	case FE_UPWARD:
-		r |= 2 << 13;
-		break;
-#endif
-#if defined(FE_DOWNWARD)
-	case FE_DOWNWARD:
-		r |= 1 << 13;
-		break;
-#endif
-#if defined(FE_TOWARDZERO)
-	case FE_TOWARDZERO:
-		r = 3 << 13;
-		break;
-#endif
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_getcsr() simde_mm_getcsr()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_setcsr(uint32_t a)
-{
-#if defined(SIMDE_X86_SSE_NATIVE)
-	_mm_setcsr(a);
-#else
-	switch ((a >> 13) & 3) {
-#if defined(FE_TONEAREST)
-	case 0:
-		fesetround(FE_TONEAREST);
-#endif
-#if defined(FE_DOWNWARD)
-		break;
-	case 1:
-		fesetround(FE_DOWNWARD);
-#endif
-#if defined(FE_UPWARD)
-		break;
-	case 2:
-		fesetround(FE_UPWARD);
-#endif
-#if defined(FE_TOWARDZERO)
-		break;
-	case 3:
-		fesetround(FE_TOWARDZERO);
-		break;
-#endif
-	}
-#endif
-}
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _mm_setcsr(a) simde_mm_setcsr(a)
-#endif
-
-#define SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
-	do { \
-		simde__m128 tmp3, tmp2, tmp1, tmp0; \
-		tmp0 = simde_mm_unpacklo_ps((row0), (row1)); \
-		tmp2 = simde_mm_unpacklo_ps((row2), (row3)); \
-		tmp1 = simde_mm_unpackhi_ps((row0), (row1)); \
-		tmp3 = simde_mm_unpackhi_ps((row2), (row3)); \
-		row0 = simde_mm_movelh_ps(tmp0, tmp2); \
-		row1 = simde_mm_movehl_ps(tmp2, tmp0); \
-		row2 = simde_mm_movelh_ps(tmp1, tmp3); \
-		row3 = simde_mm_movehl_ps(tmp3, tmp1); \
-	} while (0)
-
-#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
-#define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
-	SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3)
-#endif
-
-#if defined(_MM_EXCEPT_INVALID)
-#define SIMDE_MM_EXCEPT_INVALID _MM_EXCEPT_INVALID
-#else
-#define SIMDE_MM_EXCEPT_INVALID (0x0001)
-#endif
-#if defined(_MM_EXCEPT_DENORM)
-#define SIMDE_MM_EXCEPT_DENORM _MM_EXCEPT_DENORM
-#else
-#define SIMDE_MM_EXCEPT_DENORM (0x0002)
-#endif
-#if defined(_MM_EXCEPT_DIV_ZERO)
-#define SIMDE_MM_EXCEPT_DIV_ZERO _MM_EXCEPT_DIV_ZERO
-#else
-#define SIMDE_MM_EXCEPT_DIV_ZERO (0x0004)
-#endif
-#if defined(_MM_EXCEPT_OVERFLOW)
-#define SIMDE_MM_EXCEPT_OVERFLOW _MM_EXCEPT_OVERFLOW
-#else
-#define SIMDE_MM_EXCEPT_OVERFLOW (0x0008)
-#endif
-#if defined(_MM_EXCEPT_UNDERFLOW)
-#define SIMDE_MM_EXCEPT_UNDERFLOW _MM_EXCEPT_UNDERFLOW
-#else
-#define SIMDE_MM_EXCEPT_UNDERFLOW (0x0010)
-#endif
-#if defined(_MM_EXCEPT_INEXACT)
-#define SIMDE_MM_EXCEPT_INEXACT _MM_EXCEPT_INEXACT
-#else
-#define SIMDE_MM_EXCEPT_INEXACT (0x0020)
-#endif
-#if defined(_MM_EXCEPT_MASK)
-#define SIMDE_MM_EXCEPT_MASK _MM_EXCEPT_MASK
-#else
-#define SIMDE_MM_EXCEPT_MASK \
-	(SIMDE_MM_EXCEPT_INVALID | SIMDE_MM_EXCEPT_DENORM | \
-	 SIMDE_MM_EXCEPT_DIV_ZERO | SIMDE_MM_EXCEPT_OVERFLOW | \
-	 SIMDE_MM_EXCEPT_UNDERFLOW | SIMDE_MM_EXCEPT_INEXACT)
-#endif
-
-#if defined(_MM_MASK_INVALID)
-#define SIMDE_MM_MASK_INVALID _MM_MASK_INVALID
-#else
-#define SIMDE_MM_MASK_INVALID (0x0080)
-#endif
-#if defined(_MM_MASK_DENORM)
-#define SIMDE_MM_MASK_DENORM _MM_MASK_DENORM
-#else
-#define SIMDE_MM_MASK_DENORM (0x0100)
-#endif
-#if defined(_MM_MASK_DIV_ZERO)
-#define SIMDE_MM_MASK_DIV_ZERO _MM_MASK_DIV_ZERO
-#else
-#define SIMDE_MM_MASK_DIV_ZERO (0x0200)
-#endif
-#if defined(_MM_MASK_OVERFLOW)
-#define SIMDE_MM_MASK_OVERFLOW _MM_MASK_OVERFLOW
-#else
-#define SIMDE_MM_MASK_OVERFLOW (0x0400)
-#endif
-#if defined(_MM_MASK_UNDERFLOW)
-#define SIMDE_MM_MASK_UNDERFLOW _MM_MASK_UNDERFLOW
-#else
-#define SIMDE_MM_MASK_UNDERFLOW (0x0800)
-#endif
-#if defined(_MM_MASK_INEXACT)
-#define SIMDE_MM_MASK_INEXACT _MM_MASK_INEXACT
-#else
-#define SIMDE_MM_MASK_INEXACT (0x1000)
-#endif
-#if defined(_MM_MASK_MASK)
-#define SIMDE_MM_MASK_MASK _MM_MASK_MASK
-#else
-#define SIMDE_MM_MASK_MASK \
-	(SIMDE_MM_MASK_INVALID | SIMDE_MM_MASK_DENORM | \
-	 SIMDE_MM_MASK_DIV_ZERO | SIMDE_MM_MASK_OVERFLOW | \
-	 SIMDE_MM_MASK_UNDERFLOW | SIMDE_MM_MASK_INEXACT)
-#endif
-
-#if defined(_MM_FLUSH_ZERO_MASK)
-#define SIMDE_MM_FLUSH_ZERO_MASK _MM_FLUSH_ZERO_MASK
-#else
-#define SIMDE_MM_FLUSH_ZERO_MASK (0x8000)
-#endif
-#if defined(_MM_FLUSH_ZERO_ON)
-#define SIMDE_MM_FLUSH_ZERO_ON _MM_FLUSH_ZERO_ON
-#else
-#define SIMDE_MM_FLUSH_ZERO_ON (0x8000)
-#endif
-#if defined(_MM_FLUSH_ZERO_OFF)
-#define SIMDE_MM_FLUSH_ZERO_OFF _MM_FLUSH_ZERO_OFF
-#else
-#define SIMDE_MM_FLUSH_ZERO_OFF (0x0000)
-#endif
-
-SIMDE_END_DECLS_
-
-HEDLEY_DIAGNOSTIC_POP
-
-#endif /* !defined(SIMDE_X86_SSE_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/sse2.h Deleted

@@ -1,6214 +0,0 @@
-/* SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person
- * obtaining a copy of this software and associated documentation
- * files (the "Software"), to deal in the Software without
- * restriction, including without limitation the rights to use, copy,
- * modify, merge, publish, distribute, sublicense, and/or sell copies
- * of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be
- * included in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- *
- * Copyright:
- * 2017-2020 Evan Nemerson <evan@nemerson.com>
- * 2015-2017 John W. Ratcliff <jratcliffscarab@gmail.com>
- * 2015 Brandon Rowlett <browlett@nvidia.com>
- * 2015 Ken Fast <kfast@gdeb.com>
- * 2017 Hasindu Gamaarachchi <hasindu@unsw.edu.au>
- * 2018 Jeff Daily <jeff.daily@amd.com>
- */
-
-#if !defined(SIMDE_X86_SSE2_H)
-#define SIMDE_X86_SSE2_H
-
-#include "sse.h"
-
-#if !defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES
-#endif
-
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
-SIMDE_BEGIN_DECLS_
-
-typedef union {
-#if defined(SIMDE_VECTOR_SUBSCRIPT)
-	SIMDE_ALIGN(16) int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#if defined(SIMDE_HAVE_INT128_)
-	SIMDE_ALIGN(16) simde_int128 i128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) simde_uint128 u128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#endif
-	SIMDE_ALIGN(16) simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-
-	SIMDE_ALIGN(16) int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-	SIMDE_ALIGN(16) int8_t i8[16];
-	SIMDE_ALIGN(16) int16_t i16[8];
-	SIMDE_ALIGN(16) int32_t i32[4];
-	SIMDE_ALIGN(16) int64_t i64[2];
-	SIMDE_ALIGN(16) uint8_t u8[16];
-	SIMDE_ALIGN(16) uint16_t u16[8];
-	SIMDE_ALIGN(16) uint32_t u32[4];
-	SIMDE_ALIGN(16) uint64_t u64[2];
-#if defined(SIMDE_HAVE_INT128_)
-	SIMDE_ALIGN(16) simde_int128 i128[1];
-	SIMDE_ALIGN(16) simde_uint128 u128[1];
-#endif
-	SIMDE_ALIGN(16) simde_float32 f32[4];
-	SIMDE_ALIGN(16) simde_float64 f64[2];
-
-	SIMDE_ALIGN(16) int_fast32_t i32f[16 / sizeof(int_fast32_t)];
-	SIMDE_ALIGN(16) uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
-#endif
-
-	SIMDE_ALIGN(16) simde__m64_private m64_private[2];
-	SIMDE_ALIGN(16) simde__m64 m64[2];
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	SIMDE_ALIGN(16) __m128i n;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN(16) int8x16_t neon_i8;
-	SIMDE_ALIGN(16) int16x8_t neon_i16;
-	SIMDE_ALIGN(16) int32x4_t neon_i32;
-	SIMDE_ALIGN(16) int64x2_t neon_i64;
-	SIMDE_ALIGN(16) uint8x16_t neon_u8;
-	SIMDE_ALIGN(16) uint16x8_t neon_u16;
-	SIMDE_ALIGN(16) uint32x4_t neon_u32;
-	SIMDE_ALIGN(16) uint64x2_t neon_u64;
-	SIMDE_ALIGN(16) float32x4_t neon_f32;
-#if defined(SIMDE_ARCH_AARCH64)
-	SIMDE_ALIGN(16) float64x2_t neon_f64;
-#endif
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	SIMDE_ALIGN(16) v128_t wasm_v128;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
-#if defined(__UINT_FAST32_TYPE__)
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f;
-#else
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f;
-#endif
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
-#if defined(__UINT_FAST32_TYPE__)
-	SIMDE_ALIGN(16) vector __UINT_FAST32_TYPE__ altivec_u32f;
-#else
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f;
-#endif
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
-#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
-#endif
-#endif
-} simde__m128i_private;
-
-typedef union {
-#if defined(SIMDE_VECTOR_SUBSCRIPT)
-	SIMDE_ALIGN(16) int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-	SIMDE_ALIGN(16) uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-	SIMDE_ALIGN(16) int8_t i8[16];
-	SIMDE_ALIGN(16) int16_t i16[8];
-	SIMDE_ALIGN(16) int32_t i32[4];
-	SIMDE_ALIGN(16) int64_t i64[2];
-	SIMDE_ALIGN(16) uint8_t u8[16];
-	SIMDE_ALIGN(16) uint16_t u16[8];
-	SIMDE_ALIGN(16) uint32_t u32[4];
-	SIMDE_ALIGN(16) uint64_t u64[2];
-	SIMDE_ALIGN(16) simde_float32 f32[4];
-	SIMDE_ALIGN(16) simde_float64 f64[2];
-	SIMDE_ALIGN(16) int_fast32_t i32f[16 / sizeof(int_fast32_t)];
-	SIMDE_ALIGN(16) uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
-#endif
-
-	SIMDE_ALIGN(16) simde__m64_private m64_private[2];
-	SIMDE_ALIGN(16) simde__m64 m64[2];
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	SIMDE_ALIGN(16) __m128d n;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN(16) int8x16_t neon_i8;
-	SIMDE_ALIGN(16) int16x8_t neon_i16;
-	SIMDE_ALIGN(16) int32x4_t neon_i32;
-	SIMDE_ALIGN(16) int64x2_t neon_i64;
-	SIMDE_ALIGN(16) uint8x16_t neon_u8;
-	SIMDE_ALIGN(16) uint16x8_t neon_u16;
-	SIMDE_ALIGN(16) uint32x4_t neon_u32;
-	SIMDE_ALIGN(16) uint64x2_t neon_u64;
-	SIMDE_ALIGN(16) float32x4_t neon_f32;
-#if defined(SIMDE_ARCH_AARCH64)
-	SIMDE_ALIGN(16) float64x2_t neon_f64;
-#endif
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	SIMDE_ALIGN(16) v128_t wasm_v128;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
-#if defined(__INT_FAST32_TYPE__)
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f;
-#else
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f;
-#endif
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
-#if defined(__UINT_FAST32_TYPE__)
-	SIMDE_ALIGN(16) vector __UINT_FAST32_TYPE__ altivec_u32f;
-#else
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f;
-#endif
-	SIMDE_ALIGN(16)
-	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
-#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
-	SIMDE_ALIGN(16) SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
-#endif
-#endif
-} simde__m128d_private;
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-typedef __m128i simde__m128i;
-typedef __m128d simde__m128d;
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-typedef int64x2_t simde__m128i;
-#if defined(SIMDE_ARCH_AARCH64)
-typedef float64x2_t simde__m128d;
-#elif defined(SIMDE_VECTOR_SUBSCRIPT)
-typedef simde_float64 simde__m128d SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-typedef simde__m128d_private simde__m128d;
-#endif
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-typedef v128_t simde__m128i;
-typedef v128_t simde__m128d;
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128i;
-typedef SIMDE_POWER_ALTIVEC_VECTOR(double) simde__m128d;
-#elif defined(SIMDE_VECTOR_SUBSCRIPT)
-typedef int_fast32_t simde__m128i SIMDE_ALIGN(16)
-	SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-typedef simde_float64 simde__m128d SIMDE_ALIGN(16)
-	SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
-#else
-typedef simde__m128i_private simde__m128i;
-typedef simde__m128d_private simde__m128d;
-#endif
-
-#if !defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ENABLE_NATIVE_ALIASES)
-#define SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES
-typedef simde__m128i __m128i;
-typedef simde__m128d __m128d;
-#endif
-
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i), "simde__m128i size incorrect");
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i_private),
-		 "simde__m128i_private size incorrect");
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d), "simde__m128d size incorrect");
-HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d_private),
-		 "simde__m128d_private size incorrect");
-#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i) == 16,
-		 "simde__m128i is not 16-byte aligned");
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i_private) == 16,
-		 "simde__m128i_private is not 16-byte aligned");
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d) == 16,
-		 "simde__m128d is not 16-byte aligned");
-HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d_private) == 16,
-		 "simde__m128d_private is not 16-byte aligned");
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde__m128i_from_private(simde__m128i_private v)
-{
-	simde__m128i r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i_private simde__m128i_to_private(simde__m128i v)
-{
-	simde__m128i_private r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde__m128d_from_private(simde__m128d_private v)
-{
-	simde__m128d r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d_private simde__m128d_to_private(simde__m128d v)
-{
-	simde__m128d_private r;
-	simde_memcpy(&r, &v, sizeof(r));
-	return r;
-}
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int8x16_t, neon, i8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int16x8_t, neon, i16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int32x4_t, neon, i32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int64x2_t, neon, i64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint8x16_t, neon, u8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint16x8_t, neon, u16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint32x4_t, neon, u32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint64x2_t, neon, u64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float32x4_t, neon, f32)
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float64x2_t, neon, f64)
-#endif
-#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int8x16_t, neon, i8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int16x8_t, neon, i16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int32x4_t, neon, i32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int64x2_t, neon, i64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint8x16_t, neon, u8)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint16x8_t, neon, u16)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint32x4_t, neon, u32)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint64x2_t, neon, u64)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float32x4_t, neon, f32)
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float64x2_t, neon, f64)
-#endif
-#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_add_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vaddq_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i8 = vec_add(a_.altivec_i8, b_.altivec_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = a_.i8 + b_.i8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = a_.i8[i] + b_.i8[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_epi8(a, b) simde_mm_add_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_add_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vaddq_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = vec_add(a_.altivec_i16, b_.altivec_i16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = a_.i16 + b_.i16;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] + b_.i16[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_epi16(a, b) simde_mm_add_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_add_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vaddq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_add(a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = a_.i32 + b_.i32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] + b_.i32[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_epi32(a, b) simde_mm_add_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_add_epi64(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_epi64(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vaddq_s64(a_.neon_i64, b_.neon_i64);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i64 = vec_add(a_.altivec_i64, b_.altivec_i64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 + b_.i64;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a_.i64[i] + b_.i64[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_epi64(a, b) simde_mm_add_epi64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_add_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f64 = vaddq_f64(a_.neon_f64, b_.neon_f64);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_add(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 = vec_add(a_.altivec_f64, b_.altivec_f64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f64 = a_.f64 + b_.f64;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = a_.f64[i] + b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_pd(a, b) simde_mm_add_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_move_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_move_sd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f64 =
-		vsetq_lane_f64(vgetq_lane_f64(b_.neon_f64, 0), a_.neon_f64, 0);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
-	m = {16, 17, 18, 19, 20, 21, 22, 23, 8, 9, 10, 11, 12, 13, 14, 15};
-	r_.altivec_f64 = vec_perm(a_.altivec_f64, b_.altivec_f64, m);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 2, 1);
-#else
-	r_.f64[0] = b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_move_sd(a, b) simde_mm_move_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_add_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_add_sd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = a_.f64[0] + b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-#if defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_add_pd(a, b));
-#else
-	r_.f64[0] = a_.f64[0] + b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_sd(a, b) simde_mm_add_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_add_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_add_si64(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vadd_s64(a_.neon_i64, b_.neon_i64);
-#else
-	r_.i64[0] = a_.i64[0] + b_.i64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_add_si64(a, b) simde_mm_add_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_adds_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_adds_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vqaddq_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i8 = vec_adds(a_.altivec_i8, b_.altivec_i8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		const int32_t tmp = HEDLEY_STATIC_CAST(int16_t, a_.i8[i]) +
-				 HEDLEY_STATIC_CAST(int16_t, b_.i8[i]);
-		r_.i8[i] = HEDLEY_STATIC_CAST(
-			int8_t,
-			((tmp < INT8_MAX) ? ((tmp > INT8_MIN) ? tmp : INT8_MIN)
-					 : INT8_MAX));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_epi8(a, b) simde_mm_adds_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_adds_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_adds_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vqaddq_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = vec_adds(a_.altivec_i16, b_.altivec_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		const int32_t tmp = HEDLEY_STATIC_CAST(int32_t, a_.i16[i]) +
-				 HEDLEY_STATIC_CAST(int32_t, b_.i16[i]);
-		r_.i16[i] = HEDLEY_STATIC_CAST(
-			int16_t,
-			((tmp < INT16_MAX)
-				 ? ((tmp > INT16_MIN) ? tmp : INT16_MIN)
-				 : INT16_MAX));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_epi16(a, b) simde_mm_adds_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_adds_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_adds_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vqaddq_u8(a_.neon_u8, b_.neon_u8);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u8 = vec_adds(a_.altivec_u8, b_.altivec_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = ((UINT8_MAX - a_.u8[i]) > b_.u8[i])
-				 ? (a_.u8[i] + b_.u8[i])
-				 : UINT8_MAX;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_epu8(a, b) simde_mm_adds_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_adds_epu16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_adds_epu16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vqaddq_u16(a_.neon_u16, b_.neon_u16);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u16 = vec_adds(a_.altivec_u16, b_.altivec_u16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = ((UINT16_MAX - a_.u16[i]) > b_.u16[i])
-				 ? (a_.u16[i] + b_.u16[i])
-				 : UINT16_MAX;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_adds_epu16(a, b) simde_mm_adds_epu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_and_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_and_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vandq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 = vec_and(a_.altivec_f64, b_.altivec_f64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f & b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] & b_.i32f[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_and_pd(a, b) simde_mm_and_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_and_si128(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_and_si128(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vandq_s32(b_.neon_i32, a_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u32f = vec_and(a_.altivec_u32f, b_.altivec_u32f);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f & b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] & b_.i32f[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_and_si128(a, b) simde_mm_and_si128(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_andnot_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_andnot_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vbicq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32f = vec_andc(a_.altivec_i32f, b_.altivec_i32f);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = ~a_.i32f & b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
-		r_.u64[i] = ~a_.u64[i] & b_.u64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_andnot_pd(a, b) simde_mm_andnot_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_andnot_si128(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_andnot_si128(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_andc(b_.altivec_i32, a_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = ~a_.i32f & b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = ~(a_.i32f[i]) & b_.i32f[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_andnot_si128(a, b) simde_mm_andnot_si128(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_avg_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_avg_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vrhaddq_u8(b_.neon_u8, a_.neon_u8);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u8 = vec_avg(a_.altivec_u8, b_.altivec_u8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
-	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
-	defined(SIMDE_CONVERT_VECTOR_)
-	uint16_t wa SIMDE_VECTOR(32);
-	uint16_t wb SIMDE_VECTOR(32);
-	uint16_t wr SIMDE_VECTOR(32);
-	SIMDE_CONVERT_VECTOR_(wa, a_.u8);
-	SIMDE_CONVERT_VECTOR_(wb, b_.u8);
-	wr = (wa + wb + 1) >> 1;
-	SIMDE_CONVERT_VECTOR_(r_.u8, wr);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] + b_.u8[i] + 1) >> 1;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_avg_epu8(a, b) simde_mm_avg_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_avg_epu16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_avg_epu16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vrhaddq_u16(b_.neon_u16, a_.neon_u16);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u16 = vec_avg(a_.altivec_u16, b_.altivec_u16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
-	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
-	defined(SIMDE_CONVERT_VECTOR_)
-	uint32_t wa SIMDE_VECTOR(32);
-	uint32_t wb SIMDE_VECTOR(32);
-	uint32_t wr SIMDE_VECTOR(32);
-	SIMDE_CONVERT_VECTOR_(wa, a_.u16);
-	SIMDE_CONVERT_VECTOR_(wb, b_.u16);
-	wr = (wa + wb + 1) >> 1;
-	SIMDE_CONVERT_VECTOR_(r_.u16, wr);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = (a_.u16[i] + b_.u16[i] + 1) >> 1;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_avg_epu16(a, b) simde_mm_avg_epu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_setzero_si128(void)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setzero_si128();
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vdupq_n_s32(0);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = 0;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setzero_si128() (simde_mm_setzero_si128())
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_bslli_si128(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	if (HEDLEY_UNLIKELY((imm8 & ~15))) {
-		return simde_mm_setzero_si128();
-	}
-
-#if defined(SIMDE_HAVE_INT128_) && defined(__BYTE_ORDER__) && \
-	(__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) && 0
-	r_.u128[0] = a_.u128[0] << s;
-#else
-	r_ = simde__m128i_to_private(simde_mm_setzero_si128());
-	for (int i = imm8;
-	 i < HEDLEY_STATIC_CAST(int, sizeof(r_.i8) / sizeof(r_.i8[0]));
-	 i++) {
-		r_.i8[i] = a_.i8[i - imm8];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-#define simde_mm_bslli_si128(a, imm8) _mm_slli_si128(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__)
-#define simde_mm_bslli_si128(a, imm8) \
-	simde__m128i_from_neon_i8( \
-		((imm8) <= 0) \
-			? simde__m128i_to_neon_i8(a) \
-			: (((imm8) > 15) \
-				 ? (vdupq_n_s8(0)) \
-				 : (vextq_s8(vdupq_n_s8(0), \
-					 simde__m128i_to_neon_i8(a), \
-					 16 - (imm8)))))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_bslli_si128(a, imm8) \
-	(__extension__({ \
-		const simde__m128i_private simde__tmp_a_ = \
-			simde__m128i_to_private(a); \
-		const simde__m128i_private simde__tmp_z_ = \
-			simde__m128i_to_private(simde_mm_setzero_si128()); \
-		simde__m128i_private simde__tmp_r_; \
-		if (HEDLEY_UNLIKELY(imm8 > 15)) { \
-			simde__tmp_r_ = simde__m128i_to_private( \
-				simde_mm_setzero_si128()); \
-		} else { \
-			simde__tmp_r_.i8 = SIMDE_SHUFFLE_VECTOR_( \
-				8, 16, simde__tmp_z_.i8, (simde__tmp_a_).i8, \
-				HEDLEY_STATIC_CAST(int8_t, (16 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (17 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (18 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (19 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (20 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (21 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (22 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (23 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (24 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (25 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (26 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (27 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (28 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (29 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (30 - imm8) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (31 - imm8) & 31)); \
-		} \
-		simde__m128i_from_private(simde__tmp_r_); \
-	}))
-#endif
-#define simde_mm_slli_si128(a, imm8) simde_mm_bslli_si128(a, imm8)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_bslli_si128(a, b) simde_mm_bslli_si128(a, b)
-#define _mm_slli_si128(a, b) simde_mm_bslli_si128(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_bsrli_si128(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		const int e = HEDLEY_STATIC_CAST(int, i) + imm8;
-		r_.i8[i] = (e < 16) ? a_.i8[e] : 0;
-	}
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-#define simde_mm_bsrli_si128(a, imm8) _mm_srli_si128(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__)
-#define simde_mm_bsrli_si128(a, imm8) \
-	simde__m128i_from_neon_i8( \
-		((imm8 < 0) || (imm8 > 15)) \
-			? vdupq_n_s8(0) \
-			: (vextq_s8(simde__m128i_to_private(a).neon_i8, \
-				 vdupq_n_s8(0), \
-				 ((imm8 & 15) != 0) ? imm8 : (imm8 & 15))))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_bsrli_si128(a, imm8) \
-	(__extension__({ \
-		const simde__m128i_private simde__tmp_a_ = \
-			simde__m128i_to_private(a); \
-		const simde__m128i_private simde__tmp_z_ = \
-			simde__m128i_to_private(simde_mm_setzero_si128()); \
-		simde__m128i_private simde__tmp_r_ = \
-			simde__m128i_to_private(a); \
-		if (HEDLEY_UNLIKELY(imm8 > 15)) { \
-			simde__tmp_r_ = simde__m128i_to_private( \
-				simde_mm_setzero_si128()); \
-		} else { \
-			simde__tmp_r_.i8 = SIMDE_SHUFFLE_VECTOR_( \
-				8, 16, simde__tmp_z_.i8, (simde__tmp_a_).i8, \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 16) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 17) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 18) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 19) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 20) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 21) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 22) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 23) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 24) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 25) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 26) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 27) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 28) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 29) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 30) & 31), \
-				HEDLEY_STATIC_CAST(int8_t, (imm8 + 31) & 31)); \
-		} \
-		simde__m128i_from_private(simde__tmp_r_); \
-	}))
-#endif
-#define simde_mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_bsrli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
-#define _mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_clflush(void const *p)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_clflush(p);
-#else
-	(void)p;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_clflush(a, b) simde_mm_clflush()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comieq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comieq_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !!vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] == b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comieq_sd(a, b) simde_mm_comieq_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comige_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comige_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !!vgetq_lane_u64(vcgeq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] >= b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comige_sd(a, b) simde_mm_comige_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comigt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comigt_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !!vgetq_lane_u64(vcgtq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] > b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comigt_sd(a, b) simde_mm_comigt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comile_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comile_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !!vgetq_lane_u64(vcleq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] <= b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comile_sd(a, b) simde_mm_comile_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comilt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comilt_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !!vgetq_lane_u64(vcltq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] < b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comilt_sd(a, b) simde_mm_comilt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_comineq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_comineq_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return !vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0);
-#else
-	return a_.f64[0] != b_.f64[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_comineq_sd(a, b) simde_mm_comineq_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_castpd_ps(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castpd_ps(a);
-#else
-	simde__m128 r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castpd_ps(a) simde_mm_castpd_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_castpd_si128(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castpd_si128(a);
-#else
-	simde__m128i r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castpd_si128(a) simde_mm_castpd_si128(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_castps_pd(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castps_pd(a);
-#else
-	simde__m128d r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castps_pd(a) simde_mm_castps_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_castps_si128(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castps_si128(a);
-#else
-	simde__m128i r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castps_si128(a) simde_mm_castps_si128(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_castsi128_pd(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castsi128_pd(a);
-#else
-	simde__m128d r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castsi128_pd(a) simde_mm_castsi128_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_castsi128_ps(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_castsi128_ps(a);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	return a;
-#else
-	simde__m128 r;
-	simde_memcpy(&r, &a, sizeof(a));
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_castsi128_ps(a) simde_mm_castsi128_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpeq_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpeq_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vreinterpretq_s8_u8(vceqq_s8(b_.neon_i8, a_.neon_i8));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i8x16_eq(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i8 = (SIMDE_POWER_ALTIVEC_VECTOR(signed char))vec_cmpeq(
-		a_.altivec_i8, b_.altivec_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 == b_.i8));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = (a_.i8[i] == b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_epi8(a, b) simde_mm_cmpeq_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpeq_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpeq_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 =
-		vreinterpretq_s16_u16(vceqq_s16(b_.neon_i16, a_.neon_i16));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i16x8_eq(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = (SIMDE_POWER_ALTIVEC_VECTOR(signed short))vec_cmpeq(
-		a_.altivec_i16, b_.altivec_i16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = (a_.i16 == b_.i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] == b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_epi16(a, b) simde_mm_cmpeq_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpeq_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpeq_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 =
-		vreinterpretq_s32_u32(vceqq_s32(b_.neon_i32, a_.neon_i32));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i32x4_eq(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = (SIMDE_POWER_ALTIVEC_VECTOR(signed int))vec_cmpeq(
-		a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), a_.i32 == b_.i32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = (a_.i32[i] == b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_epi32(a, b) simde_mm_cmpeq_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpeq_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpeq_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_i64 = vreinterpretq_s64_u64(
-		vceqq_s64(vreinterpretq_s64_f64(b_.neon_f64),
-			 vreinterpretq_s64_f64(a_.neon_f64)));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_eq(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 = (SIMDE_POWER_ALTIVEC_VECTOR(double))vec_cmpeq(
-		a_.altivec_f64, b_.altivec_f64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 == b_.f64));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] == b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_pd(a, b) simde_mm_cmpeq_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpeq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpeq_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpeq_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.u64[0] == b_.u64[0]) ? ~UINT64_C(0) : 0;
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpeq_sd(a, b) simde_mm_cmpeq_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpneq_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpneq_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vreinterpretq_f32_u16(
-		vmvnq_u16(vceqq_s16(b_.neon_i16, a_.neon_i16)));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_ne(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 != b_.f64));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] != b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpneq_pd(a, b) simde_mm_cmpneq_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpneq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpneq_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpneq_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.f64[0] != b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpneq_sd(a, b) simde_mm_cmpneq_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmplt_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmplt_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vreinterpretq_s8_u8(vcltq_s8(a_.neon_i8, b_.neon_i8));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i8 = HEDLEY_REINTERPRET_CAST(
-		SIMDE_POWER_ALTIVEC_VECTOR(signed char),
-		vec_cmplt(a_.altivec_i8, b_.altivec_i8));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i8x16_lt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 < b_.i8));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = (a_.i8[i] < b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_epi8(a, b) simde_mm_cmplt_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmplt_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmplt_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 =
-		vreinterpretq_s16_u16(vcltq_s16(a_.neon_i16, b_.neon_i16));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(
-		SIMDE_POWER_ALTIVEC_VECTOR(signed short),
-		vec_cmplt(a_.altivec_i16, b_.altivec_i16));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i16x8_lt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = HEDLEY_STATIC_CAST(__typeof__(r_.i16), (a_.i16 < b_.i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_epi16(a, b) simde_mm_cmplt_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmplt_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmplt_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 =
-		vreinterpretq_s32_u32(vcltq_s32(a_.neon_i32, b_.neon_i32));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = HEDLEY_REINTERPRET_CAST(
-		SIMDE_POWER_ALTIVEC_VECTOR(signed int),
-		vec_cmplt(a_.altivec_i32, b_.altivec_i32));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i32x4_lt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.i32 < b_.i32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = (a_.i32[i] < b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_epi32(a, b) simde_mm_cmplt_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmplt_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmplt_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 < b_.f64));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_lt(a_.wasm_v128, b_.wasm_v128);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] < b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_pd(a, b) simde_mm_cmplt_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmplt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmplt_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmplt_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.f64[0] < b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmplt_sd(a, b) simde_mm_cmplt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmple_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmple_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 <= b_.f64));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_le(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 = (SIMDE_POWER_ALTIVEC_VECTOR(double))vec_cmple(
-		a_.altivec_f64, b_.altivec_f64);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] <= b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmple_pd(a, b) simde_mm_cmple_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmple_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmple_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmple_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.f64[0] <= b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmple_sd(a, b) simde_mm_cmple_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpgt_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpgt_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vreinterpretq_s8_u8(vcgtq_s8(a_.neon_i8, b_.neon_i8));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i8x16_gt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i8 = (SIMDE_POWER_ALTIVEC_VECTOR(signed char))vec_cmpgt(
-		a_.altivec_i8, b_.altivec_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 > b_.i8));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = (a_.i8[i] > b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_epi8(a, b) simde_mm_cmpgt_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpgt_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpgt_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 =
-		vreinterpretq_s16_u16(vcgtq_s16(a_.neon_i16, b_.neon_i16));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i16x8_gt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(
-		SIMDE_POWER_ALTIVEC_VECTOR(signed short),
-		vec_cmpgt(a_.altivec_i16, b_.altivec_i16));
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = HEDLEY_STATIC_CAST(__typeof__(r_.i16), (a_.i16 > b_.i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_epi16(a, b) simde_mm_cmpgt_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cmpgt_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpgt_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 =
-		vreinterpretq_s32_u32(vcgtq_s32(a_.neon_i32, b_.neon_i32));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i32x4_gt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = (SIMDE_POWER_ALTIVEC_VECTOR(signed int))vec_cmpgt(
-		a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.i32 > b_.i32));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = (a_.i32[i] > b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_epi32(a, b) simde_mm_cmpgt_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpgt_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpgt_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 > b_.f64));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_gt(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 =
-		HEDLEY_STATIC_CAST(SIMDE_POWER_ALTIVEC_VECTOR(double),
-				 vec_cmpgt(a_.altivec_f64, b_.altivec_f64));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] > b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_pd(a, b) simde_mm_cmpgt_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpgt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-	return _mm_cmpgt_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpgt_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.f64[0] > b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpgt_sd(a, b) simde_mm_cmpgt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpge_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpge_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 >= b_.f64));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_ge(a_.wasm_v128, b_.wasm_v128);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 =
-		HEDLEY_STATIC_CAST(SIMDE_POWER_ALTIVEC_VECTOR(double),
-				 vec_cmpge(a_.altivec_f64, b_.altivec_f64));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (a_.f64[i] >= b_.f64[i]) ? ~UINT64_C(0)
-						 : UINT64_C(0);
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpge_pd(a, b) simde_mm_cmpge_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpge_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-	return _mm_cmpge_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpge_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.u64[0] = (a_.f64[0] >= b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpge_sd(a, b) simde_mm_cmpge_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnge_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpnge_pd(a, b);
-#else
-	return simde_mm_cmplt_pd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnge_pd(a, b) simde_mm_cmpnge_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnge_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-	return _mm_cmpnge_sd(a, b);
-#else
-	return simde_mm_cmplt_sd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnge_sd(a, b) simde_mm_cmpnge_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnlt_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpnlt_pd(a, b);
-#else
-	return simde_mm_cmpge_pd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnlt_pd(a, b) simde_mm_cmpnlt_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnlt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpnlt_sd(a, b);
-#else
-	return simde_mm_cmpge_sd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnlt_sd(a, b) simde_mm_cmpnlt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnle_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpnle_pd(a, b);
-#else
-	return simde_mm_cmpgt_pd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnle_pd(a, b) simde_mm_cmpnle_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpnle_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpnle_sd(a, b);
-#else
-	return simde_mm_cmpgt_sd(a, b);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpnle_sd(a, b) simde_mm_cmpnle_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpord_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpord_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(simde_math_isnan)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (!simde_math_isnan(a_.f64[i]) &&
-			 !simde_math_isnan(b_.f64[i]))
-				 ? ~UINT64_C(0)
-				 : UINT64_C(0);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpord_pd(a, b) simde_mm_cmpord_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde_float64 simde_mm_cvtsd_f64(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-	return _mm_cvtsd_f64(a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	return a_.f64[0];
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsd_f64(a) simde_mm_cvtsd_f64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpord_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpord_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpord_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(simde_math_isnan)
-	r_.u64[0] =
-		(!simde_math_isnan(a_.f64[0]) && !simde_math_isnan(b_.f64[0]))
-			? ~UINT64_C(0)
-			: UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpord_sd(a, b) simde_mm_cmpord_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpunord_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpunord_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(simde_math_isnan)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.u64[i] = (simde_math_isnan(a_.f64[i]) ||
-			 simde_math_isnan(b_.f64[i]))
-				 ? ~UINT64_C(0)
-				 : UINT64_C(0);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpunord_pd(a, b) simde_mm_cmpunord_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cmpunord_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cmpunord_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_cmpunord_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(simde_math_isnan)
-	r_.u64[0] = (simde_math_isnan(a_.f64[0]) || simde_math_isnan(b_.f64[0]))
-			 ? ~UINT64_C(0)
-			 : UINT64_C(0);
-	r_.u64[1] = a_.u64[1];
-
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cmpunord_sd(a, b) simde_mm_cmpunord_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtepi32_pd(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtepi32_pd(a);
-#else
-	simde__m128d_private r_;
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f64, a_.m64_private[0].i32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = (simde_float64)a_.i32[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtepi32_pd(a) simde_mm_cvtepi32_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtepi32_ps(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtepi32_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_f32 = vcvtq_f32_s32(a_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f32 = vec_ctf(a_.altivec_i32, 0);
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f32, a_.i32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
-		r_.f32[i] = (simde_float32)a_.i32[i];
-	}
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtepi32_ps(a) simde_mm_cvtepi32_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvtpd_epi32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtpd_epi32(a);
-#else
-	simde__m128i_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].i32, a_.f64);
-	r_.m64_private[1] = simde__m64_to_private(simde_mm_setzero_si64());
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(a_.f64) / sizeof(a_.f64[0])); i++) {
-		r_.i32[i] = HEDLEY_STATIC_CAST(int32_t, a_.f64[i]);
-	}
-	simde_memset(&(r_.m64_private[1]), 0, sizeof(r_.m64_private[1]));
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpd_epi32(a) simde_mm_cvtpd_epi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvtpd_pi32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpd_pi32(a);
-#else
-	simde__m64_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f64);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = HEDLEY_STATIC_CAST(int32_t, a_.f64[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpd_pi32(a) simde_mm_cvtpd_pi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtpd_ps(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtpd_ps(a);
-#else
-	simde__m128_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, a_.f64);
-	r_.m64_private[1] = simde__m64_to_private(simde_mm_setzero_si64());
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(a_.f64) / sizeof(a_.f64[0])); i++) {
-		r_.f32[i] = (simde_float32)a_.f64[i];
-	}
-	simde_memset(&(r_.m64_private[1]), 0, sizeof(r_.m64_private[1]));
-#endif
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpd_ps(a) simde_mm_cvtpd_ps(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtpi32_pd(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvtpi32_pd(a);
-#else
-	simde__m128d_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f64, a_.i32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = (simde_float64)a_.i32[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtpi32_pd(a) simde_mm_cvtpi32_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvtps_epi32(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtps_epi32(a);
-#else
-	simde__m128i_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-/* The default rounding mode on SSE is 'round to even', which ArmV7
- does not support! It is supported on ARMv8 however. */
-#if defined(SIMDE_ARCH_AARCH64)
-	r_.neon_i32 = vcvtnq_s32_f32(a_.neon_f32);
-#else
-	uint32x4_t signmask = vdupq_n_u32(0x80000000);
-	float32x4_t half = vbslq_f32(signmask, a_.neon_f32,
-				 vdupq_n_f32(0.5f)); /* +/- 0.5 */
-	int32x4_t r_normal = vcvtq_s32_f32(
-		vaddq_f32(a_.neon_f32, half)); /* round to integer: [a + 0.5]*/
-	int32x4_t r_trunc =
-		vcvtq_s32_f32(a_.neon_f32); /* truncate to integer: [a] */
-	int32x4_t plusone = vshrq_n_s32(vnegq_s32(r_trunc), 31); /* 1 or 0 */
-	int32x4_t r_even = vbicq_s32(vaddq_s32(r_trunc, plusone),
-				 vdupq_n_s32(1)); /* ([a] + {0,1}) & ~1 */
-	float32x4_t delta = vsubq_f32(
-		a_.neon_f32,
-		vcvtq_f32_s32(r_trunc)); /* compute delta: delta = (a - [a]) */
-	uint32x4_t is_delta_half =
-		vceqq_f32(delta, half); /* delta == +/- 0.5 */
-	r_.neon_i32 = vbslq_s32(is_delta_half, r_even, r_normal);
-#endif
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_cts(a_.altivec_f32, 0);
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = HEDLEY_STATIC_CAST(int32_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtps_epi32(a) simde_mm_cvtps_epi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtps_pd(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtps_pd(a);
-#else
-	simde__m128d_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.f64, a_.m64_private[0].f32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = a_.f32[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtps_pd(a) simde_mm_cvtps_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvtsd_si32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtsd_si32(a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	return SIMDE_CONVERT_FTOI(int32_t, a_.f64[0]);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsd_si32(a) simde_mm_cvtsd_si32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvtsd_si64(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if defined(__PGI)
-	return _mm_cvtsd_si64x(a);
-#else
-	return _mm_cvtsd_si64(a);
-#endif
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	return SIMDE_CONVERT_FTOI(int64_t, a_.f64[0]);
-#endif
-}
-#define simde_mm_cvtsd_si64x(a) simde_mm_cvtsd_si64(a)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsd_si64(a) simde_mm_cvtsd_si64(a)
-#define _mm_cvtsd_si64x(a) simde_mm_cvtsd_si64x(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128 simde_mm_cvtsd_ss(simde__m128 a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtsd_ss(a, b);
-#else
-	simde__m128_private r_, a_ = simde__m128_to_private(a);
-	simde__m128d_private b_ = simde__m128d_to_private(b);
-
-	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b_.f64[0]);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 1; i < (sizeof(r_) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i];
-	}
-
-	return simde__m128_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsd_ss(a, b) simde_mm_cvtsd_ss(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvtsi128_si32(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtsi128_si32(a);
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	return vgetq_lane_s32(a_.neon_i32, 0);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-#if defined(SIMDE_BUG_GCC_95227)
-	(void)a_;
-#endif
-	return vec_extract(a_.altivec_i32, 0);
-#else
-	return a_.i32[0];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi128_si32(a) simde_mm_cvtsi128_si32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvtsi128_si64(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if defined(__PGI)
-	return _mm_cvtsi128_si64x(a);
-#else
-	return _mm_cvtsi128_si64(a);
-#endif
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-#if defined(SIMDE_POWER_ALTIVEC_P5_NATIVE) && !defined(HEDLEY_IBM_VERSION)
-	return vec_extract(a_.i64, 0);
-#endif
-	return a_.i64[0];
-#endif
-}
-#define simde_mm_cvtsi128_si64x(a) simde_mm_cvtsi128_si64(a)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi128_si64(a) simde_mm_cvtsi128_si64(a)
-#define _mm_cvtsi128_si64x(a) simde_mm_cvtsi128_si64x(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtsi32_sd(simde__m128d a, int32_t b)
-{
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtsi32_sd(a, b);
-#else
-	simde__m128d_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_ARCH_AMD64)
-	r_.neon_f64 = vsetq_lane_f64((simde_float64)b, a_.neon_f64, 0);
-#else
-	r_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b);
-	r_.i64[1] = a_.i64[1];
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi32_sd(a, b) simde_mm_cvtsi32_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvtsi32_si128(int32_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtsi32_si128(a);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vsetq_lane_s32(a, vdupq_n_s32(0), 0);
-#else
-	r_.i32[0] = a;
-	r_.i32[1] = 0;
-	r_.i32[2] = 0;
-	r_.i32[3] = 0;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi32_si128(a) simde_mm_cvtsi32_si128(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtsi64_sd(simde__m128d a, int64_t b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if !defined(__PGI)
-	return _mm_cvtsi64_sd(a, b);
-#else
-	return _mm_cvtsi64x_sd(a, b);
-#endif
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f64 = vsetq_lane_f64((simde_float64)b, a_.neon_f64, 0);
-#else
-	r_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b);
-	r_.f64[1] = a_.f64[1];
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#define simde_mm_cvtsi64x_sd(a, b) simde_mm_cvtsi64_sd(a, b)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi64_sd(a, b) simde_mm_cvtsi64_sd(a, b)
-#define _mm_cvtsi64x_sd(a, b) simde_mm_cvtsi64x_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvtsi64_si128(int64_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if !defined(__PGI)
-	return _mm_cvtsi64_si128(a);
-#else
-	return _mm_cvtsi64x_si128(a);
-#endif
-#else
-	simde__m128i_private r_;
-
-	r_.i64[0] = a;
-	r_.i64[1] = 0;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#define simde_mm_cvtsi64x_si128(a) simde_mm_cvtsi64_si128(a)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtsi64_si128(a) simde_mm_cvtsi64_si128(a)
-#define _mm_cvtsi64x_si128(a) simde_mm_cvtsi64x_si128(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_cvtss_sd(simde__m128d a, simde__m128 b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvtss_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	simde__m128_private b_ = simde__m128_to_private(b);
-
-	a_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b_.f32[0]);
-
-	return simde__m128d_from_private(a_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvtss_sd(a, b) simde_mm_cvtss_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvttpd_epi32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvttpd_epi32(a);
-#else
-	simde__m128i_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-	for (size_t i = 0; i < (sizeof(a_.f64) / sizeof(a_.f64[0])); i++) {
-		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, a_.f64[i]);
-	}
-	simde_memset(&(r_.m64_private[1]), 0, sizeof(r_.m64_private[1]));
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttpd_epi32(a) simde_mm_cvttpd_epi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_cvttpd_pi32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_cvttpd_pi32(a);
-#else
-	simde__m64_private r_;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f64);
-#else
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, a_.f64[i]);
-	}
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttpd_pi32(a) simde_mm_cvttpd_pi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_cvttps_epi32(simde__m128 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvttps_epi32(a);
-#else
-	simde__m128i_private r_;
-	simde__m128_private a_ = simde__m128_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vcvtq_s32_f32(a_.neon_f32);
-#elif defined(SIMDE_CONVERT_VECTOR_)
-	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f32);
-#else
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, a_.f32[i]);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttps_epi32(a) simde_mm_cvttps_epi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_cvttsd_si32(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_cvttsd_si32(a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	return SIMDE_CONVERT_FTOI(int32_t, a_.f64[0]);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttsd_si32(a) simde_mm_cvttsd_si32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int64_t simde_mm_cvttsd_si64(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
-#if !defined(__PGI)
-	return _mm_cvttsd_si64(a);
-#else
-	return _mm_cvttsd_si64x(a);
-#endif
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-	return SIMDE_CONVERT_FTOI(int64_t, a_.f64[0]);
-#endif
-}
-#define simde_mm_cvttsd_si64x(a) simde_mm_cvttsd_si64(a)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_cvttsd_si64(a) simde_mm_cvttsd_si64(a)
-#define _mm_cvttsd_si64x(a) simde_mm_cvttsd_si64x(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_div_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_div_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f64 = a_.f64 / b_.f64;
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_div(a_.wasm_v128, b_.wasm_v128);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = a_.f64[i] / b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_div_pd(a, b) simde_mm_div_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_div_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_div_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_div_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = a_.f64[0] / b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_div_sd(a, b) simde_mm_div_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_extract_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 7)
-{
-	uint16_t r;
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-#if defined(SIMDE_BUG_GCC_95227)
-	(void)a_;
-	(void)imm8;
-#endif
-	r = vec_extract(a_.altivec_i16, imm8);
-#else
-	r = a_.u16[imm8 & 7];
-#endif
-
-	return HEDLEY_STATIC_CAST(int32_t, r);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE) && \
-	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(4, 6, 0))
-#define simde_mm_extract_epi16(a, imm8) _mm_extract_epi16(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_extract_epi16(a, imm8) \
-	HEDLEY_STATIC_CAST(int32_t, \
-			 vgetq_lane_s16(simde__m128i_to_private(a).neon_i16, \
-					 (imm8)) & \
-				 (UINT32_C(0x0000ffff)))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_extract_epi16(a, imm8) simde_mm_extract_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_insert_epi16(simde__m128i a, int16_t i, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 7)
-{
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-	a_.i16[imm8 & 7] = i;
-	return simde__m128i_from_private(a_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-#define simde_mm_insert_epi16(a, i, imm8) _mm_insert_epi16((a), (i), (imm8))
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_insert_epi16(a, i, imm8) \
-	simde__m128i_from_neon_i16( \
-		vsetq_lane_s16((i), simde__m128i_to_neon_i16(a), (imm8)))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_insert_epi16(a, i, imm8) simde_mm_insert_epi16(a, i, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d
-simde_mm_load_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_load_pd(mem_addr);
-#else
-	simde__m128d_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 =
-		vld1q_u32(HEDLEY_REINTERPRET_CAST(uint32_t const *, mem_addr));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE) && !defined(HEDLEY_IBM_VERSION)
-	r_.altivec_f64 = vec_ld(
-		0, HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(double)
-						 const *,
-					 mem_addr));
-#else
-	r_ = *SIMDE_ALIGN_CAST(simde__m128d_private const *, mem_addr);
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_load_pd(mem_addr) simde_mm_load_pd(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_load_pd1(simde_float64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_load1_pd(mem_addr);
-#else
-	simde__m128d_private r_;
-
-	r_.f64[0] = *mem_addr;
-	r_.f64[1] = *mem_addr;
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#define simde_mm_load1_pd(mem_addr) simde_mm_load_pd1(mem_addr)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_load_pd1(mem_addr) simde_mm_load_pd1(mem_addr)
-#define _mm_load1_pd(mem_addr) simde_mm_load1_pd(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_load_sd(simde_float64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_load_sd(mem_addr);
-#else
-	simde__m128d_private r_;
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f64 = vsetq_lane_f64(*mem_addr, vdupq_n_f64(0), 0);
-#else
-	r_.f64[0] = *mem_addr;
-	r_.u64[1] = UINT64_C(0);
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_load_sd(mem_addr) simde_mm_load_sd(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_load_si128(simde__m128i const *mem_addr)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_load_si128(
-		HEDLEY_REINTERPRET_CAST(__m128i const *, mem_addr));
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	simde__m128i_private r_;
-
-#if defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_ld(
-		0, HEDLEY_REINTERPRET_CAST(
-			 SIMDE_POWER_ALTIVEC_VECTOR(int) const *, mem_addr));
-#else
-	r_.neon_i32 = vld1q_s32((int32_t const *)mem_addr);
-#endif
-
-	return simde__m128i_from_private(r_);
-#else
-	return *mem_addr;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_load_si128(mem_addr) simde_mm_load_si128(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_loadh_pd(simde__m128d a, simde_float64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadh_pd(a, mem_addr);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
-	simde_float64 t;
-
-	simde_memcpy(&t, mem_addr, sizeof(t));
-	r_.f64[0] = a_.f64[0];
-	r_.f64[1] = t;
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadh_pd(a, mem_addr) simde_mm_loadh_pd(a, mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_loadl_epi64(simde__m128i const *mem_addr)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadl_epi64(
-		HEDLEY_REINTERPRET_CAST(__m128i const *, mem_addr));
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vcombine_s32(vld1_s32((int32_t const *)mem_addr),
-				 vcreate_s32(0));
-#else
-	r_.i64[0] = *HEDLEY_REINTERPRET_CAST(int64_t const *, mem_addr);
-	r_.i64[1] = 0;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadl_epi64(mem_addr) simde_mm_loadl_epi64(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_loadl_pd(simde__m128d a, simde_float64 const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadl_pd(a, mem_addr);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
-
-	r_.f64[0] = *mem_addr;
-	r_.u64[1] = a_.u64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadl_pd(a, mem_addr) simde_mm_loadl_pd(a, mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d
-simde_mm_loadr_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadr_pd(mem_addr);
-#else
-	simde__m128d_private r_;
-
-	r_.f64[0] = mem_addr[1];
-	r_.f64[1] = mem_addr[0];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadr_pd(mem_addr) simde_mm_loadr_pd(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d
-simde_mm_loadu_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadu_pd(mem_addr);
-#else
-	simde__m128d_private r_;
-
-	simde_memcpy(&r_, mem_addr, sizeof(r_));
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadu_pd(mem_addr) simde_mm_loadu_pd(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_loadu_si128(simde__m128i const *mem_addr)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_loadu_si128(HEDLEY_STATIC_CAST(__m128i const *, mem_addr));
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vld1q_s32((int32_t const *)mem_addr);
-#else
-	simde_memcpy(&r_, mem_addr, sizeof(r_));
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_loadu_si128(mem_addr) simde_mm_loadu_si128(mem_addr)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_madd_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_madd_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int32x4_t pl =
-		vmull_s16(vget_low_s16(a_.neon_i16), vget_low_s16(b_.neon_i16));
-	int32x4_t ph = vmull_s16(vget_high_s16(a_.neon_i16),
-				 vget_high_s16(b_.neon_i16));
-	int32x2_t rl = vpadd_s32(vget_low_s32(pl), vget_high_s32(pl));
-	int32x2_t rh = vpadd_s32(vget_low_s32(ph), vget_high_s32(ph));
-	r_.neon_i32 = vcombine_s32(rl, rh);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i += 2) {
-		r_.i32[i / 2] = (a_.i16[i] * b_.i16[i]) +
-				(a_.i16[i + 1] * b_.i16[i + 1]);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_madd_epi16(a, b) simde_mm_madd_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_maskmoveu_si128(simde__m128i a, simde__m128i mask,
-			 int8_t mem_addr[HEDLEY_ARRAY_PARAM(16)])
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_maskmoveu_si128(a, mask, HEDLEY_REINTERPRET_CAST(char *, mem_addr));
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a),
-			 mask_ = simde__m128i_to_private(mask);
-
-	for (size_t i = 0; i < (sizeof(a_.i8) / sizeof(a_.i8[0])); i++) {
-		if (mask_.u8[i] & 0x80) {
-			mem_addr[i] = a_.i8[i];
-		}
-	}
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_maskmoveu_si128(a, mask, mem_addr) \
-	simde_mm_maskmoveu_si128( \
-		(a), (mask), \
-		SIMDE_CHECKED_REINTERPRET_CAST(int8_t *, char *, (mem_addr)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_movemask_epi8(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__INTEL_COMPILER)
-	/* ICC has trouble with _mm_movemask_epi8 at -O2 and above: */
-	return _mm_movemask_epi8(a);
-#else
-	int32_t r = 0;
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	uint8x16_t input = a_.neon_u8;
-	SIMDE_ALIGN_AS(16, int8x8_t)
-	static const int8_t xr[8] = {-7, -6, -5, -4, -3, -2, -1, 0};
-	uint8x8_t mask_and = vdup_n_u8(0x80);
-	int8x8_t mask_shift = vld1_s8(xr);
-
-	uint8x8_t lo = vget_low_u8(input);
-	uint8x8_t hi = vget_high_u8(input);
-
-	lo = vand_u8(lo, mask_and);
-	lo = vshl_u8(lo, mask_shift);
-
-	hi = vand_u8(hi, mask_and);
-	hi = vshl_u8(hi, mask_shift);
-
-	lo = vpadd_u8(lo, lo);
-	lo = vpadd_u8(lo, lo);
-	lo = vpadd_u8(lo, lo);
-
-	hi = vpadd_u8(hi, hi);
-	hi = vpadd_u8(hi, hi);
-	hi = vpadd_u8(hi, hi);
-
-	r = ((hi[0] << 8) | (lo[0] & 0xFF));
-#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && !defined(HEDLEY_IBM_VERSION)
-	static const SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
-		perm = {120, 112, 104, 96, 88, 80, 72, 64,
-			56, 48, 40, 32, 24, 16, 8, 0};
-	r = HEDLEY_STATIC_CAST(
-		int32_t, vec_extract(vec_vbpermq(a_.altivec_u8, perm), 1));
-#else
-	SIMDE_VECTORIZE_REDUCTION(| : r)
-	for (size_t i = 0; i < (sizeof(a_.u8) / sizeof(a_.u8[0])); i++) {
-		r |= (a_.u8[15 - i] >> 7) << (15 - i);
-	}
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_movemask_epi8(a) simde_mm_movemask_epi8(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int32_t simde_mm_movemask_pd(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_movemask_pd(a);
-#else
-	int32_t r = 0;
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(a_.u64) / sizeof(a_.u64[0])); i++) {
-		r |= (a_.u64[i] >> 63) << i;
-	}
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_movemask_pd(a) simde_mm_movemask_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_movepi64_pi64(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_movepi64_pi64(a);
-#else
-	simde__m64_private r_;
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-	r_.i64[0] = a_.i64[0];
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_movepi64_pi64(a) simde_mm_movepi64_pi64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_movpi64_epi64(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_movpi64_epi64(a);
-#else
-	simde__m128i_private r_;
-	simde__m64_private a_ = simde__m64_to_private(a);
-
-	r_.i64[0] = a_.i64[0];
-	r_.i64[1] = 0;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_movpi64_epi64(a) simde_mm_movpi64_epi64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_min_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_min_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vminq_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? a_.i16[i] : b_.i16[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_min_epi16(a, b) simde_mm_min_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_min_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_min_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vminq_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] < b_.u8[i]) ? a_.u8[i] : b_.u8[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_min_epu8(a, b) simde_mm_min_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_min_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_min_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = (a_.f64[i] < b_.f64[i]) ? a_.f64[i] : b_.f64[i];
-	}
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_min_pd(a, b) simde_mm_min_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_min_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_min_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_min_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = (a_.f64[0] < b_.f64[0]) ? a_.f64[0] : b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_min_sd(a, b) simde_mm_min_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_max_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_max_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vmaxq_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = vec_max(a_.altivec_i16, b_.altivec_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? a_.i16[i] : b_.i16[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_max_epi16(a, b) simde_mm_max_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_max_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_max_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vmaxq_u8(a_.neon_u8, b_.neon_u8);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u8 = vec_max(a_.altivec_u8, b_.altivec_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
-		r_.u8[i] = (a_.u8[i] > b_.u8[i]) ? a_.u8[i] : b_.u8[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_max_epu8(a, b) simde_mm_max_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_max_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_max_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_f64 = vec_max(a_.altivec_f64, b_.altivec_f64);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = (a_.f64[i] > b_.f64[i]) ? a_.f64[i] : b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_max_pd(a, b) simde_mm_max_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_max_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_max_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_max_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = (a_.f64[0] > b_.f64[0]) ? a_.f64[0] : b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_max_sd(a, b) simde_mm_max_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_move_epi64(simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_move_epi64(a);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vsetq_lane_s64(0, a_.neon_i64, 1);
-#else
-	r_.i64[0] = a_.i64[0];
-	r_.i64[1] = 0;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_move_epi64(a) simde_mm_move_epi64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_mul_epu32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_mul_epu32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
-		r_.u64[i] = HEDLEY_STATIC_CAST(uint64_t, a_.u32[i * 2]) *
-			 HEDLEY_STATIC_CAST(uint64_t, b_.u32[i * 2]);
-	}
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_epu32(a, b) simde_mm_mul_epu32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_mul_epi64(simde__m128i a, simde__m128i b)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 * b_.i64;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a_.i64[i] * b_.i64[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_mod_epi64(simde__m128i a, simde__m128i b)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 % b_.i64;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a_.i64[i] % b_.i64[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_mul_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_mul_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f64 = a_.f64 * b_.f64;
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_mul(a_.wasm_v128, b_.wasm_v128);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = a_.f64[i] * b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_pd(a, b) simde_mm_mul_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_mul_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_mul_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_mul_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = a_.f64[0] * b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_sd(a, b) simde_mm_mul_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_mul_su32(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
-	!defined(__PGI)
-	return _mm_mul_su32(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-	r_.u64[0] = HEDLEY_STATIC_CAST(uint64_t, a_.u32[0]) *
-		 HEDLEY_STATIC_CAST(uint64_t, b_.u32[0]);
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mul_su32(a, b) simde_mm_mul_su32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_mulhi_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_mulhi_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int16x4_t a3210 = vget_low_s16(a_.neon_i16);
-	int16x4_t b3210 = vget_low_s16(b_.neon_i16);
-	int32x4_t ab3210 = vmull_s16(a3210, b3210); /* 3333222211110000 */
-	int16x4_t a7654 = vget_high_s16(a_.neon_i16);
-	int16x4_t b7654 = vget_high_s16(b_.neon_i16);
-	int32x4_t ab7654 = vmull_s16(a7654, b7654); /* 7777666655554444 */
-	uint16x8x2_t rv = vuzpq_u16(vreinterpretq_u16_s32(ab3210),
-				 vreinterpretq_u16_s32(ab7654));
-	r_.neon_u16 = rv.val[1];
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(
-			uint16_t,
-			(HEDLEY_STATIC_CAST(
-				 uint32_t,
-				 HEDLEY_STATIC_CAST(int32_t, a_.i16[i]) *
-					 HEDLEY_STATIC_CAST(int32_t,
-							 b_.i16[i])) >>
-			 16));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mulhi_epi16(a, b) simde_mm_mulhi_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_mulhi_epu16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-	return _mm_mulhi_epu16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(
-			uint16_t,
-			HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
-					HEDLEY_STATIC_CAST(uint32_t,
-							 b_.u16[i]) >>
-				16);
-	}
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mulhi_epu16(a, b) simde_mm_mulhi_epu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_mullo_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_mullo_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vmulq_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(
-			uint16_t,
-			HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
-				HEDLEY_STATIC_CAST(uint32_t, b_.u16[i]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mullo_epi16(a, b) simde_mm_mullo_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_or_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_or_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f | b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] | b_.i32f[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_or_pd(a, b) simde_mm_or_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_or_si128(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_or_si128(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vorrq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_or(a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f | b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] | b_.i32f[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_or_si128(a, b) simde_mm_or_si128(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_packs_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_packs_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 =
-		vcombine_s8(vqmovn_s16(a_.neon_i16), vqmovn_s16(b_.neon_i16));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i8[i] = (a_.i16[i] > INT8_MAX)
-				 ? INT8_MAX
-				 : ((a_.i16[i] < INT8_MIN)
-					 ? INT8_MIN
-					 : HEDLEY_STATIC_CAST(int8_t,
-								 a_.i16[i]));
-		r_.i8[i + 8] = (b_.i16[i] > INT8_MAX)
-				 ? INT8_MAX
-				 : ((b_.i16[i] < INT8_MIN)
-						 ? INT8_MIN
-						 : HEDLEY_STATIC_CAST(
-							 int8_t, b_.i16[i]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_packs_epi16(a, b) simde_mm_packs_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_packs_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_packs_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 =
-		vcombine_s16(vqmovn_s32(a_.neon_i32), vqmovn_s32(b_.neon_i32));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i16 = vec_packs(a_.altivec_i32, b_.altivec_i32);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i16[i] = (a_.i32[i] > INT16_MAX)
-				 ? INT16_MAX
-				 : ((a_.i32[i] < INT16_MIN)
-					 ? INT16_MIN
-					 : HEDLEY_STATIC_CAST(int16_t,
-								 a_.i32[i]));
-		r_.i16[i + 4] =
-			(b_.i32[i] > INT16_MAX)
-				? INT16_MAX
-				: ((b_.i32[i] < INT16_MIN)
-					 ? INT16_MIN
-					 : HEDLEY_STATIC_CAST(int16_t,
-								b_.i32[i]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_packs_epi32(a, b) simde_mm_packs_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_packus_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_packus_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 =
-		vcombine_u8(vqmovun_s16(a_.neon_i16), vqmovun_s16(b_.neon_i16));
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_u8 = vec_packsu(a_.altivec_i16, b_.altivec_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.u8[i] = (a_.i16[i] > UINT8_MAX)
-				 ? UINT8_MAX
-				 : ((a_.i16[i] < 0)
-					 ? UINT8_C(0)
-					 : HEDLEY_STATIC_CAST(uint8_t,
-								 a_.i16[i]));
-		r_.u8[i + 8] =
-			(b_.i16[i] > UINT8_MAX)
-				? UINT8_MAX
-				: ((b_.i16[i] < 0)
-					 ? UINT8_C(0)
-					 : HEDLEY_STATIC_CAST(uint8_t,
-								b_.i16[i]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_packus_epi16(a, b) simde_mm_packus_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_pause(void)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_pause();
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_pause() (simde_mm_pause())
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sad_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sad_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		uint16_t tmp = 0;
-		SIMDE_VECTORIZE_REDUCTION(+ : tmp)
-		for (size_t j = 0; j < ((sizeof(r_.u8) / sizeof(r_.u8[0])) / 2);
-		 j++) {
-			const size_t e = j + (i * 8);
-			tmp += (a_.u8[e] > b_.u8[e]) ? (a_.u8[e] - b_.u8[e])
-						 : (b_.u8[e] - a_.u8[e]);
-		}
-		r_.i64[i] = tmp;
-	}
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sad_epu8(a, b) simde_mm_sad_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set_epi8(int8_t e15, int8_t e14, int8_t e13, int8_t e12,
-			 int8_t e11, int8_t e10, int8_t e9, int8_t e8,
-			 int8_t e7, int8_t e6, int8_t e5, int8_t e4,
-			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
-{
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5,
-			 e4, e3, e2, e1, e0);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i8x16_make(e0, e1, e2, e3, e4, e5, e6, e7, e8, e9,
-				 e10, e11, e12, e13, e14, e15);
-#else
-	r_.i8[0] = e0;
-	r_.i8[1] = e1;
-	r_.i8[2] = e2;
-	r_.i8[3] = e3;
-	r_.i8[4] = e4;
-	r_.i8[5] = e5;
-	r_.i8[6] = e6;
-	r_.i8[7] = e7;
-	r_.i8[8] = e8;
-	r_.i8[9] = e9;
-	r_.i8[10] = e10;
-	r_.i8[11] = e11;
-	r_.i8[12] = e12;
-	r_.i8[13] = e13;
-	r_.i8[14] = e14;
-	r_.i8[15] = e15;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, e4, e3, \
-		 e2, e1, e0) \
-	simde_mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, \
-			 e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set_epi16(int16_t e7, int16_t e6, int16_t e5, int16_t e4,
-				int16_t e3, int16_t e2, int16_t e1, int16_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN_AS(16, int16x8_t)
-	int16_t data[8] = {e0, e1, e2, e3, e4, e5, e6, e7};
-	r_.neon_i16 = vld1q_s16(data);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i16x8_make(e0, e1, e2, e3, e4, e5, e6, e7);
-#else
-	r_.i16[0] = e0;
-	r_.i16[1] = e1;
-	r_.i16[2] = e2;
-	r_.i16[3] = e3;
-	r_.i16[4] = e4;
-	r_.i16[5] = e5;
-	r_.i16[6] = e6;
-	r_.i16[7] = e7;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0) \
-	simde_mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set_epi32(int32_t e3, int32_t e2, int32_t e1, int32_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi32(e3, e2, e1, e0);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	SIMDE_ALIGN_AS(16, int32x4_t) int32_t data[4] = {e0, e1, e2, e3};
-	r_.neon_i32 = vld1q_s32(data);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i32x4_make(e0, e1, e2, e3);
-#else
-	r_.i32[0] = e0;
-	r_.i32[1] = e1;
-	r_.i32[2] = e2;
-	r_.i32[3] = e3;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_epi32(e3, e2, e1, e0) simde_mm_set_epi32(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set_epi64(simde__m64 e1, simde__m64 e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set_epi64(e1, e0);
-#else
-	simde__m128i_private r_;
-
-	r_.m64_private[0] = simde__m64_to_private(e0);
-	r_.m64_private[1] = simde__m64_to_private(e1);
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_epi64(e1, e0) (simde_mm_set_epi64((e1), (e0)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set_epi64x(int64_t e1, int64_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && \
-	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
-	return _mm_set_epi64x(e1, e0);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vcombine_s64(vdup_n_s64(e0), vdup_n_s64(e1));
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i64x2_make(e0, e1);
-#else
-	r_.i64[0] = e0;
-	r_.i64[1] = e1;
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_epi64x(e1, e0) simde_mm_set_epi64x(e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set_epu8(uint8_t e15, uint8_t e14, uint8_t e13,
-				 uint8_t e12, uint8_t e11, uint8_t e10,
-				 uint8_t e9, uint8_t e8, uint8_t e7, uint8_t e6,
-				 uint8_t e5, uint8_t e4, uint8_t e3, uint8_t e2,
-				 uint8_t e1, uint8_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi8(
-		HEDLEY_STATIC_CAST(char, e15), HEDLEY_STATIC_CAST(char, e14),
-		HEDLEY_STATIC_CAST(char, e13), HEDLEY_STATIC_CAST(char, e12),
-		HEDLEY_STATIC_CAST(char, e11), HEDLEY_STATIC_CAST(char, e10),
-		HEDLEY_STATIC_CAST(char, e9), HEDLEY_STATIC_CAST(char, e8),
-		HEDLEY_STATIC_CAST(char, e7), HEDLEY_STATIC_CAST(char, e6),
-		HEDLEY_STATIC_CAST(char, e5), HEDLEY_STATIC_CAST(char, e4),
-		HEDLEY_STATIC_CAST(char, e3), HEDLEY_STATIC_CAST(char, e2),
-		HEDLEY_STATIC_CAST(char, e1), HEDLEY_STATIC_CAST(char, e0));
-#else
-	simde__m128i_private r_;
-
-	r_.u8[0] = e0;
-	r_.u8[1] = e1;
-	r_.u8[2] = e2;
-	r_.u8[3] = e3;
-	r_.u8[4] = e4;
-	r_.u8[5] = e5;
-	r_.u8[6] = e6;
-	r_.u8[7] = e7;
-	r_.u8[8] = e8;
-	r_.u8[9] = e9;
-	r_.u8[10] = e10;
-	r_.u8[11] = e11;
-	r_.u8[12] = e12;
-	r_.u8[13] = e13;
-	r_.u8[14] = e14;
-	r_.u8[15] = e15;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set_epu16(uint16_t e7, uint16_t e6, uint16_t e5,
-				 uint16_t e4, uint16_t e3, uint16_t e2,
-				 uint16_t e1, uint16_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi16(
-		HEDLEY_STATIC_CAST(short, e7), HEDLEY_STATIC_CAST(short, e6),
-		HEDLEY_STATIC_CAST(short, e5), HEDLEY_STATIC_CAST(short, e4),
-		HEDLEY_STATIC_CAST(short, e3), HEDLEY_STATIC_CAST(short, e2),
-		HEDLEY_STATIC_CAST(short, e1), HEDLEY_STATIC_CAST(short, e0));
-#else
-	simde__m128i_private r_;
-
-	r_.u16[0] = e0;
-	r_.u16[1] = e1;
-	r_.u16[2] = e2;
-	r_.u16[3] = e3;
-	r_.u16[4] = e4;
-	r_.u16[5] = e5;
-	r_.u16[6] = e6;
-	r_.u16[7] = e7;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set_epu32(uint32_t e3, uint32_t e2, uint32_t e1,
-				 uint32_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_epi32(HEDLEY_STATIC_CAST(int, e3),
-			 HEDLEY_STATIC_CAST(int, e2),
-			 HEDLEY_STATIC_CAST(int, e1),
-			 HEDLEY_STATIC_CAST(int, e0));
-#else
-	simde__m128i_private r_;
-
-	r_.u32[0] = e0;
-	r_.u32[1] = e1;
-	r_.u32[2] = e2;
-	r_.u32[3] = e3;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set_epu64x(uint64_t e1, uint64_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && \
-	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
-	return _mm_set_epi64x(HEDLEY_STATIC_CAST(int64_t, e1),
-			 HEDLEY_STATIC_CAST(int64_t, e0));
-#else
-	simde__m128i_private r_;
-
-	r_.u64[0] = e0;
-	r_.u64[1] = e1;
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_set_pd(simde_float64 e1, simde_float64 e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_pd(e1, e0);
-#else
-	simde__m128d_private r_;
-
-#if defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_make(e0, e1);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_make(e0, e1);
-#else
-	r_.f64[0] = e0;
-	r_.f64[1] = e1;
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_pd(e1, e0) simde_mm_set_pd(e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_set_pd1(simde_float64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set1_pd(a);
-#else
-	simde__m128d_private r_;
-
-	r_.f64[0] = a;
-	r_.f64[1] = a;
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_pd1(a) simde_mm_set1_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_set_sd(simde_float64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set_sd(a);
-#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	return vsetq_lane_f64(a, vdupq_n_f64(SIMDE_FLOAT32_C(0.0)), 0);
-#else
-	return simde_mm_set_pd(SIMDE_FLOAT64_C(0.0), a);
-
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set_sd(a) simde_mm_set_sd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set1_epi8(int8_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set1_epi8(a);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vdupq_n_s8(a);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i8x16_splat(a);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = a;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_epi8(a) simde_mm_set1_epi8(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set1_epi16(int16_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set1_epi16(a);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vdupq_n_s16(a);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i16x8_splat(a);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_epi16(a) simde_mm_set1_epi16(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set1_epi32(int32_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set1_epi32(a);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vdupq_n_s32(a);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i32x4_splat(a);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_epi32(a) simde_mm_set1_epi32(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set1_epi64x(int64_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && \
-	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
-	return _mm_set1_epi64x(a);
-#else
-	simde__m128i_private r_;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vmovq_n_s64(a);
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_i64x2_splat(a);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_epi64x(a) simde_mm_set1_epi64x(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_set1_epi64(simde__m64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_set1_epi64(a);
-#else
-	simde__m64_private a_ = simde__m64_to_private(a);
-	return simde_mm_set1_epi64x(a_.i64[0]);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_epi64(a) simde_mm_set1_epi64(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set1_epu8(uint8_t value)
-{
-	return simde_mm_set1_epi8(HEDLEY_STATIC_CAST(int8_t, value));
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set1_epu16(uint16_t value)
-{
-	return simde_mm_set1_epi16(HEDLEY_STATIC_CAST(int16_t, value));
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set1_epu32(uint32_t value)
-{
-	return simde_mm_set1_epi32(HEDLEY_STATIC_CAST(int32_t, value));
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_set1_epu64(uint64_t value)
-{
-	return simde_mm_set1_epi64x(HEDLEY_STATIC_CAST(int64_t, value));
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_set1_pd(simde_float64 a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_set1_pd(a);
-#else
-	simde__m128d_private r_;
-
-#if defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_splat(a);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.f64[i] = a;
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_set1_pd(a) simde_mm_set1_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_setr_epi8(int8_t e15, int8_t e14, int8_t e13, int8_t e12,
-				int8_t e11, int8_t e10, int8_t e9, int8_t e8,
-				int8_t e7, int8_t e6, int8_t e5, int8_t e4,
-				int8_t e3, int8_t e2, int8_t e1, int8_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5,
-			 e4, e3, e2, e1, e0);
-#else
-	return simde_mm_set_epi8(e0, e1, e2, e3, e4, e5, e6, e7, e8, e9, e10,
-				 e11, e12, e13, e14, e15);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, e4, \
-		 e3, e2, e1, e0) \
-	simde_mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, \
-			 e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_setr_epi16(int16_t e7, int16_t e6, int16_t e5, int16_t e4,
-				 int16_t e3, int16_t e2, int16_t e1, int16_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0);
-#else
-	return simde_mm_set_epi16(e0, e1, e2, e3, e4, e5, e6, e7);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0) \
-	simde_mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_setr_epi32(int32_t e3, int32_t e2, int32_t e1, int32_t e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setr_epi32(e3, e2, e1, e0);
-#else
-	return simde_mm_set_epi32(e0, e1, e2, e3);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_epi32(e3, e2, e1, e0) simde_mm_setr_epi32(e3, e2, e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_setr_epi64(simde__m64 e1, simde__m64 e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_setr_epi64(e1, e0);
-#else
-	return simde_mm_set_epi64(e0, e1);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_epi64(e1, e0) (simde_mm_setr_epi64((e1), (e0)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_setr_pd(simde_float64 e1, simde_float64 e0)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setr_pd(e1, e0);
-#else
-	return simde_mm_set_pd(e0, e1);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setr_pd(e1, e0) simde_mm_setr_pd(e1, e0)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_setzero_pd(void)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_setzero_pd();
-#else
-	simde__m128d_private r_;
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = 0;
-	}
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_setzero_pd() simde_mm_setzero_pd()
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_undefined_pd(void)
-{
-	simde__m128d_private r_;
-
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE__HAVE_UNDEFINED128)
-	r_.n = _mm_undefined_pd();
-#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-	r_ = simde__m128d_to_private(simde_mm_setzero_pd());
-#endif
-
-	return simde__m128d_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_undefined_pd() simde_mm_undefined_pd()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_undefined_si128(void)
-{
-	simde__m128i_private r_;
-
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE__HAVE_UNDEFINED128)
-	r_.n = _mm_undefined_si128();
-#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-	r_ = simde__m128i_to_private(simde_mm_setzero_si128());
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_undefined_si128() (simde_mm_undefined_si128())
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_POP
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_x_mm_setone_pd(void)
-{
-	return simde_mm_castps_pd(simde_x_mm_setone_ps());
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_setone_si128(void)
-{
-	return simde_mm_castps_si128(simde_x_mm_setone_ps());
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_shuffle_epi32(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[(imm8 >> (i * 2)) & 3];
-	}
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_shuffle_epi32(a, imm8) _mm_shuffle_epi32((a), (imm8))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shuffle_epi32(a, imm8) \
-	(__extension__({ \
-		const simde__m128i_private simde__tmp_a_ = \
-			simde__m128i_to_private(a); \
-		simde__m128i_from_private((simde__m128i_private){ \
-			.i32 = SIMDE_SHUFFLE_VECTOR_( \
-				32, 16, (simde__tmp_a_).i32, \
-				(simde__tmp_a_).i32, ((imm8)) & 3, \
-				((imm8) >> 2) & 3, ((imm8) >> 4) & 3, \
-				((imm8) >> 6) & 3)}); \
-	}))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_shuffle_epi32(a, imm8) simde_mm_shuffle_epi32(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_shuffle_pd(simde__m128d a, simde__m128d b, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 3)
-{
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = ((imm8 & 1) == 0) ? a_.f64[0] : a_.f64[1];
-	r_.f64[1] = ((imm8 & 2) == 0) ? b_.f64[0] : b_.f64[1];
-
-	return simde__m128d_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
-#define simde_mm_shuffle_pd(a, b, imm8) _mm_shuffle_pd((a), (b), (imm8))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shuffle_pd(a, b, imm8) \
-	(__extension__({ \
-		simde__m128d_from_private((simde__m128d_private){ \
-			.f64 = SIMDE_SHUFFLE_VECTOR_( \
-				64, 16, simde__m128d_to_private(a).f64, \
-				simde__m128d_to_private(b).f64, \
-				(((imm8)) & 1), (((imm8) >> 1) & 1) + 2)}); \
-	}))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_shuffle_pd(a, b, imm8) simde_mm_shuffle_pd(a, b, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_shufflehi_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
-	 i++) {
-		r_.i16[i] = a_.i16[i];
-	}
-	for (size_t i = ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
-	 i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[((imm8 >> ((i - 4) * 2)) & 3) + 4];
-	}
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_shufflehi_epi16(a, imm8) _mm_shufflehi_epi16((a), (imm8))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shufflehi_epi16(a, imm8) \
-	(__extension__({ \
-		const simde__m128i_private simde__tmp_a_ = \
-			simde__m128i_to_private(a); \
-		simde__m128i_from_private((simde__m128i_private){ \
-			.i16 = SIMDE_SHUFFLE_VECTOR_( \
-				16, 16, (simde__tmp_a_).i16, \
-				(simde__tmp_a_).i16, 0, 1, 2, 3, \
-				(((imm8)) & 3) + 4, (((imm8) >> 2) & 3) + 4, \
-				(((imm8) >> 4) & 3) + 4, \
-				(((imm8) >> 6) & 3) + 4)}); \
-	}))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_shufflehi_epi16(a, imm8) simde_mm_shufflehi_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_shufflelo_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	for (size_t i = 0; i < ((sizeof(r_.i16) / sizeof(r_.i16[0])) / 2);
-	 i++) {
-		r_.i16[i] = a_.i16[((imm8 >> (i * 2)) & 3)];
-	}
-	SIMDE_VECTORIZE
-	for (size_t i = ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
-	 i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i];
-	}
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_shufflelo_epi16(a, imm8) _mm_shufflelo_epi16((a), (imm8))
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-#define simde_mm_shufflelo_epi16(a, imm8) \
-	(__extension__({ \
-		const simde__m128i_private simde__tmp_a_ = \
-			simde__m128i_to_private(a); \
-		simde__m128i_from_private((simde__m128i_private){ \
-			.i16 = SIMDE_SHUFFLE_VECTOR_( \
-				16, 16, (simde__tmp_a_).i16, \
-				(simde__tmp_a_).i16, (((imm8)) & 3), \
-				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3), \
-				(((imm8) >> 6) & 3), 4, 5, 6, 7)}); \
-	}))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_shufflelo_epi16(a, imm8) simde_mm_shufflelo_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sll_epi16(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sll_epi16(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	if (count_.u64[0] > 15)
-		return simde_mm_setzero_si128();
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u16 = (a_.u16 << count_.u64[0]);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t,
-					 (a_.u16[i] << count_.u64[0]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_epi16(a, count) simde_mm_sll_epi16((a), (count))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sll_epi32(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sll_epi32(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	if (count_.u64[0] > 31)
-		return simde_mm_setzero_si128();
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u32 = (a_.u32 << count_.u64[0]);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = HEDLEY_STATIC_CAST(uint32_t,
-					 (a_.u32[i] << count_.u64[0]));
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_epi32(a, count) (simde_mm_sll_epi32(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sll_epi64(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sll_epi64(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	if (count_.u64[0] > 63)
-		return simde_mm_setzero_si128();
-
-	const int_fast16_t s = HEDLEY_STATIC_CAST(int_fast16_t, count_.u64[0]);
-#if !defined(SIMDE_BUG_GCC_94488)
-	SIMDE_VECTORIZE
-#endif
-	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
-		r_.u64[i] = a_.u64[i] << s;
-	}
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sll_epi64(a, count) (simde_mm_sll_epi64(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_sqrt_pd(simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sqrt_pd(a);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	r_.neon_f64 = vsqrtq_f64(a_.neon_f64);
-#elif defined(simde_math_sqrt)
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = simde_math_sqrt(a_.f64[i]);
-	}
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sqrt_pd(a) simde_mm_sqrt_pd(a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_sqrt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sqrt_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_sqrt_pd(b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(simde_math_sqrt)
-	r_.f64[0] = simde_math_sqrt(b_.f64[0]);
-	r_.f64[1] = a_.f64[1];
-#else
-	HEDLEY_UNREACHABLE();
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sqrt_sd(a, b) simde_mm_sqrt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srl_epi16(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_srl_epi16(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	const int cnt = HEDLEY_STATIC_CAST(
-		int, (count_.i64[0] > 16 ? 16 : count_.i64[0]));
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vshlq_u16(a_.neon_u16,
-				vdupq_n_s16(HEDLEY_STATIC_CAST(int16_t, -cnt)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
-		r_.u16[i] = a_.u16[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_epi16(a, count) (simde_mm_srl_epi16(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srl_epi32(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_srl_epi32(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	const int cnt = HEDLEY_STATIC_CAST(
-		int, (count_.i64[0] > 32 ? 32 : count_.i64[0]));
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u32 = vshlq_u32(a_.neon_u32,
-				vdupq_n_s32(HEDLEY_STATIC_CAST(int32_t, -cnt)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_epi32(a, count) (simde_mm_srl_epi32(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srl_epi64(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_srl_epi64(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	const int cnt = HEDLEY_STATIC_CAST(
-		int, (count_.i64[0] > 64 ? 64 : count_.i64[0]));
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u64 = vshlq_u64(a_.neon_u64,
-				vdupq_n_s64(HEDLEY_STATIC_CAST(int64_t, -cnt)));
-#else
-#if !defined(SIMDE_BUG_GCC_94488)
-	SIMDE_VECTORIZE
-#endif
-	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
-		r_.u64[i] = a_.u64[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srl_epi64(a, count) (simde_mm_srl_epi64(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srai_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
-{
-	/* MSVC requires a range of (0, 255). */
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	const int cnt = (imm8 & ~15) ? 15 : imm8;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vshlq_s16(a_.neon_i16, vdupq_n_s16(-cnt));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_srai_epi16(a, imm8) _mm_srai_epi16((a), (imm8))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srai_epi16(a, imm8) simde_mm_srai_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srai_epi32(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
-{
-	/* MSVC requires a range of (0, 255). */
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	const int cnt = (imm8 & ~31) ? 31 : imm8;
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vshlq_s32(a_.neon_i32, vdupq_n_s32(-cnt));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_srai_epi32(a, imm8) _mm_srai_epi32((a), (imm8))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srai_epi32(a, imm8) simde_mm_srai_epi32(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sra_epi16(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sra_epi16(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	const int cnt = HEDLEY_STATIC_CAST(
-		int, (count_.i64[0] > 15 ? 15 : count_.i64[0]));
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vshlq_s16(a_.neon_i16,
-				vdupq_n_s16(HEDLEY_STATIC_CAST(int16_t, -cnt)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sra_epi16(a, count) (simde_mm_sra_epi16(a, count))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sra_epi32(simde__m128i a, simde__m128i count)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(SIMDE_BUG_GCC_BAD_MM_SRA_EPI32)
-	return _mm_sra_epi32(a, count);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 count_ = simde__m128i_to_private(count);
-
-	const int cnt = count_.u64[0] > 31
-				? 31
-				: HEDLEY_STATIC_CAST(int, count_.u64[0]);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vshlq_s32(a_.neon_i32,
-				vdupq_n_s32(HEDLEY_STATIC_CAST(int32_t, -cnt)));
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] >> cnt;
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sra_epi32(a, count) (simde_mm_sra_epi32(a, (count)))
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_slli_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i16 = a_.i16 << (imm8 & 0xff);
-#else
-	const int s =
-		(imm8 >
-		 HEDLEY_STATIC_CAST(int, sizeof(r_.i16[0]) * CHAR_BIT) - 1)
-			? 0
-			: imm8;
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = HEDLEY_STATIC_CAST(int16_t, a_.i16[i] << s);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_slli_epi16(a, imm8) _mm_slli_epi16(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_slli_epi16(a, imm8) \
-	simde__m128i_from_neon_u16( \
-		vshlq_n_u16(simde__m128i_to_neon_u16(a), (imm8)))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_epi16(a, imm8) simde_mm_slli_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_slli_epi32(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i32 = a_.i32 << imm8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] << (imm8 & 0xff);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_slli_epi32(a, imm8) _mm_slli_epi32(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_slli_epi32(a, imm8) \
-	simde__m128i_from_neon_u32( \
-		vshlq_n_u32(simde__m128i_to_neon_u32(a), (imm8)))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_epi32(a, imm8) simde_mm_slli_epi32(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_slli_epi64(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.i64 = a_.i64 << imm8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a_.i64[i] << (imm8 & 0xff);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_slli_epi64(a, imm8) _mm_slli_epi64(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_slli_epi64(a, imm8) \
-	simde__m128i_from_neon_u64( \
-		vshlq_n_u64(simde__m128i_to_neon_u64(a), (imm8)))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_slli_epi64(a, imm8) simde_mm_slli_epi64(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srli_epi16(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u16 = a_.u16 >> imm8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.u16[i] = a_.u16[i] >> (imm8 & 0xff);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_srli_epi16(a, imm8) _mm_srli_epi16(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_srli_epi16(a, imm8) \
-	simde__m128i_from_neon_u16( \
-		vshrq_n_u16(simde__m128i_to_neon_u16(a), imm8))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_epi16(a, imm8) simde_mm_srli_epi16(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srli_epi32(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
-	r_.u32 = a_.u32 >> (imm8 & 0xff);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.u32[i] = a_.u32[i] >> (imm8 & 0xff);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_srli_epi32(a, imm8) _mm_srli_epi32(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-#define simde_mm_srli_epi32(a, imm8) \
-	simde__m128i_from_neon_u32( \
-		vshrq_n_u32(simde__m128i_to_neon_u32(a), imm8))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_epi32(a, imm8) simde_mm_srli_epi32(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_srli_epi64(simde__m128i a, const int imm8)
-	SIMDE_REQUIRE_RANGE(imm8, 0, 255)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-	if (HEDLEY_UNLIKELY((imm8 & 63) != imm8))
-		return simde_mm_setzero_si128();
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u64 = vshlq_u64(a_.neon_u64, vdupq_n_s64(-imm8));
-#else
-#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && !defined(SIMDE_BUG_GCC_94488)
-	r_.u64 = a_.u64 >> imm8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.u64[i] = a_.u64[i] >> imm8;
-	}
-#endif
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-#if defined(SIMDE_X86_SSE2_NATIVE)
-#define simde_mm_srli_epi64(a, imm8) _mm_srli_epi64(a, imm8)
-#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__)
-#define simde_mm_srli_epi64(a, imm8) \
-	((imm8 == 0) ? (a) \
-		 : (simde__m128i_from_neon_u64(vshrq_n_u64( \
-			 simde__m128i_to_neon_u64(a), imm8))))
-#endif
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_srli_epi64(a, imm8) simde_mm_srli_epi64(a, imm8)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
-		 simde__m128d a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_store_pd(mem_addr, a);
-#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	vst1q_f64(mem_addr, simde__m128d_to_private(a).neon_f64);
-#else
-	simde_memcpy(mem_addr, &a, sizeof(a));
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_store_pd(mem_addr, a) \
-	simde_mm_store_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store1_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
-			simde__m128d a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_store1_pd(mem_addr, a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-	mem_addr[0] = a_.f64[0];
-	mem_addr[1] = a_.f64[0];
-#endif
-}
-#define simde_mm_store_pd1(mem_addr, a) \
-	simde_mm_store1_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_store1_pd(mem_addr, a) \
-	simde_mm_store1_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#define _mm_store_pd1(mem_addr, a) \
-	simde_mm_store_pd1(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_sd(simde_float64 *mem_addr, simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_store_sd(mem_addr, a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	simde_float64 v = vgetq_lane_f64(a_.neon_f64, 0);
-	simde_memcpy(mem_addr, &v, sizeof(simde_float64));
-#else
-	simde_float64 v = a_.f64[0];
-	simde_memcpy(mem_addr, &v, sizeof(simde_float64));
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_store_sd(mem_addr, a) \
-	simde_mm_store_sd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_store_si128(simde__m128i *mem_addr, simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_store_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_s32(HEDLEY_REINTERPRET_CAST(int32_t *, mem_addr), a_.neon_i32);
-#else
-	simde_memcpy(SIMDE_ASSUME_ALIGNED(16, mem_addr), &a_, sizeof(a_));
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_store_si128(mem_addr, a) simde_mm_store_si128(mem_addr, a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storeh_pd(simde_float64 *mem_addr, simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storeh_pd(mem_addr, a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
-	*mem_addr = vgetq_lane_f64(a_.neon_f64, 1);
-#else
-	*mem_addr = a_.f64[1];
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storeh_pd(mem_addr, a) \
-	simde_mm_storeh_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storel_epi64(simde__m128i *mem_addr, simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storel_epi64(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-	int64_t tmp;
-
-	/* memcpy to prevent aliasing, tmp because we can't take the
- * address of a vector element. */
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	tmp = vgetq_lane_s64(a_.neon_i64, 0);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-#if defined(SIMDE_BUG_GCC_95227)
-	(void)a_;
-#endif
-	tmp = vec_extract(a_.altivec_i64, 0);
-#else
-	tmp = a_.i64[0];
-#endif
-
-	simde_memcpy(mem_addr, &tmp, sizeof(tmp));
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storel_epi64(mem_addr, a) simde_mm_storel_epi64(mem_addr, a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storel_pd(simde_float64 *mem_addr, simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storel_pd(mem_addr, a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-	*mem_addr = a_.f64[0];
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storel_pd(mem_addr, a) \
-	simde_mm_storel_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storer_pd(simde_float64 mem_addr[2], simde__m128d a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storer_pd(mem_addr, a);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a);
-
-	mem_addr[0] = a_.f64[1];
-	mem_addr[1] = a_.f64[0];
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storer_pd(mem_addr, a) \
-	simde_mm_storer_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storeu_pd(simde_float64 *mem_addr, simde__m128d a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storeu_pd(mem_addr, a);
-#else
-	simde_memcpy(mem_addr, &a, sizeof(a));
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storeu_pd(mem_addr, a) \
-	simde_mm_storeu_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_storeu_si128(simde__m128i *mem_addr, simde__m128i a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_storeu_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
-#else
-	simde__m128i_private a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	vst1q_s32(HEDLEY_REINTERPRET_CAST(int32_t *, mem_addr), a_.neon_i32);
-#else
-	simde_memcpy(mem_addr, &a_, sizeof(a_));
-#endif
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_storeu_si128(mem_addr, a) simde_mm_storeu_si128(mem_addr, a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
-			simde__m128d a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_stream_pd(mem_addr, a);
-#else
-	simde_memcpy(mem_addr, &a, sizeof(a));
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_pd(mem_addr, a) \
-	simde_mm_stream_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_si128(simde__m128i *mem_addr, simde__m128i a)
-{
-	simde_assert_aligned(16, mem_addr);
-
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_stream_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
-#else
-	simde_memcpy(mem_addr, &a, sizeof(a));
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_si128(mem_addr, a) simde_mm_stream_si128(mem_addr, a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_si32(int32_t *mem_addr, int32_t a)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_stream_si32(mem_addr, a);
-#else
-	*mem_addr = a;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_si32(mem_addr, a) simde_mm_stream_si32(mem_addr, a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_stream_si64(int64_t *mem_addr, int64_t a)
-{
-	*mem_addr = a;
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_stream_si64(mem_addr, a) \
-	simde_mm_stream_si64(SIMDE_CHECKED_REINTERPRET_CAST( \
-				 int64_t *, __int64 *, mem_addr), \
-			 a)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sub_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vsubq_s8(a_.neon_i8, b_.neon_i8);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i8 = a_.i8 - b_.i8;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
-		r_.i8[i] = a_.i8[i] - b_.i8[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_epi8(a, b) simde_mm_sub_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sub_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vsubq_s16(a_.neon_i16, b_.neon_i16);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i16 = a_.i16 - b_.i16;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
-		r_.i16[i] = a_.i16[i] - b_.i16[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_epi16(a, b) simde_mm_sub_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sub_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vsubq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32 = a_.i32 - b_.i32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
-		r_.i32[i] = a_.i32[i] - b_.i32[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_epi32(a, b) simde_mm_sub_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_sub_epi64(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_epi64(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i64 = vsubq_s64(a_.neon_i64, b_.neon_i64);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 - b_.i64;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
-		r_.i64[i] = a_.i64[i] - b_.i64[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_epi64(a, b) simde_mm_sub_epi64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_sub_epu32(simde__m128i a, simde__m128i b)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.u32 = a_.u32 - b_.u32;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
-		r_.u32[i] = a_.u32[i] - b_.u32[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_sub_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.f64 = a_.f64 - b_.f64;
-#elif defined(SIMDE_WASM_SIMD128_NATIVE)
-	r_.wasm_v128 = wasm_f64x2_sub(a_.wasm_v128, b_.wasm_v128);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
-		r_.f64[i] = a_.f64[i] - b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_pd(a, b) simde_mm_sub_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_sub_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_sub_sd(a, b);
-#elif defined(SIMDE_ASSUME_VECTORIZATION)
-	return simde_mm_move_sd(a, simde_mm_sub_pd(a, b));
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-	r_.f64[0] = a_.f64[0] - b_.f64[0];
-	r_.f64[1] = a_.f64[1];
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_sd(a, b) simde_mm_sub_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m64 simde_mm_sub_si64(simde__m64 a, simde__m64 b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
-	return _mm_sub_si64(a, b);
-#else
-	simde__m64_private r_, a_ = simde__m64_to_private(a),
-			 b_ = simde__m64_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i64 = a_.i64 - b_.i64;
-#else
-	r_.i64[0] = a_.i64[0] - b_.i64[0];
-#endif
-
-	return simde__m64_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_sub_si64(a, b) simde_mm_sub_si64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_subs_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_subs_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i8 = vqsubq_s8(a_.neon_i8, b_.neon_i8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i8[0])); i++) {
-		if (((b_.i8[i]) > 0 && (a_.i8[i]) < INT8_MIN + (b_.i8[i]))) {
-			r_.i8[i] = INT8_MIN;
-		} else if ((b_.i8[i]) < 0 &&
-			 (a_.i8[i]) > INT8_MAX + (b_.i8[i])) {
-			r_.i8[i] = INT8_MAX;
-		} else {
-			r_.i8[i] = (a_.i8[i]) - (b_.i8[i]);
-		}
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_epi8(a, b) simde_mm_subs_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_subs_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_subs_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i16 = vqsubq_s16(a_.neon_i16, b_.neon_i16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
-		if (((b_.i16[i]) > 0 &&
-		 (a_.i16[i]) < INT16_MIN + (b_.i16[i]))) {
-			r_.i16[i] = INT16_MIN;
-		} else if ((b_.i16[i]) < 0 &&
-			 (a_.i16[i]) > INT16_MAX + (b_.i16[i])) {
-			r_.i16[i] = INT16_MAX;
-		} else {
-			r_.i16[i] = (a_.i16[i]) - (b_.i16[i]);
-		}
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_epi16(a, b) simde_mm_subs_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_subs_epu8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_subs_epu8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u8 = vqsubq_u8(a_.neon_u8, b_.neon_u8);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i8[0])); i++) {
-		const int32_t x = a_.u8[i] - b_.u8[i];
-		if (x < 0) {
-			r_.u8[i] = 0;
-		} else if (x > UINT8_MAX) {
-			r_.u8[i] = UINT8_MAX;
-		} else {
-			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
-		}
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_epu8(a, b) simde_mm_subs_epu8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_subs_epu16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_subs_epu16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_u16 = vqsubq_u16(a_.neon_u16, b_.neon_u16);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
-		const int32_t x = a_.u16[i] - b_.u16[i];
-		if (x < 0) {
-			r_.u16[i] = 0;
-		} else if (x > UINT16_MAX) {
-			r_.u16[i] = UINT16_MAX;
-		} else {
-			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
-		}
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_subs_epu16(a, b) simde_mm_subs_epu16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomieq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomieq_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] == b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] == b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomieq_sd(a, b) simde_mm_ucomieq_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomige_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomige_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] >= b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] >= b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomige_sd(a, b) simde_mm_ucomige_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomigt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomigt_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] > b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] > b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomigt_sd(a, b) simde_mm_ucomigt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomile_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomile_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] <= b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] <= b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomile_sd(a, b) simde_mm_ucomile_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomilt_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomilt_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] < b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] < b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomilt_sd(a, b) simde_mm_ucomilt_sd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-int simde_mm_ucomineq_sd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_ucomineq_sd(a, b);
-#else
-	simde__m128d_private a_ = simde__m128d_to_private(a),
-			 b_ = simde__m128d_to_private(b);
-	int r;
-
-#if defined(SIMDE_HAVE_FENV_H)
-	fenv_t envp;
-	int x = feholdexcept(&envp);
-	r = a_.f64[0] != b_.f64[0];
-	if (HEDLEY_LIKELY(x == 0))
-		fesetenv(&envp);
-#else
-	r = a_.f64[0] != b_.f64[0];
-#endif
-
-	return r;
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_ucomineq_sd(a, b) simde_mm_ucomineq_sd(a, b)
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
-#endif
-
-#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
-HEDLEY_DIAGNOSTIC_POP
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_lfence(void)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_lfence();
-#else
-	simde_mm_sfence();
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_lfence() simde_mm_lfence()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-void simde_mm_mfence(void)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	_mm_mfence();
-#else
-	simde_mm_sfence();
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_mfence() simde_mm_mfence()
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpackhi_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpackhi_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int8x8_t a1 = vreinterpret_s8_s16(vget_high_s16(a_.neon_i16));
-	int8x8_t b1 = vreinterpret_s8_s16(vget_high_s16(b_.neon_i16));
-	int8x8x2_t result = vzip_s8(a1, b1);
-	r_.neon_i8 = vcombine_s8(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 16, a_.i8, b_.i8, 8, 24, 9, 25, 10, 26,
-				 11, 27, 12, 28, 13, 29, 14, 30, 15, 31);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i8[0])) / 2); i++) {
-		r_.i8[(i * 2)] =
-			a_.i8[i + ((sizeof(r_) / sizeof(r_.i8[0])) / 2)];
-		r_.i8[(i * 2) + 1] =
-			b_.i8[i + ((sizeof(r_) / sizeof(r_.i8[0])) / 2)];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_epi8(a, b) simde_mm_unpackhi_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpackhi_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpackhi_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int16x4_t a1 = vget_high_s16(a_.neon_i16);
-	int16x4_t b1 = vget_high_s16(b_.neon_i16);
-	int16x4x2_t result = vzip_s16(a1, b1);
-	r_.neon_i16 = vcombine_s16(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 4, 12, 5, 13, 6,
-				 14, 7, 15);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i16[0])) / 2); i++) {
-		r_.i16[(i * 2)] =
-			a_.i16[i + ((sizeof(r_) / sizeof(r_.i16[0])) / 2)];
-		r_.i16[(i * 2) + 1] =
-			b_.i16[i + ((sizeof(r_) / sizeof(r_.i16[0])) / 2)];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_epi16(a, b) simde_mm_unpackhi_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpackhi_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpackhi_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int32x2_t a1 = vget_high_s32(a_.neon_i32);
-	int32x2_t b1 = vget_high_s32(b_.neon_i32);
-	int32x2x2_t result = vzip_s32(a1, b1);
-	r_.neon_i32 = vcombine_s32(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 2, 6, 3, 7);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i32[0])) / 2); i++) {
-		r_.i32[(i * 2)] =
-			a_.i32[i + ((sizeof(r_) / sizeof(r_.i32[0])) / 2)];
-		r_.i32[(i * 2) + 1] =
-			b_.i32[i + ((sizeof(r_) / sizeof(r_.i32[0])) / 2)];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_epi32(a, b) simde_mm_unpackhi_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpackhi_epi64(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpackhi_epi64(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.i64, b_.i64, 1, 3);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i64[0])) / 2); i++) {
-		r_.i64[(i * 2)] =
-			a_.i64[i + ((sizeof(r_) / sizeof(r_.i64[0])) / 2)];
-		r_.i64[(i * 2) + 1] =
-			b_.i64[i + ((sizeof(r_) / sizeof(r_.i64[0])) / 2)];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_epi64(a, b) simde_mm_unpackhi_epi64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_unpackhi_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpackhi_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 1, 3);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.f64[0])) / 2); i++) {
-		r_.f64[(i * 2)] =
-			a_.f64[i + ((sizeof(r_) / sizeof(r_.f64[0])) / 2)];
-		r_.f64[(i * 2) + 1] =
-			b_.f64[i + ((sizeof(r_) / sizeof(r_.f64[0])) / 2)];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpackhi_pd(a, b) simde_mm_unpackhi_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpacklo_epi8(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpacklo_epi8(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int8x8_t a1 = vreinterpret_s8_s16(vget_low_s16(a_.neon_i16));
-	int8x8_t b1 = vreinterpret_s8_s16(vget_low_s16(b_.neon_i16));
-	int8x8x2_t result = vzip_s8(a1, b1);
-	r_.neon_i8 = vcombine_s8(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 16, a_.i8, b_.i8, 0, 16, 1, 17, 2, 18,
-				 3, 19, 4, 20, 5, 21, 6, 22, 7, 23);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i8[0])) / 2); i++) {
-		r_.i8[(i * 2)] = a_.i8[i];
-		r_.i8[(i * 2) + 1] = b_.i8[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_epi8(a, b) simde_mm_unpacklo_epi8(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpacklo_epi16(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpacklo_epi16(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int16x4_t a1 = vget_low_s16(a_.neon_i16);
-	int16x4_t b1 = vget_low_s16(b_.neon_i16);
-	int16x4x2_t result = vzip_s16(a1, b1);
-	r_.neon_i16 = vcombine_s16(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 0, 8, 1, 9, 2,
-				 10, 3, 11);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i16[0])) / 2); i++) {
-		r_.i16[(i * 2)] = a_.i16[i];
-		r_.i16[(i * 2) + 1] = b_.i16[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_epi16(a, b) simde_mm_unpacklo_epi16(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpacklo_epi32(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpacklo_epi32(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	int32x2_t a1 = vget_low_s32(a_.neon_i32);
-	int32x2_t b1 = vget_low_s32(b_.neon_i32);
-	int32x2x2_t result = vzip_s32(a1, b1);
-	r_.neon_i32 = vcombine_s32(result.val[0], result.val[1]);
-#elif defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 0, 4, 1, 5);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i32[0])) / 2); i++) {
-		r_.i32[(i * 2)] = a_.i32[i];
-		r_.i32[(i * 2) + 1] = b_.i32[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_epi32(a, b) simde_mm_unpacklo_epi32(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_unpacklo_epi64(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpacklo_epi64(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.i64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.i64, b_.i64, 0, 2);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i64[0])) / 2); i++) {
-		r_.i64[(i * 2)] = a_.i64[i];
-		r_.i64[(i * 2) + 1] = b_.i64[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_epi64(a, b) simde_mm_unpacklo_epi64(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_unpacklo_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_unpacklo_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_SHUFFLE_VECTOR_)
-	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 0, 2);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.f64[0])) / 2); i++) {
-		r_.f64[(i * 2)] = a_.f64[i];
-		r_.f64[(i * 2) + 1] = b_.f64[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_unpacklo_pd(a, b) simde_mm_unpacklo_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128d simde_mm_xor_pd(simde__m128d a, simde__m128d b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_xor_pd(a, b);
-#else
-	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
-				 b_ = simde__m128d_to_private(b);
-
-#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f ^ b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] ^ b_.i32f[i];
-	}
-#endif
-
-	return simde__m128d_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_xor_pd(a, b) simde_mm_xor_pd(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_mm_xor_si128(simde__m128i a, simde__m128i b)
-{
-#if defined(SIMDE_X86_SSE2_NATIVE)
-	return _mm_xor_si128(a, b);
-#else
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
-				 b_ = simde__m128i_to_private(b);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = veorq_s32(a_.neon_i32, b_.neon_i32);
-#elif defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-	r_.altivec_i32 = vec_xor(a_.altivec_i32, b_.altivec_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = a_.i32f ^ b_.i32f;
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = a_.i32f[i] ^ b_.i32f[i];
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-#endif
-}
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _mm_xor_si128(a, b) simde_mm_xor_si128(a, b)
-#endif
-
-SIMDE_FUNCTION_ATTRIBUTES
-simde__m128i simde_x_mm_not_si128(simde__m128i a)
-{
-	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
-
-#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-	r_.neon_i32 = vmvnq_s32(a_.neon_i32);
-#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
-	r_.i32f = ~(a_.i32f);
-#else
-	SIMDE_VECTORIZE
-	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
-		r_.i32f[i] = ~(a_.i32f[i]);
-	}
-#endif
-
-	return simde__m128i_from_private(r_);
-}
-
-#define SIMDE_MM_SHUFFLE2(x, y) (((x) << 1) | (y))
-#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
-#define _MM_SHUFFLE2(x, y) SIMDE_MM_SHUFFLE2(x, y)
-#endif
-
-SIMDE_END_DECLS_
-
-HEDLEY_DIAGNOSTIC_POP
-
-#endif /* !defined(SIMDE_X86_SSE2_H) */

obs-studio-26.1.0.tar.xz/libobs/util/sse2neon.h Deleted

@@ -1,4207 +0,0 @@
-#ifndef SSE2NEON_H
-#define SSE2NEON_H
-
-// This header file provides a simple API translation layer
-// between SSE intrinsics to their corresponding Arm/Aarch64 NEON versions
-//
-// This header file does not yet translate all of the SSE intrinsics.
-//
-// Contributors to this work are:
-// John W. Ratcliff <jratcliffscarab@gmail.com>
-// Brandon Rowlett <browlett@nvidia.com>
-// Ken Fast <kfast@gdeb.com>
-// Eric van Beurden <evanbeurden@nvidia.com>
-// Alexander Potylitsin <apotylitsin@nvidia.com>
-// Hasindu Gamaarachchi <hasindu2008@gmail.com>
-// Jim Huang <jserv@biilabs.io>
-// Mark Cheng <marktwtn@biilabs.io>
-// Malcolm James MacLeod <malcolm@gulden.com>
-// Devin Hussey (easyaspi314) <husseydevin@gmail.com>
-// Sebastian Pop <spop@amazon.com>
-// Developer Ecosystem Engineering <DeveloperEcosystemEngineering@apple.com>
-// Danila Kutenin <danilak@google.com>
-
-/*
- * sse2neon is freely redistributable under the MIT License.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to deal
- * in the Software without restriction, including without limitation the rights
- * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
- * copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in
- * all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#if defined(__GNUC__) || defined(__clang__)
-#pragma push_macro("FORCE_INLINE")
-#pragma push_macro("ALIGN_STRUCT")
-#define FORCE_INLINE static inline __attribute__((always_inline))
-#define ALIGN_STRUCT(x) __attribute__((aligned(x)))
-#else
-#error "Macro name collisions may happen with unsupported compiler."
-#ifdef FORCE_INLINE
-#undef FORCE_INLINE
-#endif
-#define FORCE_INLINE static inline
-#ifndef ALIGN_STRUCT
-#define ALIGN_STRUCT(x) __declspec(align(x))
-#endif
-#endif
-
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <arm_neon.h>
-
-/* "__has_builtin" can be used to query support for built-in functions
- * provided by gcc/clang and other compilers that support it.
- */
-#ifndef __has_builtin /* GCC prior to 10 or non-clang compilers */
-/* Compatibility with gcc <= 9 */
-#if __GNUC__ <= 9
-#define __has_builtin(x) HAS##x
-#define HAS__builtin_popcount 1
-#define HAS__builtin_popcountll 1
-#else
-#define __has_builtin(x) 0
-#endif
-#endif
-
-/**
- * MACRO for shuffle parameter for _mm_shuffle_ps().
- * Argument fp3 is a digit[0123] that represents the fp from argument "b"
- * of mm_shuffle_ps that will be placed in fp3 of result. fp2 is the same
- * for fp2 in result. fp1 is a digit[0123] that represents the fp from
- * argument "a" of mm_shuffle_ps that will be places in fp1 of result.
- * fp0 is the same for fp0 of result.
- */
-#define _MM_SHUFFLE(fp3, fp2, fp1, fp0) \
-	(((fp3) << 6) | ((fp2) << 4) | ((fp1) << 2) | ((fp0)))
-
-/* indicate immediate constant argument in a given range */
-#define __constrange(a, b) const
-
-/* A few intrinsics accept traditional data types like ints or floats, but
- * most operate on data types that are specific to SSE.
- * If a vector type ends in d, it contains doubles, and if it does not have
- * a suffix, it contains floats. An integer vector type can contain any type
- * of integer, from chars to shorts to unsigned long longs.
- */
-typedef float32x2_t __m64;
-typedef float32x4_t __m128; /* 128-bit vector containing 4 floats */
-// On ARM 32-bit architecture, the float64x2_t is not supported.
-// The data type __m128d should be represented in a different way for related
-// intrinsic conversion.
-#if defined(__aarch64__)
-typedef float64x2_t __m128d; /* 128-bit vector containing 2 doubles */
-#else
-typedef float32x4_t __m128d;
-#endif
-typedef int64x1_t __m64i;
-typedef int64x2_t __m128i; /* 128-bit vector containing integers */
-
-/* type-safe casting between types */
-
-#define vreinterpretq_m128_f16(x) vreinterpretq_f32_f16(x)
-#define vreinterpretq_m128_f32(x) (x)
-#define vreinterpretq_m128_f64(x) vreinterpretq_f32_f64(x)
-
-#define vreinterpretq_m128_u8(x) vreinterpretq_f32_u8(x)
-#define vreinterpretq_m128_u16(x) vreinterpretq_f32_u16(x)
-#define vreinterpretq_m128_u32(x) vreinterpretq_f32_u32(x)
-#define vreinterpretq_m128_u64(x) vreinterpretq_f32_u64(x)
-
-#define vreinterpretq_m128_s8(x) vreinterpretq_f32_s8(x)
-#define vreinterpretq_m128_s16(x) vreinterpretq_f32_s16(x)
-#define vreinterpretq_m128_s32(x) vreinterpretq_f32_s32(x)
-#define vreinterpretq_m128_s64(x) vreinterpretq_f32_s64(x)
-
-#define vreinterpretq_f16_m128(x) vreinterpretq_f16_f32(x)
-#define vreinterpretq_f32_m128(x) (x)
-#define vreinterpretq_f64_m128(x) vreinterpretq_f64_f32(x)
-
-#define vreinterpretq_u8_m128(x) vreinterpretq_u8_f32(x)
-#define vreinterpretq_u16_m128(x) vreinterpretq_u16_f32(x)
-#define vreinterpretq_u32_m128(x) vreinterpretq_u32_f32(x)
-#define vreinterpretq_u64_m128(x) vreinterpretq_u64_f32(x)
-
-#define vreinterpretq_s8_m128(x) vreinterpretq_s8_f32(x)
-#define vreinterpretq_s16_m128(x) vreinterpretq_s16_f32(x)
-#define vreinterpretq_s32_m128(x) vreinterpretq_s32_f32(x)
-#define vreinterpretq_s64_m128(x) vreinterpretq_s64_f32(x)
-
-#define vreinterpretq_m128i_s8(x) vreinterpretq_s64_s8(x)
-#define vreinterpretq_m128i_s16(x) vreinterpretq_s64_s16(x)
-#define vreinterpretq_m128i_s32(x) vreinterpretq_s64_s32(x)
-#define vreinterpretq_m128i_s64(x) (x)
-
-#define vreinterpretq_m128i_u8(x) vreinterpretq_s64_u8(x)
-#define vreinterpretq_m128i_u16(x) vreinterpretq_s64_u16(x)
-#define vreinterpretq_m128i_u32(x) vreinterpretq_s64_u32(x)
-#define vreinterpretq_m128i_u64(x) vreinterpretq_s64_u64(x)
-
-#define vreinterpretq_s8_m128i(x) vreinterpretq_s8_s64(x)
-#define vreinterpretq_s16_m128i(x) vreinterpretq_s16_s64(x)
-#define vreinterpretq_s32_m128i(x) vreinterpretq_s32_s64(x)
-#define vreinterpretq_s64_m128i(x) (x)
-
-#define vreinterpretq_u8_m128i(x) vreinterpretq_u8_s64(x)
-#define vreinterpretq_u16_m128i(x) vreinterpretq_u16_s64(x)
-#define vreinterpretq_u32_m128i(x) vreinterpretq_u32_s64(x)
-#define vreinterpretq_u64_m128i(x) vreinterpretq_u64_s64(x)
-
-#define vreinterpret_m64i_s8(x) vreinterpret_s64_s8(x)
-#define vreinterpret_m64i_s16(x) vreinterpret_s64_s16(x)
-#define vreinterpret_m64i_s32(x) vreinterpret_s64_s32(x)
-#define vreinterpret_m64i_s64(x) (x)
-
-#define vreinterpret_m64i_u8(x) vreinterpret_s64_u8(x)
-#define vreinterpret_m64i_u16(x) vreinterpret_s64_u16(x)
-#define vreinterpret_m64i_u32(x) vreinterpret_s64_u32(x)
-#define vreinterpret_m64i_u64(x) vreinterpret_s64_u64(x)
-
-#define vreinterpret_u8_m64i(x) vreinterpret_u8_s64(x)
-#define vreinterpret_u16_m64i(x) vreinterpret_u16_s64(x)
-#define vreinterpret_u32_m64i(x) vreinterpret_u32_s64(x)
-#define vreinterpret_u64_m64i(x) vreinterpret_u64_s64(x)
-
-#define vreinterpret_s8_m64i(x) vreinterpret_s8_s64(x)
-#define vreinterpret_s16_m64i(x) vreinterpret_s16_s64(x)
-#define vreinterpret_s32_m64i(x) vreinterpret_s32_s64(x)
-#define vreinterpret_s64_m64i(x) (x)
-
-// A struct is defined in this header file called 'SIMDVec' which can be used
-// by applications which attempt to access the contents of an _m128 struct
-// directly. It is important to note that accessing the __m128 struct directly
-// is bad coding practice by Microsoft: @see:
-// https://msdn.microsoft.com/en-us/library/ayeb3ayc.aspx
-//
-// However, some legacy source code may try to access the contents of an __m128
-// struct directly so the developer can use the SIMDVec as an alias for it. Any
-// casting must be done manually by the developer, as you cannot cast or
-// otherwise alias the base NEON data type for intrinsic operations.
-//
-// union intended to allow direct access to an __m128 variable using the names
-// that the MSVC compiler provides. This union should really only be used when
-// trying to access the members of the vector as integer values. GCC/clang
-// allow native access to the float members through a simple array access
-// operator (in C since 4.6, in C++ since 4.8).
-//
-// Ideally direct accesses to SIMD vectors should not be used since it can cause
-// a performance hit. If it really is needed however, the original __m128
-// variable can be aliased with a pointer to this union and used to access
-// individual components. The use of this union should be hidden behind a macro
-// that is used throughout the codebase to access the members instead of always
-// declaring this type of variable.
-typedef union ALIGN_STRUCT(16) SIMDVec {
-	float m128_f32[4]; // as floats - DON'T USE. Added for convenience.
-	int8_t m128_i8[16]; // as signed 8-bit integers.
-	int16_t m128_i16[8]; // as signed 16-bit integers.
-	int32_t m128_i32[4]; // as signed 32-bit integers.
-	int64_t m128_i64[2]; // as signed 64-bit integers.
-	uint8_t m128_u8[16]; // as unsigned 8-bit integers.
-	uint16_t m128_u16[8]; // as unsigned 16-bit integers.
-	uint32_t m128_u32[4]; // as unsigned 32-bit integers.
-	uint64_t m128_u64[2]; // as unsigned 64-bit integers.
-} SIMDVec;
-
-// casting using SIMDVec
-#define vreinterpretq_nth_u64_m128i(x, n) (((SIMDVec *)&x)->m128_u64[n])
-#define vreinterpretq_nth_u32_m128i(x, n) (((SIMDVec *)&x)->m128_u32[n])
-
-/* Backwards compatibility for compilers with lack of specific type support */
-
-// Older gcc does not define vld1q_u8_x4 type
-#if defined(__GNUC__) && !defined(__clang__)
-#if __GNUC__ <= 9
-FORCE_INLINE uint8x16x4_t vld1q_u8_x4(const uint8_t *p)
-{
-	uint8x16x4_t ret;
-	ret.val[0] = vld1q_u8(p + 0);
-	ret.val[1] = vld1q_u8(p + 16);
-	ret.val[2] = vld1q_u8(p + 32);
-	ret.val[3] = vld1q_u8(p + 48);
-	return ret;
-}
-#endif
-#endif
-
-/* Function Naming Conventions
- * The naming convention of SSE intrinsics is straightforward. A generic SSE
- * intrinsic function is given as follows:
- * _mm_<name>_<data_type>
- *
- * The parts of this format are given as follows:
- * 1. <name> describes the operation performed by the intrinsic
- * 2. <data_type> identifies the data type of the function's primary arguments
- *
- * This last part, <data_type>, is a little complicated. It identifies the
- * content of the input values, and can be set to any of the following values:
- * + ps - vectors contain floats (ps stands for packed single-precision)
- * + pd - vectors cantain doubles (pd stands for packed double-precision)
- * + epi8/epi16/epi32/epi64 - vectors contain 8-bit/16-bit/32-bit/64-bit
- * signed integers
- * + epu8/epu16/epu32/epu64 - vectors contain 8-bit/16-bit/32-bit/64-bit
- * unsigned integers
- * + si128 - unspecified 128-bit vector or 256-bit vector
- * + m128/m128i/m128d - identifies input vector types when they are different
- * than the type of the returned vector
- *
- * For example, _mm_setzero_ps. The _mm implies that the function returns
- * a 128-bit vector. The _ps at the end implies that the argument vectors
- * contain floats.
- *
- * A complete example: Byte Shuffle - pshufb (_mm_shuffle_epi8)
- * // Set packed 16-bit integers. 128 bits, 8 short, per 16 bits
- * __m128i v_in = _mm_setr_epi16(1, 2, 3, 4, 5, 6, 7, 8);
- * // Set packed 8-bit integers
- * // 128 bits, 16 chars, per 8 bits
- * __m128i v_perm = _mm_setr_epi8(1, 0, 2, 3, 8, 9, 10, 11,
- * 4, 5, 12, 13, 6, 7, 14, 15);
- * // Shuffle packed 8-bit integers
- * __m128i v_out = _mm_shuffle_epi8(v_in, v_perm); // pshufb
- *
- * Data (Number, Binary, Byte Index):
- +------+------+-------------+------+------+-------------+
- | 1 | 2 | 3 | 4 | Number
- +------+------+------+------+------+------+------+------+
- | 0000 | 0001 | 0000 | 0010 | 0000 | 0011 | 0000 | 0100 | Binary
- +------+------+------+------+------+------+------+------+
- | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Index
- +------+------+------+------+------+------+------+------+
-
- +------+------+------+------+------+------+------+------+
- | 5 | 6 | 7 | 8 | Number
- +------+------+------+------+------+------+------+------+
- | 0000 | 0101 | 0000 | 0110 | 0000 | 0111 | 0000 | 1000 | Binary
- +------+------+------+------+------+------+------+------+
- | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | Index
- +------+------+------+------+------+------+------+------+
- * Index (Byte Index):
- +------+------+------+------+------+------+------+------+
- | 1 | 0 | 2 | 3 | 8 | 9 | 10 | 11 |
- +------+------+------+------+------+------+------+------+
-
- +------+------+------+------+------+------+------+------+
- | 4 | 5 | 12 | 13 | 6 | 7 | 14 | 15 |
- +------+------+------+------+------+------+------+------+
- * Result:
- +------+------+------+------+------+------+------+------+
- | 1 | 0 | 2 | 3 | 8 | 9 | 10 | 11 | Index
- +------+------+------+------+------+------+------+------+
- | 0001 | 0000 | 0000 | 0010 | 0000 | 0101 | 0000 | 0110 | Binary
- +------+------+------+------+------+------+------+------+
- | 256 | 2 | 5 | 6 | Number
- +------+------+------+------+------+------+------+------+
-
- +------+------+------+------+------+------+------+------+
- | 4 | 5 | 12 | 13 | 6 | 7 | 14 | 15 | Index
- +------+------+------+------+------+------+------+------+
- | 0000 | 0011 | 0000 | 0111 | 0000 | 0100 | 0000 | 1000 | Binary
- +------+------+------+------+------+------+------+------+
- | 3 | 7 | 4 | 8 | Number
- +------+------+------+------+------+------+-------------+
- */
-
-/* Set/get methods */
-
-/* Constants for use with _mm_prefetch. */
-enum _mm_hint {
-	_MM_HINT_NTA = 0, /* load data to L1 and L2 cache, mark it as NTA */
-	_MM_HINT_T0 = 1, /* load data to L1 and L2 cache */
-	_MM_HINT_T1 = 2, /* load data to L2 cache only */
-	_MM_HINT_T2 = 3, /* load data to L2 cache only, mark it as NTA */
-	_MM_HINT_ENTA = 4, /* exclusive version of _MM_HINT_NTA */
-	_MM_HINT_ET0 = 5, /* exclusive version of _MM_HINT_T0 */
-	_MM_HINT_ET1 = 6, /* exclusive version of _MM_HINT_T1 */
-	_MM_HINT_ET2 = 7 /* exclusive version of _MM_HINT_T2 */
-};
-
-// Loads one cache line of data from address p to a location closer to the
-// processor. https://msdn.microsoft.com/en-us/library/84szxsww(v=vs.100).aspx
-FORCE_INLINE void _mm_prefetch(const void *p, int i)
-{
-	(void)i;
-	__builtin_prefetch(p);
-}
-
-// extracts the lower order floating point value from the parameter :
-// https://msdn.microsoft.com/en-us/library/bb514059%28v=vs.120%29.aspx?f=255&MSPPError=-2147217396
-FORCE_INLINE float _mm_cvtss_f32(__m128 a)
-{
-	return vgetq_lane_f32(vreinterpretq_f32_m128(a), 0);
-}
-
-// Sets the 128-bit value to zero
-// https://msdn.microsoft.com/en-us/library/vstudio/ys7dw0kh(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_setzero_si128(void)
-{
-	return vreinterpretq_m128i_s32(vdupq_n_s32(0));
-}
-
-// Clears the four single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/vstudio/tk1t2tbz(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_setzero_ps(void)
-{
-	return vreinterpretq_m128_f32(vdupq_n_f32(0));
-}
-
-// Sets the four single-precision, floating-point values to w.
-//
-// r0 := r1 := r2 := r3 := w
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/2x1se8ha(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_set1_ps(float _w)
-{
-	return vreinterpretq_m128_f32(vdupq_n_f32(_w));
-}
-
-// Sets the four single-precision, floating-point values to w.
-// https://msdn.microsoft.com/en-us/library/vstudio/2x1se8ha(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_set_ps1(float _w)
-{
-	return vreinterpretq_m128_f32(vdupq_n_f32(_w));
-}
-
-// Sets the four single-precision, floating-point values to the four inputs.
-// https://msdn.microsoft.com/en-us/library/vstudio/afh0zf75(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_set_ps(float w, float z, float y, float x)
-{
-	float ALIGN_STRUCT(16) data[4] = {x, y, z, w};
-	return vreinterpretq_m128_f32(vld1q_f32(data));
-}
-
-// Copy single-precision (32-bit) floating-point element a to the lower element
-// of dst, and zero the upper 3 elements.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_set_ss&expand=4901,4895,4901
-FORCE_INLINE __m128 _mm_set_ss(float a)
-{
-	float ALIGN_STRUCT(16) data[4] = {a, 0, 0, 0};
-	return vreinterpretq_m128_f32(vld1q_f32(data));
-}
-
-// Sets the four single-precision, floating-point values to the four inputs in
-// reverse order.
-// https://msdn.microsoft.com/en-us/library/vstudio/d2172ct3(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_setr_ps(float w, float z, float y, float x)
-{
-	float ALIGN_STRUCT(16) data[4] = {w, z, y, x};
-	return vreinterpretq_m128_f32(vld1q_f32(data));
-}
-
-// Sets the 8 signed 16-bit integer values in reverse order.
-//
-// Return Value
-// r0 := w0
-// r1 := w1
-// ...
-// r7 := w7
-FORCE_INLINE __m128i _mm_setr_epi16(short w0, short w1, short w2, short w3,
-				 short w4, short w5, short w6, short w7)
-{
-	int16_t ALIGN_STRUCT(16) data[8] = {w0, w1, w2, w3, w4, w5, w6, w7};
-	return vreinterpretq_m128i_s16(vld1q_s16((int16_t *)data));
-}
-
-// Sets the 4 signed 32-bit integer values in reverse order
-// https://technet.microsoft.com/en-us/library/security/27yb3ee5(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_setr_epi32(int i3, int i2, int i1, int i0)
-{
-	int32_t ALIGN_STRUCT(16) data[4] = {i3, i2, i1, i0};
-	return vreinterpretq_m128i_s32(vld1q_s32(data));
-}
-
-// Sets the 16 signed 8-bit integer values to b.
-//
-// r0 := b
-// r1 := b
-// ...
-// r15 := b
-//
-// https://msdn.microsoft.com/en-us/library/6e14xhyf(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_set1_epi8(signed char w)
-{
-	return vreinterpretq_m128i_s8(vdupq_n_s8(w));
-}
-
-// Sets the 8 signed 16-bit integer values to w.
-//
-// r0 := w
-// r1 := w
-// ...
-// r7 := w
-//
-// https://msdn.microsoft.com/en-us/library/k0ya3x0e(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_set1_epi16(short w)
-{
-	return vreinterpretq_m128i_s16(vdupq_n_s16(w));
-}
-
-// Sets the 16 signed 8-bit integer values.
-// https://msdn.microsoft.com/en-us/library/x0cx8zd3(v=vs.90).aspx
-FORCE_INLINE __m128i
-_mm_set_epi8(signed char b15, signed char b14, signed char b13, signed char b12,
-	 signed char b11, signed char b10, signed char b9, signed char b8,
-	 signed char b7, signed char b6, signed char b5, signed char b4,
-	 signed char b3, signed char b2, signed char b1, signed char b0)
-{
-	int8_t ALIGN_STRUCT(16)
-		data[16] = {(int8_t)b0, (int8_t)b1, (int8_t)b2, (int8_t)b3,
-			 (int8_t)b4, (int8_t)b5, (int8_t)b6, (int8_t)b7,
-			 (int8_t)b8, (int8_t)b9, (int8_t)b10, (int8_t)b11,
-			 (int8_t)b12, (int8_t)b13, (int8_t)b14, (int8_t)b15};
-	return (__m128i)vld1q_s8(data);
-}
-
-// Sets the 8 signed 16-bit integer values.
-// https://msdn.microsoft.com/en-au/library/3e0fek84(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_set_epi16(short i7, short i6, short i5, short i4,
-				 short i3, short i2, short i1, short i0)
-{
-	int16_t ALIGN_STRUCT(16) data[8] = {i0, i1, i2, i3, i4, i5, i6, i7};
-	return vreinterpretq_m128i_s16(vld1q_s16(data));
-}
-
-// Sets the 16 signed 8-bit integer values in reverse order.
-// https://msdn.microsoft.com/en-us/library/2khb9c7k(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_setr_epi8(
-	signed char b0, signed char b1, signed char b2, signed char b3,
-	signed char b4, signed char b5, signed char b6, signed char b7,
-	signed char b8, signed char b9, signed char b10, signed char b11,
-	signed char b12, signed char b13, signed char b14, signed char b15)
-{
-	int8_t ALIGN_STRUCT(16)
-		data[16] = {(int8_t)b0, (int8_t)b1, (int8_t)b2, (int8_t)b3,
-			 (int8_t)b4, (int8_t)b5, (int8_t)b6, (int8_t)b7,
-			 (int8_t)b8, (int8_t)b9, (int8_t)b10, (int8_t)b11,
-			 (int8_t)b12, (int8_t)b13, (int8_t)b14, (int8_t)b15};
-	return (__m128i)vld1q_s8(data);
-}
-
-// Sets the 4 signed 32-bit integer values to i.
-//
-// r0 := i
-// r1 := i
-// r2 := i
-// r3 := I
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/h4xscxat(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_set1_epi32(int _i)
-{
-	return vreinterpretq_m128i_s32(vdupq_n_s32(_i));
-}
-
-// Sets the 2 signed 64-bit integer values to i.
-// https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2010/whtfzhzk(v=vs.100)
-FORCE_INLINE __m128i _mm_set1_epi64(int64_t _i)
-{
-	return vreinterpretq_m128i_s64(vdupq_n_s64(_i));
-}
-
-// Sets the 2 signed 64-bit integer values to i.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_set1_epi64x&expand=4961
-FORCE_INLINE __m128i _mm_set1_epi64x(int64_t _i)
-{
-	return vreinterpretq_m128i_s64(vdupq_n_s64(_i));
-}
-
-// Sets the 4 signed 32-bit integer values.
-// https://msdn.microsoft.com/en-us/library/vstudio/019beekt(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_set_epi32(int i3, int i2, int i1, int i0)
-{
-	int32_t ALIGN_STRUCT(16) data[4] = {i0, i1, i2, i3};
-	return vreinterpretq_m128i_s32(vld1q_s32(data));
-}
-
-// Returns the __m128i structure with its two 64-bit integer values
-// initialized to the values of the two 64-bit integers passed in.
-// https://msdn.microsoft.com/en-us/library/dk2sdw0h(v=vs.120).aspx
-FORCE_INLINE __m128i _mm_set_epi64x(int64_t i1, int64_t i2)
-{
-	int64_t ALIGN_STRUCT(16) data[2] = {i2, i1};
-	return vreinterpretq_m128i_s64(vld1q_s64(data));
-}
-
-// Stores four single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/vstudio/s3h4ay6y(v=vs.100).aspx
-FORCE_INLINE void _mm_store_ps(float *p, __m128 a)
-{
-	vst1q_f32(p, vreinterpretq_f32_m128(a));
-}
-
-// Stores four single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/44e30x22(v=vs.100).aspx
-FORCE_INLINE void _mm_storeu_ps(float *p, __m128 a)
-{
-	vst1q_f32(p, vreinterpretq_f32_m128(a));
-}
-
-// Stores four 32-bit integer values as (as a __m128i value) at the address p.
-// https://msdn.microsoft.com/en-us/library/vstudio/edk11s13(v=vs.100).aspx
-FORCE_INLINE void _mm_store_si128(__m128i *p, __m128i a)
-{
-	vst1q_s32((int32_t *)p, vreinterpretq_s32_m128i(a));
-}
-
-// Stores four 32-bit integer values as (as a __m128i value) at the address p.
-// https://msdn.microsoft.com/en-us/library/vstudio/edk11s13(v=vs.100).aspx
-FORCE_INLINE void _mm_storeu_si128(__m128i *p, __m128i a)
-{
-	vst1q_s32((int32_t *)p, vreinterpretq_s32_m128i(a));
-}
-
-// Stores the lower single - precision, floating - point value.
-// https://msdn.microsoft.com/en-us/library/tzz10fbx(v=vs.100).aspx
-FORCE_INLINE void _mm_store_ss(float *p, __m128 a)
-{
-	vst1q_lane_f32(p, vreinterpretq_f32_m128(a), 0);
-}
-
-// Reads the lower 64 bits of b and stores them into the lower 64 bits of a.
-// https://msdn.microsoft.com/en-us/library/hhwf428f%28v=vs.90%29.aspx
-FORCE_INLINE void _mm_storel_epi64(__m128i *a, __m128i b)
-{
-	uint64x1_t hi = vget_high_u64(vreinterpretq_u64_m128i(*a));
-	uint64x1_t lo = vget_low_u64(vreinterpretq_u64_m128i(b));
-	*a = vreinterpretq_m128i_u64(vcombine_u64(lo, hi));
-}
-
-// Stores the lower two single-precision floating point values of a to the
-// address p.
-//
-// *p0 := a0
-// *p1 := a1
-//
-// https://msdn.microsoft.com/en-us/library/h54t98ks(v=vs.90).aspx
-FORCE_INLINE void _mm_storel_pi(__m64 *p, __m128 a)
-{
-	*p = vget_low_f32(a);
-}
-
-// Stores the upper two single-precision, floating-point values of a to the
-// address p.
-//
-// *p0 := a2
-// *p1 := a3
-//
-// https://msdn.microsoft.com/en-us/library/a7525fs8(v%3dvs.90).aspx
-FORCE_INLINE void _mm_storeh_pi(__m64 *p, __m128 a)
-{
-	*p = vget_high_f32(a);
-}
-
-// Loads a single single-precision, floating-point value, copying it into all
-// four words
-// https://msdn.microsoft.com/en-us/library/vstudio/5cdkf716(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_load1_ps(const float *p)
-{
-	return vreinterpretq_m128_f32(vld1q_dup_f32(p));
-}
-#define _mm_load_ps1 _mm_load1_ps
-
-// Sets the lower two single-precision, floating-point values with 64
-// bits of data loaded from the address p; the upper two values are passed
-// through from a.
-//
-// Return Value
-// r0 := *p0
-// r1 := *p1
-// r2 := a2
-// r3 := a3
-//
-// https://msdn.microsoft.com/en-us/library/s57cyak2(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_loadl_pi(__m128 a, __m64 const *p)
-{
-	return vreinterpretq_m128_f32(
-		vcombine_f32(vld1_f32((const float32_t *)p), vget_high_f32(a)));
-}
-
-// Sets the upper two single-precision, floating-point values with 64
-// bits of data loaded from the address p; the lower two values are passed
-// through from a.
-//
-// r0 := a0
-// r1 := a1
-// r2 := *p0
-// r3 := *p1
-//
-// https://msdn.microsoft.com/en-us/library/w92wta0x(v%3dvs.100).aspx
-FORCE_INLINE __m128 _mm_loadh_pi(__m128 a, __m64 const *p)
-{
-	return vreinterpretq_m128_f32(
-		vcombine_f32(vget_low_f32(a), vld1_f32((const float32_t *)p)));
-}
-
-// Loads four single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/vstudio/zzd50xxt(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_load_ps(const float *p)
-{
-	return vreinterpretq_m128_f32(vld1q_f32(p));
-}
-
-// Loads four single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/x1b16s7z%28v=vs.90%29.aspx
-FORCE_INLINE __m128 _mm_loadu_ps(const float *p)
-{
-	// for neon, alignment doesn't matter, so _mm_load_ps and _mm_loadu_ps are
-	// equivalent for neon
-	return vreinterpretq_m128_f32(vld1q_f32(p));
-}
-
-// Loads a double-precision, floating-point value.
-// The upper double-precision, floating-point is set to zero. The address p does
-// not need to be 16-byte aligned.
-// https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2010/574w9fdd(v%3dvs.100)
-FORCE_INLINE __m128d _mm_load_sd(const double *p)
-{
-#if defined(__aarch64__)
-	return vsetq_lane_f64(*p, vdupq_n_f64(0), 0);
-#else
-	const float *fp = (const float *)p;
-	float ALIGN_STRUCT(16) data[4] = {fp[0], fp[1], 0, 0};
-	return vld1q_f32(data);
-#endif
-}
-
-// Loads an single - precision, floating - point value into the low word and
-// clears the upper three words.
-// https://msdn.microsoft.com/en-us/library/548bb9h4%28v=vs.90%29.aspx
-FORCE_INLINE __m128 _mm_load_ss(const float *p)
-{
-	return vreinterpretq_m128_f32(vsetq_lane_f32(*p, vdupq_n_f32(0), 0));
-}
-
-FORCE_INLINE __m128i _mm_loadl_epi64(__m128i const *p)
-{
-	/* Load the lower 64 bits of the value pointed to by p into the
- * lower 64 bits of the result, zeroing the upper 64 bits of the result.
- */
-	return vreinterpretq_m128i_s32(
-		vcombine_s32(vld1_s32((int32_t const *)p), vcreate_s32(0)));
-}
-
-/* Logic/Binary operations */
-
-// Compares for inequality.
-// https://msdn.microsoft.com/en-us/library/sf44thbx(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmpneq_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vmvnq_u32(vceqq_f32(
-		vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b))));
-}
-
-// Computes the bitwise AND-NOT of the four single-precision, floating-point
-// values of a and b.
-//
-// r0 := ~a0 & b0
-// r1 := ~a1 & b1
-// r2 := ~a2 & b2
-// r3 := ~a3 & b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/68h7wd02(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_andnot_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_s32(
-		vbicq_s32(vreinterpretq_s32_m128(b),
-			 vreinterpretq_s32_m128(a))); // *NOTE* argument swap
-}
-
-// Computes the bitwise AND of the 128-bit value in b and the bitwise NOT of the
-// 128-bit value in a.
-//
-// r := (~a) & b
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/1beaceh8(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_andnot_si128(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(
-		vbicq_s32(vreinterpretq_s32_m128i(b),
-			 vreinterpretq_s32_m128i(a))); // *NOTE* argument swap
-}
-
-// Computes the bitwise AND of the 128-bit value in a and the 128-bit value in
-// b.
-//
-// r := a & b
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/6d1txsa8(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_and_si128(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vandq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Computes the bitwise AND of the four single-precision, floating-point values
-// of a and b.
-//
-// r0 := a0 & b0
-// r1 := a1 & b1
-// r2 := a2 & b2
-// r3 := a3 & b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/73ck1xc5(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_and_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_s32(vandq_s32(vreinterpretq_s32_m128(a),
-						vreinterpretq_s32_m128(b)));
-}
-
-// Computes the bitwise OR of the four single-precision, floating-point values
-// of a and b.
-// https://msdn.microsoft.com/en-us/library/vstudio/7ctdsyy0(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_or_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_s32(vorrq_s32(vreinterpretq_s32_m128(a),
-						vreinterpretq_s32_m128(b)));
-}
-
-// Computes bitwise EXOR (exclusive-or) of the four single-precision,
-// floating-point values of a and b.
-// https://msdn.microsoft.com/en-us/library/ss6k3wk8(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_xor_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_s32(veorq_s32(vreinterpretq_s32_m128(a),
-						vreinterpretq_s32_m128(b)));
-}
-
-// Computes the bitwise OR of the 128-bit value in a and the 128-bit value in b.
-//
-// r := a | b
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/ew8ty0db(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_or_si128(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vorrq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Computes the bitwise XOR of the 128-bit value in a and the 128-bit value in
-// b. https://msdn.microsoft.com/en-us/library/fzt08www(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_xor_si128(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(veorq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Moves the upper two values of B into the lower two values of A.
-//
-// r3 := a3
-// r2 := a2
-// r1 := b3
-// r0 := b2
-FORCE_INLINE __m128 _mm_movehl_ps(__m128 __A, __m128 __B)
-{
-	float32x2_t a32 = vget_high_f32(vreinterpretq_f32_m128(__A));
-	float32x2_t b32 = vget_high_f32(vreinterpretq_f32_m128(__B));
-	return vreinterpretq_m128_f32(vcombine_f32(b32, a32));
-}
-
-// Moves the lower two values of B into the upper two values of A.
-//
-// r3 := b1
-// r2 := b0
-// r1 := a1
-// r0 := a0
-FORCE_INLINE __m128 _mm_movelh_ps(__m128 __A, __m128 __B)
-{
-	float32x2_t a10 = vget_low_f32(vreinterpretq_f32_m128(__A));
-	float32x2_t b10 = vget_low_f32(vreinterpretq_f32_m128(__B));
-	return vreinterpretq_m128_f32(vcombine_f32(a10, b10));
-}
-
-FORCE_INLINE __m128i _mm_abs_epi32(__m128i a)
-{
-	return vreinterpretq_m128i_s32(vabsq_s32(vreinterpretq_s32_m128i(a)));
-}
-
-FORCE_INLINE __m128i _mm_abs_epi16(__m128i a)
-{
-	return vreinterpretq_m128i_s16(vabsq_s16(vreinterpretq_s16_m128i(a)));
-}
-
-FORCE_INLINE __m128i _mm_abs_epi8(__m128i a)
-{
-	return vreinterpretq_m128i_s8(vabsq_s8(vreinterpretq_s8_m128i(a)));
-}
-
-// Takes the upper 64 bits of a and places it in the low end of the result
-// Takes the lower 64 bits of b and places it into the high end of the result.
-FORCE_INLINE __m128 _mm_shuffle_ps_1032(__m128 a, __m128 b)
-{
-	float32x2_t a32 = vget_high_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b10 = vget_low_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(vcombine_f32(a32, b10));
-}
-
-// takes the lower two 32-bit values from a and swaps them and places in high
-// end of result takes the higher two 32 bit values from b and swaps them and
-// places in low end of result.
-FORCE_INLINE __m128 _mm_shuffle_ps_2301(__m128 a, __m128 b)
-{
-	float32x2_t a01 = vrev64_f32(vget_low_f32(vreinterpretq_f32_m128(a)));
-	float32x2_t b23 = vrev64_f32(vget_high_f32(vreinterpretq_f32_m128(b)));
-	return vreinterpretq_m128_f32(vcombine_f32(a01, b23));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_0321(__m128 a, __m128 b)
-{
-	float32x2_t a21 = vget_high_f32(vextq_f32(
-		vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a), 3));
-	float32x2_t b03 = vget_low_f32(vextq_f32(vreinterpretq_f32_m128(b),
-						 vreinterpretq_f32_m128(b), 3));
-	return vreinterpretq_m128_f32(vcombine_f32(a21, b03));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_2103(__m128 a, __m128 b)
-{
-	float32x2_t a03 = vget_low_f32(vextq_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(a), 3));
-	float32x2_t b21 = vget_high_f32(vextq_f32(
-		vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b), 3));
-	return vreinterpretq_m128_f32(vcombine_f32(a03, b21));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_1010(__m128 a, __m128 b)
-{
-	float32x2_t a10 = vget_low_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b10 = vget_low_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(vcombine_f32(a10, b10));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_1001(__m128 a, __m128 b)
-{
-	float32x2_t a01 = vrev64_f32(vget_low_f32(vreinterpretq_f32_m128(a)));
-	float32x2_t b10 = vget_low_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(vcombine_f32(a01, b10));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_0101(__m128 a, __m128 b)
-{
-	float32x2_t a01 = vrev64_f32(vget_low_f32(vreinterpretq_f32_m128(a)));
-	float32x2_t b01 = vrev64_f32(vget_low_f32(vreinterpretq_f32_m128(b)));
-	return vreinterpretq_m128_f32(vcombine_f32(a01, b01));
-}
-
-// keeps the low 64 bits of b in the low and puts the high 64 bits of a in the
-// high
-FORCE_INLINE __m128 _mm_shuffle_ps_3210(__m128 a, __m128 b)
-{
-	float32x2_t a10 = vget_low_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b32 = vget_high_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(vcombine_f32(a10, b32));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_0011(__m128 a, __m128 b)
-{
-	float32x2_t a11 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(a)), 1);
-	float32x2_t b00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 0);
-	return vreinterpretq_m128_f32(vcombine_f32(a11, b00));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_0022(__m128 a, __m128 b)
-{
-	float32x2_t a22 =
-		vdup_lane_f32(vget_high_f32(vreinterpretq_f32_m128(a)), 0);
-	float32x2_t b00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 0);
-	return vreinterpretq_m128_f32(vcombine_f32(a22, b00));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_2200(__m128 a, __m128 b)
-{
-	float32x2_t a00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(a)), 0);
-	float32x2_t b22 =
-		vdup_lane_f32(vget_high_f32(vreinterpretq_f32_m128(b)), 0);
-	return vreinterpretq_m128_f32(vcombine_f32(a00, b22));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_3202(__m128 a, __m128 b)
-{
-	float32_t a0 = vgetq_lane_f32(vreinterpretq_f32_m128(a), 0);
-	float32x2_t a22 =
-		vdup_lane_f32(vget_high_f32(vreinterpretq_f32_m128(a)), 0);
-	float32x2_t a02 = vset_lane_f32(a0, a22, 1); /* TODO: use vzip ?*/
-	float32x2_t b32 = vget_high_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(vcombine_f32(a02, b32));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_1133(__m128 a, __m128 b)
-{
-	float32x2_t a33 =
-		vdup_lane_f32(vget_high_f32(vreinterpretq_f32_m128(a)), 1);
-	float32x2_t b11 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 1);
-	return vreinterpretq_m128_f32(vcombine_f32(a33, b11));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_2010(__m128 a, __m128 b)
-{
-	float32x2_t a10 = vget_low_f32(vreinterpretq_f32_m128(a));
-	float32_t b2 = vgetq_lane_f32(vreinterpretq_f32_m128(b), 2);
-	float32x2_t b00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 0);
-	float32x2_t b20 = vset_lane_f32(b2, b00, 1);
-	return vreinterpretq_m128_f32(vcombine_f32(a10, b20));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_2001(__m128 a, __m128 b)
-{
-	float32x2_t a01 = vrev64_f32(vget_low_f32(vreinterpretq_f32_m128(a)));
-	float32_t b2 = vgetq_lane_f32(b, 2);
-	float32x2_t b00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 0);
-	float32x2_t b20 = vset_lane_f32(b2, b00, 1);
-	return vreinterpretq_m128_f32(vcombine_f32(a01, b20));
-}
-
-FORCE_INLINE __m128 _mm_shuffle_ps_2032(__m128 a, __m128 b)
-{
-	float32x2_t a32 = vget_high_f32(vreinterpretq_f32_m128(a));
-	float32_t b2 = vgetq_lane_f32(b, 2);
-	float32x2_t b00 =
-		vdup_lane_f32(vget_low_f32(vreinterpretq_f32_m128(b)), 0);
-	float32x2_t b20 = vset_lane_f32(b2, b00, 1);
-	return vreinterpretq_m128_f32(vcombine_f32(a32, b20));
-}
-
-// NEON does not support a general purpose permute intrinsic
-// Selects four specific single-precision, floating-point values from a and b,
-// based on the mask i.
-// https://msdn.microsoft.com/en-us/library/vstudio/5f0858x0(v=vs.100).aspx
-#if 0 /* C version */
-FORCE_INLINE __m128 _mm_shuffle_ps_default(__m128 a,
- __m128 b,
- __constrange(0, 255) int imm)
-{
- __m128 ret;
- ret[0] = a[imm & 0x3];
- ret[1] = a[(imm >> 2) & 0x3];
- ret[2] = b[(imm >> 4) & 0x03];
- ret[3] = b[(imm >> 6) & 0x03];
- return ret;
-}
-#endif
-#define _mm_shuffle_ps_default(a, b, imm) \
-	__extension__({ \
-		float32x4_t ret; \
-		ret = vmovq_n_f32(vgetq_lane_f32(vreinterpretq_f32_m128(a), \
-						 (imm) & (0x3))); \
-		ret = vsetq_lane_f32(vgetq_lane_f32(vreinterpretq_f32_m128(a), \
-						 ((imm) >> 2) & 0x3), \
-				 ret, 1); \
-		ret = vsetq_lane_f32(vgetq_lane_f32(vreinterpretq_f32_m128(b), \
-						 ((imm) >> 4) & 0x3), \
-				 ret, 2); \
-		ret = vsetq_lane_f32(vgetq_lane_f32(vreinterpretq_f32_m128(b), \
-						 ((imm) >> 6) & 0x3), \
-				 ret, 3); \
-		vreinterpretq_m128_f32(ret); \
-	})
-
-// FORCE_INLINE __m128 _mm_shuffle_ps(__m128 a, __m128 b, __constrange(0,255)
-// int imm)
-#if __has_builtin(__builtin_shufflevector)
-#define _mm_shuffle_ps(a, b, imm) \
-	__extension__({ \
-		float32x4_t _input1 = vreinterpretq_f32_m128(a); \
-		float32x4_t _input2 = vreinterpretq_f32_m128(b); \
-		float32x4_t _shuf = __builtin_shufflevector( \
-			_input1, _input2, (imm) & (0x3), ((imm) >> 2) & 0x3, \
-			(((imm) >> 4) & 0x3) + 4, (((imm) >> 6) & 0x3) + 4); \
-		vreinterpretq_m128_f32(_shuf); \
-	})
-#else // generic
-#define _mm_shuffle_ps(a, b, imm) \
-	__extension__({ \
-		__m128 ret; \
-		switch (imm) { \
-		case _MM_SHUFFLE(1, 0, 3, 2): \
-			ret = _mm_shuffle_ps_1032((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 3, 0, 1): \
-			ret = _mm_shuffle_ps_2301((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(0, 3, 2, 1): \
-			ret = _mm_shuffle_ps_0321((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 1, 0, 3): \
-			ret = _mm_shuffle_ps_2103((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(1, 0, 1, 0): \
-			ret = _mm_movelh_ps((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(1, 0, 0, 1): \
-			ret = _mm_shuffle_ps_1001((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(0, 1, 0, 1): \
-			ret = _mm_shuffle_ps_0101((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(3, 2, 1, 0): \
-			ret = _mm_shuffle_ps_3210((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(0, 0, 1, 1): \
-			ret = _mm_shuffle_ps_0011((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(0, 0, 2, 2): \
-			ret = _mm_shuffle_ps_0022((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 2, 0, 0): \
-			ret = _mm_shuffle_ps_2200((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(3, 2, 0, 2): \
-			ret = _mm_shuffle_ps_3202((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(3, 2, 3, 2): \
-			ret = _mm_movehl_ps((b), (a)); \
-			break; \
-		case _MM_SHUFFLE(1, 1, 3, 3): \
-			ret = _mm_shuffle_ps_1133((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 0, 1, 0): \
-			ret = _mm_shuffle_ps_2010((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 0, 0, 1): \
-			ret = _mm_shuffle_ps_2001((a), (b)); \
-			break; \
-		case _MM_SHUFFLE(2, 0, 3, 2): \
-			ret = _mm_shuffle_ps_2032((a), (b)); \
-			break; \
-		default: \
-			ret = _mm_shuffle_ps_default((a), (b), (imm)); \
-			break; \
-		} \
-		ret; \
-	})
-#endif
-
-// Takes the upper 64 bits of a and places it in the low end of the result
-// Takes the lower 64 bits of a and places it into the high end of the result.
-FORCE_INLINE __m128i _mm_shuffle_epi_1032(__m128i a)
-{
-	int32x2_t a32 = vget_high_s32(vreinterpretq_s32_m128i(a));
-	int32x2_t a10 = vget_low_s32(vreinterpretq_s32_m128i(a));
-	return vreinterpretq_m128i_s32(vcombine_s32(a32, a10));
-}
-
-// takes the lower two 32-bit values from a and swaps them and places in low end
-// of result takes the higher two 32 bit values from a and swaps them and places
-// in high end of result.
-FORCE_INLINE __m128i _mm_shuffle_epi_2301(__m128i a)
-{
-	int32x2_t a01 = vrev64_s32(vget_low_s32(vreinterpretq_s32_m128i(a)));
-	int32x2_t a23 = vrev64_s32(vget_high_s32(vreinterpretq_s32_m128i(a)));
-	return vreinterpretq_m128i_s32(vcombine_s32(a01, a23));
-}
-
-// rotates the least significant 32 bits into the most signficant 32 bits, and
-// shifts the rest down
-FORCE_INLINE __m128i _mm_shuffle_epi_0321(__m128i a)
-{
-	return vreinterpretq_m128i_s32(vextq_s32(
-		vreinterpretq_s32_m128i(a), vreinterpretq_s32_m128i(a), 1));
-}
-
-// rotates the most significant 32 bits into the least signficant 32 bits, and
-// shifts the rest up
-FORCE_INLINE __m128i _mm_shuffle_epi_2103(__m128i a)
-{
-	return vreinterpretq_m128i_s32(vextq_s32(
-		vreinterpretq_s32_m128i(a), vreinterpretq_s32_m128i(a), 3));
-}
-
-// gets the lower 64 bits of a, and places it in the upper 64 bits
-// gets the lower 64 bits of a and places it in the lower 64 bits
-FORCE_INLINE __m128i _mm_shuffle_epi_1010(__m128i a)
-{
-	int32x2_t a10 = vget_low_s32(vreinterpretq_s32_m128i(a));
-	return vreinterpretq_m128i_s32(vcombine_s32(a10, a10));
-}
-
-// gets the lower 64 bits of a, swaps the 0 and 1 elements, and places it in the
-// lower 64 bits gets the lower 64 bits of a, and places it in the upper 64 bits
-FORCE_INLINE __m128i _mm_shuffle_epi_1001(__m128i a)
-{
-	int32x2_t a01 = vrev64_s32(vget_low_s32(vreinterpretq_s32_m128i(a)));
-	int32x2_t a10 = vget_low_s32(vreinterpretq_s32_m128i(a));
-	return vreinterpretq_m128i_s32(vcombine_s32(a01, a10));
-}
-
-// gets the lower 64 bits of a, swaps the 0 and 1 elements and places it in the
-// upper 64 bits gets the lower 64 bits of a, swaps the 0 and 1 elements, and
-// places it in the lower 64 bits
-FORCE_INLINE __m128i _mm_shuffle_epi_0101(__m128i a)
-{
-	int32x2_t a01 = vrev64_s32(vget_low_s32(vreinterpretq_s32_m128i(a)));
-	return vreinterpretq_m128i_s32(vcombine_s32(a01, a01));
-}
-
-FORCE_INLINE __m128i _mm_shuffle_epi_2211(__m128i a)
-{
-	int32x2_t a11 =
-		vdup_lane_s32(vget_low_s32(vreinterpretq_s32_m128i(a)), 1);
-	int32x2_t a22 =
-		vdup_lane_s32(vget_high_s32(vreinterpretq_s32_m128i(a)), 0);
-	return vreinterpretq_m128i_s32(vcombine_s32(a11, a22));
-}
-
-FORCE_INLINE __m128i _mm_shuffle_epi_0122(__m128i a)
-{
-	int32x2_t a22 =
-		vdup_lane_s32(vget_high_s32(vreinterpretq_s32_m128i(a)), 0);
-	int32x2_t a01 = vrev64_s32(vget_low_s32(vreinterpretq_s32_m128i(a)));
-	return vreinterpretq_m128i_s32(vcombine_s32(a22, a01));
-}
-
-FORCE_INLINE __m128i _mm_shuffle_epi_3332(__m128i a)
-{
-	int32x2_t a32 = vget_high_s32(vreinterpretq_s32_m128i(a));
-	int32x2_t a33 =
-		vdup_lane_s32(vget_high_s32(vreinterpretq_s32_m128i(a)), 1);
-	return vreinterpretq_m128i_s32(vcombine_s32(a32, a33));
-}
-
-// Shuffle packed 8-bit integers in a according to shuffle control mask in the
-// corresponding 8-bit element of b, and store the results in dst.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_shuffle_epi8&expand=5146
-FORCE_INLINE __m128i _mm_shuffle_epi8(__m128i a, __m128i b)
-{
-	int8x16_t tbl = vreinterpretq_s8_m128i(a); // input a
-	uint8x16_t idx = vreinterpretq_u8_m128i(b); // input b
-	uint8x16_t idx_masked =
-		vandq_u8(idx, vdupq_n_u8(0x8F)); // avoid using meaningless bits
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s8(vqtbl1q_s8(tbl, idx_masked));
-#elif defined(__GNUC__)
-	int8x16_t ret;
-	// %e and %f represent the even and odd D registers
-	// respectively.
-	__asm__ __volatile__("vtbl.8 %e[ret], {%e[tbl], %f[tbl]}, %e[idx]\n"
-			 "vtbl.8 %f[ret], {%e[tbl], %f[tbl]}, %f[idx]\n"
-			 : [ret] "=&w"(ret)
-			 : [tbl] "w"(tbl), [idx] "w"(idx_masked));
-	return vreinterpretq_m128i_s8(ret);
-#else
-	// use this line if testing on aarch64
-	int8x8x2_t a_split = {vget_low_s8(tbl), vget_high_s8(tbl)};
-	return vreinterpretq_m128i_s8(
-		vcombine_s8(vtbl2_s8(a_split, vget_low_u8(idx_masked)),
-			 vtbl2_s8(a_split, vget_high_u8(idx_masked))));
-#endif
-}
-
-#if 0 /* C version */
-FORCE_INLINE __m128i _mm_shuffle_epi32_default(__m128i a,
- __constrange(0, 255) int imm)
-{
- __m128i ret;
- ret[0] = a[imm & 0x3];
- ret[1] = a[(imm >> 2) & 0x3];
- ret[2] = a[(imm >> 4) & 0x03];
- ret[3] = a[(imm >> 6) & 0x03];
- return ret;
-}
-#endif
-#define _mm_shuffle_epi32_default(a, imm) \
-	__extension__({ \
-		int32x4_t ret; \
-		ret = vmovq_n_s32(vgetq_lane_s32(vreinterpretq_s32_m128i(a), \
-						 (imm) & (0x3))); \
-		ret = vsetq_lane_s32( \
-			vgetq_lane_s32(vreinterpretq_s32_m128i(a), \
-				 ((imm) >> 2) & 0x3), \
-			ret, 1); \
-		ret = vsetq_lane_s32( \
-			vgetq_lane_s32(vreinterpretq_s32_m128i(a), \
-				 ((imm) >> 4) & 0x3), \
-			ret, 2); \
-		ret = vsetq_lane_s32( \
-			vgetq_lane_s32(vreinterpretq_s32_m128i(a), \
-				 ((imm) >> 6) & 0x3), \
-			ret, 3); \
-		vreinterpretq_m128i_s32(ret); \
-	})
-
-// FORCE_INLINE __m128i _mm_shuffle_epi32_splat(__m128i a, __constrange(0,255)
-// int imm)
-#if defined(__aarch64__)
-#define _mm_shuffle_epi32_splat(a, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s32( \
-			vdupq_laneq_s32(vreinterpretq_s32_m128i(a), (imm))); \
-	})
-#else
-#define _mm_shuffle_epi32_splat(a, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s32(vdupq_n_s32( \
-			vgetq_lane_s32(vreinterpretq_s32_m128i(a), (imm)))); \
-	})
-#endif
-
-// Shuffles the 4 signed or unsigned 32-bit integers in a as specified by imm.
-// https://msdn.microsoft.com/en-us/library/56f67xbk%28v=vs.90%29.aspx
-// FORCE_INLINE __m128i _mm_shuffle_epi32(__m128i a,
-// __constrange(0,255) int imm)
-#if __has_builtin(__builtin_shufflevector)
-#define _mm_shuffle_epi32(a, imm) \
-	__extension__({ \
-		int32x4_t _input = vreinterpretq_s32_m128i(a); \
-		int32x4_t _shuf = __builtin_shufflevector( \
-			_input, _input, (imm) & (0x3), ((imm) >> 2) & 0x3, \
-			((imm) >> 4) & 0x3, ((imm) >> 6) & 0x3); \
-		vreinterpretq_m128i_s32(_shuf); \
-	})
-#else // generic
-#define _mm_shuffle_epi32(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		switch (imm) { \
-		case _MM_SHUFFLE(1, 0, 3, 2): \
-			ret = _mm_shuffle_epi_1032((a)); \
-			break; \
-		case _MM_SHUFFLE(2, 3, 0, 1): \
-			ret = _mm_shuffle_epi_2301((a)); \
-			break; \
-		case _MM_SHUFFLE(0, 3, 2, 1): \
-			ret = _mm_shuffle_epi_0321((a)); \
-			break; \
-		case _MM_SHUFFLE(2, 1, 0, 3): \
-			ret = _mm_shuffle_epi_2103((a)); \
-			break; \
-		case _MM_SHUFFLE(1, 0, 1, 0): \
-			ret = _mm_shuffle_epi_1010((a)); \
-			break; \
-		case _MM_SHUFFLE(1, 0, 0, 1): \
-			ret = _mm_shuffle_epi_1001((a)); \
-			break; \
-		case _MM_SHUFFLE(0, 1, 0, 1): \
-			ret = _mm_shuffle_epi_0101((a)); \
-			break; \
-		case _MM_SHUFFLE(2, 2, 1, 1): \
-			ret = _mm_shuffle_epi_2211((a)); \
-			break; \
-		case _MM_SHUFFLE(0, 1, 2, 2): \
-			ret = _mm_shuffle_epi_0122((a)); \
-			break; \
-		case _MM_SHUFFLE(3, 3, 3, 2): \
-			ret = _mm_shuffle_epi_3332((a)); \
-			break; \
-		case _MM_SHUFFLE(0, 0, 0, 0): \
-			ret = _mm_shuffle_epi32_splat((a), 0); \
-			break; \
-		case _MM_SHUFFLE(1, 1, 1, 1): \
-			ret = _mm_shuffle_epi32_splat((a), 1); \
-			break; \
-		case _MM_SHUFFLE(2, 2, 2, 2): \
-			ret = _mm_shuffle_epi32_splat((a), 2); \
-			break; \
-		case _MM_SHUFFLE(3, 3, 3, 3): \
-			ret = _mm_shuffle_epi32_splat((a), 3); \
-			break; \
-		default: \
-			ret = _mm_shuffle_epi32_default((a), (imm)); \
-			break; \
-		} \
-		ret; \
-	})
-#endif
-
-// Shuffles the lower 4 signed or unsigned 16-bit integers in a as specified
-// by imm.
-// https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2010/y41dkk37(v=vs.100)
-// FORCE_INLINE __m128i _mm_shufflelo_epi16_function(__m128i a,
-// __constrange(0,255) int
-// imm)
-#define _mm_shufflelo_epi16_function(a, imm) \
-	__extension__({ \
-		int16x8_t ret = vreinterpretq_s16_m128i(a); \
-		int16x4_t lowBits = vget_low_s16(ret); \
-		ret = vsetq_lane_s16(vget_lane_s16(lowBits, (imm) & (0x3)), \
-				 ret, 0); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(lowBits, ((imm) >> 2) & 0x3), ret, 1); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(lowBits, ((imm) >> 4) & 0x3), ret, 2); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(lowBits, ((imm) >> 6) & 0x3), ret, 3); \
-		vreinterpretq_m128i_s16(ret); \
-	})
-
-// FORCE_INLINE __m128i _mm_shufflelo_epi16(__m128i a,
-// __constrange(0,255) int imm)
-#if __has_builtin(__builtin_shufflevector)
-#define _mm_shufflelo_epi16(a, imm) \
-	__extension__({ \
-		int16x8_t _input = vreinterpretq_s16_m128i(a); \
-		int16x8_t _shuf = __builtin_shufflevector( \
-			_input, _input, ((imm) & (0x3)), (((imm) >> 2) & 0x3), \
-			(((imm) >> 4) & 0x3), (((imm) >> 6) & 0x3), 4, 5, 6, \
-			7); \
-		vreinterpretq_m128i_s16(_shuf); \
-	})
-#else // generic
-#define _mm_shufflelo_epi16(a, imm) _mm_shufflelo_epi16_function((a), (imm))
-#endif
-
-// Shuffles the upper 4 signed or unsigned 16-bit integers in a as specified
-// by imm.
-// https://msdn.microsoft.com/en-us/library/13ywktbs(v=vs.100).aspx
-// FORCE_INLINE __m128i _mm_shufflehi_epi16_function(__m128i a,
-// __constrange(0,255) int
-// imm)
-#define _mm_shufflehi_epi16_function(a, imm) \
-	__extension__({ \
-		int16x8_t ret = vreinterpretq_s16_m128i(a); \
-		int16x4_t highBits = vget_high_s16(ret); \
-		ret = vsetq_lane_s16(vget_lane_s16(highBits, (imm) & (0x3)), \
-				 ret, 4); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(highBits, ((imm) >> 2) & 0x3), ret, 5); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(highBits, ((imm) >> 4) & 0x3), ret, 6); \
-		ret = vsetq_lane_s16( \
-			vget_lane_s16(highBits, ((imm) >> 6) & 0x3), ret, 7); \
-		vreinterpretq_m128i_s16(ret); \
-	})
-
-// FORCE_INLINE __m128i _mm_shufflehi_epi16(__m128i a,
-// __constrange(0,255) int imm)
-#if __has_builtin(__builtin_shufflevector)
-#define _mm_shufflehi_epi16(a, imm) \
-	__extension__({ \
-		int16x8_t _input = vreinterpretq_s16_m128i(a); \
-		int16x8_t _shuf = __builtin_shufflevector( \
-			_input, _input, 0, 1, 2, 3, ((imm) & (0x3)) + 4, \
-			(((imm) >> 2) & 0x3) + 4, (((imm) >> 4) & 0x3) + 4, \
-			(((imm) >> 6) & 0x3) + 4); \
-		vreinterpretq_m128i_s16(_shuf); \
-	})
-#else // generic
-#define _mm_shufflehi_epi16(a, imm) _mm_shufflehi_epi16_function((a), (imm))
-#endif
-
-// Blend packed 16-bit integers from a and b using control mask imm8, and store
-// the results in dst.
-//
-// FOR j := 0 to 7
-// i := j*16
-// IF imm8[j]
-// dst[i+15:i] := b[i+15:i]
-// ELSE
-// dst[i+15:i] := a[i+15:i]
-// FI
-// ENDFOR
-// FORCE_INLINE __m128i _mm_blend_epi16(__m128i a, __m128i b,
-// __constrange(0,255) int imm)
-#define _mm_blend_epi16(a, b, imm) \
-	__extension__({ \
-		const uint16_t _mask[8] = { \
-			((imm) & (1 << 0)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 1)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 2)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 3)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 4)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 5)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 6)) ? 0xFFFF : 0x0000, \
-			((imm) & (1 << 7)) ? 0xFFFF : 0x0000}; \
-		uint16x8_t _mask_vec = vld1q_u16(_mask); \
-		uint16x8_t _a = vreinterpretq_u16_m128i(a); \
-		uint16x8_t _b = vreinterpretq_u16_m128i(b); \
-		vreinterpretq_m128i_u16(vbslq_u16(_mask_vec, _b, _a)); \
-	})
-
-// Blend packed 8-bit integers from a and b using mask, and store the results in
-// dst.
-//
-// FOR j := 0 to 15
-// i := j*8
-// IF mask[i+7]
-// dst[i+7:i] := b[i+7:i]
-// ELSE
-// dst[i+7:i] := a[i+7:i]
-// FI
-// ENDFOR
-FORCE_INLINE __m128i _mm_blendv_epi8(__m128i _a, __m128i _b, __m128i _mask)
-{
-	// Use a signed shift right to create a mask with the sign bit
-	uint8x16_t mask = vreinterpretq_u8_s8(
-		vshrq_n_s8(vreinterpretq_s8_m128i(_mask), 7));
-	uint8x16_t a = vreinterpretq_u8_m128i(_a);
-	uint8x16_t b = vreinterpretq_u8_m128i(_b);
-	return vreinterpretq_m128i_u8(vbslq_u8(mask, b, a));
-}
-
-/* Shifts */
-
-// Shifts the 4 signed 32-bit integers in a right by count bits while shifting
-// in the sign bit.
-//
-// r0 := a0 >> count
-// r1 := a1 >> count
-// r2 := a2 >> count
-// r3 := a3 >> count immediate
-FORCE_INLINE __m128i _mm_srai_epi32(__m128i a, int count)
-{
-	return (__m128i)vshlq_s32((int32x4_t)a, vdupq_n_s32(-count));
-}
-
-// Shifts the 8 signed 16-bit integers in a right by count bits while shifting
-// in the sign bit.
-//
-// r0 := a0 >> count
-// r1 := a1 >> count
-// ...
-// r7 := a7 >> count
-FORCE_INLINE __m128i _mm_srai_epi16(__m128i a, int count)
-{
-	return (__m128i)vshlq_s16((int16x8_t)a, vdupq_n_s16(-count));
-}
-
-// Shifts the 8 signed or unsigned 16-bit integers in a left by count bits while
-// shifting in zeros.
-//
-// r0 := a0 << count
-// r1 := a1 << count
-// ...
-// r7 := a7 << count
-//
-// https://msdn.microsoft.com/en-us/library/es73bcsy(v=vs.90).aspx
-#define _mm_slli_epi16(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 31) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_s16(vshlq_n_s16( \
-				vreinterpretq_s16_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 4 signed or unsigned 32-bit integers in a left by count bits while
-// shifting in zeros. :
-// https://msdn.microsoft.com/en-us/library/z2k3bbtb%28v=vs.90%29.aspx
-// FORCE_INLINE __m128i _mm_slli_epi32(__m128i a, __constrange(0,255) int imm)
-#define _mm_slli_epi32(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 31) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_s32(vshlq_n_s32( \
-				vreinterpretq_s32_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shift packed 64-bit integers in a left by imm8 while shifting in zeros, and
-// store the results in dst.
-#define _mm_slli_epi64(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 63) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_s64(vshlq_n_s64( \
-				vreinterpretq_s64_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 8 signed or unsigned 16-bit integers in a right by count bits
-// while shifting in zeros.
-//
-// r0 := srl(a0, count)
-// r1 := srl(a1, count)
-// ...
-// r7 := srl(a7, count)
-//
-// https://msdn.microsoft.com/en-us/library/6tcwd38t(v=vs.90).aspx
-#define _mm_srli_epi16(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 31) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_u16(vshrq_n_u16( \
-				vreinterpretq_u16_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 4 signed or unsigned 32-bit integers in a right by count bits
-// while shifting in zeros.
-// https://msdn.microsoft.com/en-us/library/w486zcfa(v=vs.100).aspx FORCE_INLINE
-// __m128i _mm_srli_epi32(__m128i a, __constrange(0,255) int imm)
-#define _mm_srli_epi32(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 31) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_u32(vshrq_n_u32( \
-				vreinterpretq_u32_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shift packed 64-bit integers in a right by imm8 while shifting in zeros, and
-// store the results in dst.
-#define _mm_srli_epi64(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 63) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_u64(vshrq_n_u64( \
-				vreinterpretq_u64_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 4 signed 32 - bit integers in a right by count bits while shifting
-// in the sign bit.
-// https://msdn.microsoft.com/en-us/library/z1939387(v=vs.100).aspx
-// FORCE_INLINE __m128i _mm_srai_epi32(__m128i a, __constrange(0,255) int imm)
-#define _mm_srai_epi32(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 31) { \
-			ret = vreinterpretq_m128i_s32( \
-				vshrq_n_s32(vreinterpretq_s32_m128i(a), 16)); \
-			ret = vreinterpretq_m128i_s32(vshrq_n_s32( \
-				vreinterpretq_s32_m128i(ret), 16)); \
-		} else { \
-			ret = vreinterpretq_m128i_s32(vshrq_n_s32( \
-				vreinterpretq_s32_m128i(a), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 128 - bit value in a right by imm bytes while shifting in
-// zeros.imm must be an immediate.
-//
-// r := srl(a, imm*8)
-//
-// https://msdn.microsoft.com/en-us/library/305w28yz(v=vs.100).aspx
-// FORCE_INLINE _mm_srli_si128(__m128i a, __constrange(0,255) int imm)
-#define _mm_srli_si128(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 15) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_s8( \
-				vextq_s8(vreinterpretq_s8_m128i(a), \
-					 vdupq_n_s8(0), (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 128-bit value in a left by imm bytes while shifting in zeros. imm
-// must be an immediate.
-//
-// r := a << (imm * 8)
-//
-// https://msdn.microsoft.com/en-us/library/34d3k2kt(v=vs.100).aspx
-// FORCE_INLINE __m128i _mm_slli_si128(__m128i a, __constrange(0,255) int imm)
-#define _mm_slli_si128(a, imm) \
-	__extension__({ \
-		__m128i ret; \
-		if ((imm) <= 0) { \
-			ret = a; \
-		} else if ((imm) > 15) { \
-			ret = _mm_setzero_si128(); \
-		} else { \
-			ret = vreinterpretq_m128i_s8(vextq_s8( \
-				vdupq_n_s8(0), vreinterpretq_s8_m128i(a), \
-				16 - (imm))); \
-		} \
-		ret; \
-	})
-
-// Shifts the 8 signed or unsigned 16-bit integers in a left by count bits while
-// shifting in zeros.
-//
-// r0 := a0 << count
-// r1 := a1 << count
-// ...
-// r7 := a7 << count
-//
-// https://msdn.microsoft.com/en-us/library/c79w388h(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_sll_epi16(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 15)
-		return _mm_setzero_si128();
-
-	int16x8_t vc = vdupq_n_s16((int16_t)c);
-	return vreinterpretq_m128i_s16(
-		vshlq_s16(vreinterpretq_s16_m128i(a), vc));
-}
-
-// Shifts the 4 signed or unsigned 32-bit integers in a left by count bits while
-// shifting in zeros.
-//
-// r0 := a0 << count
-// r1 := a1 << count
-// r2 := a2 << count
-// r3 := a3 << count
-//
-// https://msdn.microsoft.com/en-us/library/6fe5a6s9(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_sll_epi32(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 31)
-		return _mm_setzero_si128();
-
-	int32x4_t vc = vdupq_n_s32((int32_t)c);
-	return vreinterpretq_m128i_s32(
-		vshlq_s32(vreinterpretq_s32_m128i(a), vc));
-}
-
-// Shifts the 2 signed or unsigned 64-bit integers in a left by count bits while
-// shifting in zeros.
-//
-// r0 := a0 << count
-// r1 := a1 << count
-//
-// https://msdn.microsoft.com/en-us/library/6ta9dffd(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_sll_epi64(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 63)
-		return _mm_setzero_si128();
-
-	int64x2_t vc = vdupq_n_s64((int64_t)c);
-	return vreinterpretq_m128i_s64(
-		vshlq_s64(vreinterpretq_s64_m128i(a), vc));
-}
-
-// Shifts the 8 signed or unsigned 16-bit integers in a right by count bits
-// while shifting in zeros.
-//
-// r0 := srl(a0, count)
-// r1 := srl(a1, count)
-// ...
-// r7 := srl(a7, count)
-//
-// https://msdn.microsoft.com/en-us/library/wd5ax830(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_srl_epi16(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 15)
-		return _mm_setzero_si128();
-
-	int16x8_t vc = vdupq_n_s16(-(int16_t)c);
-	return vreinterpretq_m128i_u16(
-		vshlq_u16(vreinterpretq_u16_m128i(a), vc));
-}
-
-// Shifts the 4 signed or unsigned 32-bit integers in a right by count bits
-// while shifting in zeros.
-//
-// r0 := srl(a0, count)
-// r1 := srl(a1, count)
-// r2 := srl(a2, count)
-// r3 := srl(a3, count)
-//
-// https://msdn.microsoft.com/en-us/library/a9cbttf4(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_srl_epi32(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 31)
-		return _mm_setzero_si128();
-
-	int32x4_t vc = vdupq_n_s32(-(int32_t)c);
-	return vreinterpretq_m128i_u32(
-		vshlq_u32(vreinterpretq_u32_m128i(a), vc));
-}
-
-// Shifts the 2 signed or unsigned 64-bit integers in a right by count bits
-// while shifting in zeros.
-//
-// r0 := srl(a0, count)
-// r1 := srl(a1, count)
-//
-// https://msdn.microsoft.com/en-us/library/yf6cf9k8(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_srl_epi64(__m128i a, __m128i count)
-{
-	uint64_t c = vreinterpretq_nth_u64_m128i(count, 0);
-	if (c > 63)
-		return _mm_setzero_si128();
-
-	int64x2_t vc = vdupq_n_s64(-(int64_t)c);
-	return vreinterpretq_m128i_u64(
-		vshlq_u64(vreinterpretq_u64_m128i(a), vc));
-}
-
-// NEON does not provide a version of this function.
-// Creates a 16-bit mask from the most significant bits of the 16 signed or
-// unsigned 8-bit integers in a and zero extends the upper bits.
-// https://msdn.microsoft.com/en-us/library/vstudio/s090c8fk(v=vs.100).aspx
-FORCE_INLINE int _mm_movemask_epi8(__m128i a)
-{
-#if defined(__aarch64__)
-	uint8x16_t input = vreinterpretq_u8_m128i(a);
-	const int8_t ALIGN_STRUCT(16) xr[16] = {-7, -6, -5, -4, -3, -2, -1, 0,
-						-7, -6, -5, -4, -3, -2, -1, 0};
-	const uint8x16_t mask_and = vdupq_n_u8(0x80);
-	const int8x16_t mask_shift = vld1q_s8(xr);
-	const uint8x16_t mask_result =
-		vshlq_u8(vandq_u8(input, mask_and), mask_shift);
-	uint8x8_t lo = vget_low_u8(mask_result);
-	uint8x8_t hi = vget_high_u8(mask_result);
-
-	return vaddv_u8(lo) + (vaddv_u8(hi) << 8);
-#else
-	// Use increasingly wide shifts+adds to collect the sign bits
-	// together.
-	// Since the widening shifts would be rather confusing to follow in little
-	// endian, everything will be illustrated in big endian order instead. This
-	// has a different result - the bits would actually be reversed on a big
-	// endian machine.
-
-	// Starting input (only half the elements are shown):
-	// 89 ff 1d c0 00 10 99 33
-	uint8x16_t input = vreinterpretq_u8_m128i(a);
-
-	// Shift out everything but the sign bits with an unsigned shift right.
-	//
-	// Bytes of the vector::
-	// 89 ff 1d c0 00 10 99 33
-	// \ \ \ \ \ \ \ \ high_bits = (uint16x4_t)(input >> 7)
-	// | | | | | | | |
-	// 01 01 00 01 00 00 01 00
-	//
-	// Bits of first important lane(s):
-	// 10001001 (89)
-	// \______
-	// |
-	// 00000001 (01)
-	uint16x8_t high_bits = vreinterpretq_u16_u8(vshrq_n_u8(input, 7));
-
-	// Merge the even lanes together with a 16-bit unsigned shift right + add.
-	// 'xx' represents garbage data which will be ignored in the final result.
-	// In the important bytes, the add functions like a binary OR.
-	//
-	// 01 01 00 01 00 00 01 00
-	// \_ | \_ | \_ | \_ | paired16 = (uint32x4_t)(input + (input >> 7))
-	// \| \| \| \|
-	// xx 03 xx 01 xx 00 xx 02
-	//
-	// 00000001 00000001 (01 01)
-	// \_______ |
-	// \|
-	// xxxxxxxx xxxxxx11 (xx 03)
-	uint32x4_t paired16 =
-		vreinterpretq_u32_u16(vsraq_n_u16(high_bits, high_bits, 7));
-
-	// Repeat with a wider 32-bit shift + add.
-	// xx 03 xx 01 xx 00 xx 02
-	// \____ | \____ | paired32 = (uint64x1_t)(paired16 + (paired16 >>
-	// 14))
-	// \| \|
-	// xx xx xx 0d xx xx xx 02
-	//
-	// 00000011 00000001 (03 01)
-	// \\_____ ||
-	// '----.\||
-	// xxxxxxxx xxxx1101 (xx 0d)
-	uint64x2_t paired32 =
-		vreinterpretq_u64_u32(vsraq_n_u32(paired16, paired16, 14));
-
-	// Last, an even wider 64-bit shift + add to get our result in the low 8 bit
-	// lanes. xx xx xx 0d xx xx xx 02
-	// \_________ | paired64 = (uint8x8_t)(paired32 + (paired32 >>
-	// 28))
-	// \|
-	// xx xx xx xx xx xx xx d2
-	//
-	// 00001101 00000010 (0d 02)
-	// \ \___ | |
-	// '---. \| |
-	// xxxxxxxx 11010010 (xx d2)
-	uint8x16_t paired64 =
-		vreinterpretq_u8_u64(vsraq_n_u64(paired32, paired32, 28));
-
-	// Extract the low 8 bits from each 64-bit lane with 2 8-bit extracts.
-	// xx xx xx xx xx xx xx d2
-	// || return paired64[0]
-	// d2
-	// Note: Little endian would return the correct value 4b (01001011) instead.
-	return vgetq_lane_u8(paired64, 0) |
-	 ((int)vgetq_lane_u8(paired64, 8) << 8);
-#endif
-}
-
-// NEON does not provide this method
-// Creates a 4-bit mask from the most significant bits of the four
-// single-precision, floating-point values.
-// https://msdn.microsoft.com/en-us/library/vstudio/4490ys29(v=vs.100).aspx
-FORCE_INLINE int _mm_movemask_ps(__m128 a)
-{
-	uint32x4_t input = vreinterpretq_u32_m128(a);
-#if defined(__aarch64__)
-	static const int32x4_t shift = {-31, -30, -29, -28};
-	static const uint32x4_t highbit = {0x80000000, 0x80000000, 0x80000000,
-					 0x80000000};
-	return vaddvq_u32(vshlq_u32(vandq_u32(input, highbit), shift));
-#else
-	// Uses the exact same method as _mm_movemask_epi8, see that for details.
-	// Shift out everything but the sign bits with a 32-bit unsigned shift
-	// right.
-	uint64x2_t high_bits = vreinterpretq_u64_u32(vshrq_n_u32(input, 31));
-	// Merge the two pairs together with a 64-bit unsigned shift right + add.
-	uint8x16_t paired =
-		vreinterpretq_u8_u64(vsraq_n_u64(high_bits, high_bits, 31));
-	// Extract the result.
-	return vgetq_lane_u8(paired, 0) | (vgetq_lane_u8(paired, 8) << 2);
-#endif
-}
-
-// Compute the bitwise AND of 128 bits (representing integer data) in a and
-// mask, and return 1 if the result is zero, otherwise return 0.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_test_all_zeros&expand=5871
-FORCE_INLINE int _mm_test_all_zeros(__m128i a, __m128i mask)
-{
-	int64x2_t a_and_mask = vandq_s64(vreinterpretq_s64_m128i(a),
-					 vreinterpretq_s64_m128i(mask));
-	return (vgetq_lane_s64(a_and_mask, 0) | vgetq_lane_s64(a_and_mask, 1))
-		 ? 0
-		 : 1;
-}
-
-/* Math operations */
-
-// Subtracts the four single-precision, floating-point values of a and b.
-//
-// r0 := a0 - b0
-// r1 := a1 - b1
-// r2 := a2 - b2
-// r3 := a3 - b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/1zad2k61(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_sub_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_f32(vsubq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Subtract 2 packed 64-bit integers in b from 2 packed 64-bit integers in a,
-// and store the results in dst.
-// r0 := a0 - b0
-// r1 := a1 - b1
-FORCE_INLINE __m128i _mm_sub_epi64(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s64(vsubq_s64(vreinterpretq_s64_m128i(a),
-						 vreinterpretq_s64_m128i(b)));
-}
-
-// Subtracts the 4 signed or unsigned 32-bit integers of b from the 4 signed or
-// unsigned 32-bit integers of a.
-//
-// r0 := a0 - b0
-// r1 := a1 - b1
-// r2 := a2 - b2
-// r3 := a3 - b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/fhh866h0(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_sub_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vsubq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-FORCE_INLINE __m128i _mm_sub_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vsubq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-FORCE_INLINE __m128i _mm_sub_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s8(
-		vsubq_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b)));
-}
-
-// Subtracts the 8 unsigned 16-bit integers of bfrom the 8 unsigned 16-bit
-// integers of a and saturates..
-// https://technet.microsoft.com/en-us/subscriptions/index/f44y0s19(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_subs_epu16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(vqsubq_u16(vreinterpretq_u16_m128i(a),
-						 vreinterpretq_u16_m128i(b)));
-}
-
-// Subtracts the 16 unsigned 8-bit integers of b from the 16 unsigned 8-bit
-// integers of a and saturates.
-//
-// r0 := UnsignedSaturate(a0 - b0)
-// r1 := UnsignedSaturate(a1 - b1)
-// ...
-// r15 := UnsignedSaturate(a15 - b15)
-//
-// https://technet.microsoft.com/en-us/subscriptions/yadkxc18(v=vs.90)
-FORCE_INLINE __m128i _mm_subs_epu8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(vqsubq_u8(vreinterpretq_u8_m128i(a),
-						vreinterpretq_u8_m128i(b)));
-}
-
-// Subtracts the 16 signed 8-bit integers of b from the 16 signed 8-bit integers
-// of a and saturates.
-//
-// r0 := SignedSaturate(a0 - b0)
-// r1 := SignedSaturate(a1 - b1)
-// ...
-// r15 := SignedSaturate(a15 - b15)
-//
-// https://technet.microsoft.com/en-us/subscriptions/by7kzks1(v=vs.90)
-FORCE_INLINE __m128i _mm_subs_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s8(vqsubq_s8(vreinterpretq_s8_m128i(a),
-						vreinterpretq_s8_m128i(b)));
-}
-
-// Subtracts the 8 signed 16-bit integers of b from the 8 signed 16-bit integers
-// of a and saturates.
-//
-// r0 := SignedSaturate(a0 - b0)
-// r1 := SignedSaturate(a1 - b1)
-// ...
-// r7 := SignedSaturate(a7 - b7)
-//
-// https://technet.microsoft.com/en-us/subscriptions/3247z5b8(v=vs.90)
-FORCE_INLINE __m128i _mm_subs_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vqsubq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-FORCE_INLINE __m128i _mm_adds_epu16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(vqaddq_u16(vreinterpretq_u16_m128i(a),
-						 vreinterpretq_u16_m128i(b)));
-}
-
-// Negate packed 8-bit integers in a when the corresponding signed
-// 8-bit integer in b is negative, and store the results in dst.
-// Element in dst are zeroed out when the corresponding element
-// in b is zero.
-//
-// for i in 0..15
-// if b[i] < 0
-// r[i] := -a[i]
-// else if b[i] == 0
-// r[i] := 0
-// else
-// r[i] := a[i]
-// fi
-// done
-FORCE_INLINE __m128i _mm_sign_epi8(__m128i _a, __m128i _b)
-{
-	int8x16_t a = vreinterpretq_s8_m128i(_a);
-	int8x16_t b = vreinterpretq_s8_m128i(_b);
-
-	int8x16_t zero = vdupq_n_s8(0);
-	// signed shift right: faster than vclt
-	// (b < 0) ? 0xFF : 0
-	uint8x16_t ltMask = vreinterpretq_u8_s8(vshrq_n_s8(b, 7));
-	// (b == 0) ? 0xFF : 0
-	int8x16_t zeroMask = vreinterpretq_s8_u8(vceqq_s8(b, zero));
-	// -a
-	int8x16_t neg = vnegq_s8(a);
-	// bitwise select either a or neg based on ltMask
-	int8x16_t masked = vbslq_s8(ltMask, a, neg);
-	// res = masked & (~zeroMask)
-	int8x16_t res = vbicq_s8(masked, zeroMask);
-	return vreinterpretq_m128i_s8(res);
-}
-
-// Negate packed 16-bit integers in a when the corresponding signed
-// 16-bit integer in b is negative, and store the results in dst.
-// Element in dst are zeroed out when the corresponding element
-// in b is zero.
-//
-// for i in 0..7
-// if b[i] < 0
-// r[i] := -a[i]
-// else if b[i] == 0
-// r[i] := 0
-// else
-// r[i] := a[i]
-// fi
-// done
-FORCE_INLINE __m128i _mm_sign_epi16(__m128i _a, __m128i _b)
-{
-	int16x8_t a = vreinterpretq_s16_m128i(_a);
-	int16x8_t b = vreinterpretq_s16_m128i(_b);
-
-	int16x8_t zero = vdupq_n_s16(0);
-	// signed shift right: faster than vclt
-	// (b < 0) ? 0xFFFF : 0
-	uint16x8_t ltMask = vreinterpretq_u16_s16(vshrq_n_s16(b, 15));
-	// (b == 0) ? 0xFFFF : 0
-	int16x8_t zeroMask = vreinterpretq_s16_u16(vceqq_s16(b, zero));
-	// -a
-	int16x8_t neg = vnegq_s16(a);
-	// bitwise select either a or neg based on ltMask
-	int16x8_t masked = vbslq_s16(ltMask, a, neg);
-	// res = masked & (~zeroMask)
-	int16x8_t res = vbicq_s16(masked, zeroMask);
-	return vreinterpretq_m128i_s16(res);
-}
-
-// Negate packed 32-bit integers in a when the corresponding signed
-// 32-bit integer in b is negative, and store the results in dst.
-// Element in dst are zeroed out when the corresponding element
-// in b is zero.
-//
-// for i in 0..3
-// if b[i] < 0
-// r[i] := -a[i]
-// else if b[i] == 0
-// r[i] := 0
-// else
-// r[i] := a[i]
-// fi
-// done
-FORCE_INLINE __m128i _mm_sign_epi32(__m128i _a, __m128i _b)
-{
-	int32x4_t a = vreinterpretq_s32_m128i(_a);
-	int32x4_t b = vreinterpretq_s32_m128i(_b);
-
-	int32x4_t zero = vdupq_n_s32(0);
-	// signed shift right: faster than vclt
-	// (b < 0) ? 0xFFFFFFFF : 0
-	uint32x4_t ltMask = vreinterpretq_u32_s32(vshrq_n_s32(b, 31));
-	// (b == 0) ? 0xFFFFFFFF : 0
-	int32x4_t zeroMask = vreinterpretq_s32_u32(vceqq_s32(b, zero));
-	// neg = -a
-	int32x4_t neg = vnegq_s32(a);
-	// bitwise select either a or neg based on ltMask
-	int32x4_t masked = vbslq_s32(ltMask, a, neg);
-	// res = masked & (~zeroMask)
-	int32x4_t res = vbicq_s32(masked, zeroMask);
-	return vreinterpretq_m128i_s32(res);
-}
-
-// Computes the average of the 16 unsigned 8-bit integers in a and the 16
-// unsigned 8-bit integers in b and rounds.
-//
-// r0 := (a0 + b0) / 2
-// r1 := (a1 + b1) / 2
-// ...
-// r15 := (a15 + b15) / 2
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/8zwh554a(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_avg_epu8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(vrhaddq_u8(vreinterpretq_u8_m128i(a),
-						 vreinterpretq_u8_m128i(b)));
-}
-
-// Computes the average of the 8 unsigned 16-bit integers in a and the 8
-// unsigned 16-bit integers in b and rounds.
-//
-// r0 := (a0 + b0) / 2
-// r1 := (a1 + b1) / 2
-// ...
-// r7 := (a7 + b7) / 2
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/y13ca3c8(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_avg_epu16(__m128i a, __m128i b)
-{
-	return (__m128i)vrhaddq_u16(vreinterpretq_u16_m128i(a),
-				 vreinterpretq_u16_m128i(b));
-}
-
-// Adds the four single-precision, floating-point values of a and b.
-//
-// r0 := a0 + b0
-// r1 := a1 + b1
-// r2 := a2 + b2
-// r3 := a3 + b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/c9848chc(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_add_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_f32(vaddq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// adds the scalar single-precision floating point values of a and b.
-// https://msdn.microsoft.com/en-us/library/be94x2y6(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_add_ss(__m128 a, __m128 b)
-{
-	float32_t b0 = vgetq_lane_f32(vreinterpretq_f32_m128(b), 0);
-	float32x4_t value = vsetq_lane_f32(b0, vdupq_n_f32(0), 0);
-	// the upper values in the result must be the remnants of <a>.
-	return vreinterpretq_m128_f32(vaddq_f32(a, value));
-}
-
-// Adds the 4 signed or unsigned 64-bit integers in a to the 4 signed or
-// unsigned 32-bit integers in b.
-// https://msdn.microsoft.com/en-us/library/vstudio/09xs4fkk(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_add_epi64(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s64(vaddq_s64(vreinterpretq_s64_m128i(a),
-						 vreinterpretq_s64_m128i(b)));
-}
-
-// Adds the 4 signed or unsigned 32-bit integers in a to the 4 signed or
-// unsigned 32-bit integers in b.
-//
-// r0 := a0 + b0
-// r1 := a1 + b1
-// r2 := a2 + b2
-// r3 := a3 + b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/09xs4fkk(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_add_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vaddq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Adds the 8 signed or unsigned 16-bit integers in a to the 8 signed or
-// unsigned 16-bit integers in b.
-// https://msdn.microsoft.com/en-us/library/fceha5k4(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_add_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vaddq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Adds the 16 signed or unsigned 8-bit integers in a to the 16 signed or
-// unsigned 8-bit integers in b.
-// https://technet.microsoft.com/en-us/subscriptions/yc7tcyzs(v=vs.90)
-FORCE_INLINE __m128i _mm_add_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s8(
-		vaddq_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b)));
-}
-
-// Adds the 8 signed 16-bit integers in a to the 8 signed 16-bit integers in b
-// and saturates.
-//
-// r0 := SignedSaturate(a0 + b0)
-// r1 := SignedSaturate(a1 + b1)
-// ...
-// r7 := SignedSaturate(a7 + b7)
-//
-// https://msdn.microsoft.com/en-us/library/1a306ef8(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_adds_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vqaddq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Adds the 16 unsigned 8-bit integers in a to the 16 unsigned 8-bit integers in
-// b and saturates..
-// https://msdn.microsoft.com/en-us/library/9hahyddy(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_adds_epu8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(vqaddq_u8(vreinterpretq_u8_m128i(a),
-						vreinterpretq_u8_m128i(b)));
-}
-
-// Multiplies the 8 signed or unsigned 16-bit integers from a by the 8 signed or
-// unsigned 16-bit integers from b.
-//
-// r0 := (a0 * b0)[15:0]
-// r1 := (a1 * b1)[15:0]
-// ...
-// r7 := (a7 * b7)[15:0]
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/9ks1472s(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_mullo_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vmulq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Multiplies the 4 signed or unsigned 32-bit integers from a by the 4 signed or
-// unsigned 32-bit integers from b.
-// https://msdn.microsoft.com/en-us/library/vstudio/bb531409(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_mullo_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vmulq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Multiplies the four single-precision, floating-point values of a and b.
-//
-// r0 := a0 * b0
-// r1 := a1 * b1
-// r2 := a2 * b2
-// r3 := a3 * b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/22kbk6t9(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_mul_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_f32(vmulq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Multiply the low unsigned 32-bit integers from each packed 64-bit element in
-// a and b, and store the unsigned 64-bit results in dst.
-//
-// r0 := (a0 & 0xFFFFFFFF) * (b0 & 0xFFFFFFFF)
-// r1 := (a2 & 0xFFFFFFFF) * (b2 & 0xFFFFFFFF)
-FORCE_INLINE __m128i _mm_mul_epu32(__m128i a, __m128i b)
-{
-	// vmull_u32 upcasts instead of masking, so we downcast.
-	uint32x2_t a_lo = vmovn_u64(vreinterpretq_u64_m128i(a));
-	uint32x2_t b_lo = vmovn_u64(vreinterpretq_u64_m128i(b));
-	return vreinterpretq_m128i_u64(vmull_u32(a_lo, b_lo));
-}
-
-// Multiply the low signed 32-bit integers from each packed 64-bit element in
-// a and b, and store the signed 64-bit results in dst.
-//
-// r0 := (int64_t)(int32_t)a0 * (int64_t)(int32_t)b0
-// r1 := (int64_t)(int32_t)a2 * (int64_t)(int32_t)b2
-FORCE_INLINE __m128i _mm_mul_epi32(__m128i a, __m128i b)
-{
-	// vmull_s32 upcasts instead of masking, so we downcast.
-	int32x2_t a_lo = vmovn_s64(vreinterpretq_s64_m128i(a));
-	int32x2_t b_lo = vmovn_s64(vreinterpretq_s64_m128i(b));
-	return vreinterpretq_m128i_s64(vmull_s32(a_lo, b_lo));
-}
-
-// Multiplies the 8 signed 16-bit integers from a by the 8 signed 16-bit
-// integers from b.
-//
-// r0 := (a0 * b0) + (a1 * b1)
-// r1 := (a2 * b2) + (a3 * b3)
-// r2 := (a4 * b4) + (a5 * b5)
-// r3 := (a6 * b6) + (a7 * b7)
-// https://msdn.microsoft.com/en-us/library/yht36sa6(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_madd_epi16(__m128i a, __m128i b)
-{
-	int32x4_t low = vmull_s16(vget_low_s16(vreinterpretq_s16_m128i(a)),
-				 vget_low_s16(vreinterpretq_s16_m128i(b)));
-	int32x4_t high = vmull_s16(vget_high_s16(vreinterpretq_s16_m128i(a)),
-				 vget_high_s16(vreinterpretq_s16_m128i(b)));
-
-	int32x2_t low_sum = vpadd_s32(vget_low_s32(low), vget_high_s32(low));
-	int32x2_t high_sum = vpadd_s32(vget_low_s32(high), vget_high_s32(high));
-
-	return vreinterpretq_m128i_s32(vcombine_s32(low_sum, high_sum));
-}
-
-// Multiply packed signed 16-bit integers in a and b, producing intermediate
-// signed 32-bit integers. Shift right by 15 bits while rounding up, and store
-// the packed 16-bit integers in dst.
-//
-// r0 := Round(((int32_t)a0 * (int32_t)b0) >> 15)
-// r1 := Round(((int32_t)a1 * (int32_t)b1) >> 15)
-// r2 := Round(((int32_t)a2 * (int32_t)b2) >> 15)
-// ...
-// r7 := Round(((int32_t)a7 * (int32_t)b7) >> 15)
-FORCE_INLINE __m128i _mm_mulhrs_epi16(__m128i a, __m128i b)
-{
-	// Has issues due to saturation
-	// return vreinterpretq_m128i_s16(vqrdmulhq_s16(a, b));
-
-	// Multiply
-	int32x4_t mul_lo = vmull_s16(vget_low_s16(vreinterpretq_s16_m128i(a)),
-				 vget_low_s16(vreinterpretq_s16_m128i(b)));
-	int32x4_t mul_hi = vmull_s16(vget_high_s16(vreinterpretq_s16_m128i(a)),
-				 vget_high_s16(vreinterpretq_s16_m128i(b)));
-
-	// Rounding narrowing shift right
-	// narrow = (int16_t)((mul + 16384) >> 15);
-	int16x4_t narrow_lo = vrshrn_n_s32(mul_lo, 15);
-	int16x4_t narrow_hi = vrshrn_n_s32(mul_hi, 15);
-
-	// Join together
-	return vreinterpretq_m128i_s16(vcombine_s16(narrow_lo, narrow_hi));
-}
-
-// Vertically multiply each unsigned 8-bit integer from a with the corresponding
-// signed 8-bit integer from b, producing intermediate signed 16-bit integers.
-// Horizontally add adjacent pairs of intermediate signed 16-bit integers,
-// and pack the saturated results in dst.
-//
-// FOR j := 0 to 7
-// i := j*16
-// dst[i+15:i] := Saturate_To_Int16( a[i+15:i+8]*b[i+15:i+8] +
-// a[i+7:i]*b[i+7:i] )
-// ENDFOR
-FORCE_INLINE __m128i _mm_maddubs_epi16(__m128i _a, __m128i _b)
-{
-	// This would be much simpler if x86 would choose to zero extend OR sign
-	// extend, not both. This could probably be optimized better.
-	uint16x8_t a = vreinterpretq_u16_m128i(_a);
-	int16x8_t b = vreinterpretq_s16_m128i(_b);
-
-	// Zero extend a
-	int16x8_t a_odd = vreinterpretq_s16_u16(vshrq_n_u16(a, 8));
-	int16x8_t a_even =
-		vreinterpretq_s16_u16(vbicq_u16(a, vdupq_n_u16(0xff00)));
-
-	// Sign extend by shifting left then shifting right.
-	int16x8_t b_even = vshrq_n_s16(vshlq_n_s16(b, 8), 8);
-	int16x8_t b_odd = vshrq_n_s16(b, 8);
-
-	// multiply
-	int16x8_t prod1 = vmulq_s16(a_even, b_even);
-	int16x8_t prod2 = vmulq_s16(a_odd, b_odd);
-
-	// saturated add
-	return vreinterpretq_m128i_s16(vqaddq_s16(prod1, prod2));
-}
-
-// Computes the absolute difference of the 16 unsigned 8-bit integers from a
-// and the 16 unsigned 8-bit integers from b.
-//
-// Return Value
-// Sums the upper 8 differences and lower 8 differences and packs the
-// resulting 2 unsigned 16-bit integers into the upper and lower 64-bit
-// elements.
-//
-// r0 := abs(a0 - b0) + abs(a1 - b1) +...+ abs(a7 - b7)
-// r1 := 0x0
-// r2 := 0x0
-// r3 := 0x0
-// r4 := abs(a8 - b8) + abs(a9 - b9) +...+ abs(a15 - b15)
-// r5 := 0x0
-// r6 := 0x0
-// r7 := 0x0
-FORCE_INLINE __m128i _mm_sad_epu8(__m128i a, __m128i b)
-{
-	uint16x8_t t = vpaddlq_u8(vabdq_u8((uint8x16_t)a, (uint8x16_t)b));
-	uint16_t r0 = t[0] + t[1] + t[2] + t[3];
-	uint16_t r4 = t[4] + t[5] + t[6] + t[7];
-	uint16x8_t r = vsetq_lane_u16(r0, vdupq_n_u16(0), 0);
-	return (__m128i)vsetq_lane_u16(r4, r, 4);
-}
-
-// Divides the four single-precision, floating-point values of a and b.
-//
-// r0 := a0 / b0
-// r1 := a1 / b1
-// r2 := a2 / b2
-// r3 := a3 / b3
-//
-// https://msdn.microsoft.com/en-us/library/edaw8147(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_div_ps(__m128 a, __m128 b)
-{
-	float32x4_t recip0 = vrecpeq_f32(vreinterpretq_f32_m128(b));
-	float32x4_t recip1 = vmulq_f32(
-		recip0, vrecpsq_f32(recip0, vreinterpretq_f32_m128(b)));
-	return vreinterpretq_m128_f32(
-		vmulq_f32(vreinterpretq_f32_m128(a), recip1));
-}
-
-// Divides the scalar single-precision floating point value of a by b.
-// https://msdn.microsoft.com/en-us/library/4y73xa49(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_div_ss(__m128 a, __m128 b)
-{
-	float32_t value =
-		vgetq_lane_f32(vreinterpretq_f32_m128(_mm_div_ps(a, b)), 0);
-	return vreinterpretq_m128_f32(
-		vsetq_lane_f32(value, vreinterpretq_f32_m128(a), 0));
-}
-
-// Computes the approximations of reciprocals of the four single-precision,
-// floating-point values of a.
-// https://msdn.microsoft.com/en-us/library/vstudio/796k1tty(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_rcp_ps(__m128 in)
-{
-	float32x4_t recip = vrecpeq_f32(vreinterpretq_f32_m128(in));
-	recip = vmulq_f32(recip,
-			 vrecpsq_f32(recip, vreinterpretq_f32_m128(in)));
-	return vreinterpretq_m128_f32(recip);
-}
-
-// Computes the approximations of square roots of the four single-precision,
-// floating-point values of a. First computes reciprocal square roots and then
-// reciprocals of the four values.
-//
-// r0 := sqrt(a0)
-// r1 := sqrt(a1)
-// r2 := sqrt(a2)
-// r3 := sqrt(a3)
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/8z67bwwk(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_sqrt_ps(__m128 in)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128_f32(vsqrtq_f32(vreinterpretq_f32_m128(in)));
-#else
-	float32x4_t recipsq = vrsqrteq_f32(vreinterpretq_f32_m128(in));
-	float32x4_t sq = vrecpeq_f32(recipsq);
-	// ??? use step versions of both sqrt and recip for better accuracy?
-	return vreinterpretq_m128_f32(sq);
-#endif
-}
-
-// Computes the approximation of the square root of the scalar single-precision
-// floating point value of in.
-// https://msdn.microsoft.com/en-us/library/ahfsc22d(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_sqrt_ss(__m128 in)
-{
-	float32_t value =
-		vgetq_lane_f32(vreinterpretq_f32_m128(_mm_sqrt_ps(in)), 0);
-	return vreinterpretq_m128_f32(
-		vsetq_lane_f32(value, vreinterpretq_f32_m128(in), 0));
-}
-
-// Computes the approximations of the reciprocal square roots of the four
-// single-precision floating point values of in.
-// https://msdn.microsoft.com/en-us/library/22hfsh53(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_rsqrt_ps(__m128 in)
-{
-	return vreinterpretq_m128_f32(vrsqrteq_f32(vreinterpretq_f32_m128(in)));
-}
-
-// Compute the approximate reciprocal square root of the lower single-precision
-// (32-bit) floating-point element in a, store the result in the lower element
-// of dst, and copy the upper 3 packed elements from a to the upper elements of
-// dst.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_rsqrt_ss
-FORCE_INLINE __m128 _mm_rsqrt_ss(__m128 in)
-{
-	return vsetq_lane_f32(vgetq_lane_f32(_mm_rsqrt_ps(in), 0), in, 0);
-}
-
-// Computes the maximums of the four single-precision, floating-point values of
-// a and b.
-// https://msdn.microsoft.com/en-us/library/vstudio/ff5d607a(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_max_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_f32(vmaxq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Computes the minima of the four single-precision, floating-point values of a
-// and b.
-// https://msdn.microsoft.com/en-us/library/vstudio/wh13kadz(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_min_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_f32(vminq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Computes the maximum of the two lower scalar single-precision floating point
-// values of a and b.
-// https://msdn.microsoft.com/en-us/library/s6db5esz(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_max_ss(__m128 a, __m128 b)
-{
-	float32_t value = vgetq_lane_f32(vmaxq_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)),
-					 0);
-	return vreinterpretq_m128_f32(
-		vsetq_lane_f32(value, vreinterpretq_f32_m128(a), 0));
-}
-
-// Computes the minimum of the two lower scalar single-precision floating point
-// values of a and b.
-// https://msdn.microsoft.com/en-us/library/0a9y7xaa(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_min_ss(__m128 a, __m128 b)
-{
-	float32_t value = vgetq_lane_f32(vminq_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)),
-					 0);
-	return vreinterpretq_m128_f32(
-		vsetq_lane_f32(value, vreinterpretq_f32_m128(a), 0));
-}
-
-// Computes the pairwise maxima of the 16 unsigned 8-bit integers from a and the
-// 16 unsigned 8-bit integers from b.
-// https://msdn.microsoft.com/en-us/library/st6634za(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_max_epu8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vmaxq_u8(vreinterpretq_u8_m128i(a), vreinterpretq_u8_m128i(b)));
-}
-
-// Computes the pairwise minima of the 16 unsigned 8-bit integers from a and the
-// 16 unsigned 8-bit integers from b.
-// https://msdn.microsoft.com/ko-kr/library/17k8cf58(v=vs.100).aspxx
-FORCE_INLINE __m128i _mm_min_epu8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vminq_u8(vreinterpretq_u8_m128i(a), vreinterpretq_u8_m128i(b)));
-}
-
-// Computes the pairwise minima of the 8 signed 16-bit integers from a and the 8
-// signed 16-bit integers from b.
-// https://msdn.microsoft.com/en-us/library/vstudio/6te997ew(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_min_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vminq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Computes the pairwise maxima of the 8 signed 16-bit integers from a and the 8
-// signed 16-bit integers from b.
-// https://msdn.microsoft.com/en-us/LIBRary/3x060h7c(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_max_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(vmaxq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// epi versions of min/max
-// Computes the pariwise maximums of the four signed 32-bit integer values of a
-// and b.
-//
-// A 128-bit parameter that can be defined with the following equations:
-// r0 := (a0 > b0) ? a0 : b0
-// r1 := (a1 > b1) ? a1 : b1
-// r2 := (a2 > b2) ? a2 : b2
-// r3 := (a3 > b3) ? a3 : b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/bb514055(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_max_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vmaxq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Computes the pariwise minima of the four signed 32-bit integer values of a
-// and b.
-//
-// A 128-bit parameter that can be defined with the following equations:
-// r0 := (a0 < b0) ? a0 : b0
-// r1 := (a1 < b1) ? a1 : b1
-// r2 := (a2 < b2) ? a2 : b2
-// r3 := (a3 < b3) ? a3 : b3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/bb531476(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_min_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s32(vminq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Multiplies the 8 signed 16-bit integers from a by the 8 signed 16-bit
-// integers from b.
-//
-// r0 := (a0 * b0)[31:16]
-// r1 := (a1 * b1)[31:16]
-// ...
-// r7 := (a7 * b7)[31:16]
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/59hddw1d(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_mulhi_epi16(__m128i a, __m128i b)
-{
-	/* FIXME: issue with large values because of result saturation */
-	// int16x8_t ret = vqdmulhq_s16(vreinterpretq_s16_m128i(a),
-	// vreinterpretq_s16_m128i(b)); /* =2*a*b */ return
-	// vreinterpretq_m128i_s16(vshrq_n_s16(ret, 1));
-	int16x4_t a3210 = vget_low_s16(vreinterpretq_s16_m128i(a));
-	int16x4_t b3210 = vget_low_s16(vreinterpretq_s16_m128i(b));
-	int32x4_t ab3210 = vmull_s16(a3210, b3210); /* 3333222211110000 */
-	int16x4_t a7654 = vget_high_s16(vreinterpretq_s16_m128i(a));
-	int16x4_t b7654 = vget_high_s16(vreinterpretq_s16_m128i(b));
-	int32x4_t ab7654 = vmull_s16(a7654, b7654); /* 7777666655554444 */
-	uint16x8x2_t r = vuzpq_u16(vreinterpretq_u16_s32(ab3210),
-				 vreinterpretq_u16_s32(ab7654));
-	return vreinterpretq_m128i_u16(r.val[1]);
-}
-
-// Computes pairwise add of each argument as single-precision, floating-point
-// values a and b.
-// https://msdn.microsoft.com/en-us/library/yd9wecaa.aspx
-FORCE_INLINE __m128 _mm_hadd_ps(__m128 a, __m128 b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128_f32(vpaddq_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)));
-#else
-	float32x2_t a10 = vget_low_f32(vreinterpretq_f32_m128(a));
-	float32x2_t a32 = vget_high_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b10 = vget_low_f32(vreinterpretq_f32_m128(b));
-	float32x2_t b32 = vget_high_f32(vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_f32(
-		vcombine_f32(vpadd_f32(a10, a32), vpadd_f32(b10, b32)));
-#endif
-}
-
-// Computes pairwise add of each argument as a 16-bit signed or unsigned integer
-// values a and b.
-FORCE_INLINE __m128i _mm_hadd_epi16(__m128i _a, __m128i _b)
-{
-	int16x8_t a = vreinterpretq_s16_m128i(_a);
-	int16x8_t b = vreinterpretq_s16_m128i(_b);
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s16(vpaddq_s16(a, b));
-#else
-	return vreinterpretq_m128i_s16(
-		vcombine_s16(vpadd_s16(vget_low_s16(a), vget_high_s16(a)),
-			 vpadd_s16(vget_low_s16(b), vget_high_s16(b))));
-#endif
-}
-
-// Computes pairwise difference of each argument as a 16-bit signed or unsigned
-// integer values a and b.
-FORCE_INLINE __m128i _mm_hsub_epi16(__m128i _a, __m128i _b)
-{
-	int32x4_t a = vreinterpretq_s32_m128i(_a);
-	int32x4_t b = vreinterpretq_s32_m128i(_b);
-	// Interleave using vshrn/vmovn
-	// [a0|a2|a4|a6|b0|b2|b4|b6]
-	// [a1|a3|a5|a7|b1|b3|b5|b7]
-	int16x8_t ab0246 = vcombine_s16(vmovn_s32(a), vmovn_s32(b));
-	int16x8_t ab1357 = vcombine_s16(vshrn_n_s32(a, 16), vshrn_n_s32(b, 16));
-	// Subtract
-	return vreinterpretq_m128i_s16(vsubq_s16(ab0246, ab1357));
-}
-
-// Computes saturated pairwise sub of each argument as a 16-bit signed
-// integer values a and b.
-FORCE_INLINE __m128i _mm_hadds_epi16(__m128i _a, __m128i _b)
-{
-	int32x4_t a = vreinterpretq_s32_m128i(_a);
-	int32x4_t b = vreinterpretq_s32_m128i(_b);
-	// Interleave using vshrn/vmovn
-	// [a0|a2|a4|a6|b0|b2|b4|b6]
-	// [a1|a3|a5|a7|b1|b3|b5|b7]
-	int16x8_t ab0246 = vcombine_s16(vmovn_s32(a), vmovn_s32(b));
-	int16x8_t ab1357 = vcombine_s16(vshrn_n_s32(a, 16), vshrn_n_s32(b, 16));
-	// Saturated add
-	return vreinterpretq_m128i_s16(vqaddq_s16(ab0246, ab1357));
-}
-
-// Computes saturated pairwise difference of each argument as a 16-bit signed
-// integer values a and b.
-FORCE_INLINE __m128i _mm_hsubs_epi16(__m128i _a, __m128i _b)
-{
-	int32x4_t a = vreinterpretq_s32_m128i(_a);
-	int32x4_t b = vreinterpretq_s32_m128i(_b);
-	// Interleave using vshrn/vmovn
-	// [a0|a2|a4|a6|b0|b2|b4|b6]
-	// [a1|a3|a5|a7|b1|b3|b5|b7]
-	int16x8_t ab0246 = vcombine_s16(vmovn_s32(a), vmovn_s32(b));
-	int16x8_t ab1357 = vcombine_s16(vshrn_n_s32(a, 16), vshrn_n_s32(b, 16));
-	// Saturated subtract
-	return vreinterpretq_m128i_s16(vqsubq_s16(ab0246, ab1357));
-}
-
-// Computes pairwise add of each argument as a 32-bit signed or unsigned integer
-// values a and b.
-FORCE_INLINE __m128i _mm_hadd_epi32(__m128i _a, __m128i _b)
-{
-	int32x4_t a = vreinterpretq_s32_m128i(_a);
-	int32x4_t b = vreinterpretq_s32_m128i(_b);
-	return vreinterpretq_m128i_s32(
-		vcombine_s32(vpadd_s32(vget_low_s32(a), vget_high_s32(a)),
-			 vpadd_s32(vget_low_s32(b), vget_high_s32(b))));
-}
-
-// Computes pairwise difference of each argument as a 32-bit signed or unsigned
-// integer values a and b.
-FORCE_INLINE __m128i _mm_hsub_epi32(__m128i _a, __m128i _b)
-{
-	int64x2_t a = vreinterpretq_s64_m128i(_a);
-	int64x2_t b = vreinterpretq_s64_m128i(_b);
-	// Interleave using vshrn/vmovn
-	// [a0|a2|b0|b2]
-	// [a1|a2|b1|b3]
-	int32x4_t ab02 = vcombine_s32(vmovn_s64(a), vmovn_s64(b));
-	int32x4_t ab13 = vcombine_s32(vshrn_n_s64(a, 32), vshrn_n_s64(b, 32));
-	// Subtract
-	return vreinterpretq_m128i_s32(vsubq_s32(ab02, ab13));
-}
-
-/* Compare operations */
-
-// Compares for less than
-// https://msdn.microsoft.com/en-us/library/vstudio/f330yhc8(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmplt_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vcltq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Compares for greater than.
-//
-// r0 := (a0 > b0) ? 0xffffffff : 0x0
-// r1 := (a1 > b1) ? 0xffffffff : 0x0
-// r2 := (a2 > b2) ? 0xffffffff : 0x0
-// r3 := (a3 > b3) ? 0xffffffff : 0x0
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/11dy102s(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmpgt_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vcgtq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Compares for greater than or equal.
-// https://msdn.microsoft.com/en-us/library/vstudio/fs813y2t(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmpge_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vcgeq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Compares for less than or equal.
-//
-// r0 := (a0 <= b0) ? 0xffffffff : 0x0
-// r1 := (a1 <= b1) ? 0xffffffff : 0x0
-// r2 := (a2 <= b2) ? 0xffffffff : 0x0
-// r3 := (a3 <= b3) ? 0xffffffff : 0x0
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/1s75w83z(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmple_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vcleq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Compares for equality.
-// https://msdn.microsoft.com/en-us/library/vstudio/36aectz5(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cmpeq_ps(__m128 a, __m128 b)
-{
-	return vreinterpretq_m128_u32(vceqq_f32(vreinterpretq_f32_m128(a),
-						vreinterpretq_f32_m128(b)));
-}
-
-// Compares the 16 signed or unsigned 8-bit integers in a and the 16 signed or
-// unsigned 8-bit integers in b for equality.
-// https://msdn.microsoft.com/en-us/library/windows/desktop/bz5xk21a(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_cmpeq_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vceqq_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b)));
-}
-
-// Compares the 8 signed or unsigned 16-bit integers in a and the 8 signed or
-// unsigned 16-bit integers in b for equality.
-// https://msdn.microsoft.com/en-us/library/2ay060te(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmpeq_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(vceqq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Compare packed 32-bit integers in a and b for equality, and store the results
-// in dst
-FORCE_INLINE __m128i _mm_cmpeq_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u32(vceqq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Compare packed 64-bit integers in a and b for equality, and store the results
-// in dst
-FORCE_INLINE __m128i _mm_cmpeq_epi64(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_u64(vceqq_u64(vreinterpretq_u64_m128i(a),
-						 vreinterpretq_u64_m128i(b)));
-#else
-	// ARMv7 lacks vceqq_u64
-	// (a == b) -> (a_lo == b_lo) && (a_hi == b_hi)
-	uint32x4_t cmp = vceqq_u32(vreinterpretq_u32_m128i(a),
-				 vreinterpretq_u32_m128i(b));
-	uint32x4_t swapped = vrev64q_u32(cmp);
-	return vreinterpretq_m128i_u32(vandq_u32(cmp, swapped));
-#endif
-}
-
-// Compares the 16 signed 8-bit integers in a and the 16 signed 8-bit integers
-// in b for lesser than.
-// https://msdn.microsoft.com/en-us/library/windows/desktop/9s46csht(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_cmplt_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vcltq_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b)));
-}
-
-// Compares the 16 signed 8-bit integers in a and the 16 signed 8-bit integers
-// in b for greater than.
-//
-// r0 := (a0 > b0) ? 0xff : 0x0
-// r1 := (a1 > b1) ? 0xff : 0x0
-// ...
-// r15 := (a15 > b15) ? 0xff : 0x0
-//
-// https://msdn.microsoft.com/zh-tw/library/wf45zt2b(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmpgt_epi8(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vcgtq_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b)));
-}
-
-// Compares the 8 signed 16-bit integers in a and the 8 signed 16-bit integers
-// in b for less than.
-//
-// r0 := (a0 < b0) ? 0xffff : 0x0
-// r1 := (a1 < b1) ? 0xffff : 0x0
-// ...
-// r7 := (a7 < b7) ? 0xffff : 0x0
-//
-// https://technet.microsoft.com/en-us/library/t863edb2(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmplt_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(vcltq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Compares the 8 signed 16-bit integers in a and the 8 signed 16-bit integers
-// in b for greater than.
-//
-// r0 := (a0 > b0) ? 0xffff : 0x0
-// r1 := (a1 > b1) ? 0xffff : 0x0
-// ...
-// r7 := (a7 > b7) ? 0xffff : 0x0
-//
-// https://technet.microsoft.com/en-us/library/xd43yfsa(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmpgt_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(vcgtq_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-}
-
-// Compares the 4 signed 32-bit integers in a and the 4 signed 32-bit integers
-// in b for less than.
-// https://msdn.microsoft.com/en-us/library/vstudio/4ak0bf5d(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmplt_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u32(vcltq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Compares the 4 signed 32-bit integers in a and the 4 signed 32-bit integers
-// in b for greater than.
-// https://msdn.microsoft.com/en-us/library/vstudio/1s9f2z0y(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cmpgt_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u32(vcgtq_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-}
-
-// Compares the 2 signed 64-bit integers in a and the 2 signed 64-bit integers
-// in b for greater than.
-FORCE_INLINE __m128i _mm_cmpgt_epi64(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_u64(vcgtq_s64(vreinterpretq_s64_m128i(a),
-						 vreinterpretq_s64_m128i(b)));
-#else
-	// ARMv7 lacks vcgtq_s64.
-	// This is based off of Clang's SSE2 polyfill:
-	// (a > b) -> ((a_hi > b_hi) || (a_lo > b_lo && a_hi == b_hi))
-
-	// Mask the sign bit out since we need a signed AND an unsigned comparison
-	// and it is ugly to try and split them.
-	int32x4_t mask = vreinterpretq_s32_s64(vdupq_n_s64(0x80000000ull));
-	int32x4_t a_mask = veorq_s32(vreinterpretq_s32_m128i(a), mask);
-	int32x4_t b_mask = veorq_s32(vreinterpretq_s32_m128i(b), mask);
-	// Check if a > b
-	int64x2_t greater = vreinterpretq_s64_u32(vcgtq_s32(a_mask, b_mask));
-	// Copy upper mask to lower mask
-	// a_hi > b_hi
-	int64x2_t gt_hi = vshrq_n_s64(greater, 63);
-	// Copy lower mask to upper mask
-	// a_lo > b_lo
-	int64x2_t gt_lo = vsliq_n_s64(greater, greater, 32);
-	// Compare for equality
-	int64x2_t equal = vreinterpretq_s64_u32(vceqq_s32(a_mask, b_mask));
-	// Copy upper mask to lower mask
-	// a_hi == b_hi
-	int64x2_t eq_hi = vshrq_n_s64(equal, 63);
-	// a_hi > b_hi || (a_lo > b_lo && a_hi == b_hi)
-	int64x2_t ret = vorrq_s64(gt_hi, vandq_s64(gt_lo, eq_hi));
-	return vreinterpretq_m128i_s64(ret);
-#endif
-}
-
-// Compares the four 32-bit floats in a and b to check if any values are NaN.
-// Ordered compare between each value returns true for "orderable" and false for
-// "not orderable" (NaN).
-// https://msdn.microsoft.com/en-us/library/vstudio/0h9w00fx(v=vs.100).aspx see
-// also:
-// http://stackoverflow.com/questions/8627331/what-does-ordered-unordered-comparison-mean
-// http://stackoverflow.com/questions/29349621/neon-isnanval-intrinsics
-FORCE_INLINE __m128 _mm_cmpord_ps(__m128 a, __m128 b)
-{
-	// Note: NEON does not have ordered compare builtin
-	// Need to compare a eq a and b eq b to check for NaN
-	// Do AND of results to get final
-	uint32x4_t ceqaa =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t ceqbb =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	return vreinterpretq_m128_u32(vandq_u32(ceqaa, ceqbb));
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using a less than operation. :
-// https://msdn.microsoft.com/en-us/library/2kwe606b(v=vs.90).aspx Important
-// note!! The documentation on MSDN is incorrect! If either of the values is a
-// NAN the docs say you will get a one, but in fact, it will return a zero!!
-FORCE_INLINE int _mm_comilt_ss(__m128 a, __m128 b)
-{
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_lt_b =
-		vcltq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b));
-	return (vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_lt_b), 0) != 0) ? 1
-									 : 0;
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using a greater than operation. :
-// https://msdn.microsoft.com/en-us/library/b0738e0t(v=vs.100).aspx
-FORCE_INLINE int _mm_comigt_ss(__m128 a, __m128 b)
-{
-	// return vgetq_lane_u32(vcgtq_f32(vreinterpretq_f32_m128(a),
-	// vreinterpretq_f32_m128(b)), 0);
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_gt_b =
-		vcgtq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b));
-	return (vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0) ? 1
-									 : 0;
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using a less than or equal operation. :
-// https://msdn.microsoft.com/en-us/library/1w4t7c57(v=vs.90).aspx
-FORCE_INLINE int _mm_comile_ss(__m128 a, __m128 b)
-{
-	// return vgetq_lane_u32(vcleq_f32(vreinterpretq_f32_m128(a),
-	// vreinterpretq_f32_m128(b)), 0);
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_le_b =
-		vcleq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b));
-	return (vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_le_b), 0) != 0) ? 1
-									 : 0;
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using a greater than or equal operation. :
-// https://msdn.microsoft.com/en-us/library/8t80des6(v=vs.100).aspx
-FORCE_INLINE int _mm_comige_ss(__m128 a, __m128 b)
-{
-	// return vgetq_lane_u32(vcgeq_f32(vreinterpretq_f32_m128(a),
-	// vreinterpretq_f32_m128(b)), 0);
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_ge_b =
-		vcgeq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b));
-	return (vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0) ? 1
-									 : 0;
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using an equality operation. :
-// https://msdn.microsoft.com/en-us/library/93yx2h2b(v=vs.100).aspx
-FORCE_INLINE int _mm_comieq_ss(__m128 a, __m128 b)
-{
-	// return vgetq_lane_u32(vceqq_f32(vreinterpretq_f32_m128(a),
-	// vreinterpretq_f32_m128(b)), 0);
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
-	uint32x4_t a_eq_b =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b));
-	return (vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_eq_b), 0) != 0) ? 1
-									 : 0;
-}
-
-// Compares the lower single-precision floating point scalar values of a and b
-// using an inequality operation. :
-// https://msdn.microsoft.com/en-us/library/bafh5e0a(v=vs.90).aspx
-FORCE_INLINE int _mm_comineq_ss(__m128 a, __m128 b)
-{
-	// return !vgetq_lane_u32(vceqq_f32(vreinterpretq_f32_m128(a),
-	// vreinterpretq_f32_m128(b)), 0);
-	uint32x4_t a_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(a));
-	uint32x4_t b_not_nan =
-		vceqq_f32(vreinterpretq_f32_m128(b), vreinterpretq_f32_m128(b));
-	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
-	uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)));
-	return (vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_neq_b), 0) != 0) ? 1 : 0;
-}
-
-// according to the documentation, these intrinsics behave the same as the
-// non-'u' versions. We'll just alias them here.
-#define _mm_ucomilt_ss _mm_comilt_ss
-#define _mm_ucomile_ss _mm_comile_ss
-#define _mm_ucomigt_ss _mm_comigt_ss
-#define _mm_ucomige_ss _mm_comige_ss
-#define _mm_ucomieq_ss _mm_comieq_ss
-#define _mm_ucomineq_ss _mm_comineq_ss
-
-/* Conversions */
-
-// Converts the four single-precision, floating-point values of a to signed
-// 32-bit integer values using truncate.
-// https://msdn.microsoft.com/en-us/library/vstudio/1h005y6x(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_cvttps_epi32(__m128 a)
-{
-	return vreinterpretq_m128i_s32(
-		vcvtq_s32_f32(vreinterpretq_f32_m128(a)));
-}
-
-// Converts the four signed 32-bit integer values of a to single-precision,
-// floating-point values
-// https://msdn.microsoft.com/en-us/library/vstudio/36bwxcx5(v=vs.100).aspx
-FORCE_INLINE __m128 _mm_cvtepi32_ps(__m128i a)
-{
-	return vreinterpretq_m128_f32(
-		vcvtq_f32_s32(vreinterpretq_s32_m128i(a)));
-}
-
-// Converts the four unsigned 8-bit integers in the lower 16 bits to four
-// unsigned 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepu8_epi16(__m128i a)
-{
-	uint8x16_t u8x16 = vreinterpretq_u8_m128i(a); /* xxxx xxxx xxxx DCBA */
-	uint16x8_t u16x8 =
-		vmovl_u8(vget_low_u8(u8x16)); /* 0x0x 0x0x 0D0C 0B0A */
-	return vreinterpretq_m128i_u16(u16x8);
-}
-
-// Converts the four unsigned 8-bit integers in the lower 32 bits to four
-// unsigned 32-bit integers.
-// https://msdn.microsoft.com/en-us/library/bb531467%28v=vs.100%29.aspx
-FORCE_INLINE __m128i _mm_cvtepu8_epi32(__m128i a)
-{
-	uint8x16_t u8x16 = vreinterpretq_u8_m128i(a); /* xxxx xxxx xxxx DCBA */
-	uint16x8_t u16x8 =
-		vmovl_u8(vget_low_u8(u8x16)); /* 0x0x 0x0x 0D0C 0B0A */
-	uint32x4_t u32x4 =
-		vmovl_u16(vget_low_u16(u16x8)); /* 000D 000C 000B 000A */
-	return vreinterpretq_m128i_u32(u32x4);
-}
-
-// Converts the two unsigned 8-bit integers in the lower 16 bits to two
-// unsigned 64-bit integers.
-FORCE_INLINE __m128i _mm_cvtepu8_epi64(__m128i a)
-{
-	uint8x16_t u8x16 = vreinterpretq_u8_m128i(a); /* xxxx xxxx xxxx xxBA */
-	uint16x8_t u16x8 =
-		vmovl_u8(vget_low_u8(u8x16)); /* 0x0x 0x0x 0x0x 0B0A */
-	uint32x4_t u32x4 =
-		vmovl_u16(vget_low_u16(u16x8)); /* 000x 000x 000B 000A */
-	uint64x2_t u64x2 =
-		vmovl_u32(vget_low_u32(u32x4)); /* 0000 000B 0000 000A */
-	return vreinterpretq_m128i_u64(u64x2);
-}
-
-// Converts the four unsigned 8-bit integers in the lower 16 bits to four
-// unsigned 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi8_epi16(__m128i a)
-{
-	int8x16_t s8x16 = vreinterpretq_s8_m128i(a); /* xxxx xxxx xxxx DCBA */
-	int16x8_t s16x8 =
-		vmovl_s8(vget_low_s8(s8x16)); /* 0x0x 0x0x 0D0C 0B0A */
-	return vreinterpretq_m128i_s16(s16x8);
-}
-
-// Converts the four unsigned 8-bit integers in the lower 32 bits to four
-// unsigned 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi8_epi32(__m128i a)
-{
-	int8x16_t s8x16 = vreinterpretq_s8_m128i(a); /* xxxx xxxx xxxx DCBA */
-	int16x8_t s16x8 =
-		vmovl_s8(vget_low_s8(s8x16)); /* 0x0x 0x0x 0D0C 0B0A */
-	int32x4_t s32x4 =
-		vmovl_s16(vget_low_s16(s16x8)); /* 000D 000C 000B 000A */
-	return vreinterpretq_m128i_s32(s32x4);
-}
-
-// Converts the two signed 8-bit integers in the lower 32 bits to four
-// signed 64-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi8_epi64(__m128i a)
-{
-	int8x16_t s8x16 = vreinterpretq_s8_m128i(a); /* xxxx xxxx xxxx xxBA */
-	int16x8_t s16x8 =
-		vmovl_s8(vget_low_s8(s8x16)); /* 0x0x 0x0x 0x0x 0B0A */
-	int32x4_t s32x4 =
-		vmovl_s16(vget_low_s16(s16x8)); /* 000x 000x 000B 000A */
-	int64x2_t s64x2 =
-		vmovl_s32(vget_low_s32(s32x4)); /* 0000 000B 0000 000A */
-	return vreinterpretq_m128i_s64(s64x2);
-}
-
-// Converts the four signed 16-bit integers in the lower 64 bits to four signed
-// 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi16_epi32(__m128i a)
-{
-	return vreinterpretq_m128i_s32(
-		vmovl_s16(vget_low_s16(vreinterpretq_s16_m128i(a))));
-}
-
-// Converts the two signed 16-bit integers in the lower 32 bits two signed
-// 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi16_epi64(__m128i a)
-{
-	int16x8_t s16x8 = vreinterpretq_s16_m128i(a); /* xxxx xxxx xxxx 0B0A */
-	int32x4_t s32x4 =
-		vmovl_s16(vget_low_s16(s16x8)); /* 000x 000x 000B 000A */
-	int64x2_t s64x2 =
-		vmovl_s32(vget_low_s32(s32x4)); /* 0000 000B 0000 000A */
-	return vreinterpretq_m128i_s64(s64x2);
-}
-
-// Converts the four unsigned 16-bit integers in the lower 64 bits to four
-// unsigned 32-bit integers.
-FORCE_INLINE __m128i _mm_cvtepu16_epi32(__m128i a)
-{
-	return vreinterpretq_m128i_u32(
-		vmovl_u16(vget_low_u16(vreinterpretq_u16_m128i(a))));
-}
-
-// Converts the two unsigned 16-bit integers in the lower 32 bits to two
-// unsigned 64-bit integers.
-FORCE_INLINE __m128i _mm_cvtepu16_epi64(__m128i a)
-{
-	uint16x8_t u16x8 = vreinterpretq_u16_m128i(a); /* xxxx xxxx xxxx 0B0A */
-	uint32x4_t u32x4 =
-		vmovl_u16(vget_low_u16(u16x8)); /* 000x 000x 000B 000A */
-	uint64x2_t u64x2 =
-		vmovl_u32(vget_low_u32(u32x4)); /* 0000 000B 0000 000A */
-	return vreinterpretq_m128i_u64(u64x2);
-}
-
-// Converts the two unsigned 32-bit integers in the lower 64 bits to two
-// unsigned 64-bit integers.
-FORCE_INLINE __m128i _mm_cvtepu32_epi64(__m128i a)
-{
-	return vreinterpretq_m128i_u64(
-		vmovl_u32(vget_low_u32(vreinterpretq_u32_m128i(a))));
-}
-
-// Converts the two signed 32-bit integers in the lower 64 bits to two signed
-// 64-bit integers.
-FORCE_INLINE __m128i _mm_cvtepi32_epi64(__m128i a)
-{
-	return vreinterpretq_m128i_s64(
-		vmovl_s32(vget_low_s32(vreinterpretq_s32_m128i(a))));
-}
-
-// Converts the four single-precision, floating-point values of a to signed
-// 32-bit integer values.
-//
-// r0 := (int) a0
-// r1 := (int) a1
-// r2 := (int) a2
-// r3 := (int) a3
-//
-// https://msdn.microsoft.com/en-us/library/vstudio/xdc42k5e(v=vs.100).aspx
-// *NOTE*. The default rounding mode on SSE is 'round to even', which ARMv7-A
-// does not support! It is supported on ARMv8-A however.
-FORCE_INLINE __m128i _mm_cvtps_epi32(__m128 a)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s32(vcvtnq_s32_f32(a));
-#else
-	uint32x4_t signmask = vdupq_n_u32(0x80000000);
-	float32x4_t half = vbslq_f32(signmask, vreinterpretq_f32_m128(a),
-				 vdupq_n_f32(0.5f)); /* +/- 0.5 */
-	int32x4_t r_normal =
-		vcvtq_s32_f32(vaddq_f32(vreinterpretq_f32_m128(a),
-					half)); /* round to integer: [a + 0.5]*/
-	int32x4_t r_trunc = vcvtq_s32_f32(
-		vreinterpretq_f32_m128(a)); /* truncate to integer: [a] */
-	int32x4_t plusone = vreinterpretq_s32_u32(vshrq_n_u32(
-		vreinterpretq_u32_s32(vnegq_s32(r_trunc)), 31)); /* 1 or 0 */
-	int32x4_t r_even = vbicq_s32(vaddq_s32(r_trunc, plusone),
-				 vdupq_n_s32(1)); /* ([a] + {0,1}) & ~1 */
-	float32x4_t delta = vsubq_f32(
-		vreinterpretq_f32_m128(a),
-		vcvtq_f32_s32(r_trunc)); /* compute delta: delta = (a - [a]) */
-	uint32x4_t is_delta_half =
-		vceqq_f32(delta, half); /* delta == +/- 0.5 */
-	return vreinterpretq_m128i_s32(
-		vbslq_s32(is_delta_half, r_even, r_normal));
-#endif
-}
-
-// Moves the least significant 32 bits of a to a 32-bit integer.
-// https://msdn.microsoft.com/en-us/library/5z7a9642%28v=vs.90%29.aspx
-FORCE_INLINE int _mm_cvtsi128_si32(__m128i a)
-{
-	return vgetq_lane_s32(vreinterpretq_s32_m128i(a), 0);
-}
-
-// Extracts the low order 64-bit integer from the parameter.
-// https://msdn.microsoft.com/en-us/library/bb531384(v=vs.120).aspx
-FORCE_INLINE uint64_t _mm_cvtsi128_si64(__m128i a)
-{
-	return vgetq_lane_s64(vreinterpretq_s64_m128i(a), 0);
-}
-
-// Moves 32-bit integer a to the least significant 32 bits of an __m128 object,
-// zero extending the upper bits.
-//
-// r0 := a
-// r1 := 0x0
-// r2 := 0x0
-// r3 := 0x0
-//
-// https://msdn.microsoft.com/en-us/library/ct3539ha%28v=vs.90%29.aspx
-FORCE_INLINE __m128i _mm_cvtsi32_si128(int a)
-{
-	return vreinterpretq_m128i_s32(vsetq_lane_s32(a, vdupq_n_s32(0), 0));
-}
-
-// Moves 64-bit integer a to the least significant 64 bits of an __m128 object,
-// zero extending the upper bits.
-//
-// r0 := a
-// r1 := 0x0
-FORCE_INLINE __m128i _mm_cvtsi64_si128(int64_t a)
-{
-	return vreinterpretq_m128i_s64(vsetq_lane_s64(a, vdupq_n_s64(0), 0));
-}
-
-// Applies a type cast to reinterpret four 32-bit floating point values passed
-// in as a 128-bit parameter as packed 32-bit integers.
-// https://msdn.microsoft.com/en-us/library/bb514099.aspx
-FORCE_INLINE __m128i _mm_castps_si128(__m128 a)
-{
-	return vreinterpretq_m128i_s32(vreinterpretq_s32_m128(a));
-}
-
-// Applies a type cast to reinterpret four 32-bit integers passed in as a
-// 128-bit parameter as packed 32-bit floating point values.
-// https://msdn.microsoft.com/en-us/library/bb514029.aspx
-FORCE_INLINE __m128 _mm_castsi128_ps(__m128i a)
-{
-	return vreinterpretq_m128_s32(vreinterpretq_s32_m128i(a));
-}
-
-// Loads 128-bit value. :
-// https://msdn.microsoft.com/en-us/library/atzzad1h(v=vs.80).aspx
-FORCE_INLINE __m128i _mm_load_si128(const __m128i *p)
-{
-	return vreinterpretq_m128i_s32(vld1q_s32((const int32_t *)p));
-}
-
-// Loads 128-bit value. :
-// https://msdn.microsoft.com/zh-cn/library/f4k12ae8(v=vs.90).aspx
-FORCE_INLINE __m128i _mm_loadu_si128(const __m128i *p)
-{
-	return vreinterpretq_m128i_s32(vld1q_s32((const int32_t *)p));
-}
-
-// _mm_lddqu_si128 functions the same as _mm_loadu_si128.
-#define _mm_lddqu_si128 _mm_loadu_si128
-
-/* Miscellaneous Operations */
-
-// Shifts the 8 signed 16-bit integers in a right by count bits while shifting
-// in the sign bit.
-//
-// r0 := a0 >> count
-// r1 := a1 >> count
-// ...
-// r7 := a7 >> count
-//
-// https://msdn.microsoft.com/en-us/library/3c9997dk(v%3dvs.90).aspx
-FORCE_INLINE __m128i _mm_sra_epi16(__m128i a, __m128i count)
-{
-	int64_t c = (int64_t)vget_low_s64((int64x2_t)count);
-	if (c > 15)
-		return _mm_cmplt_epi16(a, _mm_setzero_si128());
-	return vreinterpretq_m128i_s16(
-		vshlq_s16((int16x8_t)a, vdupq_n_s16(-c)));
-}
-
-// Shifts the 4 signed 32-bit integers in a right by count bits while shifting
-// in the sign bit.
-//
-// r0 := a0 >> count
-// r1 := a1 >> count
-// r2 := a2 >> count
-// r3 := a3 >> count
-//
-// https://msdn.microsoft.com/en-us/library/ce40009e(v%3dvs.100).aspx
-FORCE_INLINE __m128i _mm_sra_epi32(__m128i a, __m128i count)
-{
-	int64_t c = (int64_t)vget_low_s64((int64x2_t)count);
-	if (c > 31)
-		return _mm_cmplt_epi32(a, _mm_setzero_si128());
-	return vreinterpretq_m128i_s32(
-		vshlq_s32((int32x4_t)a, vdupq_n_s32(-c)));
-}
-
-// Packs the 16 signed 16-bit integers from a and b into 8-bit integers and
-// saturates.
-// https://msdn.microsoft.com/en-us/library/k4y4f7w5%28v=vs.90%29.aspx
-FORCE_INLINE __m128i _mm_packs_epi16(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s8(
-		vcombine_s8(vqmovn_s16(vreinterpretq_s16_m128i(a)),
-			 vqmovn_s16(vreinterpretq_s16_m128i(b))));
-}
-
-// Packs the 16 signed 16 - bit integers from a and b into 8 - bit unsigned
-// integers and saturates.
-//
-// r0 := UnsignedSaturate(a0)
-// r1 := UnsignedSaturate(a1)
-// ...
-// r7 := UnsignedSaturate(a7)
-// r8 := UnsignedSaturate(b0)
-// r9 := UnsignedSaturate(b1)
-// ...
-// r15 := UnsignedSaturate(b7)
-//
-// https://msdn.microsoft.com/en-us/library/07ad1wx4(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_packus_epi16(const __m128i a, const __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vcombine_u8(vqmovun_s16(vreinterpretq_s16_m128i(a)),
-			 vqmovun_s16(vreinterpretq_s16_m128i(b))));
-}
-
-// Packs the 8 signed 32-bit integers from a and b into signed 16-bit integers
-// and saturates.
-//
-// r0 := SignedSaturate(a0)
-// r1 := SignedSaturate(a1)
-// r2 := SignedSaturate(a2)
-// r3 := SignedSaturate(a3)
-// r4 := SignedSaturate(b0)
-// r5 := SignedSaturate(b1)
-// r6 := SignedSaturate(b2)
-// r7 := SignedSaturate(b3)
-//
-// https://msdn.microsoft.com/en-us/library/393t56f9%28v=vs.90%29.aspx
-FORCE_INLINE __m128i _mm_packs_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_s16(
-		vcombine_s16(vqmovn_s32(vreinterpretq_s32_m128i(a)),
-			 vqmovn_s32(vreinterpretq_s32_m128i(b))));
-}
-
-// Packs the 8 unsigned 32-bit integers from a and b into unsigned 16-bit
-// integers and saturates.
-//
-// r0 := UnsignedSaturate(a0)
-// r1 := UnsignedSaturate(a1)
-// r2 := UnsignedSaturate(a2)
-// r3 := UnsignedSaturate(a3)
-// r4 := UnsignedSaturate(b0)
-// r5 := UnsignedSaturate(b1)
-// r6 := UnsignedSaturate(b2)
-// r7 := UnsignedSaturate(b3)
-FORCE_INLINE __m128i _mm_packus_epi32(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u16(
-		vcombine_u16(vqmovn_u32(vreinterpretq_u32_m128i(a)),
-			 vqmovn_u32(vreinterpretq_u32_m128i(b))));
-}
-
-// Interleaves the lower 8 signed or unsigned 8-bit integers in a with the lower
-// 8 signed or unsigned 8-bit integers in b.
-//
-// r0 := a0
-// r1 := b0
-// r2 := a1
-// r3 := b1
-// ...
-// r14 := a7
-// r15 := b7
-//
-// https://msdn.microsoft.com/en-us/library/xf7k860c%28v=vs.90%29.aspx
-FORCE_INLINE __m128i _mm_unpacklo_epi8(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s8(vzip1q_s8(vreinterpretq_s8_m128i(a),
-						vreinterpretq_s8_m128i(b)));
-#else
-	int8x8_t a1 =
-		vreinterpret_s8_s16(vget_low_s16(vreinterpretq_s16_m128i(a)));
-	int8x8_t b1 =
-		vreinterpret_s8_s16(vget_low_s16(vreinterpretq_s16_m128i(b)));
-	int8x8x2_t result = vzip_s8(a1, b1);
-	return vreinterpretq_m128i_s8(
-		vcombine_s8(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the lower 4 signed or unsigned 16-bit integers in a with the
-// lower 4 signed or unsigned 16-bit integers in b.
-//
-// r0 := a0
-// r1 := b0
-// r2 := a1
-// r3 := b1
-// r4 := a2
-// r5 := b2
-// r6 := a3
-// r7 := b3
-//
-// https://msdn.microsoft.com/en-us/library/btxb17bw%28v=vs.90%29.aspx
-FORCE_INLINE __m128i _mm_unpacklo_epi16(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s16(vzip1q_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-#else
-	int16x4_t a1 = vget_low_s16(vreinterpretq_s16_m128i(a));
-	int16x4_t b1 = vget_low_s16(vreinterpretq_s16_m128i(b));
-	int16x4x2_t result = vzip_s16(a1, b1);
-	return vreinterpretq_m128i_s16(
-		vcombine_s16(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the lower 2 signed or unsigned 32 - bit integers in a with the
-// lower 2 signed or unsigned 32 - bit integers in b.
-//
-// r0 := a0
-// r1 := b0
-// r2 := a1
-// r3 := b1
-//
-// https://msdn.microsoft.com/en-us/library/x8atst9d(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_unpacklo_epi32(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s32(vzip1q_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-#else
-	int32x2_t a1 = vget_low_s32(vreinterpretq_s32_m128i(a));
-	int32x2_t b1 = vget_low_s32(vreinterpretq_s32_m128i(b));
-	int32x2x2_t result = vzip_s32(a1, b1);
-	return vreinterpretq_m128i_s32(
-		vcombine_s32(result.val[0], result.val[1]));
-#endif
-}
-
-FORCE_INLINE __m128i _mm_unpacklo_epi64(__m128i a, __m128i b)
-{
-	int64x1_t a_l = vget_low_s64(vreinterpretq_s64_m128i(a));
-	int64x1_t b_l = vget_low_s64(vreinterpretq_s64_m128i(b));
-	return vreinterpretq_m128i_s64(vcombine_s64(a_l, b_l));
-}
-
-// Selects and interleaves the lower two single-precision, floating-point values
-// from a and b.
-//
-// r0 := a0
-// r1 := b0
-// r2 := a1
-// r3 := b1
-//
-// https://msdn.microsoft.com/en-us/library/25st103b%28v=vs.90%29.aspx
-FORCE_INLINE __m128 _mm_unpacklo_ps(__m128 a, __m128 b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128_f32(vzip1q_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)));
-#else
-	float32x2_t a1 = vget_low_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b1 = vget_low_f32(vreinterpretq_f32_m128(b));
-	float32x2x2_t result = vzip_f32(a1, b1);
-	return vreinterpretq_m128_f32(
-		vcombine_f32(result.val[0], result.val[1]));
-#endif
-}
-
-// Selects and interleaves the upper two single-precision, floating-point values
-// from a and b.
-//
-// r0 := a2
-// r1 := b2
-// r2 := a3
-// r3 := b3
-//
-// https://msdn.microsoft.com/en-us/library/skccxx7d%28v=vs.90%29.aspx
-FORCE_INLINE __m128 _mm_unpackhi_ps(__m128 a, __m128 b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128_f32(vzip2q_f32(vreinterpretq_f32_m128(a),
-						 vreinterpretq_f32_m128(b)));
-#else
-	float32x2_t a1 = vget_high_f32(vreinterpretq_f32_m128(a));
-	float32x2_t b1 = vget_high_f32(vreinterpretq_f32_m128(b));
-	float32x2x2_t result = vzip_f32(a1, b1);
-	return vreinterpretq_m128_f32(
-		vcombine_f32(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the upper 8 signed or unsigned 8-bit integers in a with the upper
-// 8 signed or unsigned 8-bit integers in b.
-//
-// r0 := a8
-// r1 := b8
-// r2 := a9
-// r3 := b9
-// ...
-// r14 := a15
-// r15 := b15
-//
-// https://msdn.microsoft.com/en-us/library/t5h7783k(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_unpackhi_epi8(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s8(vzip2q_s8(vreinterpretq_s8_m128i(a),
-						vreinterpretq_s8_m128i(b)));
-#else
-	int8x8_t a1 =
-		vreinterpret_s8_s16(vget_high_s16(vreinterpretq_s16_m128i(a)));
-	int8x8_t b1 =
-		vreinterpret_s8_s16(vget_high_s16(vreinterpretq_s16_m128i(b)));
-	int8x8x2_t result = vzip_s8(a1, b1);
-	return vreinterpretq_m128i_s8(
-		vcombine_s8(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the upper 4 signed or unsigned 16-bit integers in a with the
-// upper 4 signed or unsigned 16-bit integers in b.
-//
-// r0 := a4
-// r1 := b4
-// r2 := a5
-// r3 := b5
-// r4 := a6
-// r5 := b6
-// r6 := a7
-// r7 := b7
-//
-// https://msdn.microsoft.com/en-us/library/03196cz7(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_unpackhi_epi16(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s16(vzip2q_s16(vreinterpretq_s16_m128i(a),
-						 vreinterpretq_s16_m128i(b)));
-#else
-	int16x4_t a1 = vget_high_s16(vreinterpretq_s16_m128i(a));
-	int16x4_t b1 = vget_high_s16(vreinterpretq_s16_m128i(b));
-	int16x4x2_t result = vzip_s16(a1, b1);
-	return vreinterpretq_m128i_s16(
-		vcombine_s16(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the upper 2 signed or unsigned 32-bit integers in a with the
-// upper 2 signed or unsigned 32-bit integers in b.
-// https://msdn.microsoft.com/en-us/library/65sa7cbs(v=vs.100).aspx
-FORCE_INLINE __m128i _mm_unpackhi_epi32(__m128i a, __m128i b)
-{
-#if defined(__aarch64__)
-	return vreinterpretq_m128i_s32(vzip2q_s32(vreinterpretq_s32_m128i(a),
-						 vreinterpretq_s32_m128i(b)));
-#else
-	int32x2_t a1 = vget_high_s32(vreinterpretq_s32_m128i(a));
-	int32x2_t b1 = vget_high_s32(vreinterpretq_s32_m128i(b));
-	int32x2x2_t result = vzip_s32(a1, b1);
-	return vreinterpretq_m128i_s32(
-		vcombine_s32(result.val[0], result.val[1]));
-#endif
-}
-
-// Interleaves the upper signed or unsigned 64-bit integer in a with the
-// upper signed or unsigned 64-bit integer in b.
-//
-// r0 := a1
-// r1 := b1
-FORCE_INLINE __m128i _mm_unpackhi_epi64(__m128i a, __m128i b)
-{
-	int64x1_t a_h = vget_high_s64(vreinterpretq_s64_m128i(a));
-	int64x1_t b_h = vget_high_s64(vreinterpretq_s64_m128i(b));
-	return vreinterpretq_m128i_s64(vcombine_s64(a_h, b_h));
-}
-
-// Horizontally compute the minimum amongst the packed unsigned 16-bit integers
-// in a, store the minimum and index in dst, and zero the remaining bits in dst.
-//
-// index[2:0] := 0
-// min[15:0] := a[15:0]
-// FOR j := 0 to 7
-// i := j*16
-// IF a[i+15:i] < min[15:0]
-// index[2:0] := j
-// min[15:0] := a[i+15:i]
-// FI
-// ENDFOR
-// dst[15:0] := min[15:0]
-// dst[18:16] := index[2:0]
-// dst[127:19] := 0
-//
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_minpos_epu16&expand=3789
-FORCE_INLINE __m128i _mm_minpos_epu16(__m128i a)
-{
-	__m128i dst;
-	uint16_t min, idx = 0;
-	// Find the minimum value
-#if defined(__aarch64__)
-	min = vminvq_u16(vreinterpretq_u16_m128i(a));
-#else
-	__m64i tmp;
-	tmp = vreinterpret_m64i_u16(
-		vmin_u16(vget_low_u16(vreinterpretq_u16_m128i(a)),
-			 vget_high_u16(vreinterpretq_u16_m128i(a))));
-	tmp = vreinterpret_m64i_u16(vpmin_u16(vreinterpret_u16_m64i(tmp),
-					 vreinterpret_u16_m64i(tmp)));
-	tmp = vreinterpret_m64i_u16(vpmin_u16(vreinterpret_u16_m64i(tmp),
-					 vreinterpret_u16_m64i(tmp)));
-	min = vget_lane_u16(vreinterpret_u16_m64i(tmp), 0);
-#endif
-	// Get the index of the minimum value
-	int i;
-	for (i = 0; i < 8; i++) {
-		if (min == vgetq_lane_u16(vreinterpretq_u16_m128i(a), 0)) {
-			idx = (uint16_t)i;
-			break;
-		}
-		a = _mm_srli_si128(a, 2);
-	}
-	// Generate result
-	dst = _mm_setzero_si128();
-	dst = vreinterpretq_m128i_u16(
-		vsetq_lane_u16(min, vreinterpretq_u16_m128i(dst), 0));
-	dst = vreinterpretq_m128i_u16(
-		vsetq_lane_u16(idx, vreinterpretq_u16_m128i(dst), 1));
-	return dst;
-}
-
-// shift to right
-// https://msdn.microsoft.com/en-us/library/bb514041(v=vs.120).aspx
-// http://blog.csdn.net/hemmingway/article/details/44828303
-// Clang requires a macro here, as it is extremely picky about c being a
-// literal.
-#define _mm_alignr_epi8(a, b, c) \
-	((__m128i)vextq_s8((int8x16_t)(b), (int8x16_t)(a), (c)))
-
-// Extracts the selected signed or unsigned 8-bit integer from a and zero
-// extends.
-// FORCE_INLINE int _mm_extract_epi8(__m128i a, __constrange(0,16) int imm)
-#define _mm_extract_epi8(a, imm) vgetq_lane_u8(vreinterpretq_u8_m128i(a), (imm))
-
-// Inserts the least significant 8 bits of b into the selected 8-bit integer
-// of a.
-// FORCE_INLINE __m128i _mm_insert_epi8(__m128i a, int b,
-// __constrange(0,16) int imm)
-#define _mm_insert_epi8(a, b, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s8( \
-			vsetq_lane_s8((b), vreinterpretq_s8_m128i(a), (imm))); \
-	})
-
-// Extracts the selected signed or unsigned 16-bit integer from a and zero
-// extends.
-// https://msdn.microsoft.com/en-us/library/6dceta0c(v=vs.100).aspx
-// FORCE_INLINE int _mm_extract_epi16(__m128i a, __constrange(0,8) int imm)
-#define _mm_extract_epi16(a, imm) \
-	vgetq_lane_u16(vreinterpretq_u16_m128i(a), (imm))
-
-// Inserts the least significant 16 bits of b into the selected 16-bit integer
-// of a.
-// https://msdn.microsoft.com/en-us/library/kaze8hz1%28v=vs.100%29.aspx
-// FORCE_INLINE __m128i _mm_insert_epi16(__m128i a, int b,
-// __constrange(0,8) int imm)
-#define _mm_insert_epi16(a, b, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s16(vsetq_lane_s16( \
-			(b), vreinterpretq_s16_m128i(a), (imm))); \
-	})
-
-// Extracts the selected signed or unsigned 32-bit integer from a and zero
-// extends.
-// FORCE_INLINE int _mm_extract_epi32(__m128i a, __constrange(0,4) int imm)
-#define _mm_extract_epi32(a, imm) \
-	vgetq_lane_s32(vreinterpretq_s32_m128i(a), (imm))
-
-// Extracts the selected single-precision (32-bit) floating-point from a.
-// FORCE_INLINE int _mm_extract_ps(__m128 a, __constrange(0,4) int imm)
-#define _mm_extract_ps(a, imm) vgetq_lane_s32(vreinterpretq_s32_m128(a), (imm))
-
-// Inserts the least significant 32 bits of b into the selected 32-bit integer
-// of a.
-// FORCE_INLINE __m128i _mm_insert_epi32(__m128i a, int b,
-// __constrange(0,4) int imm)
-#define _mm_insert_epi32(a, b, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s32(vsetq_lane_s32( \
-			(b), vreinterpretq_s32_m128i(a), (imm))); \
-	})
-
-// Extracts the selected signed or unsigned 64-bit integer from a and zero
-// extends.
-// FORCE_INLINE __int64 _mm_extract_epi64(__m128i a, __constrange(0,2) int imm)
-#define _mm_extract_epi64(a, imm) \
-	vgetq_lane_s64(vreinterpretq_s64_m128i(a), (imm))
-
-// Inserts the least significant 64 bits of b into the selected 64-bit integer
-// of a.
-// FORCE_INLINE __m128i _mm_insert_epi64(__m128i a, __int64 b,
-// __constrange(0,2) int imm)
-#define _mm_insert_epi64(a, b, imm) \
-	__extension__({ \
-		vreinterpretq_m128i_s64(vsetq_lane_s64( \
-			(b), vreinterpretq_s64_m128i(a), (imm))); \
-	})
-
-// Count the number of bits set to 1 in unsigned 32-bit integer a, and
-// return that count in dst.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_popcnt_u32
-FORCE_INLINE int _mm_popcnt_u32(unsigned int a)
-{
-#if defined(__aarch64__)
-#if __has_builtin(__builtin_popcount)
-	return __builtin_popcount(a);
-#else
-	return (int)vaddlv_u8(vcnt_u8(vcreate_u8((uint64_t)a)));
-#endif
-#else
-	uint32_t count = 0;
-	uint8x8_t input_val, count8x8_val;
-	uint16x4_t count16x4_val;
-	uint32x2_t count32x2_val;
-
-	input_val = vld1_u8((uint8_t *)&a);
-	count8x8_val = vcnt_u8(input_val);
-	count16x4_val = vpaddl_u8(count8x8_val);
-	count32x2_val = vpaddl_u16(count16x4_val);
-
-	vst1_u32(&count, count32x2_val);
-	return count;
-#endif
-}
-
-// Count the number of bits set to 1 in unsigned 64-bit integer a, and
-// return that count in dst.
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_popcnt_u64
-FORCE_INLINE int64_t _mm_popcnt_u64(uint64_t a)
-{
-#if defined(__aarch64__)
-#if __has_builtin(__builtin_popcountll)
-	return __builtin_popcountll(a);
-#else
-	return (int64_t)vaddlv_u8(vcnt_u8(vcreate_u8(a)));
-#endif
-#else
-	uint64_t count = 0;
-	uint8x8_t input_val, count8x8_val;
-	uint16x4_t count16x4_val;
-	uint32x2_t count32x2_val;
-	uint64x1_t count64x1_val;
-
-	input_val = vld1_u8((uint8_t *)&a);
-	count8x8_val = vcnt_u8(input_val);
-	count16x4_val = vpaddl_u8(count8x8_val);
-	count32x2_val = vpaddl_u16(count16x4_val);
-	count64x1_val = vpaddl_u32(count32x2_val);
-	vst1_u64(&count, count64x1_val);
-	return count;
-#endif
-}
-
-// Macro: Transpose the 4x4 matrix formed by the 4 rows of single-precision
-// (32-bit) floating-point elements in row0, row1, row2, and row3, and store the
-// transposed matrix in these vectors (row0 now contains column 0, etc.).
-// https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=MM_TRANSPOSE4_PS&expand=5949
-#define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
-	do { \
-		__m128 tmp0, tmp1, tmp2, tmp3; \
-		tmp0 = _mm_unpacklo_ps(row0, row1); \
-		tmp2 = _mm_unpacklo_ps(row2, row3); \
-		tmp1 = _mm_unpackhi_ps(row0, row1); \
-		tmp3 = _mm_unpackhi_ps(row2, row3); \
-		row0 = _mm_movelh_ps(tmp0, tmp2); \
-		row1 = _mm_movehl_ps(tmp2, tmp0); \
-		row2 = _mm_movelh_ps(tmp1, tmp3); \
-		row3 = _mm_movehl_ps(tmp3, tmp1); \
-	} while (0)
-
-/* Crypto Extensions */
-
-#if defined(__ARM_FEATURE_CRYPTO)
-// Wraps vmull_p64
-FORCE_INLINE uint64x2_t _sse2neon_vmull_p64(uint64x1_t _a, uint64x1_t _b)
-{
-	poly64_t a = vget_lane_p64(vreinterpret_p64_u64(_a), 0);
-	poly64_t b = vget_lane_p64(vreinterpret_p64_u64(_b), 0);
-	return vreinterpretq_u64_p128(vmull_p64(a, b));
-}
-#else // ARMv7 polyfill
-// ARMv7/some A64 lacks vmull_p64, but it has vmull_p8.
-//
-// vmull_p8 calculates 8 8-bit->16-bit polynomial multiplies, but we need a
-// 64-bit->128-bit polynomial multiply.
-//
-// It needs some work and is somewhat slow, but it is still faster than all
-// known scalar methods.
-//
-// Algorithm adapted to C from
-// https://www.workofard.com/2017/07/ghash-for-low-end-cores/, which is adapted
-// from "Fast Software Polynomial Multiplication on ARM Processors Using the
-// NEON Engine" by Danilo Camara, Conrado Gouvea, Julio Lopez and Ricardo Dahab
-// (https://hal.inria.fr/hal-01506572)
-static uint64x2_t _sse2neon_vmull_p64(uint64x1_t _a, uint64x1_t _b)
-{
-	poly8x8_t a = vreinterpret_p8_u64(_a);
-	poly8x8_t b = vreinterpret_p8_u64(_b);
-
-	// Masks
-	uint8x16_t k48_32 = vcombine_u8(vcreate_u8(0x0000ffffffffffff),
-					vcreate_u8(0x00000000ffffffff));
-	uint8x16_t k16_00 = vcombine_u8(vcreate_u8(0x000000000000ffff),
-					vcreate_u8(0x0000000000000000));
-
-	// Do the multiplies, rotating with vext to get all combinations
-	uint8x16_t d = vreinterpretq_u8_p16(vmull_p8(a, b)); // D = A0 * B0
-	uint8x16_t e = vreinterpretq_u8_p16(
-		vmull_p8(a, vext_p8(b, b, 1))); // E = A0 * B1
-	uint8x16_t f = vreinterpretq_u8_p16(
-		vmull_p8(vext_p8(a, a, 1), b)); // F = A1 * B0
-	uint8x16_t g = vreinterpretq_u8_p16(
-		vmull_p8(a, vext_p8(b, b, 2))); // G = A0 * B2
-	uint8x16_t h = vreinterpretq_u8_p16(
-		vmull_p8(vext_p8(a, a, 2), b)); // H = A2 * B0
-	uint8x16_t i = vreinterpretq_u8_p16(
-		vmull_p8(a, vext_p8(b, b, 3))); // I = A0 * B3
-	uint8x16_t j = vreinterpretq_u8_p16(
-		vmull_p8(vext_p8(a, a, 3), b)); // J = A3 * B0
-	uint8x16_t k = vreinterpretq_u8_p16(
-		vmull_p8(a, vext_p8(b, b, 4))); // L = A0 * B4
-
-	// Add cross products
-	uint8x16_t l = veorq_u8(e, f); // L = E + F
-	uint8x16_t m = veorq_u8(g, h); // M = G + H
-	uint8x16_t n = veorq_u8(i, j); // N = I + J
-
-	// Interleave. Using vzip1 and vzip2 prevents Clang from emitting TBL
-	// instructions.
-#if defined(__aarch64__)
-	uint8x16_t lm_p0 = vreinterpretq_u8_u64(
-		vzip1q_u64(vreinterpretq_u64_u8(l), vreinterpretq_u64_u8(m)));
-	uint8x16_t lm_p1 = vreinterpretq_u8_u64(
-		vzip2q_u64(vreinterpretq_u64_u8(l), vreinterpretq_u64_u8(m)));
-	uint8x16_t nk_p0 = vreinterpretq_u8_u64(
-		vzip1q_u64(vreinterpretq_u64_u8(n), vreinterpretq_u64_u8(k)));
-	uint8x16_t nk_p1 = vreinterpretq_u8_u64(
-		vzip2q_u64(vreinterpretq_u64_u8(n), vreinterpretq_u64_u8(k)));
-#else
-	uint8x16_t lm_p0 = vcombine_u8(vget_low_u8(l), vget_low_u8(m));
-	uint8x16_t lm_p1 = vcombine_u8(vget_high_u8(l), vget_high_u8(m));
-	uint8x16_t nk_p0 = vcombine_u8(vget_low_u8(n), vget_low_u8(k));
-	uint8x16_t nk_p1 = vcombine_u8(vget_high_u8(n), vget_high_u8(k));
-#endif
-	// t0 = (L) (P0 + P1) << 8
-	// t1 = (M) (P2 + P3) << 16
-	uint8x16_t t0t1_tmp = veorq_u8(lm_p0, lm_p1);
-	uint8x16_t t0t1_h = vandq_u8(lm_p1, k48_32);
-	uint8x16_t t0t1_l = veorq_u8(t0t1_tmp, t0t1_h);
-
-	// t2 = (N) (P4 + P5) << 24
-	// t3 = (K) (P6 + P7) << 32
-	uint8x16_t t2t3_tmp = veorq_u8(nk_p0, nk_p1);
-	uint8x16_t t2t3_h = vandq_u8(nk_p1, k16_00);
-	uint8x16_t t2t3_l = veorq_u8(t2t3_tmp, t2t3_h);
-
-	// De-interleave
-#if defined(__aarch64__)
-	uint8x16_t t0 = vreinterpretq_u8_u64(vuzp1q_u64(
-		vreinterpretq_u64_u8(t0t1_l), vreinterpretq_u64_u8(t0t1_h)));
-	uint8x16_t t1 = vreinterpretq_u8_u64(vuzp2q_u64(
-		vreinterpretq_u64_u8(t0t1_l), vreinterpretq_u64_u8(t0t1_h)));
-	uint8x16_t t2 = vreinterpretq_u8_u64(vuzp1q_u64(
-		vreinterpretq_u64_u8(t2t3_l), vreinterpretq_u64_u8(t2t3_h)));
-	uint8x16_t t3 = vreinterpretq_u8_u64(vuzp2q_u64(
-		vreinterpretq_u64_u8(t2t3_l), vreinterpretq_u64_u8(t2t3_h)));
-#else
-	uint8x16_t t1 = vcombine_u8(vget_high_u8(t0t1_l), vget_high_u8(t0t1_h));
-	uint8x16_t t0 = vcombine_u8(vget_low_u8(t0t1_l), vget_low_u8(t0t1_h));
-	uint8x16_t t3 = vcombine_u8(vget_high_u8(t2t3_l), vget_high_u8(t2t3_h));
-	uint8x16_t t2 = vcombine_u8(vget_low_u8(t2t3_l), vget_low_u8(t2t3_h));
-#endif
-	// Shift the cross products
-	uint8x16_t t0_shift = vextq_u8(t0, t0, 15); // t0 << 8
-	uint8x16_t t1_shift = vextq_u8(t1, t1, 14); // t1 << 16
-	uint8x16_t t2_shift = vextq_u8(t2, t2, 13); // t2 << 24
-	uint8x16_t t3_shift = vextq_u8(t3, t3, 12); // t3 << 32
-
-	// Accumulate the products
-	uint8x16_t cross1 = veorq_u8(t0_shift, t1_shift);
-	uint8x16_t cross2 = veorq_u8(t2_shift, t3_shift);
-	uint8x16_t mix = veorq_u8(d, cross1);
-	uint8x16_t r = veorq_u8(mix, cross2);
-	return vreinterpretq_u64_u8(r);
-}
-#endif // ARMv7 polyfill
-
-FORCE_INLINE __m128i _mm_clmulepi64_si128(__m128i _a, __m128i _b, const int imm)
-{
-	uint64x2_t a = vreinterpretq_u64_m128i(_a);
-	uint64x2_t b = vreinterpretq_u64_m128i(_b);
-	switch (imm & 0x11) {
-	case 0x00:
-		return vreinterpretq_m128i_u64(
-			_sse2neon_vmull_p64(vget_low_u64(a), vget_low_u64(b)));
-	case 0x01:
-		return vreinterpretq_m128i_u64(
-			_sse2neon_vmull_p64(vget_high_u64(a), vget_low_u64(b)));
-	case 0x10:
-		return vreinterpretq_m128i_u64(
-			_sse2neon_vmull_p64(vget_low_u64(a), vget_high_u64(b)));
-	case 0x11:
-		return vreinterpretq_m128i_u64(_sse2neon_vmull_p64(
-			vget_high_u64(a), vget_high_u64(b)));
-	default:
-		abort();
-	}
-}
-
-#if !defined(__ARM_FEATURE_CRYPTO) && defined(__aarch64__)
-// In the absence of crypto extensions, implement aesenc using regular neon
-// intrinsics instead. See:
-// https://www.workofard.com/2017/01/accelerated-aes-for-the-arm64-linux-kernel/
-// https://www.workofard.com/2017/07/ghash-for-low-end-cores/ and
-// https://github.com/ColinIanKing/linux-next-mirror/blob/b5f466091e130caaf0735976648f72bd5e09aa84/crypto/aegis128-neon-inner.c#L52
-// for more information Reproduced with permission of the author.
-FORCE_INLINE __m128i _mm_aesenc_si128(__m128i EncBlock, __m128i RoundKey)
-{
-	static const uint8_t crypto_aes_sbox[256] = {
-		0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5, 0x30, 0x01,
-		0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76, 0xca, 0x82, 0xc9, 0x7d,
-		0xfa, 0x59, 0x47, 0xf0, 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4,
-		0x72, 0xc0, 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
-		0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15, 0x04, 0xc7,
-		0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a, 0x07, 0x12, 0x80, 0xe2,
-		0xeb, 0x27, 0xb2, 0x75, 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e,
-		0x5a, 0xa0, 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
-		0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b, 0x6a, 0xcb,
-		0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf, 0xd0, 0xef, 0xaa, 0xfb,
-		0x43, 0x4d, 0x33, 0x85, 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c,
-		0x9f, 0xa8, 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
-		0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2, 0xcd, 0x0c,
-		0x13, 0xec, 0x5f, 0x97, 0x44, 0x17, 0xc4, 0xa7, 0x7e, 0x3d,
-		0x64, 0x5d, 0x19, 0x73, 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a,
-		0x90, 0x88, 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
-		0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c, 0xc2, 0xd3,
-		0xac, 0x62, 0x91, 0x95, 0xe4, 0x79, 0xe7, 0xc8, 0x37, 0x6d,
-		0x8d, 0xd5, 0x4e, 0xa9, 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a,
-		0xae, 0x08, 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
-		0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a, 0x70, 0x3e,
-		0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e, 0x61, 0x35, 0x57, 0xb9,
-		0x86, 0xc1, 0x1d, 0x9e, 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9,
-		0x8e, 0x94, 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
-		0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68, 0x41, 0x99,
-		0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16};
-	static const uint8_t shift_rows[] = {0x0, 0x5, 0xa, 0xf, 0x4, 0x9,
-					 0xe, 0x3, 0x8, 0xd, 0x2, 0x7,
-					 0xc, 0x1, 0x6, 0xb};
-	static const uint8_t ror32by8[] = {0x1, 0x2, 0x3, 0x0, 0x5, 0x6,
-					 0x7, 0x4, 0x9, 0xa, 0xb, 0x8,
-					 0xd, 0xe, 0xf, 0xc};
-
-	uint8x16_t v;
-	uint8x16_t w = vreinterpretq_u8_m128i(EncBlock);
-
-	// shift rows
-	w = vqtbl1q_u8(w, vld1q_u8(shift_rows));
-
-	// sub bytes
-	v = vqtbl4q_u8(vld1q_u8_x4(crypto_aes_sbox), w);
-	v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0x40), w - 0x40);
-	v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0x80), w - 0x80);
-	v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0xc0), w - 0xc0);
-
-	// mix columns
-	w = (v << 1) ^ (uint8x16_t)(((int8x16_t)v >> 7) & 0x1b);
-	w ^= (uint8x16_t)vrev32q_u16((uint16x8_t)v);
-	w ^= vqtbl1q_u8(v ^ w, vld1q_u8(ror32by8));
-
-	// add round key
-	return vreinterpretq_m128i_u8(w) ^ RoundKey;
-}
-#elif defined(__ARM_FEATURE_CRYPTO)
-// Implements equivalent of 'aesenc' by combining AESE (with an empty key) and
-// AESMC and then manually applying the real key as an xor operation This
-// unfortunately means an additional xor op; the compiler should be able to
-// optimise this away for repeated calls however See
-// https://blog.michaelbrase.com/2018/05/08/emulating-x86-aes-intrinsics-on-armv8-a
-// for more details.
-inline __m128i _mm_aesenc_si128(__m128i a, __m128i b)
-{
-	return vreinterpretq_m128i_u8(
-		vaesmcq_u8(
-			vaeseq_u8(vreinterpretq_u8_m128i(a), vdupq_n_u8(0))) ^
-		vreinterpretq_u8_m128i(b));
-}
-#endif
-
-/* Streaming Extensions */
-
-// Guarantees that every preceding store is globally visible before any
-// subsequent store.
-// https://msdn.microsoft.com/en-us/library/5h2w73d1%28v=vs.90%29.aspx
-FORCE_INLINE void _mm_sfence(void)
-{
-	__sync_synchronize();
-}
-
-// Stores the data in a to the address p without polluting the caches. If the
-// cache line containing address p is already in the cache, the cache will be
-// updated.Address p must be 16 - byte aligned.
-// https://msdn.microsoft.com/en-us/library/ba08y07y%28v=vs.90%29.aspx
-FORCE_INLINE void _mm_stream_si128(__m128i *p, __m128i a)
-{
-#if __has_builtin(__builtin_nontemporal_store)
-	__builtin_nontemporal_store(a, p);
-#else
-	vst1q_s64((int64_t *)p, vreinterpretq_s64_m128i(a));
-#endif
-}
-
-// Cache line containing p is flushed and invalidated from all caches in the
-// coherency domain. :
-// https://msdn.microsoft.com/en-us/library/ba08y07y(v=vs.100).aspx
-FORCE_INLINE void _mm_clflush(void const *p)
-{
-	(void)p;
-	// no corollary for Neon?
-}
-
-// Allocate aligned blocks of memory.
-// https://software.intel.com/en-us/
-// cpp-compiler-developer-guide-and-reference-allocating-and-freeing-aligned-memory-blocks
-FORCE_INLINE void *_mm_malloc(size_t size, size_t align)
-{
-	void *ptr;
-	if (align == 1)
-		return malloc(size);
-	if (align == 2 || (sizeof(void *) == 8 && align == 4))
-		align = sizeof(void *);
-	if (!posix_memalign(&ptr, align, size))
-		return ptr;
-	return NULL;
-}
-
-FORCE_INLINE void _mm_free(void *addr)
-{
-	free(addr);
-}
-
-// Starting with the initial value in crc, accumulates a CRC32 value for
-// unsigned 8-bit integer v.
-// https://msdn.microsoft.com/en-us/library/bb514036(v=vs.100)
-FORCE_INLINE uint32_t _mm_crc32_u8(uint32_t crc, uint8_t v)
-{
-#if defined(__aarch64__) && defined(__ARM_FEATURE_CRC32)
-	__asm__ __volatile__("crc32cb %w[c], %w[c], %w[v]\n\t"
-			 : [c] "+r"(crc)
-			 : [v] "r"(v));
-#else
-	crc ^= v;
-	for (int bit = 0; bit < 8; bit++) {
-		if (crc & 1)
-			crc = (crc >> 1) ^ UINT32_C(0x82f63b78);
-		else
-			crc = (crc >> 1);
-	}
-#endif
-	return crc;
-}
-
-// Starting with the initial value in crc, accumulates a CRC32 value for
-// unsigned 16-bit integer v.
-// https://msdn.microsoft.com/en-us/library/bb531411(v=vs.100)
-FORCE_INLINE uint32_t _mm_crc32_u16(uint32_t crc, uint16_t v)
-{
-#if defined(__aarch64__) && defined(__ARM_FEATURE_CRC32)
-	__asm__ __volatile__("crc32ch %w[c], %w[c], %w[v]\n\t"
-			 : [c] "+r"(crc)
-			 : [v] "r"(v));
-#else
-	crc = _mm_crc32_u8(crc, v & 0xff);
-	crc = _mm_crc32_u8(crc, (v >> 8) & 0xff);
-#endif
-	return crc;
-}
-
-// Starting with the initial value in crc, accumulates a CRC32 value for
-// unsigned 32-bit integer v.
-// https://msdn.microsoft.com/en-us/library/bb531394(v=vs.100)
-FORCE_INLINE uint32_t _mm_crc32_u32(uint32_t crc, uint32_t v)
-{
-#if defined(__aarch64__) && defined(__ARM_FEATURE_CRC32)
-	__asm__ __volatile__("crc32cw %w[c], %w[c], %w[v]\n\t"
-			 : [c] "+r"(crc)
-			 : [v] "r"(v));
-#else
-	crc = _mm_crc32_u16(crc, v & 0xffff);
-	crc = _mm_crc32_u16(crc, (v >> 16) & 0xffff);
-#endif
-	return crc;
-}
-
-// Starting with the initial value in crc, accumulates a CRC32 value for
-// unsigned 64-bit integer v.
-// https://msdn.microsoft.com/en-us/library/bb514033(v=vs.100)
-FORCE_INLINE uint64_t _mm_crc32_u64(uint64_t crc, uint64_t v)
-{
-#if defined(__aarch64__) && defined(__ARM_FEATURE_CRC32)
-	__asm__ __volatile__("crc32cx %w[c], %w[c], %x[v]\n\t"
-			 : [c] "+r"(crc)
-			 : [v] "r"(v));
-#else
-	crc = _mm_crc32_u32((uint32_t)(crc), v & 0xffffffff);
-	crc = _mm_crc32_u32((uint32_t)(crc), (v >> 32) & 0xffffffff);
-#endif
-	return crc;
-}
-
-#if defined(__GNUC__) || defined(__clang__)
-#pragma pop_macro("ALIGN_STRUCT")
-#pragma pop_macro("FORCE_INLINE")
-#endif
-
-#endif

obs-studio-26.1.0.tar.xz/.github/workflows/main.yml -> obs-studio-26.1.1.tar.xz/.github/workflows/main.yml Changed

@@ -24,7 +24,7 @@
     runs-on: [macos-latest]
     env:
       MIN_MACOS_VERSION: '10.13'
-      MACOS_DEPS_VERSION: '2020-12-11'
+      MACOS_DEPS_VERSION: '2020-12-22'
       VLC_VERSION: '3.0.8'
       SPARKLE_VERSION: '1.23.0'
       QT_VERSION: '5.15.2'
@@ -54,13 +54,17 @@
         shell: bash
         run: |
           if [ -d /usr/local/opt/openssl@1.0.2t ]; then
-              brew uninstall openssl@1.0.2t
-              brew untap local/openssl
+            brew uninstall openssl@1.0.2t
+            brew untap local/openssl
           fi
 
           if [ -d /usr/local/opt/python@2.7.17 ]; then
-              brew uninstall python@2.7.17
-              brew untap local/python2
+            brew uninstall python@2.7.17
+            brew untap local/python2
+          fi
+
+          if [ -d /usr/local/opt/speexdsp ]; then
+            brew unlink speexdsp
           fi
           brew bundle --file ./CI/scripts/macos/Brewfile
       - name: 'Restore Chromium Embedded Framework from cache'
@@ -136,7 +140,8 @@
         run: |
           mkdir ./build
           cd ./build
-          cmake -DENABLE_UNIT_TESTS=YES -DENABLE_SPARKLE_UPDATER=ON -DDISABLE_PYTHON=ON -DCMAKE_OSX_DEPLOYMENT_TARGET=${{ env.MIN_MACOS_VERSION }} -DQTDIR="/tmp/obsdeps" -DSWIGDIR="/tmp/obsdeps" -DDepsPath="/tmp/obsdeps" -DVLCPath="${{ github.workspace }}/cmbuild/vlc-${{ env.VLC_VERSION }}" -DENABLE_VLC=ON -DBUILD_BROWSER=ON -DBROWSER_DEPLOY=ON -DWITH_RTMPS=ON -DCEF_ROOT_DIR="${{ github.workspace }}/cmbuild/cef_binary_${{ env.CEF_BUILD_VERSION }}_macosx64" ..
+          LEGACY_BROWSER="$(test "${{ env.CEF_BUILD_VERSION }}" -le 3770 && echo "ON" || echo "OFF")"
+          cmake -DENABLE_UNIT_TESTS=YES -DENABLE_SPARKLE_UPDATER=ON -DDISABLE_PYTHON=ON -DCMAKE_OSX_DEPLOYMENT_TARGET=${{ env.MIN_MACOS_VERSION }} -DQTDIR="/tmp/obsdeps" -DSWIGDIR="/tmp/obsdeps" -DDepsPath="/tmp/obsdeps" -DVLCPath="${{ github.workspace }}/cmbuild/vlc-${{ env.VLC_VERSION }}" -DENABLE_VLC=ON -DBUILD_BROWSER=ON -DBROWSER_LEGACY=$LEGACY_BROWSER -DWITH_RTMPS=ON -DCEF_ROOT_DIR="${{ github.workspace }}/cmbuild/cef_binary_${{ env.CEF_BUILD_VERSION }}_macosx64" ..
       - name: 'Build'
         shell: bash
         working-directory: ${{ github.workspace }}/build
@@ -169,9 +174,16 @@
           mkdir -p OBS.app/Contents/MacOS
           mkdir OBS.app/Contents/PlugIns
           mkdir OBS.app/Contents/Resources
+          mkdir OBS.app/Contents/Frameworks
 
           cp rundir/RelWithDebInfo/bin/obs ./OBS.app/Contents/MacOS
           cp rundir/RelWithDebInfo/bin/obs-ffmpeg-mux ./OBS.app/Contents/MacOS
+          if ! [ "${{ env.CEF_BUILD_VERSION }}" -le 3770 ]; then
+            cp -R "rundir/RelWithDebInfo/bin/OBS Helper.app" "./OBS.app/Contents/Frameworks/OBS Helper.app"
+            cp -R "rundir/RelWithDebInfo/bin/OBS Helper (GPU).app" "./OBS.app/Contents/Frameworks/OBS Helper (GPU).app"
+            cp -R "rundir/RelWithDebInfo/bin/OBS Helper (Plugin).app" "./OBS.app/Contents/Frameworks/OBS Helper (Plugin).app"
+            cp -R "rundir/RelWithDebInfo/bin/OBS Helper (Renderer).app" "./OBS.app/Contents/Frameworks/OBS Helper (Renderer).app"
+          fi
           cp rundir/RelWithDebInfo/bin/libobsglad.0.dylib ./OBS.app/Contents/MacOS
           cp -R rundir/RelWithDebInfo/data ./OBS.app/Contents/Resources
           cp ../CI/scripts/macos/app/AppIcon.icns ./OBS.app/Contents/Resources
@@ -185,35 +197,45 @@
             rm -rf ./OBS.app/Contents/Resources/data/obs-scripting/
           fi
 
+          BUNDLE_PLUGINS=(
+            ./OBS.app/Contents/PlugIns/coreaudio-encoder.so
+            ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so
+            ./OBS.app/Contents/PlugIns/decklink-captions.so
+            ./OBS.app/Contents/PlugIns/frontend-tools.so
+            ./OBS.app/Contents/PlugIns/image-source.so
+            ./OBS.app/Contents/PlugIns/mac-avcapture.so
+            ./OBS.app/Contents/PlugIns/mac-capture.so
+            ./OBS.app/Contents/PlugIns/mac-decklink.so
+            ./OBS.app/Contents/PlugIns/mac-syphon.so
+            ./OBS.app/Contents/PlugIns/mac-vth264.so
+            ./OBS.app/Contents/PlugIns/mac-virtualcam.so
+            ./OBS.app/Contents/PlugIns/obs-browser.so
+            ./OBS.app/Contents/PlugIns/obs-ffmpeg.so
+            ./OBS.app/Contents/PlugIns/obs-filters.so
+            ./OBS.app/Contents/PlugIns/obs-transitions.so
+            ./OBS.app/Contents/PlugIns/obs-vst.so
+            ./OBS.app/Contents/PlugIns/rtmp-services.so
+            ./OBS.app/Contents/MacOS/obs-ffmpeg-mux
+            ./OBS.app/Contents/MacOS/obslua.so
+            ./OBS.app/Contents/PlugIns/obs-x264.so
+            ./OBS.app/Contents/PlugIns/text-freetype2.so
+            ./OBS.app/Contents/PlugIns/obs-outputs.so
+          )
+
+          if ! [ "${{ env.CEF_BUILD_VERSION }}" -le 3770 ]; then
           ../CI/scripts/macos/app/dylibBundler -cd -of -a ./OBS.app -q -f \
             -s ./OBS.app/Contents/MacOS \
             -s "${{ github.workspace }}/cmbuild/sparkle/Sparkle.framework" \
             -s ./rundir/RelWithDebInfo/bin \
-            -x ./OBS.app/Contents/PlugIns/coreaudio-encoder.so \
-            -x ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so \
-            -x ./OBS.app/Contents/PlugIns/decklink-captions.so \
-            -x ./OBS.app/Contents/PlugIns/frontend-tools.so \
-            -x ./OBS.app/Contents/PlugIns/image-source.so \
-            -x ./OBS.app/Contents/PlugIns/linux-jack.so \
-            -x ./OBS.app/Contents/PlugIns/mac-avcapture.so \
-            -x ./OBS.app/Contents/PlugIns/mac-capture.so \
-            -x ./OBS.app/Contents/PlugIns/mac-decklink.so \
-            -x ./OBS.app/Contents/PlugIns/mac-syphon.so \
-            -x ./OBS.app/Contents/PlugIns/mac-vth264.so \
-            -x ./OBS.app/Contents/PlugIns/mac-virtualcam.so \
-            -x ./OBS.app/Contents/PlugIns/obs-browser.so \
-            -x ./OBS.app/Contents/PlugIns/obs-browser-page \
-            -x ./OBS.app/Contents/PlugIns/obs-ffmpeg.so \
-            -x ./OBS.app/Contents/PlugIns/obs-filters.so \
-            -x ./OBS.app/Contents/PlugIns/obs-transitions.so \
-            -x ./OBS.app/Contents/PlugIns/obs-vst.so \
-            -x ./OBS.app/Contents/PlugIns/rtmp-services.so \
-            -x ./OBS.app/Contents/MacOS/obs-ffmpeg-mux \
-            -x ./OBS.app/Contents/MacOS/obslua.so \
-            -x ./OBS.app/Contents/PlugIns/obs-x264.so \
-            -x ./OBS.app/Contents/PlugIns/text-freetype2.so \
-            -x ./OBS.app/Contents/PlugIns/obs-libfdk.so \
-            -x ./OBS.app/Contents/PlugIns/obs-outputs.so
+            $(echo "${BUNDLE_PLUGINS[@]/#/-x }")
+          else
+          ../CI/scripts/macos/app/dylibBundler -cd -of -a ./OBS.app -q -f \
+            -s ./OBS.app/Contents/MacOS \
+            -s "${{ github.workspace }}/cmbuild/sparkle/Sparkle.framework" \
+            -s ./rundir/RelWithDebInfo/bin \
+            $(echo "${BUNDLE_PLUGINS[@]/#/-x }") \
+            -x ./OBS.app/Contents/PlugIns/obs-browser-page
+          fi
 
           mv ./libobs-opengl/libobs-opengl.so ./OBS.app/Contents/Frameworks
 
@@ -250,12 +272,22 @@
           codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libswiftshader_libEGL.dylib"
           codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libGLESv2.dylib"
           codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libswiftshader_libGLESv2.dylib"
+          if ! [ "${{ env.CEF_BUILD_VERSION }}" -le 3770 ]; then
+            codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libvk_swiftshader.dylib"
+          fi
           codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" --deep "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework"
 
           codesign --force --options runtime --deep --sign "${SIGN_IDENTITY:--}" "./OBS.app/Contents/Resources/data/obs-mac-virtualcam.plugin"
 
           codesign --force --options runtime --entitlements "../CI/scripts/macos/app/entitlements.plist" --sign "${SIGN_IDENTITY:--}" --deep ./OBS.app
 
+          if ! [ "${{ env.CEF_BUILD_VERSION }}" -le 3770 ]; then
+            codesign --force --options runtime --sign "${SIGN_IDENTITY:--}" --deep "./OBS.app/Contents/Frameworks/OBS Helper.app"
+            codesign --force --options runtime --entitlements "../CI/scripts/macos/helpers/helper-gpu-entitlements.plist" --sign "${SIGN_IDENTITY:--}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (GPU).app"
+            codesign --force --options runtime --entitlements "../CI/scripts/macos/helpers/helper-plugin-entitlements.plist" --sign "${SIGN_IDENTITY:--}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (Plugin).app"
+            codesign --force --options runtime --entitlements "../CI/scripts/macos/helpers/helper-renderer-entitlements.plist" --sign "${SIGN_IDENTITY:--}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (Renderer).app"
+          fi
+
           codesign -dvv ./OBS.app
       - name: 'Package'
         if: success() && (github.event_name != 'pull_request' || env.SEEKING_TESTERS == '1')

obs-studio-26.1.0.tar.xz/CI/full-build-macos.sh -> obs-studio-26.1.1.tar.xz/CI/full-build-macos.sh Changed

@@ -47,6 +47,7 @@
 CI_SPARKLE_VERSION=$(cat ${CI_WORKFLOW} | sed -En "s/[ ]+SPARKLE_VERSION: '([0-9\.]+)'/\1/p")
 CI_QT_VERSION=$(cat ${CI_WORKFLOW} | sed -En "s/[ ]+QT_VERSION: '([0-9\.]+)'/\1/p" | head -1)
 CI_MIN_MACOS_VERSION=$(cat ${CI_WORKFLOW} | sed -En "s/[ ]+MIN_MACOS_VERSION: '([0-9\.]+)'/\1/p")
+NPROC="${NPROC:-$(sysctl -n hw.ncpu)}"
 
 BUILD_DEPS=(
     "obs-deps ${MACOS_DEPS_VERSION:-${CI_DEPS_VERSION}}"
@@ -226,7 +227,7 @@
         -DCMAKE_OSX_DEPLOYMENT_TARGET=${MIN_MACOS_VERSION:-${CI_MIN_MACOS_VERSION}} \
         ..
     step "Build..."
-    make -j4
+    make -j${NPROC}
     if [ ! -d libcef_dll ]; then mkdir libcef_dll; fi
 }
 
@@ -277,7 +278,7 @@
         -DDepsPath="/tmp/obsdeps" \
         -DVLCPath="${DEPS_BUILD_DIR}/vlc-${VLC_VERSION:-${CI_VLC_VERSION}}" \
         -DBUILD_BROWSER=ON \
-        -DBROWSER_DEPLOY=ON \
+        -DBROWSER_LEGACY="$(test "${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}" -le 3770 && echo "ON" || echo "OFF")" \
         -DWITH_RTMPS=ON \
         -DCEF_ROOT_DIR="${DEPS_BUILD_DIR}/cef_binary_${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}_macosx64" \
         -DCMAKE_BUILD_TYPE="${BUILD_CONFIG}" \
@@ -288,7 +289,7 @@
 run_obs_build() {
     ensure_dir "${CHECKOUT_DIR}/${BUILD_DIR}"
     hr "Build OBS..."
-    make -j4
+    make -j${NPROC}
 }
 
 ## OBS BUNDLE AS MACOS APPLICATION ##
@@ -303,37 +304,47 @@
     hr "Bundle dylibs for macOS application"
 
     step "Run dylibBundler.."
-    ${CI_SCRIPTS}/app/dylibbundler -cd -of -a ./OBS.app -q -f \
-        -s ./OBS.app/Contents/MacOS \
-        -s "${DEPS_BUILD_DIR}/sparkle/Sparkle.framework" \
-        -s ./rundir/${BUILD_CONFIG}/bin/ \
-        -x ./OBS.app/Contents/PlugIns/coreaudio-encoder.so \
-        -x ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so \
-        -x ./OBS.app/Contents/PlugIns/decklink-captions.so \
-        -x ./OBS.app/Contents/PlugIns/frontend-tools.so \
-        -x ./OBS.app/Contents/PlugIns/image-source.so \
-        -x ./OBS.app/Contents/PlugIns/linux-jack.so \
-        -x ./OBS.app/Contents/PlugIns/mac-avcapture.so \
-        -x ./OBS.app/Contents/PlugIns/mac-capture.so \
-        -x ./OBS.app/Contents/PlugIns/mac-decklink.so \
-        -x ./OBS.app/Contents/PlugIns/mac-syphon.so \
-        -x ./OBS.app/Contents/PlugIns/mac-vth264.so \
-        -x ./OBS.app/Contents/PlugIns/mac-virtualcam.so \
-        -x ./OBS.app/Contents/PlugIns/obs-browser.so \
-        -x ./OBS.app/Contents/PlugIns/obs-browser-page \
-        -x ./OBS.app/Contents/PlugIns/obs-ffmpeg.so \
-        -x ./OBS.app/Contents/PlugIns/obs-filters.so \
-        -x ./OBS.app/Contents/PlugIns/obs-transitions.so \
-        -x ./OBS.app/Contents/PlugIns/obs-vst.so \
-        -x ./OBS.app/Contents/PlugIns/rtmp-services.so \
-        -x ./OBS.app/Contents/MacOS/obs-ffmpeg-mux \
-        -x ./OBS.app/Contents/MacOS/obslua.so \
-        -x ./OBS.app/Contents/PlugIns/obs-x264.so \
-        -x ./OBS.app/Contents/PlugIns/text-freetype2.so \
-        -x ./OBS.app/Contents/PlugIns/obs-libfdk.so \
-        -x ./OBS.app/Contents/PlugIns/obs-outputs.so
-    step "Move libobs-opengl to final destination"
 
+    BUNDLE_PLUGINS=(
+        ./OBS.app/Contents/PlugIns/coreaudio-encoder.so
+        ./OBS.app/Contents/PlugIns/decklink-ouput-ui.so
+        ./OBS.app/Contents/PlugIns/decklink-captions.so
+        ./OBS.app/Contents/PlugIns/frontend-tools.so
+        ./OBS.app/Contents/PlugIns/image-source.so
+        ./OBS.app/Contents/PlugIns/mac-avcapture.so
+        ./OBS.app/Contents/PlugIns/mac-capture.so
+        ./OBS.app/Contents/PlugIns/mac-decklink.so
+        ./OBS.app/Contents/PlugIns/mac-syphon.so
+        ./OBS.app/Contents/PlugIns/mac-vth264.so
+        ./OBS.app/Contents/PlugIns/mac-virtualcam.so
+        ./OBS.app/Contents/PlugIns/obs-browser.so
+        ./OBS.app/Contents/PlugIns/obs-ffmpeg.so
+        ./OBS.app/Contents/PlugIns/obs-filters.so
+        ./OBS.app/Contents/PlugIns/obs-transitions.so
+        ./OBS.app/Contents/PlugIns/obs-vst.so
+        ./OBS.app/Contents/PlugIns/rtmp-services.so
+        ./OBS.app/Contents/MacOS/obs-ffmpeg-mux
+        ./OBS.app/Contents/MacOS/obslua.so
+        ./OBS.app/Contents/PlugIns/obs-x264.so
+        ./OBS.app/Contents/PlugIns/text-freetype2.so
+        ./OBS.app/Contents/PlugIns/obs-outputs.so
+        )
+    if ! [ "${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}" -le 3770 ]; then
+        ${CI_SCRIPTS}/app/dylibbundler -cd -of -a ./OBS.app -q -f \
+            -s ./OBS.app/Contents/MacOS \
+            -s "${DEPS_BUILD_DIR}/sparkle/Sparkle.framework" \
+            -s ./rundir/${BUILD_CONFIG}/bin/ \
+            $(echo "${BUNDLE_PLUGINS[@]/#/-x }")
+    else
+        ${CI_SCRIPTS}/app/dylibbundler -cd -of -a ./OBS.app -q -f \
+            -s ./OBS.app/Contents/MacOS \
+            -s "${DEPS_BUILD_DIR}/sparkle/Sparkle.framework" \
+            -s ./rundir/${BUILD_CONFIG}/bin/ \
+            $(echo "${BUNDLE_PLUGINS[@]/#/-x }") \
+            -x ./OBS.app/Contents/PlugIns/obs-browser-page
+    fi
+
+    step "Move libobs-opengl to final destination"
     if [ -f "./libobs-opengl/libobs-opengl.so" ]; then
         cp ./libobs-opengl/libobs-opengl.so ./OBS.app/Contents/Frameworks
     else
@@ -379,10 +390,17 @@
     mkdir -p OBS.app/Contents/MacOS
     mkdir OBS.app/Contents/PlugIns
     mkdir OBS.app/Contents/Resources
+    mkdir OBS.app/Contents/Frameworks
 
     cp rundir/${BUILD_CONFIG}/bin/obs ./OBS.app/Contents/MacOS
     cp rundir/${BUILD_CONFIG}/bin/obs-ffmpeg-mux ./OBS.app/Contents/MacOS
     cp rundir/${BUILD_CONFIG}/bin/libobsglad.0.dylib ./OBS.app/Contents/MacOS
+    if ! [ "${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}" -le 3770 ]; then
+        cp -R "rundir/${BUILD_CONFIG}/bin/OBS Helper.app" "./OBS.app/Contents/Frameworks/OBS Helper.app"
+        cp -R "rundir/${BUILD_CONFIG}/bin/OBS Helper (GPU).app" "./OBS.app/Contents/Frameworks/OBS Helper (GPU).app"
+        cp -R "rundir/${BUILD_CONFIG}/bin/OBS Helper (Plugin).app" "./OBS.app/Contents/Frameworks/OBS Helper (Plugin).app"
+        cp -R "rundir/${BUILD_CONFIG}/bin/OBS Helper (Renderer).app" "./OBS.app/Contents/Frameworks/OBS Helper (Renderer).app"
+    fi
     cp -R rundir/${BUILD_CONFIG}/data ./OBS.app/Contents/Resources
     cp ${CI_SCRIPTS}/app/AppIcon.icns ./OBS.app/Contents/Resources
     cp -R rundir/${BUILD_CONFIG}/obs-plugins/ ./OBS.app/Contents/PlugIns
@@ -506,7 +524,10 @@
     codesign --force --options runtime --sign "${CODESIGN_IDENT}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libswiftshader_libEGL.dylib"
     codesign --force --options runtime --sign "${CODESIGN_IDENT}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libGLESv2.dylib"
     codesign --force --options runtime --sign "${CODESIGN_IDENT}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libswiftshader_libGLESv2.dylib"
-    codesign --force --options runtime --sign "${CODESIGN_IDENT}" --deep "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework"
+    if ! [ "${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}" -le 3770 ]; then
+        codesign --force --options runtime --sign "${CODESIGN_IDENT}" "./OBS.app/Contents/Frameworks/Chromium Embedded Framework.framework/Libraries/libvk_swiftshader.dylib"
+    fi
+
     echo -n "${COLOR_RESET}"
 
     step "Code-sign DAL Plugin..."
@@ -518,6 +539,17 @@
     echo -n "${COLOR_ORANGE}"
     codesign --force --options runtime --entitlements "${CI_SCRIPTS}/app/entitlements.plist" --sign "${CODESIGN_IDENT}" --deep ./OBS.app
     echo -n "${COLOR_RESET}"
+
+    if ! [ "${CEF_BUILD_VERSION:-${CI_CEF_VERSION}}" -le 3770 ]; then
+        step "Code-sign CEF helper apps..."
+        echo -n "${COLOR_ORANGE}"
+        codesign --force --options runtime --sign "${CODESIGN_IDENT}" --deep "./OBS.app/Contents/Frameworks/OBS Helper.app"
+        codesign --force --options runtime --entitlements "${CI_SCRIPTS}/helpers/helper-gpu-entitlements.plist" --sign "${CODESIGN_IDENT}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (GPU).app"
+        codesign --force --options runtime --entitlements "${CI_SCRIPTS}/helpers/helper-plugin-entitlements.plist" --sign "${CODESIGN_IDENT}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (Plugin).app"
+        codesign --force --options runtime --entitlements "${CI_SCRIPTS}/helpers/helper-renderer-entitlements.plist" --sign "${CODESIGN_IDENT}" --deep "./OBS.app/Contents/Frameworks/OBS Helper (Renderer).app"
+        echo -n "${COLOR_RESET}"
+    fi
+
     step "Check code-sign result..."
     codesign -dvv ./OBS.app
 }

obs-studio-26.1.0.tar.xz/CI/scripts/macos/Brewfile -> obs-studio-26.1.1.tar.xz/CI/scripts/macos/Brewfile Changed

obs-studio-26.1.1.tar.xz/CI/scripts/macos/helpers Added

obs-studio-26.1.1.tar.xz/CI/scripts/macos/helpers/helper-gpu-entitlements.plist Added

obs-studio-26.1.1.tar.xz/CI/scripts/macos/helpers/helper-plugin-entitlements.plist Added

obs-studio-26.1.1.tar.xz/CI/scripts/macos/helpers/helper-renderer-entitlements.plist Added

obs-studio-26.1.0.tar.xz/CMakeLists.txt -> obs-studio-26.1.1.tar.xz/CMakeLists.txt Changed

obs-studio-26.1.0.tar.xz/UI/data/locale.ini -> obs-studio-26.1.1.tar.xz/UI/data/locale.ini Changed

obs-studio-26.1.0.tar.xz/UI/data/themes/Acri.qss -> obs-studio-26.1.1.tar.xz/UI/data/themes/Acri.qss Changed

obs-studio-26.1.0.tar.xz/UI/installer/mp-installer.nsi -> obs-studio-26.1.1.tar.xz/UI/installer/mp-installer.nsi Changed

obs-studio-26.1.0.tar.xz/UI/win-update/updater/updater.cpp -> obs-studio-26.1.1.tar.xz/UI/win-update/updater/updater.cpp Changed

@@ -298,6 +298,8 @@
 			} else {
 				DeleteFile(outputPath.c_str());
 			}
+			if (state == STATE_INSTALL_FAILED)
+				DeleteFile(tempPath.c_str());
 		} else if (state == STATE_DOWNLOADED) {
 			DeleteFile(tempPath.c_str());
 		}
@@ -337,7 +339,10 @@
 
 bool DownloadWorkerThread()
 {
-	const DWORD tlsProtocols = WINHTTP_FLAG_SECURE_PROTOCOL_TLS1_2;
+	const DWORD tlsProtocols = WINHTTP_FLAG_SECURE_PROTOCOL_TLS1_2 |
+				   WINHTTP_FLAG_SECURE_PROTOCOL_TLS1_3;
+
+	const DWORD enableHTTP2Flag = WINHTTP_PROTOCOL_FLAG_HTTP2;
 
 	HttpHandle hSession = WinHttpOpen(L"OBS Studio Updater/2.1",
 					  WINHTTP_ACCESS_TYPE_DEFAULT_PROXY,
@@ -352,6 +357,9 @@
 	WinHttpSetOption(hSession, WINHTTP_OPTION_SECURE_PROTOCOLS,
 			 (LPVOID)&tlsProtocols, sizeof(tlsProtocols));
 
+	WinHttpSetOption(hSession, WINHTTP_OPTION_ENABLE_HTTP_PROTOCOL,
+			 (LPVOID)&enableHTTP2Flag, sizeof(enableHTTP2Flag));
+
 	HttpHandle hConnect = WinHttpConnect(hSession,
 					     L"cdn-fastly.obsproject.com",
 					     INTERNET_DEFAULT_HTTPS_PORT, 0);
@@ -784,6 +792,41 @@
 	}
 }
 
+static bool MoveInUseFileAway(update_t &file)
+{
+	_TCHAR deleteMeName[MAX_PATH];
+	_TCHAR randomStr[MAX_PATH];
+
+	BYTE junk[40];
+	BYTE hash[BLAKE2_HASH_LENGTH];
+
+	CryptGenRandom(hProvider, sizeof(junk), junk);
+	blake2b(hash, sizeof(hash), junk, sizeof(junk), NULL, 0);
+	HashToString(hash, randomStr);
+	randomStr[8] = 0;
+
+	StringCbCopy(deleteMeName, sizeof(deleteMeName),
+		     file.outputPath.c_str());
+
+	StringCbCat(deleteMeName, sizeof(deleteMeName), L".");
+	StringCbCat(deleteMeName, sizeof(deleteMeName), randomStr);
+	StringCbCat(deleteMeName, sizeof(deleteMeName), L".deleteme");
+
+	if (MoveFile(file.outputPath.c_str(), deleteMeName)) {
+
+		if (MyCopyFile(deleteMeName, file.outputPath.c_str())) {
+			MoveFileEx(deleteMeName, NULL,
+				   MOVEFILE_DELAY_UNTIL_REBOOT);
+
+			return true;
+		} else {
+			MoveFile(deleteMeName, file.outputPath.c_str());
+		}
+	}
+
+	return false;
+}
+
 static bool UpdateFile(update_t &file)
 {
 	wchar_t oldFileRenamedPath[MAX_PATH];
@@ -836,6 +879,9 @@
 
 		int error_code;
 		bool installed_ok;
+		bool already_tried_to_move = false;
+
+	retryAfterMovingFile:
 
 		if (file.patchable) {
 			error_code = ApplyPatch(file.tempPath.c_str(),
@@ -875,15 +921,23 @@
 			int is_sharing_violation =
 				(error_code == ERROR_SHARING_VIOLATION);
 
-			if (is_sharing_violation)
+			if (is_sharing_violation) {
+				if (!already_tried_to_move) {
+					already_tried_to_move = true;
+
+					if (MoveInUseFileAway(file))
+						goto retryAfterMovingFile;
+				}
+
 				Status(L"Update failed: %s is still in use.  "
 				       L"Close all "
 				       L"programs and try again.",
 				       curFileName);
-			else
+			} else {
 				Status(L"Update failed: Couldn't update %s "
 				       L"(error %d)",
 				       curFileName, GetLastError());
+			}
 
 			file.state = STATE_INSTALL_FAILED;
 			return false;
@@ -1390,7 +1444,7 @@
 	/* ------------------------------------- *
 	 * Download Updates                      */
 
-	if (!RunDownloadWorkers(2))
+	if (!RunDownloadWorkers(4))
 		return false;
 
 	if ((size_t)completedUpdates != updates.size()) {

obs-studio-26.1.0.tar.xz/UI/window-basic-auto-config.cpp -> obs-studio-26.1.1.tar.xz/UI/window-basic-auto-config.cpp Changed

@@ -373,8 +373,6 @@
 	if (!wiz->customServer) {
 		if (wiz->serviceName == "Twitch")
 			wiz->service = AutoConfig::Service::Twitch;
-		else if (wiz->serviceName == "Smashcast")
-			wiz->service = AutoConfig::Service::Smashcast;
 		else
 			wiz->service = AutoConfig::Service::Other;
 	} else {
@@ -504,7 +502,7 @@
 		return;
 
 	std::string service = QT_TO_UTF8(ui->service->currentText());
-	bool regionBased = service == "Twitch" || service == "Smashcast";
+	bool regionBased = service == "Twitch";
 	bool testBandwidth = ui->doBandwidthTest->isChecked();
 	bool custom = IsCustomService();
 
@@ -928,21 +926,6 @@
 		} else if (regionOther) {
 			return true;
 		}
-	} else if (service == Service::Smashcast) {
-		if (strcmp(server, "Default") == 0) {
-			return true;
-		} else if (astrcmp_n(server, "US-West:", 8) == 0 ||
-			   astrcmp_n(server, "US-East:", 8) == 0) {
-			return regionUS;
-		} else if (astrcmp_n(server, "EU-", 3) == 0) {
-			return regionEU;
-		} else if (astrcmp_n(server, "South Korea:", 12) == 0 ||
-			   astrcmp_n(server, "Asia:", 5) == 0 ||
-			   astrcmp_n(server, "China:", 6) == 0) {
-			return regionAsia;
-		} else if (regionOther) {
-			return true;
-		}
 	} else {
 		return true;
 	}

obs-studio-26.1.0.tar.xz/UI/window-basic-auto-config.hpp -> obs-studio-26.1.1.tar.xz/UI/window-basic-auto-config.hpp Changed

obs-studio-26.1.0.tar.xz/UI/window-basic-main.cpp -> obs-studio-26.1.1.tar.xz/UI/window-basic-main.cpp Changed

obs-studio-26.1.0.tar.xz/cmake/Modules/FindLibcurl.cmake -> obs-studio-26.1.1.tar.xz/cmake/Modules/FindLibcurl.cmake Changed

@@ -36,29 +36,45 @@
 	PATH_SUFFIXES
 		include)
 
-find_library(CURL_LIB
-	NAMES ${_CURL_LIBRARIES} curl libcurl
-	HINTS
-		ENV curlPath${_lib_suffix}
-		ENV curlPath
-		ENV DepsPath${_lib_suffix}
-		ENV DepsPath
-		${curlPath${_lib_suffix}}
-		${curlPath}
-		${DepsPath${_lib_suffix}}
-		${DepsPath}
-		${_CURL_LIBRARY_DIRS}
-	PATHS
-		/usr/lib /usr/local/lib /opt/local/lib /sw/lib
-	PATH_SUFFIXES
-		lib${_lib_suffix} lib
-		libs${_lib_suffix} libs
-		bin${_lib_suffix} bin
-		../lib${_lib_suffix} ../lib
-		../libs${_lib_suffix} ../libs
-		../bin${_lib_suffix} ../bin
-		"build/Win${_lib_suffix}/VC12/DLL Release - DLL Windows SSPI"
-		"../build/Win${_lib_suffix}/VC12/DLL Release - DLL Windows SSPI")
+if(APPLE)
+	find_library(CURL_LIB
+		NAMES ${_CURL_LIBRARIES} curl libcurl
+		HINTS
+			ENV curlPath${_lib_suffix}
+			ENV curlPath
+			ENV DepsPath${_lib_suffix}
+			ENV DepsPath
+			${curlPath${_lib_suffix}}
+			${curlPath}
+			${DepsPath${_lib_suffix}}
+			${DepsPath}
+			${_CURL_LIBRARY_DIRS}
+		)
+else()
+	find_library(CURL_LIB
+		NAMES ${_CURL_LIBRARIES} curl libcurl
+		HINTS
+			ENV curlPath${_lib_suffix}
+			ENV curlPath
+			ENV DepsPath${_lib_suffix}
+			ENV DepsPath
+			${curlPath${_lib_suffix}}
+			${curlPath}
+			${DepsPath${_lib_suffix}}
+			${DepsPath}
+			${_CURL_LIBRARY_DIRS}
+		PATHS
+			/usr/lib /usr/local/lib /opt/local/lib /sw/lib
+		PATH_SUFFIXES
+			lib${_lib_suffix} lib
+			libs${_lib_suffix} libs
+			bin${_lib_suffix} bin
+			../lib${_lib_suffix} ../lib
+			../libs${_lib_suffix} ../libs
+			../bin${_lib_suffix} ../bin
+			"build/Win${_lib_suffix}/VC12/DLL Release - DLL Windows SSPI"
+			"../build/Win${_lib_suffix}/VC12/DLL Release - DLL Windows SSPI")
+endif()
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(Libcurl DEFAULT_MSG CURL_LIB CURL_INCLUDE_DIR)

obs-studio-26.1.0.tar.xz/docs/sphinx/reference-frontend-api.rst -> obs-studio-26.1.1.tar.xz/docs/sphinx/reference-frontend-api.rst Changed

obs-studio-26.1.0.tar.xz/libobs/CMakeLists.txt -> obs-studio-26.1.1.tar.xz/libobs/CMakeLists.txt Changed

@@ -188,20 +188,8 @@
 		util/pipe-posix.c
 		util/platform-nix.c)
 
-	if(NEEDS_SIMDE)
-		set(libobs_PLATFORM_HEADERS
-			util/simde/check.h
-			util/simde/hedley.h
-			util/simde/mmx.h
-			util/simde/simde-arch.h
-			util/simde/simde-common.h
-			util/simde/sse.h
-			util/simde/sse2.h
-			util/threading-posix.h)
-	else()
-		set(libobs_PLATFORM_HEADERS
-			util/threading-posix.h)
-	endif()
+	set(libobs_PLATFORM_HEADERS
+		util/threading-posix.h)
 
 	if(HAVE_PULSEAUDIO)
 		set(libobs_audio_monitoring_HEADERS
@@ -369,7 +357,6 @@
 set(libobs_util_HEADERS
 	util/curl/curl-helper.h
 	util/sse-intrin.h
-	util/sse2neon.h
 	util/array-serializer.h
 	util/file-serializer.h
 	util/utf8.h
@@ -419,6 +406,20 @@
 	obs-video-gpu-encode.c
 	obs-video.c)
 set(libobs_libobs_HEADERS
+	util/simde/check.h
+	util/simde/debug-trap.h
+	util/simde/hedley.h
+	util/simde/simde-align.h
+	util/simde/simde-arch.h
+	util/simde/simde-common.h
+	util/simde/simde-constify.h
+	util/simde/simde-detect-clang.h
+	util/simde/simde-diagnostic.h
+	util/simde/simde-features.h
+	util/simde/simde-math.h
+	util/simde/x86/mmx.h
+	util/simde/x86/sse2.h
+	util/simde/x86/sse.h
 	${libobs_PLATFORM_HEADERS}
 	obs-audio-controls.h
 	obs-defs.h
@@ -499,6 +500,10 @@
 	PUBLIC
 		HAVE_OBSCONFIG_H)
 
+target_compile_definitions(libobs
+	PUBLIC
+	${ARCH_SIMD_DEFINES})
+
 target_compile_options(libobs
 	PUBLIC
 	${ARCH_SIMD_FLAGS})

obs-studio-26.1.0.tar.xz/libobs/media-io/media-remux.c -> obs-studio-26.1.1.tar.xz/libobs/media-io/media-remux.c Changed

obs-studio-26.1.0.tar.xz/libobs/obs-config.h -> obs-studio-26.1.1.tar.xz/libobs/obs-config.h Changed

obs-studio-26.1.0.tar.xz/libobs/obs-scene.c -> obs-studio-26.1.1.tar.xz/libobs/obs-scene.c Changed

@@ -951,19 +951,12 @@
 }
 
 static void apply_scene_item_audio_actions(struct obs_scene_item *item,
-					   float **p_buf, uint64_t ts,
+					   float *buf, uint64_t ts,
 					   size_t sample_rate)
 {
 	bool cur_visible = item->visible;
 	uint64_t frame_num = 0;
 	size_t deref_count = 0;
-	float *buf = NULL;
-
-	if (p_buf) {
-		if (!*p_buf)
-			*p_buf = malloc(AUDIO_OUTPUT_FRAMES * sizeof(float));
-		buf = *p_buf;
-	}
 
 	pthread_mutex_lock(&item->actions_mutex);
 
@@ -1010,7 +1003,7 @@
 	}
 }
 
-static bool apply_scene_item_volume(struct obs_scene_item *item, float **buf,
+static bool apply_scene_item_volume(struct obs_scene_item *item, float *buf,
 				    uint64_t ts, size_t sample_rate)
 {
 	bool actions_pending;
@@ -1074,7 +1067,7 @@
 			       size_t sample_rate)
 {
 	uint64_t timestamp = 0;
-	float *buf = NULL;
+	float buf[AUDIO_OUTPUT_FRAMES];
 	struct obs_source_audio_mix child_audio;
 	struct obs_scene *scene = data;
 	struct obs_scene_item *item;
@@ -1113,7 +1106,7 @@
 		size_t pos, count;
 		bool apply_buf;
 
-		apply_buf = apply_scene_item_volume(item, &buf, timestamp,
+		apply_buf = apply_scene_item_volume(item, buf, timestamp,
 						    sample_rate);
 
 		if (obs_source_audio_pending(item->source)) {
@@ -1159,7 +1152,6 @@
 	*ts_out = timestamp;
 	audio_unlock(scene);
 
-	free(buf);
 	return true;
 }
 
@@ -1300,6 +1292,7 @@
 	}
 
 	obs_sceneitem_set_crop(dst, &src->crop);
+	obs_sceneitem_set_locked(dst, src->locked);
 
 	if (defer_texture_update) {
 		os_atomic_set_bool(&dst->update_transform, true);

obs-studio-26.1.0.tar.xz/libobs/obs-source.c -> obs-studio-26.1.1.tar.xz/libobs/obs-source.c Changed

obs-studio-26.1.0.tar.xz/libobs/obsconfig.h.in -> obs-studio-26.1.1.tar.xz/libobs/obsconfig.h.in Changed

obs-studio-26.1.0.tar.xz/libobs/util/simde/README.libobs -> obs-studio-26.1.1.tar.xz/libobs/util/simde/README.libobs Changed

obs-studio-26.1.0.tar.xz/libobs/util/simde/check.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/check.h Changed

obs-studio-26.1.0.tar.xz/libobs/util/simde/hedley.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/hedley.h Changed

@@ -10,11 +10,11 @@
 * SPDX-License-Identifier: CC0-1.0
 */
 
-#if !defined(HEDLEY_VERSION) || (HEDLEY_VERSION < 12)
+#if !defined(HEDLEY_VERSION) || (HEDLEY_VERSION < 14)
 #if defined(HEDLEY_VERSION)
 #undef HEDLEY_VERSION
 #endif
-#define HEDLEY_VERSION 12
+#define HEDLEY_VERSION 14
 
 #if defined(HEDLEY_STRINGIFY_EX)
 #undef HEDLEY_STRINGIFY_EX
@@ -36,6 +36,16 @@
 #endif
 #define HEDLEY_CONCAT(a, b) HEDLEY_CONCAT_EX(a, b)
 
+#if defined(HEDLEY_CONCAT3_EX)
+#undef HEDLEY_CONCAT3_EX
+#endif
+#define HEDLEY_CONCAT3_EX(a, b, c) a##b##c
+
+#if defined(HEDLEY_CONCAT3)
+#undef HEDLEY_CONCAT3
+#endif
+#define HEDLEY_CONCAT3(a, b, c) HEDLEY_CONCAT3_EX(a, b, c)
+
 #if defined(HEDLEY_VERSION_ENCODE)
 #undef HEDLEY_VERSION_ENCODE
 #endif
@@ -80,17 +90,17 @@
 #if defined(HEDLEY_MSVC_VERSION)
 #undef HEDLEY_MSVC_VERSION
 #endif
-#if defined(_MSC_FULL_VER) && (_MSC_FULL_VER >= 140000000)
+#if defined(_MSC_FULL_VER) && (_MSC_FULL_VER >= 140000000) && !defined(__ICL)
 #define HEDLEY_MSVC_VERSION \
 	HEDLEY_VERSION_ENCODE(_MSC_FULL_VER / 10000000, \
 			 (_MSC_FULL_VER % 10000000) / 100000, \
 			 (_MSC_FULL_VER % 100000) / 100)
-#elif defined(_MSC_FULL_VER)
+#elif defined(_MSC_FULL_VER) && !defined(__ICL)
 #define HEDLEY_MSVC_VERSION \
 	HEDLEY_VERSION_ENCODE(_MSC_FULL_VER / 1000000, \
 			 (_MSC_FULL_VER % 1000000) / 10000, \
 			 (_MSC_FULL_VER % 10000) / 10)
-#elif defined(_MSC_VER)
+#elif defined(_MSC_VER) && !defined(__ICL)
 #define HEDLEY_MSVC_VERSION \
 	HEDLEY_VERSION_ENCODE(_MSC_VER / 100, _MSC_VER % 100, 0)
 #endif
@@ -98,7 +108,7 @@
 #if defined(HEDLEY_MSVC_VERSION_CHECK)
 #undef HEDLEY_MSVC_VERSION_CHECK
 #endif
-#if !defined(_MSC_VER)
+#if !defined(HEDLEY_MSVC_VERSION)
 #define HEDLEY_MSVC_VERSION_CHECK(major, minor, patch) (0)
 #elif defined(_MSC_VER) && (_MSC_VER >= 1400)
 #define HEDLEY_MSVC_VERSION_CHECK(major, minor, patch) \
@@ -114,11 +124,12 @@
 #if defined(HEDLEY_INTEL_VERSION)
 #undef HEDLEY_INTEL_VERSION
 #endif
-#if defined(__INTEL_COMPILER) && defined(__INTEL_COMPILER_UPDATE)
+#if defined(__INTEL_COMPILER) && defined(__INTEL_COMPILER_UPDATE) && \
+	!defined(__ICL)
 #define HEDLEY_INTEL_VERSION \
 	HEDLEY_VERSION_ENCODE(__INTEL_COMPILER / 100, __INTEL_COMPILER % 100, \
 			 __INTEL_COMPILER_UPDATE)
-#elif defined(__INTEL_COMPILER)
+#elif defined(__INTEL_COMPILER) && !defined(__ICL)
 #define HEDLEY_INTEL_VERSION \
 	HEDLEY_VERSION_ENCODE(__INTEL_COMPILER / 100, __INTEL_COMPILER % 100, 0)
 #endif
@@ -133,6 +144,25 @@
 #define HEDLEY_INTEL_VERSION_CHECK(major, minor, patch) (0)
 #endif
 
+#if defined(HEDLEY_INTEL_CL_VERSION)
+#undef HEDLEY_INTEL_CL_VERSION
+#endif
+#if defined(__INTEL_COMPILER) && defined(__INTEL_COMPILER_UPDATE) && \
+	defined(__ICL)
+#define HEDLEY_INTEL_CL_VERSION \
+	HEDLEY_VERSION_ENCODE(__INTEL_COMPILER, __INTEL_COMPILER_UPDATE, 0)
+#endif
+
+#if defined(HEDLEY_INTEL_CL_VERSION_CHECK)
+#undef HEDLEY_INTEL_CL_VERSION_CHECK
+#endif
+#if defined(HEDLEY_INTEL_CL_VERSION)
+#define HEDLEY_INTEL_CL_VERSION_CHECK(major, minor, patch) \
+	(HEDLEY_INTEL_CL_VERSION >= HEDLEY_VERSION_ENCODE(major, minor, patch))
+#else
+#define HEDLEY_INTEL_CL_VERSION_CHECK(major, minor, patch) (0)
+#endif
+
 #if defined(HEDLEY_PGI_VERSION)
 #undef HEDLEY_PGI_VERSION
 #endif
@@ -788,6 +818,68 @@
 	HEDLEY_GCC_VERSION_CHECK(major, minor, patch)
 #endif
 
+#if (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)) || \
+	defined(__clang__) || HEDLEY_GCC_VERSION_CHECK(3, 0, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
+	HEDLEY_IAR_VERSION_CHECK(8, 0, 0) || \
+	HEDLEY_PGI_VERSION_CHECK(18, 4, 0) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
+	HEDLEY_TI_ARMCL_VERSION_CHECK(4, 7, 0) || \
+	HEDLEY_TI_CL430_VERSION_CHECK(2, 0, 1) || \
+	HEDLEY_TI_CL2000_VERSION_CHECK(6, 1, 0) || \
+	HEDLEY_TI_CL6X_VERSION_CHECK(7, 0, 0) || \
+	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
+	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0) || \
+	HEDLEY_CRAY_VERSION_CHECK(5, 0, 0) || \
+	HEDLEY_TINYC_VERSION_CHECK(0, 9, 17) || \
+	HEDLEY_SUNPRO_VERSION_CHECK(8, 0, 0) || \
+	(HEDLEY_IBM_VERSION_CHECK(10, 1, 0) && defined(__C99_PRAGMA_OPERATOR))
+#define HEDLEY_PRAGMA(value) _Pragma(#value)
+#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0)
+#define HEDLEY_PRAGMA(value) __pragma(value)
+#else
+#define HEDLEY_PRAGMA(value)
+#endif
+
+#if defined(HEDLEY_DIAGNOSTIC_PUSH)
+#undef HEDLEY_DIAGNOSTIC_PUSH
+#endif
+#if defined(HEDLEY_DIAGNOSTIC_POP)
+#undef HEDLEY_DIAGNOSTIC_POP
+#endif
+#if defined(__clang__)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("clang diagnostic push")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("clang diagnostic pop")
+#elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("warning(push)")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("warning(pop)")
+#elif HEDLEY_GCC_VERSION_CHECK(4, 6, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("GCC diagnostic push")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("GCC diagnostic pop")
+#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH __pragma(warning(push))
+#define HEDLEY_DIAGNOSTIC_POP __pragma(warning(pop))
+#elif HEDLEY_ARM_VERSION_CHECK(5, 6, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("push")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("pop")
+#elif HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
+	HEDLEY_TI_ARMCL_VERSION_CHECK(5, 2, 0) || \
+	HEDLEY_TI_CL430_VERSION_CHECK(4, 4, 0) || \
+	HEDLEY_TI_CL6X_VERSION_CHECK(8, 1, 0) || \
+	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
+	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("diag_push")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("diag_pop")
+#elif HEDLEY_PELLES_VERSION_CHECK(2, 90, 0)
+#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("warning(push)")
+#define HEDLEY_DIAGNOSTIC_POP _Pragma("warning(pop)")
+#else
+#define HEDLEY_DIAGNOSTIC_PUSH
+#define HEDLEY_DIAGNOSTIC_POP
+#endif
+
 /* HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_ is for
 HEDLEY INTERNAL USE ONLY. API subject to change without notice. */
 #if defined(HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_)
@@ -796,11 +888,20 @@
 #if defined(__cplusplus)
 #if HEDLEY_HAS_WARNING("-Wc++98-compat")
 #if HEDLEY_HAS_WARNING("-Wc++17-extensions")
+#if HEDLEY_HAS_WARNING("-Wc++1z-extensions")
+#define HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_(xpr) \
+	HEDLEY_DIAGNOSTIC_PUSH \
+	_Pragma("clang diagnostic ignored \"-Wc++98-compat\"") _Pragma( \
+		"clang diagnostic ignored \"-Wc++17-extensions\"") \
+		_Pragma("clang diagnostic ignored \"-Wc++1z-extensions\"") \
+			xpr HEDLEY_DIAGNOSTIC_POP
+#else
 #define HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_(xpr) \
 	HEDLEY_DIAGNOSTIC_PUSH \
 	_Pragma("clang diagnostic ignored \"-Wc++98-compat\"") \
 		_Pragma("clang diagnostic ignored \"-Wc++17-extensions\"") \
 			xpr HEDLEY_DIAGNOSTIC_POP
+#endif
 #else
 #define HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_(xpr) \
 	HEDLEY_DIAGNOSTIC_PUSH \
@@ -861,74 +962,14 @@
 #elif HEDLEY_IAR_VERSION_CHECK(8, 3, 0)
 #define HEDLEY_CPP_CAST(T, expr) \
 	HEDLEY_DIAGNOSTIC_PUSH \
-	_Pragma("diag_suppress=Pe137") HEDLEY_DIAGNOSTIC_POP #else
+	_Pragma("diag_suppress=Pe137") HEDLEY_DIAGNOSTIC_POP
+#else
 #define HEDLEY_CPP_CAST(T, expr) ((T)(expr))
 #endif
 #else
 #define HEDLEY_CPP_CAST(T, expr) (expr)
 #endif
 
-#if (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)) || \
-	defined(__clang__) || HEDLEY_GCC_VERSION_CHECK(3, 0, 0) || \
-	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
-	HEDLEY_IAR_VERSION_CHECK(8, 0, 0) || \
-	HEDLEY_PGI_VERSION_CHECK(18, 4, 0) || \
-	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
-	HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
-	HEDLEY_TI_ARMCL_VERSION_CHECK(4, 7, 0) || \
-	HEDLEY_TI_CL430_VERSION_CHECK(2, 0, 1) || \
-	HEDLEY_TI_CL2000_VERSION_CHECK(6, 1, 0) || \
-	HEDLEY_TI_CL6X_VERSION_CHECK(7, 0, 0) || \
-	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
-	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0) || \
-	HEDLEY_CRAY_VERSION_CHECK(5, 0, 0) || \
-	HEDLEY_TINYC_VERSION_CHECK(0, 9, 17) || \
-	HEDLEY_SUNPRO_VERSION_CHECK(8, 0, 0) || \
-	(HEDLEY_IBM_VERSION_CHECK(10, 1, 0) && defined(__C99_PRAGMA_OPERATOR))
-#define HEDLEY_PRAGMA(value) _Pragma(#value)
-#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0)
-#define HEDLEY_PRAGMA(value) __pragma(value)
-#else
-#define HEDLEY_PRAGMA(value)
-#endif
-
-#if defined(HEDLEY_DIAGNOSTIC_PUSH)
-#undef HEDLEY_DIAGNOSTIC_PUSH
-#endif
-#if defined(HEDLEY_DIAGNOSTIC_POP)
-#undef HEDLEY_DIAGNOSTIC_POP
-#endif
-#if defined(__clang__)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("clang diagnostic push")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("clang diagnostic pop")
-#elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("warning(push)")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("warning(pop)")
-#elif HEDLEY_GCC_VERSION_CHECK(4, 6, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("GCC diagnostic push")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("GCC diagnostic pop")
-#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH __pragma(warning(push))
-#define HEDLEY_DIAGNOSTIC_POP __pragma(warning(pop))
-#elif HEDLEY_ARM_VERSION_CHECK(5, 6, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("push")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("pop")
-#elif HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
-	HEDLEY_TI_ARMCL_VERSION_CHECK(5, 2, 0) || \
-	HEDLEY_TI_CL430_VERSION_CHECK(4, 4, 0) || \
-	HEDLEY_TI_CL6X_VERSION_CHECK(8, 1, 0) || \
-	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
-	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("diag_push")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("diag_pop")
-#elif HEDLEY_PELLES_VERSION_CHECK(2, 90, 0)
-#define HEDLEY_DIAGNOSTIC_PUSH _Pragma("warning(push)")
-#define HEDLEY_DIAGNOSTIC_POP _Pragma("warning(pop)")
-#else
-#define HEDLEY_DIAGNOSTIC_PUSH
-#define HEDLEY_DIAGNOSTIC_POP
-#endif
-
 #if defined(HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED)
 #undef HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED
 #endif
@@ -938,6 +979,12 @@
 #elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED \
 	_Pragma("warning(disable:1478 1786)")
+#elif HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
+#define HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED \
+	__pragma(warning(disable : 1478 1786))
+#elif HEDLEY_PGI_VERSION_CHECK(20, 7, 0)
+#define HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED \
+	_Pragma("diag_suppress 1215,1216,1444,1445")
 #elif HEDLEY_PGI_VERSION_CHECK(17, 10, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_DEPRECATED _Pragma("diag_suppress 1215,1444")
 #elif HEDLEY_GCC_VERSION_CHECK(4, 3, 0)
@@ -985,6 +1032,9 @@
 #elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_PRAGMAS \
 	_Pragma("warning(disable:161)")
+#elif HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
+#define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_PRAGMAS \
+	__pragma(warning(disable : 161))
 #elif HEDLEY_PGI_VERSION_CHECK(17, 10, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_PRAGMAS _Pragma("diag_suppress 1675")
 #elif HEDLEY_GCC_VERSION_CHECK(4, 3, 0)
@@ -1018,9 +1068,15 @@
 #elif HEDLEY_INTEL_VERSION_CHECK(17, 0, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_CPP_ATTRIBUTES \
 	_Pragma("warning(disable:1292)")
+#elif HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
+#define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_CPP_ATTRIBUTES \
+	__pragma(warning(disable : 1292))
 #elif HEDLEY_MSVC_VERSION_CHECK(19, 0, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_CPP_ATTRIBUTES \
 	__pragma(warning(disable : 5030))
+#elif HEDLEY_PGI_VERSION_CHECK(20, 7, 0)
+#define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_CPP_ATTRIBUTES \
+	_Pragma("diag_suppress 1097,1098")
 #elif HEDLEY_PGI_VERSION_CHECK(17, 10, 0)
 #define HEDLEY_DIAGNOSTIC_DISABLE_UNKNOWN_CPP_ATTRIBUTES \
 	_Pragma("diag_suppress 1097")
@@ -1061,13 +1117,11 @@
 #if defined(HEDLEY_DEPRECATED_FOR)
 #undef HEDLEY_DEPRECATED_FOR
 #endif
-#if defined(__cplusplus) && (__cplusplus >= 201402L)
-#define HEDLEY_DEPRECATED(since) \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
-		[[deprecated("Since " #since)]])
-#define HEDLEY_DEPRECATED_FOR(since, replacement) \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
-		[[deprecated("Since " #since "; use " #replacement)]])
+#if HEDLEY_MSVC_VERSION_CHECK(14, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
+#define HEDLEY_DEPRECATED(since) __declspec(deprecated("Since " #since))
+#define HEDLEY_DEPRECATED_FOR(since, replacement) \
+	__declspec(deprecated("Since " #since "; use " #replacement))
 #elif HEDLEY_HAS_EXTENSION(attribute_deprecated_with_message) || \
 	HEDLEY_GCC_VERSION_CHECK(4, 5, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
@@ -1083,6 +1137,13 @@
 	__attribute__((__deprecated__("Since " #since)))
 #define HEDLEY_DEPRECATED_FOR(since, replacement) \
 	__attribute__((__deprecated__("Since " #since "; use " #replacement)))
+#elif defined(__cplusplus) && (__cplusplus >= 201402L)
+#define HEDLEY_DEPRECATED(since) \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
+		[[deprecated("Since " #since)]])
+#define HEDLEY_DEPRECATED_FOR(since, replacement) \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
+		[[deprecated("Since " #since "; use " #replacement)]])
 #elif HEDLEY_HAS_ATTRIBUTE(deprecated) || HEDLEY_GCC_VERSION_CHECK(3, 1, 0) || \
 	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
 	HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
@@ -1103,12 +1164,9 @@
 #define HEDLEY_DEPRECATED(since) __attribute__((__deprecated__))
 #define HEDLEY_DEPRECATED_FOR(since, replacement) \
 	__attribute__((__deprecated__))
-#elif HEDLEY_MSVC_VERSION_CHECK(14, 0, 0)
-#define HEDLEY_DEPRECATED(since) __declspec(deprecated("Since " #since))
-#define HEDLEY_DEPRECATED_FOR(since, replacement) \
-	__declspec(deprecated("Since " #since "; use " #replacement))
-#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || \
-	HEDLEY_PELLES_VERSION_CHECK(6, 50, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || \
+	HEDLEY_PELLES_VERSION_CHECK(6, 50, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_DEPRECATED(since) __declspec(deprecated)
 #define HEDLEY_DEPRECATED_FOR(since, replacement) __declspec(deprecated)
 #elif HEDLEY_IAR_VERSION_CHECK(8, 0, 0)
@@ -1136,17 +1194,7 @@
 #if defined(HEDLEY_WARN_UNUSED_RESULT_MSG)
 #undef HEDLEY_WARN_UNUSED_RESULT_MSG
 #endif
-#if (HEDLEY_HAS_CPP_ATTRIBUTE(nodiscard) >= 201907L)
-#define HEDLEY_WARN_UNUSED_RESULT \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
-#define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard(msg)]])
-#elif HEDLEY_HAS_CPP_ATTRIBUTE(nodiscard)
-#define HEDLEY_WARN_UNUSED_RESULT \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
-#define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) \
-	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
-#elif HEDLEY_HAS_ATTRIBUTE(warn_unused_result) || \
+#if HEDLEY_HAS_ATTRIBUTE(warn_unused_result) || \
 	HEDLEY_GCC_VERSION_CHECK(3, 4, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
 	HEDLEY_TI_VERSION_CHECK(15, 12, 0) || \
@@ -1169,6 +1217,16 @@
 #define HEDLEY_WARN_UNUSED_RESULT __attribute__((__warn_unused_result__))
 #define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) \
 	__attribute__((__warn_unused_result__))
+#elif (HEDLEY_HAS_CPP_ATTRIBUTE(nodiscard) >= 201907L)
+#define HEDLEY_WARN_UNUSED_RESULT \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
+#define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard(msg)]])
+#elif HEDLEY_HAS_CPP_ATTRIBUTE(nodiscard)
+#define HEDLEY_WARN_UNUSED_RESULT \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
+#define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_([[nodiscard]])
 #elif defined(_Check_return_) /* SAL */
 #define HEDLEY_WARN_UNUSED_RESULT _Check_return_
 #define HEDLEY_WARN_UNUSED_RESULT_MSG(msg) _Check_return_
@@ -1222,7 +1280,8 @@
 #define HEDLEY_NO_RETURN __attribute__((__noreturn__))
 #elif HEDLEY_SUNPRO_VERSION_CHECK(5, 10, 0)
 #define HEDLEY_NO_RETURN _Pragma("does_not_return")
-#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_NO_RETURN __declspec(noreturn)
 #elif HEDLEY_TI_CL6X_VERSION_CHECK(6, 0, 0) && defined(__cplusplus)
 #define HEDLEY_NO_RETURN _Pragma("FUNC_NEVER_RETURNS;")
@@ -1252,7 +1311,9 @@
 #if defined(HEDLEY_ASSUME)
 #undef HEDLEY_ASSUME
 #endif
-#if HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+#if HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_ASSUME(expr) __assume(expr)
 #elif HEDLEY_HAS_BUILTIN(__builtin_assume)
 #define HEDLEY_ASSUME(expr) __builtin_assume(expr)
@@ -1389,7 +1450,8 @@
 #if HEDLEY_HAS_BUILTIN(__builtin_unpredictable)
 #define HEDLEY_UNPREDICTABLE(expr) __builtin_unpredictable((expr))
 #endif
-#if HEDLEY_HAS_BUILTIN(__builtin_expect_with_probability) || \
+#if (HEDLEY_HAS_BUILTIN(__builtin_expect_with_probability) && \
+ !defined(HEDLEY_PGI_VERSION)) || \
 	HEDLEY_GCC_VERSION_CHECK(9, 0, 0)
 #define HEDLEY_PREDICT(expr, value, probability) \
 	__builtin_expect_with_probability((expr), (value), (probability))
@@ -1399,7 +1461,8 @@
 	__builtin_expect_with_probability(!!(expr), 0, (probability))
 #define HEDLEY_LIKELY(expr) __builtin_expect(!!(expr), 1)
 #define HEDLEY_UNLIKELY(expr) __builtin_expect(!!(expr), 0)
-#elif HEDLEY_HAS_BUILTIN(__builtin_expect) || \
+#elif (HEDLEY_HAS_BUILTIN(__builtin_expect) && \
+ !defined(HEDLEY_INTEL_CL_VERSION)) || \
 	HEDLEY_GCC_VERSION_CHECK(3, 0, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
 	(HEDLEY_SUNPRO_VERSION_CHECK(5, 15, 0) && defined(__cplusplus)) || \
@@ -1476,7 +1539,8 @@
 #define HEDLEY_MALLOC __attribute__((__malloc__))
 #elif HEDLEY_SUNPRO_VERSION_CHECK(5, 10, 0)
 #define HEDLEY_MALLOC _Pragma("returns_new_memory")
-#elif HEDLEY_MSVC_VERSION_CHECK(14, 0, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(14, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_MALLOC __declspec(restrict)
 #else
 #define HEDLEY_MALLOC
@@ -1557,6 +1621,7 @@
 #elif HEDLEY_GCC_VERSION_CHECK(3, 1, 0) || \
 	HEDLEY_MSVC_VERSION_CHECK(14, 0, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0) || \
 	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
 	HEDLEY_IBM_VERSION_CHECK(10, 1, 0) || \
 	HEDLEY_PGI_VERSION_CHECK(17, 10, 0) || \
@@ -1581,13 +1646,14 @@
 #define HEDLEY_INLINE inline
 #elif defined(HEDLEY_GCC_VERSION) || HEDLEY_ARM_VERSION_CHECK(6, 2, 0)
 #define HEDLEY_INLINE __inline__
-#elif HEDLEY_MSVC_VERSION_CHECK(12, 0, 0) || \
-	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
-	HEDLEY_TI_ARMCL_VERSION_CHECK(5, 1, 0) || \
-	HEDLEY_TI_CL430_VERSION_CHECK(3, 1, 0) || \
-	HEDLEY_TI_CL2000_VERSION_CHECK(6, 2, 0) || \
-	HEDLEY_TI_CL6X_VERSION_CHECK(8, 0, 0) || \
-	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
+#elif HEDLEY_MSVC_VERSION_CHECK(12, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_TI_ARMCL_VERSION_CHECK(5, 1, 0) || \
+	HEDLEY_TI_CL430_VERSION_CHECK(3, 1, 0) || \
+	HEDLEY_TI_CL2000_VERSION_CHECK(6, 2, 0) || \
+	HEDLEY_TI_CL6X_VERSION_CHECK(8, 0, 0) || \
+	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
 	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0)
 #define HEDLEY_INLINE __inline
 #else
@@ -1619,7 +1685,8 @@
 	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
 	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0)
 #define HEDLEY_ALWAYS_INLINE __attribute__((__always_inline__)) HEDLEY_INLINE
-#elif HEDLEY_MSVC_VERSION_CHECK(12, 0, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(12, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_ALWAYS_INLINE __forceinline
 #elif defined(__cplusplus) && (HEDLEY_TI_ARMCL_VERSION_CHECK(5, 2, 0) || \
 			 HEDLEY_TI_CL430_VERSION_CHECK(4, 3, 0) || \
@@ -1658,7 +1725,8 @@
 	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
 	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 1, 0)
 #define HEDLEY_NEVER_INLINE __attribute__((__noinline__))
-#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(13, 10, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_NEVER_INLINE __declspec(noinline)
 #elif HEDLEY_PGI_VERSION_CHECK(10, 2, 0)
 #define HEDLEY_NEVER_INLINE _Pragma("noinline")
@@ -1711,7 +1779,9 @@
 #if HEDLEY_HAS_ATTRIBUTE(nothrow) || HEDLEY_GCC_VERSION_CHECK(3, 3, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
 #define HEDLEY_NO_THROW __attribute__((__nothrow__))
-#elif HEDLEY_MSVC_VERSION_CHECK(13, 1, 0) || HEDLEY_ARM_VERSION_CHECK(4, 1, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(13, 1, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0)
 #define HEDLEY_NO_THROW __declspec(nothrow)
 #else
 #define HEDLEY_NO_THROW
@@ -1720,8 +1790,7 @@
 #if defined(HEDLEY_FALL_THROUGH)
 #undef HEDLEY_FALL_THROUGH
 #endif
-#if HEDLEY_GNUC_HAS_ATTRIBUTE(fallthrough, 7, 0, 0) && \
-	!defined(HEDLEY_PGI_VERSION)
+#if HEDLEY_HAS_ATTRIBUTE(fallthrough) || HEDLEY_GCC_VERSION_CHECK(7, 0, 0)
 #define HEDLEY_FALL_THROUGH __attribute__((__fallthrough__))
 #elif HEDLEY_HAS_CPP_ATTRIBUTE_NS(clang, fallthrough)
 #define HEDLEY_FALL_THROUGH \
@@ -1866,12 +1935,14 @@
 #endif
 #if !defined(__cplusplus) && \
 	((defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)) || \
-	 HEDLEY_HAS_FEATURE(c_static_assert) || \
+	 (HEDLEY_HAS_FEATURE(c_static_assert) && \
+	 !defined(HEDLEY_INTEL_CL_VERSION)) || \
 	 HEDLEY_GCC_VERSION_CHECK(6, 0, 0) || \
 	 HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || defined(_Static_assert))
 #define HEDLEY_STATIC_ASSERT(expr, message) _Static_assert(expr, message)
 #elif (defined(__cplusplus) && (__cplusplus >= 201103L)) || \
-	HEDLEY_MSVC_VERSION_CHECK(16, 0, 0)
+	HEDLEY_MSVC_VERSION_CHECK(16, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_STATIC_ASSERT(expr, message) \
 	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
 		static_assert(expr, message))
@@ -1930,7 +2001,8 @@
 	HEDLEY_PGI_VERSION_CHECK(18, 4, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
 #define HEDLEY_WARNING(msg) HEDLEY_PRAGMA(GCC warning msg)
-#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0)
+#elif HEDLEY_MSVC_VERSION_CHECK(15, 0, 0) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_WARNING(msg) HEDLEY_PRAGMA(message(msg))
 #else
 #define HEDLEY_WARNING(msg) HEDLEY_MESSAGE(msg)
@@ -1970,6 +2042,8 @@
 #endif
 #if HEDLEY_HAS_ATTRIBUTE(flag_enum)
 #define HEDLEY_FLAGS __attribute__((__flag_enum__))
+#else
+#define HEDLEY_FLAGS
 #endif
 
 #if defined(HEDLEY_FLAGS_CAST)
@@ -1989,8 +2063,9 @@
 #if defined(HEDLEY_EMPTY_BASES)
 #undef HEDLEY_EMPTY_BASES
 #endif
-#if HEDLEY_MSVC_VERSION_CHECK(19, 0, 23918) && \
-	!HEDLEY_MSVC_VERSION_CHECK(20, 0, 0)
+#if (HEDLEY_MSVC_VERSION_CHECK(19, 0, 23918) && \
+ !HEDLEY_MSVC_VERSION_CHECK(20, 0, 0)) || \
+	HEDLEY_INTEL_CL_VERSION_CHECK(2021, 1, 0)
 #define HEDLEY_EMPTY_BASES __declspec(empty_bases)
 #else
 #define HEDLEY_EMPTY_BASES

obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-align.h Added

@@ -0,0 +1,481 @@
+/* Alignment
+ * Created by Evan Nemerson <evan@nemerson.com>
+ *
+ * To the extent possible under law, the authors have waived all
+ * copyright and related or neighboring rights to this code. For
+ * details, see the Creative Commons Zero 1.0 Universal license at
+ * <https://creativecommons.org/publicdomain/zero/1.0/>
+ *
+ * SPDX-License-Identifier: CC0-1.0
+ *
+ **********************************************************************
+ *
+ * This is portability layer which should help iron out some
+ * differences across various compilers, as well as various verisons of
+ * C and C++.
+ *
+ * It was originally developed for SIMD Everywhere
+ * (<https://github.com/simd-everywhere/simde>), but since its only
+ * dependency is Hedley (<https://nemequ.github.io/hedley>, also CC0)
+ * it can easily be used in other projects, so please feel free to do
+ * so.
+ *
+ * If you do use this in your project, please keep a link to SIMDe in
+ * your code to remind you where to report any bugs and/or check for
+ * updated versions.
+ *
+ * # API Overview
+ *
+ * The API has several parts, and most macros have a few variations.
+ * There are APIs for declaring aligned fields/variables, optimization
+ * hints, and run-time alignment checks.
+ *
+ * Briefly, macros ending with "_TO" take numeric values and are great
+ * when you know the value you would like to use. Macros ending with
+ * "_LIKE", on the other hand, accept a type and are used when you want
+ * to use the alignment of a type instead of hardcoding a value.
+ *
+ * Documentation for each section of the API is inline.
+ *
+ * True to form, MSVC is the main problem and imposes several
+ * limitations on the effectiveness of the APIs. Detailed descriptions
+ * of the limitations of each macro are inline, but in general:
+ *
+ * * On C11+ or C++11+ code written using this API will work. The
+ * ASSUME macros may or may not generate a hint to the compiler, but
+ * that is only an optimization issue and will not actually cause
+ * failures.
+ * * If you're using pretty much any compiler other than MSVC,
+ * everything should basically work as well as in C11/C++11.
+ */
+
+#if !defined(SIMDE_ALIGN_H)
+#define SIMDE_ALIGN_H
+
+#include "hedley.h"
+
+/* I know this seems a little silly, but some non-hosted compilers
+ * don't have stddef.h, so we try to accomodate them. */
+#if !defined(SIMDE_ALIGN_SIZE_T_)
+#if defined(__SIZE_TYPE__)
+#define SIMDE_ALIGN_SIZE_T_ __SIZE_TYPE__
+#elif defined(__SIZE_T_TYPE__)
+#define SIMDE_ALIGN_SIZE_T_ __SIZE_TYPE__
+#elif defined(__cplusplus)
+#include <cstddef>
+#define SIMDE_ALIGN_SIZE_T_ size_t
+#else
+#include <stddef.h>
+#define SIMDE_ALIGN_SIZE_T_ size_t
+#endif
+#endif
+
+#if !defined(SIMDE_ALIGN_INTPTR_T_)
+#if defined(__INTPTR_TYPE__)
+#define SIMDE_ALIGN_INTPTR_T_ __INTPTR_TYPE__
+#elif defined(__PTRDIFF_TYPE__)
+#define SIMDE_ALIGN_INTPTR_T_ __PTRDIFF_TYPE__
+#elif defined(__PTRDIFF_T_TYPE__)
+#define SIMDE_ALIGN_INTPTR_T_ __PTRDIFF_T_TYPE__
+#elif defined(__cplusplus)
+#include <cstddef>
+#define SIMDE_ALIGN_INTPTR_T_ ptrdiff_t
+#else
+#include <stddef.h>
+#define SIMDE_ALIGN_INTPTR_T_ ptrdiff_t
+#endif
+#endif
+
+#if defined(SIMDE_ALIGN_DEBUG)
+#if defined(__cplusplus)
+#include <cstdio>
+#else
+#include <stdio.h>
+#endif
+#endif
+
+/* SIMDE_ALIGN_OF(Type)
+ *
+ * The SIMDE_ALIGN_OF macro works like alignof, or _Alignof, or
+ * __alignof, or __alignof__, or __ALIGNOF__, depending on the compiler.
+ * It isn't defined everywhere (only when the compiler has some alignof-
+ * like feature we can use to implement it), but it should work in most
+ * modern compilers, as well as C11 and C++11.
+ *
+ * If we can't find an implementation for SIMDE_ALIGN_OF then the macro
+ * will not be defined, so if you can handle that situation sensibly
+ * you may need to sprinkle some ifdefs into your code.
+ */
+#if (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)) || \
+	(0 && HEDLEY_HAS_FEATURE(c_alignof))
+#define SIMDE_ALIGN_OF(Type) _Alignof(Type)
+#elif (defined(__cplusplus) && (__cplusplus >= 201103L)) || \
+	(0 && HEDLEY_HAS_FEATURE(cxx_alignof))
+#define SIMDE_ALIGN_OF(Type) alignof(Type)
+#elif HEDLEY_GCC_VERSION_CHECK(2, 95, 0) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
+	HEDLEY_SUNPRO_VERSION_CHECK(5, 13, 0) || \
+	HEDLEY_TINYC_VERSION_CHECK(0, 9, 24) || \
+	HEDLEY_PGI_VERSION_CHECK(19, 10, 0) || \
+	HEDLEY_CRAY_VERSION_CHECK(10, 0, 0) || \
+	HEDLEY_TI_ARMCL_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CL2000_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CL6X_VERSION_CHECK(8, 0, 0) || \
+	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
+	HEDLEY_TI_CL430_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 3, 2) || defined(__IBM__ALIGNOF__) || \
+	defined(__clang__)
+#define SIMDE_ALIGN_OF(Type) __alignof__(Type)
+#elif HEDLEY_IAR_VERSION_CHECK(8, 40, 0)
+#define SIMDE_ALIGN_OF(Type) __ALIGNOF__(Type)
+#elif HEDLEY_MSVC_VERSION_CHECK(19, 0, 0)
+/* Probably goes back much further, but MS takes down their old docs.
+ * If you can verify that this works in earlier versions please let
+ * me know! */
+#define SIMDE_ALIGN_OF(Type) __alignof(Type)
+#endif
+
+/* SIMDE_ALIGN_MAXIMUM:
+ *
+ * This is the maximum alignment that the compiler supports. You can
+ * define the value prior to including SIMDe if necessary, but in that
+ * case *please* submit an issue so we can add the platform to the
+ * detection code.
+ *
+ * Most compilers are okay with types which are aligned beyond what
+ * they think is the maximum, as long as the alignment is a power
+ * of two. MSVC is the exception (of course), so we need to cap the
+ * alignment requests at values that the implementation supports.
+ *
+ * XL C/C++ will accept values larger than 16 (which is the alignment
+ * of an AltiVec vector), but will not reliably align to the larger
+ * value, so so we cap the value at 16 there.
+ *
+ * If the compiler accepts any power-of-two value within reason then
+ * this macro should be left undefined, and the SIMDE_ALIGN_CAP
+ * macro will just return the value passed to it. */
+#if !defined(SIMDE_ALIGN_MAXIMUM)
+#if defined(HEDLEY_MSVC_VERSION)
+#if defined(_M_IX86) || defined(_M_AMD64)
+#if HEDLEY_MSVC_VERSION_CHECK(19, 14, 0)
+#define SIMDE_ALIGN_PLATFORM_MAXIMUM 64
+#elif HEDLEY_MSVC_VERSION_CHECK(16, 0, 0)
+/* VS 2010 is really a guess based on Wikipedia; if anyone can
+ * test with old VS versions I'd really appreciate it. */
+#define SIMDE_ALIGN_PLATFORM_MAXIMUM 32
+#else
+#define SIMDE_ALIGN_PLATFORM_MAXIMUM 16
+#endif
+#elif defined(_M_ARM) || defined(_M_ARM64)
+#define SIMDE_ALIGN_PLATFORM_MAXIMUM 8
+#endif
+#elif defined(HEDLEY_IBM_VERSION)
+#define SIMDE_ALIGN_PLATFORM_MAXIMUM 16
+#endif
+#endif
+
+/* You can mostly ignore these; they're intended for internal use.
+ * If you do need to use them please let me know; if they fulfill
+ * a common use case I'll probably drop the trailing underscore
+ * and make them part of the public API. */
+#if defined(SIMDE_ALIGN_PLATFORM_MAXIMUM)
+#if SIMDE_ALIGN_PLATFORM_MAXIMUM >= 64
+#define SIMDE_ALIGN_64_ 64
+#define SIMDE_ALIGN_32_ 32
+#define SIMDE_ALIGN_16_ 16
+#define SIMDE_ALIGN_8_ 8
+#elif SIMDE_ALIGN_PLATFORM_MAXIMUM >= 32
+#define SIMDE_ALIGN_64_ 32
+#define SIMDE_ALIGN_32_ 32
+#define SIMDE_ALIGN_16_ 16
+#define SIMDE_ALIGN_8_ 8
+#elif SIMDE_ALIGN_PLATFORM_MAXIMUM >= 16
+#define SIMDE_ALIGN_64_ 16
+#define SIMDE_ALIGN_32_ 16
+#define SIMDE_ALIGN_16_ 16
+#define SIMDE_ALIGN_8_ 8
+#elif SIMDE_ALIGN_PLATFORM_MAXIMUM >= 8
+#define SIMDE_ALIGN_64_ 8
+#define SIMDE_ALIGN_32_ 8
+#define SIMDE_ALIGN_16_ 8
+#define SIMDE_ALIGN_8_ 8
+#else
+#error Max alignment expected to be >= 8
+#endif
+#else
+#define SIMDE_ALIGN_64_ 64
+#define SIMDE_ALIGN_32_ 32
+#define SIMDE_ALIGN_16_ 16
+#define SIMDE_ALIGN_8_ 8
+#endif
+
+/**
+ * SIMDE_ALIGN_CAP(Alignment)
+ *
+ * Returns the minimum of Alignment or SIMDE_ALIGN_MAXIMUM.
+ */
+#if defined(SIMDE_ALIGN_MAXIMUM)
+#define SIMDE_ALIGN_CAP(Alignment) \
+	(((Alignment) < (SIMDE_ALIGN_PLATFORM_MAXIMUM)) \
+		 ? (Alignment) \
+		 : (SIMDE_ALIGN_PLATFORM_MAXIMUM))
+#else
+#define SIMDE_ALIGN_CAP(Alignment) (Alignment)
+#endif
+
+/* SIMDE_ALIGN_TO(Alignment)
+ *
+ * SIMDE_ALIGN_TO is used to declare types or variables. It basically
+ * maps to the align attribute in most compilers, the align declspec
+ * in MSVC, or _Alignas/alignas in C11/C++11.
+ *
+ * Example:
+ *
+ * struct i32x4 {
+ * SIMDE_ALIGN_TO(16) int32_t values[4];
+ * }
+ *
+ * Limitations:
+ *
+ * MSVC requires that the Alignment parameter be numeric; you can't do
+ * something like `SIMDE_ALIGN_TO(SIMDE_ALIGN_OF(int))`. This is
+ * unfortunate because that's really how the LIKE macros are
+ * implemented, and I am not aware of a way to get anything like this
+ * to work without using the C11/C++11 keywords.
+ *
+ * It also means that we can't use SIMDE_ALIGN_CAP to limit the
+ * alignment to the value specified, which MSVC also requires, so on
+ * MSVC you should use the `SIMDE_ALIGN_TO_8/16/32/64` macros instead.
+ * They work like `SIMDE_ALIGN_TO(SIMDE_ALIGN_CAP(Alignment))` would,
+ * but should be safe to use on MSVC.
+ *
+ * All this is to say that, if you want your code to work on MSVC, you
+ * should use the SIMDE_ALIGN_TO_8/16/32/64 macros below instead of
+ * SIMDE_ALIGN_TO(8/16/32/64).
+ */
+#if HEDLEY_HAS_ATTRIBUTE(aligned) || HEDLEY_GCC_VERSION_CHECK(2, 95, 0) || \
+	HEDLEY_CRAY_VERSION_CHECK(8, 4, 0) || \
+	HEDLEY_IBM_VERSION_CHECK(11, 1, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
+	HEDLEY_PGI_VERSION_CHECK(19, 4, 0) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_TINYC_VERSION_CHECK(0, 9, 24) || \
+	HEDLEY_TI_ARMCL_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CL2000_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CL6X_VERSION_CHECK(8, 0, 0) || \
+	HEDLEY_TI_CL7X_VERSION_CHECK(1, 2, 0) || \
+	HEDLEY_TI_CL430_VERSION_CHECK(16, 9, 0) || \
+	HEDLEY_TI_CLPRU_VERSION_CHECK(2, 3, 2)
+#define SIMDE_ALIGN_TO(Alignment) \
+	__attribute__((__aligned__(SIMDE_ALIGN_CAP(Alignment))))
+#elif (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L))
+#define SIMDE_ALIGN_TO(Alignment) _Alignas(SIMDE_ALIGN_CAP(Alignment))
+#elif (defined(__cplusplus) && (__cplusplus >= 201103L))
+#define SIMDE_ALIGN_TO(Alignment) alignas(SIMDE_ALIGN_CAP(Alignment))
+#elif defined(HEDLEY_MSVC_VERSION)
+#define SIMDE_ALIGN_TO(Alignment) __declspec(align(Alignment))
+/* Unfortunately MSVC can't handle __declspec(align(__alignof(Type)));
+ * the alignment passed to the declspec has to be an integer. */
+#define SIMDE_ALIGN_OF_UNUSABLE_FOR_LIKE
+#endif
+#define SIMDE_ALIGN_TO_64 SIMDE_ALIGN_TO(SIMDE_ALIGN_64_)
+#define SIMDE_ALIGN_TO_32 SIMDE_ALIGN_TO(SIMDE_ALIGN_32_)
+#define SIMDE_ALIGN_TO_16 SIMDE_ALIGN_TO(SIMDE_ALIGN_16_)
+#define SIMDE_ALIGN_TO_8 SIMDE_ALIGN_TO(SIMDE_ALIGN_8_)
+
+/* SIMDE_ALIGN_ASSUME_TO(Pointer, Alignment)
+ *
+ * SIMDE_ALIGN_ASSUME_TO is semantically similar to C++20's
+ * std::assume_aligned, or __builtin_assume_aligned. It tells the
+ * compiler to assume that the provided pointer is aligned to an
+ * `Alignment`-byte boundary.
+ *
+ * If you define SIMDE_ALIGN_DEBUG prior to including this header then
+ * SIMDE_ALIGN_ASSUME_TO will turn into a runtime check. We don't
+ * integrate with NDEBUG in this header, but it may be a good idea to
+ * put something like this in your code:
+ *
+ * #if !defined(NDEBUG)
+ * #define SIMDE_ALIGN_DEBUG
+ * #endif
+ * #include <.../simde-align.h>
+ */
+#if HEDLEY_HAS_BUILTIN(__builtin_assume_aligned) || \
+	HEDLEY_GCC_VERSION_CHECK(4, 7, 0)
+#define SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment) \
+	HEDLEY_REINTERPRET_CAST( \
+		__typeof__(Pointer), \
+		__builtin_assume_aligned( \
+			HEDLEY_CONST_CAST( \
+				void *, HEDLEY_REINTERPRET_CAST(const void *, \
+								Pointer)), \
+			Alignment))
+#elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+#define SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment) \
+	(__extension__({ \
+		__typeof__(v) simde_assume_aligned_t_ = (Pointer); \
+		__assume_aligned(simde_assume_aligned_t_, Alignment); \
+		simde_assume_aligned_t_; \
+	}))
+#elif defined(__cplusplus) && (__cplusplus > 201703L)
+#include <memory>
+#define SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment) \
+	std::assume_aligned<Alignment>(Pointer)
+#else
+#if defined(__cplusplus)
+template<typename T>
+HEDLEY_ALWAYS_INLINE static T *
+simde_align_assume_to_unchecked(T *ptr, const size_t alignment)
+#else
+HEDLEY_ALWAYS_INLINE static void *
+simde_align_assume_to_unchecked(void *ptr, const size_t alignment)
+#endif
+{
+	HEDLEY_ASSUME((HEDLEY_REINTERPRET_CAST(size_t, (ptr)) %
+		 SIMDE_ALIGN_CAP(alignment)) == 0);
+	return ptr;
+}
+#if defined(__cplusplus)
+#define SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment) \
+	simde_align_assume_to_unchecked((Pointer), (Alignment))
+#else
+#define SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment) \
+	simde_align_assume_to_unchecked( \
+		HEDLEY_CONST_CAST(void *, HEDLEY_REINTERPRET_CAST( \
+						 const void *, Pointer)), \
+		(Alignment))
+#endif
+#endif
+
+#if !defined(SIMDE_ALIGN_DEBUG)
+#define SIMDE_ALIGN_ASSUME_TO(Pointer, Alignment) \
+	SIMDE_ALIGN_ASSUME_TO_UNCHECKED(Pointer, Alignment)
+#else
+#include <stdio.h>
+#if defined(__cplusplus)
+template<typename T>
+static HEDLEY_ALWAYS_INLINE T *
+simde_align_assume_to_checked_uncapped(T *ptr, const size_t alignment,
+				 const char *file, int line,
+				 const char *ptrname)
+#else
+static HEDLEY_ALWAYS_INLINE void *
+simde_align_assume_to_checked_uncapped(void *ptr, const size_t alignment,
+				 const char *file, int line,
+				 const char *ptrname)
+#endif
+{
+	if (HEDLEY_UNLIKELY(
+		 (HEDLEY_REINTERPRET_CAST(SIMDE_ALIGN_INTPTR_T_, (ptr)) %
+		 HEDLEY_STATIC_CAST(SIMDE_ALIGN_INTPTR_T_,
+					SIMDE_ALIGN_CAP(alignment))) != 0)) {
+		fprintf(stderr,
+			"%s:%d: alignment check failed for `%s' (%p %% %u == %u)\n",
+			file, line, ptrname,
+			HEDLEY_REINTERPRET_CAST(const void *, ptr),
+			HEDLEY_STATIC_CAST(unsigned int,
+					 SIMDE_ALIGN_CAP(alignment)),
+			HEDLEY_STATIC_CAST(
+				unsigned int,
+				HEDLEY_REINTERPRET_CAST(SIMDE_ALIGN_INTPTR_T_,
+							(ptr)) %
+					HEDLEY_STATIC_CAST(
+						SIMDE_ALIGN_INTPTR_T_,
+						SIMDE_ALIGN_CAP(alignment))));
+	}
+
+	return ptr;
+}
+
+#if defined(__cplusplus)
+#define SIMDE_ALIGN_ASSUME_TO(Pointer, Alignment) \
+	simde_align_assume_to_checked_uncapped((Pointer), (Alignment), \
+					 __FILE__, __LINE__, #Pointer)
+#else
+#define SIMDE_ALIGN_ASSUME_TO(Pointer, Alignment) \
+	simde_align_assume_to_checked_uncapped( \
+		HEDLEY_CONST_CAST(void *, HEDLEY_REINTERPRET_CAST( \
+						 const void *, Pointer)), \
+		(Alignment), __FILE__, __LINE__, #Pointer)
+#endif
+#endif
+
+/* SIMDE_ALIGN_LIKE(Type)
+ * SIMDE_ALIGN_LIKE_#(Type)
+ *
+ * The SIMDE_ALIGN_LIKE macros are similar to the SIMDE_ALIGN_TO macros
+ * except instead of an integer they take a type; basically, it's just
+ * a more convenient way to do something like:
+ *
+ * SIMDE_ALIGN_TO(SIMDE_ALIGN_OF(Type))
+ *
+ * The versions with a numeric suffix will fall back on using a numeric
+ * value in the event we can't use SIMDE_ALIGN_OF(Type). This is
+ * mainly for MSVC, where __declspec(align()) can't handle anything
+ * other than hard-coded numeric values.
+ */
+#if defined(SIMDE_ALIGN_OF) && defined(SIMDE_ALIGN_TO) && \
+	!defined(SIMDE_ALIGN_OF_UNUSABLE_FOR_LIKE)
+#define SIMDE_ALIGN_LIKE(Type) SIMDE_ALIGN_TO(SIMDE_ALIGN_OF(Type))
+#define SIMDE_ALIGN_LIKE_64(Type) SIMDE_ALIGN_LIKE(Type)
+#define SIMDE_ALIGN_LIKE_32(Type) SIMDE_ALIGN_LIKE(Type)
+#define SIMDE_ALIGN_LIKE_16(Type) SIMDE_ALIGN_LIKE(Type)
+#define SIMDE_ALIGN_LIKE_8(Type) SIMDE_ALIGN_LIKE(Type)
+#else
+#define SIMDE_ALIGN_LIKE_64(Type) SIMDE_ALIGN_TO_64
+#define SIMDE_ALIGN_LIKE_32(Type) SIMDE_ALIGN_TO_32
+#define SIMDE_ALIGN_LIKE_16(Type) SIMDE_ALIGN_TO_16
+#define SIMDE_ALIGN_LIKE_8(Type) SIMDE_ALIGN_TO_8
+#endif
+
+/* SIMDE_ALIGN_ASSUME_LIKE(Pointer, Type)
+ *
+ * Tihs is similar to SIMDE_ALIGN_ASSUME_TO, except that it takes a
+ * type instead of a numeric value. */
+#if defined(SIMDE_ALIGN_OF) && defined(SIMDE_ALIGN_ASSUME_TO)
+#define SIMDE_ALIGN_ASSUME_LIKE(Pointer, Type) \
+	SIMDE_ALIGN_ASSUME_TO(Pointer, SIMDE_ALIGN_OF(Type))
+#endif
+
+/* SIMDE_ALIGN_CAST(Type, Pointer)
+ *
+ * SIMDE_ALIGN_CAST is like C++'s reinterpret_cast, but it will try
+ * to silence warnings that some compilers may produce if you try
+ * to assign to a type with increased alignment requirements.
+ *
+ * Note that it does *not* actually attempt to tell the compiler that
+ * the pointer is aligned like the destination should be; that's the
+ * job of the next macro. This macro is necessary for stupid APIs
+ * like _mm_loadu_si128 where the input is a __m128i* but the function
+ * is specifically for data which isn't necessarily aligned to
+ * _Alignof(__m128i).
+ */
+#if HEDLEY_HAS_WARNING("-Wcast-align") || defined(__clang__) || \
+	HEDLEY_GCC_VERSION_CHECK(3, 4, 0)
+#define SIMDE_ALIGN_CAST(Type, Pointer) \
+	(__extension__({ \
+		HEDLEY_DIAGNOSTIC_PUSH \
+		_Pragma("GCC diagnostic ignored \"-Wcast-align\"") \
+			Type simde_r_ = \
+				HEDLEY_REINTERPRET_CAST(Type, Pointer); \
+		HEDLEY_DIAGNOSTIC_POP \
+		simde_r_; \
+	}))
+#else
+#define SIMDE_ALIGN_CAST(Type, Pointer) HEDLEY_REINTERPRET_CAST(Type, Pointer)
+#endif
+
+/* SIMDE_ALIGN_ASSUME_CAST(Type, Pointer)
+ *
+ * This is sort of like a combination of a reinterpret_cast and a
+ * SIMDE_ALIGN_ASSUME_LIKE. It uses SIMDE_ALIGN_ASSUME_LIKE to tell
+ * the compiler that the pointer is aligned like the specified type
+ * and casts the pointer to the specified type while suppressing any
+ * warnings from the compiler about casting to a type with greater
+ * alignment requirements.
+ */
+#define SIMDE_ALIGN_ASSUME_CAST(Type, Pointer) \
+	SIMDE_ALIGN_ASSUME_LIKE(SIMDE_ALIGN_CAST(Type, Pointer), Type)
+
+#endif /* !defined(SIMDE_ALIGN_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/simde-arch.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-arch.h Changed

@@ -27,14 +27,14 @@
 * an undefined macro being used (e.g., GCC with -Wundef).
 *
 * This was originally created for SIMDe
- * <https://github.com/nemequ/simde> (hence the prefix), but this
+ * <https://github.com/simd-everywhere/simde> (hence the prefix), but this
 * header has no dependencies and may be used anywhere. It is
 * originally based on information from
 * <https://sourceforge.net/p/predef/wiki/Architectures/>, though it
 * has been enhanced with additional information.
 *
 * If you improve this file, or find a bug, please file the issue at
- * <https://github.com/nemequ/simde/issues>. If you copy this into
+ * <https://github.com/simd-everywhere/simde/issues>. If you copy this into
 * your project, even if you change the prefix, please keep the links
 * to SIMDe intact so others know where to report issues, submit
 * enhancements, and find the latest version. */
@@ -70,7 +70,7 @@
 /* AMD64 / x86_64
 <https://en.wikipedia.org/wiki/X86-64> */
 #if defined(__amd64__) || defined(__amd64) || defined(__x86_64__) || \
-	defined(__x86_64) || defined(_M_X66) || defined(_M_AMD64)
+	defined(__x86_64) || defined(_M_X64) || defined(_M_AMD64)
 #define SIMDE_ARCH_AMD64 1000
 #endif
 
@@ -125,6 +125,9 @@
 #define SIMDE_ARCH_ARM_NEON SIMDE_ARCH_ARM
 #endif
 #endif
+#if defined(__ARM_FEATURE_SVE)
+#define SIMDE_ARCH_ARM_SVE
+#endif
 
 /* Blackfin
 <https://en.wikipedia.org/wiki/Blackfin> */
@@ -276,6 +279,12 @@
 #define SIMDE_ARCH_X86_AVX 1
 #endif
 #endif
+#if defined(__AVX512VP2INTERSECT__)
+#define SIMDE_ARCH_X86_AVX512VP2INTERSECT 1
+#endif
+#if defined(__AVX512VBMI__)
+#define SIMDE_ARCH_X86_AVX512VBMI 1
+#endif
 #if defined(__AVX512BW__)
 #define SIMDE_ARCH_X86_AVX512BW 1
 #endif
@@ -294,6 +303,12 @@
 #if defined(__GFNI__)
 #define SIMDE_ARCH_X86_GFNI 1
 #endif
+#if defined(__PCLMUL__)
+#define SIMDE_ARCH_X86_PCLMUL 1
+#endif
+#if defined(__VPCLMULQDQ__)
+#define SIMDE_ARCH_X86_VPCLMULQDQ 1
+#endif
 #endif
 
 /* Itanium
@@ -363,6 +378,10 @@
 #define SIMDE_ARCH_MIPS_CHECK(version) (0)
 #endif
 
+#if defined(__mips_loongson_mmi)
+#define SIMDE_ARCH_MIPS_LOONGSON_MMI 1
+#endif
+
 /* Matsushita MN10300
 <https://en.wikipedia.org/wiki/MN103> */
 #if defined(__MN10300__) || defined(__mn10300__)

obs-studio-26.1.0.tar.xz/libobs/util/simde/simde-common.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-common.h Changed

@@ -30,63 +30,104 @@
 #include "hedley.h"
 
 #define SIMDE_VERSION_MAJOR 0
-#define SIMDE_VERSION_MINOR 5
-#define SIMDE_VERSION_MICRO 0
+#define SIMDE_VERSION_MINOR 7
+#define SIMDE_VERSION_MICRO 1
 #define SIMDE_VERSION \
 	HEDLEY_VERSION_ENCODE(SIMDE_VERSION_MAJOR, SIMDE_VERSION_MINOR, \
 			 SIMDE_VERSION_MICRO)
 
-#include "simde-arch.h"
-#include "simde-features.h"
-#include "simde-diagnostic.h"
-
 #include <stddef.h>
 #include <stdint.h>
 
-#if HEDLEY_HAS_ATTRIBUTE(aligned) || HEDLEY_GCC_VERSION_CHECK(2, 95, 0) || \
-	HEDLEY_CRAY_VERSION_CHECK(8, 4, 0) || \
-	HEDLEY_IBM_VERSION_CHECK(11, 1, 0) || \
-	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
-	HEDLEY_PGI_VERSION_CHECK(19, 4, 0) || \
-	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
-	HEDLEY_TINYC_VERSION_CHECK(0, 9, 24) || \
-	HEDLEY_TI_VERSION_CHECK(8, 1, 0)
-#define SIMDE_ALIGN(alignment) __attribute__((aligned(alignment)))
-#elif defined(_MSC_VER) && !(defined(_M_ARM) && !defined(_M_ARM64))
-#define SIMDE_ALIGN(alignment) __declspec(align(alignment))
-#elif defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)
-#define SIMDE_ALIGN(alignment) _Alignas(alignment)
-#elif defined(__cplusplus) && (__cplusplus >= 201103L)
-#define SIMDE_ALIGN(alignment) alignas(alignment)
-#else
-#define SIMDE_ALIGN(alignment)
-#endif
-
-#if HEDLEY_GNUC_VERSION_CHECK(2, 95, 0) || \
-	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
-	HEDLEY_IBM_VERSION_CHECK(11, 1, 0)
-#define SIMDE_ALIGN_OF(T) (__alignof__(T))
-#elif (defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)) || \
-	HEDLEY_HAS_FEATURE(c11_alignof)
-#define SIMDE_ALIGN_OF(T) (_Alignof(T))
-#elif (defined(__cplusplus) && (__cplusplus >= 201103L)) || \
-	HEDLEY_HAS_FEATURE(cxx_alignof)
-#define SIMDE_ALIGN_OF(T) (alignof(T))
-#endif
-
-#if defined(SIMDE_ALIGN_OF)
-#define SIMDE_ALIGN_AS(N, T) SIMDE_ALIGN(SIMDE_ALIGN_OF(T))
-#else
-#define SIMDE_ALIGN_AS(N, T) SIMDE_ALIGN(N)
+#include "simde-detect-clang.h"
+#include "simde-arch.h"
+#include "simde-features.h"
+#include "simde-diagnostic.h"
+#include "simde-math.h"
+#include "simde-constify.h"
+#include "simde-align.h"
+
+/* In some situations, SIMDe has to make large performance sacrifices
+ * for small increases in how faithfully it reproduces an API, but
+ * only a relatively small number of users will actually need the API
+ * to be completely accurate. The SIMDE_FAST_* options can be used to
+ * disable these trade-offs.
+ *
+ * They can be enabled by passing -DSIMDE_FAST_MATH to the compiler, or
+ * the individual defines (e.g., -DSIMDE_FAST_NANS) if you only want to
+ * enable some optimizations. Using -ffast-math and/or
+ * -ffinite-math-only will also enable the relevant options. If you
+ * don't want that you can pass -DSIMDE_NO_FAST_* to disable them. */
+
+/* Most programs avoid NaNs by never passing values which can result in
+ * a NaN; for example, if you only pass non-negative values to the sqrt
+ * functions, it won't generate a NaN. On some platforms, similar
+ * functions handle NaNs differently; for example, the _mm_min_ps SSE
+ * function will return 0.0 if you pass it (0.0, NaN), but the NEON
+ * vminq_f32 function will return NaN. Making them behave like one
+ * another is expensive; it requires generating a mask of all lanes
+ * with NaNs, then performing the operation (e.g., vminq_f32), then
+ * blending together the result with another vector using the mask.
+ *
+ * If you don't want SIMDe to worry about the differences between how
+ * NaNs are handled on the two platforms, define this (or pass
+ * -ffinite-math-only) */
+#if !defined(SIMDE_FAST_MATH) && !defined(SIMDE_NO_FAST_MATH) && \
+	defined(__FAST_MATH__)
+#define SIMDE_FAST_MATH
+#endif
+
+#if !defined(SIMDE_FAST_NANS) && !defined(SIMDE_NO_FAST_NANS)
+#if defined(SIMDE_FAST_MATH)
+#define SIMDE_FAST_NANS
+#elif defined(__FINITE_MATH_ONLY__)
+#if __FINITE_MATH_ONLY__
+#define SIMDE_FAST_NANS
+#endif
+#endif
+#endif
+
+/* Many functions are defined as using the current rounding mode
+ * (i.e., the SIMD version of fegetround()) when converting to
+ * an integer. For example, _mm_cvtpd_epi32. Unfortunately,
+ * on some platforms (such as ARMv8+ where round-to-nearest is
+ * always used, regardless of the FPSCR register) this means we
+ * have to first query the current rounding mode, then choose
+ * the proper function (rounnd
+ , ceil, floor, etc.) */
+#if !defined(SIMDE_FAST_ROUND_MODE) && !defined(SIMDE_NO_FAST_ROUND_MODE) && \
+	defined(SIMDE_FAST_MATH)
+#define SIMDE_FAST_ROUND_MODE
+#endif
+
+/* This controls how ties are rounded. For example, does 10.5 round to
+ * 10 or 11? IEEE 754 specifies round-towards-even, but ARMv7 (for
+ * example) doesn't support it and it must be emulated (which is rather
+ * slow). If you're okay with just using the default for whatever arch
+ * you're on, you should definitely define this.
+ *
+ * Note that we don't use this macro to avoid correct implementations
+ * in functions which are explicitly about rounding (such as vrnd* on
+ * NEON, _mm_round_* on x86, etc.); it is only used for code where
+ * rounding is a component in another function, and even then it isn't
+ * usually a problem since such functions will use the current rounding
+ * mode. */
+#if !defined(SIMDE_FAST_ROUND_TIES) && !defined(SIMDE_NO_FAST_ROUND_TIES) && \
+	defined(SIMDE_FAST_MATH)
+#define SIMDE_FAST_ROUND_TIES
+#endif
+
+/* For functions which convert from one type to another (mostly from
+ * floating point to integer types), sometimes we need to do a range
+ * check and potentially return a different result if the value
+ * falls outside that range. Skipping this check can provide a
+ * performance boost, at the expense of faithfulness to the API we're
+ * emulating. */
+#if !defined(SIMDE_FAST_CONVERSION_RANGE) && \
+	!defined(SIMDE_NO_FAST_CONVERSION_RANGE) && defined(SIMDE_FAST_MATH)
+#define SIMDE_FAST_CONVERSION_RANGE
 #endif
 
-#define simde_assert_aligned(alignment, val) \
-	simde_assert_int(HEDLEY_REINTERPRET_CAST( \
-				 uintptr_t, HEDLEY_REINTERPRET_CAST( \
-						 const void *, (val))) % \
-				 (alignment), \
-			 ==, 0)
-
 #if HEDLEY_HAS_BUILTIN(__builtin_constant_p) || \
 	HEDLEY_GCC_VERSION_CHECK(3, 4, 0) || \
 	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
@@ -102,15 +143,21 @@
 #define SIMDE_CHECK_CONSTANT_(expr) (std::is_constant_evaluated())
 #endif
 
-/* diagnose_if + __builtin_constant_p was broken until clang 9,
- * which is when __FILE_NAME__ was added. */
-#if defined(SIMDE_CHECK_CONSTANT_) && defined(__FILE_NAME__)
+#if !defined(SIMDE_NO_CHECK_IMMEDIATE_CONSTANT)
+#if defined(SIMDE_CHECK_CONSTANT_) && \
+	SIMDE_DETECT_CLANG_VERSION_CHECK(9, 0, 0) && \
+	(!defined(__apple_build_version__) || \
+	 ((__apple_build_version__ < 11000000) || \
+	 (__apple_build_version__ >= 12000000)))
 #define SIMDE_REQUIRE_CONSTANT(arg) \
 	HEDLEY_REQUIRE_MSG(SIMDE_CHECK_CONSTANT_(arg), \
 			 "`" #arg "' must be constant")
 #else
 #define SIMDE_REQUIRE_CONSTANT(arg)
 #endif
+#else
+#define SIMDE_REQUIRE_CONSTANT(arg)
+#endif
 
 #define SIMDE_REQUIRE_RANGE(arg, min, max) \
 	HEDLEY_REQUIRE_MSG((((arg) >= (min)) && ((arg) <= (max))), \
@@ -120,39 +167,20 @@
 	SIMDE_REQUIRE_CONSTANT(arg) \
 	SIMDE_REQUIRE_RANGE(arg, min, max)
 
-/* SIMDE_ASSUME_ALIGNED allows you to (try to) tell the compiler
- * that a pointer is aligned to an `alignment`-byte boundary. */
-#if HEDLEY_HAS_BUILTIN(__builtin_assume_aligned) || \
-	HEDLEY_GCC_VERSION_CHECK(4, 7, 0)
-#define SIMDE_ASSUME_ALIGNED(alignment, v) \
-	HEDLEY_REINTERPRET_CAST(__typeof__(v), \
-				__builtin_assume_aligned(v, alignment))
-#elif defined(__cplusplus) && (__cplusplus > 201703L)
-#define SIMDE_ASSUME_ALIGNED(alignment, v) std::assume_aligned<alignment>(v)
-#elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
-#define SIMDE_ASSUME_ALIGNED(alignment, v) \
-	(__extension__({ \
-		__typeof__(v) simde_assume_aligned_t_ = (v); \
-		__assume_aligned(simde_assume_aligned_t_, alignment); \
-		simde_assume_aligned_t_; \
-	}))
-#else
-#define SIMDE_ASSUME_ALIGNED(alignment, v) (v)
-#endif
-
-/* SIMDE_ALIGN_CAST allows you to convert to a type with greater
- * aligment requirements without triggering a warning. */
-#if HEDLEY_HAS_WARNING("-Wcast-align")
-#define SIMDE_ALIGN_CAST(T, v) \
-	(__extension__({ \
-		HEDLEY_DIAGNOSTIC_PUSH \
-		_Pragma("clang diagnostic ignored \"-Wcast-align\"") \
-			T simde_r_ = HEDLEY_REINTERPRET_CAST(T, v); \
-		HEDLEY_DIAGNOSTIC_POP \
-		simde_r_; \
-	}))
-#else
-#define SIMDE_ALIGN_CAST(T, v) HEDLEY_REINTERPRET_CAST(T, v)
+/* A copy of HEDLEY_STATIC_ASSERT, except we don't define an empty
+ * fallback if we can't find an implementation; instead we have to
+ * check if SIMDE_STATIC_ASSERT is defined before using it. */
+#if !defined(__cplusplus) && \
+	((defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)) || \
+	 HEDLEY_HAS_FEATURE(c_static_assert) || \
+	 HEDLEY_GCC_VERSION_CHECK(6, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || defined(_Static_assert))
+#define SIMDE_STATIC_ASSERT(expr, message) _Static_assert(expr, message)
+#elif (defined(__cplusplus) && (__cplusplus >= 201103L)) || \
+	HEDLEY_MSVC_VERSION_CHECK(16, 0, 0)
+#define SIMDE_STATIC_ASSERT(expr, message) \
+	HEDLEY_DIAGNOSTIC_DISABLE_CPP98_COMPAT_WRAP_( \
+		static_assert(expr, message))
 #endif
 
 #if (HEDLEY_HAS_ATTRIBUTE(may_alias) && !defined(HEDLEY_SUNPRO_VERSION)) || \
@@ -170,6 +198,7 @@
 
 * SIMDE_VECTOR - Declaring a vector.
 * SIMDE_VECTOR_OPS - basic operations (binary and unary).
+ * SIMDE_VECTOR_NEGATE - negating a vector
 * SIMDE_VECTOR_SCALAR - For binary operators, the second argument
 can be a scalar, in which case the result is as if that scalar
 had been broadcast to all lanes of a vector.
@@ -182,11 +211,13 @@
 #if HEDLEY_GCC_VERSION_CHECK(4, 8, 0)
 #define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
 #define SIMDE_VECTOR_OPS
+#define SIMDE_VECTOR_NEGATE
 #define SIMDE_VECTOR_SCALAR
 #define SIMDE_VECTOR_SUBSCRIPT
 #elif HEDLEY_INTEL_VERSION_CHECK(16, 0, 0)
 #define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
 #define SIMDE_VECTOR_OPS
+#define SIMDE_VECTOR_NEGATE
 /* ICC only supports SIMDE_VECTOR_SCALAR for constants */
 #define SIMDE_VECTOR_SUBSCRIPT
 #elif HEDLEY_GCC_VERSION_CHECK(4, 1, 0) || HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
@@ -197,8 +228,9 @@
 #elif HEDLEY_HAS_ATTRIBUTE(vector_size)
 #define SIMDE_VECTOR(size) __attribute__((__vector_size__(size)))
 #define SIMDE_VECTOR_OPS
+#define SIMDE_VECTOR_NEGATE
 #define SIMDE_VECTOR_SUBSCRIPT
-#if HEDLEY_HAS_ATTRIBUTE(diagnose_if) /* clang 4.0 */
+#if SIMDE_DETECT_CLANG_VERSION_CHECK(5, 0, 0)
 #define SIMDE_VECTOR_SCALAR
 #endif
 #endif
@@ -281,27 +313,34 @@
 #endif
 
 #if defined(SIMDE_ENABLE_OPENMP)
-#define SIMDE_VECTORIZE _Pragma("omp simd")
+#define SIMDE_VECTORIZE HEDLEY_PRAGMA(omp simd)
 #define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(omp simd safelen(l))
+#if defined(__clang__)
+#define SIMDE_VECTORIZE_REDUCTION(r) \
+	HEDLEY_DIAGNOSTIC_PUSH \
+	_Pragma("clang diagnostic ignored \"-Wsign-conversion\"") \
+		HEDLEY_PRAGMA(omp simd reduction(r)) HEDLEY_DIAGNOSTIC_POP
+#else
 #define SIMDE_VECTORIZE_REDUCTION(r) HEDLEY_PRAGMA(omp simd reduction(r))
+#endif
 #define SIMDE_VECTORIZE_ALIGNED(a) HEDLEY_PRAGMA(omp simd aligned(a))
 #elif defined(SIMDE_ENABLE_CILKPLUS)
-#define SIMDE_VECTORIZE _Pragma("simd")
+#define SIMDE_VECTORIZE HEDLEY_PRAGMA(simd)
 #define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(simd vectorlength(l))
 #define SIMDE_VECTORIZE_REDUCTION(r) HEDLEY_PRAGMA(simd reduction(r))
 #define SIMDE_VECTORIZE_ALIGNED(a) HEDLEY_PRAGMA(simd aligned(a))
 #elif defined(__clang__) && !defined(HEDLEY_IBM_VERSION)
-#define SIMDE_VECTORIZE _Pragma("clang loop vectorize(enable)")
+#define SIMDE_VECTORIZE HEDLEY_PRAGMA(clang loop vectorize(enable))
 #define SIMDE_VECTORIZE_SAFELEN(l) HEDLEY_PRAGMA(clang loop vectorize_width(l))
 #define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
 #define SIMDE_VECTORIZE_ALIGNED(a)
 #elif HEDLEY_GCC_VERSION_CHECK(4, 9, 0)
-#define SIMDE_VECTORIZE _Pragma("GCC ivdep")
+#define SIMDE_VECTORIZE HEDLEY_PRAGMA(GCC ivdep)
 #define SIMDE_VECTORIZE_SAFELEN(l) SIMDE_VECTORIZE
 #define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
 #define SIMDE_VECTORIZE_ALIGNED(a)
 #elif HEDLEY_CRAY_VERSION_CHECK(5, 0, 0)
-#define SIMDE_VECTORIZE _Pragma("_CRI ivdep")
+#define SIMDE_VECTORIZE HEDLEY_PRAGMA(_CRI ivdep)
 #define SIMDE_VECTORIZE_SAFELEN(l) SIMDE_VECTORIZE
 #define SIMDE_VECTORIZE_REDUCTION(r) SIMDE_VECTORIZE
 #define SIMDE_VECTORIZE_ALIGNED(a)
@@ -350,20 +389,10 @@
 	HEDLEY_DIAGNOSTIC_POP
 #endif
 
-#if HEDLEY_HAS_WARNING("-Wpedantic")
-#define SIMDE_DIAGNOSTIC_DISABLE_INT128 \
-	_Pragma("clang diagnostic ignored \"-Wpedantic\"")
-#elif defined(HEDLEY_GCC_VERSION)
-#define SIMDE_DIAGNOSTIC_DISABLE_INT128 \
-	_Pragma("GCC diagnostic ignored \"-Wpedantic\"")
-#else
-#define SIMDE_DIAGNOSTIC_DISABLE_INT128
-#endif
-
 #if defined(__SIZEOF_INT128__)
 #define SIMDE_HAVE_INT128_
 HEDLEY_DIAGNOSTIC_PUSH
-SIMDE_DIAGNOSTIC_DISABLE_INT128
+SIMDE_DIAGNOSTIC_DISABLE_PEDANTIC_
 typedef __int128 simde_int128;
 typedef unsigned __int128 simde_uint128;
 HEDLEY_DIAGNOSTIC_POP
@@ -488,39 +517,6 @@
 #endif
 typedef SIMDE_FLOAT64_TYPE simde_float64;
 
-/* Whether to assume that the compiler can auto-vectorize reasonably
- well. This will cause SIMDe to attempt to compose vector
- operations using more simple vector operations instead of minimize
- serial work.
-
- As an example, consider the _mm_add_ss(a, b) function from SSE,
- which returns { a0 + b0, a1, a2, a3 }. This pattern is repeated
- for other operations (sub, mul, etc.).
-
- The naïve implementation would result in loading a0 and b0, adding
- them into a temporary variable, then splicing that value into a new
- vector with the remaining elements from a.
-
- On platforms which support vectorization, it's generally faster to
- simply perform the operation on the entire vector to avoid having
- to move data between SIMD registers and non-SIMD registers.
- Basically, instead of the temporary variable being (a0 + b0) it
- would be a vector of (a + b), which is then combined with a to form
- the result.
-
- By default, SIMDe will prefer the pure-vector versions if we detect
- a vector ISA extension, but this can be overridden by defining
- SIMDE_NO_ASSUME_VECTORIZATION. You can also define
- SIMDE_ASSUME_VECTORIZATION if you want to force SIMDe to use the
- vectorized version. */
-#if !defined(SIMDE_NO_ASSUME_VECTORIZATION) && \
-	!defined(SIMDE_ASSUME_VECTORIZATION)
-#if defined(__SSE__) || defined(__ARM_NEON) || defined(__mips_msa) || \
-	defined(__ALTIVEC__) || defined(__wasm_simd128__)
-#define SIMDE_ASSUME_VECTORIZATION
-#endif
-#endif
-
 #if HEDLEY_HAS_WARNING("-Wbad-function-cast")
 #define SIMDE_CONVERT_FTOI(T, v) \
 	HEDLEY_DIAGNOSTIC_PUSH \
@@ -530,11 +526,18 @@
 #define SIMDE_CONVERT_FTOI(T, v) ((T)(v))
 #endif
 
+/* TODO: detect compilers which support this outside of C11 mode */
 #if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)
 #define SIMDE_CHECKED_REINTERPRET_CAST(to, from, value) \
-	(_Generic((value), to : (value), from : ((to)(value))))
+	_Generic((value), to \
+		 : (value), default \
+		 : (_Generic((value), from \
+			 : ((to)(value)))))
 #define SIMDE_CHECKED_STATIC_CAST(to, from, value) \
-	(_Generic((value), to : (value), from : ((to)(value))))
+	_Generic((value), to \
+		 : (value), default \
+		 : (_Generic((value), from \
+			 : ((to)(value)))))
 #else
 #define SIMDE_CHECKED_REINTERPRET_CAST(to, from, value) \
 	HEDLEY_REINTERPRET_CAST(to, value)
@@ -564,7 +567,7 @@
 #if defined(__STDC_HOSTED__)
 #define SIMDE_STDC_HOSTED __STDC_HOSTED__
 #else
-#if defined(HEDLEY_PGI_VERSION_CHECK) || defined(HEDLEY_MSVC_VERSION_CHECK)
+#if defined(HEDLEY_PGI_VERSION) || defined(HEDLEY_MSVC_VERSION)
 #define SIMDE_STDC_HOSTED 1
 #else
 #define SIMDE_STDC_HOSTED 0
@@ -572,23 +575,34 @@
 #endif
 
 /* Try to deal with environments without a standard library. */
-#if !defined(simde_memcpy) || !defined(simde_memset)
-#if !defined(SIMDE_NO_STRING_H) && defined(__has_include)
-#if __has_include(<string.h>)
-#include <string.h>
 #if !defined(simde_memcpy)
-#define simde_memcpy(dest, src, n) memcpy(dest, src, n)
+#if HEDLEY_HAS_BUILTIN(__builtin_memcpy)
+#define simde_memcpy(dest, src, n) __builtin_memcpy(dest, src, n)
+#endif
 #endif
 #if !defined(simde_memset)
-#define simde_memset(s, c, n) memset(s, c, n)
+#if HEDLEY_HAS_BUILTIN(__builtin_memset)
+#define simde_memset(s, c, n) __builtin_memset(s, c, n)
 #endif
-#else
+#endif
+#if !defined(simde_memcmp)
+#if HEDLEY_HAS_BUILTIN(__builtin_memcmp)
+#define simde_memcmp(s1, s2, n) __builtin_memcmp(s1, s2, n)
+#endif
+#endif
+
+#if !defined(simde_memcpy) || !defined(simde_memset) || !defined(simde_memcmp)
+#if !defined(SIMDE_NO_STRING_H)
+#if defined(__has_include)
+#if !__has_include(<string.h>)
 #define SIMDE_NO_STRING_H
 #endif
+#elif (SIMDE_STDC_HOSTED == 0)
+#define SIMDE_NO_STRING_H
 #endif
 #endif
-#if !defined(simde_memcpy) || !defined(simde_memset)
-#if !defined(SIMDE_NO_STRING_H) && (SIMDE_STDC_HOSTED == 1)
+
+#if !defined(SIMDE_NO_STRING_H)
 #include <string.h>
 #if !defined(simde_memcpy)
 #define simde_memcpy(dest, src, n) memcpy(dest, src, n)
@@ -596,14 +610,8 @@
 #if !defined(simde_memset)
 #define simde_memset(s, c, n) memset(s, c, n)
 #endif
-#elif (HEDLEY_HAS_BUILTIN(__builtin_memcpy) && \
- HEDLEY_HAS_BUILTIN(__builtin_memset)) || \
-	HEDLEY_GCC_VERSION_CHECK(4, 2, 0)
-#if !defined(simde_memcpy)
-#define simde_memcpy(dest, src, n) __builtin_memcpy(dest, src, n)
-#endif
-#if !defined(simde_memset)
-#define simde_memset(s, c, n) __builtin_memset(s, c, n)
+#if !defined(simde_memcmp)
+#define simde_memcmp(s1, s2, n) memcmp(s1, s2, n)
 #endif
 #else
 /* These are meant to be portable, not fast. If you're hitting them you
@@ -637,10 +645,24 @@
 }
 #define simde_memset(s, c, n) simde_memset_(s, c, n)
 #endif
-#endif /* !defined(SIMDE_NO_STRING_H) && (SIMDE_STDC_HOSTED == 1) */
-#endif /* !defined(simde_memcpy) || !defined(simde_memset) */
 
-#include "simde-math.h"
+#if !defined(simde_memcmp)
+SIMDE_FUCTION_ATTRIBUTES
+int simde_memcmp_(const void *s1, const void *s2, size_t n)
+{
+	unsigned char *s1_ = HEDLEY_STATIC_CAST(unsigned char *, s1);
+	unsigned char *s2_ = HEDLEY_STATIC_CAST(unsigned char *, s2);
+	for (size_t i = 0; i < len; i++) {
+		if (s1_[i] != s2_[i]) {
+			return (int)(s1_[i] - s2_[i]);
+		}
+	}
+	return 0;
+}
+#define simde_memcmp(s1, s2, n) simde_memcmp_(s1, s2, n)
+#endif
+#endif
+#endif
 
 #if defined(FE_ALL_EXCEPT)
 #define SIMDE_HAVE_FENV_H
@@ -682,6 +704,105 @@
 
 #include "check.h"
 
+/* GCC/clang have a bunch of functionality in builtins which we would
+ * like to access, but the suffixes indicate whether the operate on
+ * int, long, or long long, not fixed width types (e.g., int32_t).
+ * we use these macros to attempt to map from fixed-width to the
+ * names GCC uses. Note that you should still cast the input(s) and
+ * return values (to/from SIMDE_BUILTIN_TYPE_*_) since often even if
+ * types are the same size they may not be compatible according to the
+ * compiler. For example, on x86 long and long lonsg are generally
+ * both 64 bits, but platforms vary on whether an int64_t is mapped
+ * to a long or long long. */
+
+#include <limits.h>
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_
+
+#if (INT8_MAX == INT_MAX) && (INT8_MIN == INT_MIN)
+#define SIMDE_BUILTIN_SUFFIX_8_
+#define SIMDE_BUILTIN_TYPE_8_ int
+#elif (INT8_MAX == LONG_MAX) && (INT8_MIN == LONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_8_ l
+#define SIMDE_BUILTIN_TYPE_8_ long
+#elif (INT8_MAX == LLONG_MAX) && (INT8_MIN == LLONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_8_ ll
+#define SIMDE_BUILTIN_TYPE_8_ long long
+#endif
+
+#if (INT16_MAX == INT_MAX) && (INT16_MIN == INT_MIN)
+#define SIMDE_BUILTIN_SUFFIX_16_
+#define SIMDE_BUILTIN_TYPE_16_ int
+#elif (INT16_MAX == LONG_MAX) && (INT16_MIN == LONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_16_ l
+#define SIMDE_BUILTIN_TYPE_16_ long
+#elif (INT16_MAX == LLONG_MAX) && (INT16_MIN == LLONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_16_ ll
+#define SIMDE_BUILTIN_TYPE_16_ long long
+#endif
+
+#if (INT32_MAX == INT_MAX) && (INT32_MIN == INT_MIN)
+#define SIMDE_BUILTIN_SUFFIX_32_
+#define SIMDE_BUILTIN_TYPE_32_ int
+#elif (INT32_MAX == LONG_MAX) && (INT32_MIN == LONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_32_ l
+#define SIMDE_BUILTIN_TYPE_32_ long
+#elif (INT32_MAX == LLONG_MAX) && (INT32_MIN == LLONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_32_ ll
+#define SIMDE_BUILTIN_TYPE_32_ long long
+#endif
+
+#if (INT64_MAX == INT_MAX) && (INT64_MIN == INT_MIN)
+#define SIMDE_BUILTIN_SUFFIX_64_
+#define SIMDE_BUILTIN_TYPE_64_ int
+#elif (INT64_MAX == LONG_MAX) && (INT64_MIN == LONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_64_ l
+#define SIMDE_BUILTIN_TYPE_64_ long
+#elif (INT64_MAX == LLONG_MAX) && (INT64_MIN == LLONG_MIN)
+#define SIMDE_BUILTIN_SUFFIX_64_ ll
+#define SIMDE_BUILTIN_TYPE_64_ long long
+#endif
+
+#if defined(SIMDE_BUILTIN_SUFFIX_8_)
+#define SIMDE_BUILTIN_8_(name) \
+	HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_8_)
+#define SIMDE_BUILTIN_HAS_8_(name) \
+	HEDLEY_HAS_BUILTIN( \
+		HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_8_))
+#else
+#define SIMDE_BUILTIN_HAS_8_(name) 0
+#endif
+#if defined(SIMDE_BUILTIN_SUFFIX_16_)
+#define SIMDE_BUILTIN_16_(name) \
+	HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_16_)
+#define SIMDE_BUILTIN_HAS_16_(name) \
+	HEDLEY_HAS_BUILTIN( \
+		HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_16_))
+#else
+#define SIMDE_BUILTIN_HAS_16_(name) 0
+#endif
+#if defined(SIMDE_BUILTIN_SUFFIX_32_)
+#define SIMDE_BUILTIN_32_(name) \
+	HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_32_)
+#define SIMDE_BUILTIN_HAS_32_(name) \
+	HEDLEY_HAS_BUILTIN( \
+		HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_32_))
+#else
+#define SIMDE_BUILTIN_HAS_32_(name) 0
+#endif
+#if defined(SIMDE_BUILTIN_SUFFIX_64_)
+#define SIMDE_BUILTIN_64_(name) \
+	HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_64_)
+#define SIMDE_BUILTIN_HAS_64_(name) \
+	HEDLEY_HAS_BUILTIN( \
+		HEDLEY_CONCAT3(__builtin_, name, SIMDE_BUILTIN_SUFFIX_64_))
+#else
+#define SIMDE_BUILTIN_HAS_64_(name) 0
+#endif
+
+HEDLEY_DIAGNOSTIC_POP
+
 /* Sometimes we run into problems with specific versions of compilers
 which make the native versions unusable for us. Often this is due
 to missing functions, sometimes buggy implementations, etc. These
@@ -712,29 +833,75 @@
 #if defined(SIMDE_ARCH_X86) && !defined(SIMDE_ARCH_AMD64)
 #define SIMDE_BUG_GCC_94482
 #endif
+#if (defined(SIMDE_ARCH_X86) && !defined(SIMDE_ARCH_AMD64)) || \
+	defined(SIMDE_ARCH_SYSTEMZ)
+#define SIMDE_BUG_GCC_53784
+#endif
+#if defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
+#if HEDLEY_GCC_VERSION_CHECK(4, 3, 0) /* -Wsign-conversion */
+#define SIMDE_BUG_GCC_95144
+#endif
+#endif
 #if !HEDLEY_GCC_VERSION_CHECK(9, 4, 0) && defined(SIMDE_ARCH_AARCH64)
 #define SIMDE_BUG_GCC_94488
 #endif
-#if defined(SIMDE_ARCH_POWER)
+#if defined(SIMDE_ARCH_ARM)
+#define SIMDE_BUG_GCC_95399
+#define SIMDE_BUG_GCC_95471
+#elif defined(SIMDE_ARCH_POWER)
 #define SIMDE_BUG_GCC_95227
+#define SIMDE_BUG_GCC_95782
+#elif defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
+#if !HEDLEY_GCC_VERSION_CHECK(10, 2, 0) && !defined(__OPTIMIZE__)
+#define SIMDE_BUG_GCC_96174
+#endif
 #endif
 #define SIMDE_BUG_GCC_95399
 #elif defined(__clang__)
 #if defined(SIMDE_ARCH_AARCH64)
 #define SIMDE_BUG_CLANG_45541
+#define SIMDE_BUG_CLANG_46844
+#define SIMDE_BUG_CLANG_48257
+#if SIMDE_DETECT_CLANG_VERSION_CHECK(10, 0, 0) && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(11, 0, 0)
+#define SIMDE_BUG_CLANG_BAD_VI64_OPS
+#endif
+#endif
+#if defined(SIMDE_ARCH_POWER)
+#define SIMDE_BUG_CLANG_46770
+#endif
+#if defined(_ARCH_PWR9) && !SIMDE_DETECT_CLANG_VERSION_CHECK(12, 0, 0) && \
+	!defined(__OPTIMIZE__)
+#define SIMDE_BUG_CLANG_POWER9_16x4_BAD_SHIFT
+#endif
+#if defined(SIMDE_ARCH_X86) || defined(SIMDE_ARCH_AMD64)
+#if HEDLEY_HAS_WARNING("-Wsign-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(11, 0, 0)
+#define SIMDE_BUG_CLANG_45931
+#endif
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(11, 0, 0)
+#define SIMDE_BUG_CLANG_44589
+#endif
 #endif
+#define SIMDE_BUG_CLANG_45959
+#elif defined(HEDLEY_MSVC_VERSION)
+#if defined(SIMDE_ARCH_X86)
+#define SIMDE_BUG_MSVC_ROUND_EXTRACT
 #endif
-#if defined(HEDLEY_EMSCRIPTEN_VERSION)
-#define SIMDE_BUG_EMSCRIPTEN_MISSING_IMPL /* Placeholder for (as yet) unfiled issues. */
-#define SIMDE_BUG_EMSCRIPTEN_5242
+#elif defined(HEDLEY_INTEL_VERSION)
+#define SIMDE_BUG_INTEL_857088
 #endif
 #endif
 
 /* GCC and Clang both have the same issue:
 * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95144
 * https://bugs.llvm.org/show_bug.cgi?id=45931
+ * This is just an easy way to work around it.
 */
-#if HEDLEY_HAS_WARNING("-Wsign-conversion") || HEDLEY_GCC_VERSION_CHECK(4, 3, 0)
+#if (HEDLEY_HAS_WARNING("-Wsign-conversion") && \
+ SIMDE_DETECT_CLANG_VERSION_NOT(11, 0, 0)) || \
+	HEDLEY_GCC_VERSION_CHECK(4, 3, 0)
 #define SIMDE_BUG_IGNORE_SIGN_CONVERSION(expr) \
 	(__extension__({ \
 		HEDLEY_DIAGNOSTIC_PUSH \

obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-constify.h Added

@@ -0,0 +1,925 @@
+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ * 2020 Evan Nemerson <evan@nemerson.com>
+ */
+
+/* Constify macros. For internal use only.
+ *
+ * These are used to make it possible to call a function which takes
+ * an Integer Constant Expression (ICE) using a compile time constant.
+ * Technically it would also be possible to use a value not trivially
+ * known by the compiler, but there would be a siginficant performance
+ * hit (a switch switch is used).
+ *
+ * The basic idea is pretty simple; we just emit a do while loop which
+ * contains a switch with a case for every possible value of the
+ * constant.
+ *
+ * As long as the value you pass to the function in constant, pretty
+ * much any copmiler shouldn't have a problem generating exactly the
+ * same code as if you had used an ICE.
+ *
+ * This is intended to be used in the SIMDe implementations of
+ * functions the compilers require to be an ICE, but the other benefit
+ * is that if we also disable the warnings from
+ * SIMDE_REQUIRE_CONSTANT_RANGE we can actually just allow the tests
+ * to use non-ICE parameters
+ */
+
+#if !defined(SIMDE_CONSTIFY_H)
+#define SIMDE_CONSTIFY_H
+
+#include "simde-diagnostic.h"
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_VARIADIC_MACROS_
+SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_
+
+#define SIMDE_CONSTIFY_2_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_4_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			result = func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			result = func_name(__VA_ARGS__, 3); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_8_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			result = func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			result = func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			result = func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			result = func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			result = func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			result = func_name(__VA_ARGS__, 7); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_16_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			result = func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			result = func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			result = func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			result = func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			result = func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			result = func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			result = func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			result = func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			result = func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			result = func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			result = func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			result = func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			result = func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			result = func_name(__VA_ARGS__, 15); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_32_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			result = func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			result = func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			result = func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			result = func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			result = func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			result = func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			result = func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			result = func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			result = func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			result = func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			result = func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			result = func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			result = func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			result = func_name(__VA_ARGS__, 15); \
+			break; \
+		case 16: \
+			result = func_name(__VA_ARGS__, 16); \
+			break; \
+		case 17: \
+			result = func_name(__VA_ARGS__, 17); \
+			break; \
+		case 18: \
+			result = func_name(__VA_ARGS__, 18); \
+			break; \
+		case 19: \
+			result = func_name(__VA_ARGS__, 19); \
+			break; \
+		case 20: \
+			result = func_name(__VA_ARGS__, 20); \
+			break; \
+		case 21: \
+			result = func_name(__VA_ARGS__, 21); \
+			break; \
+		case 22: \
+			result = func_name(__VA_ARGS__, 22); \
+			break; \
+		case 23: \
+			result = func_name(__VA_ARGS__, 23); \
+			break; \
+		case 24: \
+			result = func_name(__VA_ARGS__, 24); \
+			break; \
+		case 25: \
+			result = func_name(__VA_ARGS__, 25); \
+			break; \
+		case 26: \
+			result = func_name(__VA_ARGS__, 26); \
+			break; \
+		case 27: \
+			result = func_name(__VA_ARGS__, 27); \
+			break; \
+		case 28: \
+			result = func_name(__VA_ARGS__, 28); \
+			break; \
+		case 29: \
+			result = func_name(__VA_ARGS__, 29); \
+			break; \
+		case 30: \
+			result = func_name(__VA_ARGS__, 30); \
+			break; \
+		case 31: \
+			result = func_name(__VA_ARGS__, 31); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_64_(func_name, result, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			result = func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			result = func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			result = func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			result = func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			result = func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			result = func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			result = func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			result = func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			result = func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			result = func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			result = func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			result = func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			result = func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			result = func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			result = func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			result = func_name(__VA_ARGS__, 15); \
+			break; \
+		case 16: \
+			result = func_name(__VA_ARGS__, 16); \
+			break; \
+		case 17: \
+			result = func_name(__VA_ARGS__, 17); \
+			break; \
+		case 18: \
+			result = func_name(__VA_ARGS__, 18); \
+			break; \
+		case 19: \
+			result = func_name(__VA_ARGS__, 19); \
+			break; \
+		case 20: \
+			result = func_name(__VA_ARGS__, 20); \
+			break; \
+		case 21: \
+			result = func_name(__VA_ARGS__, 21); \
+			break; \
+		case 22: \
+			result = func_name(__VA_ARGS__, 22); \
+			break; \
+		case 23: \
+			result = func_name(__VA_ARGS__, 23); \
+			break; \
+		case 24: \
+			result = func_name(__VA_ARGS__, 24); \
+			break; \
+		case 25: \
+			result = func_name(__VA_ARGS__, 25); \
+			break; \
+		case 26: \
+			result = func_name(__VA_ARGS__, 26); \
+			break; \
+		case 27: \
+			result = func_name(__VA_ARGS__, 27); \
+			break; \
+		case 28: \
+			result = func_name(__VA_ARGS__, 28); \
+			break; \
+		case 29: \
+			result = func_name(__VA_ARGS__, 29); \
+			break; \
+		case 30: \
+			result = func_name(__VA_ARGS__, 30); \
+			break; \
+		case 31: \
+			result = func_name(__VA_ARGS__, 31); \
+			break; \
+		case 32: \
+			result = func_name(__VA_ARGS__, 32); \
+			break; \
+		case 33: \
+			result = func_name(__VA_ARGS__, 33); \
+			break; \
+		case 34: \
+			result = func_name(__VA_ARGS__, 34); \
+			break; \
+		case 35: \
+			result = func_name(__VA_ARGS__, 35); \
+			break; \
+		case 36: \
+			result = func_name(__VA_ARGS__, 36); \
+			break; \
+		case 37: \
+			result = func_name(__VA_ARGS__, 37); \
+			break; \
+		case 38: \
+			result = func_name(__VA_ARGS__, 38); \
+			break; \
+		case 39: \
+			result = func_name(__VA_ARGS__, 39); \
+			break; \
+		case 40: \
+			result = func_name(__VA_ARGS__, 40); \
+			break; \
+		case 41: \
+			result = func_name(__VA_ARGS__, 41); \
+			break; \
+		case 42: \
+			result = func_name(__VA_ARGS__, 42); \
+			break; \
+		case 43: \
+			result = func_name(__VA_ARGS__, 43); \
+			break; \
+		case 44: \
+			result = func_name(__VA_ARGS__, 44); \
+			break; \
+		case 45: \
+			result = func_name(__VA_ARGS__, 45); \
+			break; \
+		case 46: \
+			result = func_name(__VA_ARGS__, 46); \
+			break; \
+		case 47: \
+			result = func_name(__VA_ARGS__, 47); \
+			break; \
+		case 48: \
+			result = func_name(__VA_ARGS__, 48); \
+			break; \
+		case 49: \
+			result = func_name(__VA_ARGS__, 49); \
+			break; \
+		case 50: \
+			result = func_name(__VA_ARGS__, 50); \
+			break; \
+		case 51: \
+			result = func_name(__VA_ARGS__, 51); \
+			break; \
+		case 52: \
+			result = func_name(__VA_ARGS__, 52); \
+			break; \
+		case 53: \
+			result = func_name(__VA_ARGS__, 53); \
+			break; \
+		case 54: \
+			result = func_name(__VA_ARGS__, 54); \
+			break; \
+		case 55: \
+			result = func_name(__VA_ARGS__, 55); \
+			break; \
+		case 56: \
+			result = func_name(__VA_ARGS__, 56); \
+			break; \
+		case 57: \
+			result = func_name(__VA_ARGS__, 57); \
+			break; \
+		case 58: \
+			result = func_name(__VA_ARGS__, 58); \
+			break; \
+		case 59: \
+			result = func_name(__VA_ARGS__, 59); \
+			break; \
+		case 60: \
+			result = func_name(__VA_ARGS__, 60); \
+			break; \
+		case 61: \
+			result = func_name(__VA_ARGS__, 61); \
+			break; \
+		case 62: \
+			result = func_name(__VA_ARGS__, 62); \
+			break; \
+		case 63: \
+			result = func_name(__VA_ARGS__, 63); \
+			break; \
+		default: \
+			result = default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_2_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_4_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			func_name(__VA_ARGS__, 3); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_8_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			func_name(__VA_ARGS__, 7); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_16_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			func_name(__VA_ARGS__, 15); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_32_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			func_name(__VA_ARGS__, 15); \
+			break; \
+		case 16: \
+			func_name(__VA_ARGS__, 16); \
+			break; \
+		case 17: \
+			func_name(__VA_ARGS__, 17); \
+			break; \
+		case 18: \
+			func_name(__VA_ARGS__, 18); \
+			break; \
+		case 19: \
+			func_name(__VA_ARGS__, 19); \
+			break; \
+		case 20: \
+			func_name(__VA_ARGS__, 20); \
+			break; \
+		case 21: \
+			func_name(__VA_ARGS__, 21); \
+			break; \
+		case 22: \
+			func_name(__VA_ARGS__, 22); \
+			break; \
+		case 23: \
+			func_name(__VA_ARGS__, 23); \
+			break; \
+		case 24: \
+			func_name(__VA_ARGS__, 24); \
+			break; \
+		case 25: \
+			func_name(__VA_ARGS__, 25); \
+			break; \
+		case 26: \
+			func_name(__VA_ARGS__, 26); \
+			break; \
+		case 27: \
+			func_name(__VA_ARGS__, 27); \
+			break; \
+		case 28: \
+			func_name(__VA_ARGS__, 28); \
+			break; \
+		case 29: \
+			func_name(__VA_ARGS__, 29); \
+			break; \
+		case 30: \
+			func_name(__VA_ARGS__, 30); \
+			break; \
+		case 31: \
+			func_name(__VA_ARGS__, 31); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+#define SIMDE_CONSTIFY_64_NO_RESULT_(func_name, default_case, imm, ...) \
+	do { \
+		switch (imm) { \
+		case 0: \
+			func_name(__VA_ARGS__, 0); \
+			break; \
+		case 1: \
+			func_name(__VA_ARGS__, 1); \
+			break; \
+		case 2: \
+			func_name(__VA_ARGS__, 2); \
+			break; \
+		case 3: \
+			func_name(__VA_ARGS__, 3); \
+			break; \
+		case 4: \
+			func_name(__VA_ARGS__, 4); \
+			break; \
+		case 5: \
+			func_name(__VA_ARGS__, 5); \
+			break; \
+		case 6: \
+			func_name(__VA_ARGS__, 6); \
+			break; \
+		case 7: \
+			func_name(__VA_ARGS__, 7); \
+			break; \
+		case 8: \
+			func_name(__VA_ARGS__, 8); \
+			break; \
+		case 9: \
+			func_name(__VA_ARGS__, 9); \
+			break; \
+		case 10: \
+			func_name(__VA_ARGS__, 10); \
+			break; \
+		case 11: \
+			func_name(__VA_ARGS__, 11); \
+			break; \
+		case 12: \
+			func_name(__VA_ARGS__, 12); \
+			break; \
+		case 13: \
+			func_name(__VA_ARGS__, 13); \
+			break; \
+		case 14: \
+			func_name(__VA_ARGS__, 14); \
+			break; \
+		case 15: \
+			func_name(__VA_ARGS__, 15); \
+			break; \
+		case 16: \
+			func_name(__VA_ARGS__, 16); \
+			break; \
+		case 17: \
+			func_name(__VA_ARGS__, 17); \
+			break; \
+		case 18: \
+			func_name(__VA_ARGS__, 18); \
+			break; \
+		case 19: \
+			func_name(__VA_ARGS__, 19); \
+			break; \
+		case 20: \
+			func_name(__VA_ARGS__, 20); \
+			break; \
+		case 21: \
+			func_name(__VA_ARGS__, 21); \
+			break; \
+		case 22: \
+			func_name(__VA_ARGS__, 22); \
+			break; \
+		case 23: \
+			func_name(__VA_ARGS__, 23); \
+			break; \
+		case 24: \
+			func_name(__VA_ARGS__, 24); \
+			break; \
+		case 25: \
+			func_name(__VA_ARGS__, 25); \
+			break; \
+		case 26: \
+			func_name(__VA_ARGS__, 26); \
+			break; \
+		case 27: \
+			func_name(__VA_ARGS__, 27); \
+			break; \
+		case 28: \
+			func_name(__VA_ARGS__, 28); \
+			break; \
+		case 29: \
+			func_name(__VA_ARGS__, 29); \
+			break; \
+		case 30: \
+			func_name(__VA_ARGS__, 30); \
+			break; \
+		case 31: \
+			func_name(__VA_ARGS__, 31); \
+			break; \
+		case 32: \
+			func_name(__VA_ARGS__, 32); \
+			break; \
+		case 33: \
+			func_name(__VA_ARGS__, 33); \
+			break; \
+		case 34: \
+			func_name(__VA_ARGS__, 34); \
+			break; \
+		case 35: \
+			func_name(__VA_ARGS__, 35); \
+			break; \
+		case 36: \
+			func_name(__VA_ARGS__, 36); \
+			break; \
+		case 37: \
+			func_name(__VA_ARGS__, 37); \
+			break; \
+		case 38: \
+			func_name(__VA_ARGS__, 38); \
+			break; \
+		case 39: \
+			func_name(__VA_ARGS__, 39); \
+			break; \
+		case 40: \
+			func_name(__VA_ARGS__, 40); \
+			break; \
+		case 41: \
+			func_name(__VA_ARGS__, 41); \
+			break; \
+		case 42: \
+			func_name(__VA_ARGS__, 42); \
+			break; \
+		case 43: \
+			func_name(__VA_ARGS__, 43); \
+			break; \
+		case 44: \
+			func_name(__VA_ARGS__, 44); \
+			break; \
+		case 45: \
+			func_name(__VA_ARGS__, 45); \
+			break; \
+		case 46: \
+			func_name(__VA_ARGS__, 46); \
+			break; \
+		case 47: \
+			func_name(__VA_ARGS__, 47); \
+			break; \
+		case 48: \
+			func_name(__VA_ARGS__, 48); \
+			break; \
+		case 49: \
+			func_name(__VA_ARGS__, 49); \
+			break; \
+		case 50: \
+			func_name(__VA_ARGS__, 50); \
+			break; \
+		case 51: \
+			func_name(__VA_ARGS__, 51); \
+			break; \
+		case 52: \
+			func_name(__VA_ARGS__, 52); \
+			break; \
+		case 53: \
+			func_name(__VA_ARGS__, 53); \
+			break; \
+		case 54: \
+			func_name(__VA_ARGS__, 54); \
+			break; \
+		case 55: \
+			func_name(__VA_ARGS__, 55); \
+			break; \
+		case 56: \
+			func_name(__VA_ARGS__, 56); \
+			break; \
+		case 57: \
+			func_name(__VA_ARGS__, 57); \
+			break; \
+		case 58: \
+			func_name(__VA_ARGS__, 58); \
+			break; \
+		case 59: \
+			func_name(__VA_ARGS__, 59); \
+			break; \
+		case 60: \
+			func_name(__VA_ARGS__, 60); \
+			break; \
+		case 61: \
+			func_name(__VA_ARGS__, 61); \
+			break; \
+		case 62: \
+			func_name(__VA_ARGS__, 62); \
+			break; \
+		case 63: \
+			func_name(__VA_ARGS__, 63); \
+			break; \
+		default: \
+			default_case; \
+			break; \
+		} \
+	} while (0)
+
+HEDLEY_DIAGNOSTIC_POP
+
+#endif

obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-detect-clang.h Added

@@ -0,0 +1,114 @@
+/* Detect Clang Version
+ * Created by Evan Nemerson <evan@nemerson.com>
+ *
+ * To the extent possible under law, the author(s) have dedicated all
+ * copyright and related and neighboring rights to this software to
+ * the public domain worldwide. This software is distributed without
+ * any warranty.
+ *
+ * For details, see <http://creativecommons.org/publicdomain/zero/1.0/>.
+ * SPDX-License-Identifier: CC0-1.0
+ */
+
+/* This file was originally part of SIMDe
+ * (<https://github.com/simd-everywhere/simde>). You're free to do with it as
+ * you please, but I do have a few small requests:
+ *
+ * * If you make improvements, please submit them back to SIMDe
+ * (at <https://github.com/simd-everywhere/simde/issues>) so others can
+ * benefit from them.
+ * * Please keep a link to SIMDe intact so people know where to submit
+ * improvements.
+ * * If you expose it publicly, please change the SIMDE_ prefix to
+ * something specific to your project.
+ *
+ * The version numbers clang exposes (in the ___clang_major__,
+ * __clang_minor__, and __clang_patchlevel__ macros) are unreliable.
+ * Vendors such as Apple will define these values to their version
+ * numbers; for example, "Apple Clang 4.0" is really clang 3.1, but
+ * __clang_major__ and __clang_minor__ are defined to 4 and 0
+ * respectively, instead of 3 and 1.
+ *
+ * The solution is *usually* to use clang's feature detection macros
+ * (<https://clang.llvm.org/docs/LanguageExtensions.html#feature-checking-macros>)
+ * to determine if the feature you're interested in is available. This
+ * generally works well, and it should probably be the first thing you
+ * try. Unfortunately, it's not possible to check for everything. In
+ * particular, compiler bugs.
+ *
+ * This file just uses the feature checking macros to detect features
+ * added in specific versions of clang to identify which version of
+ * clang the compiler is based on.
+ *
+ * Right now it only goes back to 3.6, but I'm happy to accept patches
+ * to go back further. And, of course, newer versions are welcome if
+ * they're not already present, and if you find a way to detect a point
+ * release that would be great, too!
+ */
+
+#if !defined(SIMDE_DETECT_CLANG_H)
+#define SIMDE_DETECT_CLANG_H 1
+
+/* Attempt to detect the upstream clang version number. I usually only
+ * worry about major version numbers (at least for 4.0+), but if you
+ * need more resolution I'm happy to accept patches that are able to
+ * detect minor versions as well. That said, you'll probably have a
+ * hard time with detection since AFAIK most minor releases don't add
+ * anything we can detect. */
+
+#if defined(__clang__) && !defined(SIMDE_DETECT_CLANG_VERSION)
+#if __has_warning("-Wformat-insufficient-args")
+#define SIMDE_DETECT_CLANG_VERSION 120000
+#elif __has_warning("-Wimplicit-const-int-float-conversion")
+#define SIMDE_DETECT_CLANG_VERSION 110000
+#elif __has_warning("-Wmisleading-indentation")
+#define SIMDE_DETECT_CLANG_VERSION 100000
+#elif defined(__FILE_NAME__)
+#define SIMDE_DETECT_CLANG_VERSION 90000
+#elif __has_warning("-Wextra-semi-stmt") || \
+	__has_builtin(__builtin_rotateleft32)
+#define SIMDE_DETECT_CLANG_VERSION 80000
+#elif __has_warning("-Wc++98-compat-extra-semi")
+#define SIMDE_DETECT_CLANG_VERSION 70000
+#elif __has_warning("-Wpragma-pack")
+#define SIMDE_DETECT_CLANG_VERSION 60000
+#elif __has_warning("-Wbitfield-enum-conversion")
+#define SIMDE_DETECT_CLANG_VERSION 50000
+#elif __has_attribute(diagnose_if)
+#define SIMDE_DETECT_CLANG_VERSION 40000
+#elif __has_warning("-Wcast-calling-convention")
+#define SIMDE_DETECT_CLANG_VERSION 30900
+#elif __has_warning("-WCL4")
+#define SIMDE_DETECT_CLANG_VERSION 30800
+#elif __has_warning("-WIndependentClass-attribute")
+#define SIMDE_DETECT_CLANG_VERSION 30700
+#elif __has_warning("-Wambiguous-ellipsis")
+#define SIMDE_DETECT_CLANG_VERSION 30600
+#else
+#define SIMDE_DETECT_CLANG_VERSION 1
+#endif
+#endif /* defined(__clang__) && !defined(SIMDE_DETECT_CLANG_VERSION) */
+
+/* The SIMDE_DETECT_CLANG_VERSION_CHECK macro is pretty
+ * straightforward; it returns true if the compiler is a derivative
+ * of clang >= the specified version.
+ *
+ * Since this file is often (primarily?) useful for working around bugs
+ * it is also helpful to have a macro which returns true if only if the
+ * compiler is a version of clang *older* than the specified version to
+ * make it a bit easier to ifdef regions to add code for older versions,
+ * such as pragmas to disable a specific warning. */
+
+#if defined(SIMDE_DETECT_CLANG_VERSION)
+#define SIMDE_DETECT_CLANG_VERSION_CHECK(major, minor, revision) \
+	(SIMDE_DETECT_CLANG_VERSION >= \
+	 ((major * 10000) + (minor * 1000) + (revision)))
+#define SIMDE_DETECT_CLANG_VERSION_NOT(major, minor, revision) \
+	(SIMDE_DETECT_CLANG_VERSION < \
+	 ((major * 10000) + (minor * 1000) + (revision)))
+#else
+#define SIMDE_DETECT_CLANG_VERSION_CHECK(major, minor, revision) (0)
+#define SIMDE_DETECT_CLANG_VERSION_NOT(major, minor, revision) (1)
+#endif
+
+#endif /* !defined(SIMDE_DETECT_CLANG_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/simde-diagnostic.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-diagnostic.h Changed

@@ -45,8 +45,10 @@
 */
 
 #if !defined(SIMDE_DIAGNOSTIC_H)
+#define SIMDE_DIAGNOSTIC_H
 
 #include "hedley.h"
+#include "simde-detect-clang.h"
 
 /* This is only to help us implement functions like _mm_undefined_ps. */
 #if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
@@ -119,6 +121,9 @@
 #define SIMDE_DIAGNOSTIC_DISABLE_SIMD_PRAGMA_DEPRECATED_
 #endif
 
+/* MSVC emits a diagnostic when we call a function (like
+ * simde_mm_set_epi32) while initializing a struct. We currently do
+ * this a *lot* in the tests. */
 #if defined(HEDLEY_MSVC_VERSION)
 #define SIMDE_DIAGNOSTIC_DISABLE_NON_CONSTANT_AGGREGATE_INITIALIZER_ \
 	__pragma(warning(disable : 4204))
@@ -183,6 +188,32 @@
 #define SIMDE_DIAGNOSTIC_DISABLE_VARIADIC_MACROS_
 #endif
 
+/* emscripten requires us to use a __wasm_unimplemented_simd128__ macro
+ * before we can access certain SIMD intrinsics, but this diagnostic
+ * warns about it being a reserved name. It is a reserved name, but
+ * it's reserved for the compiler and we are using it to convey
+ * information to the compiler.
+ *
+ * This is also used when enabling native aliases since we don't get to
+ * choose the macro names. */
+#if HEDLEY_HAS_WARNING("-Wdouble-promotion")
+#define SIMDE_DIAGNOSTIC_DISABLE_RESERVED_ID_MACRO_ \
+	_Pragma("clang diagnostic ignored \"-Wreserved-id-macro\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_RESERVED_ID_MACRO_
+#endif
+
+/* clang 3.8 warns about the packed attribute being unnecessary when
+ * used in the _mm_loadu_* functions. That *may* be true for version
+ * 3.8, but for later versions it is crucial in order to make unaligned
+ * access safe. */
+#if HEDLEY_HAS_WARNING("-Wpacked")
+#define SIMDE_DIAGNOSTIC_DISABLE_PACKED_ \
+	_Pragma("clang diagnostic ignored \"-Wpacked\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_PACKED_
+#endif
+
 /* Triggered when assigning a float to a double implicitly. We use
 * explicit casts in SIMDe, this is only used in the test suite. */
 #if HEDLEY_HAS_WARNING("-Wdouble-promotion")
@@ -194,7 +225,7 @@
 
 /* Several compilers treat conformant array parameters as VLAs. We
 * test to make sure we're in C mode (C++ doesn't support CAPs), and
- * that the version of the standard supports CAPs. We also blacklist
+ * that the version of the standard supports CAPs. We also reject
 * some buggy compilers like MSVC (the logic is in Hedley if you want
 * to take a look), but with certain warnings enabled some compilers
 * still like to emit a diagnostic. */
@@ -221,6 +252,9 @@
 #elif HEDLEY_GCC_VERSION_CHECK(3, 4, 0)
 #define SIMDE_DIAGNOSTIC_DISABLE_UNUSED_FUNCTION_ \
 	_Pragma("GCC diagnostic ignored \"-Wunused-function\"")
+#elif HEDLEY_MSVC_VERSION_CHECK(19, 0, 0) /* Likely goes back further */
+#define SIMDE_DIAGNOSTIC_DISABLE_UNUSED_FUNCTION_ \
+	__pragma(warning(disable : 4505))
 #else
 #define SIMDE_DIAGNOSTIC_DISABLE_UNUSED_FUNCTION_
 #endif
@@ -232,13 +266,63 @@
 #define SIMDE_DIAGNOSTIC_DISABLE_PASS_FAILED_
 #endif
 
-/* https://github.com/nemequ/simde/issues/277 */
+#if HEDLEY_HAS_WARNING("-Wpadded")
+#define SIMDE_DIAGNOSTIC_DISABLE_PADDED_ \
+	_Pragma("clang diagnostic ignored \"-Wpadded\"")
+#elif HEDLEY_MSVC_VERSION_CHECK(19, 0, 0) /* Likely goes back further */
+#define SIMDE_DIAGNOSTIC_DISABLE_PADDED_ __pragma(warning(disable : 4324))
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_PADDED_
+#endif
+
+#if HEDLEY_HAS_WARNING("-Wzero-as-null-pointer-constant")
+#define SIMDE_DIAGNOSTIC_DISABLE_ZERO_AS_NULL_POINTER_CONSTANT_ \
+	_Pragma("clang diagnostic ignored \"-Wzero-as-null-pointer-constant\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_ZERO_AS_NULL_POINTER_CONSTANT_
+#endif
+
+#if HEDLEY_HAS_WARNING("-Wold-style-cast")
+#define SIMDE_DIAGNOSTIC_DISABLE_OLD_STYLE_CAST_ \
+	_Pragma("clang diagnostic ignored \"-Wold-style-cast\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_OLD_STYLE_CAST_
+#endif
+
+#if HEDLEY_HAS_WARNING("-Wcast-function-type") || \
+	HEDLEY_GCC_VERSION_CHECK(8, 0, 0)
+#define SIMDE_DIAGNOSTIC_DISABLE_CAST_FUNCTION_TYPE_ \
+	_Pragma("GCC diagnostic ignored \"-Wcast-function-type\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_CAST_FUNCTION_TYPE_
+#endif
+
+/* clang will emit this warning when we use C99 extensions whan not in
+ * C99 mode, even though it does support this. In such cases we check
+ * the compiler and version first, so we know it's not a problem. */
+#if HEDLEY_HAS_WARNING("-Wc99-extensions")
+#define SIMDE_DIAGNOSTIC_DISABLE_C99_EXTENSIONS_ \
+	_Pragma("clang diagnostic ignored \"-Wc99-extensions\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_C99_EXTENSIONS_
+#endif
+
+/* https://github.com/simd-everywhere/simde/issues/277 */
 #if defined(HEDLEY_GCC_VERSION) && HEDLEY_GCC_VERSION_CHECK(4, 6, 0) && \
-	!HEDLEY_GCC_VERSION_CHECK(6, 0, 0) && defined(__cplusplus)
-#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE \
+	!HEDLEY_GCC_VERSION_CHECK(6, 4, 0) && defined(__cplusplus)
+#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE_ \
 	_Pragma("GCC diagnostic ignored \"-Wunused-but-set-variable\"")
 #else
-#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE
+#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE_
+#endif
+
+/* This is the warning that you normally define _CRT_SECURE_NO_WARNINGS
+ * to silence, but you have to do that before including anything and
+ * that would require reordering includes. */
+#if defined(_MSC_VER)
+#define SIMDE_DIAGNOSTIC_DISABLE_ANNEX_K_ __pragma(warning(disable : 4996))
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_ANNEX_K_
 #endif
 
 /* Some compilers, such as clang, may use `long long` for 64-bit
@@ -246,13 +330,104 @@
 * -Wc++98-compat-pedantic which says 'long long' is incompatible with
 * C++98. */
 #if HEDLEY_HAS_WARNING("-Wc++98-compat-pedantic")
-#define SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC \
+#define SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_ \
 	_Pragma("clang diagnostic ignored \"-Wc++98-compat-pedantic\"")
 #else
-#define SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC
+#define SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_
+#endif
+
+/* Some problem as above */
+#if HEDLEY_HAS_WARNING("-Wc++11-long-long")
+#define SIMDE_DIAGNOSTIC_DISABLE_CPP11_LONG_LONG_ \
+	_Pragma("clang diagnostic ignored \"-Wc++11-long-long\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_CPP11_LONG_LONG_
+#endif
+
+/* emscripten emits this whenever stdin/stdout/stderr is used in a
+ * macro. */
+#if HEDLEY_HAS_WARNING("-Wdisabled-macro-expansion")
+#define SIMDE_DIAGNOSTIC_DISABLE_DISABLED_MACRO_EXPANSION_ \
+	_Pragma("clang diagnostic ignored \"-Wdisabled-macro-expansion\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_DISABLED_MACRO_EXPANSION_
+#endif
+
+/* Clang uses C11 generic selections to implement some AltiVec
+ * functions, which triggers this diagnostic when not compiling
+ * in C11 mode */
+#if HEDLEY_HAS_WARNING("-Wc11-extensions")
+#define SIMDE_DIAGNOSTIC_DISABLE_C11_EXTENSIONS_ \
+	_Pragma("clang diagnostic ignored \"-Wc11-extensions\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_C11_EXTENSIONS_
+#endif
+
+/* Clang sometimes triggers this warning in macros in the AltiVec and
+ * NEON headers, or due to missing functions. */
+#if HEDLEY_HAS_WARNING("-Wvector-conversion")
+#define SIMDE_DIAGNOSTIC_DISABLE_VECTOR_CONVERSION_ \
+	_Pragma("clang diagnostic ignored \"-Wvector-conversion\"")
+/* For NEON, the situation with -Wvector-conversion in clang < 10 is
+ * bad enough that we just disable the warning altogether. */
+#if defined(SIMDE_ARCH_ARM) && SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_VECTOR_CONVERSION_ \
+	SIMDE_DIAGNOSTIC_DISABLE_VECTOR_CONVERSION_
+#endif
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_VECTOR_CONVERSION_
+#endif
+#if !defined(SIMDE_DIAGNOSTIC_DISABLE_BUGGY_VECTOR_CONVERSION_)
+#define SIMDE_DIAGNOSTIC_DISABLE_BUGGY_VECTOR_CONVERSION_
+#endif
+
+/* SLEEF triggers this a *lot* in their headers */
+#if HEDLEY_HAS_WARNING("-Wignored-qualifiers")
+#define SIMDE_DIAGNOSTIC_DISABLE_IGNORED_QUALIFIERS_ \
+	_Pragma("clang diagnostic ignored \"-Wignored-qualifiers\"")
+#elif HEDLEY_GCC_VERSION_CHECK(4, 3, 0)
+#define SIMDE_DIAGNOSTIC_DISABLE_IGNORED_QUALIFIERS_ \
+	_Pragma("GCC diagnostic ignored \"-Wignored-qualifiers\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_IGNORED_QUALIFIERS_
+#endif
+
+/* GCC emits this under some circumstances when using __int128 */
+#if HEDLEY_GCC_VERSION_CHECK(4, 8, 0)
+#define SIMDE_DIAGNOSTIC_DISABLE_PEDANTIC_ \
+	_Pragma("GCC diagnostic ignored \"-Wpedantic\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_PEDANTIC_
+#endif
+
+/* MSVC doesn't like (__assume(0), code) and will warn about code being
+ * unreachable, but we want it there because not all compilers
+ * understand the unreachable macro and will complain if it is missing.
+ * I'm planning on adding a new macro to Hedley to handle this a bit
+ * more elegantly, but until then... */
+#if defined(HEDLEY_MSVC_VERSION)
+#define SIMDE_DIAGNOSTIC_DISABLE_UNREACHABLE_ __pragma(warning(disable : 4702))
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_UNREACHABLE_
+#endif
+
+/* This is a false positive from GCC in a few places. */
+#if HEDLEY_GCC_VERSION_CHECK(4, 7, 0)
+#define SIMDE_DIAGNOSTIC_DISABLE_MAYBE_UNINITIAZILED_ \
+	_Pragma("GCC diagnostic ignored \"-Wmaybe-uninitialized\"")
+#else
+#define SIMDE_DIAGNOSTIC_DISABLE_MAYBE_UNINITIAZILED_
+#endif
+
+#if defined(SIMDE_ENABLE_NATIVE_ALIASES)
+#define SIMDE_DISABLE_UNWANTED_DIAGNOSTICS_NATIVE_ALIASES_ \
+	SIMDE_DIAGNOSTIC_DISABLE_RESERVED_ID_MACRO_
+#else
+#define SIMDE_DISABLE_UNWANTED_DIAGNOSTICS_NATIVE_ALIASES_
 #endif
 
 #define SIMDE_DISABLE_UNWANTED_DIAGNOSTICS \
+	SIMDE_DISABLE_UNWANTED_DIAGNOSTICS_NATIVE_ALIASES_ \
 	SIMDE_DIAGNOSTIC_DISABLE_PSABI_ \
 	SIMDE_DIAGNOSTIC_DISABLE_NO_EMMS_INSTRUCTION_ \
 	SIMDE_DIAGNOSTIC_DISABLE_SIMD_PRAGMA_DEPRECATED_ \
@@ -264,7 +439,9 @@
 	SIMDE_DIAGNOSTIC_DISABLE_USED_BUT_MARKED_UNUSED_ \
 	SIMDE_DIAGNOSTIC_DISABLE_UNUSED_FUNCTION_ \
 	SIMDE_DIAGNOSTIC_DISABLE_PASS_FAILED_ \
-	SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC \
-	SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE
+	SIMDE_DIAGNOSTIC_DISABLE_CPP98_COMPAT_PEDANTIC_ \
+	SIMDE_DIAGNOSTIC_DISABLE_CPP11_LONG_LONG_ \
+	SIMDE_DIAGNOSTIC_DISABLE_BUGGY_UNUSED_BUT_SET_VARIBALE_ \
+	SIMDE_DIAGNOSTIC_DISABLE_BUGGY_VECTOR_CONVERSION_
 
-#endif
+#endif /* !defined(SIMDE_DIAGNOSTIC_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/simde-features.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-features.h Changed

@@ -32,6 +32,7 @@
 #define SIMDE_FEATURES_H
 
 #include "simde-arch.h"
+#include "simde-diagnostic.h"
 
 #if !defined(SIMDE_X86_SVML_NATIVE) && !defined(SIMDE_X86_SVML_NO_NATIVE) && \
 	!defined(SIMDE_NO_NATIVE)
@@ -43,6 +44,28 @@
 #define SIMDE_X86_AVX512F_NATIVE
 #endif
 
+#if !defined(SIMDE_X86_AVX512VP2INTERSECT_NATIVE) && \
+	!defined(SIMDE_X86_AVX512VP2INTERSECT_NO_NATIVE) && \
+	!defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_X86_AVX512VP2INTERSECT)
+#define SIMDE_X86_AVX512VP2INTERSECT_NATIVE
+#endif
+#endif
+#if defined(SIMDE_X86_AVX512VP2INTERSECT_NATIVE) && \
+	!defined(SIMDE_X86_AVX512F_NATIVE)
+#define SIMDE_X86_AVX512F_NATIVE
+#endif
+
+#if !defined(SIMDE_X86_AVX512VBMI_NATIVE) && \
+	!defined(SIMDE_X86_AVX512VBMI_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_X86_AVX512VBMI)
+#define SIMDE_X86_AVX512VBMI_NATIVE
+#endif
+#endif
+#if defined(SIMDE_X86_AVX512VBMI_NATIVE) && !defined(SIMDE_X86_AVX512F_NATIVE)
+#define SIMDE_X86_AVX512F_NATIVE
+#endif
+
 #if !defined(SIMDE_X86_AVX512CD_NATIVE) && \
 	!defined(SIMDE_X86_AVX512CD_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
 #if defined(SIMDE_ARCH_X86_AVX512CD)
@@ -194,6 +217,20 @@
 #endif
 #endif
 
+#if !defined(SIMDE_X86_PCLMUL_NATIVE) && \
+	!defined(SIMDE_X86_PCLMUL_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_X86_PCLMUL)
+#define SIMDE_X86_PCLMUL_NATIVE
+#endif
+#endif
+
+#if !defined(SIMDE_X86_VPCLMULQDQ_NATIVE) && \
+	!defined(SIMDE_X86_VPCLMULQDQ_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_X86_VPCLMULQDQ)
+#define SIMDE_X86_VPCLMULQDQ_NATIVE
+#endif
+#endif
+
 #if !defined(SIMDE_X86_SVML_NATIVE) && !defined(SIMDE_X86_SVML_NO_NATIVE) && \
 	!defined(SIMDE_NO_NATIVE)
 #if defined(__INTEL_COMPILER)
@@ -206,8 +243,7 @@
 #pragma warning(disable : 4799)
 #endif
 
-#if defined(SIMDE_X86_AVX_NATIVE) || defined(SIMDE_X86_GFNI_NATIVE) || \
-	defined(SIMDE_X86_SVML_NATIVE)
+#if defined(SIMDE_X86_AVX_NATIVE) || defined(SIMDE_X86_GFNI_NATIVE)
 #include <immintrin.h>
 #elif defined(SIMDE_X86_SSE4_2_NATIVE)
 #include <nmmintrin.h>
@@ -243,7 +279,8 @@
 
 #if !defined(SIMDE_ARM_NEON_A32V8_NATIVE) && \
 	!defined(SIMDE_ARM_NEON_A32V8_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
-#if defined(SIMDE_ARCH_ARM_NEON) && SIMDE_ARCH_ARM_CHECK(80)
+#if defined(SIMDE_ARCH_ARM_NEON) && SIMDE_ARCH_ARM_CHECK(80) && \
+	(__ARM_NEON_FP & 0x02)
 #define SIMDE_ARM_NEON_A32V8_NATIVE
 #endif
 #endif
@@ -262,6 +299,14 @@
 #include <arm_neon.h>
 #endif
 
+#if !defined(SIMDE_ARM_SVE_NATIVE) && !defined(SIMDE_ARM_SVE_NO_NATIVE) && \
+	!defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_ARM_SVE)
+#define SIMDE_ARM_SVE_NATIVE
+#include <arm_sve.h>
+#endif
+#endif
+
 #if !defined(SIMDE_WASM_SIMD128_NATIVE) && \
 	!defined(SIMDE_WASM_SIMD128_NO_NATIVE) && !defined(SIMDE_NO_NATIVE)
 #if defined(SIMDE_ARCH_WASM_SIMD128)
@@ -270,7 +315,10 @@
 #endif
 #if defined(SIMDE_WASM_SIMD128_NATIVE)
 #if !defined(__wasm_unimplemented_simd128__)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_RESERVED_ID_MACRO_
 #define __wasm_unimplemented_simd128__
+HEDLEY_DIAGNOSTIC_POP
 #endif
 #include <wasm_simd128.h>
 #endif
@@ -326,15 +374,28 @@
 #define SIMDE_POWER_ALTIVEC_P5_NATIVE
 #endif
 #endif
-#if defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
-/* stdbool.h conflicts with the bool in altivec.h */
-#if defined(bool) && !defined(SIMDE_POWER_ALTIVEC_NO_UNDEF_BOOL_)
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+/* AltiVec conflicts with lots of stuff. The bool keyword conflicts
+ * with the bool keyword in C++ and the bool macro in C99+ (defined
+ * in stdbool.h). The vector keyword conflicts with std::vector in
+ * C++ if you are `using std;`.
+ *
+ * Luckily AltiVec allows you to use `__vector`/`__bool`/`__pixel`
+ * instead, but altivec.h will unconditionally define
+ * `vector`/`bool`/`pixel` so we need to work around that.
+ *
+ * Unfortunately this means that if your code uses AltiVec directly
+ * it may break. If this is the case you'll want to define
+ * `SIMDE_POWER_ALTIVEC_NO_UNDEF` before including SIMDe. Or, even
+ * better, port your code to use the double-underscore versions. */
+#if defined(bool)
 #undef bool
 #endif
+
 #include <altivec.h>
-/* GCC allows you to undefine these macros to prevent conflicts with
- * standard types as they become context-sensitive keywords. */
-#if defined(__cplusplus)
+
+#if !defined(SIMDE_POWER_ALTIVEC_NO_UNDEF)
 #if defined(vector)
 #undef vector
 #endif
@@ -344,14 +405,146 @@
 #if defined(bool)
 #undef bool
 #endif
-#define SIMDE_POWER_ALTIVEC_VECTOR(T) vector T
-#define SIMDE_POWER_ALTIVEC_PIXEL pixel
-#define SIMDE_POWER_ALTIVEC_BOOL bool
-#else
+#endif /* !defined(SIMDE_POWER_ALTIVEC_NO_UNDEF) */
+
+/* Use these intsead of vector/pixel/bool in SIMDe. */
 #define SIMDE_POWER_ALTIVEC_VECTOR(T) __vector T
 #define SIMDE_POWER_ALTIVEC_PIXEL __pixel
 #define SIMDE_POWER_ALTIVEC_BOOL __bool
-#endif /* defined(__cplusplus) */
+
+/* Re-define bool if we're using stdbool.h */
+#if !defined(__cplusplus) && defined(__bool_true_false_are_defined) && \
+	!defined(SIMDE_POWER_ALTIVEC_NO_UNDEF)
+#define bool _Bool
+#endif
+#endif
+
+#if !defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE) && \
+	!defined(SIMDE_MIPS_LOONGSON_MMI_NO_NATIVE) && \
+	!defined(SIMDE_NO_NATIVE)
+#if defined(SIMDE_ARCH_MIPS_LOONGSON_MMI)
+#define SIMDE_MIPS_LOONGSON_MMI_NATIVE 1
+#endif
+#endif
+#if defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+#include <loongson-mmiintrin.h>
+#endif
+
+/* This is used to determine whether or not to fall back on a vector
+ * function in an earlier ISA extensions, as well as whether
+ * we expected any attempts at vectorization to be fruitful or if we
+ * expect to always be running serial code. */
+
+#if !defined(SIMDE_NATURAL_VECTOR_SIZE)
+#if defined(SIMDE_X86_AVX512F_NATIVE)
+#define SIMDE_NATURAL_VECTOR_SIZE (512)
+#elif defined(SIMDE_X86_AVX_NATIVE)
+#define SIMDE_NATURAL_VECTOR_SIZE (256)
+#elif defined(SIMDE_X86_SSE_NATIVE) || defined(SIMDE_ARM_NEON_A32V7_NATIVE) || \
+	defined(SIMDE_WASM_SIMD128_NATIVE) || \
+	defined(SIMDE_POWER_ALTIVEC_P5_NATIVE)
+#define SIMDE_NATURAL_VECTOR_SIZE (128)
+#endif
+
+#if !defined(SIMDE_NATURAL_VECTOR_SIZE)
+#define SIMDE_NATURAL_VECTOR_SIZE (0)
+#endif
+#endif
+
+#define SIMDE_NATURAL_VECTOR_SIZE_LE(x) \
+	((SIMDE_NATURAL_VECTOR_SIZE > 0) && (SIMDE_NATURAL_VECTOR_SIZE <= (x)))
+#define SIMDE_NATURAL_VECTOR_SIZE_GE(x) \
+	((SIMDE_NATURAL_VECTOR_SIZE > 0) && (SIMDE_NATURAL_VECTOR_SIZE >= (x)))
+
+/* Native aliases */
+#if defined(SIMDE_ENABLE_NATIVE_ALIASES)
+#if !defined(SIMDE_X86_MMX_NATIVE)
+#define SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSE_NATIVE)
+#define SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSE2_NATIVE)
+#define SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSE3_NATIVE)
+#define SIMDE_X86_SSE3_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSSE3_NATIVE)
+#define SIMDE_X86_SSSE3_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSE4_1_NATIVE)
+#define SIMDE_X86_SSE4_1_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_SSE4_2_NATIVE)
+#define SIMDE_X86_SSE4_2_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX_NATIVE)
+#define SIMDE_X86_AVX_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX2_NATIVE)
+#define SIMDE_X86_AVX2_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_FMA_NATIVE)
+#define SIMDE_X86_FMA_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX512F_NATIVE)
+#define SIMDE_X86_AVX512F_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX512VL_NATIVE)
+#define SIMDE_X86_AVX512VL_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX512BW_NATIVE)
+#define SIMDE_X86_AVX512BW_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX512DQ_NATIVE)
+#define SIMDE_X86_AVX512DQ_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_AVX512CD_NATIVE)
+#define SIMDE_X86_AVX512CD_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_GFNI_NATIVE)
+#define SIMDE_X86_GFNI_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_PCLMUL_NATIVE)
+#define SIMDE_X86_PCLMUL_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_X86_VPCLMULQDQ_NATIVE)
+#define SIMDE_X86_VPCLMULQDQ_ENABLE_NATIVE_ALIASES
+#endif
+
+#if !defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define SIMDE_ARM_NEON_A32V7_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+#define SIMDE_ARM_NEON_A32V8_ENABLE_NATIVE_ALIASES
+#endif
+#if !defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+#define SIMDE_ARM_NEON_A64V8_ENABLE_NATIVE_ALIASES
+#endif
+#endif
+
+/* Are floating point values stored using IEEE 754? Knowing
+ * this at during preprocessing is a bit tricky, mostly because what
+ * we're curious about is how values are stored and not whether the
+ * implementation is fully conformant in terms of rounding, NaN
+ * handling, etc.
+ *
+ * For example, if you use -ffast-math or -Ofast on
+ * GCC or clang IEEE 754 isn't strictly followed, therefore IEE 754
+ * support is not advertised (by defining __STDC_IEC_559__).
+ *
+ * However, what we care about is whether it is safe to assume that
+ * floating point values are stored in IEEE 754 format, in which case
+ * we can provide faster implementations of some functions.
+ *
+ * Luckily every vaugely modern architecture I'm aware of uses IEEE 754-
+ * so we just assume IEEE 754 for now. There is a test which verifies
+ * this, if that test fails sowewhere please let us know and we'll add
+ * an exception for that platform. Meanwhile, you can define
+ * SIMDE_NO_IEEE754_STORAGE. */
+#if !defined(SIMDE_IEEE754_STORAGE) && !defined(SIMDE_NO_IEE754_STORAGE)
+#define SIMDE_IEEE754_STORAGE
 #endif
 
 #endif /* !defined(SIMDE_FEATURES_H) */

obs-studio-26.1.0.tar.xz/libobs/util/simde/simde-math.h -> obs-studio-26.1.1.tar.xz/libobs/util/simde/simde-math.h Changed

@@ -34,6 +34,58 @@
 #include "hedley.h"
 #include "simde-features.h"
 
+#include <stdint.h>
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+#include <arm_neon.h>
+#endif
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+
+/* SLEEF support
+ * https://sleef.org/
+ *
+ * If you include <sleef.h> prior to including SIMDe, SIMDe will use
+ * SLEEF. You can also define SIMDE_MATH_SLEEF_ENABLE prior to
+ * including SIMDe to force the issue.
+ *
+ * Note that SLEEF does requires linking to libsleef.
+ *
+ * By default, SIMDe will use the 1 ULP functions, but if you use
+ * SIMDE_ACCURACY_PREFERENCE of 0 we will use up to 4 ULP. This is
+ * only the case for the simde_math_* functions; for code in other
+ * SIMDe headers which calls SLEEF directly we may use functions with
+ * greater error if the API we're implementing is less precise (for
+ * example, SVML guarantees 4 ULP, so we will generally use the 3.5
+ * ULP functions from SLEEF). */
+#if !defined(SIMDE_MATH_SLEEF_DISABLE)
+#if defined(__SLEEF_H__)
+#define SIMDE_MATH_SLEEF_ENABLE
+#endif
+#endif
+
+#if defined(SIMDE_MATH_SLEEF_ENABLE) && !defined(__SLEEF_H__)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_IGNORED_QUALIFIERS_
+#include <sleef.h>
+HEDLEY_DIAGNOSTIC_POP
+#endif
+
+#if defined(SIMDE_MATH_SLEEF_ENABLE) && defined(__SLEEF_H__)
+#if defined(SLEEF_VERSION_MAJOR)
+#define SIMDE_MATH_SLEEF_VERSION_CHECK(major, minor, patch) \
+	(HEDLEY_VERSION_ENCODE(SLEEF_VERSION_MAJOR, SLEEF_VERSION_MINOR, \
+			 SLEEF_VERSION_PATCHLEVEL) >= \
+	 HEDLEY_VERSION_ENCODE(major, minor, patch))
+#else
+#define SIMDE_MATH_SLEEF_VERSION_CHECK(major, minor, patch) \
+	(HEDLEY_VERSION_ENCODE(3, 0, 0) >= \
+	 HEDLEY_VERSION_ENCODE(major, minor, patch))
+#endif
+#else
+#define SIMDE_MATH_SLEEF_VERSION_CHECK(major, minor, patch) (0)
+#endif
+
 #if defined(__has_builtin)
 #define SIMDE_MATH_BUILTIN_LIBM(func) __has_builtin(__builtin_##func)
 #elif HEDLEY_INTEL_VERSION_CHECK(13, 0, 0) || \
@@ -82,11 +134,35 @@
 #endif
 #endif
 
-#if !defined(__cplusplus)
-/* If this is a problem we *might* be able to avoid including
- * <complex.h> on some compilers (gcc, clang, and others which
- * implement builtins like __builtin_cexpf). If you don't have
- * a <complex.h> please file an issue and we'll take a look. */
+/* Try to avoid including <complex> since it pulls in a *lot* of code. */
+#if HEDLEY_HAS_BUILTIN(__builtin_creal) || \
+	HEDLEY_GCC_VERSION_CHECK(4, 7, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_C99_EXTENSIONS_
+typedef __complex__ float simde_cfloat32;
+typedef __complex__ double simde_cfloat64;
+HEDLEY_DIAGNOSTIC_POP
+#define SIMDE_MATH_CMPLX(x, y) \
+	(HEDLEY_STATIC_CAST(double, x) + \
+	 HEDLEY_STATIC_CAST(double, y) * (__extension__ 1.0j))
+#define SIMDE_MATH_CMPLXF(x, y) \
+	(HEDLEY_STATIC_CAST(float, x) + \
+	 HEDLEY_STATIC_CAST(float, y) * (__extension__ 1.0fj))
+
+#if !defined(simde_math_creal)
+#define simde_math_crealf(z) __builtin_crealf(z)
+#endif
+#if !defined(simde_math_crealf)
+#define simde_math_creal(z) __builtin_creal(z)
+#endif
+#if !defined(simde_math_cimag)
+#define simde_math_cimagf(z) __builtin_cimagf(z)
+#endif
+#if !defined(simde_math_cimagf)
+#define simde_math_cimag(z) __builtin_cimag(z)
+#endif
+#elif !defined(__cplusplus)
 #include <complex.h>
 
 #if !defined(HEDLEY_MSVC_VERSION)
@@ -96,20 +172,14 @@
 typedef _Fcomplex simde_cfloat32;
 typedef _Dcomplex simde_cfloat64;
 #endif
-#if HEDLEY_HAS_BUILTIN(__builtin_complex) || \
-	HEDLEY_GCC_VERSION_CHECK(4, 7, 0) || \
-	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
-#define SIMDE_MATH_CMPLX(x, y) __builtin_complex((double)(x), (double)(y))
-#define SIMDE_MATH_CMPLXF(x, y) __builtin_complex((float)(x), (float)(y))
-#elif defined(HEDLEY_MSVC_VERSION)
+
+#if defined(HEDLEY_MSVC_VERSION)
 #define SIMDE_MATH_CMPLX(x, y) ((simde_cfloat64){(x), (y)})
 #define SIMDE_MATH_CMPLXF(x, y) ((simde_cfloat32){(x), (y)})
 #elif defined(CMPLX) && defined(CMPLXF)
 #define SIMDE_MATH_CMPLX(x, y) CMPLX(x, y)
 #define SIMDE_MATH_CMPLXF(x, y) CMPLXF(x, y)
 #else
-/* CMPLX / CMPLXF are in C99, but these seem to be necessary in
- * some compilers that aren't even MSVC. */
 #define SIMDE_MATH_CMPLX(x, y) \
 	(HEDLEY_STATIC_CAST(double, x) + HEDLEY_STATIC_CAST(double, y) * I)
 #define SIMDE_MATH_CMPLXF(x, y) \
@@ -117,38 +187,18 @@
 #endif
 
 #if !defined(simde_math_creal)
-#if SIMDE_MATH_BUILTIN_LIBM(creal)
-#define simde_math_creal(z) __builtin_creal(z)
-#else
 #define simde_math_creal(z) creal(z)
 #endif
-#endif
-
 #if !defined(simde_math_crealf)
-#if SIMDE_MATH_BUILTIN_LIBM(crealf)
-#define simde_math_crealf(z) __builtin_crealf(z)
-#else
 #define simde_math_crealf(z) crealf(z)
 #endif
-#endif
-
 #if !defined(simde_math_cimag)
-#if SIMDE_MATH_BUILTIN_LIBM(cimag)
-#define simde_math_cimag(z) __builtin_cimag(z)
-#else
 #define simde_math_cimag(z) cimag(z)
 #endif
-#endif
-
 #if !defined(simde_math_cimagf)
-#if SIMDE_MATH_BUILTIN_LIBM(cimagf)
-#define simde_math_cimagf(z) __builtin_cimagf(z)
-#else
 #define simde_math_cimagf(z) cimagf(z)
 #endif
-#endif
 #else
-
 HEDLEY_DIAGNOSTIC_PUSH
 #if defined(HEDLEY_MSVC_VERSION)
 #pragma warning(disable : 4530)
@@ -240,6 +290,26 @@
 #endif
 #endif
 
+#if !defined(SIMDE_MATH_PI_OVER_180)
+#define SIMDE_MATH_PI_OVER_180 \
+	0.0174532925199432957692369076848861271344287188854172545609719144
+#endif
+
+#if !defined(SIMDE_MATH_PI_OVER_180F)
+#define SIMDE_MATH_PI_OVER_180F \
+	0.0174532925199432957692369076848861271344287188854172545609719144f
+#endif
+
+#if !defined(SIMDE_MATH_180_OVER_PI)
+#define SIMDE_MATH_180_OVER_PI \
+	57.295779513082320876798154814105170332405472466564321549160243861
+#endif
+
+#if !defined(SIMDE_MATH_180_OVER_PIF)
+#define SIMDE_MATH_180_OVER_PIF \
+	57.295779513082320876798154814105170332405472466564321549160243861f
+#endif
+
 #if !defined(SIMDE_MATH_FLT_MIN)
 #if defined(FLT_MIN)
 #define SIMDE_MATH_FLT_MIN FLT_MIN
@@ -341,6 +411,36 @@
 #endif
 #endif
 
+/*** Manipulation functions ***/
+
+#if !defined(simde_math_nextafter)
+#if (HEDLEY_HAS_BUILTIN(__builtin_nextafter) && \
+ !defined(HEDLEY_IBM_VERSION)) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_GCC_VERSION_CHECK(3, 4, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+#define simde_math_nextafter(x, y) __builtin_nextafter(x, y)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_nextafter(x, y) std::nextafter(x, y)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_nextafter(x, y) nextafter(x, y)
+#endif
+#endif
+
+#if !defined(simde_math_nextafterf)
+#if (HEDLEY_HAS_BUILTIN(__builtin_nextafterf) && \
+ !defined(HEDLEY_IBM_VERSION)) || \
+	HEDLEY_ARM_VERSION_CHECK(4, 1, 0) || \
+	HEDLEY_GCC_VERSION_CHECK(3, 4, 0) || \
+	HEDLEY_INTEL_VERSION_CHECK(13, 0, 0)
+#define simde_math_nextafterf(x, y) __builtin_nextafterf(x, y)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_nextafterf(x, y) std::nextafter(x, y)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_nextafterf(x, y) nextafterf(x, y)
+#endif
+#endif
+
 /*** Functions from C99 ***/
 
 #if !defined(simde_math_abs)
@@ -353,13 +453,13 @@
 #endif
 #endif
 
-#if !defined(simde_math_absf)
-#if SIMDE_MATH_BUILTIN_LIBM(absf)
-#define simde_math_absf(v) __builtin_absf(v)
+#if !defined(simde_math_fabsf)
+#if SIMDE_MATH_BUILTIN_LIBM(fabsf)
+#define simde_math_fabsf(v) __builtin_fabsf(v)
 #elif defined(SIMDE_MATH_HAVE_CMATH)
-#define simde_math_absf(v) std::abs(v)
+#define simde_math_fabsf(v) std::abs(v)
 #elif defined(SIMDE_MATH_HAVE_MATH_H)
-#define simde_math_absf(v) absf(v)
+#define simde_math_fabsf(v) fabsf(v)
 #endif
 #endif
 
@@ -574,7 +674,13 @@
 #endif
 
 #if !defined(simde_math_cosf)
-#if SIMDE_MATH_BUILTIN_LIBM(cosf)
+#if defined(SIMDE_MATH_SLEEF_ENABLE)
+#if SIMDE_ACCURACY_PREFERENCE < 1
+#define simde_math_cosf(v) Sleef_cosf_u35(v)
+#else
+#define simde_math_cosf(v) Sleef_cosf_u10(v)
+#endif
+#elif SIMDE_MATH_BUILTIN_LIBM(cosf)
 #define simde_math_cosf(v) __builtin_cosf(v)
 #elif defined(SIMDE_MATH_HAVE_CMATH)
 #define simde_math_cosf(v) std::cos(v)
@@ -755,6 +861,46 @@
 #endif
 #endif
 
+#if !defined(simde_math_fma)
+#if SIMDE_MATH_BUILTIN_LIBM(fma)
+#define simde_math_fma(x, y, z) __builtin_fma(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_fma(x, y, z) std::fma(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_fma(x, y, z) fma(x, y, z)
+#endif
+#endif
+
+#if !defined(simde_math_fmaf)
+#if SIMDE_MATH_BUILTIN_LIBM(fmaf)
+#define simde_math_fmaf(x, y, z) __builtin_fmaf(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_fmaf(x, y, z) std::fma(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_fmaf(x, y, z) fmaf(x, y, z)
+#endif
+#endif
+
+#if !defined(simde_math_fmax)
+#if SIMDE_MATH_BUILTIN_LIBM(fmax)
+#define simde_math_fmax(x, y, z) __builtin_fmax(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_fmax(x, y, z) std::fmax(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_fmax(x, y, z) fmax(x, y, z)
+#endif
+#endif
+
+#if !defined(simde_math_fmaxf)
+#if SIMDE_MATH_BUILTIN_LIBM(fmaxf)
+#define simde_math_fmaxf(x, y, z) __builtin_fmaxf(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_fmaxf(x, y, z) std::fmax(x, y, z)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_fmaxf(x, y, z) fmaxf(x, y, z)
+#endif
+#endif
+
 #if !defined(simde_math_hypot)
 #if SIMDE_MATH_BUILTIN_LIBM(hypot)
 #define simde_math_hypot(y, x) __builtin_hypot(y, x)
@@ -875,6 +1021,26 @@
 #endif
 #endif
 
+#if !defined(simde_math_modf)
+#if SIMDE_MATH_BUILTIN_LIBM(modf)
+#define simde_math_modf(x, iptr) __builtin_modf(x, iptr)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_modf(x, iptr) std::modf(x, iptr)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_modf(x, iptr) modf(x, iptr)
+#endif
+#endif
+
+#if !defined(simde_math_modff)
+#if SIMDE_MATH_BUILTIN_LIBM(modff)
+#define simde_math_modff(x, iptr) __builtin_modff(x, iptr)
+#elif defined(SIMDE_MATH_HAVE_CMATH)
+#define simde_math_modff(x, iptr) std::modf(x, iptr)
+#elif defined(SIMDE_MATH_HAVE_MATH_H)
+#define simde_math_modff(x, iptr) modff(x, iptr)
+#endif
+#endif
+
 #if !defined(simde_math_nearbyint)
 #if SIMDE_MATH_BUILTIN_LIBM(nearbyint)
 #define simde_math_nearbyint(v) __builtin_nearbyint(v)
@@ -955,6 +1121,44 @@
 #endif
 #endif
 
+#if !defined(simde_math_roundeven)
+#if HEDLEY_HAS_BUILTIN(__builtin_roundeven) || \
+	HEDLEY_GCC_VERSION_CHECK(10, 0, 0)
+#define simde_math_roundeven(v) __builtin_roundeven(v)
+#elif defined(simde_math_round) && defined(simde_math_fabs)
+static HEDLEY_INLINE double simde_math_roundeven(double v)
+{
+	double rounded = simde_math_round(v);
+	double diff = rounded - v;
+	if (HEDLEY_UNLIKELY(simde_math_fabs(diff) == 0.5) &&
+	 (HEDLEY_STATIC_CAST(int64_t, rounded) & 1)) {
+		rounded = v - diff;
+	}
+	return rounded;
+}
+#define simde_math_roundeven simde_math_roundeven
+#endif
+#endif
+
+#if !defined(simde_math_roundevenf)
+#if HEDLEY_HAS_BUILTIN(__builtin_roundevenf) || \
+	HEDLEY_GCC_VERSION_CHECK(10, 0, 0)
+#define simde_math_roundevenf(v) __builtin_roundevenf(v)
+#elif defined(simde_math_roundf) && defined(simde_math_fabsf)
+static HEDLEY_INLINE float simde_math_roundevenf(float v)
+{
+	float rounded = simde_math_roundf(v);
+	float diff = rounded - v;
+	if (HEDLEY_UNLIKELY(simde_math_fabsf(diff) == 0.5f) &&
+	 (HEDLEY_STATIC_CAST(int32_t, rounded) & 1)) {
+		rounded = v - diff;
+	}
+	return rounded;
+}
+#define simde_math_roundevenf simde_math_roundevenf
+#endif
+#endif
+
 #if !defined(simde_math_sin)
 #if SIMDE_MATH_BUILTIN_LIBM(sin)
 #define simde_math_sin(v) __builtin_sin(v)
@@ -1078,20 +1282,20 @@
 /*** Complex functions ***/
 
 #if !defined(simde_math_cexp)
-#if defined(__cplusplus)
-#define simde_math_cexp(v) std::cexp(v)
-#elif SIMDE_MATH_BUILTIN_LIBM(cexp)
+#if SIMDE_MATH_BUILTIN_LIBM(cexp)
 #define simde_math_cexp(v) __builtin_cexp(v)
+#elif defined(__cplusplus)
+#define simde_math_cexp(v) std::cexp(v)
 #elif defined(SIMDE_MATH_HAVE_MATH_H)
 #define simde_math_cexp(v) cexp(v)
 #endif
 #endif
 
 #if !defined(simde_math_cexpf)
-#if defined(__cplusplus)
-#define simde_math_cexpf(v) std::exp(v)
-#elif SIMDE_MATH_BUILTIN_LIBM(cexpf)
+#if SIMDE_MATH_BUILTIN_LIBM(cexpf)
 #define simde_math_cexpf(v) __builtin_cexpf(v)
+#elif defined(__cplusplus)
+#define simde_math_cexpf(v) std::exp(v)
 #elif defined(SIMDE_MATH_HAVE_MATH_H)
 #define simde_math_cexpf(v) cexpf(v)
 #endif
@@ -1393,22 +1597,262 @@
 
 static HEDLEY_INLINE double simde_math_rad2deg(double radians)
 {
-	return radians * (180.0 / SIMDE_MATH_PI);
+	return radians * SIMDE_MATH_180_OVER_PI;
 }
 
 static HEDLEY_INLINE float simde_math_rad2degf(float radians)
 {
-	return radians * (180.0f / SIMDE_MATH_PIF);
+	return radians * SIMDE_MATH_180_OVER_PIF;
 }
 
 static HEDLEY_INLINE double simde_math_deg2rad(double degrees)
 {
-	return degrees * (SIMDE_MATH_PI / 180.0);
+	return degrees * SIMDE_MATH_PI_OVER_180;
 }
 
 static HEDLEY_INLINE float simde_math_deg2radf(float degrees)
 {
-	return degrees * (SIMDE_MATH_PIF / 180.0f);
+	return degrees * (SIMDE_MATH_PI_OVER_180F);
 }
 
+/*** Saturated arithmetic ***/
+
+static HEDLEY_INLINE int8_t simde_math_adds_i8(int8_t a, int8_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddb_s8(a, b);
+#else
+	uint8_t a_ = HEDLEY_STATIC_CAST(uint8_t, a);
+	uint8_t b_ = HEDLEY_STATIC_CAST(uint8_t, b);
+	uint8_t r_ = a_ + b_;
+
+	a_ = (a_ >> ((8 * sizeof(r_)) - 1)) + INT8_MAX;
+	if (HEDLEY_STATIC_CAST(int8_t, ((a_ ^ b_) | ~(b_ ^ r_))) >= 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int8_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int16_t simde_math_adds_i16(int16_t a, int16_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddh_s16(a, b);
+#else
+	uint16_t a_ = HEDLEY_STATIC_CAST(uint16_t, a);
+	uint16_t b_ = HEDLEY_STATIC_CAST(uint16_t, b);
+	uint16_t r_ = a_ + b_;
+
+	a_ = (a_ >> ((8 * sizeof(r_)) - 1)) + INT16_MAX;
+	if (HEDLEY_STATIC_CAST(int16_t, ((a_ ^ b_) | ~(b_ ^ r_))) >= 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int16_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int32_t simde_math_adds_i32(int32_t a, int32_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqadds_s32(a, b);
+#else
+	uint32_t a_ = HEDLEY_STATIC_CAST(uint32_t, a);
+	uint32_t b_ = HEDLEY_STATIC_CAST(uint32_t, b);
+	uint32_t r_ = a_ + b_;
+
+	a_ = (a_ >> ((8 * sizeof(r_)) - 1)) + INT32_MAX;
+	if (HEDLEY_STATIC_CAST(int32_t, ((a_ ^ b_) | ~(b_ ^ r_))) >= 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int32_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int64_t simde_math_adds_i64(int64_t a, int64_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddd_s64(a, b);
+#else
+	uint64_t a_ = HEDLEY_STATIC_CAST(uint64_t, a);
+	uint64_t b_ = HEDLEY_STATIC_CAST(uint64_t, b);
+	uint64_t r_ = a_ + b_;
+
+	a_ = (a_ >> ((8 * sizeof(r_)) - 1)) + INT64_MAX;
+	if (HEDLEY_STATIC_CAST(int64_t, ((a_ ^ b_) | ~(b_ ^ r_))) >= 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int64_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE uint8_t simde_math_adds_u8(uint8_t a, uint8_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddb_u8(a, b);
+#else
+	uint8_t r = a + b;
+	r |= -(r < a);
+	return r;
+#endif
+}
+
+static HEDLEY_INLINE uint16_t simde_math_adds_u16(uint16_t a, uint16_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddh_u16(a, b);
+#else
+	uint16_t r = a + b;
+	r |= -(r < a);
+	return r;
+#endif
+}
+
+static HEDLEY_INLINE uint32_t simde_math_adds_u32(uint32_t a, uint32_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqadds_u32(a, b);
+#else
+	uint32_t r = a + b;
+	r |= -(r < a);
+	return r;
+#endif
+}
+
+static HEDLEY_INLINE uint64_t simde_math_adds_u64(uint64_t a, uint64_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqaddd_u64(a, b);
+#else
+	uint64_t r = a + b;
+	r |= -(r < a);
+	return r;
+#endif
+}
+
+static HEDLEY_INLINE int8_t simde_math_subs_i8(int8_t a, int8_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubb_s8(a, b);
+#else
+	uint8_t a_ = HEDLEY_STATIC_CAST(uint8_t, a);
+	uint8_t b_ = HEDLEY_STATIC_CAST(uint8_t, b);
+	uint8_t r_ = a_ - b_;
+
+	a_ = (a_ >> 7) + INT8_MAX;
+
+	if (HEDLEY_STATIC_CAST(int8_t, (a_ ^ b_) & (a_ ^ r_)) < 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int8_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int16_t simde_math_subs_i16(int16_t a, int16_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubh_s16(a, b);
+#else
+	uint16_t a_ = HEDLEY_STATIC_CAST(uint16_t, a);
+	uint16_t b_ = HEDLEY_STATIC_CAST(uint16_t, b);
+	uint16_t r_ = a_ - b_;
+
+	a_ = (a_ >> 15) + INT16_MAX;
+
+	if (HEDLEY_STATIC_CAST(int16_t, (a_ ^ b_) & (a_ ^ r_)) < 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int16_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int32_t simde_math_subs_i32(int32_t a, int32_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubs_s32(a, b);
+#else
+	uint32_t a_ = HEDLEY_STATIC_CAST(uint32_t, a);
+	uint32_t b_ = HEDLEY_STATIC_CAST(uint32_t, b);
+	uint32_t r_ = a_ - b_;
+
+	a_ = (a_ >> 31) + INT32_MAX;
+
+	if (HEDLEY_STATIC_CAST(int32_t, (a_ ^ b_) & (a_ ^ r_)) < 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int32_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE int64_t simde_math_subs_i64(int64_t a, int64_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubd_s64(a, b);
+#else
+	uint64_t a_ = HEDLEY_STATIC_CAST(uint64_t, a);
+	uint64_t b_ = HEDLEY_STATIC_CAST(uint64_t, b);
+	uint64_t r_ = a_ - b_;
+
+	a_ = (a_ >> 63) + INT64_MAX;
+
+	if (HEDLEY_STATIC_CAST(int64_t, (a_ ^ b_) & (a_ ^ r_)) < 0) {
+		r_ = a_;
+	}
+
+	return HEDLEY_STATIC_CAST(int64_t, r_);
+#endif
+}
+
+static HEDLEY_INLINE uint8_t simde_math_subs_u8(uint8_t a, uint8_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubb_u8(a, b);
+#else
+	uint8_t res = a - b;
+	res &= -(res <= a);
+	return res;
+#endif
+}
+
+static HEDLEY_INLINE uint16_t simde_math_subs_u16(uint16_t a, uint16_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubh_u16(a, b);
+#else
+	uint16_t res = a - b;
+	res &= -(res <= a);
+	return res;
+#endif
+}
+
+static HEDLEY_INLINE uint32_t simde_math_subs_u32(uint32_t a, uint32_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubs_u32(a, b);
+#else
+	uint32_t res = a - b;
+	res &= -(res <= a);
+	return res;
+#endif
+}
+
+static HEDLEY_INLINE uint64_t simde_math_subs_u64(uint64_t a, uint64_t b)
+{
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vqsubd_u64(a, b);
+#else
+	uint64_t res = a - b;
+	res &= -(res <= a);
+	return res;
+#endif
+}
+
+HEDLEY_DIAGNOSTIC_POP
+
 #endif /* !defined(SIMDE_MATH_H) */

obs-studio-26.1.1.tar.xz/libobs/util/simde/x86 Added

obs-studio-26.1.1.tar.xz/libobs/util/simde/x86/mmx.h Added

@@ -0,0 +1,2456 @@
+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ * 2017-2020 Evan Nemerson <evan@nemerson.com>
+ */
+
+#if !defined(SIMDE_X86_MMX_H)
+#define SIMDE_X86_MMX_H
+
+#include "../simde-common.h"
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+
+#if defined(SIMDE_X86_MMX_NATIVE)
+#define SIMDE_X86_MMX_USE_NATIVE_TYPE
+#elif defined(SIMDE_X86_SSE_NATIVE)
+#define SIMDE_X86_MMX_USE_NATIVE_TYPE
+#endif
+
+#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
+#include <mmintrin.h>
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#include <arm_neon.h>
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+#include <loongson-mmiintrin.h>
+#endif
+
+#include <stdint.h>
+#include <limits.h>
+
+SIMDE_BEGIN_DECLS_
+
+typedef union {
+#if defined(SIMDE_VECTOR_SUBSCRIPT)
+	SIMDE_ALIGN_TO_8 int8_t i8 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 int16_t i16 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 int32_t i32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 int64_t i64 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 uint8_t u8 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 uint16_t u16 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 uint32_t u32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 uint64_t u64 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 simde_float32 f32 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 int_fast32_t i32f SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_8 uint_fast32_t u32f SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+#else
+	SIMDE_ALIGN_TO_8 int8_t i8[8];
+	SIMDE_ALIGN_TO_8 int16_t i16[4];
+	SIMDE_ALIGN_TO_8 int32_t i32[2];
+	SIMDE_ALIGN_TO_8 int64_t i64[1];
+	SIMDE_ALIGN_TO_8 uint8_t u8[8];
+	SIMDE_ALIGN_TO_8 uint16_t u16[4];
+	SIMDE_ALIGN_TO_8 uint32_t u32[2];
+	SIMDE_ALIGN_TO_8 uint64_t u64[1];
+	SIMDE_ALIGN_TO_8 simde_float32 f32[2];
+	SIMDE_ALIGN_TO_8 int_fast32_t i32f[8 / sizeof(int_fast32_t)];
+	SIMDE_ALIGN_TO_8 uint_fast32_t u32f[8 / sizeof(uint_fast32_t)];
+#endif
+
+#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
+	__m64 n;
+#endif
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int8x8_t neon_i8;
+	int16x4_t neon_i16;
+	int32x2_t neon_i32;
+	int64x1_t neon_i64;
+	uint8x8_t neon_u8;
+	uint16x4_t neon_u16;
+	uint32x2_t neon_u32;
+	uint64x1_t neon_u64;
+	float32x2_t neon_f32;
+#endif
+#if defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	int8x8_t mmi_i8;
+	int16x4_t mmi_i16;
+	int32x2_t mmi_i32;
+	int64_t mmi_i64;
+	uint8x8_t mmi_u8;
+	uint16x4_t mmi_u16;
+	uint32x2_t mmi_u32;
+	uint64_t mmi_u64;
+#endif
+} simde__m64_private;
+
+#if defined(SIMDE_X86_MMX_USE_NATIVE_TYPE)
+typedef __m64 simde__m64;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+typedef int32x2_t simde__m64;
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+typedef int32x2_t simde__m64;
+#elif defined(SIMDE_VECTOR_SUBSCRIPT)
+typedef int32_t simde__m64 SIMDE_ALIGN_TO_8 SIMDE_VECTOR(8) SIMDE_MAY_ALIAS;
+#else
+typedef simde__m64_private simde__m64;
+#endif
+
+#if !defined(SIMDE_X86_MMX_USE_NATIVE_TYPE) && \
+	defined(SIMDE_ENABLE_NATIVE_ALIASES)
+#define SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES
+typedef simde__m64 __m64;
+#endif
+
+HEDLEY_STATIC_ASSERT(8 == sizeof(simde__m64), "__m64 size incorrect");
+HEDLEY_STATIC_ASSERT(8 == sizeof(simde__m64_private), "__m64 size incorrect");
+#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m64) == 8,
+		 "simde__m64 is not 8-byte aligned");
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m64_private) == 8,
+		 "simde__m64_private is not 8-byte aligned");
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde__m64_from_private(simde__m64_private v)
+{
+	simde__m64 r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64_private simde__m64_to_private(simde__m64 v)
+{
+	simde__m64_private r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+#define SIMDE_X86_GENERATE_CONVERSION_FUNCTION(simde_type, source_type, isax, \
+					 fragment) \
+	SIMDE_FUNCTION_ATTRIBUTES \
+	simde__##simde_type simde__##simde_type##_from_##isax##_##fragment( \
+		source_type value) \
+	{ \
+		simde__##simde_type##_private r_; \
+		r_.isax##_##fragment = value; \
+		return simde__##simde_type##_from_private(r_); \
+	} \
+ \
+	SIMDE_FUNCTION_ATTRIBUTES \
+	source_type simde__##simde_type##_to_##isax##_##fragment( \
+		simde__##simde_type value) \
+	{ \
+		simde__##simde_type##_private r_ = \
+			simde__##simde_type##_to_private(value); \
+		return r_.isax##_##fragment; \
+	}
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int8x8_t, neon, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int16x4_t, neon, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int32x2_t, neon, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int64x1_t, neon, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint8x8_t, neon, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint16x4_t, neon, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint32x2_t, neon, u32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint64x1_t, neon, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, float32x2_t, neon, f32)
+#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
+
+#if defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int8x8_t, mmi, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int16x4_t, mmi, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int32x2_t, mmi, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, int64_t, mmi, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint8x8_t, mmi, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint16x4_t, mmi, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint32x2_t, mmi, u32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m64, uint64_t, mmi, u64)
+#endif /* defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE) */
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_add_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_add_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vadd_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = paddb_s(a_.mmi_i8, b_.mmi_i8);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = a_.i8 + b_.i8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = a_.i8[i] + b_.i8[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddb(a, b) simde_mm_add_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_add_pi8(a, b) simde_mm_add_pi8(a, b)
+#define _m_paddb(a, b) simde_m_paddb(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_add_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_add_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vadd_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = paddh_s(a_.mmi_i16, b_.mmi_i16);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = a_.i16 + b_.i16;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] + b_.i16[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddw(a, b) simde_mm_add_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_add_pi16(a, b) simde_mm_add_pi16(a, b)
+#define _m_paddw(a, b) simde_mm_add_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_add_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_add_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vadd_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = paddw_s(a_.mmi_i32, b_.mmi_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 + b_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] + b_.i32[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddd(a, b) simde_mm_add_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_add_pi32(a, b) simde_mm_add_pi32(a, b)
+#define _m_paddd(a, b) simde_mm_add_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_adds_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_adds_pi8(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vqadd_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = paddsb(a_.mmi_i8, b_.mmi_i8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		if ((((b_.i8[i]) > 0) &&
+		 ((a_.i8[i]) > (INT8_MAX - (b_.i8[i]))))) {
+			r_.i8[i] = INT8_MAX;
+		} else if ((((b_.i8[i]) < 0) &&
+			 ((a_.i8[i]) < (INT8_MIN - (b_.i8[i]))))) {
+			r_.i8[i] = INT8_MIN;
+		} else {
+			r_.i8[i] = (a_.i8[i]) + (b_.i8[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddsb(a, b) simde_mm_adds_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_pi8(a, b) simde_mm_adds_pi8(a, b)
+#define _m_paddsb(a, b) simde_mm_adds_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_adds_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_adds_pu8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vqadd_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_u8 = paddusb(a_.mmi_u8, b_.mmi_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		const uint_fast16_t x =
+			HEDLEY_STATIC_CAST(uint_fast16_t, a_.u8[i]) +
+			HEDLEY_STATIC_CAST(uint_fast16_t, b_.u8[i]);
+		if (x > UINT8_MAX)
+			r_.u8[i] = UINT8_MAX;
+		else
+			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddusb(a, b) simde_mm_adds_pu8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_pu8(a, b) simde_mm_adds_pu8(a, b)
+#define _m_paddusb(a, b) simde_mm_adds_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_adds_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_adds_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vqadd_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = paddsh(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if ((((b_.i16[i]) > 0) &&
+		 ((a_.i16[i]) > (INT16_MAX - (b_.i16[i]))))) {
+			r_.i16[i] = INT16_MAX;
+		} else if ((((b_.i16[i]) < 0) &&
+			 ((a_.i16[i]) < (SHRT_MIN - (b_.i16[i]))))) {
+			r_.i16[i] = SHRT_MIN;
+		} else {
+			r_.i16[i] = (a_.i16[i]) + (b_.i16[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddsw(a, b) simde_mm_adds_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_pi16(a, b) simde_mm_adds_pi16(a, b)
+#define _m_paddsw(a, b) simde_mm_adds_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_adds_pu16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_adds_pu16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vqadd_u16(a_.neon_u16, b_.neon_u16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_u16 = paddush(a_.mmi_u16, b_.mmi_u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		const uint32_t x = a_.u16[i] + b_.u16[i];
+		if (x > UINT16_MAX)
+			r_.u16[i] = UINT16_MAX;
+		else
+			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_paddusw(a, b) simde_mm_adds_pu16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_pu16(a, b) simde_mm_adds_pu16(a, b)
+#define _m_paddusw(a, b) simde_mm_adds_pu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_and_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_and_si64(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vand_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 & b_.i64;
+#else
+	r_.i64[0] = a_.i64[0] & b_.i64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pand(a, b) simde_mm_and_si64(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_and_si64(a, b) simde_mm_and_si64(a, b)
+#define _m_pand(a, b) simde_mm_and_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_andnot_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_andnot_si64(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vbic_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = pandn_sw(a_.mmi_i32, b_.mmi_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = ~a_.i32f & b_.i32f;
+#else
+	r_.u64[0] = (~(a_.u64[0])) & (b_.u64[0]);
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pandn(a, b) simde_mm_andnot_si64(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_andnot_si64(a, b) simde_mm_andnot_si64(a, b)
+#define _m_pandn(a, b) simde_mm_andnot_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpeq_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpeq_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vceq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = pcmpeqb_s(a_.mmi_i8, b_.mmi_i8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = (a_.i8[i] == b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpeqb(a, b) simde_mm_cmpeq_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_pi8(a, b) simde_mm_cmpeq_pi8(a, b)
+#define _m_pcmpeqb(a, b) simde_mm_cmpeq_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpeq_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpeq_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vceq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = pcmpeqh_s(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] == b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpeqw(a, b) simde_mm_cmpeq_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_pi16(a, b) simde_mm_cmpeq_pi16(a, b)
+#define _m_pcmpeqw(a, b) simde_mm_cmpeq_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpeq_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpeq_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vceq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = pcmpeqw_s(a_.mmi_i32, b_.mmi_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = (a_.i32[i] == b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpeqd(a, b) simde_mm_cmpeq_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_pi32(a, b) simde_mm_cmpeq_pi32(a, b)
+#define _m_pcmpeqd(a, b) simde_mm_cmpeq_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpgt_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpgt_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vcgt_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = pcmpgtb_s(a_.mmi_i8, b_.mmi_i8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = (a_.i8[i] > b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpgtb(a, b) simde_mm_cmpgt_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_pi8(a, b) simde_mm_cmpgt_pi8(a, b)
+#define _m_pcmpgtb(a, b) simde_mm_cmpgt_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpgt_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpgt_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vcgt_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = pcmpgth_s(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpgtw(a, b) simde_mm_cmpgt_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_pi16(a, b) simde_mm_cmpgt_pi16(a, b)
+#define _m_pcmpgtw(a, b) simde_mm_cmpgt_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cmpgt_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cmpgt_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcgt_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = pcmpgtw_s(a_.mmi_i32, b_.mmi_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = (a_.i32[i] > b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pcmpgtd(a, b) simde_mm_cmpgt_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_pi32(a, b) simde_mm_cmpgt_pi32(a, b)
+#define _m_pcmpgtd(a, b) simde_mm_cmpgt_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvtm64_si64(simde__m64 a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
+	!defined(__PGI)
+	return _mm_cvtm64_si64(a);
+#else
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#pragma clang diagnostic ignored "-Wvector-conversion"
+#endif
+	return vget_lane_s64(a_.neon_i64, 0);
+	HEDLEY_DIAGNOSTIC_POP
+#else
+	return a_.i64[0];
+#endif
+#endif
+}
+#define simde_m_to_int64(a) simde_mm_cvtm64_si64(a)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtm64_si64(a) simde_mm_cvtm64_si64(a)
+#define _m_to_int64(a) simde_mm_cvtm64_si64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtsi32_si64(int32_t a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtsi32_si64(a);
+#else
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int32_t av[sizeof(r_.neon_i32) / sizeof(r_.neon_i32[0])] = {a, 0};
+	r_.neon_i32 = vld1_s32(av);
+#else
+	r_.i32[0] = a;
+	r_.i32[1] = 0;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_from_int(a) simde_mm_cvtsi32_si64(a)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi32_si64(a) simde_mm_cvtsi32_si64(a)
+#define _m_from_int(a) simde_mm_cvtsi32_si64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtsi64_m64(int64_t a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
+	!defined(__PGI)
+	return _mm_cvtsi64_m64(a);
+#else
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vld1_s64(&a);
+#else
+	r_.i64[0] = a;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_from_int64(a) simde_mm_cvtsi64_m64(a)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi64_m64(a) simde_mm_cvtsi64_m64(a)
+#define _m_from_int64(a) simde_mm_cvtsi64_m64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvtsi64_si32(simde__m64 a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtsi64_si32(a);
+#else
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#pragma clang diagnostic ignored "-Wvector-conversion"
+#endif
+	return vget_lane_s32(a_.neon_i32, 0);
+	HEDLEY_DIAGNOSTIC_POP
+#else
+	return a_.i32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi64_si32(a) simde_mm_cvtsi64_si32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_empty(void)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	_mm_empty();
+#else
+	/* noop */
+#endif
+}
+#define simde_m_empty() simde_mm_empty()
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_empty() simde_mm_empty()
+#define _m_empty() simde_mm_empty()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_madd_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_madd_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int32x4_t i1 = vmull_s16(a_.neon_i16, b_.neon_i16);
+	r_.neon_i32 = vpadd_s32(vget_low_s32(i1), vget_high_s32(i1));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = pmaddhw(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i += 2) {
+		r_.i32[i / 2] = (a_.i16[i] * b_.i16[i]) +
+				(a_.i16[i + 1] * b_.i16[i + 1]);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmaddwd(a, b) simde_mm_madd_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_madd_pi16(a, b) simde_mm_madd_pi16(a, b)
+#define _m_pmaddwd(a, b) simde_mm_madd_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_mulhi_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_mulhi_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int32x4_t t1 = vmull_s16(a_.neon_i16, b_.neon_i16);
+	const uint32x4_t t2 = vshrq_n_u32(vreinterpretq_u32_s32(t1), 16);
+	const uint16x4_t t3 = vmovn_u32(t2);
+	r_.neon_u16 = t3;
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = pmulhh(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = HEDLEY_STATIC_CAST(int16_t,
+					 ((a_.i16[i] * b_.i16[i]) >> 16));
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmulhw(a, b) simde_mm_mulhi_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_mulhi_pi16(a, b) simde_mm_mulhi_pi16(a, b)
+#define _m_pmulhw(a, b) simde_mm_mulhi_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_mullo_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_mullo_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int32x4_t t1 = vmull_s16(a_.neon_i16, b_.neon_i16);
+	const uint16x4_t t2 = vmovn_u32(vreinterpretq_u32_s32(t1));
+	r_.neon_u16 = t2;
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = pmullh(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = HEDLEY_STATIC_CAST(
+			int16_t, ((a_.i16[i] * b_.i16[i]) & 0xffff));
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmullw(a, b) simde_mm_mullo_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_mullo_pi16(a, b) simde_mm_mullo_pi16(a, b)
+#define _m_pmullw(a, b) simde_mm_mullo_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_or_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_or_si64(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vorr_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 | b_.i64;
+#else
+	r_.i64[0] = a_.i64[0] | b_.i64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_por(a, b) simde_mm_or_si64(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_or_si64(a, b) simde_mm_or_si64(a, b)
+#define _m_por(a, b) simde_mm_or_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_packs_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_packs_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vqmovn_s16(vcombine_s16(a_.neon_i16, b_.neon_i16));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = packsshb(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if (a_.i16[i] < INT8_MIN) {
+			r_.i8[i] = INT8_MIN;
+		} else if (a_.i16[i] > INT8_MAX) {
+			r_.i8[i] = INT8_MAX;
+		} else {
+			r_.i8[i] = HEDLEY_STATIC_CAST(int8_t, a_.i16[i]);
+		}
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if (b_.i16[i] < INT8_MIN) {
+			r_.i8[i + 4] = INT8_MIN;
+		} else if (b_.i16[i] > INT8_MAX) {
+			r_.i8[i + 4] = INT8_MAX;
+		} else {
+			r_.i8[i + 4] = HEDLEY_STATIC_CAST(int8_t, b_.i16[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_packsswb(a, b) simde_mm_packs_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_packs_pi16(a, b) simde_mm_packs_pi16(a, b)
+#define _m_packsswb(a, b) simde_mm_packs_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_packs_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_packs_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vqmovn_s32(vcombine_s32(a_.neon_i32, b_.neon_i32));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = packsswh(a_.mmi_i32, b_.mmi_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (8 / sizeof(a_.i32[0])); i++) {
+		if (a_.i32[i] < SHRT_MIN) {
+			r_.i16[i] = SHRT_MIN;
+		} else if (a_.i32[i] > INT16_MAX) {
+			r_.i16[i] = INT16_MAX;
+		} else {
+			r_.i16[i] = HEDLEY_STATIC_CAST(int16_t, a_.i32[i]);
+		}
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (8 / sizeof(b_.i32[0])); i++) {
+		if (b_.i32[i] < SHRT_MIN) {
+			r_.i16[i + 2] = SHRT_MIN;
+		} else if (b_.i32[i] > INT16_MAX) {
+			r_.i16[i + 2] = INT16_MAX;
+		} else {
+			r_.i16[i + 2] = HEDLEY_STATIC_CAST(int16_t, b_.i32[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_packssdw(a, b) simde_mm_packs_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_packs_pi32(a, b) simde_mm_packs_pi32(a, b)
+#define _m_packssdw(a, b) simde_mm_packs_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_packs_pu16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_packs_pu16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	const int16x8_t t1 = vcombine_s16(a_.neon_i16, b_.neon_i16);
+
+	/* Set elements which are < 0 to 0 */
+	const int16x8_t t2 =
+		vandq_s16(t1, vreinterpretq_s16_u16(vcgezq_s16(t1)));
+
+	/* Vector with all s16 elements set to UINT8_MAX */
+	const int16x8_t vmax =
+		vmovq_n_s16(HEDLEY_STATIC_CAST(int16_t, UINT8_MAX));
+
+	/* Elements which are within the acceptable range */
+	const int16x8_t le_max =
+		vandq_s16(t2, vreinterpretq_s16_u16(vcleq_s16(t2, vmax)));
+	const int16x8_t gt_max =
+		vandq_s16(vmax, vreinterpretq_s16_u16(vcgtq_s16(t2, vmax)));
+
+	/* Final values as 16-bit integers */
+	const int16x8_t values = vorrq_s16(le_max, gt_max);
+
+	r_.neon_u8 = vmovn_u16(vreinterpretq_u16_s16(values));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_u8 = packushb(a_.mmi_u16, b_.mmi_u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if (a_.i16[i] > UINT8_MAX) {
+			r_.u8[i] = UINT8_MAX;
+		} else if (a_.i16[i] < 0) {
+			r_.u8[i] = 0;
+		} else {
+			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, a_.i16[i]);
+		}
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if (b_.i16[i] > UINT8_MAX) {
+			r_.u8[i + 4] = UINT8_MAX;
+		} else if (b_.i16[i] < 0) {
+			r_.u8[i + 4] = 0;
+		} else {
+			r_.u8[i + 4] = HEDLEY_STATIC_CAST(uint8_t, b_.i16[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_packuswb(a, b) simde_mm_packs_pu16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_packs_pu16(a, b) simde_mm_packs_pu16(a, b)
+#define _m_packuswb(a, b) simde_mm_packs_pu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set_pi8(int8_t e7, int8_t e6, int8_t e5, int8_t e4,
+			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0);
+#else
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int8_t v[sizeof(r_.i8) / sizeof(r_.i8[0])] = {e0, e1, e2, e3,
+							 e4, e5, e6, e7};
+	r_.neon_i8 = vld1_s8(v);
+#else
+	r_.i8[0] = e0;
+	r_.i8[1] = e1;
+	r_.i8[2] = e2;
+	r_.i8[3] = e3;
+	r_.i8[4] = e4;
+	r_.i8[5] = e5;
+	r_.i8[6] = e6;
+	r_.i8[7] = e7;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0) \
+	simde_mm_set_pi8(e7, e6, e5, e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_set_pu8(uint8_t e7, uint8_t e6, uint8_t e5, uint8_t e4,
+			 uint8_t e3, uint8_t e2, uint8_t e1, uint8_t e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_X86_MMX_NATIVE)
+	r_.n = _mm_set_pi8(
+		HEDLEY_STATIC_CAST(int8_t, e7), HEDLEY_STATIC_CAST(int8_t, e6),
+		HEDLEY_STATIC_CAST(int8_t, e5), HEDLEY_STATIC_CAST(int8_t, e4),
+		HEDLEY_STATIC_CAST(int8_t, e3), HEDLEY_STATIC_CAST(int8_t, e2),
+		HEDLEY_STATIC_CAST(int8_t, e1), HEDLEY_STATIC_CAST(int8_t, e0));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint8_t v[sizeof(r_.u8) / sizeof(r_.u8[0])] = {e0, e1, e2, e3,
+							 e4, e5, e6, e7};
+	r_.neon_u8 = vld1_u8(v);
+#else
+	r_.u8[0] = e0;
+	r_.u8[1] = e1;
+	r_.u8[2] = e2;
+	r_.u8[3] = e3;
+	r_.u8[4] = e4;
+	r_.u8[5] = e5;
+	r_.u8[6] = e6;
+	r_.u8[7] = e7;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set_pi16(int16_t e3, int16_t e2, int16_t e1, int16_t e0)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set_pi16(e3, e2, e1, e0);
+#else
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int16_t v[sizeof(r_.i16) / sizeof(r_.i16[0])] = {e0, e1, e2, e3};
+	r_.neon_i16 = vld1_s16(v);
+#else
+	r_.i16[0] = e0;
+	r_.i16[1] = e1;
+	r_.i16[2] = e2;
+	r_.i16[3] = e3;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set_pi16(e3, e2, e1, e0) simde_mm_set_pi16(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_set_pu16(uint16_t e3, uint16_t e2, uint16_t e1,
+			 uint16_t e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_X86_MMX_NATIVE)
+	r_.n = _mm_set_pi16(HEDLEY_STATIC_CAST(int16_t, e3),
+			 HEDLEY_STATIC_CAST(int16_t, e2),
+			 HEDLEY_STATIC_CAST(int16_t, e1),
+			 HEDLEY_STATIC_CAST(int16_t, e0));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint16_t v[sizeof(r_.u16) / sizeof(r_.u16[0])] = {e0, e1, e2, e3};
+	r_.neon_u16 = vld1_u16(v);
+#else
+	r_.u16[0] = e0;
+	r_.u16[1] = e1;
+	r_.u16[2] = e2;
+	r_.u16[3] = e3;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_set_pu32(uint32_t e1, uint32_t e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_X86_MMX_NATIVE)
+	r_.n = _mm_set_pi32(HEDLEY_STATIC_CAST(int32_t, e1),
+			 HEDLEY_STATIC_CAST(int32_t, e0));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint32_t v[sizeof(r_.u32) / sizeof(r_.u32[0])] = {e0, e1};
+	r_.neon_u32 = vld1_u32(v);
+#else
+	r_.u32[0] = e0;
+	r_.u32[1] = e1;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set_pi32(int32_t e1, int32_t e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_X86_MMX_NATIVE)
+	r_.n = _mm_set_pi32(e1, e0);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int32_t v[sizeof(r_.i32) / sizeof(r_.i32[0])] = {e0, e1};
+	r_.neon_i32 = vld1_s32(v);
+#else
+	r_.i32[0] = e0;
+	r_.i32[1] = e1;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set_pi32(e1, e0) simde_mm_set_pi32(e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_set_pi64(int64_t e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int64_t v[sizeof(r_.i64) / sizeof(r_.i64[0])] = {e0};
+	r_.neon_i64 = vld1_s64(v);
+#else
+	r_.i64[0] = e0;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_set_f32x2(simde_float32 e1, simde_float32 e0)
+{
+	simde__m64_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const simde_float32 v[sizeof(r_.f32) / sizeof(r_.f32[0])] = {e0, e1};
+	r_.neon_f32 = vld1_f32(v);
+#else
+	r_.f32[0] = e0;
+	r_.f32[1] = e1;
+#endif
+
+	return simde__m64_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set1_pi8(int8_t a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set1_pi8(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	simde__m64_private r_;
+	r_.neon_i8 = vmov_n_s8(a);
+	return simde__m64_from_private(r_);
+#else
+	return simde_mm_set_pi8(a, a, a, a, a, a, a, a);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_pi8(a) simde_mm_set1_pi8(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set1_pi16(int16_t a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set1_pi16(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	simde__m64_private r_;
+	r_.neon_i16 = vmov_n_s16(a);
+	return simde__m64_from_private(r_);
+#else
+	return simde_mm_set_pi16(a, a, a, a);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_pi16(a) simde_mm_set1_pi16(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_set1_pi32(int32_t a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set1_pi32(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	simde__m64_private r_;
+	r_.neon_i32 = vmov_n_s32(a);
+	return simde__m64_from_private(r_);
+#else
+	return simde_mm_set_pi32(a, a);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_pi32(a) simde_mm_set1_pi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_setr_pi8(int8_t e7, int8_t e6, int8_t e5, int8_t e4,
+			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0);
+#else
+	return simde_mm_set_pi8(e0, e1, e2, e3, e4, e5, e6, e7);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0) \
+	simde_mm_setr_pi8(e7, e6, e5, e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_setr_pi16(int16_t e3, int16_t e2, int16_t e1, int16_t e0)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_setr_pi16(e3, e2, e1, e0);
+#else
+	return simde_mm_set_pi16(e0, e1, e2, e3);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_pi16(e3, e2, e1, e0) simde_mm_setr_pi16(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_setr_pi32(int32_t e1, int32_t e0)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_setr_pi32(e1, e0);
+#else
+	return simde_mm_set_pi32(e0, e1);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_pi32(e1, e0) simde_mm_setr_pi32(e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_setzero_si64(void)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_setzero_si64();
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	simde__m64_private r_;
+	r_.neon_u32 = vmov_n_u32(0);
+	return simde__m64_from_private(r_);
+#else
+	return simde_mm_set_pi32(0, 0);
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_setzero_si64() simde_mm_setzero_si64()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_load_si64(const void *mem_addr)
+{
+	simde__m64 r;
+	simde_memcpy(&r, SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m64),
+		 sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_loadu_si64(const void *mem_addr)
+{
+	simde__m64 r;
+	simde_memcpy(&r, mem_addr, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_x_mm_store_si64(void *mem_addr, simde__m64 value)
+{
+	simde_memcpy(SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m64), &value,
+		 sizeof(value));
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_x_mm_storeu_si64(void *mem_addr, simde__m64 value)
+{
+	simde_memcpy(mem_addr, &value, sizeof(value));
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_x_mm_setone_si64(void)
+{
+	return simde_mm_set1_pi32(~INT32_C(0));
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sll_pi16(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sll_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#pragma clang diagnostic ignored "-Wvector-conversion"
+#endif
+	r_.neon_i16 =
+		vshl_s16(a_.neon_i16,
+			 vmov_n_s16(HEDLEY_STATIC_CAST(
+				 int16_t, vget_lane_u64(count_.neon_u64, 0))));
+	HEDLEY_DIAGNOSTIC_POP
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_BUG_CLANG_POWER9_16x4_BAD_SHIFT)
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 15))
+		return simde_mm_setzero_si64();
+
+	r_.i16 = a_.i16 << HEDLEY_STATIC_CAST(int16_t, count_.u64[0]);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i16 = a_.i16 << count_.u64[0];
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 15)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t,
+					 a_.u16[i] << count_.u64[0]);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psllw(a, count) simde_mm_sll_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_pi16(a, count) simde_mm_sll_pi16(a, count)
+#define _m_psllw(a, count) simde_mm_sll_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sll_pi32(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sll_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#pragma clang diagnostic ignored "-Wvector-conversion"
+#endif
+	r_.neon_i32 =
+		vshl_s32(a_.neon_i32,
+			 vmov_n_s32(HEDLEY_STATIC_CAST(
+				 int32_t, vget_lane_u64(count_.neon_u64, 0))));
+	HEDLEY_DIAGNOSTIC_POP
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i32 = a_.i32 << count_.u64[0];
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 31)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] << count_.u64[0];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pslld(a, count) simde_mm_sll_pi32(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_pi32(a, count) simde_mm_sll_pi32(a, count)
+#define _m_pslld(a, count) simde_mm_sll_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_slli_pi16(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_slli_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_BUG_CLANG_POWER9_16x4_BAD_SHIFT)
+	if (HEDLEY_UNLIKELY(count > 15))
+		return simde_mm_setzero_si64();
+
+	r_.i16 = a_.i16 << HEDLEY_STATIC_CAST(int16_t, count);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i16 = a_.i16 << count;
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vshl_s16(a_.neon_i16, vmov_n_s16((int16_t)count));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = psllh_s(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, a_.u16[i] << count);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psllwi(a, count) simde_mm_slli_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_pi16(a, count) simde_mm_slli_pi16(a, count)
+#define _m_psllwi(a, count) simde_mm_slli_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_slli_pi32(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_slli_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i32 = a_.i32 << count;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vshl_s32(a_.neon_i32, vmov_n_s32((int32_t)count));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = psllw_s(a_.mmi_i32, b_.mmi_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] << count;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pslldi(a, b) simde_mm_slli_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_pi32(a, count) simde_mm_slli_pi32(a, count)
+#define _m_pslldi(a, count) simde_mm_slli_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_slli_si64(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_slli_si64(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i64 = a_.i64 << count;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vshl_s64(a_.neon_i64, vmov_n_s64((int64_t)count));
+#else
+	r_.u64[0] = a_.u64[0] << count;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psllqi(a, count) simde_mm_slli_si64(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_si64(a, count) simde_mm_slli_si64(a, count)
+#define _m_psllqi(a, count) simde_mm_slli_si64(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sll_si64(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sll_si64(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vshl_s64(a_.neon_i64, count_.neon_i64);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 << count_.i64;
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 63)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	r_.u64[0] = a_.u64[0] << count_.u64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psllq(a, count) simde_mm_sll_si64(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_si64(a, count) simde_mm_sll_si64(a, count)
+#define _m_psllq(a, count) simde_mm_sll_si64(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srl_pi16(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_srl_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_BUG_CLANG_POWER9_16x4_BAD_SHIFT)
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 15))
+		return simde_mm_setzero_si64();
+
+	r_.i16 = a_.i16 >> HEDLEY_STATIC_CAST(int16_t, count_.u64[0]);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u16 = a_.u16 >> count_.u64[0];
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vshl_u16(
+		a_.neon_u16,
+		vmov_n_s16(-((int16_t)vget_lane_u64(count_.neon_u64, 0))));
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 15)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < sizeof(r_.u16) / sizeof(r_.u16[0]); i++) {
+		r_.u16[i] = a_.u16[i] >> count_.u64[0];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrlw(a, count) simde_mm_srl_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_pi16(a, count) simde_mm_srl_pi16(a, count)
+#define _m_psrlw(a, count) simde_mm_srl_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srl_pi32(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_srl_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u32 = a_.u32 >> count_.u64[0];
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vshl_u32(
+		a_.neon_u32,
+		vmov_n_s32(-((int32_t)vget_lane_u64(count_.neon_u64, 0))));
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 31)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < sizeof(r_.u32) / sizeof(r_.u32[0]); i++) {
+		r_.u32[i] = a_.u32[i] >> count_.u64[0];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrld(a, count) simde_mm_srl_pi32(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_pi32(a, count) simde_mm_srl_pi32(a, count)
+#define _m_psrld(a, count) simde_mm_srl_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srli_pi16(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_srli_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u16 = a_.u16 >> count;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vshl_u16(a_.neon_u16, vmov_n_s16(-((int16_t)count)));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = psrlh_s(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = a_.u16[i] >> count;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrlwi(a, count) simde_mm_srli_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_pi16(a, count) simde_mm_srli_pi16(a, count)
+#define _m_psrlwi(a, count) simde_mm_srli_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srli_pi32(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_srli_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u32 = a_.u32 >> count;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vshl_u32(a_.neon_u32, vmov_n_s32(-((int32_t)count)));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = psrlw_s(a_.mmi_i32, b_.mmi_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] >> count;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrldi(a, count) simde_mm_srli_pi32(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_pi32(a, count) simde_mm_srli_pi32(a, count)
+#define _m_psrldi(a, count) simde_mm_srli_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srli_si64(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_srli_si64(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u64 = vshl_u64(a_.neon_u64, vmov_n_s64(-count));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u64 = a_.u64 >> count;
+#else
+	r_.u64[0] = a_.u64[0] >> count;
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrlqi(a, count) simde_mm_srli_si64(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_si64(a, count) simde_mm_srli_si64(a, count)
+#define _m_psrlqi(a, count) simde_mm_srli_si64(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srl_si64(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_srl_si64(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vshl_u64(a_.neon_u64, vneg_s64(count_.neon_i64));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.u64 = a_.u64 >> count_.u64;
+#else
+	if (HEDLEY_UNLIKELY(count_.u64[0] > 63)) {
+		simde_memset(&r_, 0, sizeof(r_));
+		return simde__m64_from_private(r_);
+	}
+
+	r_.u64[0] = a_.u64[0] >> count_.u64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrlq(a, count) simde_mm_srl_si64(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_si64(a, count) simde_mm_srl_si64(a, count)
+#define _m_psrlq(a, count) simde_mm_srl_si64(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srai_pi16(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_srai_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i16 = a_.i16 >> (count & 0xff);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vshl_s16(a_.neon_i16,
+			 vmov_n_s16(-HEDLEY_STATIC_CAST(int16_t, count)));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = psrah_s(a_.mmi_i16, count);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] >> (count & 0xff);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrawi(a, count) simde_mm_srai_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srai_pi16(a, count) simde_mm_srai_pi16(a, count)
+#define _m_psrawi(a, count) simde_mm_srai_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_srai_pi32(simde__m64 a, int count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE) && !defined(__PGI)
+	return _mm_srai_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i32 = a_.i32 >> (count & 0xff);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vshl_s32(a_.neon_i32,
+			 vmov_n_s32(-HEDLEY_STATIC_CAST(int32_t, count)));
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = psraw_s(a_.mmi_i32, count);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] >> (count & 0xff);
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psradi(a, count) simde_mm_srai_pi32(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_srai_pi32(a, count) simde_mm_srai_pi32(a, count)
+#define _m_psradi(a, count) simde_mm_srai_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sra_pi16(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sra_pi16(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+	const int cnt = HEDLEY_STATIC_CAST(
+		int, (count_.i64[0] > 15 ? 15 : count_.i64[0]));
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i16 = a_.i16 >> cnt;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 =
+		vshl_s16(a_.neon_i16,
+			 vmov_n_s16(-HEDLEY_STATIC_CAST(
+				 int16_t, vget_lane_u64(count_.neon_u64, 0))));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] >> cnt;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psraw(a, count) simde_mm_sra_pi16(a, count)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sra_pi16(a, count) simde_mm_sra_pi16(a, count)
+#define _m_psraw(a, count) simde_mm_sra_pi16(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sra_pi32(simde__m64 a, simde__m64 count)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sra_pi32(a, count);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private count_ = simde__m64_to_private(count);
+	const int32_t cnt =
+		(count_.u64[0] > 31)
+			? 31
+			: HEDLEY_STATIC_CAST(int32_t, count_.u64[0]);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i32 = a_.i32 >> cnt;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 =
+		vshl_s32(a_.neon_i32,
+			 vmov_n_s32(-HEDLEY_STATIC_CAST(
+				 int32_t, vget_lane_u64(count_.neon_u64, 0))));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] >> cnt;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psrad(a, b) simde_mm_sra_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sra_pi32(a, count) simde_mm_sra_pi32(a, count)
+#define _m_psrad(a, count) simde_mm_sra_pi32(a, count)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sub_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sub_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vsub_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = psubb_s(a_.mmi_i8, b_.mmi_i8);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = a_.i8 - b_.i8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = a_.i8[i] - b_.i8[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubb(a, b) simde_mm_sub_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_pi8(a, b) simde_mm_sub_pi8(a, b)
+#define _m_psubb(a, b) simde_mm_sub_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sub_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sub_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vsub_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = psubh_s(a_.mmi_i16, b_.mmi_i16);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = a_.i16 - b_.i16;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] - b_.i16[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubw(a, b) simde_mm_sub_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_pi16(a, b) simde_mm_sub_pi16(a, b)
+#define _m_psubw(a, b) simde_mm_sub_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sub_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sub_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vsub_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = psubw_s(a_.mmi_i32, b_.mmi_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 - b_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] - b_.i32[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubd(a, b) simde_mm_sub_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_pi32(a, b) simde_mm_sub_pi32(a, b)
+#define _m_psubd(a, b) simde_mm_sub_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_subs_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_subs_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vqsub_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = psubsb(a_.mmi_i8, b_.mmi_i8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		if (((b_.i8[i]) > 0 && (a_.i8[i]) < INT8_MIN + (b_.i8[i]))) {
+			r_.i8[i] = INT8_MIN;
+		} else if ((b_.i8[i]) < 0 &&
+			 (a_.i8[i]) > INT8_MAX + (b_.i8[i])) {
+			r_.i8[i] = INT8_MAX;
+		} else {
+			r_.i8[i] = (a_.i8[i]) - (b_.i8[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubsb(a, b) simde_mm_subs_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_pi8(a, b) simde_mm_subs_pi8(a, b)
+#define _m_psubsb(a, b) simde_mm_subs_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_subs_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_subs_pu8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vqsub_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_u8 = psubusb(a_.mmi_u8, b_.mmi_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		const int32_t x = a_.u8[i] - b_.u8[i];
+		if (x < 0) {
+			r_.u8[i] = 0;
+		} else if (x > UINT8_MAX) {
+			r_.u8[i] = UINT8_MAX;
+		} else {
+			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubusb(a, b) simde_mm_subs_pu8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_pu8(a, b) simde_mm_subs_pu8(a, b)
+#define _m_psubusb(a, b) simde_mm_subs_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_subs_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_subs_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vqsub_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = psubsh(a_.mmi_i16, b_.mmi_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		if (((b_.i16[i]) > 0 && (a_.i16[i]) < SHRT_MIN + (b_.i16[i]))) {
+			r_.i16[i] = SHRT_MIN;
+		} else if ((b_.i16[i]) < 0 &&
+			 (a_.i16[i]) > INT16_MAX + (b_.i16[i])) {
+			r_.i16[i] = INT16_MAX;
+		} else {
+			r_.i16[i] = (a_.i16[i]) - (b_.i16[i]);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubsw(a, b) simde_mm_subs_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_pi16(a, b) simde_mm_subs_pi16(a, b)
+#define _m_psubsw(a, b) simde_mm_subs_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_subs_pu16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_subs_pu16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vqsub_u16(a_.neon_u16, b_.neon_u16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_u16 = psubush(a_.mmi_u16, b_.mmi_u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		const int x = a_.u16[i] - b_.u16[i];
+		if (x < 0) {
+			r_.u16[i] = 0;
+		} else if (x > UINT16_MAX) {
+			r_.u16[i] = UINT16_MAX;
+		} else {
+			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
+		}
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psubusw(a, b) simde_mm_subs_pu16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_pu16(a, b) simde_mm_subs_pu16(a, b)
+#define _m_psubusw(a, b) simde_mm_subs_pu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpackhi_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpackhi_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i8 = vzip2_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 8, a_.i8, b_.i8, 4, 12, 5, 13, 6, 14,
+				 7, 15);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = punpckhbh_s(a_.mmi_i8, b_.mmi_i8);
+#else
+	r_.i8[0] = a_.i8[4];
+	r_.i8[1] = b_.i8[4];
+	r_.i8[2] = a_.i8[5];
+	r_.i8[3] = b_.i8[5];
+	r_.i8[4] = a_.i8[6];
+	r_.i8[5] = b_.i8[6];
+	r_.i8[6] = a_.i8[7];
+	r_.i8[7] = b_.i8[7];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpckhbw(a, b) simde_mm_unpackhi_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_pi8(a, b) simde_mm_unpackhi_pi8(a, b)
+#define _m_punpckhbw(a, b) simde_mm_unpackhi_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpackhi_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpackhi_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i16 = vzip2_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = punpckhhw_s(a_.mmi_i16, b_.mmi_i16);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 8, a_.i16, b_.i16, 2, 6, 3, 7);
+#else
+	r_.i16[0] = a_.i16[2];
+	r_.i16[1] = b_.i16[2];
+	r_.i16[2] = a_.i16[3];
+	r_.i16[3] = b_.i16[3];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpckhwd(a, b) simde_mm_unpackhi_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_pi16(a, b) simde_mm_unpackhi_pi16(a, b)
+#define _m_punpckhwd(a, b) simde_mm_unpackhi_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpackhi_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpackhi_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i32 = vzip2_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = punpckhwd_s(a_.mmi_i32, b_.mmi_i32);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 8, a_.i32, b_.i32, 1, 3);
+#else
+	r_.i32[0] = a_.i32[1];
+	r_.i32[1] = b_.i32[1];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpckhdq(a, b) simde_mm_unpackhi_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_pi32(a, b) simde_mm_unpackhi_pi32(a, b)
+#define _m_punpckhdq(a, b) simde_mm_unpackhi_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpacklo_pi8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpacklo_pi8(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i8 = vzip1_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i8 = punpcklbh_s(a_.mmi_i8, b_.mmi_i8);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 8, a_.i8, b_.i8, 0, 8, 1, 9, 2, 10, 3,
+				 11);
+#else
+	r_.i8[0] = a_.i8[0];
+	r_.i8[1] = b_.i8[0];
+	r_.i8[2] = a_.i8[1];
+	r_.i8[3] = b_.i8[1];
+	r_.i8[4] = a_.i8[2];
+	r_.i8[5] = b_.i8[2];
+	r_.i8[6] = a_.i8[3];
+	r_.i8[7] = b_.i8[3];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpcklbw(a, b) simde_mm_unpacklo_pi8(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_pi8(a, b) simde_mm_unpacklo_pi8(a, b)
+#define _m_punpcklbw(a, b) simde_mm_unpacklo_pi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpacklo_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpacklo_pi16(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i16 = vzip1_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i16 = punpcklhw_s(a_.mmi_i16, b_.mmi_i16);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 8, a_.i16, b_.i16, 0, 4, 1, 5);
+#else
+	r_.i16[0] = a_.i16[0];
+	r_.i16[1] = b_.i16[0];
+	r_.i16[2] = a_.i16[1];
+	r_.i16[3] = b_.i16[1];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpcklwd(a, b) simde_mm_unpacklo_pi16(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_pi16(a, b) simde_mm_unpacklo_pi16(a, b)
+#define _m_punpcklwd(a, b) simde_mm_unpacklo_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_unpacklo_pi32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_unpacklo_pi32(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i32 = vzip1_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_MIPS_LOONGSON_MMI_NATIVE)
+	r_.mmi_i32 = punpcklwd_s(a_.mmi_i32, b_.mmi_i32);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 8, a_.i32, b_.i32, 0, 2);
+#else
+	r_.i32[0] = a_.i32[0];
+	r_.i32[1] = b_.i32[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_punpckldq(a, b) simde_mm_unpacklo_pi32(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_pi32(a, b) simde_mm_unpacklo_pi32(a, b)
+#define _m_punpckldq(a, b) simde_mm_unpacklo_pi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_xor_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_xor_si64(a, b);
+#else
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = veor_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f ^ b_.i32f;
+#else
+	r_.u64[0] = a_.u64[0] ^ b_.u64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pxor(a, b) simde_mm_xor_si64(a, b)
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _mm_xor_si64(a, b) simde_mm_xor_si64(a, b)
+#define _m_pxor(a, b) simde_mm_xor_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_m_to_int(simde__m64 a)
+{
+#if defined(SIMDE_X86_MMX_NATIVE)
+	return _m_to_int(a);
+#else
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wvector-conversion") && \
+	SIMDE_DETECT_CLANG_VERSION_NOT(10, 0, 0)
+#pragma clang diagnostic ignored "-Wvector-conversion"
+#endif
+	return vget_lane_s32(a_.neon_i32, 0);
+	HEDLEY_DIAGNOSTIC_POP
+#else
+	return a_.i32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_MMX_ENABLE_NATIVE_ALIASES)
+#define _m_to_int(a) simde_m_to_int(a)
+#endif
+
+SIMDE_END_DECLS_
+
+HEDLEY_DIAGNOSTIC_POP
+
+#endif /* !defined(SIMDE_X86_MMX_H) */

obs-studio-26.1.1.tar.xz/libobs/util/simde/x86/sse.h Added

@@ -0,0 +1,4479 @@
+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ * 2017-2020 Evan Nemerson <evan@nemerson.com>
+ * 2015-2017 John W. Ratcliff <jratcliffscarab@gmail.com>
+ * 2015 Brandon Rowlett <browlett@nvidia.com>
+ * 2015 Ken Fast <kfast@gdeb.com>
+ */
+
+#if !defined(SIMDE_X86_SSE_H)
+#define SIMDE_X86_SSE_H
+
+#include "mmx.h"
+
+#if defined(_WIN32)
+#include <windows.h>
+#endif
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+SIMDE_BEGIN_DECLS_
+
+typedef union {
+#if defined(SIMDE_VECTOR_SUBSCRIPT)
+	SIMDE_ALIGN_TO_16 int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#if defined(SIMDE_HAVE_INT128_)
+	SIMDE_ALIGN_TO_16 simde_int128 i128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 simde_uint128 u128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#endif
+	SIMDE_ALIGN_TO_16 simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+	SIMDE_ALIGN_TO_16 int8_t i8[16];
+	SIMDE_ALIGN_TO_16 int16_t i16[8];
+	SIMDE_ALIGN_TO_16 int32_t i32[4];
+	SIMDE_ALIGN_TO_16 int64_t i64[2];
+	SIMDE_ALIGN_TO_16 uint8_t u8[16];
+	SIMDE_ALIGN_TO_16 uint16_t u16[8];
+	SIMDE_ALIGN_TO_16 uint32_t u32[4];
+	SIMDE_ALIGN_TO_16 uint64_t u64[2];
+#if defined(SIMDE_HAVE_INT128_)
+	SIMDE_ALIGN_TO_16 simde_int128 i128[1];
+	SIMDE_ALIGN_TO_16 simde_uint128 u128[1];
+#endif
+	SIMDE_ALIGN_TO_16 simde_float32 f32[4];
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f[16 / sizeof(int_fast32_t)];
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
+#endif
+
+	SIMDE_ALIGN_TO_16 simde__m64_private m64_private[2];
+	SIMDE_ALIGN_TO_16 simde__m64 m64[2];
+
+#if defined(SIMDE_X86_SSE_NATIVE)
+	SIMDE_ALIGN_TO_16 __m128 n;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_TO_16 int8x16_t neon_i8;
+	SIMDE_ALIGN_TO_16 int16x8_t neon_i16;
+	SIMDE_ALIGN_TO_16 int32x4_t neon_i32;
+	SIMDE_ALIGN_TO_16 int64x2_t neon_i64;
+	SIMDE_ALIGN_TO_16 uint8x16_t neon_u8;
+	SIMDE_ALIGN_TO_16 uint16x8_t neon_u16;
+	SIMDE_ALIGN_TO_16 uint32x4_t neon_u32;
+	SIMDE_ALIGN_TO_16 uint64x2_t neon_u64;
+	SIMDE_ALIGN_TO_16 float32x4_t neon_f32;
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	SIMDE_ALIGN_TO_16 float64x2_t neon_f64;
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	SIMDE_ALIGN_TO_16 v128_t wasm_v128;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
+#endif
+#endif
+} simde__m128_private;
+
+#if defined(SIMDE_X86_SSE_NATIVE)
+typedef __m128 simde__m128;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+typedef float32x4_t simde__m128;
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+typedef v128_t simde__m128;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128;
+#elif defined(SIMDE_VECTOR_SUBSCRIPT)
+typedef simde_float32
+	simde__m128 SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+typedef simde__m128_private simde__m128;
+#endif
+
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+typedef simde__m128 __m128;
+#endif
+
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128), "simde__m128 size incorrect");
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128_private),
+		 "simde__m128_private size incorrect");
+#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128) == 16,
+		 "simde__m128 is not 16-byte aligned");
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128_private) == 16,
+		 "simde__m128_private is not 16-byte aligned");
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde__m128_from_private(simde__m128_private v)
+{
+	simde__m128 r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128_private simde__m128_to_private(simde__m128 v)
+{
+	simde__m128_private r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int8x16_t, neon, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int16x8_t, neon, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int32x4_t, neon, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, int64x2_t, neon, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint8x16_t, neon, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint16x8_t, neon, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint32x4_t, neon, u32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, uint64x2_t, neon, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float32x4_t, neon, f32)
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, float64x2_t, neon, f64)
+#endif
+#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+				 altivec, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+				 altivec, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+				 altivec, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned char), altivec, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned short), altivec, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128,
+				 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int),
+				 altivec, u32)
+
+#if defined(SIMDE_BUG_GCC_95782)
+SIMDE_FUNCTION_ATTRIBUTES
+SIMDE_POWER_ALTIVEC_VECTOR(float)
+simde__m128_to_altivec_f32(simde__m128 value)
+{
+	simde__m128_private r_ = simde__m128_to_private(value);
+	return r_.altivec_f32;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde__m128_from_altivec_f32(SIMDE_POWER_ALTIVEC_VECTOR(float)
+						 value)
+{
+	simde__m128_private r_;
+	r_.altivec_f32 = value;
+	return simde__m128_from_private(r_);
+}
+#else
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, SIMDE_POWER_ALTIVEC_VECTOR(float),
+				 altivec, f32)
+#endif
+
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128, SIMDE_POWER_ALTIVEC_VECTOR(signed long long), altivec, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128, SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long), altivec, u64)
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128, v128_t, wasm, v128);
+#endif /* defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) */
+
+enum {
+#if defined(SIMDE_X86_SSE_NATIVE)
+	SIMDE_MM_ROUND_NEAREST = _MM_ROUND_NEAREST,
+	SIMDE_MM_ROUND_DOWN = _MM_ROUND_DOWN,
+	SIMDE_MM_ROUND_UP = _MM_ROUND_UP,
+	SIMDE_MM_ROUND_TOWARD_ZERO = _MM_ROUND_TOWARD_ZERO
+#else
+	SIMDE_MM_ROUND_NEAREST = 0x0000,
+	SIMDE_MM_ROUND_DOWN = 0x2000,
+	SIMDE_MM_ROUND_UP = 0x4000,
+	SIMDE_MM_ROUND_TOWARD_ZERO = 0x6000
+#endif
+};
+
+#if defined(_MM_FROUND_TO_NEAREST_INT)
+#define SIMDE_MM_FROUND_TO_NEAREST_INT _MM_FROUND_TO_NEAREST_INT
+#define SIMDE_MM_FROUND_TO_NEG_INF _MM_FROUND_TO_NEG_INF
+#define SIMDE_MM_FROUND_TO_POS_INF _MM_FROUND_TO_POS_INF
+#define SIMDE_MM_FROUND_TO_ZERO _MM_FROUND_TO_ZERO
+#define SIMDE_MM_FROUND_CUR_DIRECTION _MM_FROUND_CUR_DIRECTION
+
+#define SIMDE_MM_FROUND_RAISE_EXC _MM_FROUND_RAISE_EXC
+#define SIMDE_MM_FROUND_NO_EXC _MM_FROUND_NO_EXC
+#else
+#define SIMDE_MM_FROUND_TO_NEAREST_INT 0x00
+#define SIMDE_MM_FROUND_TO_NEG_INF 0x01
+#define SIMDE_MM_FROUND_TO_POS_INF 0x02
+#define SIMDE_MM_FROUND_TO_ZERO 0x03
+#define SIMDE_MM_FROUND_CUR_DIRECTION 0x04
+
+#define SIMDE_MM_FROUND_RAISE_EXC 0x00
+#define SIMDE_MM_FROUND_NO_EXC 0x08
+#endif
+
+#define SIMDE_MM_FROUND_NINT \
+	(SIMDE_MM_FROUND_TO_NEAREST_INT | SIMDE_MM_FROUND_RAISE_EXC)
+#define SIMDE_MM_FROUND_FLOOR \
+	(SIMDE_MM_FROUND_TO_NEG_INF | SIMDE_MM_FROUND_RAISE_EXC)
+#define SIMDE_MM_FROUND_CEIL \
+	(SIMDE_MM_FROUND_TO_POS_INF | SIMDE_MM_FROUND_RAISE_EXC)
+#define SIMDE_MM_FROUND_TRUNC \
+	(SIMDE_MM_FROUND_TO_ZERO | SIMDE_MM_FROUND_RAISE_EXC)
+#define SIMDE_MM_FROUND_RINT \
+	(SIMDE_MM_FROUND_CUR_DIRECTION | SIMDE_MM_FROUND_RAISE_EXC)
+#define SIMDE_MM_FROUND_NEARBYINT \
+	(SIMDE_MM_FROUND_CUR_DIRECTION | SIMDE_MM_FROUND_NO_EXC)
+
+#if defined(SIMDE_X86_SSE4_1_ENABLE_NATIVE_ALIASES) && \
+	!defined(_MM_FROUND_TO_NEAREST_INT)
+#define _MM_FROUND_TO_NEAREST_INT SIMDE_MM_FROUND_TO_NEAREST_INT
+#define _MM_FROUND_TO_NEG_INF SIMDE_MM_FROUND_TO_NEG_INF
+#define _MM_FROUND_TO_POS_INF SIMDE_MM_FROUND_TO_POS_INF
+#define _MM_FROUND_TO_ZERO SIMDE_MM_FROUND_TO_ZERO
+#define _MM_FROUND_CUR_DIRECTION SIMDE_MM_FROUND_CUR_DIRECTION
+#define _MM_FROUND_RAISE_EXC SIMDE_MM_FROUND_RAISE_EXC
+#define _MM_FROUND_NINT SIMDE_MM_FROUND_NINT
+#define _MM_FROUND_FLOOR SIMDE_MM_FROUND_FLOOR
+#define _MM_FROUND_CEIL SIMDE_MM_FROUND_CEIL
+#define _MM_FROUND_TRUNC SIMDE_MM_FROUND_TRUNC
+#define _MM_FROUND_RINT SIMDE_MM_FROUND_RINT
+#define _MM_FROUND_NEARBYINT SIMDE_MM_FROUND_NEARBYINT
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+unsigned int SIMDE_MM_GET_ROUNDING_MODE(void)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _MM_GET_ROUNDING_MODE();
+#elif defined(SIMDE_HAVE_FENV_H)
+	unsigned int vfe_mode;
+
+	switch (fegetround()) {
+#if defined(FE_TONEAREST)
+	case FE_TONEAREST:
+		vfe_mode = SIMDE_MM_ROUND_NEAREST;
+		break;
+#endif
+
+#if defined(FE_TOWARDZERO)
+	case FE_TOWARDZERO:
+		vfe_mode = SIMDE_MM_ROUND_DOWN;
+		break;
+#endif
+
+#if defined(FE_UPWARD)
+	case FE_UPWARD:
+		vfe_mode = SIMDE_MM_ROUND_UP;
+		break;
+#endif
+
+#if defined(FE_DOWNWARD)
+	case FE_DOWNWARD:
+		vfe_mode = SIMDE_MM_ROUND_TOWARD_ZERO;
+		break;
+#endif
+
+	default:
+		vfe_mode = SIMDE_MM_ROUND_NEAREST;
+		break;
+	}
+
+	return vfe_mode;
+#else
+	return SIMDE_MM_ROUND_NEAREST;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _MM_GET_ROUNDING_MODE() SIMDE_MM_GET_ROUNDING_MODE()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void SIMDE_MM_SET_ROUNDING_MODE(unsigned int a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_MM_SET_ROUNDING_MODE(a);
+#elif defined(SIMDE_HAVE_FENV_H)
+	int fe_mode = FE_TONEAREST;
+
+	switch (a) {
+#if defined(FE_TONEAREST)
+	case SIMDE_MM_ROUND_NEAREST:
+		fe_mode = FE_TONEAREST;
+		break;
+#endif
+
+#if defined(FE_TOWARDZERO)
+	case SIMDE_MM_ROUND_TOWARD_ZERO:
+		fe_mode = FE_TOWARDZERO;
+		break;
+#endif
+
+#if defined(FE_DOWNWARD)
+	case SIMDE_MM_ROUND_DOWN:
+		fe_mode = FE_DOWNWARD;
+		break;
+#endif
+
+#if defined(FE_UPWARD)
+	case SIMDE_MM_ROUND_UP:
+		fe_mode = FE_UPWARD;
+		break;
+#endif
+
+	default:
+		return;
+	}
+
+	fesetround(fe_mode);
+#else
+	(void)a;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _MM_SET_ROUNDING_MODE(a) SIMDE_MM_SET_ROUNDING_MODE(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+uint32_t simde_mm_getcsr(void)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_getcsr();
+#else
+	return SIMDE_MM_GET_ROUNDING_MODE();
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_getcsr() simde_mm_getcsr()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_setcsr(uint32_t a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_setcsr(a);
+#else
+	SIMDE_MM_SET_ROUNDING_MODE(HEDLEY_STATIC_CAST(unsigned int, a));
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_setcsr(a) simde_mm_setcsr(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_round_ps(simde__m128 a, int rounding, int lax_rounding)
+	SIMDE_REQUIRE_CONSTANT_RANGE(rounding, 0, 15)
+		SIMDE_REQUIRE_CONSTANT_RANGE(lax_rounding, 0, 1)
+{
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+	(void)lax_rounding;
+
+/* For architectures which lack a current direction SIMD instruction.
+ *
+ * Note that NEON actually has a current rounding mode instruction,
+ * but in ARMv8+ the rounding mode is ignored and nearest is always
+ * used, so we treat ARMv7 as having a rounding mode but ARMv8 as
+ * not. */
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) || defined(SIMDE_ARM_NEON_A32V8)
+	if ((rounding & 7) == SIMDE_MM_FROUND_CUR_DIRECTION)
+		rounding = HEDLEY_STATIC_CAST(int, SIMDE_MM_GET_ROUNDING_MODE())
+			 << 13;
+#endif
+
+	switch (rounding & ~SIMDE_MM_FROUND_NO_EXC) {
+	case SIMDE_MM_FROUND_CUR_DIRECTION:
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+		r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(float),
+			vec_round(a_.altivec_f32));
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
+		r_.neon_f32 = vrndiq_f32(a_.neon_f32);
+#elif defined(simde_math_nearbyintf)
+		SIMDE_VECTORIZE
+		for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0]));
+		 i++) {
+			r_.f32[i] = simde_math_nearbyintf(a_.f32[i]);
+		}
+#else
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+#endif
+		break;
+
+	case SIMDE_MM_FROUND_TO_NEAREST_INT:
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+		r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(float),
+			vec_rint(a_.altivec_f32));
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+		r_.neon_f32 = vrndnq_f32(a_.neon_f32);
+#elif defined(simde_math_roundevenf)
+		SIMDE_VECTORIZE
+		for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0]));
+		 i++) {
+			r_.f32[i] = simde_math_roundevenf(a_.f32[i]);
+		}
+#else
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+#endif
+		break;
+
+	case SIMDE_MM_FROUND_TO_NEG_INF:
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+		r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(float),
+			vec_floor(a_.altivec_f32));
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+		r_.neon_f32 = vrndmq_f32(a_.neon_f32);
+#elif defined(simde_math_floorf)
+		SIMDE_VECTORIZE
+		for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0]));
+		 i++) {
+			r_.f32[i] = simde_math_floorf(a_.f32[i]);
+		}
+#else
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+#endif
+		break;
+
+	case SIMDE_MM_FROUND_TO_POS_INF:
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+		r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(float),
+			vec_ceil(a_.altivec_f32));
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+		r_.neon_f32 = vrndpq_f32(a_.neon_f32);
+#elif defined(simde_math_ceilf)
+		SIMDE_VECTORIZE
+		for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0]));
+		 i++) {
+			r_.f32[i] = simde_math_ceilf(a_.f32[i]);
+		}
+#else
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+#endif
+		break;
+
+	case SIMDE_MM_FROUND_TO_ZERO:
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+		r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(float),
+			vec_trunc(a_.altivec_f32));
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+		r_.neon_f32 = vrndq_f32(a_.neon_f32);
+#elif defined(simde_math_truncf)
+		SIMDE_VECTORIZE
+		for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0]));
+		 i++) {
+			r_.f32[i] = simde_math_truncf(a_.f32[i]);
+		}
+#else
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+#endif
+		break;
+
+	default:
+		HEDLEY_UNREACHABLE_RETURN(simde_mm_undefined_pd());
+	}
+
+	return simde__m128_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE4_1_NATIVE)
+#define simde_mm_round_ps(a, rounding) _mm_round_ps((a), (rounding))
+#else
+#define simde_mm_round_ps(a, rounding) simde_x_mm_round_ps((a), (rounding), 0)
+#endif
+#if defined(SIMDE_X86_SSE4_1_ENABLE_NATIVE_ALIASES)
+#define _mm_round_ps(a, rounding) simde_mm_round_ps((a), (rounding))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_set_ps(simde_float32 e3, simde_float32 e2,
+			 simde_float32 e1, simde_float32 e0)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_set_ps(e3, e2, e1, e0);
+#else
+	simde__m128_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_TO_16 simde_float32 data[4] = {e0, e1, e2, e3};
+	r_.neon_f32 = vld1q_f32(data);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_make(e0, e1, e2, e3);
+#else
+	r_.f32[0] = e0;
+	r_.f32[1] = e1;
+	r_.f32[2] = e2;
+	r_.f32[3] = e3;
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_set_ps(e3, e2, e1, e0) simde_mm_set_ps(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_set_ps1(simde_float32 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_set_ps1(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vdupq_n_f32(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	(void)a;
+	return vec_splats(a);
+#else
+	return simde_mm_set_ps(a, a, a, a);
+#endif
+}
+#define simde_mm_set1_ps(a) simde_mm_set_ps1(a)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_set_ps1(a) simde_mm_set_ps1(a)
+#define _mm_set1_ps(a) simde_mm_set1_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_move_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_move_ss(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vsetq_lane_f32(vgetq_lane_f32(b_.neon_f32, 0), a_.neon_f32, 0);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
+	m = {16, 17, 18, 19, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
+	r_.altivec_f32 = vec_perm(a_.altivec_f32, b_.altivec_f32, m);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v8x16_shuffle(b_.wasm_v128, a_.wasm_v128, 0, 1, 2,
+					 3, 20, 21, 22, 23, 24, 25, 26, 27, 28,
+					 29, 30, 31);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 4, 1, 2, 3);
+#else
+	r_.f32[0] = b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_move_ss(a, b) simde_mm_move_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_add_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_add_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vaddq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_add(a_.altivec_f32, b_.altivec_f32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f32 = a_.f32 + b_.f32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = a_.f32[i] + b_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_add_ps(a, b) simde_mm_add_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_add_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_add_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_add_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32_t b0 = vgetq_lane_f32(b_.neon_f32, 0);
+	float32x4_t value = vsetq_lane_f32(b0, vdupq_n_f32(0), 0);
+	// the upper values in the result must be the remnants of <a>.
+	r_.neon_f32 = vaddq_f32(a_.neon_f32, value);
+#else
+	r_.f32[0] = a_.f32[0] + b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_add_ss(a, b) simde_mm_add_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_and_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_and_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vandq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 & b_.i32;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_and(a_.altivec_f32, b_.altivec_f32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] & b_.i32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_and_ps(a, b) simde_mm_and_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_andnot_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_andnot_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_andc(b_.altivec_f32, a_.altivec_f32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = ~a_.i32 & b_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = ~(a_.i32[i]) & b_.i32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_andnot_ps(a, b) simde_mm_andnot_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_xor_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_xor_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = veorq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_xor(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_xor(a_.altivec_i32, b_.altivec_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f ^ b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] ^ b_.u32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_xor_ps(a, b) simde_mm_xor_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_or_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_or_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vorrq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_or(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_or(a_.altivec_i32, b_.altivec_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f | b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] | b_.u32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_or_ps(a, b) simde_mm_or_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_not_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_AVX512VL_NATIVE)
+	__m128i ai = _mm_castps_si128(a);
+	return _mm_castsi128_ps(_mm_ternarylogic_epi32(ai, ai, ai, 0x55));
+#elif defined(SIMDE_X86_SSE2_NATIVE)
+	/* Note: we use ints instead of floats because we don't want cmpeq
+ * to return false for (NaN, NaN) */
+	__m128i ai = _mm_castps_si128(a);
+	return _mm_castsi128_ps(_mm_andnot_si128(ai, _mm_cmpeq_epi32(ai, ai)));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vmvnq_s32(a_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_nor(a_.altivec_i32, a_.altivec_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_not(a_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = ~a_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = ~(a_.i32[i]);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_select_ps(simde__m128 a, simde__m128 b, simde__m128 mask)
+{
+/* This function is for when you want to blend two elements together
+ * according to a mask. It is similar to _mm_blendv_ps, except that
+ * it is undefined whether the blend is based on the highest bit in
+ * each lane (like blendv) or just bitwise operations. This allows
+ * us to implement the function efficiently everywhere.
+ *
+ * Basically, you promise that all the lanes in mask are either 0 or
+ * ~0. */
+#if defined(SIMDE_X86_SSE4_1_NATIVE)
+	return _mm_blendv_ps(a, b, mask);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b),
+				mask_ = simde__m128_to_private(mask);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vbslq_s32(mask_.neon_u32, b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_bitselect(b_.wasm_v128, a_.wasm_v128,
+					 mask_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 =
+		vec_sel(a_.altivec_i32, b_.altivec_i32, mask_.altivec_u32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 ^ ((a_.i32 ^ b_.i32) & mask_.i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] ^
+			 ((a_.i32[i] ^ b_.i32[i]) & mask_.i32[i]);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_avg_pu16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_avg_pu16(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vrhadd_u16(b_.neon_u16, a_.neon_u16);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
+	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_CONVERT_VECTOR_)
+	uint32_t wa SIMDE_VECTOR(16);
+	uint32_t wb SIMDE_VECTOR(16);
+	uint32_t wr SIMDE_VECTOR(16);
+	SIMDE_CONVERT_VECTOR_(wa, a_.u16);
+	SIMDE_CONVERT_VECTOR_(wb, b_.u16);
+	wr = (wa + wb + 1) >> 1;
+	SIMDE_CONVERT_VECTOR_(r_.u16, wr);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = (a_.u16[i] + b_.u16[i] + 1) >> 1;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pavgw(a, b) simde_mm_avg_pu16(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_avg_pu16(a, b) simde_mm_avg_pu16(a, b)
+#define _m_pavgw(a, b) simde_mm_avg_pu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_avg_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_avg_pu8(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vrhadd_u8(b_.neon_u8, a_.neon_u8);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
+	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_CONVERT_VECTOR_)
+	uint16_t wa SIMDE_VECTOR(16);
+	uint16_t wb SIMDE_VECTOR(16);
+	uint16_t wr SIMDE_VECTOR(16);
+	SIMDE_CONVERT_VECTOR_(wa, a_.u8);
+	SIMDE_CONVERT_VECTOR_(wb, b_.u8);
+	wr = (wa + wb + 1) >> 1;
+	SIMDE_CONVERT_VECTOR_(r_.u8, wr);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] + b_.u8[i] + 1) >> 1;
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pavgb(a, b) simde_mm_avg_pu8(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_avg_pu8(a, b) simde_mm_avg_pu8(a, b)
+#define _m_pavgb(a, b) simde_mm_avg_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_abs_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_AVX512F_NATIVE) && \
+	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(7, 1, 0))
+	return _mm512_castps512_ps128(_mm512_abs_ps(_mm512_castps128_ps512(a)));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vabsq_f32(a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_abs(a_.altivec_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_abs(a_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = simde_math_fabsf(a_.f32[i]);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpeq_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpeq_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vceqq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_eq(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmpeq(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), a_.f32 == b_.f32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] == b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_ps(a, b) simde_mm_cmpeq_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpeq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpeq_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpeq_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] == b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_ss(a, b) simde_mm_cmpeq_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpge_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpge_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcgeq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_ge(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmpge(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 >= b_.f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] >= b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpge_ps(a, b) simde_mm_cmpge_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpge_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
+	return _mm_cmpge_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpge_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] >= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpge_ss(a, b) simde_mm_cmpge_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpgt_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpgt_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcgtq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_gt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmpgt(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 > b_.f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] > b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_ps(a, b) simde_mm_cmpgt_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpgt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
+	return _mm_cmpgt_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpgt_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] > b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_ss(a, b) simde_mm_cmpgt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmple_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmple_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcleq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_le(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmple(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 <= b_.f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] <= b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmple_ps(a, b) simde_mm_cmple_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmple_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmple_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmple_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] <= b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmple_ss(a, b) simde_mm_cmple_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmplt_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmplt_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcltq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_lt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmplt(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 < b_.f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] < b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_ps(a, b) simde_mm_cmplt_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmplt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmplt_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmplt_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] < b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_ss(a, b) simde_mm_cmplt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpneq_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpneq_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_ne(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P9_NATIVE) && SIMDE_ARCH_POWER_CHECK(900) && \
+	!defined(HEDLEY_IBM_VERSION)
+	/* vec_cmpne(SIMDE_POWER_ALTIVEC_VECTOR(float), SIMDE_POWER_ALTIVEC_VECTOR(float))
+ is missing from XL C/C++ v16.1.1,
+ though the documentation (table 89 on page 432 of the IBM XL C/C++ for
+ Linux Compiler Reference, Version 16.1.1) shows that it should be
+ present. Both GCC and clang support it. */
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmpne(a_.altivec_f32, b_.altivec_f32));
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_cmpeq(a_.altivec_f32, b_.altivec_f32));
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_nor(r_.altivec_f32, r_.altivec_f32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.f32 != b_.f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (a_.f32[i] != b_.f32[i]) ? ~UINT32_C(0)
+						 : UINT32_C(0);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpneq_ps(a, b) simde_mm_cmpneq_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpneq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpneq_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpneq_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.u32[0] = (a_.f32[0] != b_.f32[0]) ? ~UINT32_C(0) : UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpneq_ss(a, b) simde_mm_cmpneq_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnge_ps(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmplt_ps(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnge_ps(a, b) simde_mm_cmpnge_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnge_ss(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmplt_ss(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnge_ss(a, b) simde_mm_cmpnge_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpngt_ps(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmple_ps(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpngt_ps(a, b) simde_mm_cmpngt_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpngt_ss(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmple_ss(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpngt_ss(a, b) simde_mm_cmpngt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnle_ps(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmpgt_ps(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnle_ps(a, b) simde_mm_cmpnle_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnle_ss(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmpgt_ss(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnle_ss(a, b) simde_mm_cmpnle_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnlt_ps(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmpge_ps(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnlt_ps(a, b) simde_mm_cmpnlt_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpnlt_ss(simde__m128 a, simde__m128 b)
+{
+	return simde_mm_cmpge_ss(a, b);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnlt_ss(a, b) simde_mm_cmpnlt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpord_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpord_ps(a, b);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_v128_and(wasm_f32x4_eq(a, a), wasm_f32x4_eq(b, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	/* Note: NEON does not have ordered compare builtin
+ Need to compare a eq a and b eq b to check for NaN
+ Do AND of results to get final */
+	uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	r_.neon_u32 = vandq_u32(ceqaa, ceqbb);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_and(wasm_f32x4_eq(a_.wasm_v128, a_.wasm_v128),
+				 wasm_f32x4_eq(b_.wasm_v128, b_.wasm_v128));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_and(vec_cmpeq(a_.altivec_f32, a_.altivec_f32),
+			vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
+#elif defined(simde_math_isnanf)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (simde_math_isnanf(a_.f32[i]) ||
+			 simde_math_isnanf(b_.f32[i]))
+				 ? UINT32_C(0)
+				 : ~UINT32_C(0);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpord_ps(a, b) simde_mm_cmpord_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpunord_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpunord_ps(a, b);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_v128_or(wasm_f32x4_ne(a, a), wasm_f32x4_ne(b, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t ceqaa = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t ceqbb = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	r_.neon_u32 = vmvnq_u32(vandq_u32(ceqaa, ceqbb));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_or(wasm_f32x4_ne(a_.wasm_v128, a_.wasm_v128),
+				 wasm_f32x4_ne(b_.wasm_v128, b_.wasm_v128));
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_nand(vec_cmpeq(a_.altivec_f32, a_.altivec_f32),
+			 vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_and(vec_cmpeq(a_.altivec_f32, a_.altivec_f32),
+			vec_cmpeq(b_.altivec_f32, b_.altivec_f32)));
+	r_.altivec_f32 = vec_nor(r_.altivec_f32, r_.altivec_f32);
+#elif defined(simde_math_isnanf)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = (simde_math_isnanf(a_.f32[i]) ||
+			 simde_math_isnanf(b_.f32[i]))
+				 ? ~UINT32_C(0)
+				 : UINT32_C(0);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpunord_ps(a, b) simde_mm_cmpunord_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpunord_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
+	return _mm_cmpunord_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpunord_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(simde_math_isnanf)
+	r_.u32[0] =
+		(simde_math_isnanf(a_.f32[0]) || simde_math_isnanf(b_.f32[0]))
+			? ~UINT32_C(0)
+			: UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpunord_ss(a, b) simde_mm_cmpunord_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comieq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comieq_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_eq_b = vceqq_f32(a_.neon_f32, b_.neon_f32);
+	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_eq_b), 0) != 0);
+#else
+	return a_.f32[0] == b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comieq_ss(a, b) simde_mm_comieq_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comige_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comige_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_ge_b = vcgeq_f32(a_.neon_f32, b_.neon_f32);
+	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0);
+#else
+	return a_.f32[0] >= b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comige_ss(a, b) simde_mm_comige_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comigt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comigt_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_gt_b = vcgtq_f32(a_.neon_f32, b_.neon_f32);
+	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0);
+#else
+	return a_.f32[0] > b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comigt_ss(a, b) simde_mm_comigt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comile_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comile_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_le_b = vcleq_f32(a_.neon_f32, b_.neon_f32);
+	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_le_b), 0) != 0);
+#else
+	return a_.f32[0] <= b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comile_ss(a, b) simde_mm_comile_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comilt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comilt_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_lt_b = vcltq_f32(a_.neon_f32, b_.neon_f32);
+	return !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_lt_b), 0) != 0);
+#else
+	return a_.f32[0] < b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comilt_ss(a, b) simde_mm_comilt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comineq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_comineq_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
+	return !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_neq_b), 0) != 0);
+#else
+	return a_.f32[0] != b_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_comineq_ss(a, b) simde_mm_comineq_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_copysign_ps(simde__m128 dest, simde__m128 src)
+{
+	simde__m128_private r_, dest_ = simde__m128_to_private(dest),
+				src_ = simde__m128_to_private(src);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint32x4_t sign_pos =
+		vreinterpretq_u32_f32(vdupq_n_f32(-SIMDE_FLOAT32_C(0.0)));
+	r_.neon_u32 = vbslq_u32(sign_pos, src_.neon_u32, dest_.neon_u32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	const v128_t sign_pos = wasm_f32x4_splat(-0.0f);
+	r_.wasm_v128 =
+		wasm_v128_bitselect(src_.wasm_v128, dest_.wasm_v128, sign_pos);
+#elif defined(SIMDE_POWER_ALTIVEC_P9_NATIVE)
+#if !defined(HEDLEY_IBM_VERSION)
+	r_.altivec_f32 = vec_cpsgn(dest_.altivec_f32, src_.altivec_f32);
+#else
+	r_.altivec_f32 = vec_cpsgn(src_.altivec_f32, dest_.altivec_f32);
+#endif
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	const SIMDE_POWER_ALTIVEC_VECTOR(unsigned int)
+		sign_pos = HEDLEY_REINTERPRET_CAST(
+			SIMDE_POWER_ALTIVEC_VECTOR(unsigned int),
+			vec_splats(-0.0f));
+	r_.altivec_f32 = vec_sel(dest_.altivec_f32, src_.altivec_f32, sign_pos);
+#elif defined(SIMDE_IEEE754_STORAGE)
+	(void)src_;
+	(void)dest_;
+	simde__m128 sign_pos = simde_mm_set1_ps(-0.0f);
+	r_ = simde__m128_to_private(simde_mm_xor_ps(
+		dest, simde_mm_and_ps(simde_mm_xor_ps(dest, src), sign_pos)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = simde_math_copysignf(dest_.f32[i], src_.f32[i]);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_xorsign_ps(simde__m128 dest, simde__m128 src)
+{
+	return simde_mm_xor_ps(simde_mm_and_ps(simde_mm_set1_ps(-0.0f), src),
+			 dest);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvt_pi2ps(simde__m128 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvt_pi2ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32),
+				 vget_high_f32(a_.neon_f32));
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
+	r_.m64_private[1] = a_.m64_private[1];
+#else
+	r_.f32[0] = (simde_float32)b_.i32[0];
+	r_.f32[1] = (simde_float32)b_.i32[1];
+	r_.i32[2] = a_.i32[2];
+	r_.i32[3] = a_.i32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvt_pi2ps(a, b) simde_mm_cvt_pi2ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvt_ps2pi(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvt_ps2pi(a);
+#else
+	simde__m64_private r_;
+	simde__m128_private a_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	a_ = simde__m128_to_private(
+		simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
+	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
+#elif defined(SIMDE_CONVERT_VECTOR_) && SIMDE_NATURAL_VECTOR_SIZE_GE(128)
+	a_ = simde__m128_to_private(
+		simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
+	SIMDE_CONVERT_VECTOR_(r_.i32, a_.m64_private[0].f32);
+#else
+	a_ = simde__m128_to_private(a);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = HEDLEY_STATIC_CAST(
+			int32_t, simde_math_nearbyintf(a_.f32[i]));
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvt_ps2pi(a) simde_mm_cvt_ps2pi((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvt_si2ss(simde__m128 a, int32_t b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cvt_si2ss(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vsetq_lane_f32(HEDLEY_STATIC_CAST(float, b), a_.neon_f32, 0);
+#else
+	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
+	r_.i32[1] = a_.i32[1];
+	r_.i32[2] = a_.i32[2];
+	r_.i32[3] = a_.i32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvt_si2ss(a, b) simde_mm_cvt_si2ss((a), b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvt_ss2si(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cvt_ss2si(a);
+#elif defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
+	return vgetq_lane_s32(vcvtnq_s32_f32(simde__m128_to_neon_f32(a)), 0);
+#else
+	simde__m128_private a_ = simde__m128_to_private(
+		simde_mm_round_ps(a, SIMDE_MM_FROUND_CUR_DIRECTION));
+	return SIMDE_CONVERT_FTOI(int32_t, a_.f32[0]);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvt_ss2si(a) simde_mm_cvt_ss2si((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpi16_ps(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpi16_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcvtq_f32_s32(vmovl_s16(a_.neon_i16));
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f32, a_.i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		simde_float32 v = a_.i16[i];
+		r_.f32[i] = v;
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpi16_ps(a) simde_mm_cvtpi16_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpi32_ps(simde__m128 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpi32_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+	simde__m64_private b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcombine_f32(vcvt_f32_s32(b_.neon_i32),
+				 vget_high_f32(a_.neon_f32));
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, b_.i32);
+	r_.m64_private[1] = a_.m64_private[1];
+#else
+	r_.f32[0] = (simde_float32)b_.i32[0];
+	r_.f32[1] = (simde_float32)b_.i32[1];
+	r_.i32[2] = a_.i32[2];
+	r_.i32[3] = a_.i32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpi32_ps(a, b) simde_mm_cvtpi32_ps((a), b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpi32x2_ps(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpi32x2_ps(a, b);
+#else
+	simde__m128_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcvtq_f32_s32(vcombine_s32(a_.neon_i32, b_.neon_i32));
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, a_.i32);
+	SIMDE_CONVERT_VECTOR_(r_.m64_private[1].f32, b_.i32);
+#else
+	r_.f32[0] = (simde_float32)a_.i32[0];
+	r_.f32[1] = (simde_float32)a_.i32[1];
+	r_.f32[2] = (simde_float32)b_.i32[0];
+	r_.f32[3] = (simde_float32)b_.i32[1];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpi32x2_ps(a, b) simde_mm_cvtpi32x2_ps(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpi8_ps(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpi8_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vcvtq_f32_s32(vmovl_s16(vget_low_s16(vmovl_s8(a_.neon_i8))));
+#else
+	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[0]);
+	r_.f32[1] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[1]);
+	r_.f32[2] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[2]);
+	r_.f32[3] = HEDLEY_STATIC_CAST(simde_float32, a_.i8[3]);
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpi8_ps(a) simde_mm_cvtpi8_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtps_pi16(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtps_pi16(a);
+#else
+	simde__m64_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95399)
+	r_.neon_i16 = vmovn_s32(vcvtq_s32_f32(vrndiq_f32(a_.neon_f32)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = SIMDE_CONVERT_FTOI(int16_t,
+					 simde_math_roundf(a_.f32[i]));
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtps_pi16(a) simde_mm_cvtps_pi16((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtps_pi32(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtps_pi32(a);
+#else
+	simde__m64_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && \
+	defined(SIMDE_FAST_CONVERSION_RANGE) && !defined(SIMDE_BUG_GCC_95399)
+	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(vrndiq_f32(a_.neon_f32)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		simde_float32 v = simde_math_roundf(a_.f32[i]);
+#if !defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#else
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#endif
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtps_pi32(a) simde_mm_cvtps_pi32((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtps_pi8(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtps_pi8(a);
+#else
+	simde__m64_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V8_NATIVE) && !defined(SIMDE_BUG_GCC_95471)
+	/* Clamp the input to [INT8_MIN, INT8_MAX], round, convert to i32, narrow to
+ * i16, combine with an all-zero vector of i16 (which will become the upper
+ * half), narrow to i8. */
+	float32x4_t max =
+		vdupq_n_f32(HEDLEY_STATIC_CAST(simde_float32, INT8_MAX));
+	float32x4_t min =
+		vdupq_n_f32(HEDLEY_STATIC_CAST(simde_float32, INT8_MIN));
+	float32x4_t values =
+		vrndnq_f32(vmaxq_f32(vminq_f32(max, a_.neon_f32), min));
+	r_.neon_i8 = vmovn_s16(
+		vcombine_s16(vmovn_s32(vcvtq_s32_f32(values)), vdup_n_s16(0)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(a_.f32) / sizeof(a_.f32[0])); i++) {
+		if (a_.f32[i] > HEDLEY_STATIC_CAST(simde_float32, INT8_MAX))
+			r_.i8[i] = INT8_MAX;
+		else if (a_.f32[i] <
+			 HEDLEY_STATIC_CAST(simde_float32, INT8_MIN))
+			r_.i8[i] = INT8_MIN;
+		else
+			r_.i8[i] = SIMDE_CONVERT_FTOI(
+				int8_t, simde_math_roundf(a_.f32[i]));
+	}
+	/* Note: the upper half is undefined */
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtps_pi8(a) simde_mm_cvtps_pi8((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpu16_ps(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpu16_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcvtq_f32_u32(vmovl_u16(a_.neon_u16));
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f32, a_.u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = (simde_float32)a_.u16[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpu16_ps(a) simde_mm_cvtpu16_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpu8_ps(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpu8_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vcvtq_f32_u32(vmovl_u16(vget_low_u16(vmovl_u8(a_.neon_u8))));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = HEDLEY_STATIC_CAST(simde_float32, a_.u8[i]);
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpu8_ps(a) simde_mm_cvtpu8_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtsi32_ss(simde__m128 a, int32_t b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cvtsi32_ss(a, b);
+#else
+	simde__m128_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vsetq_lane_f32(HEDLEY_STATIC_CAST(float32_t, b),
+				 a_.neon_f32, 0);
+#else
+	r_ = a_;
+	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi32_ss(a, b) simde_mm_cvtsi32_ss((a), b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtsi64_ss(simde__m128 a, int64_t b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if !defined(__PGI)
+	return _mm_cvtsi64_ss(a, b);
+#else
+	return _mm_cvtsi64x_ss(a, b);
+#endif
+#else
+	simde__m128_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vsetq_lane_f32(HEDLEY_STATIC_CAST(float32_t, b),
+				 a_.neon_f32, 0);
+#else
+	r_ = a_;
+	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b);
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi64_ss(a, b) simde_mm_cvtsi64_ss((a), b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float32 simde_mm_cvtss_f32(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cvtss_f32(a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vgetq_lane_f32(a_.neon_f32, 0);
+#else
+	return a_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtss_f32(a) simde_mm_cvtss_f32((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvtss_si32(simde__m128 a)
+{
+	return simde_mm_cvt_ss2si(a);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtss_si32(a) simde_mm_cvtss_si32((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvtss_si64(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if !defined(__PGI)
+	return _mm_cvtss_si64(a);
+#else
+	return _mm_cvtss_si64x(a);
+#endif
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return SIMDE_CONVERT_FTOI(
+		int64_t, simde_math_roundf(vgetq_lane_f32(a_.neon_f32, 0)));
+#else
+	return SIMDE_CONVERT_FTOI(int64_t, simde_math_roundf(a_.f32[0]));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtss_si64(a) simde_mm_cvtss_si64((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtt_ps2pi(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtt_ps2pi(a);
+#else
+	simde__m64_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	r_.neon_i32 = vcvt_s32_f32(vget_low_f32(a_.neon_f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		simde_float32 v = a_.f32[i];
+#if !defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#else
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#endif
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_mm_cvttps_pi32(a) simde_mm_cvtt_ps2pi(a)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtt_ps2pi(a) simde_mm_cvtt_ps2pi((a))
+#define _mm_cvttps_pi32(a) simde_mm_cvttps_pi32((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvtt_ss2si(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cvtt_ss2si(a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	return SIMDE_CONVERT_FTOI(int32_t, vgetq_lane_f32(a_.neon_f32, 0));
+#else
+	simde_float32 v = a_.f32[0];
+#if !defined(SIMDE_FAST_CONVERSION_RANGE)
+	return ((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
+		(v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX)))
+		 ? SIMDE_CONVERT_FTOI(int32_t, v)
+		 : INT32_MIN;
+#else
+	return SIMDE_CONVERT_FTOI(int32_t, v);
+#endif
+#endif
+#endif
+}
+#define simde_mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtt_ss2si(a) simde_mm_cvtt_ss2si((a))
+#define _mm_cvttss_si32(a) simde_mm_cvtt_ss2si((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvttss_si64(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
+	!defined(_MSC_VER)
+#if defined(__PGI)
+	return _mm_cvttss_si64x(a);
+#else
+	return _mm_cvttss_si64(a);
+#endif
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return SIMDE_CONVERT_FTOI(int64_t, vgetq_lane_f32(a_.neon_f32, 0));
+#else
+	return SIMDE_CONVERT_FTOI(int64_t, a_.f32[0]);
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttss_si64(a) simde_mm_cvttss_si64((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cmpord_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_cmpord_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_cmpord_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(simde_math_isnanf)
+	r_.u32[0] = (simde_math_isnanf(simde_mm_cvtss_f32(a)) ||
+		 simde_math_isnanf(simde_mm_cvtss_f32(b)))
+			 ? UINT32_C(0)
+			 : ~UINT32_C(0);
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.u32[i] = a_.u32[i];
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpord_ss(a, b) simde_mm_cmpord_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_div_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_div_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vdivq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x4_t recip0 = vrecpeq_f32(b_.neon_f32);
+	float32x4_t recip1 =
+		vmulq_f32(recip0, vrecpsq_f32(recip0, b_.neon_f32));
+	r_.neon_f32 = vmulq_f32(a_.neon_f32, recip1);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_div(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = vec_div(a_.altivec_f32, b_.altivec_f32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f32 = a_.f32 / b_.f32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = a_.f32[i] / b_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_div_ps(a, b) simde_mm_div_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_div_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_div_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_div_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32_t value = vgetq_lane_f32(
+		simde__m128_to_private(simde_mm_div_ps(a, b)).neon_f32, 0);
+	r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
+#else
+	r_.f32[0] = a_.f32[0] / b_.f32[0];
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = a_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_div_ss(a, b) simde_mm_div_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int16_t simde_mm_extract_pi16(simde__m64 a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 3)
+{
+	simde__m64_private a_ = simde__m64_to_private(a);
+	return a_.i16[imm8];
+}
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
+	!defined(HEDLEY_PGI_VERSION)
+#if defined(SIMDE_BUG_CLANG_44589)
+#define simde_mm_extract_pi16(a, imm8) \
+	(HEDLEY_DIAGNOSTIC_PUSH _Pragma( \
+		"clang diagnostic ignored \"-Wvector-conversion\"") \
+		 HEDLEY_STATIC_CAST(int16_t, _mm_extract_pi16((a), (imm8))) \
+			 HEDLEY_DIAGNOSTIC_POP)
+#else
+#define simde_mm_extract_pi16(a, imm8) \
+	HEDLEY_STATIC_CAST(int16_t, _mm_extract_pi16(a, imm8))
+#endif
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_extract_pi16(a, imm8) \
+	vget_lane_s16(simde__m64_to_private(a).neon_i16, imm8)
+#endif
+#define simde_m_pextrw(a, imm8) simde_mm_extract_pi16(a, imm8)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_extract_pi16(a, imm8) simde_mm_extract_pi16((a), (imm8))
+#define _m_pextrw(a, imm8) simde_mm_extract_pi16((a), (imm8))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_insert_pi16(simde__m64 a, int16_t i, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 3)
+{
+	simde__m64_private r_, a_ = simde__m64_to_private(a);
+
+	r_.i64[0] = a_.i64[0];
+	r_.i16[imm8] = i;
+
+	return simde__m64_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
+	!defined(__PGI)
+#if defined(SIMDE_BUG_CLANG_44589)
+#define ssimde_mm_insert_pi16(a, i, imm8) \
+	(HEDLEY_DIAGNOSTIC_PUSH _Pragma( \
+		"clang diagnostic ignored \"-Wvector-conversion\"")( \
+		_mm_insert_pi16((a), (i), (imm8))) HEDLEY_DIAGNOSTIC_POP)
+#else
+#define simde_mm_insert_pi16(a, i, imm8) _mm_insert_pi16(a, i, imm8)
+#endif
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_insert_pi16(a, i, imm8) \
+	simde__m64_from_neon_i16( \
+		vset_lane_s16((i), simde__m64_to_neon_i16(a), (imm8)))
+#endif
+#define simde_m_pinsrw(a, i, imm8) (simde_mm_insert_pi16(a, i, imm8))
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_insert_pi16(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
+#define _m_pinsrw(a, i, imm8) simde_mm_insert_pi16(a, i, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128
+simde_mm_load_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_load_ps(mem_addr);
+#else
+	simde__m128_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vld1q_f32(mem_addr);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = vec_vsx_ld(0, mem_addr);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_ld(0, mem_addr);
+#else
+	simde_memcpy(&r_, SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128),
+		 sizeof(r_));
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_load_ps(mem_addr) simde_mm_load_ps(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_load1_ps(simde_float32 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_load_ps1(mem_addr);
+#else
+	simde__m128_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vld1q_dup_f32(mem_addr);
+#else
+	r_ = simde__m128_to_private(simde_mm_set1_ps(*mem_addr));
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#define simde_mm_load_ps1(mem_addr) simde_mm_load1_ps(mem_addr)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_load_ps1(mem_addr) simde_mm_load1_ps(mem_addr)
+#define _mm_load1_ps(mem_addr) simde_mm_load1_ps(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_load_ss(simde_float32 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_load_ss(mem_addr);
+#else
+	simde__m128_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vsetq_lane_f32(*mem_addr, vdupq_n_f32(0), 0);
+#else
+	r_.f32[0] = *mem_addr;
+	r_.i32[1] = 0;
+	r_.i32[2] = 0;
+	r_.i32[3] = 0;
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_load_ss(mem_addr) simde_mm_load_ss(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_loadh_pi(simde__m128 a, simde__m64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_loadh_pi(a,
+			 HEDLEY_REINTERPRET_CAST(__m64 const *, mem_addr));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcombine_f32(
+		vget_low_f32(a_.neon_f32),
+		vld1_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr)));
+#else
+	simde__m64_private b_ =
+		*HEDLEY_REINTERPRET_CAST(simde__m64_private const *, mem_addr);
+	r_.f32[0] = a_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = b_.f32[0];
+	r_.f32[3] = b_.f32[1];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#if HEDLEY_HAS_WARNING("-Wold-style-cast")
+#define _mm_loadh_pi(a, mem_addr) \
+	simde_mm_loadh_pi((a), HEDLEY_REINTERPRET_CAST(simde__m64 const *, \
+						 (mem_addr)))
+#else
+#define _mm_loadh_pi(a, mem_addr) \
+	simde_mm_loadh_pi((a), (simde__m64 const *)(mem_addr))
+#endif
+#endif
+
+/* The SSE documentation says that there are no alignment requirements
+ for mem_addr. Unfortunately they used the __m64 type for the argument
+ which is supposed to be 8-byte aligned, so some compilers (like clang
+ with -Wcast-align) will generate a warning if you try to cast, say,
+ a simde_float32* to a simde__m64* for this function.
+
+ I think the choice of argument type is unfortunate, but I do think we
+ need to stick to it here. If there is demand I can always add something
+ like simde_x_mm_loadl_f32(simde__m128, simde_float32 mem_addr[2]) */
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_loadl_pi(simde__m128 a, simde__m64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_loadl_pi(a,
+			 HEDLEY_REINTERPRET_CAST(__m64 const *, mem_addr));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcombine_f32(
+		vld1_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr)),
+		vget_high_f32(a_.neon_f32));
+#else
+	simde__m64_private b_;
+	simde_memcpy(&b_, mem_addr, sizeof(b_));
+	r_.i32[0] = b_.i32[0];
+	r_.i32[1] = b_.i32[1];
+	r_.i32[2] = a_.i32[2];
+	r_.i32[3] = a_.i32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#if HEDLEY_HAS_WARNING("-Wold-style-cast")
+#define _mm_loadl_pi(a, mem_addr) \
+	simde_mm_loadl_pi((a), HEDLEY_REINTERPRET_CAST(simde__m64 const *, \
+						 (mem_addr)))
+#else
+#define _mm_loadl_pi(a, mem_addr) \
+	simde_mm_loadl_pi((a), (simde__m64 const *)(mem_addr))
+#endif
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128
+simde_mm_loadr_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_loadr_ps(mem_addr);
+#else
+	simde__m128_private r_,
+		v_ = simde__m128_to_private(simde_mm_load_ps(mem_addr));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vrev64q_f32(v_.neon_f32);
+	r_.neon_f32 = vextq_f32(r_.neon_f32, r_.neon_f32, 2);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) && defined(__PPC64__)
+	r_.altivec_f32 = vec_reve(v_.altivec_f32);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, v_.f32, v_.f32, 3, 2, 1, 0);
+#else
+	r_.f32[0] = v_.f32[3];
+	r_.f32[1] = v_.f32[2];
+	r_.f32[2] = v_.f32[1];
+	r_.f32[3] = v_.f32[0];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_loadr_ps(mem_addr) simde_mm_loadr_ps(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128
+simde_mm_loadu_ps(simde_float32 const mem_addr[HEDLEY_ARRAY_PARAM(4)])
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_loadu_ps(mem_addr);
+#else
+	simde__m128_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vld1q_f32(HEDLEY_REINTERPRET_CAST(const float32_t *, mem_addr));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_load(mem_addr);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) && defined(__PPC64__)
+	r_.altivec_f32 = vec_vsx_ld(0, mem_addr);
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_ps(mem_addr) simde_mm_loadu_ps(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_maskmove_si64(simde__m64 a, simde__m64 mask, int8_t *mem_addr)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	_mm_maskmove_si64(a, mask, HEDLEY_REINTERPRET_CAST(char *, mem_addr));
+#else
+	simde__m64_private a_ = simde__m64_to_private(a),
+			 mask_ = simde__m64_to_private(mask);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(a_.i8) / sizeof(a_.i8[0])); i++)
+		if (mask_.i8[i] < 0)
+			mem_addr[i] = a_.i8[i];
+#endif
+}
+#define simde_m_maskmovq(a, mask, mem_addr) \
+	simde_mm_maskmove_si64(a, mask, mem_addr)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_maskmove_si64(a, mask, mem_addr) \
+	simde_mm_maskmove_si64( \
+		(a), (mask), \
+		SIMDE_CHECKED_REINTERPRET_CAST(int8_t *, char *, (mem_addr)))
+#define _m_maskmovq(a, mask, mem_addr) \
+	simde_mm_maskmove_si64( \
+		(a), (mask), \
+		SIMDE_CHECKED_REINTERPRET_CAST(int8_t *, char *, (mem_addr)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_max_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_max_pi16(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vmax_s16(a_.neon_i16, b_.neon_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? a_.i16[i] : b_.i16[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_max_pi16(a, b) simde_mm_max_pi16(a, b)
+#define _m_pmaxsw(a, b) simde_mm_max_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_max_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_max_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_NANS)
+	r_.neon_f32 = vmaxq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vbslq_f32(vcgtq_f32(a_.neon_f32, b_.neon_f32),
+				a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE) && defined(SIMDE_FAST_NANS)
+	r_.wasm_v128 = wasm_f32x4_max(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 =
+		wasm_v128_bitselect(a_.wasm_v128, b_.wasm_v128,
+				 wasm_f32x4_gt(a_.wasm_v128, b_.wasm_v128));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && defined(SIMDE_FAST_NANS)
+	r_.altivec_f32 = vec_max(a_.altivec_f32, b_.altivec_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_sel(b_.altivec_f32, a_.altivec_f32,
+				 vec_cmpgt(a_.altivec_f32, b_.altivec_f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = (a_.f32[i] > b_.f32[i]) ? a_.f32[i] : b_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_max_ps(a, b) simde_mm_max_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_max_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_max_pu8(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vmax_u8(a_.neon_u8, b_.neon_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] > b_.u8[i]) ? a_.u8[i] : b_.u8[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmaxub(a, b) simde_mm_max_pu8(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_max_pu8(a, b) simde_mm_max_pu8(a, b)
+#define _m_pmaxub(a, b) simde_mm_max_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_max_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_max_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_max_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32_t value = vgetq_lane_f32(maxq_f32(a_.neon_f32, b_.neon_f32), 0);
+	r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
+#else
+	r_.f32[0] = (a_.f32[0] > b_.f32[0]) ? a_.f32[0] : b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_max_ss(a, b) simde_mm_max_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_min_pi16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_min_pi16(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vmin_s16(a_.neon_i16, b_.neon_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? a_.i16[i] : b_.i16[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pminsw(a, b) simde_mm_min_pi16(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_min_pi16(a, b) simde_mm_min_pi16(a, b)
+#define _m_pminsw(a, b) simde_mm_min_pi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_min_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_min_ps(a, b);
+#elif defined(SIMDE_FAST_NANS) && defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return simde__m128_from_neon_f32(vminq_f32(simde__m128_to_neon_f32(a),
+						 simde__m128_to_neon_f32(b)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+#if defined(SIMDE_FAST_NANS)
+	r_.wasm_v128 = wasm_f32x4_min(a_.wasm_v128, b_.wasm_v128);
+#else
+	r_.wasm_v128 =
+		wasm_v128_bitselect(a_.wasm_v128, b_.wasm_v128,
+				 wasm_f32x4_lt(a_.wasm_v128, b_.wasm_v128));
+#endif
+	return simde__m128_from_private(r_);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_FAST_NANS)
+	r_.altivec_f32 = vec_min(a_.altivec_f32, b_.altivec_f32);
+#else
+	r_.altivec_f32 = vec_sel(b_.altivec_f32, a_.altivec_f32,
+				 vec_cmpgt(b_.altivec_f32, a_.altivec_f32));
+#endif
+
+	return simde__m128_from_private(r_);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	simde__m128 mask = simde_mm_cmplt_ps(a, b);
+	return simde_mm_or_ps(simde_mm_and_ps(mask, a),
+			 simde_mm_andnot_ps(mask, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = (a_.f32[i] < b_.f32[i]) ? a_.f32[i] : b_.f32[i];
+	}
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_min_ps(a, b) simde_mm_min_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_min_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_min_pu8(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vmin_u8(a_.neon_u8, b_.neon_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] < b_.u8[i]) ? a_.u8[i] : b_.u8[i];
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pminub(a, b) simde_mm_min_pu8(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_min_pu8(a, b) simde_mm_min_pu8(a, b)
+#define _m_pminub(a, b) simde_mm_min_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_min_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_min_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_min_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32_t value =
+		vgetq_lane_f32(vminq_f32(a_.neon_f32, b_.neon_f32), 0);
+	r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
+#else
+	r_.f32[0] = (a_.f32[0] < b_.f32[0]) ? a_.f32[0] : b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_min_ss(a, b) simde_mm_min_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_movehl_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_movehl_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x2_t a32 = vget_high_f32(a_.neon_f32);
+	float32x2_t b32 = vget_high_f32(b_.neon_f32);
+	r_.neon_f32 = vcombine_f32(b32, a32);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_mergel(b_.altivec_i64, a_.altivec_i64));
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 6, 7, 2, 3);
+#else
+	r_.f32[0] = b_.f32[2];
+	r_.f32[1] = b_.f32[3];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_movehl_ps(a, b) simde_mm_movehl_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_movelh_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_movelh_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x2_t a10 = vget_low_f32(a_.neon_f32);
+	float32x2_t b10 = vget_low_f32(b_.neon_f32);
+	r_.neon_f32 = vcombine_f32(a10, b10);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 1, 4, 5);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(float),
+		vec_mergeh(a_.altivec_i64, b_.altivec_i64));
+#else
+	r_.f32[0] = a_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = b_.f32[0];
+	r_.f32[3] = b_.f32[1];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_movelh_ps(a, b) simde_mm_movelh_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_movemask_pi8(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_movemask_pi8(a);
+#else
+	simde__m64_private a_ = simde__m64_to_private(a);
+	int r = 0;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint8x8_t input = a_.neon_u8;
+	const int8_t xr[8] = {-7, -6, -5, -4, -3, -2, -1, 0};
+	const uint8x8_t mask_and = vdup_n_u8(0x80);
+	const int8x8_t mask_shift = vld1_s8(xr);
+	const uint8x8_t mask_result =
+		vshl_u8(vand_u8(input, mask_and), mask_shift);
+	uint8x8_t lo = mask_result;
+	r = vaddv_u8(lo);
+#else
+	const size_t nmemb = sizeof(a_.i8) / sizeof(a_.i8[0]);
+	SIMDE_VECTORIZE_REDUCTION(| : r)
+	for (size_t i = 0; i < nmemb; i++) {
+		r |= (a_.u8[nmemb - 1 - i] >> 7) << (nmemb - 1 - i);
+	}
+#endif
+
+	return r;
+#endif
+}
+#define simde_m_pmovmskb(a) simde_mm_movemask_pi8(a)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_movemask_pi8(a) simde_mm_movemask_pi8(a)
+#define _m_pmovmskb(a) simde_mm_movemask_pi8(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_movemask_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_movemask_ps(a);
+#else
+	int r = 0;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	static const int32_t shift_amount[] = {0, 1, 2, 3};
+	const int32x4_t shift = vld1q_s32(shift_amount);
+	uint32x4_t tmp = vshrq_n_u32(a_.neon_u32, 31);
+	return HEDLEY_STATIC_CAST(int, vaddvq_u32(vshlq_u32(tmp, shift)));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	// Shift out everything but the sign bits with a 32-bit unsigned shift right.
+	uint64x2_t high_bits =
+		vreinterpretq_u64_u32(vshrq_n_u32(a_.neon_u32, 31));
+	// Merge the two pairs together with a 64-bit unsigned shift right + add.
+	uint8x16_t paired =
+		vreinterpretq_u8_u64(vsraq_n_u64(high_bits, high_bits, 31));
+	// Extract the result.
+	return vgetq_lane_u8(paired, 0) | (vgetq_lane_u8(paired, 8) << 2);
+#else
+	SIMDE_VECTORIZE_REDUCTION(| : r)
+	for (size_t i = 0; i < sizeof(a_.u32) / sizeof(a_.u32[0]); i++) {
+		r |= (a_.u32[i] >> ((sizeof(a_.u32[i]) * CHAR_BIT) - 1)) << i;
+	}
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_movemask_ps(a) simde_mm_movemask_ps((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_mul_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_mul_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vmulq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_mul(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f32 = a_.f32 * b_.f32;
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = vec_mul(a_.altivec_f32, b_.altivec_f32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = a_.f32[i] * b_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_ps(a, b) simde_mm_mul_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_mul_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_mul_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_mul_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.f32[0] = a_.f32[0] * b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_ss(a, b) simde_mm_mul_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_mulhi_pu16(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_mulhi_pu16(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint32x4_t t1 = vmull_u16(a_.neon_u16, b_.neon_u16);
+	const uint32x4_t t2 = vshrq_n_u32(t1, 16);
+	const uint16x4_t t3 = vmovn_u32(t2);
+	r_.neon_u16 = t3;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(
+			uint16_t, ((HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
+				 HEDLEY_STATIC_CAST(uint32_t, b_.u16[i])) >>
+				 UINT32_C(16)));
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_pmulhuw(a, b) simde_mm_mulhi_pu16(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_mulhi_pu16(a, b) simde_mm_mulhi_pu16(a, b)
+#define _m_pmulhuw(a, b) simde_mm_mulhi_pu16(a, b)
+#endif
+
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(HEDLEY_GCC_VERSION)
+#define SIMDE_MM_HINT_NTA HEDLEY_STATIC_CAST(enum _mm_hint, 0)
+#define SIMDE_MM_HINT_T0 HEDLEY_STATIC_CAST(enum _mm_hint, 1)
+#define SIMDE_MM_HINT_T1 HEDLEY_STATIC_CAST(enum _mm_hint, 2)
+#define SIMDE_MM_HINT_T2 HEDLEY_STATIC_CAST(enum _mm_hint, 3)
+#define SIMDE_MM_HINT_ENTA HEDLEY_STATIC_CAST(enum _mm_hint, 4)
+#define SIMDE_MM_HINT_ET0 HEDLEY_STATIC_CAST(enum _mm_hint, 5)
+#define SIMDE_MM_HINT_ET1 HEDLEY_STATIC_CAST(enum _mm_hint, 6)
+#define SIMDE_MM_HINT_ET2 HEDLEY_STATIC_CAST(enum _mm_hint, 7)
+#else
+#define SIMDE_MM_HINT_NTA 0
+#define SIMDE_MM_HINT_T0 1
+#define SIMDE_MM_HINT_T1 2
+#define SIMDE_MM_HINT_T2 3
+#define SIMDE_MM_HINT_ENTA 4
+#define SIMDE_MM_HINT_ET0 5
+#define SIMDE_MM_HINT_ET1 6
+#define SIMDE_MM_HINT_ET2 7
+#endif
+
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wreserved-id-macro")
+_Pragma("clang diagnostic ignored \"-Wreserved-id-macro\"")
+#endif
+#undef _MM_HINT_NTA
+#define _MM_HINT_NTA SIMDE_MM_HINT_NTA
+#undef _MM_HINT_T0
+#define _MM_HINT_T0 SIMDE_MM_HINT_T0
+#undef _MM_HINT_T1
+#define _MM_HINT_T1 SIMDE_MM_HINT_T1
+#undef _MM_HINT_T2
+#define _MM_HINT_T2 SIMDE_MM_HINT_T2
+#undef _MM_HINT_ETNA
+#define _MM_HINT_ETNA SIMDE_MM_HINT_ETNA
+#undef _MM_HINT_ET0
+#define _MM_HINT_ET0 SIMDE_MM_HINT_ET0
+#undef _MM_HINT_ET1
+#define _MM_HINT_ET1 SIMDE_MM_HINT_ET1
+#undef _MM_HINT_ET1
+#define _MM_HINT_ET2 SIMDE_MM_HINT_ET2
+	HEDLEY_DIAGNOSTIC_POP
+#endif
+
+	SIMDE_FUNCTION_ATTRIBUTES void simde_mm_prefetch(char const *p, int i)
+{
+#if defined(HEDLEY_GCC_VERSION)
+	__builtin_prefetch(p);
+#else
+	(void)p;
+#endif
+
+	(void)i;
+}
+#if defined(SIMDE_X86_SSE_NATIVE)
+#if defined(__clang__) && \
+	!SIMDE_DETECT_CLANG_VERSION_CHECK( \
+		10, 0, 0) /* https://reviews.llvm.org/D71718 */
+#define simde_mm_prefetch(p, i) \
+	(__extension__({ \
+		HEDLEY_DIAGNOSTIC_PUSH \
+		HEDLEY_DIAGNOSTIC_DISABLE_CAST_QUAL \
+		_mm_prefetch((p), (i)); \
+		HEDLEY_DIAGNOSTIC_POP \
+	}))
+#else
+#define simde_mm_prefetch(p, i) _mm_prefetch(p, i)
+#endif
+#endif
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_prefetch(p, i) simde_mm_prefetch(p, i)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_negate_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return simde_mm_xor_ps(a, _mm_set1_ps(SIMDE_FLOAT32_C(-0.0)));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && \
+	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(8, 1, 0))
+	r_.altivec_f32 = vec_neg(a_.altivec_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vnegq_f32(a_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_neg(a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f32 = vec_neg(a_.altivec_f32);
+#elif defined(SIMDE_VECTOR_NEGATE)
+	r_.f32 = -a_.f32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = -a_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_rcp_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_rcp_ps(a);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x4_t recip = vrecpeq_f32(a_.neon_f32);
+
+#if SIMDE_ACCURACY_PREFERENCE > 0
+	for (int i = 0; i < SIMDE_ACCURACY_PREFERENCE; ++i) {
+		recip = vmulq_f32(recip, vrecpsq_f32(recip, a_.neon_f32));
+	}
+#endif
+
+	r_.neon_f32 = recip;
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_div(simde_mm_set1_ps(1.0f), a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_re(a_.altivec_f32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.f32 = 1.0f / a_.f32;
+#elif defined(SIMDE_IEEE754_STORAGE)
+	/* https://stackoverflow.com/questions/12227126/division-as-multiply-and-lut-fast-float-division-reciprocal/12228234#12228234 */
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		int32_t ix;
+		simde_float32 fx = a_.f32[i];
+		simde_memcpy(&ix, &fx, sizeof(ix));
+		int32_t x = INT32_C(0x7EF311C3) - ix;
+		simde_float32 temp;
+		simde_memcpy(&temp, &x, sizeof(temp));
+		r_.f32[i] = temp * (SIMDE_FLOAT32_C(2.0) - temp * fx);
+	}
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = 1.0f / a_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_rcp_ps(a) simde_mm_rcp_ps((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_rcp_ss(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_rcp_ss(a);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_rcp_ps(a));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+	r_.f32[0] = 1.0f / a_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_rcp_ss(a) simde_mm_rcp_ss((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_rsqrt_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_rsqrt_ps(a);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vrsqrteq_f32(a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_rsqrte(a_.altivec_f32);
+#elif defined(SIMDE_IEEE754_STORAGE)
+	/* https://basesandframes.files.wordpress.com/2020/04/even_faster_math_functions_green_2020.pdf
+ Pages 100 - 103 */
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+#if SIMDE_ACCURACY_PREFERENCE <= 0
+		r_.i32[i] = INT32_C(0x5F37624F) - (a_.i32[i] >> 1);
+#else
+		simde_float32 x = a_.f32[i];
+		simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
+		int32_t ix;
+
+		simde_memcpy(&ix, &x, sizeof(ix));
+
+#if SIMDE_ACCURACY_PREFERENCE == 1
+		ix = INT32_C(0x5F375A82) - (ix >> 1);
+#else
+		ix = INT32_C(0x5F37599E) - (ix >> 1);
+#endif
+
+		simde_memcpy(&x, &ix, sizeof(x));
+
+#if SIMDE_ACCURACY_PREFERENCE >= 2
+		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
+#endif
+		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
+
+		r_.f32[i] = x;
+#endif
+	}
+#elif defined(simde_math_sqrtf)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = 1.0f / simde_math_sqrtf(a_.f32[i]);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_rsqrt_ps(a) simde_mm_rsqrt_ps((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_rsqrt_ss(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_rsqrt_ss(a);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_rsqrt_ps(a));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 =
+		vsetq_lane_f32(vgetq_lane_f32(simde_mm_rsqrt_ps(a).neon_f32, 0),
+			 a_.neon_f32, 0);
+#elif defined(SIMDE_IEEE754_STORAGE)
+	{
+#if SIMDE_ACCURACY_PREFERENCE <= 0
+		r_.i32[0] = INT32_C(0x5F37624F) - (a_.i32[0] >> 1);
+#else
+		simde_float32 x = a_.f32[0];
+		simde_float32 xhalf = SIMDE_FLOAT32_C(0.5) * x;
+		int32_t ix;
+
+		simde_memcpy(&ix, &x, sizeof(ix));
+
+#if SIMDE_ACCURACY_PREFERENCE == 1
+		ix = INT32_C(0x5F375A82) - (ix >> 1);
+#else
+		ix = INT32_C(0x5F37599E) - (ix >> 1);
+#endif
+
+		simde_memcpy(&x, &ix, sizeof(x));
+
+#if SIMDE_ACCURACY_PREFERENCE >= 2
+		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
+#endif
+		x = x * (SIMDE_FLOAT32_C(1.5008909) - xhalf * x * x);
+
+		r_.f32[0] = x;
+#endif
+	}
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#elif defined(simde_math_sqrtf)
+	r_.f32[0] = 1.0f / simde_math_sqrtf(a_.f32[0]);
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_rsqrt_ss(a) simde_mm_rsqrt_ss((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sad_pu8(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sad_pu8(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint16x4_t t = vpaddl_u8(vabd_u8(a_.neon_u8, b_.neon_u8));
+	uint16_t r0 = t[0] + t[1] + t[2] + t[3];
+	r_.neon_u16 = vset_lane_u16(r0, vdup_n_u16(0), 0);
+#else
+	uint16_t sum = 0;
+
+#if defined(SIMDE_HAVE_STDLIB_H)
+	SIMDE_VECTORIZE_REDUCTION(+ : sum)
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		sum += HEDLEY_STATIC_CAST(uint8_t, abs(a_.u8[i] - b_.u8[i]));
+	}
+
+	r_.i16[0] = HEDLEY_STATIC_CAST(int16_t, sum);
+	r_.i16[1] = 0;
+	r_.i16[2] = 0;
+	r_.i16[3] = 0;
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#define simde_m_psadbw(a, b) simde_mm_sad_pu8(a, b)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sad_pu8(a, b) simde_mm_sad_pu8(a, b)
+#define _m_psadbw(a, b) simde_mm_sad_pu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_set_ss(simde_float32 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_set_ss(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vsetq_lane_f32(a, vdupq_n_f32(SIMDE_FLOAT32_C(0.0)), 0);
+#else
+	return simde_mm_set_ps(SIMDE_FLOAT32_C(0.0), SIMDE_FLOAT32_C(0.0),
+			 SIMDE_FLOAT32_C(0.0), a);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_set_ss(a) simde_mm_set_ss(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_setr_ps(simde_float32 e3, simde_float32 e2,
+			 simde_float32 e1, simde_float32 e0)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_setr_ps(e3, e2, e1, e0);
+#else
+	return simde_mm_set_ps(e0, e1, e2, e3);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_ps(e3, e2, e1, e0) simde_mm_setr_ps(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_setzero_ps(void)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_setzero_ps();
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vdupq_n_f32(SIMDE_FLOAT32_C(0.0));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	return vec_splats(SIMDE_FLOAT32_C(0.0));
+#else
+	simde__m128 r;
+	simde_memset(&r, 0, sizeof(r));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_setzero_ps() simde_mm_setzero_ps()
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_undefined_ps(void)
+{
+	simde__m128_private r_;
+
+#if defined(SIMDE_HAVE_UNDEFINED128)
+	r_.n = _mm_undefined_ps();
+#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+	r_ = simde__m128_to_private(simde_mm_setzero_ps());
+#endif
+
+	return simde__m128_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_undefined_ps() simde_mm_undefined_ps()
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_POP
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_x_mm_setone_ps(void)
+{
+	simde__m128 t = simde_mm_setzero_ps();
+	return simde_mm_cmpeq_ps(t, t);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_sfence(void)
+{
+	/* TODO: Use Hedley. */
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_sfence();
+#elif defined(__GNUC__) && \
+	((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 7))
+	__atomic_thread_fence(__ATOMIC_SEQ_CST);
+#elif !defined(__INTEL_COMPILER) && defined(__STDC_VERSION__) && \
+	(__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_ATOMICS__)
+#if defined(__GNUC__) && (__GNUC__ == 4) && (__GNUC_MINOR__ < 9)
+	__atomic_thread_fence(__ATOMIC_SEQ_CST);
+#else
+	atomic_thread_fence(memory_order_seq_cst);
+#endif
+#elif defined(_MSC_VER)
+	MemoryBarrier();
+#elif HEDLEY_HAS_EXTENSION(c_atomic)
+	__c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
+#elif defined(__GNUC__) && \
+	((__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 1))
+	__sync_synchronize();
+#elif defined(_OPENMP)
+#pragma omp critical(simde_mm_sfence_)
+	{
+	}
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sfence() simde_mm_sfence()
+#endif
+
+#define SIMDE_MM_SHUFFLE(z, y, x, w) \
+	(((z) << 6) | ((y) << 4) | ((x) << 2) | (w))
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _MM_SHUFFLE(z, y, x, w) SIMDE_MM_SHUFFLE(z, y, x, w)
+#endif
+
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
+	!defined(__PGI)
+#define simde_mm_shuffle_pi16(a, imm8) _mm_shuffle_pi16(a, imm8)
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shuffle_pi16(a, imm8) \
+	(__extension__({ \
+		const simde__m64_private simde__tmp_a_ = \
+			simde__m64_to_private(a); \
+		simde__m64_from_private((simde__m64_private){ \
+			.i16 = SIMDE_SHUFFLE_VECTOR_( \
+				16, 8, (simde__tmp_a_).i16, \
+				(simde__tmp_a_).i16, (((imm8)) & 3), \
+				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3), \
+				(((imm8) >> 6) & 3))}); \
+	}))
+#else
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_shuffle_pi16(simde__m64 a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m64_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+	for (size_t i = 0; i < sizeof(r_.i16) / sizeof(r_.i16[0]); i++) {
+		r_.i16[i] = a_.i16[(imm8 >> (i * 2)) & 3];
+	}
+
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wconditional-uninitialized")
+#pragma clang diagnostic ignored "-Wconditional-uninitialized"
+#endif
+	return simde__m64_from_private(r_);
+	HEDLEY_DIAGNOSTIC_POP
+}
+#endif
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
+	!defined(__PGI)
+#define simde_m_pshufw(a, imm8) _m_pshufw(a, imm8)
+#else
+#define simde_m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
+#endif
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_shuffle_pi16(a, imm8) simde_mm_shuffle_pi16(a, imm8)
+#define _m_pshufw(a, imm8) simde_mm_shuffle_pi16(a, imm8)
+#endif
+
+#if defined(SIMDE_X86_SSE_NATIVE) && !defined(__PGI)
+#define simde_mm_shuffle_ps(a, b, imm8) _mm_shuffle_ps(a, b, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_shuffle_ps(a, b, imm8) \
+	__extension__({ \
+		float32x4_t ret; \
+		ret = vmovq_n_f32(vgetq_lane_f32(a, (imm8) & (0x3))); \
+		ret = vsetq_lane_f32(vgetq_lane_f32(a, ((imm8) >> 2) & 0x3), \
+				 ret, 1); \
+		ret = vsetq_lane_f32(vgetq_lane_f32(b, ((imm8) >> 4) & 0x3), \
+				 ret, 2); \
+		ret = vsetq_lane_f32(vgetq_lane_f32(b, ((imm8) >> 6) & 0x3), \
+				 ret, 3); \
+	})
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shuffle_ps(a, b, imm8) \
+	(__extension__({ \
+		simde__m128_from_private((simde__m128_private){ \
+			.f32 = SIMDE_SHUFFLE_VECTOR_( \
+				32, 16, simde__m128_to_private(a).f32, \
+				simde__m128_to_private(b).f32, (((imm8)) & 3), \
+				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3) + 4, \
+				(((imm8) >> 6) & 3) + 4)}); \
+	}))
+#else
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_shuffle_ps(simde__m128 a, simde__m128 b, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.f32[0] = a_.f32[(imm8 >> 0) & 3];
+	r_.f32[1] = a_.f32[(imm8 >> 2) & 3];
+	r_.f32[2] = b_.f32[(imm8 >> 4) & 3];
+	r_.f32[3] = b_.f32[(imm8 >> 6) & 3];
+
+	return simde__m128_from_private(r_);
+}
+#endif
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_shuffle_ps(a, b, imm8) simde_mm_shuffle_ps((a), (b), imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_sqrt_ps(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_sqrt_ps(a);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vsqrtq_f32(a_.neon_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x4_t est = vrsqrteq_f32(a_.neon_f32);
+	for (int i = 0; i <= SIMDE_ACCURACY_PREFERENCE; i++) {
+		est = vmulq_f32(vrsqrtsq_f32(vmulq_f32(a_.neon_f32, est), est),
+				est);
+	}
+	r_.neon_f32 = vmulq_f32(a_.neon_f32, est);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_sqrt(a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = vec_sqrt(a_.altivec_f32);
+#elif defined(simde_math_sqrt)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < sizeof(r_.f32) / sizeof(r_.f32[0]); i++) {
+		r_.f32[i] = simde_math_sqrtf(a_.f32[i]);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sqrt_ps(a) simde_mm_sqrt_ps((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_sqrt_ss(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_sqrt_ss(a);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_sqrt_ps(a));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32_t value = vgetq_lane_f32(
+		simde__m128_to_private(simde_mm_sqrt_ps(a)).neon_f32, 0);
+	r_.neon_f32 = vsetq_lane_f32(value, a_.neon_f32, 0);
+#elif defined(simde_math_sqrtf)
+	r_.f32[0] = simde_math_sqrtf(a_.f32[0]);
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sqrt_ss(a) simde_mm_sqrt_ss((a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store_ps(simde_float32 mem_addr[4], simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_store_ps(mem_addr, a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_f32(mem_addr, a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	vec_st(a_.altivec_f32, 0, mem_addr);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	wasm_v128_store(mem_addr, a_.wasm_v128);
+#else
+	simde_memcpy(mem_addr, &a_, sizeof(a));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_store_ps(mem_addr, a) \
+	simde_mm_store_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store1_ps(simde_float32 mem_addr[4], simde__m128 a)
+{
+	simde_float32 *mem_addr_ =
+		SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128);
+
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_store_ps1(mem_addr_, a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_f32(mem_addr_, vdupq_lane_f32(vget_low_f32(a_.neon_f32), 0));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	wasm_v128_store(mem_addr_,
+			wasm_v32x4_shuffle(a_.wasm_v128, a_.wasm_v128, 0, 0, 0,
+					 0));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	vec_st(vec_splat(a_.altivec_f32, 0), 0, mem_addr_);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	simde__m128_private tmp_;
+	tmp_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 0, 0, 0, 0);
+	simde_mm_store_ps(mem_addr_, tmp_.f32);
+#else
+	SIMDE_VECTORIZE_ALIGNED(mem_addr_ : 16)
+	for (size_t i = 0; i < sizeof(a_.f32) / sizeof(a_.f32[0]); i++) {
+		mem_addr_[i] = a_.f32[0];
+	}
+#endif
+#endif
+}
+#define simde_mm_store_ps1(mem_addr, a) simde_mm_store1_ps(mem_addr, a)
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_store_ps1(mem_addr, a) \
+	simde_mm_store1_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#define _mm_store1_ps(mem_addr, a) \
+	simde_mm_store1_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store_ss(simde_float32 *mem_addr, simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_store_ss(mem_addr, a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_lane_f32(mem_addr, a_.neon_f32, 0);
+#else
+	*mem_addr = a_.f32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_store_ss(mem_addr, a) \
+	simde_mm_store_ss(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeh_pi(simde__m64 *mem_addr, simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_storeh_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1_f32(HEDLEY_REINTERPRET_CAST(float32_t *, mem_addr),
+		 vget_high_f32(a_.neon_f32));
+#else
+	simde_memcpy(mem_addr, &(a_.m64[1]), sizeof(a_.m64[1]));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_storeh_pi(mem_addr, a) simde_mm_storeh_pi(mem_addr, (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storel_pi(simde__m64 *mem_addr, simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_storel_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
+#else
+	simde__m64_private *dest_ =
+		HEDLEY_REINTERPRET_CAST(simde__m64_private *, mem_addr);
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	dest_->neon_f32 = vget_low_f32(a_.neon_f32);
+#else
+	dest_->f32[0] = a_.f32[0];
+	dest_->f32[1] = a_.f32[1];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_storel_pi(mem_addr, a) simde_mm_storel_pi(mem_addr, (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storer_ps(simde_float32 mem_addr[4], simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_storer_ps(mem_addr, a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	vec_st(vec_reve(a_.altivec_f32), 0, mem_addr);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x4_t tmp = vrev64q_f32(a_.neon_f32);
+	vst1q_f32(mem_addr, vextq_f32(tmp, tmp, 2));
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	a_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, a_.f32, 3, 2, 1, 0);
+	simde_mm_store_ps(mem_addr, simde__m128_from_private(a_));
+#else
+	SIMDE_VECTORIZE_ALIGNED(mem_addr : 16)
+	for (size_t i = 0; i < sizeof(a_.f32) / sizeof(a_.f32[0]); i++) {
+		mem_addr[i] =
+			a_.f32[((sizeof(a_.f32) / sizeof(a_.f32[0])) - 1) - i];
+	}
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_storer_ps(mem_addr, a) \
+	simde_mm_storer_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_ps(simde_float32 mem_addr[4], simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_storeu_ps(mem_addr, a);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_f32(mem_addr, a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	vec_vsx_st(a_.altivec_f32, 0, mem_addr);
+#else
+	simde_memcpy(mem_addr, &a_, sizeof(a_));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_ps(mem_addr, a) \
+	simde_mm_storeu_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_sub_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_sub_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vsubq_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_sub(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_sub(a_.altivec_f32, b_.altivec_f32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f32 = a_.f32 - b_.f32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = a_.f32[i] - b_.f32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_ps(a, b) simde_mm_sub_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_sub_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_sub_ss(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_ss(a, simde_mm_sub_ps(a, b));
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+	r_.f32[0] = a_.f32[0] - b_.f32[0];
+	r_.f32[1] = a_.f32[1];
+	r_.f32[2] = a_.f32[2];
+	r_.f32[3] = a_.f32[3];
+
+	return simde__m128_from_private(r_);
+#endif
+}
+
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_ss(a, b) simde_mm_sub_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomieq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomieq_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_eq_b = vceqq_f32(a_.neon_f32, b_.neon_f32);
+	r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_eq_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] == b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] == b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomieq_ss(a, b) simde_mm_ucomieq_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomige_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomige_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_ge_b = vcgeq_f32(a_.neon_f32, b_.neon_f32);
+	r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_ge_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] >= b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] >= b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomige_ss(a, b) simde_mm_ucomige_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomigt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomigt_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_gt_b = vcgtq_f32(a_.neon_f32, b_.neon_f32);
+	r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_gt_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] > b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] > b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomigt_ss(a, b) simde_mm_ucomigt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomile_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomile_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_le_b = vcleq_f32(a_.neon_f32, b_.neon_f32);
+	r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_le_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] <= b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] <= b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomile_ss(a, b) simde_mm_ucomile_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomilt_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomilt_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_or_b_nan = vmvnq_u32(vandq_u32(a_not_nan, b_not_nan));
+	uint32x4_t a_lt_b = vcltq_f32(a_.neon_f32, b_.neon_f32);
+	r = !!(vgetq_lane_u32(vorrq_u32(a_or_b_nan, a_lt_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] < b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] < b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomilt_ss(a, b) simde_mm_ucomilt_ss((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomineq_ss(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_ucomineq_ss(a, b);
+#else
+	simde__m128_private a_ = simde__m128_to_private(a),
+			 b_ = simde__m128_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x4_t a_not_nan = vceqq_f32(a_.neon_f32, a_.neon_f32);
+	uint32x4_t b_not_nan = vceqq_f32(b_.neon_f32, b_.neon_f32);
+	uint32x4_t a_and_b_not_nan = vandq_u32(a_not_nan, b_not_nan);
+	uint32x4_t a_neq_b = vmvnq_u32(vceqq_f32(a_.neon_f32, b_.neon_f32));
+	r = !!(vgetq_lane_u32(vandq_u32(a_and_b_not_nan, a_neq_b), 0) != 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f32[0] != b_.f32[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f32[0] != b_.f32[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomineq_ss(a, b) simde_mm_ucomineq_ss((a), (b))
+#endif
+
+#if defined(SIMDE_X86_SSE_NATIVE)
+#if defined(__has_builtin)
+#if __has_builtin(__builtin_ia32_undef128)
+#define SIMDE_HAVE_UNDEFINED128
+#endif
+#elif !defined(__PGI) && !defined(SIMDE_BUG_GCC_REV_208793) && \
+	!defined(_MSC_VER)
+#define SIMDE_HAVE_UNDEFINED128
+#endif
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_unpackhi_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_unpackhi_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vzip2q_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x2_t a1 = vget_high_f32(a_.neon_f32);
+	float32x2_t b1 = vget_high_f32(b_.neon_f32);
+	float32x2x2_t result = vzip_f32(a1, b1);
+	r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 2, 6, 3, 7);
+#else
+	r_.f32[0] = a_.f32[2];
+	r_.f32[1] = b_.f32[2];
+	r_.f32[2] = a_.f32[3];
+	r_.f32[3] = b_.f32[3];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_ps(a, b) simde_mm_unpackhi_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_unpacklo_ps(simde__m128 a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return _mm_unpacklo_ps(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a),
+				b_ = simde__m128_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vzip1q_f32(a_.neon_f32, b_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f32 = vec_mergeh(a_.altivec_f32, b_.altivec_f32);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.f32, b_.f32, 0, 4, 1, 5);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	float32x2_t a1 = vget_low_f32(a_.neon_f32);
+	float32x2_t b1 = vget_low_f32(b_.neon_f32);
+	float32x2x2_t result = vzip_f32(a1, b1);
+	r_.neon_f32 = vcombine_f32(result.val[0], result.val[1]);
+#else
+	r_.f32[0] = a_.f32[0];
+	r_.f32[1] = b_.f32[0];
+	r_.f32[2] = a_.f32[1];
+	r_.f32[3] = b_.f32[1];
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_ps(a, b) simde_mm_unpacklo_ps((a), (b))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_pi(simde__m64 *mem_addr, simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	_mm_stream_pi(HEDLEY_REINTERPRET_CAST(__m64 *, mem_addr), a);
+#else
+	simde__m64_private *dest = HEDLEY_REINTERPRET_CAST(simde__m64_private *,
+							 mem_addr),
+			 a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	dest->i64[0] = vget_lane_s64(a_.neon_i64, 0);
+#else
+	dest->i64[0] = a_.i64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_pi(mem_addr, a) simde_mm_stream_pi(mem_addr, (a))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_ps(simde_float32 mem_addr[4], simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	_mm_stream_ps(mem_addr, a);
+#elif HEDLEY_HAS_BUILTIN(__builtin_nontemporal_store) && \
+	defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	simde__m128_private a_ = simde__m128_to_private(a);
+	__builtin_nontemporal_store(
+		a_.f32, SIMDE_ALIGN_CAST(__typeof__(a_.f32) *, mem_addr));
+#else
+	simde_mm_store_ps(mem_addr, a);
+#endif
+}
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_ps(mem_addr, a) \
+	simde_mm_stream_ps(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 float *, simde_float32 *, mem_addr), \
+			 (a))
+#endif
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
+	do { \
+		float32x4x2_t ROW01 = vtrnq_f32(row0, row1); \
+		float32x4x2_t ROW23 = vtrnq_f32(row2, row3); \
+		row0 = vcombine_f32(vget_low_f32(ROW01.val[0]), \
+				 vget_low_f32(ROW23.val[0])); \
+		row1 = vcombine_f32(vget_low_f32(ROW01.val[1]), \
+				 vget_low_f32(ROW23.val[1])); \
+		row2 = vcombine_f32(vget_high_f32(ROW01.val[0]), \
+				 vget_high_f32(ROW23.val[0])); \
+		row3 = vcombine_f32(vget_high_f32(ROW01.val[1]), \
+				 vget_high_f32(ROW23.val[1])); \
+	} while (0)
+#else
+#define SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
+	do { \
+		simde__m128 tmp3, tmp2, tmp1, tmp0; \
+		tmp0 = simde_mm_unpacklo_ps((row0), (row1)); \
+		tmp2 = simde_mm_unpacklo_ps((row2), (row3)); \
+		tmp1 = simde_mm_unpackhi_ps((row0), (row1)); \
+		tmp3 = simde_mm_unpackhi_ps((row2), (row3)); \
+		row0 = simde_mm_movelh_ps(tmp0, tmp2); \
+		row1 = simde_mm_movehl_ps(tmp2, tmp0); \
+		row2 = simde_mm_movelh_ps(tmp1, tmp3); \
+		row3 = simde_mm_movehl_ps(tmp3, tmp1); \
+	} while (0)
+#endif
+#if defined(SIMDE_X86_SSE_ENABLE_NATIVE_ALIASES)
+#define _MM_TRANSPOSE4_PS(row0, row1, row2, row3) \
+	SIMDE_MM_TRANSPOSE4_PS(row0, row1, row2, row3)
+#endif
+
+#if defined(_MM_EXCEPT_INVALID)
+#define SIMDE_MM_EXCEPT_INVALID _MM_EXCEPT_INVALID
+#else
+#define SIMDE_MM_EXCEPT_INVALID (0x0001)
+#endif
+#if defined(_MM_EXCEPT_DENORM)
+#define SIMDE_MM_EXCEPT_DENORM _MM_EXCEPT_DENORM
+#else
+#define SIMDE_MM_EXCEPT_DENORM (0x0002)
+#endif
+#if defined(_MM_EXCEPT_DIV_ZERO)
+#define SIMDE_MM_EXCEPT_DIV_ZERO _MM_EXCEPT_DIV_ZERO
+#else
+#define SIMDE_MM_EXCEPT_DIV_ZERO (0x0004)
+#endif
+#if defined(_MM_EXCEPT_OVERFLOW)
+#define SIMDE_MM_EXCEPT_OVERFLOW _MM_EXCEPT_OVERFLOW
+#else
+#define SIMDE_MM_EXCEPT_OVERFLOW (0x0008)
+#endif
+#if defined(_MM_EXCEPT_UNDERFLOW)
+#define SIMDE_MM_EXCEPT_UNDERFLOW _MM_EXCEPT_UNDERFLOW
+#else
+#define SIMDE_MM_EXCEPT_UNDERFLOW (0x0010)
+#endif
+#if defined(_MM_EXCEPT_INEXACT)
+#define SIMDE_MM_EXCEPT_INEXACT _MM_EXCEPT_INEXACT
+#else
+#define SIMDE_MM_EXCEPT_INEXACT (0x0020)
+#endif
+#if defined(_MM_EXCEPT_MASK)
+#define SIMDE_MM_EXCEPT_MASK _MM_EXCEPT_MASK
+#else
+#define SIMDE_MM_EXCEPT_MASK \
+	(SIMDE_MM_EXCEPT_INVALID | SIMDE_MM_EXCEPT_DENORM | \
+	 SIMDE_MM_EXCEPT_DIV_ZERO | SIMDE_MM_EXCEPT_OVERFLOW | \
+	 SIMDE_MM_EXCEPT_UNDERFLOW | SIMDE_MM_EXCEPT_INEXACT)
+#endif
+
+#if defined(_MM_MASK_INVALID)
+#define SIMDE_MM_MASK_INVALID _MM_MASK_INVALID
+#else
+#define SIMDE_MM_MASK_INVALID (0x0080)
+#endif
+#if defined(_MM_MASK_DENORM)
+#define SIMDE_MM_MASK_DENORM _MM_MASK_DENORM
+#else
+#define SIMDE_MM_MASK_DENORM (0x0100)
+#endif
+#if defined(_MM_MASK_DIV_ZERO)
+#define SIMDE_MM_MASK_DIV_ZERO _MM_MASK_DIV_ZERO
+#else
+#define SIMDE_MM_MASK_DIV_ZERO (0x0200)
+#endif
+#if defined(_MM_MASK_OVERFLOW)
+#define SIMDE_MM_MASK_OVERFLOW _MM_MASK_OVERFLOW
+#else
+#define SIMDE_MM_MASK_OVERFLOW (0x0400)
+#endif
+#if defined(_MM_MASK_UNDERFLOW)
+#define SIMDE_MM_MASK_UNDERFLOW _MM_MASK_UNDERFLOW
+#else
+#define SIMDE_MM_MASK_UNDERFLOW (0x0800)
+#endif
+#if defined(_MM_MASK_INEXACT)
+#define SIMDE_MM_MASK_INEXACT _MM_MASK_INEXACT
+#else
+#define SIMDE_MM_MASK_INEXACT (0x1000)
+#endif
+#if defined(_MM_MASK_MASK)
+#define SIMDE_MM_MASK_MASK _MM_MASK_MASK
+#else
+#define SIMDE_MM_MASK_MASK \
+	(SIMDE_MM_MASK_INVALID | SIMDE_MM_MASK_DENORM | \
+	 SIMDE_MM_MASK_DIV_ZERO | SIMDE_MM_MASK_OVERFLOW | \
+	 SIMDE_MM_MASK_UNDERFLOW | SIMDE_MM_MASK_INEXACT)
+#endif
+
+#if defined(_MM_FLUSH_ZERO_MASK)
+#define SIMDE_MM_FLUSH_ZERO_MASK _MM_FLUSH_ZERO_MASK
+#else
+#define SIMDE_MM_FLUSH_ZERO_MASK (0x8000)
+#endif
+#if defined(_MM_FLUSH_ZERO_ON)
+#define SIMDE_MM_FLUSH_ZERO_ON _MM_FLUSH_ZERO_ON
+#else
+#define SIMDE_MM_FLUSH_ZERO_ON (0x8000)
+#endif
+#if defined(_MM_FLUSH_ZERO_OFF)
+#define SIMDE_MM_FLUSH_ZERO_OFF _MM_FLUSH_ZERO_OFF
+#else
+#define SIMDE_MM_FLUSH_ZERO_OFF (0x0000)
+#endif
+
+SIMDE_END_DECLS_
+
+HEDLEY_DIAGNOSTIC_POP
+
+#endif /* !defined(SIMDE_X86_SSE_H) */

obs-studio-26.1.1.tar.xz/libobs/util/simde/x86/sse2.h Added

@@ -0,0 +1,7549 @@
+/* SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy,
+ * modify, merge, publish, distribute, sublicense, and/or sell copies
+ * of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Copyright:
+ * 2017-2020 Evan Nemerson <evan@nemerson.com>
+ * 2015-2017 John W. Ratcliff <jratcliffscarab@gmail.com>
+ * 2015 Brandon Rowlett <browlett@nvidia.com>
+ * 2015 Ken Fast <kfast@gdeb.com>
+ * 2017 Hasindu Gamaarachchi <hasindu@unsw.edu.au>
+ * 2018 Jeff Daily <jeff.daily@amd.com>
+ */
+
+#if !defined(SIMDE_X86_SSE2_H)
+#define SIMDE_X86_SSE2_H
+
+#include "sse.h"
+
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DISABLE_UNWANTED_DIAGNOSTICS
+SIMDE_BEGIN_DECLS_
+
+typedef union {
+#if defined(SIMDE_VECTOR_SUBSCRIPT)
+	SIMDE_ALIGN_TO_16 int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#if defined(SIMDE_HAVE_INT128_)
+	SIMDE_ALIGN_TO_16 simde_int128 i128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 simde_uint128 u128 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#endif
+	SIMDE_ALIGN_TO_16 simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+	SIMDE_ALIGN_TO_16 int8_t i8[16];
+	SIMDE_ALIGN_TO_16 int16_t i16[8];
+	SIMDE_ALIGN_TO_16 int32_t i32[4];
+	SIMDE_ALIGN_TO_16 int64_t i64[2];
+	SIMDE_ALIGN_TO_16 uint8_t u8[16];
+	SIMDE_ALIGN_TO_16 uint16_t u16[8];
+	SIMDE_ALIGN_TO_16 uint32_t u32[4];
+	SIMDE_ALIGN_TO_16 uint64_t u64[2];
+#if defined(SIMDE_HAVE_INT128_)
+	SIMDE_ALIGN_TO_16 simde_int128 i128[1];
+	SIMDE_ALIGN_TO_16 simde_uint128 u128[1];
+#endif
+	SIMDE_ALIGN_TO_16 simde_float32 f32[4];
+	SIMDE_ALIGN_TO_16 simde_float64 f64[2];
+
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f[16 / sizeof(int_fast32_t)];
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
+#endif
+
+	SIMDE_ALIGN_TO_16 simde__m64_private m64_private[2];
+	SIMDE_ALIGN_TO_16 simde__m64 m64[2];
+
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	SIMDE_ALIGN_TO_16 __m128i n;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_TO_16 int8x16_t neon_i8;
+	SIMDE_ALIGN_TO_16 int16x8_t neon_i16;
+	SIMDE_ALIGN_TO_16 int32x4_t neon_i32;
+	SIMDE_ALIGN_TO_16 int64x2_t neon_i64;
+	SIMDE_ALIGN_TO_16 uint8x16_t neon_u8;
+	SIMDE_ALIGN_TO_16 uint16x8_t neon_u16;
+	SIMDE_ALIGN_TO_16 uint32x4_t neon_u32;
+	SIMDE_ALIGN_TO_16 uint64x2_t neon_u64;
+	SIMDE_ALIGN_TO_16 float32x4_t neon_f32;
+#if defined(SIMDE_ARCH_AARCH64)
+	SIMDE_ALIGN_TO_16 float64x2_t neon_f64;
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	SIMDE_ALIGN_TO_16 v128_t wasm_v128;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
+#if defined(__UINT_FAST32_TYPE__) && defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f;
+#else
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f;
+#endif
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
+#if defined(__UINT_FAST32_TYPE__) && defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(__UINT_FAST32_TYPE__) altivec_u32f;
+#else
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f;
+#endif
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
+#endif
+#endif
+} simde__m128i_private;
+
+typedef union {
+#if defined(SIMDE_VECTOR_SUBSCRIPT)
+	SIMDE_ALIGN_TO_16 int8_t i8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int16_t i16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int32_t i32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int64_t i64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint8_t u8 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint16_t u16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint32_t u32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint64_t u64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 simde_float32 f32 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 simde_float64 f64 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+	SIMDE_ALIGN_TO_16 int8_t i8[16];
+	SIMDE_ALIGN_TO_16 int16_t i16[8];
+	SIMDE_ALIGN_TO_16 int32_t i32[4];
+	SIMDE_ALIGN_TO_16 int64_t i64[2];
+	SIMDE_ALIGN_TO_16 uint8_t u8[16];
+	SIMDE_ALIGN_TO_16 uint16_t u16[8];
+	SIMDE_ALIGN_TO_16 uint32_t u32[4];
+	SIMDE_ALIGN_TO_16 uint64_t u64[2];
+	SIMDE_ALIGN_TO_16 simde_float32 f32[4];
+	SIMDE_ALIGN_TO_16 simde_float64 f64[2];
+	SIMDE_ALIGN_TO_16 int_fast32_t i32f[16 / sizeof(int_fast32_t)];
+	SIMDE_ALIGN_TO_16 uint_fast32_t u32f[16 / sizeof(uint_fast32_t)];
+#endif
+
+	SIMDE_ALIGN_TO_16 simde__m64_private m64_private[2];
+	SIMDE_ALIGN_TO_16 simde__m64 m64[2];
+
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	SIMDE_ALIGN_TO_16 __m128d n;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_TO_16 int8x16_t neon_i8;
+	SIMDE_ALIGN_TO_16 int16x8_t neon_i16;
+	SIMDE_ALIGN_TO_16 int32x4_t neon_i32;
+	SIMDE_ALIGN_TO_16 int64x2_t neon_i64;
+	SIMDE_ALIGN_TO_16 uint8x16_t neon_u8;
+	SIMDE_ALIGN_TO_16 uint16x8_t neon_u16;
+	SIMDE_ALIGN_TO_16 uint32x4_t neon_u32;
+	SIMDE_ALIGN_TO_16 uint64x2_t neon_u64;
+	SIMDE_ALIGN_TO_16 float32x4_t neon_f32;
+#if defined(SIMDE_ARCH_AARCH64)
+	SIMDE_ALIGN_TO_16 float64x2_t neon_f64;
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	SIMDE_ALIGN_TO_16 v128_t wasm_v128;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed char) altivec_i8;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed short) altivec_i16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32;
+#if defined(__INT_FAST32_TYPE__) && defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(__INT_FAST32_TYPE__) altivec_i32f;
+#else
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(signed int) altivec_i32f;
+#endif
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned char) altivec_u8;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned short) altivec_u16;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32;
+#if defined(__UINT_FAST32_TYPE__) && defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(__UINT_FAST32_TYPE__) altivec_u32f;
+#else
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int) altivec_u32f;
+#endif
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(float) altivec_f32;
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(signed long long) altivec_i64;
+	SIMDE_ALIGN_TO_16
+	SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long) altivec_u64;
+	SIMDE_ALIGN_TO_16 SIMDE_POWER_ALTIVEC_VECTOR(double) altivec_f64;
+#endif
+#endif
+} simde__m128d_private;
+
+#if defined(SIMDE_X86_SSE2_NATIVE)
+typedef __m128i simde__m128i;
+typedef __m128d simde__m128d;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+typedef int64x2_t simde__m128i;
+#if defined(SIMDE_ARCH_AARCH64)
+typedef float64x2_t simde__m128d;
+#elif defined(SIMDE_VECTOR_SUBSCRIPT)
+typedef simde_float64 simde__m128d SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+typedef simde__m128d_private simde__m128d;
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+typedef v128_t simde__m128i;
+typedef v128_t simde__m128d;
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+typedef SIMDE_POWER_ALTIVEC_VECTOR(float) simde__m128i;
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+typedef SIMDE_POWER_ALTIVEC_VECTOR(double) simde__m128d;
+#else
+typedef simde__m128d_private simde__m128d;
+#endif
+#elif defined(SIMDE_VECTOR_SUBSCRIPT)
+typedef int64_t simde__m128i SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+typedef simde_float64
+	simde__m128d SIMDE_ALIGN_TO_16 SIMDE_VECTOR(16) SIMDE_MAY_ALIAS;
+#else
+typedef simde__m128i_private simde__m128i;
+typedef simde__m128d_private simde__m128d;
+#endif
+
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+typedef simde__m128i __m128i;
+typedef simde__m128d __m128d;
+#endif
+
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i), "simde__m128i size incorrect");
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128i_private),
+		 "simde__m128i_private size incorrect");
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d), "simde__m128d size incorrect");
+HEDLEY_STATIC_ASSERT(16 == sizeof(simde__m128d_private),
+		 "simde__m128d_private size incorrect");
+#if defined(SIMDE_CHECK_ALIGNMENT) && defined(SIMDE_ALIGN_OF)
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i) == 16,
+		 "simde__m128i is not 16-byte aligned");
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128i_private) == 16,
+		 "simde__m128i_private is not 16-byte aligned");
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d) == 16,
+		 "simde__m128d is not 16-byte aligned");
+HEDLEY_STATIC_ASSERT(SIMDE_ALIGN_OF(simde__m128d_private) == 16,
+		 "simde__m128d_private is not 16-byte aligned");
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde__m128i_from_private(simde__m128i_private v)
+{
+	simde__m128i r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i_private simde__m128i_to_private(simde__m128i v)
+{
+	simde__m128i_private r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde__m128d_from_private(simde__m128d_private v)
+{
+	simde__m128d r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d_private simde__m128d_to_private(simde__m128d v)
+{
+	simde__m128d_private r;
+	simde_memcpy(&r, &v, sizeof(r));
+	return r;
+}
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int8x16_t, neon, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int16x8_t, neon, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int32x4_t, neon, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, int64x2_t, neon, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint8x16_t, neon, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint16x8_t, neon, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint32x4_t, neon, u32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, uint64x2_t, neon, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float32x4_t, neon, f32)
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, float64x2_t, neon, f64)
+#endif
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+				 altivec, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+				 altivec, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+				 altivec, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned char), altivec, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned short), altivec, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i,
+				 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int),
+				 altivec, u32)
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128i, SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long), altivec, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128i, SIMDE_POWER_ALTIVEC_VECTOR(signed long long), altivec, i64)
+#endif
+#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int8x16_t, neon, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int16x8_t, neon, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int32x4_t, neon, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, int64x2_t, neon, i64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint8x16_t, neon, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint16x8_t, neon, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint32x4_t, neon, u32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, uint64x2_t, neon, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float32x4_t, neon, f32)
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, float64x2_t, neon, f64)
+#endif
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+				 altivec, i8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+				 altivec, i16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d,
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+				 altivec, i32)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned char), altivec, u8)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned short), altivec, u16)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d,
+				 SIMDE_POWER_ALTIVEC_VECTOR(unsigned int),
+				 altivec, u32)
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128d, SIMDE_POWER_ALTIVEC_VECTOR(unsigned long long), altivec, u64)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(
+	m128d, SIMDE_POWER_ALTIVEC_VECTOR(signed long long), altivec, i64)
+#if defined(SIMDE_BUG_GCC_95782)
+SIMDE_FUNCTION_ATTRIBUTES
+SIMDE_POWER_ALTIVEC_VECTOR(double)
+simde__m128d_to_altivec_f64(simde__m128d value)
+{
+	simde__m128d_private r_ = simde__m128d_to_private(value);
+	return r_.altivec_f64;
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde__m128d_from_altivec_f64(SIMDE_POWER_ALTIVEC_VECTOR(double)
+						 value)
+{
+	simde__m128d_private r_;
+	r_.altivec_f64 = value;
+	return simde__m128d_from_private(r_);
+}
+#else
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d,
+				 SIMDE_POWER_ALTIVEC_VECTOR(double),
+				 altivec, f64)
+#endif
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128d, v128_t, wasm, v128);
+SIMDE_X86_GENERATE_CONVERSION_FUNCTION(m128i, v128_t, wasm, v128);
+#endif /* defined(SIMDE_ARM_NEON_A32V7_NATIVE) */
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_set_pd(simde_float64 e1, simde_float64 e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_pd(e1, e0);
+#else
+	simde__m128d_private r_;
+
+#if defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_make(e0, e1);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	SIMDE_ALIGN_TO_16 simde_float64 data[2] = {e0, e1};
+	r_.neon_f64 = vld1q_f64(data);
+#else
+	r_.f64[0] = e0;
+	r_.f64[1] = e1;
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_pd(e1, e0) simde_mm_set_pd(e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_set1_pd(simde_float64 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set1_pd(a);
+#else
+	simde__m128d_private r_;
+
+#if defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_splat(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vdupq_n_f64(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f64 = vec_splats(HEDLEY_STATIC_CAST(double, a));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.f64[i] = a;
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#define simde_mm_set_pd1(a) simde_mm_set1_pd(a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_pd(a) simde_mm_set1_pd(a)
+#define _mm_set_pd1(a) simde_mm_set1_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_abs_pd(simde__m128d a)
+{
+#if defined(SIMDE_X86_AVX512F_NATIVE) && \
+	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(7, 4, 0))
+	return _mm512_castpd512_pd128(_mm512_abs_pd(_mm512_castpd128_pd512(a)));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V8_NATIVE)
+	r_.neon_f32 = vabsq_f32(a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f32 = vec_abs(a_.altivec_f32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = simde_math_fabs(a_.f64[i]);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_not_pd(simde__m128d a)
+{
+#if defined(SIMDE_X86_AVX512VL_NATIVE)
+	__m128i ai = _mm_castpd_si128(a);
+	return _mm_castsi128_pd(_mm_ternarylogic_epi64(ai, ai, ai, 0x55));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vmvnq_s32(a_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f64 = vec_nor(a_.altivec_f64, a_.altivec_f64);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_nor(a_.altivec_i32, a_.altivec_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_not(a_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = ~a_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = ~(a_.i32f[i]);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_select_pd(simde__m128d a, simde__m128d b,
+				 simde__m128d mask)
+{
+/* This function is for when you want to blend two elements together
+ * according to a mask. It is similar to _mm_blendv_pd, except that
+ * it is undefined whether the blend is based on the highest bit in
+ * each lane (like blendv) or just bitwise operations. This allows
+ * us to implement the function efficiently everywhere.
+ *
+ * Basically, you promise that all the lanes in mask are either 0 or
+ * ~0. */
+#if defined(SIMDE_X86_SSE4_1_NATIVE)
+	return _mm_blendv_pd(a, b, mask);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b),
+				 mask_ = simde__m128d_to_private(mask);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 ^ ((a_.i64 ^ b_.i64) & mask_.i64);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vbslq_s64(mask_.neon_u64, b_.neon_i64, a_.neon_i64);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] ^
+			 ((a_.i64[i] ^ b_.i64[i]) & mask_.i64[i]);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_add_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vaddq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i8 = vec_add(a_.altivec_i8, b_.altivec_i8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = a_.i8 + b_.i8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = a_.i8[i] + b_.i8[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_epi8(a, b) simde_mm_add_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_add_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vaddq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = vec_add(a_.altivec_i16, b_.altivec_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = a_.i16 + b_.i16;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] + b_.i16[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_epi16(a, b) simde_mm_add_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_add_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vaddq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_add(a_.altivec_i32, b_.altivec_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 + b_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] + b_.i32[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_epi32(a, b) simde_mm_add_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_add_epi64(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_epi64(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vaddq_s64(a_.neon_i64, b_.neon_i64);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_i64 = vec_add(a_.altivec_i64, b_.altivec_i64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i64x2_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 + b_.i64;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] + b_.i64[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_epi64(a, b) simde_mm_add_epi64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_add_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vaddq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f64 = vec_add(a_.altivec_f64, b_.altivec_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_add(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f64 = a_.f64 + b_.f64;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = a_.f64[i] + b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_pd(a, b) simde_mm_add_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_move_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_move_sd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 =
+		vsetq_lane_f64(vgetq_lane_f64(b_.neon_f64, 0), a_.neon_f64, 0);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+#if defined(HEDLEY_IBM_VERSION)
+	r_.altivec_f64 = vec_xxpermdi(a_.altivec_f64, b_.altivec_f64, 1);
+#else
+	r_.altivec_f64 = vec_xxpermdi(b_.altivec_f64, a_.altivec_f64, 1);
+#endif
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v64x2_shuffle(a_.wasm_v128, b_.wasm_v128, 2, 1);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 2, 1);
+#else
+	r_.f64[0] = b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_move_sd(a, b) simde_mm_move_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_add_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_add_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_add_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.f64[0] = a_.f64[0] + b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_sd(a, b) simde_mm_add_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_add_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_add_si64(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vadd_s64(a_.neon_i64, b_.neon_i64);
+#else
+	r_.i64[0] = a_.i64[0] + b_.i64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_add_si64(a, b) simde_mm_add_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_adds_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_adds_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vqaddq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_add_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i8 = vec_adds(a_.altivec_i8, b_.altivec_i8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		const int_fast16_t tmp =
+			HEDLEY_STATIC_CAST(int_fast16_t, a_.i8[i]) +
+			HEDLEY_STATIC_CAST(int_fast16_t, b_.i8[i]);
+		r_.i8[i] = HEDLEY_STATIC_CAST(
+			int8_t,
+			((tmp < INT8_MAX) ? ((tmp > INT8_MIN) ? tmp : INT8_MIN)
+					 : INT8_MAX));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_epi8(a, b) simde_mm_adds_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_adds_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_adds_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vqaddq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_add_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = vec_adds(a_.altivec_i16, b_.altivec_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		const int_fast32_t tmp =
+			HEDLEY_STATIC_CAST(int_fast32_t, a_.i16[i]) +
+			HEDLEY_STATIC_CAST(int_fast32_t, b_.i16[i]);
+		r_.i16[i] = HEDLEY_STATIC_CAST(
+			int16_t,
+			((tmp < INT16_MAX)
+				 ? ((tmp > INT16_MIN) ? tmp : INT16_MIN)
+				 : INT16_MAX));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_epi16(a, b) simde_mm_adds_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_adds_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_adds_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vqaddq_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u8x16_add_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_u8 = vec_adds(a_.altivec_u8, b_.altivec_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = ((UINT8_MAX - a_.u8[i]) > b_.u8[i])
+				 ? (a_.u8[i] + b_.u8[i])
+				 : UINT8_MAX;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_epu8(a, b) simde_mm_adds_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_adds_epu16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_adds_epu16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vqaddq_u16(a_.neon_u16, b_.neon_u16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u16x8_add_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u16 = vec_adds(a_.altivec_u16, b_.altivec_u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = ((UINT16_MAX - a_.u16[i]) > b_.u16[i])
+				 ? (a_.u16[i] + b_.u16[i])
+				 : UINT16_MAX;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_adds_epu16(a, b) simde_mm_adds_epu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_and_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_and_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vandq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_and(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f64 = vec_and(a_.altivec_f64, b_.altivec_f64);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f & b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] & b_.i32f[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_and_pd(a, b) simde_mm_and_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_and_si128(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_and_si128(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vandq_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u32f = vec_and(a_.altivec_u32f, b_.altivec_u32f);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f & b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] & b_.i32f[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_and_si128(a, b) simde_mm_and_si128(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_andnot_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_andnot_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_andnot(b_.wasm_v128, a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f64 = vec_andc(b_.altivec_f64, a_.altivec_f64);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32f = vec_andc(b_.altivec_i32f, a_.altivec_i32f);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = ~a_.i32f & b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
+		r_.u64[i] = ~a_.u64[i] & b_.u64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_andnot_pd(a, b) simde_mm_andnot_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_andnot_si128(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_andnot_si128(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vbicq_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_andc(b_.altivec_i32, a_.altivec_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = ~a_.i32f & b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = ~(a_.i32f[i]) & b_.i32f[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_andnot_si128(a, b) simde_mm_andnot_si128(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_xor_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_xor_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f ^ b_.i32f;
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_xor(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = veorq_s64(a_.neon_i64, b_.neon_i64);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] ^ b_.i32f[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_xor_pd(a, b) simde_mm_xor_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_avg_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_avg_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vrhaddq_u8(b_.neon_u8, a_.neon_u8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u8x16_avgr(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u8 = vec_avg(a_.altivec_u8, b_.altivec_u8);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
+	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_CONVERT_VECTOR_)
+	uint16_t wa SIMDE_VECTOR(32);
+	uint16_t wb SIMDE_VECTOR(32);
+	uint16_t wr SIMDE_VECTOR(32);
+	SIMDE_CONVERT_VECTOR_(wa, a_.u8);
+	SIMDE_CONVERT_VECTOR_(wb, b_.u8);
+	wr = (wa + wb + 1) >> 1;
+	SIMDE_CONVERT_VECTOR_(r_.u8, wr);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] + b_.u8[i] + 1) >> 1;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_avg_epu8(a, b) simde_mm_avg_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_avg_epu16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_avg_epu16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vrhaddq_u16(b_.neon_u16, a_.neon_u16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u16x8_avgr(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u16 = vec_avg(a_.altivec_u16, b_.altivec_u16);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS) && \
+	defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && \
+	defined(SIMDE_CONVERT_VECTOR_)
+	uint32_t wa SIMDE_VECTOR(32);
+	uint32_t wb SIMDE_VECTOR(32);
+	uint32_t wr SIMDE_VECTOR(32);
+	SIMDE_CONVERT_VECTOR_(wa, a_.u16);
+	SIMDE_CONVERT_VECTOR_(wb, b_.u16);
+	wr = (wa + wb + 1) >> 1;
+	SIMDE_CONVERT_VECTOR_(r_.u16, wr);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = (a_.u16[i] + b_.u16[i] + 1) >> 1;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_avg_epu16(a, b) simde_mm_avg_epu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_setzero_si128(void)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setzero_si128();
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vdupq_n_s32(0);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_splats(HEDLEY_STATIC_CAST(signed int, 0));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_splat(INT32_C(0));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT)
+	r_.i32 = __extension__(__typeof__(r_.i32)){0, 0, 0, 0};
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = 0;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setzero_si128() (simde_mm_setzero_si128())
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_bslli_si128(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	if (HEDLEY_UNLIKELY((imm8 & ~15))) {
+		return simde_mm_setzero_si128();
+	}
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && defined(SIMDE_ENDIAN_ORDER)
+	r_.altivec_i8 =
+#if (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE)
+		vec_slo
+#else /* SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG */
+		vec_sro
+#endif
+		(a_.altivec_i8,
+		 vec_splats(HEDLEY_STATIC_CAST(unsigned char, imm8 * 8)));
+#elif defined(SIMDE_HAVE_INT128_) && (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE)
+	r_.u128[0] = a_.u128[0] << (imm8 * 8);
+#else
+	r_ = simde__m128i_to_private(simde_mm_setzero_si128());
+	for (int i = imm8;
+	 i < HEDLEY_STATIC_CAST(int, sizeof(r_.i8) / sizeof(r_.i8[0]));
+	 i++) {
+		r_.i8[i] = a_.i8[i - imm8];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+#define simde_mm_bslli_si128(a, imm8) _mm_slli_si128(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__)
+#define simde_mm_bslli_si128(a, imm8) \
+	simde__m128i_from_neon_i8( \
+		((imm8) <= 0) \
+			? simde__m128i_to_neon_i8(a) \
+			: (((imm8) > 15) \
+				 ? (vdupq_n_s8(0)) \
+				 : (vextq_s8(vdupq_n_s8(0), \
+					 simde__m128i_to_neon_i8(a), \
+					 16 - (imm8)))))
+#elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+#define simde_mm_bslli_si128(a, imm8) \
+	(__extension__({ \
+		const simde__m128i_private simde__tmp_a_ = \
+			simde__m128i_to_private(a); \
+		const simde__m128i_private simde__tmp_z_ = \
+			simde__m128i_to_private(simde_mm_setzero_si128()); \
+		simde__m128i_private simde__tmp_r_; \
+		if (HEDLEY_UNLIKELY(imm8 > 15)) { \
+			simde__tmp_r_ = simde__m128i_to_private( \
+				simde_mm_setzero_si128()); \
+		} else { \
+			simde__tmp_r_.i8 = SIMDE_SHUFFLE_VECTOR_( \
+				8, 16, simde__tmp_z_.i8, (simde__tmp_a_).i8, \
+				HEDLEY_STATIC_CAST(int8_t, (16 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (17 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (18 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (19 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (20 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (21 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (22 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (23 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (24 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (25 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (26 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (27 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (28 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (29 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (30 - imm8) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (31 - imm8) & 31)); \
+		} \
+		simde__m128i_from_private(simde__tmp_r_); \
+	}))
+#endif
+#define simde_mm_slli_si128(a, imm8) simde_mm_bslli_si128(a, imm8)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_bslli_si128(a, imm8) simde_mm_bslli_si128(a, imm8)
+#define _mm_slli_si128(a, imm8) simde_mm_bslli_si128(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_bsrli_si128(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	if (HEDLEY_UNLIKELY((imm8 & ~15))) {
+		return simde_mm_setzero_si128();
+	}
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && defined(SIMDE_ENDIAN_ORDER)
+	r_.altivec_i8 =
+#if (SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE)
+		vec_sro
+#else /* SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG */
+		vec_slo
+#endif
+		(a_.altivec_i8,
+		 vec_splats(HEDLEY_STATIC_CAST(unsigned char, imm8 * 8)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		const int e = HEDLEY_STATIC_CAST(int, i) + imm8;
+		r_.i8[i] = (e < 16) ? a_.i8[e] : 0;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+#define simde_mm_bsrli_si128(a, imm8) _mm_srli_si128(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && !defined(__clang__)
+#define simde_mm_bsrli_si128(a, imm8) \
+	simde__m128i_from_neon_i8( \
+		((imm8 < 0) || (imm8 > 15)) \
+			? vdupq_n_s8(0) \
+			: (vextq_s8(simde__m128i_to_private(a).neon_i8, \
+				 vdupq_n_s8(0), \
+				 ((imm8 & 15) != 0) ? imm8 : (imm8 & 15))))
+#elif defined(SIMDE_SHUFFLE_VECTOR_) && !defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+#define simde_mm_bsrli_si128(a, imm8) \
+	(__extension__({ \
+		const simde__m128i_private simde__tmp_a_ = \
+			simde__m128i_to_private(a); \
+		const simde__m128i_private simde__tmp_z_ = \
+			simde__m128i_to_private(simde_mm_setzero_si128()); \
+		simde__m128i_private simde__tmp_r_ = \
+			simde__m128i_to_private(a); \
+		if (HEDLEY_UNLIKELY(imm8 > 15)) { \
+			simde__tmp_r_ = simde__m128i_to_private( \
+				simde_mm_setzero_si128()); \
+		} else { \
+			simde__tmp_r_.i8 = SIMDE_SHUFFLE_VECTOR_( \
+				8, 16, simde__tmp_z_.i8, (simde__tmp_a_).i8, \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 16) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 17) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 18) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 19) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 20) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 21) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 22) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 23) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 24) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 25) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 26) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 27) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 28) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 29) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 30) & 31), \
+				HEDLEY_STATIC_CAST(int8_t, (imm8 + 31) & 31)); \
+		} \
+		simde__m128i_from_private(simde__tmp_r_); \
+	}))
+#endif
+#define simde_mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_bsrli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
+#define _mm_srli_si128(a, imm8) simde_mm_bsrli_si128((a), (imm8))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_clflush(void const *p)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_clflush(p);
+#else
+	(void)p;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_clflush(a, b) simde_mm_clflush()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comieq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comieq_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !!vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) ==
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] == b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comieq_sd(a, b) simde_mm_comieq_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comige_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comige_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !!vgetq_lane_u64(vcgeq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) >=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] >= b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comige_sd(a, b) simde_mm_comige_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comigt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comigt_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !!vgetq_lane_u64(vcgtq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) >
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] > b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comigt_sd(a, b) simde_mm_comigt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comile_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comile_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !!vgetq_lane_u64(vcleq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) <=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] <= b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comile_sd(a, b) simde_mm_comile_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comilt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comilt_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !!vgetq_lane_u64(vcltq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) <
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] < b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comilt_sd(a, b) simde_mm_comilt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_comineq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_comineq_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return !vgetq_lane_u64(vceqq_f64(a_.neon_f64, b_.neon_f64), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) !=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#else
+	return a_.f64[0] != b_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_comineq_sd(a, b) simde_mm_comineq_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_copysign_pd(simde__m128d dest, simde__m128d src)
+{
+	simde__m128d_private r_, dest_ = simde__m128d_to_private(dest),
+				 src_ = simde__m128d_to_private(src);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t sign_pos =
+		vreinterpretq_u64_f64(vdupq_n_f64(-SIMDE_FLOAT64_C(0.0)));
+#else
+	simde_float64 dbl_nz = -SIMDE_FLOAT64_C(0.0);
+	uint64_t u64_nz;
+	simde_memcpy(&u64_nz, &dbl_nz, sizeof(u64_nz));
+	uint64x2_t sign_pos = vdupq_n_u64(u64_nz);
+#endif
+	r_.neon_u64 = vbslq_u64(sign_pos, src_.neon_u64, dest_.neon_u64);
+#elif defined(SIMDE_POWER_ALTIVEC_P9_NATIVE)
+#if !defined(HEDLEY_IBM_VERSION)
+	r_.altivec_f64 = vec_cpsgn(dest_.altivec_f64, src_.altivec_f64);
+#else
+	r_.altivec_f64 = vec_cpsgn(src_.altivec_f64, dest_.altivec_f64);
+#endif
+#elif defined(simde_math_copysign)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = simde_math_copysign(dest_.f64[i], src_.f64[i]);
+	}
+#else
+	simde__m128d sgnbit = simde_mm_set1_pd(-SIMDE_FLOAT64_C(0.0));
+	return simde_mm_xor_pd(simde_mm_and_pd(sgnbit, src),
+			 simde_mm_andnot_pd(sgnbit, dest));
+#endif
+
+	return simde__m128d_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_xorsign_pd(simde__m128d dest, simde__m128d src)
+{
+	return simde_mm_xor_pd(simde_mm_and_pd(simde_mm_set1_pd(-0.0), src),
+			 dest);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_castpd_ps(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castpd_ps(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vreinterpretq_f32_f64(a);
+#else
+	simde__m128 r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castpd_ps(a) simde_mm_castpd_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_castpd_si128(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castpd_si128(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vreinterpretq_s64_f64(a);
+#else
+	simde__m128i r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castpd_si128(a) simde_mm_castpd_si128(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_castps_pd(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castps_pd(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vreinterpretq_f64_f32(a);
+#else
+	simde__m128d r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castps_pd(a) simde_mm_castps_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_castps_si128(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castps_si128(a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return simde__m128i_from_neon_i32(simde__m128_to_private(a).neon_i32);
+#else
+	simde__m128i r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castps_si128(a) simde_mm_castps_si128(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_castsi128_pd(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castsi128_pd(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vreinterpretq_f64_s64(a);
+#else
+	simde__m128d r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castsi128_pd(a) simde_mm_castsi128_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_castsi128_ps(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_castsi128_ps(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	return HEDLEY_REINTERPRET_CAST(SIMDE_POWER_ALTIVEC_VECTOR(float), a);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return simde__m128_from_neon_i32(simde__m128i_to_private(a).neon_i32);
+#else
+	simde__m128 r;
+	simde_memcpy(&r, &a, sizeof(a));
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_castsi128_ps(a) simde_mm_castsi128_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpeq_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpeq_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vceqq_s8(b_.neon_i8, a_.neon_i8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_eq(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i8 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+		vec_cmpeq(a_.altivec_i8, b_.altivec_i8));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 == b_.i8));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = (a_.i8[i] == b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_epi8(a, b) simde_mm_cmpeq_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpeq_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpeq_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vceqq_s16(b_.neon_i16, a_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_eq(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+		vec_cmpeq(a_.altivec_i16, b_.altivec_i16));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = (a_.i16 == b_.i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] == b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_epi16(a, b) simde_mm_cmpeq_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpeq_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpeq_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vceqq_s32(b_.neon_i32, a_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_eq(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+		vec_cmpeq(a_.altivec_i32, b_.altivec_i32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), a_.i32 == b_.i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = (a_.i32[i] == b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_epi32(a, b) simde_mm_cmpeq_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpeq_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpeq_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vceqq_s64(b_.neon_i64, a_.neon_i64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_eq(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f64 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(double),
+		vec_cmpeq(a_.altivec_f64, b_.altivec_f64));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 == b_.f64));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] == b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_pd(a, b) simde_mm_cmpeq_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpeq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpeq_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpeq_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.u64[0] == b_.u64[0]) ? ~UINT64_C(0) : 0;
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpeq_sd(a, b) simde_mm_cmpeq_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpneq_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpneq_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u32 = vmvnq_u32(
+		vreinterpretq_u32_u64(vceqq_f64(b_.neon_f64, a_.neon_f64)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_ne(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 != b_.f64));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] != b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpneq_pd(a, b) simde_mm_cmpneq_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpneq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpneq_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpneq_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.f64[0] != b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpneq_sd(a, b) simde_mm_cmpneq_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmplt_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmplt_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vcltq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i8 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+		vec_cmplt(a_.altivec_i8, b_.altivec_i8));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_lt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 < b_.i8));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = (a_.i8[i] < b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_epi8(a, b) simde_mm_cmplt_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmplt_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmplt_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vcltq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+		vec_cmplt(a_.altivec_i16, b_.altivec_i16));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_lt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = HEDLEY_STATIC_CAST(__typeof__(r_.i16), (a_.i16 < b_.i16));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_epi16(a, b) simde_mm_cmplt_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmplt_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmplt_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcltq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+		vec_cmplt(a_.altivec_i32, b_.altivec_i32));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_lt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.i32 < b_.i32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = (a_.i32[i] < b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_epi32(a, b) simde_mm_cmplt_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmplt_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmplt_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 < b_.f64));
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vcltq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_lt(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] < b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_pd(a, b) simde_mm_cmplt_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmplt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmplt_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmplt_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.f64[0] < b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmplt_sd(a, b) simde_mm_cmplt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmple_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmple_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 <= b_.f64));
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vcleq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_le(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f64 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(double),
+		vec_cmple(a_.altivec_f64, b_.altivec_f64));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] <= b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmple_pd(a, b) simde_mm_cmple_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmple_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmple_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmple_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.f64[0] <= b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmple_sd(a, b) simde_mm_cmple_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpgt_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpgt_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vcgtq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_gt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i8 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed char),
+		vec_cmpgt(a_.altivec_i8, b_.altivec_i8));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = HEDLEY_STATIC_CAST(__typeof__(r_.i8), (a_.i8 > b_.i8));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = (a_.i8[i] > b_.i8[i]) ? ~INT8_C(0) : INT8_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_epi8(a, b) simde_mm_cmpgt_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpgt_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpgt_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vcgtq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_gt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed short),
+		vec_cmpgt(a_.altivec_i16, b_.altivec_i16));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = HEDLEY_STATIC_CAST(__typeof__(r_.i16), (a_.i16 > b_.i16));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? ~INT16_C(0) : INT16_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_epi16(a, b) simde_mm_cmpgt_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cmpgt_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpgt_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vcgtq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_gt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = HEDLEY_REINTERPRET_CAST(
+		SIMDE_POWER_ALTIVEC_VECTOR(signed int),
+		vec_cmpgt(a_.altivec_i32, b_.altivec_i32));
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = HEDLEY_STATIC_CAST(__typeof__(r_.i32), (a_.i32 > b_.i32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = (a_.i32[i] > b_.i32[i]) ? ~INT32_C(0) : INT32_C(0);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_epi32(a, b) simde_mm_cmpgt_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpgt_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpgt_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 > b_.f64));
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vcgtq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_gt(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f64 =
+		HEDLEY_STATIC_CAST(SIMDE_POWER_ALTIVEC_VECTOR(double),
+				 vec_cmpgt(a_.altivec_f64, b_.altivec_f64));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] > b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_pd(a, b) simde_mm_cmpgt_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpgt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_cmpgt_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpgt_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.f64[0] > b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpgt_sd(a, b) simde_mm_cmpgt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpge_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpge_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = HEDLEY_STATIC_CAST(__typeof__(r_.i64), (a_.f64 >= b_.f64));
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_u64 = vcgeq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_ge(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_f64 =
+		HEDLEY_STATIC_CAST(SIMDE_POWER_ALTIVEC_VECTOR(double),
+				 vec_cmpge(a_.altivec_f64, b_.altivec_f64));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (a_.f64[i] >= b_.f64[i]) ? ~UINT64_C(0)
+						 : UINT64_C(0);
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpge_pd(a, b) simde_mm_cmpge_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpge_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_cmpge_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpge_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.u64[0] = (a_.f64[0] >= b_.f64[0]) ? ~UINT64_C(0) : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpge_sd(a, b) simde_mm_cmpge_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpngt_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpngt_pd(a, b);
+#else
+	return simde_mm_cmple_pd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpngt_pd(a, b) simde_mm_cmpngt_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpngt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_cmpngt_sd(a, b);
+#else
+	return simde_mm_cmple_sd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpngt_sd(a, b) simde_mm_cmpngt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnge_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpnge_pd(a, b);
+#else
+	return simde_mm_cmplt_pd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnge_pd(a, b) simde_mm_cmpnge_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnge_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_cmpnge_sd(a, b);
+#else
+	return simde_mm_cmplt_sd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnge_sd(a, b) simde_mm_cmpnge_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnlt_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpnlt_pd(a, b);
+#else
+	return simde_mm_cmpge_pd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnlt_pd(a, b) simde_mm_cmpnlt_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnlt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpnlt_sd(a, b);
+#else
+	return simde_mm_cmpge_sd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnlt_sd(a, b) simde_mm_cmpnlt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnle_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpnle_pd(a, b);
+#else
+	return simde_mm_cmpgt_pd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnle_pd(a, b) simde_mm_cmpnle_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpnle_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpnle_sd(a, b);
+#else
+	return simde_mm_cmpgt_sd(a, b);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpnle_sd(a, b) simde_mm_cmpnle_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpord_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpord_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	/* Note: NEON does not have ordered compare builtin
+ Need to compare a eq a and b eq b to check for NaN
+ Do AND of results to get final */
+	uint64x2_t ceqaa = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t ceqbb = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	r_.neon_u64 = vandq_u64(ceqaa, ceqbb);
+#elif defined(simde_math_isnan)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (!simde_math_isnan(a_.f64[i]) &&
+			 !simde_math_isnan(b_.f64[i]))
+				 ? ~UINT64_C(0)
+				 : UINT64_C(0);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpord_pd(a, b) simde_mm_cmpord_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde_float64 simde_mm_cvtsd_f64(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_cvtsd_f64(a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return HEDLEY_STATIC_CAST(simde_float64,
+				 vgetq_lane_f64(a_.neon_f64, 0));
+#else
+	return a_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsd_f64(a) simde_mm_cvtsd_f64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpord_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpord_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpord_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(simde_math_isnan)
+	r_.u64[0] =
+		(!simde_math_isnan(a_.f64[0]) && !simde_math_isnan(b_.f64[0]))
+			? ~UINT64_C(0)
+			: UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpord_sd(a, b) simde_mm_cmpord_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpunord_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpunord_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t ceqaa = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t ceqbb = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	r_.neon_u64 = vreinterpretq_u64_u32(
+		vmvnq_u32(vreinterpretq_u32_u64(vandq_u64(ceqaa, ceqbb))));
+#elif defined(simde_math_isnan)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.u64[i] = (simde_math_isnan(a_.f64[i]) ||
+			 simde_math_isnan(b_.f64[i]))
+				 ? ~UINT64_C(0)
+				 : UINT64_C(0);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpunord_pd(a, b) simde_mm_cmpunord_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cmpunord_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cmpunord_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_cmpunord_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(simde_math_isnan)
+	r_.u64[0] = (simde_math_isnan(a_.f64[0]) || simde_math_isnan(b_.f64[0]))
+			 ? ~UINT64_C(0)
+			 : UINT64_C(0);
+	r_.u64[1] = a_.u64[1];
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cmpunord_sd(a, b) simde_mm_cmpunord_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtepi32_pd(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtepi32_pd(a);
+#else
+	simde__m128d_private r_;
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f64, a_.m64_private[0].i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = (simde_float64)a_.i32[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtepi32_pd(a) simde_mm_cvtepi32_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtepi32_ps(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtepi32_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_f32 = vcvtq_f32_s32(a_.neon_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f32x4_convert_i32x4(a_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	HEDLEY_DIAGNOSTIC_PUSH
+#if HEDLEY_HAS_WARNING("-Wc11-extensions")
+#pragma clang diagnostic ignored "-Wc11-extensions"
+#endif
+	r_.altivec_f32 = vec_ctf(a_.altivec_i32, 0);
+	HEDLEY_DIAGNOSTIC_POP
+#elif defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f32, a_.i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f32) / sizeof(r_.f32[0])); i++) {
+		r_.f32[i] = (simde_float32)a_.i32[i];
+	}
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtepi32_ps(a) simde_mm_cvtepi32_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvtpd_pi32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpd_pi32(a);
+#else
+	simde__m64_private r_;
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		simde_float64 v = simde_math_round(a_.f64[i]);
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float64, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float64, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#endif
+	}
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpd_pi32(a) simde_mm_cvtpd_pi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvtpd_epi32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtpd_epi32(a);
+#else
+	simde__m128i_private r_;
+
+	r_.m64[0] = simde_mm_cvtpd_pi32(a);
+	r_.m64[1] = simde_mm_setzero_si64();
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpd_epi32(a) simde_mm_cvtpd_epi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtpd_ps(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtpd_ps(a);
+#else
+	simde__m128_private r_;
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.m64_private[0].f32, a_.f64);
+	r_.m64_private[1] = simde__m64_to_private(simde_mm_setzero_si64());
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vreinterpretq_f32_f64(
+		vcombine_f64(vreinterpret_f64_f32(vcvtx_f32_f64(a_.neon_f64)),
+			 vdup_n_f64(0)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(a_.f64) / sizeof(a_.f64[0])); i++) {
+		r_.f32[i] = (simde_float32)a_.f64[i];
+	}
+	simde_memset(&(r_.m64_private[1]), 0, sizeof(r_.m64_private[1]));
+#endif
+
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpd_ps(a) simde_mm_cvtpd_ps(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtpi32_pd(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvtpi32_pd(a);
+#else
+	simde__m128d_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f64, a_.i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = (simde_float64)a_.i32[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtpi32_pd(a) simde_mm_cvtpi32_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvtps_epi32(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtps_epi32(a);
+#else
+	simde__m128i_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	r_.neon_i32 = vcvtnq_s32_f32(a_.neon_f32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE) && \
+	defined(SIMDE_FAST_CONVERSION_RANGE) && defined(SIMDE_FAST_ROUND_TIES)
+	r_.neon_i32 = vcvtnq_s32_f32(a_.neon_f32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE) && \
+	defined(SIMDE_FAST_CONVERSION_RANGE) && defined(SIMDE_FAST_ROUND_TIES)
+	HEDLEY_DIAGNOSTIC_PUSH
+	SIMDE_DIAGNOSTIC_DISABLE_C11_EXTENSIONS_
+	SIMDE_DIAGNOSTIC_DISABLE_VECTOR_CONVERSION_
+	r_.altivec_i32 = vec_cts(a_.altivec_f32, 1);
+	HEDLEY_DIAGNOSTIC_POP
+#else
+	a_ = simde__m128_to_private(
+		simde_x_mm_round_ps(a, SIMDE_MM_FROUND_TO_NEAREST_INT, 1));
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		simde_float32 v = simde_math_roundf(a_.f32[i]);
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#endif
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtps_epi32(a) simde_mm_cvtps_epi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtps_pd(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtps_pd(a);
+#else
+	simde__m128d_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_CONVERT_VECTOR_)
+	SIMDE_CONVERT_VECTOR_(r_.f64, a_.m64_private[0].f32);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vcvt_f64_f32(vget_low_f32(a_.neon_f32));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = a_.f32[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtps_pd(a) simde_mm_cvtps_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvtsd_si32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtsd_si32(a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+	simde_float64 v = simde_math_round(a_.f64[0]);
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+	return SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+	return ((v > HEDLEY_STATIC_CAST(simde_float64, INT32_MIN)) &&
+		(v < HEDLEY_STATIC_CAST(simde_float64, INT32_MAX)))
+		 ? SIMDE_CONVERT_FTOI(int32_t, v)
+		 : INT32_MIN;
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsd_si32(a) simde_mm_cvtsd_si32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvtsd_si64(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if defined(__PGI)
+	return _mm_cvtsd_si64x(a);
+#else
+	return _mm_cvtsd_si64(a);
+#endif
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+	return SIMDE_CONVERT_FTOI(int64_t, simde_math_round(a_.f64[0]));
+#endif
+}
+#define simde_mm_cvtsd_si64x(a) simde_mm_cvtsd_si64(a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsd_si64(a) simde_mm_cvtsd_si64(a)
+#define _mm_cvtsd_si64x(a) simde_mm_cvtsd_si64x(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128 simde_mm_cvtsd_ss(simde__m128 a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtsd_ss(a, b);
+#else
+	simde__m128_private r_, a_ = simde__m128_to_private(a);
+	simde__m128d_private b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f32 = vsetq_lane_f32(
+		vcvtxd_f32_f64(vgetq_lane_f64(b_.neon_f64, 0)), a_.neon_f32, 0);
+#else
+	r_.f32[0] = HEDLEY_STATIC_CAST(simde_float32, b_.f64[0]);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 1; i < (sizeof(r_) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i];
+	}
+#endif
+	return simde__m128_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsd_ss(a, b) simde_mm_cvtsd_ss(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int16_t simde_x_mm_cvtsi128_si16(simde__m128i a)
+{
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vgetq_lane_s16(a_.neon_i16, 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return HEDLEY_STATIC_CAST(int16_t,
+				 wasm_i16x8_extract_lane(a_.wasm_v128, 0));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+#if defined(SIMDE_BUG_GCC_95227)
+	(void)a_;
+#endif
+	return vec_extract(a_.altivec_i16, 0);
+#else
+	return a_.i16[0];
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvtsi128_si32(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtsi128_si32(a);
+#else
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vgetq_lane_s32(a_.neon_i32, 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return HEDLEY_STATIC_CAST(int32_t,
+				 wasm_i32x4_extract_lane(a_.wasm_v128, 0));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+#if defined(SIMDE_BUG_GCC_95227)
+	(void)a_;
+#endif
+	return vec_extract(a_.altivec_i32, 0);
+#else
+	return a_.i32[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi128_si32(a) simde_mm_cvtsi128_si32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvtsi128_si64(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if defined(__PGI)
+	return _mm_cvtsi128_si64x(a);
+#else
+	return _mm_cvtsi128_si64(a);
+#endif
+#else
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE) && !defined(HEDLEY_IBM_VERSION)
+	return vec_extract(HEDLEY_REINTERPRET_CAST(
+				 SIMDE_POWER_ALTIVEC_VECTOR(signed long long),
+				 a_.i64),
+			 0);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	return vgetq_lane_s64(a_.neon_i64, 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return HEDLEY_STATIC_CAST(int64_t,
+				 wasm_i64x2_extract_lane(a_.wasm_v128, 0));
+#endif
+	return a_.i64[0];
+#endif
+}
+#define simde_mm_cvtsi128_si64x(a) simde_mm_cvtsi128_si64(a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi128_si64(a) simde_mm_cvtsi128_si64(a)
+#define _mm_cvtsi128_si64x(a) simde_mm_cvtsi128_si64x(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtsi32_sd(simde__m128d a, int32_t b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtsi32_sd(a, b);
+#else
+	simde__m128d_private r_;
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_ARCH_AMD64)
+	r_.neon_f64 = vsetq_lane_f64(HEDLEY_STATIC_CAST(float64_t, b),
+				 a_.neon_f64, 0);
+#else
+	r_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b);
+	r_.i64[1] = a_.i64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi32_sd(a, b) simde_mm_cvtsi32_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_cvtsi16_si128(int16_t a)
+{
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vsetq_lane_s16(a, vdupq_n_s16(0), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_make(a, 0, 0, 0, 0, 0, 0, 0);
+#else
+	r_.i16[0] = a;
+	r_.i16[1] = 0;
+	r_.i16[2] = 0;
+	r_.i16[3] = 0;
+	r_.i16[4] = 0;
+	r_.i16[5] = 0;
+	r_.i16[6] = 0;
+	r_.i16[7] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvtsi32_si128(int32_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtsi32_si128(a);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vsetq_lane_s32(a, vdupq_n_s32(0), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_make(a, 0, 0, 0);
+#else
+	r_.i32[0] = a;
+	r_.i32[1] = 0;
+	r_.i32[2] = 0;
+	r_.i32[3] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi32_si128(a) simde_mm_cvtsi32_si128(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtsi64_sd(simde__m128d a, int64_t b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if !defined(__PGI)
+	return _mm_cvtsi64_sd(a, b);
+#else
+	return _mm_cvtsi64x_sd(a, b);
+#endif
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vsetq_lane_f64(HEDLEY_STATIC_CAST(float64_t, b),
+				 a_.neon_f64, 0);
+#else
+	r_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b);
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#define simde_mm_cvtsi64x_sd(a, b) simde_mm_cvtsi64_sd(a, b)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi64_sd(a, b) simde_mm_cvtsi64_sd(a, b)
+#define _mm_cvtsi64x_sd(a, b) simde_mm_cvtsi64x_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvtsi64_si128(int64_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if !defined(__PGI)
+	return _mm_cvtsi64_si128(a);
+#else
+	return _mm_cvtsi64x_si128(a);
+#endif
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vsetq_lane_s64(a, vdupq_n_s64(0), 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i64x2_make(a, 0);
+#else
+	r_.i64[0] = a;
+	r_.i64[1] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#define simde_mm_cvtsi64x_si128(a) simde_mm_cvtsi64_si128(a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtsi64_si128(a) simde_mm_cvtsi64_si128(a)
+#define _mm_cvtsi64x_si128(a) simde_mm_cvtsi64x_si128(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_cvtss_sd(simde__m128d a, simde__m128 b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvtss_sd(a, b);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x2_t temp = vcvt_f64_f32(vset_lane_f32(
+		vgetq_lane_f32(simde__m128_to_private(b).neon_f32, 0),
+		vdup_n_f32(0), 0));
+	return vsetq_lane_f64(
+		vgetq_lane_f64(simde__m128d_to_private(a).neon_f64, 1), temp,
+		1);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+	simde__m128_private b_ = simde__m128_to_private(b);
+
+	a_.f64[0] = HEDLEY_STATIC_CAST(simde_float64, b_.f32[0]);
+
+	return simde__m128d_from_private(a_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvtss_sd(a, b) simde_mm_cvtss_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_cvttpd_pi32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_cvttpd_pi32(a);
+#else
+	simde__m64_private r_;
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_CONVERT_VECTOR_) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f64);
+#else
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		simde_float64 v = a_.f64[i];
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float64, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float64, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#endif
+	}
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttpd_pi32(a) simde_mm_cvttpd_pi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvttpd_epi32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvttpd_epi32(a);
+#else
+	simde__m128i_private r_;
+
+	r_.m64[0] = simde_mm_cvttpd_pi32(a);
+	r_.m64[1] = simde_mm_setzero_si64();
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttpd_epi32(a) simde_mm_cvttpd_epi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_cvttps_epi32(simde__m128 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvttps_epi32(a);
+#else
+	simde__m128i_private r_;
+	simde__m128_private a_ = simde__m128_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	r_.neon_i32 = vcvtq_s32_f32(a_.neon_f32);
+#elif defined(SIMDE_CONVERT_VECTOR_) && defined(SIMDE_FAST_CONVERSION_RANGE)
+	SIMDE_CONVERT_VECTOR_(r_.i32, a_.f32);
+#else
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		simde_float32 v = a_.f32[i];
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+		r_.i32[i] = SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+		r_.i32[i] =
+			((v > HEDLEY_STATIC_CAST(simde_float32, INT32_MIN)) &&
+			 (v < HEDLEY_STATIC_CAST(simde_float32, INT32_MAX)))
+				? SIMDE_CONVERT_FTOI(int32_t, v)
+				: INT32_MIN;
+#endif
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttps_epi32(a) simde_mm_cvttps_epi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_cvttsd_si32(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_cvttsd_si32(a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+	simde_float64 v = a_.f64[0];
+#if defined(SIMDE_FAST_CONVERSION_RANGE)
+	return SIMDE_CONVERT_FTOI(int32_t, v);
+#else
+	return ((v > HEDLEY_STATIC_CAST(simde_float64, INT32_MIN)) &&
+		(v < HEDLEY_STATIC_CAST(simde_float64, INT32_MAX)))
+		 ? SIMDE_CONVERT_FTOI(int32_t, v)
+		 : INT32_MIN;
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttsd_si32(a) simde_mm_cvttsd_si32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int64_t simde_mm_cvttsd_si64(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+#if !defined(__PGI)
+	return _mm_cvttsd_si64(a);
+#else
+	return _mm_cvttsd_si64x(a);
+#endif
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+	return SIMDE_CONVERT_FTOI(int64_t, a_.f64[0]);
+#endif
+}
+#define simde_mm_cvttsd_si64x(a) simde_mm_cvttsd_si64(a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_cvttsd_si64(a) simde_mm_cvttsd_si64(a)
+#define _mm_cvttsd_si64x(a) simde_mm_cvttsd_si64x(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_div_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_div_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f64 = a_.f64 / b_.f64;
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vdivq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_div(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = a_.f64[i] / b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_div_pd(a, b) simde_mm_div_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_div_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_div_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_div_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x2_t temp = vdivq_f64(a_.neon_f64, b_.neon_f64);
+	r_.neon_f64 = vsetq_lane_f64(vgetq_lane(a_.neon_f64, 1), temp, 1);
+#else
+	r_.f64[0] = a_.f64[0] / b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_div_sd(a, b) simde_mm_div_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_extract_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 7)
+{
+	uint16_t r;
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+#if defined(SIMDE_BUG_GCC_95227)
+	(void)a_;
+	(void)imm8;
+#endif
+	r = HEDLEY_STATIC_CAST(uint16_t, vec_extract(a_.altivec_i16, imm8));
+#else
+	r = a_.u16[imm8 & 7];
+#endif
+
+	return HEDLEY_STATIC_CAST(int32_t, r);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(4, 6, 0))
+#define simde_mm_extract_epi16(a, imm8) _mm_extract_epi16(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_extract_epi16(a, imm8) \
+	(HEDLEY_STATIC_CAST( \
+		 int32_t, vgetq_lane_s16(simde__m128i_to_private(a).neon_i16, \
+					 (imm8))) & \
+	 (INT32_C(0x0000ffff)))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_extract_epi16(a, imm8) simde_mm_extract_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_insert_epi16(simde__m128i a, int16_t i, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 7)
+{
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+	a_.i16[imm8 & 7] = i;
+	return simde__m128i_from_private(a_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+#define simde_mm_insert_epi16(a, i, imm8) _mm_insert_epi16((a), (i), (imm8))
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_insert_epi16(a, i, imm8) \
+	simde__m128i_from_neon_i16( \
+		vsetq_lane_s16((i), simde__m128i_to_neon_i16(a), (imm8)))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_insert_epi16(a, i, imm8) simde_mm_insert_epi16(a, i, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d
+simde_mm_load_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_load_pd(mem_addr);
+#else
+	simde__m128d_private r_;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vld1q_f64(mem_addr);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 =
+		vld1q_u32(HEDLEY_REINTERPRET_CAST(uint32_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128d),
+		 sizeof(r_));
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_load_pd(mem_addr) simde_mm_load_pd(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_load1_pd(simde_float64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_load1_pd(mem_addr);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return simde__m128d_from_neon_f64(vld1q_dup_f64(mem_addr));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return simde__m128d_from_wasm_v128(wasm_v64x2_load_splat(mem_addr));
+#else
+	return simde_mm_set1_pd(*mem_addr);
+#endif
+}
+#define simde_mm_load_pd1(mem_addr) simde_mm_load1_pd(mem_addr)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_load_pd1(mem_addr) simde_mm_load1_pd(mem_addr)
+#define _mm_load1_pd(mem_addr) simde_mm_load1_pd(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_load_sd(simde_float64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_load_sd(mem_addr);
+#else
+	simde__m128d_private r_;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vsetq_lane_f64(*mem_addr, vdupq_n_f64(0), 0);
+#else
+	r_.f64[0] = *mem_addr;
+	r_.u64[1] = UINT64_C(0);
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_load_sd(mem_addr) simde_mm_load_sd(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_load_si128(simde__m128i const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_load_si128(
+		HEDLEY_REINTERPRET_CAST(__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_ld(
+		0, HEDLEY_REINTERPRET_CAST(
+			 SIMDE_POWER_ALTIVEC_VECTOR(int) const *, mem_addr));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 =
+		vld1q_s32(HEDLEY_REINTERPRET_CAST(int32_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128i),
+		 sizeof(simde__m128i));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_load_si128(mem_addr) simde_mm_load_si128(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_loadh_pd(simde__m128d a, simde_float64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadh_pd(a, mem_addr);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vcombine_f64(
+		vget_low_f64(a_.neon_f64),
+		vld1_f64(HEDLEY_REINTERPRET_CAST(const float64_t *, mem_addr)));
+#else
+	simde_float64 t;
+
+	simde_memcpy(&t, mem_addr, sizeof(t));
+	r_.f64[0] = a_.f64[0];
+	r_.f64[1] = t;
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadh_pd(a, mem_addr) simde_mm_loadh_pd(a, mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_loadl_epi64(simde__m128i const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadl_epi64(mem_addr);
+#else
+	simde__m128i_private r_;
+
+	int64_t value;
+	simde_memcpy(&value, mem_addr, sizeof(value));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vcombine_s64(
+		vld1_s64(HEDLEY_REINTERPRET_CAST(int64_t const *, mem_addr)),
+		vdup_n_s64(0));
+#else
+	r_.i64[0] = value;
+	r_.i64[1] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadl_epi64(mem_addr) simde_mm_loadl_epi64(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_loadl_pd(simde__m128d a, simde_float64 const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadl_pd(a, mem_addr);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vcombine_f64(
+		vld1_f64(HEDLEY_REINTERPRET_CAST(const float64_t *, mem_addr)),
+		vget_high_f64(a_.neon_f64));
+#else
+	r_.f64[0] = *mem_addr;
+	r_.u64[1] = a_.u64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadl_pd(a, mem_addr) simde_mm_loadl_pd(a, mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d
+simde_mm_loadr_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadr_pd(mem_addr);
+#else
+	simde__m128d_private r_;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vld1q_f64(mem_addr);
+	r_.neon_f64 = vextq_f64(r_.neon_f64, r_.neon_f64, 1);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 =
+		vld1q_s64(HEDLEY_REINTERPRET_CAST(int64_t const *, mem_addr));
+	r_.neon_i64 = vextq_s64(r_.neon_i64, r_.neon_i64, 1);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	v128_t tmp = wasm_v128_load(mem_addr);
+	r_.wasm_v128 = wasm_v64x2_shuffle(tmp, tmp, 1, 0);
+#else
+	r_.f64[0] = mem_addr[1];
+	r_.f64[1] = mem_addr[0];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadr_pd(mem_addr) simde_mm_loadr_pd(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d
+simde_mm_loadu_pd(simde_float64 const mem_addr[HEDLEY_ARRAY_PARAM(2)])
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_pd(mem_addr);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vld1q_f64(mem_addr);
+#else
+	simde__m128d_private r_;
+
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_pd(mem_addr) simde_mm_loadu_pd(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_loadu_epi8(int8_t const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_si128(
+		SIMDE_ALIGN_CAST(simde__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 =
+		vld1q_s8(HEDLEY_REINTERPRET_CAST(int8_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_loadu_epi16(int16_t const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_si128(
+		SIMDE_ALIGN_CAST(simde__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 =
+		vld1q_s16(HEDLEY_REINTERPRET_CAST(int16_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_loadu_epi32(int32_t const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_si128(
+		SIMDE_ALIGN_CAST(simde__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 =
+		vld1q_s32(HEDLEY_REINTERPRET_CAST(int32_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_loadu_epi64(int64_t const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_si128(
+		SIMDE_ALIGN_CAST(simde__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 =
+		vld1q_s64(HEDLEY_REINTERPRET_CAST(int64_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_loadu_si128(void const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_loadu_si128(HEDLEY_STATIC_CAST(__m128i const *, mem_addr));
+#else
+	simde__m128i_private r_;
+
+#if HEDLEY_GNUC_HAS_ATTRIBUTE(may_alias, 3, 3, 0)
+	HEDLEY_DIAGNOSTIC_PUSH
+	SIMDE_DIAGNOSTIC_DISABLE_PACKED_
+	struct simde_mm_loadu_si128_s {
+		__typeof__(r_) v;
+	} __attribute__((__packed__, __may_alias__));
+	r_ = HEDLEY_REINTERPRET_CAST(const struct simde_mm_loadu_si128_s *,
+				 mem_addr)
+		 ->v;
+	HEDLEY_DIAGNOSTIC_POP
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	/* Note that this is a lower priority than the struct above since
+ * clang assumes mem_addr is aligned (since it is a __m128i*). */
+	r_.neon_i32 =
+		vld1q_s32(HEDLEY_REINTERPRET_CAST(int32_t const *, mem_addr));
+#else
+	simde_memcpy(&r_, mem_addr, sizeof(r_));
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_si128(mem_addr) simde_mm_loadu_si128(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_madd_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_madd_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	int32x4_t pl =
+		vmull_s16(vget_low_s16(a_.neon_i16), vget_low_s16(b_.neon_i16));
+	int32x4_t ph = vmull_high_s16(a_.neon_i16, b_.neon_i16);
+	r_.neon_i32 = vpaddq_s32(pl, ph);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int32x4_t pl =
+		vmull_s16(vget_low_s16(a_.neon_i16), vget_low_s16(b_.neon_i16));
+	int32x4_t ph = vmull_s16(vget_high_s16(a_.neon_i16),
+				 vget_high_s16(b_.neon_i16));
+	int32x2_t rl = vpadd_s32(vget_low_s32(pl), vget_high_s32(pl));
+	int32x2_t rh = vpadd_s32(vget_low_s32(ph), vget_high_s32(ph));
+	r_.neon_i32 = vcombine_s32(rl, rh);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	static const SIMDE_POWER_ALTIVEC_VECTOR(int) tz = {0, 0, 0, 0};
+	r_.altivec_i32 = vec_msum(a_.altivec_i16, b_.altivec_i16, tz);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i += 2) {
+		r_.i32[i / 2] = (a_.i16[i] * b_.i16[i]) +
+				(a_.i16[i + 1] * b_.i16[i + 1]);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_madd_epi16(a, b) simde_mm_madd_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_maskmoveu_si128(simde__m128i a, simde__m128i mask,
+			 int8_t mem_addr[HEDLEY_ARRAY_PARAM(16)])
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_maskmoveu_si128(a, mask, HEDLEY_REINTERPRET_CAST(char *, mem_addr));
+#else
+	simde__m128i_private a_ = simde__m128i_to_private(a),
+			 mask_ = simde__m128i_to_private(mask);
+
+	for (size_t i = 0; i < (sizeof(a_.i8) / sizeof(a_.i8[0])); i++) {
+		if (mask_.u8[i] & 0x80) {
+			mem_addr[i] = a_.i8[i];
+		}
+	}
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_maskmoveu_si128(a, mask, mem_addr) \
+	simde_mm_maskmoveu_si128( \
+		(a), (mask), \
+		SIMDE_CHECKED_REINTERPRET_CAST(int8_t *, char *, (mem_addr)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_movemask_epi8(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__INTEL_COMPILER)
+	/* ICC has trouble with _mm_movemask_epi8 at -O2 and above: */
+	return _mm_movemask_epi8(a);
+#else
+	int32_t r = 0;
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint8x16_t input = a_.neon_u8;
+	const int8_t xr[16] = {-7, -6, -5, -4, -3, -2, -1, 0,
+			 -7, -6, -5, -4, -3, -2, -1, 0};
+	const uint8x16_t mask_and = vdupq_n_u8(0x80);
+	const int8x16_t mask_shift = vld1q_s8(xr);
+	const uint8x16_t mask_result =
+		vshlq_u8(vandq_u8(input, mask_and), mask_shift);
+	uint8x8_t lo = vget_low_u8(mask_result);
+	uint8x8_t hi = vget_high_u8(mask_result);
+	r = vaddv_u8(lo) + (vaddv_u8(hi) << 8);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	// Use increasingly wide shifts+adds to collect the sign bits
+	// together.
+	// Since the widening shifts would be rather confusing to follow in little endian, everything
+	// will be illustrated in big endian order instead. This has a different result - the bits
+	// would actually be reversed on a big endian machine.
+
+	// Starting input (only half the elements are shown):
+	// 89 ff 1d c0 00 10 99 33
+	uint8x16_t input = a_.neon_u8;
+
+	// Shift out everything but the sign bits with an unsigned shift right.
+	//
+	// Bytes of the vector::
+	// 89 ff 1d c0 00 10 99 33
+	// \ \ \ \ \ \ \ \ high_bits = (uint16x4_t)(input >> 7)
+	// | | | | | | | |
+	// 01 01 00 01 00 00 01 00
+	//
+	// Bits of first important lane(s):
+	// 10001001 (89)
+	// \______
+	// |
+	// 00000001 (01)
+	uint16x8_t high_bits = vreinterpretq_u16_u8(vshrq_n_u8(input, 7));
+
+	// Merge the even lanes together with a 16-bit unsigned shift right + add.
+	// 'xx' represents garbage data which will be ignored in the final result.
+	// In the important bytes, the add functions like a binary OR.
+	//
+	// 01 01 00 01 00 00 01 00
+	// \_ | \_ | \_ | \_ | paired16 = (uint32x4_t)(input + (input >> 7))
+	// \| \| \| \|
+	// xx 03 xx 01 xx 00 xx 02
+	//
+	// 00000001 00000001 (01 01)
+	// \_______ |
+	// \|
+	// xxxxxxxx xxxxxx11 (xx 03)
+	uint32x4_t paired16 =
+		vreinterpretq_u32_u16(vsraq_n_u16(high_bits, high_bits, 7));
+
+	// Repeat with a wider 32-bit shift + add.
+	// xx 03 xx 01 xx 00 xx 02
+	// \____ | \____ | paired32 = (uint64x1_t)(paired16 + (paired16 >> 14))
+	// \| \|
+	// xx xx xx 0d xx xx xx 02
+	//
+	// 00000011 00000001 (03 01)
+	// \\_____ ||
+	// '----.\||
+	// xxxxxxxx xxxx1101 (xx 0d)
+	uint64x2_t paired32 =
+		vreinterpretq_u64_u32(vsraq_n_u32(paired16, paired16, 14));
+
+	// Last, an even wider 64-bit shift + add to get our result in the low 8 bit lanes.
+	// xx xx xx 0d xx xx xx 02
+	// \_________ | paired64 = (uint8x8_t)(paired32 + (paired32 >> 28))
+	// \|
+	// xx xx xx xx xx xx xx d2
+	//
+	// 00001101 00000010 (0d 02)
+	// \ \___ | |
+	// '---. \| |
+	// xxxxxxxx 11010010 (xx d2)
+	uint8x16_t paired64 =
+		vreinterpretq_u8_u64(vsraq_n_u64(paired32, paired32, 28));
+
+	// Extract the low 8 bits from each 64-bit lane with 2 8-bit extracts.
+	// xx xx xx xx xx xx xx d2
+	// || return paired64[0]
+	// d2
+	// Note: Little endian would return the correct value 4b (01001011) instead.
+	r = vgetq_lane_u8(paired64, 0) |
+	 (HEDLEY_STATIC_CAST(int32_t, vgetq_lane_u8(paired64, 8)) << 8);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && \
+	!defined(HEDLEY_IBM_VERSION) && \
+	(SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE)
+	static const SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
+		perm = {120, 112, 104, 96, 88, 80, 72, 64,
+			56, 48, 40, 32, 24, 16, 8, 0};
+	r = HEDLEY_STATIC_CAST(
+		int32_t, vec_extract(vec_vbpermq(a_.altivec_u8, perm), 1));
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && \
+	!defined(HEDLEY_IBM_VERSION) && \
+	(SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_BIG)
+	static const SIMDE_POWER_ALTIVEC_VECTOR(unsigned char)
+		perm = {120, 112, 104, 96, 88, 80, 72, 64,
+			56, 48, 40, 32, 24, 16, 8, 0};
+	r = HEDLEY_STATIC_CAST(
+		int32_t, vec_extract(vec_vbpermq(a_.altivec_u8, perm), 14));
+#else
+	SIMDE_VECTORIZE_REDUCTION(| : r)
+	for (size_t i = 0; i < (sizeof(a_.u8) / sizeof(a_.u8[0])); i++) {
+		r |= (a_.u8[15 - i] >> 7) << (15 - i);
+	}
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_movemask_epi8(a) simde_mm_movemask_epi8(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int32_t simde_mm_movemask_pd(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_movemask_pd(a);
+#else
+	int32_t r = 0;
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	static const int64_t shift_amount[] = {0, 1};
+	const int64x2_t shift = vld1q_s64(shift_amount);
+	uint64x2_t tmp = vshrq_n_u64(a_.neon_u64, 63);
+	return HEDLEY_STATIC_CAST(int32_t, vaddvq_u64(vshlq_u64(tmp, shift)));
+#else
+	SIMDE_VECTORIZE_REDUCTION(| : r)
+	for (size_t i = 0; i < (sizeof(a_.u64) / sizeof(a_.u64[0])); i++) {
+		r |= (a_.u64[i] >> 63) << i;
+	}
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_movemask_pd(a) simde_mm_movemask_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_movepi64_pi64(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_movepi64_pi64(a);
+#else
+	simde__m64_private r_;
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i64 = vget_low_s64(a_.neon_i64);
+#else
+	r_.i64[0] = a_.i64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_movepi64_pi64(a) simde_mm_movepi64_pi64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_movpi64_epi64(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_movpi64_epi64(a);
+#else
+	simde__m128i_private r_;
+	simde__m64_private a_ = simde__m64_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vcombine_s64(a_.neon_i64, vdup_n_s64(0));
+#else
+	r_.i64[0] = a_.i64[0];
+	r_.i64[1] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_movpi64_epi64(a) simde_mm_movpi64_epi64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_min_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_min_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vminq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_min(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = vec_min(a_.altivec_i16, b_.altivec_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] < b_.i16[i]) ? a_.i16[i] : b_.i16[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_min_epi16(a, b) simde_mm_min_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_min_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_min_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vminq_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u8x16_min(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u8 = vec_min(a_.altivec_u8, b_.altivec_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] < b_.u8[i]) ? a_.u8[i] : b_.u8[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_min_epu8(a, b) simde_mm_min_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_min_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_min_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_f64 = vec_min(a_.altivec_f64, b_.altivec_f64);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vminq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_min(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = (a_.f64[i] < b_.f64[i]) ? a_.f64[i] : b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_min_pd(a, b) simde_mm_min_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_min_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_min_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_min_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x2_t temp = vminq_f64(a_.neon_f64, b_.neon_f64);
+	r_.neon_f64 = vsetq_lane_f64(vgetq_lane(a_.neon_f64, 1), temp, 1);
+#else
+	r_.f64[0] = (a_.f64[0] < b_.f64[0]) ? a_.f64[0] : b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_min_sd(a, b) simde_mm_min_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_max_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_max_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vmaxq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_max(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = vec_max(a_.altivec_i16, b_.altivec_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = (a_.i16[i] > b_.i16[i]) ? a_.i16[i] : b_.i16[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_max_epi16(a, b) simde_mm_max_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_max_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_max_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vmaxq_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u8x16_max(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u8 = vec_max(a_.altivec_u8, b_.altivec_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u8) / sizeof(r_.u8[0])); i++) {
+		r_.u8[i] = (a_.u8[i] > b_.u8[i]) ? a_.u8[i] : b_.u8[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_max_epu8(a, b) simde_mm_max_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_max_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_max_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+	r_.altivec_f64 = vec_max(a_.altivec_f64, b_.altivec_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_max(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vmaxq_f64(a_.neon_f64, b_.neon_f64);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = (a_.f64[i] > b_.f64[i]) ? a_.f64[i] : b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_max_pd(a, b) simde_mm_max_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_max_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_max_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_max_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x2_t temp = vmaxq_f64(a_.neon_f64, b_.neon_f64);
+	r_.neon_f64 = vsetq_lane_f64(vgetq_lane(a_.neon_f64, 1), temp, 1);
+#else
+	r_.f64[0] = (a_.f64[0] > b_.f64[0]) ? a_.f64[0] : b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_max_sd(a, b) simde_mm_max_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_move_epi64(simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_move_epi64(a);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vsetq_lane_s64(0, a_.neon_i64, 1);
+#else
+	r_.i64[0] = a_.i64[0];
+	r_.i64[1] = 0;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_move_epi64(a) simde_mm_move_epi64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_mul_epu32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_mul_epu32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint32x2_t a_lo = vmovn_u64(a_.neon_u64);
+	uint32x2_t b_lo = vmovn_u64(b_.neon_u64);
+	r_.neon_u64 = vmull_u32(a_lo, b_lo);
+#elif defined(SIMDE_SHUFFLE_VECTOR_) && \
+	(SIMDE_ENDIAN_ORDER == SIMDE_ENDIAN_LITTLE)
+	__typeof__(a_.u32) z = {
+		0,
+	};
+	a_.u32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.u32, z, 0, 4, 2, 6);
+	b_.u32 = SIMDE_SHUFFLE_VECTOR_(32, 16, b_.u32, z, 0, 4, 2, 6);
+	r_.u64 = HEDLEY_REINTERPRET_CAST(__typeof__(r_.u64), a_.u32) *
+		 HEDLEY_REINTERPRET_CAST(__typeof__(r_.u64), b_.u32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
+		r_.u64[i] = HEDLEY_STATIC_CAST(uint64_t, a_.u32[i * 2]) *
+			 HEDLEY_STATIC_CAST(uint64_t, b_.u32[i * 2]);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_epu32(a, b) simde_mm_mul_epu32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_mul_epi64(simde__m128i a, simde__m128i b)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 * b_.i64;
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vmulq_s64(a_.neon_f64, b_.neon_f64);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] * b_.i64[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_mod_epi64(simde__m128i a, simde__m128i b)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 % b_.i64;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] % b_.i64[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_mul_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_mul_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f64 = a_.f64 * b_.f64;
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vmulq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_mul(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = a_.f64[i] * b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_pd(a, b) simde_mm_mul_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_mul_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_mul_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_mul_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x2_t temp = vmulq_f64(a_.neon_f64, b_.neon_f64);
+	r_.neon_f64 = vsetq_lane_f64(vgetq_lane(a_.neon_f64, 1), temp, 1);
+#else
+	r_.f64[0] = a_.f64[0] * b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_sd(a, b) simde_mm_mul_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_mul_su32(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE) && \
+	!defined(__PGI)
+	return _mm_mul_su32(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.u64[0] = vget_lane_u64(
+		vget_low_u64(vmull_u32(vreinterpret_u32_s64(a_.neon_i64),
+				 vreinterpret_u32_s64(b_.neon_i64))),
+		0);
+#else
+	r_.u64[0] = HEDLEY_STATIC_CAST(uint64_t, a_.u32[0]) *
+		 HEDLEY_STATIC_CAST(uint64_t, b_.u32[0]);
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mul_su32(a, b) simde_mm_mul_su32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_mulhi_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_mulhi_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int16x4_t a3210 = vget_low_s16(a_.neon_i16);
+	int16x4_t b3210 = vget_low_s16(b_.neon_i16);
+	int32x4_t ab3210 = vmull_s16(a3210, b3210); /* 3333222211110000 */
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	int32x4_t ab7654 = vmull_high_s16(a_.neon_i16, b_.neon_i16);
+	r_.neon_i16 = vuzp2q_s16(vreinterpretq_s16_s32(ab3210),
+				 vreinterpretq_s16_s32(ab7654));
+#else
+	int16x4_t a7654 = vget_high_s16(a_.neon_i16);
+	int16x4_t b7654 = vget_high_s16(b_.neon_i16);
+	int32x4_t ab7654 = vmull_s16(a7654, b7654); /* 7777666655554444 */
+	uint16x8x2_t rv = vuzpq_u16(vreinterpretq_u16_s32(ab3210),
+				 vreinterpretq_u16_s32(ab7654));
+	r_.neon_u16 = rv.val[1];
+#endif
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(
+			uint16_t,
+			(HEDLEY_STATIC_CAST(
+				 uint32_t,
+				 HEDLEY_STATIC_CAST(int32_t, a_.i16[i]) *
+					 HEDLEY_STATIC_CAST(int32_t,
+							 b_.i16[i])) >>
+			 16));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mulhi_epi16(a, b) simde_mm_mulhi_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_mulhi_epu16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+	return _mm_mulhi_epu16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	uint16x4_t a3210 = vget_low_u16(a_.neon_u16);
+	uint16x4_t b3210 = vget_low_u16(b_.neon_u16);
+	uint32x4_t ab3210 = vmull_u16(a3210, b3210); /* 3333222211110000 */
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint32x4_t ab7654 = vmull_high_u16(a_.neon_u16, b_.neon_u16);
+	r_.neon_u16 = vuzp2q_u16(vreinterpretq_u16_u32(ab3210),
+				 vreinterpretq_u16_u32(ab7654));
+#else
+	uint16x4_t a7654 = vget_high_u16(a_.neon_u16);
+	uint16x4_t b7654 = vget_high_u16(b_.neon_u16);
+	uint32x4_t ab7654 = vmull_u16(a7654, b7654); /* 7777666655554444 */
+	uint16x8x2_t neon_r = vuzpq_u16(vreinterpretq_u16_u32(ab3210),
+					vreinterpretq_u16_u32(ab7654));
+	r_.neon_u16 = neon_r.val[1];
+#endif
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(
+			uint16_t,
+			HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
+					HEDLEY_STATIC_CAST(uint32_t,
+							 b_.u16[i]) >>
+				16);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mulhi_epu16(a, b) simde_mm_mulhi_epu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_mullo_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_mullo_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vmulq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	(void)a_;
+	(void)b_;
+	r_.altivec_i16 = vec_mul(a_.altivec_i16, b_.altivec_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(
+			uint16_t,
+			HEDLEY_STATIC_CAST(uint32_t, a_.u16[i]) *
+				HEDLEY_STATIC_CAST(uint32_t, b_.u16[i]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mullo_epi16(a, b) simde_mm_mullo_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_or_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_or_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f | b_.i32f;
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_or(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vorrq_s64(a_.neon_i64, b_.neon_i64);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] | b_.i32f[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_or_pd(a, b) simde_mm_or_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_or_si128(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_or_si128(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vorrq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_or(a_.altivec_i32, b_.altivec_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f | b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] | b_.i32f[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_or_si128(a, b) simde_mm_or_si128(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_packs_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_packs_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 =
+		vcombine_s8(vqmovn_s16(a_.neon_i16), vqmovn_s16(b_.neon_i16));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i8[i] = (a_.i16[i] > INT8_MAX)
+				 ? INT8_MAX
+				 : ((a_.i16[i] < INT8_MIN)
+					 ? INT8_MIN
+					 : HEDLEY_STATIC_CAST(int8_t,
+								 a_.i16[i]));
+		r_.i8[i + 8] = (b_.i16[i] > INT8_MAX)
+				 ? INT8_MAX
+				 : ((b_.i16[i] < INT8_MIN)
+						 ? INT8_MIN
+						 : HEDLEY_STATIC_CAST(
+							 int8_t, b_.i16[i]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_packs_epi16(a, b) simde_mm_packs_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_packs_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_packs_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 =
+		vcombine_s16(vqmovn_s32(a_.neon_i32), vqmovn_s32(b_.neon_i32));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i16 = vec_packs(a_.altivec_i32, b_.altivec_i32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i16[i] = (a_.i32[i] > INT16_MAX)
+				 ? INT16_MAX
+				 : ((a_.i32[i] < INT16_MIN)
+					 ? INT16_MIN
+					 : HEDLEY_STATIC_CAST(int16_t,
+								 a_.i32[i]));
+		r_.i16[i + 4] =
+			(b_.i32[i] > INT16_MAX)
+				? INT16_MAX
+				: ((b_.i32[i] < INT16_MIN)
+					 ? INT16_MIN
+					 : HEDLEY_STATIC_CAST(int16_t,
+								b_.i32[i]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_packs_epi32(a, b) simde_mm_packs_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_packus_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_packus_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 =
+		vcombine_u8(vqmovun_s16(a_.neon_i16), vqmovun_s16(b_.neon_i16));
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u8 = vec_packsu(a_.altivec_i16, b_.altivec_i16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.u8[i] = (a_.i16[i] > UINT8_MAX)
+				 ? UINT8_MAX
+				 : ((a_.i16[i] < 0)
+					 ? UINT8_C(0)
+					 : HEDLEY_STATIC_CAST(uint8_t,
+								 a_.i16[i]));
+		r_.u8[i + 8] =
+			(b_.i16[i] > UINT8_MAX)
+				? UINT8_MAX
+				: ((b_.i16[i] < 0)
+					 ? UINT8_C(0)
+					 : HEDLEY_STATIC_CAST(uint8_t,
+								b_.i16[i]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_packus_epi16(a, b) simde_mm_packus_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_pause(void)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_pause();
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_pause() (simde_mm_pause())
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sad_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sad_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const uint16x8_t t = vpaddlq_u8(vabdq_u8(a_.neon_u8, b_.neon_u8));
+	r_.neon_u64 = vcombine_u64(vpaddl_u32(vpaddl_u16(vget_low_u16(t))),
+				 vpaddl_u32(vpaddl_u16(vget_high_u16(t))));
+#else
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		uint16_t tmp = 0;
+		SIMDE_VECTORIZE_REDUCTION(+ : tmp)
+		for (size_t j = 0; j < ((sizeof(r_.u8) / sizeof(r_.u8[0])) / 2);
+		 j++) {
+			const size_t e = j + (i * 8);
+			tmp += (a_.u8[e] > b_.u8[e]) ? (a_.u8[e] - b_.u8[e])
+						 : (b_.u8[e] - a_.u8[e]);
+		}
+		r_.i64[i] = tmp;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sad_epu8(a, b) simde_mm_sad_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set_epi8(int8_t e15, int8_t e14, int8_t e13, int8_t e12,
+			 int8_t e11, int8_t e10, int8_t e9, int8_t e8,
+			 int8_t e7, int8_t e6, int8_t e5, int8_t e4,
+			 int8_t e3, int8_t e2, int8_t e1, int8_t e0)
+{
+
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5,
+			 e4, e3, e2, e1, e0);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_make(e0, e1, e2, e3, e4, e5, e6, e7, e8, e9,
+				 e10, e11, e12, e13, e14, e15);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(int8x16_t)
+	int8_t data[16] = {e0, e1, e2, e3, e4, e5, e6, e7,
+			 e8, e9, e10, e11, e12, e13, e14, e15};
+	r_.neon_i8 = vld1q_s8(data);
+#else
+	r_.i8[0] = e0;
+	r_.i8[1] = e1;
+	r_.i8[2] = e2;
+	r_.i8[3] = e3;
+	r_.i8[4] = e4;
+	r_.i8[5] = e5;
+	r_.i8[6] = e6;
+	r_.i8[7] = e7;
+	r_.i8[8] = e8;
+	r_.i8[9] = e9;
+	r_.i8[10] = e10;
+	r_.i8[11] = e11;
+	r_.i8[12] = e12;
+	r_.i8[13] = e13;
+	r_.i8[14] = e14;
+	r_.i8[15] = e15;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, e4, e3, \
+		 e2, e1, e0) \
+	simde_mm_set_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, \
+			 e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set_epi16(int16_t e7, int16_t e6, int16_t e5, int16_t e4,
+				int16_t e3, int16_t e2, int16_t e1, int16_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(int16x8_t)
+	int16_t data[8] = {e0, e1, e2, e3, e4, e5, e6, e7};
+	r_.neon_i16 = vld1q_s16(data);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_make(e0, e1, e2, e3, e4, e5, e6, e7);
+#else
+	r_.i16[0] = e0;
+	r_.i16[1] = e1;
+	r_.i16[2] = e2;
+	r_.i16[3] = e3;
+	r_.i16[4] = e4;
+	r_.i16[5] = e5;
+	r_.i16[6] = e6;
+	r_.i16[7] = e7;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0) \
+	simde_mm_set_epi16(e7, e6, e5, e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_loadu_si16(void const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	return _mm_loadu_si16(mem_addr);
+#else
+	int16_t val;
+	simde_memcpy(&val, mem_addr, sizeof(val));
+	return simde_x_mm_cvtsi16_si128(val);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_si16(mem_addr) simde_mm_loadu_si16(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set_epi32(int32_t e3, int32_t e2, int32_t e1, int32_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi32(e3, e2, e1, e0);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(int32x4_t) int32_t data[4] = {e0, e1, e2, e3};
+	r_.neon_i32 = vld1q_s32(data);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_make(e0, e1, e2, e3);
+#else
+	r_.i32[0] = e0;
+	r_.i32[1] = e1;
+	r_.i32[2] = e2;
+	r_.i32[3] = e3;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_epi32(e3, e2, e1, e0) simde_mm_set_epi32(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_loadu_si32(void const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	return _mm_loadu_si32(mem_addr);
+#else
+	int32_t val;
+	simde_memcpy(&val, mem_addr, sizeof(val));
+	return simde_mm_cvtsi32_si128(val);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_si32(mem_addr) simde_mm_loadu_si32(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set_epi64(simde__m64 e1, simde__m64 e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set_epi64(e1, e0);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vcombine_s64(simde__m64_to_neon_i64(e0),
+				 simde__m64_to_neon_i64(e1));
+#else
+	r_.m64[0] = e0;
+	r_.m64[1] = e1;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_epi64(e1, e0) (simde_mm_set_epi64((e1), (e0)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set_epi64x(int64_t e1, int64_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
+	return _mm_set_epi64x(e1, e0);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(int64x2_t) int64_t data[2] = {e0, e1};
+	r_.neon_i64 = vld1q_s64(data);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i64x2_make(e0, e1);
+#else
+	r_.i64[0] = e0;
+	r_.i64[1] = e1;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_epi64x(e1, e0) simde_mm_set_epi64x(e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_loadu_si64(void const *mem_addr)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	return _mm_loadu_si64(mem_addr);
+#else
+	int64_t val;
+	simde_memcpy(&val, mem_addr, sizeof(val));
+	return simde_mm_cvtsi64_si128(val);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_loadu_si64(mem_addr) simde_mm_loadu_si64(mem_addr)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set_epu8(uint8_t e15, uint8_t e14, uint8_t e13,
+				 uint8_t e12, uint8_t e11, uint8_t e10,
+				 uint8_t e9, uint8_t e8, uint8_t e7, uint8_t e6,
+				 uint8_t e5, uint8_t e4, uint8_t e3, uint8_t e2,
+				 uint8_t e1, uint8_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi8(
+		HEDLEY_STATIC_CAST(char, e15), HEDLEY_STATIC_CAST(char, e14),
+		HEDLEY_STATIC_CAST(char, e13), HEDLEY_STATIC_CAST(char, e12),
+		HEDLEY_STATIC_CAST(char, e11), HEDLEY_STATIC_CAST(char, e10),
+		HEDLEY_STATIC_CAST(char, e9), HEDLEY_STATIC_CAST(char, e8),
+		HEDLEY_STATIC_CAST(char, e7), HEDLEY_STATIC_CAST(char, e6),
+		HEDLEY_STATIC_CAST(char, e5), HEDLEY_STATIC_CAST(char, e4),
+		HEDLEY_STATIC_CAST(char, e3), HEDLEY_STATIC_CAST(char, e2),
+		HEDLEY_STATIC_CAST(char, e1), HEDLEY_STATIC_CAST(char, e0));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(uint8x16_t)
+	uint8_t data[16] = {e0, e1, e2, e3, e4, e5, e6, e7,
+			 e8, e9, e10, e11, e12, e13, e14, e15};
+	r_.neon_u8 = vld1q_u8(data);
+#else
+	r_.u8[0] = e0;
+	r_.u8[1] = e1;
+	r_.u8[2] = e2;
+	r_.u8[3] = e3;
+	r_.u8[4] = e4;
+	r_.u8[5] = e5;
+	r_.u8[6] = e6;
+	r_.u8[7] = e7;
+	r_.u8[8] = e8;
+	r_.u8[9] = e9;
+	r_.u8[10] = e10;
+	r_.u8[11] = e11;
+	r_.u8[12] = e12;
+	r_.u8[13] = e13;
+	r_.u8[14] = e14;
+	r_.u8[15] = e15;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set_epu16(uint16_t e7, uint16_t e6, uint16_t e5,
+				 uint16_t e4, uint16_t e3, uint16_t e2,
+				 uint16_t e1, uint16_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi16(
+		HEDLEY_STATIC_CAST(short, e7), HEDLEY_STATIC_CAST(short, e6),
+		HEDLEY_STATIC_CAST(short, e5), HEDLEY_STATIC_CAST(short, e4),
+		HEDLEY_STATIC_CAST(short, e3), HEDLEY_STATIC_CAST(short, e2),
+		HEDLEY_STATIC_CAST(short, e1), HEDLEY_STATIC_CAST(short, e0));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(uint16x8_t)
+	uint16_t data[8] = {e0, e1, e2, e3, e4, e5, e6, e7};
+	r_.neon_u16 = vld1q_u16(data);
+#else
+	r_.u16[0] = e0;
+	r_.u16[1] = e1;
+	r_.u16[2] = e2;
+	r_.u16[3] = e3;
+	r_.u16[4] = e4;
+	r_.u16[5] = e5;
+	r_.u16[6] = e6;
+	r_.u16[7] = e7;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set_epu32(uint32_t e3, uint32_t e2, uint32_t e1,
+				 uint32_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_epi32(HEDLEY_STATIC_CAST(int, e3),
+			 HEDLEY_STATIC_CAST(int, e2),
+			 HEDLEY_STATIC_CAST(int, e1),
+			 HEDLEY_STATIC_CAST(int, e0));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(uint32x4_t) uint32_t data[4] = {e0, e1, e2, e3};
+	r_.neon_u32 = vld1q_u32(data);
+#else
+	r_.u32[0] = e0;
+	r_.u32[1] = e1;
+	r_.u32[2] = e2;
+	r_.u32[3] = e3;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set_epu64x(uint64_t e1, uint64_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
+	return _mm_set_epi64x(HEDLEY_STATIC_CAST(int64_t, e1),
+			 HEDLEY_STATIC_CAST(int64_t, e0));
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	SIMDE_ALIGN_LIKE_16(uint64x2_t) uint64_t data[2] = {e0, e1};
+	r_.neon_u64 = vld1q_u64(data);
+#else
+	r_.u64[0] = e0;
+	r_.u64[1] = e1;
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_set_sd(simde_float64 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set_sd(a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	return vsetq_lane_f64(a, vdupq_n_f64(SIMDE_FLOAT64_C(0.0)), 0);
+#else
+	return simde_mm_set_pd(SIMDE_FLOAT64_C(0.0), a);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set_sd(a) simde_mm_set_sd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set1_epi8(int8_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set1_epi8(a);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vdupq_n_s8(a);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_splat(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_i8 = vec_splats(HEDLEY_STATIC_CAST(signed char, a));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = a;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_epi8(a) simde_mm_set1_epi8(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set1_epi16(int16_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set1_epi16(a);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vdupq_n_s16(a);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_splat(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_i16 = vec_splats(HEDLEY_STATIC_CAST(signed short, a));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_epi16(a) simde_mm_set1_epi16(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set1_epi32(int32_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_set1_epi32(a);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vdupq_n_s32(a);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_splat(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_i32 = vec_splats(HEDLEY_STATIC_CAST(signed int, a));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_epi32(a) simde_mm_set1_epi32(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set1_epi64x(int64_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(!defined(HEDLEY_MSVC_VERSION) || HEDLEY_MSVC_VERSION_CHECK(19, 0, 0))
+	return _mm_set1_epi64x(a);
+#else
+	simde__m128i_private r_;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vdupq_n_s64(a);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i64x2_splat(a);
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	r_.altivec_i64 = vec_splats(HEDLEY_STATIC_CAST(signed long long, a));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_epi64x(a) simde_mm_set1_epi64x(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_set1_epi64(simde__m64 a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_set1_epi64(a);
+#else
+	simde__m64_private a_ = simde__m64_to_private(a);
+	return simde_mm_set1_epi64x(a_.i64[0]);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_set1_epi64(a) simde_mm_set1_epi64(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set1_epu8(uint8_t value)
+{
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	return simde__m128i_from_altivec_u8(
+		vec_splats(HEDLEY_STATIC_CAST(unsigned char, value)));
+#else
+	return simde_mm_set1_epi8(HEDLEY_STATIC_CAST(int8_t, value));
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set1_epu16(uint16_t value)
+{
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	return simde__m128i_from_altivec_u16(
+		vec_splats(HEDLEY_STATIC_CAST(unsigned short, value)));
+#else
+	return simde_mm_set1_epi16(HEDLEY_STATIC_CAST(int16_t, value));
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set1_epu32(uint32_t value)
+{
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	return simde__m128i_from_altivec_u32(
+		vec_splats(HEDLEY_STATIC_CAST(unsigned int, value)));
+#else
+	return simde_mm_set1_epi32(HEDLEY_STATIC_CAST(int32_t, value));
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_set1_epu64(uint64_t value)
+{
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+	return simde__m128i_from_altivec_u64(
+		vec_splats(HEDLEY_STATIC_CAST(unsigned long long, value)));
+#else
+	return simde_mm_set1_epi64x(HEDLEY_STATIC_CAST(int64_t, value));
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_setr_epi8(int8_t e15, int8_t e14, int8_t e13, int8_t e12,
+				int8_t e11, int8_t e10, int8_t e9, int8_t e8,
+				int8_t e7, int8_t e6, int8_t e5, int8_t e4,
+				int8_t e3, int8_t e2, int8_t e1, int8_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5,
+			 e4, e3, e2, e1, e0);
+#else
+	return simde_mm_set_epi8(e0, e1, e2, e3, e4, e5, e6, e7, e8, e9, e10,
+				 e11, e12, e13, e14, e15);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, e4, \
+		 e3, e2, e1, e0) \
+	simde_mm_setr_epi8(e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, \
+			 e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_setr_epi16(int16_t e7, int16_t e6, int16_t e5, int16_t e4,
+				 int16_t e3, int16_t e2, int16_t e1, int16_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0);
+#else
+	return simde_mm_set_epi16(e0, e1, e2, e3, e4, e5, e6, e7);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0) \
+	simde_mm_setr_epi16(e7, e6, e5, e4, e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_setr_epi32(int32_t e3, int32_t e2, int32_t e1, int32_t e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setr_epi32(e3, e2, e1, e0);
+#else
+	return simde_mm_set_epi32(e0, e1, e2, e3);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_epi32(e3, e2, e1, e0) simde_mm_setr_epi32(e3, e2, e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_setr_epi64(simde__m64 e1, simde__m64 e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_setr_epi64(e1, e0);
+#else
+	return simde_mm_set_epi64(e0, e1);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_epi64(e1, e0) (simde_mm_setr_epi64((e1), (e0)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_setr_pd(simde_float64 e1, simde_float64 e0)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setr_pd(e1, e0);
+#else
+	return simde_mm_set_pd(e0, e1);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setr_pd(e1, e0) simde_mm_setr_pd(e1, e0)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_setzero_pd(void)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_setzero_pd();
+#else
+	return simde_mm_castsi128_pd(simde_mm_setzero_si128());
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_setzero_pd() simde_mm_setzero_pd()
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_undefined_pd(void)
+{
+	simde__m128d_private r_;
+
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE__HAVE_UNDEFINED128)
+	r_.n = _mm_undefined_pd();
+#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+	r_ = simde__m128d_to_private(simde_mm_setzero_pd());
+#endif
+
+	return simde__m128d_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_undefined_pd() simde_mm_undefined_pd()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_undefined_si128(void)
+{
+	simde__m128i_private r_;
+
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE__HAVE_UNDEFINED128)
+	r_.n = _mm_undefined_si128();
+#elif !defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+	r_ = simde__m128i_to_private(simde_mm_setzero_si128());
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_undefined_si128() (simde_mm_undefined_si128())
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_POP
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_setone_pd(void)
+{
+	return simde_mm_castps_pd(simde_x_mm_setone_ps());
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_setone_si128(void)
+{
+	return simde_mm_castps_si128(simde_x_mm_setone_ps());
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_shuffle_epi32(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[(imm8 >> (i * 2)) & 3];
+	}
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_shuffle_epi32(a, imm8) _mm_shuffle_epi32((a), (imm8))
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_shuffle_epi32(a, imm8) \
+	__extension__({ \
+		int32x4_t ret; \
+		ret = vmovq_n_s32(vgetq_lane_s32(vreinterpretq_s32_s64(a), \
+						 (imm8) & (0x3))); \
+		ret = vsetq_lane_s32(vgetq_lane_s32(vreinterpretq_s32_s64(a), \
+						 ((imm8) >> 2) & 0x3), \
+				 ret, 1); \
+		ret = vsetq_lane_s32(vgetq_lane_s32(vreinterpretq_s32_s64(a), \
+						 ((imm8) >> 4) & 0x3), \
+				 ret, 2); \
+		ret = vsetq_lane_s32(vgetq_lane_s32(vreinterpretq_s32_s64(a), \
+						 ((imm8) >> 6) & 0x3), \
+				 ret, 3); \
+		vreinterpretq_s64_s32(ret); \
+	})
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shuffle_epi32(a, imm8) \
+	(__extension__({ \
+		const simde__m128i_private simde__tmp_a_ = \
+			simde__m128i_to_private(a); \
+		simde__m128i_from_private((simde__m128i_private){ \
+			.i32 = SIMDE_SHUFFLE_VECTOR_( \
+				32, 16, (simde__tmp_a_).i32, \
+				(simde__tmp_a_).i32, ((imm8)) & 3, \
+				((imm8) >> 2) & 3, ((imm8) >> 4) & 3, \
+				((imm8) >> 6) & 3)}); \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_shuffle_epi32(a, imm8) simde_mm_shuffle_epi32(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_shuffle_pd(simde__m128d a, simde__m128d b, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 3)
+{
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.f64[0] = ((imm8 & 1) == 0) ? a_.f64[0] : a_.f64[1];
+	r_.f64[1] = ((imm8 & 2) == 0) ? b_.f64[0] : b_.f64[1];
+
+	return simde__m128d_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(__PGI)
+#define simde_mm_shuffle_pd(a, b, imm8) _mm_shuffle_pd((a), (b), (imm8))
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shuffle_pd(a, b, imm8) \
+	(__extension__({ \
+		simde__m128d_from_private((simde__m128d_private){ \
+			.f64 = SIMDE_SHUFFLE_VECTOR_( \
+				64, 16, simde__m128d_to_private(a).f64, \
+				simde__m128d_to_private(b).f64, \
+				(((imm8)) & 1), (((imm8) >> 1) & 1) + 2)}); \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_shuffle_pd(a, b, imm8) simde_mm_shuffle_pd(a, b, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_shufflehi_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
+	 i++) {
+		r_.i16[i] = a_.i16[i];
+	}
+	for (size_t i = ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
+	 i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[((imm8 >> ((i - 4) * 2)) & 3) + 4];
+	}
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_shufflehi_epi16(a, imm8) _mm_shufflehi_epi16((a), (imm8))
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_shufflehi_epi16(a, imm8) \
+	__extension__({ \
+		int16x8_t ret = vreinterpretq_s16_s64(a); \
+		int16x4_t highBits = vget_high_s16(ret); \
+		ret = vsetq_lane_s16(vget_lane_s16(highBits, (imm8) & (0x3)), \
+				 ret, 4); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(highBits, ((imm8) >> 2) & 0x3), ret, 5); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(highBits, ((imm8) >> 4) & 0x3), ret, 6); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(highBits, ((imm8) >> 6) & 0x3), ret, 7); \
+		vreinterpretq_s64_s16(ret); \
+	})
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shufflehi_epi16(a, imm8) \
+	(__extension__({ \
+		const simde__m128i_private simde__tmp_a_ = \
+			simde__m128i_to_private(a); \
+		simde__m128i_from_private((simde__m128i_private){ \
+			.i16 = SIMDE_SHUFFLE_VECTOR_( \
+				16, 16, (simde__tmp_a_).i16, \
+				(simde__tmp_a_).i16, 0, 1, 2, 3, \
+				(((imm8)) & 3) + 4, (((imm8) >> 2) & 3) + 4, \
+				(((imm8) >> 4) & 3) + 4, \
+				(((imm8) >> 6) & 3) + 4)}); \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_shufflehi_epi16(a, imm8) simde_mm_shufflehi_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_shufflelo_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	for (size_t i = 0; i < ((sizeof(r_.i16) / sizeof(r_.i16[0])) / 2);
+	 i++) {
+		r_.i16[i] = a_.i16[((imm8 >> (i * 2)) & 3)];
+	}
+	SIMDE_VECTORIZE
+	for (size_t i = ((sizeof(a_.i16) / sizeof(a_.i16[0])) / 2);
+	 i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i];
+	}
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_shufflelo_epi16(a, imm8) _mm_shufflelo_epi16((a), (imm8))
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_shufflelo_epi16(a, imm8) \
+	__extension__({ \
+		int16x8_t ret = vreinterpretq_s16_s64(a); \
+		int16x4_t lowBits = vget_low_s16(ret); \
+		ret = vsetq_lane_s16(vget_lane_s16(lowBits, (imm8) & (0x3)), \
+				 ret, 0); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(lowBits, ((imm8) >> 2) & 0x3), ret, 1); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(lowBits, ((imm8) >> 4) & 0x3), ret, 2); \
+		ret = vsetq_lane_s16( \
+			vget_lane_s16(lowBits, ((imm8) >> 6) & 0x3), ret, 3); \
+		vreinterpretq_s64_s16(ret); \
+	})
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+#define simde_mm_shufflelo_epi16(a, imm8) \
+	(__extension__({ \
+		const simde__m128i_private simde__tmp_a_ = \
+			simde__m128i_to_private(a); \
+		simde__m128i_from_private((simde__m128i_private){ \
+			.i16 = SIMDE_SHUFFLE_VECTOR_( \
+				16, 16, (simde__tmp_a_).i16, \
+				(simde__tmp_a_).i16, (((imm8)) & 3), \
+				(((imm8) >> 2) & 3), (((imm8) >> 4) & 3), \
+				(((imm8) >> 6) & 3), 4, 5, 6, 7)}); \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_shufflelo_epi16(a, imm8) simde_mm_shufflelo_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sll_epi16(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sll_epi16(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	if (count_.u64[0] > 15)
+		return simde_mm_setzero_si128();
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u16 = (a_.u16 << count_.u64[0]);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vshlq_u16(a_.neon_u16, vdupq_n_s16(HEDLEY_STATIC_CAST(
+						 int16_t, count_.u64[0])));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 =
+		((wasm_i64x2_extract_lane(count_.wasm_v128, 0) < 16)
+			 ? wasm_i16x8_shl(a_.wasm_v128,
+					 HEDLEY_STATIC_CAST(
+						 int32_t,
+						 wasm_i64x2_extract_lane(
+							 count_.wasm_v128, 0)))
+			 : wasm_i16x8_const(0, 0, 0, 0, 0, 0, 0, 0));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t,
+					 (a_.u16[i] << count_.u64[0]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_epi16(a, count) simde_mm_sll_epi16((a), (count))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sll_epi32(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sll_epi32(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	if (count_.u64[0] > 31)
+		return simde_mm_setzero_si128();
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u32 = (a_.u32 << count_.u64[0]);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vshlq_u32(a_.neon_u32, vdupq_n_s32(HEDLEY_STATIC_CAST(
+						 int32_t, count_.u64[0])));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 =
+		((wasm_i64x2_extract_lane(count_.wasm_v128, 0) < 32)
+			 ? wasm_i32x4_shl(a_.wasm_v128,
+					 HEDLEY_STATIC_CAST(
+						 int32_t,
+						 wasm_i64x2_extract_lane(
+							 count_.wasm_v128, 0)))
+			 : wasm_i32x4_const(0, 0, 0, 0));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = HEDLEY_STATIC_CAST(uint32_t,
+					 (a_.u32[i] << count_.u64[0]));
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_epi32(a, count) (simde_mm_sll_epi32(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sll_epi64(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sll_epi64(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	if (count_.u64[0] > 63)
+		return simde_mm_setzero_si128();
+
+	const int_fast16_t s = HEDLEY_STATIC_CAST(int_fast16_t, count_.u64[0]);
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u64 = vshlq_u64(a_.neon_u64,
+				vdupq_n_s64(HEDLEY_STATIC_CAST(int64_t, s)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = (s < 64) ? wasm_i64x2_shl(a_.wasm_v128, s)
+				: wasm_i64x2_const(0, 0);
+#else
+#if !defined(SIMDE_BUG_GCC_94488)
+	SIMDE_VECTORIZE
+#endif
+	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
+		r_.u64[i] = a_.u64[i] << s;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sll_epi64(a, count) (simde_mm_sll_epi64(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_sqrt_pd(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sqrt_pd(a);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vsqrtq_f64(a_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_sqrt(a_.wasm_v128);
+#elif defined(simde_math_sqrt)
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = simde_math_sqrt(a_.f64[i]);
+	}
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sqrt_pd(a) simde_mm_sqrt_pd(a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_sqrt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sqrt_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_sqrt_pd(b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(simde_math_sqrt)
+	r_.f64[0] = simde_math_sqrt(b_.f64[0]);
+	r_.f64[1] = a_.f64[1];
+#else
+	HEDLEY_UNREACHABLE();
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sqrt_sd(a, b) simde_mm_sqrt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srl_epi16(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_srl_epi16(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	const int cnt = HEDLEY_STATIC_CAST(
+		int, (count_.i64[0] > 16 ? 16 : count_.i64[0]));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vshlq_u16(a_.neon_u16,
+				vdupq_n_s16(HEDLEY_STATIC_CAST(int16_t, -cnt)));
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u16) / sizeof(r_.u16[0])); i++) {
+		r_.u16[i] = a_.u16[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_epi16(a, count) (simde_mm_srl_epi16(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srl_epi32(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_srl_epi32(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	const int cnt = HEDLEY_STATIC_CAST(
+		int, (count_.i64[0] > 32 ? 32 : count_.i64[0]));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vshlq_u32(a_.neon_u32,
+				vdupq_n_s32(HEDLEY_STATIC_CAST(int32_t, -cnt)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u32x4_shr(a_.wasm_v128, cnt);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_epi32(a, count) (simde_mm_srl_epi32(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srl_epi64(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_srl_epi64(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	const int cnt = HEDLEY_STATIC_CAST(
+		int, (count_.i64[0] > 64 ? 64 : count_.i64[0]));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u64 = vshlq_u64(a_.neon_u64,
+				vdupq_n_s64(HEDLEY_STATIC_CAST(int64_t, -cnt)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u64x2_shr(a_.wasm_v128, cnt);
+#else
+#if !defined(SIMDE_BUG_GCC_94488)
+	SIMDE_VECTORIZE
+#endif
+	for (size_t i = 0; i < (sizeof(r_.u64) / sizeof(r_.u64[0])); i++) {
+		r_.u64[i] = a_.u64[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srl_epi64(a, count) (simde_mm_srl_epi64(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srai_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	/* MSVC requires a range of (0, 255). */
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	const int cnt = (imm8 & ~15) ? 15 : imm8;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vshlq_s16(a_.neon_i16,
+				vdupq_n_s16(HEDLEY_STATIC_CAST(int16_t, -cnt)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_shr(a_.wasm_v128, cnt);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_srai_epi16(a, imm8) _mm_srai_epi16((a), (imm8))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srai_epi16(a, imm8) simde_mm_srai_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srai_epi32(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	/* MSVC requires a range of (0, 255). */
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	const int cnt = (imm8 & ~31) ? 31 : imm8;
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vshlq_s32(a_.neon_i32, vdupq_n_s32(-cnt));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_shr(a_.wasm_v128, cnt);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_srai_epi32(a, imm8) _mm_srai_epi32((a), (imm8))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srai_epi32(a, imm8) simde_mm_srai_epi32(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sra_epi16(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sra_epi16(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	const int cnt = HEDLEY_STATIC_CAST(
+		int, (count_.i64[0] > 15 ? 15 : count_.i64[0]));
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vshlq_s16(a_.neon_i16,
+				vdupq_n_s16(HEDLEY_STATIC_CAST(int16_t, -cnt)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_shr(a_.wasm_v128, cnt);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sra_epi16(a, count) (simde_mm_sra_epi16(a, count))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sra_epi32(simde__m128i a, simde__m128i count)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && !defined(SIMDE_BUG_GCC_BAD_MM_SRA_EPI32)
+	return _mm_sra_epi32(a, count);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 count_ = simde__m128i_to_private(count);
+
+	const int cnt = count_.u64[0] > 31
+				? 31
+				: HEDLEY_STATIC_CAST(int, count_.u64[0]);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vshlq_s32(a_.neon_i32,
+				vdupq_n_s32(HEDLEY_STATIC_CAST(int32_t, -cnt)));
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i32x4_shr(a_.wasm_v128, cnt);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] >> cnt;
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sra_epi32(a, count) (simde_mm_sra_epi32(a, (count)))
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_slli_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	if (HEDLEY_UNLIKELY((imm8 > 15))) {
+		return simde_mm_setzero_si128();
+	}
+
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i16 = a_.i16 << (imm8 & 0xff);
+#else
+	const int s =
+		(imm8 >
+		 HEDLEY_STATIC_CAST(int, sizeof(r_.i16[0]) * CHAR_BIT) - 1)
+			? 0
+			: imm8;
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = HEDLEY_STATIC_CAST(int16_t, a_.i16[i] << s);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_slli_epi16(a, imm8) _mm_slli_epi16(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_slli_epi16(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 15) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_i16(vshlq_n_s16( \
+				simde__m128i_to_neon_i16(a), ((imm8)&15))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_slli_epi16(a, imm8) \
+	((imm8 < 16) \
+		 ? wasm_i16x8_shl(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i16x8_const(0, 0, 0, 0, 0, 0, 0, 0))
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+#define simde_mm_slli_epi16(a, imm8) \
+	((imm8 & ~15) ? simde_mm_setzero_si128() \
+		 : simde__m128i_from_altivec_i16( \
+				vec_sl(simde__m128i_to_altivec_i16(a), \
+				 vec_splat_u16(HEDLEY_STATIC_CAST( \
+					 unsigned short, imm8)))))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_epi16(a, imm8) simde_mm_slli_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_slli_epi32(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	if (HEDLEY_UNLIKELY((imm8 > 31))) {
+		return simde_mm_setzero_si128();
+	}
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i32 = a_.i32 << imm8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] << (imm8 & 0xff);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_slli_epi32(a, imm8) _mm_slli_epi32(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_slli_epi32(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 31) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_i32(vshlq_n_s32( \
+				simde__m128i_to_neon_i32(a), ((imm8)&31))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_slli_epi32(a, imm8) \
+	((imm8 < 32) \
+		 ? wasm_i32x4_shl(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i32x4_const(0, 0, 0, 0))
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+#define simde_mm_slli_epi32(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 31) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_altivec_i32( \
+				vec_sl(simde__m128i_to_altivec_i32(a), \
+				 vec_splats(HEDLEY_STATIC_CAST( \
+					 unsigned int, (imm8)&31)))); \
+		} \
+		ret; \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_epi32(a, imm8) simde_mm_slli_epi32(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_slli_epi64(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	if (HEDLEY_UNLIKELY((imm8 > 63))) {
+		return simde_mm_setzero_si128();
+	}
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.i64 = a_.i64 << imm8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] << (imm8 & 0xff);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_slli_epi64(a, imm8) _mm_slli_epi64(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_slli_epi64(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 63) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_i64(vshlq_n_s64( \
+				simde__m128i_to_neon_i64(a), ((imm8)&63))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_slli_epi64(a, imm8) \
+	((imm8 < 64) \
+		 ? wasm_i64x2_shl(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i64x2_const(0, 0))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_slli_epi64(a, imm8) simde_mm_slli_epi64(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srli_epi16(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	if (HEDLEY_UNLIKELY((imm8 > 15))) {
+		return simde_mm_setzero_si128();
+	}
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u16 = a_.u16 >> imm8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.u16[i] = a_.u16[i] >> (imm8 & 0xff);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_srli_epi16(a, imm8) _mm_srli_epi16(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_srli_epi16(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 15) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_u16(vshrq_n_u16( \
+				simde__m128i_to_neon_u16(a), \
+				(((imm8)&15) | (((imm8)&15) == 0)))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_srli_epi16(a, imm8) \
+	((imm8 < 16) \
+		 ? wasm_u16x8_shr(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i16x8_const(0, 0, 0, 0, 0, 0, 0, 0))
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+#define simde_mm_srli_epi16(a, imm8) \
+	((imm8 & ~15) ? simde_mm_setzero_si128() \
+		 : simde__m128i_from_altivec_i16( \
+				vec_sr(simde__m128i_to_altivec_i16(a), \
+				 vec_splat_u16(HEDLEY_STATIC_CAST( \
+					 unsigned short, imm8)))))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_epi16(a, imm8) simde_mm_srli_epi16(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srli_epi32(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	if (HEDLEY_UNLIKELY((imm8 > 31))) {
+		return simde_mm_setzero_si128();
+	}
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR)
+	r_.u32 = a_.u32 >> (imm8 & 0xff);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.u32[i] = a_.u32[i] >> (imm8 & 0xff);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_srli_epi32(a, imm8) _mm_srli_epi32(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_srli_epi32(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 31) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_u32(vshrq_n_u32( \
+				simde__m128i_to_neon_u32(a), \
+				(((imm8)&31) | (((imm8)&31) == 0)))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_srli_epi32(a, imm8) \
+	((imm8 < 32) \
+		 ? wasm_u32x4_shr(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i32x4_const(0, 0, 0, 0))
+#elif defined(SIMDE_POWER_ALTIVEC_P8_NATIVE)
+#define simde_mm_srli_epi32(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 31) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_altivec_i32( \
+				vec_sr(simde__m128i_to_altivec_i32(a), \
+				 vec_splats(HEDLEY_STATIC_CAST( \
+					 unsigned int, (imm8)&31)))); \
+		} \
+		ret; \
+	}))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_epi32(a, imm8) simde_mm_srli_epi32(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_srli_epi64(simde__m128i a, const int imm8)
+	SIMDE_REQUIRE_CONSTANT_RANGE(imm8, 0, 255)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+	if (HEDLEY_UNLIKELY((imm8 & 63) != imm8))
+		return simde_mm_setzero_si128();
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u64 = vshlq_u64(a_.neon_u64, vdupq_n_s64(-imm8));
+#else
+#if defined(SIMDE_VECTOR_SUBSCRIPT_SCALAR) && !defined(SIMDE_BUG_GCC_94488)
+	r_.u64 = a_.u64 >> imm8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.u64[i] = a_.u64[i] >> imm8;
+	}
+#endif
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+#if defined(SIMDE_X86_SSE2_NATIVE)
+#define simde_mm_srli_epi64(a, imm8) _mm_srli_epi64(a, imm8)
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+#define simde_mm_srli_epi64(a, imm8) \
+	(__extension__({ \
+		simde__m128i ret; \
+		if ((imm8) <= 0) { \
+			ret = a; \
+		} else if ((imm8) > 63) { \
+			ret = simde_mm_setzero_si128(); \
+		} else { \
+			ret = simde__m128i_from_neon_u64(vshrq_n_u64( \
+				simde__m128i_to_neon_u64(a), \
+				(((imm8)&63) | (((imm8)&63) == 0)))); \
+		} \
+		ret; \
+	}))
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+#define simde_mm_srli_epi64(a, imm8) \
+	((imm8 < 64) \
+		 ? wasm_u64x2_shr(simde__m128i_to_private(a).wasm_v128, imm8) \
+		 : wasm_i64x2_const(0, 0))
+#endif
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_srli_epi64(a, imm8) simde_mm_srli_epi64(a, imm8)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
+		 simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_store_pd(mem_addr, a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	vst1q_f64(mem_addr, simde__m128d_to_private(a).neon_f64);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_s64(HEDLEY_REINTERPRET_CAST(int64_t *, mem_addr),
+		 simde__m128d_to_private(a).neon_i64);
+#else
+	simde_memcpy(SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128d), &a,
+		 sizeof(a));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_store_pd(mem_addr, a) \
+	simde_mm_store_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store1_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
+			simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_store1_pd(mem_addr, a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	vst1q_f64(mem_addr, vdupq_laneq_f64(a_.neon_f64, 0));
+#else
+	mem_addr[0] = a_.f64[0];
+	mem_addr[1] = a_.f64[0];
+#endif
+#endif
+}
+#define simde_mm_store_pd1(mem_addr, a) \
+	simde_mm_store1_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_store1_pd(mem_addr, a) \
+	simde_mm_store1_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#define _mm_store_pd1(mem_addr, a) \
+	simde_mm_store_pd1(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store_sd(simde_float64 *mem_addr, simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_store_sd(mem_addr, a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	const simde_float64 v = vgetq_lane_f64(a_.neon_f64, 0);
+	simde_memcpy(mem_addr, &v, sizeof(v));
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	const int64_t v = vgetq_lane_s64(a_.neon_i64, 0);
+	simde_memcpy(HEDLEY_REINTERPRET_CAST(int64_t *, mem_addr), &v,
+		 sizeof(v));
+#else
+	simde_float64 v = a_.f64[0];
+	simde_memcpy(mem_addr, &v, sizeof(simde_float64));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_store_sd(mem_addr, a) \
+	simde_mm_store_sd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_store_si128(simde__m128i *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_store_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
+#else
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_s32(HEDLEY_REINTERPRET_CAST(int32_t *, mem_addr), a_.neon_i32);
+#else
+	simde_memcpy(SIMDE_ALIGN_ASSUME_LIKE(mem_addr, simde__m128i), &a_,
+		 sizeof(a_));
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_store_si128(mem_addr, a) simde_mm_store_si128(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeh_pd(simde_float64 *mem_addr, simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storeh_pd(mem_addr, a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	*mem_addr = vgetq_lane_f64(a_.neon_f64, 1);
+#else
+	*mem_addr = a_.f64[1];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeh_pd(mem_addr, a) \
+	simde_mm_storeh_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storel_epi64(simde__m128i *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storel_epi64(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
+#else
+	simde__m128i_private a_ = simde__m128i_to_private(a);
+	int64_t tmp;
+
+	/* memcpy to prevent aliasing, tmp because we can't take the
+ * address of a vector element. */
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	tmp = vgetq_lane_s64(a_.neon_i64, 0);
+#elif defined(SIMDE_POWER_ALTIVEC_P7_NATIVE)
+#if defined(SIMDE_BUG_GCC_95227)
+	(void)a_;
+#endif
+	tmp = vec_extract(a_.altivec_i64, 0);
+#else
+	tmp = a_.i64[0];
+#endif
+
+	simde_memcpy(mem_addr, &tmp, sizeof(tmp));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storel_epi64(mem_addr, a) simde_mm_storel_epi64(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storel_pd(simde_float64 *mem_addr, simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storel_pd(mem_addr, a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+	simde_float64 tmp;
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	tmp = vgetq_lane_f64(a_.neon_f64, 0);
+#else
+	tmp = a_.f64[0];
+#endif
+	simde_memcpy(mem_addr, &tmp, sizeof(tmp));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storel_pd(mem_addr, a) \
+	simde_mm_storel_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storer_pd(simde_float64 mem_addr[2], simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storer_pd(mem_addr, a);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	vst1q_s64(HEDLEY_REINTERPRET_CAST(int64_t *, mem_addr),
+		 vextq_s64(a_.neon_i64, a_.neon_i64, 1));
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	a_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, a_.f64, 1, 0);
+	simde_mm_store_pd(mem_addr, simde__m128d_from_private(a_));
+#else
+	mem_addr[0] = a_.f64[1];
+	mem_addr[1] = a_.f64[0];
+#endif
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storer_pd(mem_addr, a) \
+	simde_mm_storer_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_pd(simde_float64 *mem_addr, simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storeu_pd(mem_addr, a);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	vst1q_f64(mem_addr, simde__m128d_to_private(a).neon_f64);
+#else
+	simde_memcpy(mem_addr, &a, sizeof(a));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_pd(mem_addr, a) \
+	simde_mm_storeu_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_si128(simde__m128i *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_storeu_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
+#else
+	simde_memcpy(mem_addr, &a, sizeof(a));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_si128(mem_addr, a) simde_mm_storeu_si128(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_si16(void *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	_mm_storeu_si16(mem_addr, a);
+#else
+	int16_t val = simde_x_mm_cvtsi128_si16(a);
+	simde_memcpy(mem_addr, &val, sizeof(val));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_si16(mem_addr, a) simde_mm_storeu_si16(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_si32(void *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	_mm_storeu_si32(mem_addr, a);
+#else
+	int32_t val = simde_mm_cvtsi128_si32(a);
+	simde_memcpy(mem_addr, &val, sizeof(val));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_si32(mem_addr, a) simde_mm_storeu_si32(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_storeu_si64(void *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && \
+	(SIMDE_DETECT_CLANG_VERSION_CHECK(8, 0, 0) || \
+	 HEDLEY_GCC_VERSION_CHECK(11, 0, 0) || \
+	 HEDLEY_INTEL_VERSION_CHECK(20, 21, 1))
+	_mm_storeu_si64(mem_addr, a);
+#else
+	int64_t val = simde_mm_cvtsi128_si64(a);
+	simde_memcpy(mem_addr, &val, sizeof(val));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_storeu_si64(mem_addr, a) simde_mm_storeu_si64(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_pd(simde_float64 mem_addr[HEDLEY_ARRAY_PARAM(2)],
+			simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_stream_pd(mem_addr, a);
+#else
+	simde_memcpy(mem_addr, &a, sizeof(a));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_pd(mem_addr, a) \
+	simde_mm_stream_pd(HEDLEY_REINTERPRET_CAST(double *, mem_addr), a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_si128(simde__m128i *mem_addr, simde__m128i a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64)
+	_mm_stream_si128(HEDLEY_STATIC_CAST(__m128i *, mem_addr), a);
+#else
+	simde_memcpy(mem_addr, &a, sizeof(a));
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_si128(mem_addr, a) simde_mm_stream_si128(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_si32(int32_t *mem_addr, int32_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_stream_si32(mem_addr, a);
+#else
+	*mem_addr = a;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_si32(mem_addr, a) simde_mm_stream_si32(mem_addr, a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_stream_si64(int64_t *mem_addr, int64_t a)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_ARCH_AMD64) && \
+	!defined(HEDLEY_MSVC_VERSION)
+	_mm_stream_si64(SIMDE_CHECKED_REINTERPRET_CAST(long long int *,
+						 int64_t *, mem_addr),
+			a);
+#else
+	*mem_addr = a;
+#endif
+}
+#define simde_mm_stream_si64x(mem_addr, a) simde_mm_stream_si64(mem_addr, a)
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_stream_si64(mem_addr, a) \
+	simde_mm_stream_si64(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 int64_t *, __int64 *, mem_addr), \
+			 a)
+#define _mm_stream_si64x(mem_addr, a) \
+	simde_mm_stream_si64(SIMDE_CHECKED_REINTERPRET_CAST( \
+				 int64_t *, __int64 *, mem_addr), \
+			 a)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sub_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vsubq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i8 = a_.i8 - b_.i8;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i8) / sizeof(r_.i8[0])); i++) {
+		r_.i8[i] = a_.i8[i] - b_.i8[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_epi8(a, b) simde_mm_sub_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sub_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vsubq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i16 = a_.i16 - b_.i16;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i16) / sizeof(r_.i16[0])); i++) {
+		r_.i16[i] = a_.i16[i] - b_.i16[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_epi16(a, b) simde_mm_sub_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sub_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vsubq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32 = a_.i32 - b_.i32;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32) / sizeof(r_.i32[0])); i++) {
+		r_.i32[i] = a_.i32[i] - b_.i32[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_epi32(a, b) simde_mm_sub_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_sub_epi64(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_epi64(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vsubq_s64(a_.neon_i64, b_.neon_i64);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 - b_.i64;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i64) / sizeof(r_.i64[0])); i++) {
+		r_.i64[i] = a_.i64[i] - b_.i64[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_epi64(a, b) simde_mm_sub_epi64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_sub_epu32(simde__m128i a, simde__m128i b)
+{
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.u32 = a_.u32 - b_.u32;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u32 = vsubq_u32(a_.neon_u32, b_.neon_u32);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.u32) / sizeof(r_.u32[0])); i++) {
+		r_.u32[i] = a_.u32[i] - b_.u32[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_sub_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.f64 = a_.f64 - b_.f64;
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vsubq_f64(a_.neon_f64, b_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_sub(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = a_.f64[i] - b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_pd(a, b) simde_mm_sub_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_sub_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_sub_sd(a, b);
+#elif (SIMDE_NATURAL_VECTOR_SIZE > 0)
+	return simde_mm_move_sd(a, simde_mm_sub_pd(a, b));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+	r_.f64[0] = a_.f64[0] - b_.f64[0];
+	r_.f64[1] = a_.f64[1];
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_sd(a, b) simde_mm_sub_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m64 simde_mm_sub_si64(simde__m64 a, simde__m64 b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
+	return _mm_sub_si64(a, b);
+#else
+	simde__m64_private r_, a_ = simde__m64_to_private(a),
+			 b_ = simde__m64_to_private(b);
+
+#if defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i64 = a_.i64 - b_.i64;
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i64 = vsub_s64(a_.neon_i64, b_.neon_i64);
+#else
+	r_.i64[0] = a_.i64[0] - b_.i64[0];
+#endif
+
+	return simde__m64_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_sub_si64(a, b) simde_mm_sub_si64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_subs_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_subs_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i8 = vqsubq_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i8x16_sub_saturate(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i8[0])); i++) {
+		if (((b_.i8[i]) > 0 && (a_.i8[i]) < INT8_MIN + (b_.i8[i]))) {
+			r_.i8[i] = INT8_MIN;
+		} else if ((b_.i8[i]) < 0 &&
+			 (a_.i8[i]) > INT8_MAX + (b_.i8[i])) {
+			r_.i8[i] = INT8_MAX;
+		} else {
+			r_.i8[i] = (a_.i8[i]) - (b_.i8[i]);
+		}
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_epi8(a, b) simde_mm_subs_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_subs_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_subs_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i16 = vqsubq_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_i16x8_sub_saturate(a_.wasm_v128, b_.wasm_v128);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
+		if (((b_.i16[i]) > 0 &&
+		 (a_.i16[i]) < INT16_MIN + (b_.i16[i]))) {
+			r_.i16[i] = INT16_MIN;
+		} else if ((b_.i16[i]) < 0 &&
+			 (a_.i16[i]) > INT16_MAX + (b_.i16[i])) {
+			r_.i16[i] = INT16_MAX;
+		} else {
+			r_.i16[i] = (a_.i16[i]) - (b_.i16[i]);
+		}
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_epi16(a, b) simde_mm_subs_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_subs_epu8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_subs_epu8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u8 = vqsubq_u8(a_.neon_u8, b_.neon_u8);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u8x16_sub_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u8 = vec_subs(a_.altivec_u8, b_.altivec_u8);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i8[0])); i++) {
+		const int32_t x = a_.u8[i] - b_.u8[i];
+		if (x < 0) {
+			r_.u8[i] = 0;
+		} else if (x > UINT8_MAX) {
+			r_.u8[i] = UINT8_MAX;
+		} else {
+			r_.u8[i] = HEDLEY_STATIC_CAST(uint8_t, x);
+		}
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_epu8(a, b) simde_mm_subs_epu8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_subs_epu16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_subs_epu16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_u16 = vqsubq_u16(a_.neon_u16, b_.neon_u16);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_u16x8_sub_saturate(a_.wasm_v128, b_.wasm_v128);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_u16 = vec_subs(a_.altivec_u16, b_.altivec_u16);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_) / sizeof(r_.i16[0])); i++) {
+		const int32_t x = a_.u16[i] - b_.u16[i];
+		if (x < 0) {
+			r_.u16[i] = 0;
+		} else if (x > UINT16_MAX) {
+			r_.u16[i] = UINT16_MAX;
+		} else {
+			r_.u16[i] = HEDLEY_STATIC_CAST(uint16_t, x);
+		}
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_subs_epu16(a, b) simde_mm_subs_epu16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomieq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomieq_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_or_b_nan = vreinterpretq_u64_u32(vmvnq_u32(
+		vreinterpretq_u32_u64(vandq_u64(a_not_nan, b_not_nan))));
+	uint64x2_t a_eq_b = vceqq_f64(a_.neon_f64, b_.neon_f64);
+	r = !!(vgetq_lane_u64(vorrq_u64(a_or_b_nan, a_eq_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) ==
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] == b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] == b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomieq_sd(a, b) simde_mm_ucomieq_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomige_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomige_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_and_b_not_nan = vandq_u64(a_not_nan, b_not_nan);
+	uint64x2_t a_ge_b = vcgeq_f64(a_.neon_f64, b_.neon_f64);
+	r = !!(vgetq_lane_u64(vandq_u64(a_and_b_not_nan, a_ge_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) >=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] >= b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] >= b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomige_sd(a, b) simde_mm_ucomige_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomigt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomigt_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_and_b_not_nan = vandq_u64(a_not_nan, b_not_nan);
+	uint64x2_t a_gt_b = vcgtq_f64(a_.neon_f64, b_.neon_f64);
+	r = !!(vgetq_lane_u64(vandq_u64(a_and_b_not_nan, a_gt_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) >
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] > b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] > b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomigt_sd(a, b) simde_mm_ucomigt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomile_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomile_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_or_b_nan = vreinterpretq_u64_u32(vmvnq_u32(
+		vreinterpretq_u32_u64(vandq_u64(a_not_nan, b_not_nan))));
+	uint64x2_t a_le_b = vcleq_f64(a_.neon_f64, b_.neon_f64);
+	r = !!(vgetq_lane_u64(vorrq_u64(a_or_b_nan, a_le_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) <=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] <= b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] <= b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomile_sd(a, b) simde_mm_ucomile_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomilt_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomilt_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_or_b_nan = vreinterpretq_u64_u32(vmvnq_u32(
+		vreinterpretq_u32_u64(vandq_u64(a_not_nan, b_not_nan))));
+	uint64x2_t a_lt_b = vcltq_f64(a_.neon_f64, b_.neon_f64);
+	r = !!(vgetq_lane_u64(vorrq_u64(a_or_b_nan, a_lt_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) <
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] < b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] < b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomilt_sd(a, b) simde_mm_ucomilt_sd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+int simde_mm_ucomineq_sd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_ucomineq_sd(a, b);
+#else
+	simde__m128d_private a_ = simde__m128d_to_private(a),
+			 b_ = simde__m128d_to_private(b);
+	int r;
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	uint64x2_t a_not_nan = vceqq_f64(a_.neon_f64, a_.neon_f64);
+	uint64x2_t b_not_nan = vceqq_f64(b_.neon_f64, b_.neon_f64);
+	uint64x2_t a_and_b_not_nan = vandq_u64(a_not_nan, b_not_nan);
+	uint64x2_t a_neq_b = vreinterpretq_u64_u32(vmvnq_u32(
+		vreinterpretq_u32_u64(vceqq_f64(a_.neon_f64, b_.neon_f64))));
+	r = !!(vgetq_lane_u64(vandq_u64(a_and_b_not_nan, a_neq_b), 0) != 0);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	return wasm_f64x2_extract_lane(a_.wasm_v128, 0) !=
+	 wasm_f64x2_extract_lane(b_.wasm_v128, 0);
+#elif defined(SIMDE_HAVE_FENV_H)
+	fenv_t envp;
+	int x = feholdexcept(&envp);
+	r = a_.f64[0] != b_.f64[0];
+	if (HEDLEY_LIKELY(x == 0))
+		fesetenv(&envp);
+#else
+	r = a_.f64[0] != b_.f64[0];
+#endif
+
+	return r;
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_ucomineq_sd(a, b) simde_mm_ucomineq_sd(a, b)
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_PUSH
+SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_
+#endif
+
+#if defined(SIMDE_DIAGNOSTIC_DISABLE_UNINITIALIZED_)
+HEDLEY_DIAGNOSTIC_POP
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_lfence(void)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_lfence();
+#else
+	simde_mm_sfence();
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_lfence() simde_mm_lfence()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+void simde_mm_mfence(void)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	_mm_mfence();
+#else
+	simde_mm_sfence();
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_mfence() simde_mm_mfence()
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpackhi_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpackhi_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i8 = vzip2q_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int8x8_t a1 = vreinterpret_s8_s16(vget_high_s16(a_.neon_i16));
+	int8x8_t b1 = vreinterpret_s8_s16(vget_high_s16(b_.neon_i16));
+	int8x8x2_t result = vzip_s8(a1, b1);
+	r_.neon_i8 = vcombine_s8(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 16, a_.i8, b_.i8, 8, 24, 9, 25, 10, 26,
+				 11, 27, 12, 28, 13, 29, 14, 30, 15, 31);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i8[0])) / 2); i++) {
+		r_.i8[(i * 2)] =
+			a_.i8[i + ((sizeof(r_) / sizeof(r_.i8[0])) / 2)];
+		r_.i8[(i * 2) + 1] =
+			b_.i8[i + ((sizeof(r_) / sizeof(r_.i8[0])) / 2)];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_epi8(a, b) simde_mm_unpackhi_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpackhi_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpackhi_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i16 = vzip2q_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int16x4_t a1 = vget_high_s16(a_.neon_i16);
+	int16x4_t b1 = vget_high_s16(b_.neon_i16);
+	int16x4x2_t result = vzip_s16(a1, b1);
+	r_.neon_i16 = vcombine_s16(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 4, 12, 5, 13, 6,
+				 14, 7, 15);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i16[0])) / 2); i++) {
+		r_.i16[(i * 2)] =
+			a_.i16[i + ((sizeof(r_) / sizeof(r_.i16[0])) / 2)];
+		r_.i16[(i * 2) + 1] =
+			b_.i16[i + ((sizeof(r_) / sizeof(r_.i16[0])) / 2)];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_epi16(a, b) simde_mm_unpackhi_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpackhi_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpackhi_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i32 = vzip2q_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int32x2_t a1 = vget_high_s32(a_.neon_i32);
+	int32x2_t b1 = vget_high_s32(b_.neon_i32);
+	int32x2x2_t result = vzip_s32(a1, b1);
+	r_.neon_i32 = vcombine_s32(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 2, 6, 3, 7);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i32[0])) / 2); i++) {
+		r_.i32[(i * 2)] =
+			a_.i32[i + ((sizeof(r_) / sizeof(r_.i32[0])) / 2)];
+		r_.i32[(i * 2) + 1] =
+			b_.i32[i + ((sizeof(r_) / sizeof(r_.i32[0])) / 2)];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_epi32(a, b) simde_mm_unpackhi_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpackhi_epi64(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpackhi_epi64(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int64x1_t a_h = vget_high_s64(a_.neon_i64);
+	int64x1_t b_h = vget_high_s64(b_.neon_i64);
+	r_.neon_i64 = vcombine_s64(a_h, b_h);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.i64, b_.i64, 1, 3);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i64[0])) / 2); i++) {
+		r_.i64[(i * 2)] =
+			a_.i64[i + ((sizeof(r_) / sizeof(r_.i64[0])) / 2)];
+		r_.i64[(i * 2) + 1] =
+			b_.i64[i + ((sizeof(r_) / sizeof(r_.i64[0])) / 2)];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_epi64(a, b) simde_mm_unpackhi_epi64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_unpackhi_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpackhi_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x1_t a_l = vget_high_f64(a_.f64);
+	float64x1_t b_l = vget_high_f64(b_.f64);
+	r_.neon_f64 = vcombine_f64(a_l, b_l);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v64x2_shuffle(a_.wasm_v128, b_.wasm_v128, 1, 3);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 1, 3);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.f64[0])) / 2); i++) {
+		r_.f64[(i * 2)] =
+			a_.f64[i + ((sizeof(r_) / sizeof(r_.f64[0])) / 2)];
+		r_.f64[(i * 2) + 1] =
+			b_.f64[i + ((sizeof(r_) / sizeof(r_.f64[0])) / 2)];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpackhi_pd(a, b) simde_mm_unpackhi_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpacklo_epi8(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpacklo_epi8(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i8 = vzip1q_s8(a_.neon_i8, b_.neon_i8);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int8x8_t a1 = vreinterpret_s8_s16(vget_low_s16(a_.neon_i16));
+	int8x8_t b1 = vreinterpret_s8_s16(vget_low_s16(b_.neon_i16));
+	int8x8x2_t result = vzip_s8(a1, b1);
+	r_.neon_i8 = vcombine_s8(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i8 = SIMDE_SHUFFLE_VECTOR_(8, 16, a_.i8, b_.i8, 0, 16, 1, 17, 2, 18,
+				 3, 19, 4, 20, 5, 21, 6, 22, 7, 23);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i8[0])) / 2); i++) {
+		r_.i8[(i * 2)] = a_.i8[i];
+		r_.i8[(i * 2) + 1] = b_.i8[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_epi8(a, b) simde_mm_unpacklo_epi8(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpacklo_epi16(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpacklo_epi16(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i16 = vzip1q_s16(a_.neon_i16, b_.neon_i16);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int16x4_t a1 = vget_low_s16(a_.neon_i16);
+	int16x4_t b1 = vget_low_s16(b_.neon_i16);
+	int16x4x2_t result = vzip_s16(a1, b1);
+	r_.neon_i16 = vcombine_s16(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i16 = SIMDE_SHUFFLE_VECTOR_(16, 16, a_.i16, b_.i16, 0, 8, 1, 9, 2,
+				 10, 3, 11);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i16[0])) / 2); i++) {
+		r_.i16[(i * 2)] = a_.i16[i];
+		r_.i16[(i * 2) + 1] = b_.i16[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_epi16(a, b) simde_mm_unpacklo_epi16(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpacklo_epi32(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpacklo_epi32(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_i32 = vzip1q_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int32x2_t a1 = vget_low_s32(a_.neon_i32);
+	int32x2_t b1 = vget_low_s32(b_.neon_i32);
+	int32x2x2_t result = vzip_s32(a1, b1);
+	r_.neon_i32 = vcombine_s32(result.val[0], result.val[1]);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i32 = SIMDE_SHUFFLE_VECTOR_(32, 16, a_.i32, b_.i32, 0, 4, 1, 5);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i32[0])) / 2); i++) {
+		r_.i32[(i * 2)] = a_.i32[i];
+		r_.i32[(i * 2) + 1] = b_.i32[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_epi32(a, b) simde_mm_unpacklo_epi32(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_unpacklo_epi64(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpacklo_epi64(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	int64x1_t a_l = vget_low_s64(a_.i64);
+	int64x1_t b_l = vget_low_s64(b_.i64);
+	r_.neon_i64 = vcombine_s64(a_l, b_l);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.i64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.i64, b_.i64, 0, 2);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.i64[0])) / 2); i++) {
+		r_.i64[(i * 2)] = a_.i64[i];
+		r_.i64[(i * 2) + 1] = b_.i64[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_epi64(a, b) simde_mm_unpacklo_epi64(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_mm_unpacklo_pd(simde__m128d a, simde__m128d b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_unpacklo_pd(a, b);
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a),
+				 b_ = simde__m128d_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	float64x1_t a_l = vget_low_f64(a_.f64);
+	float64x1_t b_l = vget_low_f64(b_.f64);
+	r_.neon_f64 = vcombine_f64(a_l, b_l);
+#elif defined(SIMDE_SHUFFLE_VECTOR_)
+	r_.f64 = SIMDE_SHUFFLE_VECTOR_(64, 16, a_.f64, b_.f64, 0, 2);
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < ((sizeof(r_) / sizeof(r_.f64[0])) / 2); i++) {
+		r_.f64[(i * 2)] = a_.f64[i];
+		r_.f64[(i * 2) + 1] = b_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_unpacklo_pd(a, b) simde_mm_unpacklo_pd(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128d simde_x_mm_negate_pd(simde__m128d a)
+{
+#if defined(SIMDE_X86_SSE_NATIVE)
+	return simde_mm_xor_pd(a, _mm_set1_pd(SIMDE_FLOAT64_C(-0.0)));
+#else
+	simde__m128d_private r_, a_ = simde__m128d_to_private(a);
+
+#if defined(SIMDE_POWER_ALTIVEC_P8_NATIVE) && \
+	(!defined(HEDLEY_GCC_VERSION) || HEDLEY_GCC_VERSION_CHECK(8, 1, 0))
+	r_.altivec_f64 = vec_neg(a_.altivec_f64);
+#elif defined(SIMDE_ARM_NEON_A64V8_NATIVE)
+	r_.neon_f64 = vnegq_f64(a_.neon_f64);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_f64x2_neg(a_.wasm_v128);
+#elif defined(SIMDE_VECTOR_NEGATE)
+	r_.f64 = -a_.f64;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.f64) / sizeof(r_.f64[0])); i++) {
+		r_.f64[i] = -a_.f64[i];
+	}
+#endif
+
+	return simde__m128d_from_private(r_);
+#endif
+}
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_mm_xor_si128(simde__m128i a, simde__m128i b)
+{
+#if defined(SIMDE_X86_SSE2_NATIVE)
+	return _mm_xor_si128(a, b);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a),
+				 b_ = simde__m128i_to_private(b);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = veorq_s32(a_.neon_i32, b_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_xor(a_.altivec_i32, b_.altivec_i32);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = a_.i32f ^ b_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = a_.i32f[i] ^ b_.i32f[i];
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _mm_xor_si128(a, b) simde_mm_xor_si128(a, b)
+#endif
+
+SIMDE_FUNCTION_ATTRIBUTES
+simde__m128i simde_x_mm_not_si128(simde__m128i a)
+{
+#if defined(SIMDE_X86_AVX512VL_NATIVE)
+	return _mm_ternarylogic_epi32(a, a, a, 0x55);
+#else
+	simde__m128i_private r_, a_ = simde__m128i_to_private(a);
+
+#if defined(SIMDE_ARM_NEON_A32V7_NATIVE)
+	r_.neon_i32 = vmvnq_s32(a_.neon_i32);
+#elif defined(SIMDE_POWER_ALTIVEC_P6_NATIVE)
+	r_.altivec_i32 = vec_nor(a_.altivec_i32, a_.altivec_i32);
+#elif defined(SIMDE_WASM_SIMD128_NATIVE)
+	r_.wasm_v128 = wasm_v128_not(a_.wasm_v128);
+#elif defined(SIMDE_VECTOR_SUBSCRIPT_OPS)
+	r_.i32f = ~a_.i32f;
+#else
+	SIMDE_VECTORIZE
+	for (size_t i = 0; i < (sizeof(r_.i32f) / sizeof(r_.i32f[0])); i++) {
+		r_.i32f[i] = ~(a_.i32f[i]);
+	}
+#endif
+
+	return simde__m128i_from_private(r_);
+#endif
+}
+
+#define SIMDE_MM_SHUFFLE2(x, y) (((x) << 1) | (y))
+#if defined(SIMDE_X86_SSE2_ENABLE_NATIVE_ALIASES)
+#define _MM_SHUFFLE2(x, y) SIMDE_MM_SHUFFLE2(x, y)
+#endif
+
+SIMDE_END_DECLS_
+
+HEDLEY_DIAGNOSTIC_POP
+
+#endif /* !defined(SIMDE_X86_SSE2_H) */

obs-studio-26.1.0.tar.xz/libobs/util/sse-intrin.h -> obs-studio-26.1.1.tar.xz/libobs/util/sse-intrin.h Changed

@@ -17,55 +17,9 @@
 
 #pragma once
 
-#if NEEDS_SIMDE
-
-#include "simde/sse2.h"
-
-#define __m128 simde__m128
-#define _mm_setzero_ps simde_mm_setzero_ps
-#define _mm_set_ps simde_mm_set_ps
-#define _mm_add_ps simde_mm_add_ps
-#define _mm_sub_ps simde_mm_sub_ps
-#define _mm_mul_ps simde_mm_mul_ps
-#define _mm_div_ps simde_mm_div_ps
-#define _mm_set1_ps simde_mm_set1_ps
-#define _mm_movehl_ps simde_mm_movehl_ps
-#define _mm_shuffle_ps simde_mm_shuffle_ps
-#define _mm_min_ps simde_mm_min_ps
-#define _mm_max_ps simde_mm_max_ps
-#define _mm_movelh_ps simde_mm_movelh_ps
-#define _mm_unpacklo_ps simde_mm_unpacklo_ps
-#define _mm_unpackhi_ps simde_mm_unpackhi_ps
-#define _mm_load_ps simde_mm_load_ps
-#define _mm_andnot_ps simde_mm_andnot_ps
-#define _mm_storeu_ps simde_mm_storeu_ps
-#define _mm_loadu_ps simde_mm_loadu_ps
-
-#define __m128i simde__m128i
-#define _mm_set1_epi32 simde_mm_set1_epi32
-#define _mm_set1_epi16 simde_mm_set1_epi16
-#define _mm_load_si128 simde_mm_load_si128
-#define _mm_packs_epi32 simde_mm_packs_epi32
-#define _mm_srli_si128 simde_mm_srli_si128
-#define _mm_and_si128 simde_mm_and_si128
-#define _mm_packus_epi16 simde_mm_packus_epi16
-#define _mm_add_epi64 simde_mm_add_epi64
-#define _mm_shuffle_epi32 simde_mm_shuffle_epi32
-#define _mm_srai_epi16 simde_mm_srai_epi16
-#define _mm_shufflelo_epi16 simde_mm_shufflelo_epi16
-#define _mm_storeu_si128 simde_mm_storeu_si128
-
-#define _MM_SHUFFLE SIMDE_MM_SHUFFLE
-#define _MM_TRANSPOSE4_PS SIMDE_MM_TRANSPOSE4_PS
-
-#else
-
-#if defined(__aarch64__) || defined(__arm__)
-#include <arm_neon.h>
-#include "sse2neon.h"
-#else
-#include <xmmintrin.h>
+#if defined(_MSC_VER)
 #include <emmintrin.h>
-#endif
-
+#else
+#define SIMDE_ENABLE_NATIVE_ALIASES
+#include "simde/x86/sse2.h"
 #endif

obs-studio-26.1.0.tar.xz/plugins/coreaudio-encoder/CMakeLists.txt -> obs-studio-26.1.1.tar.xz/plugins/coreaudio-encoder/CMakeLists.txt Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/DecklinkInput.cpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/DecklinkInput.cpp Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/DecklinkInput.hpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/DecklinkInput.hpp Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/OBSVideoFrame.cpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/OBSVideoFrame.cpp Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/OBSVideoFrame.h -> obs-studio-26.1.1.tar.xz/plugins/decklink/OBSVideoFrame.h Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/const.h -> obs-studio-26.1.1.tar.xz/plugins/decklink/const.h Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/data/locale/en-US.ini -> obs-studio-26.1.1.tar.xz/plugins/decklink/data/locale/en-US.ini Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/decklink-device-instance.cpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/decklink-device-instance.cpp Changed

@@ -24,10 +24,10 @@
 		return VIDEO_FORMAT_BGRX;
 
 	default:
-	case bmdFormat8BitYUV:;
+	case bmdFormat8BitYUV:
+	case bmdFormat10BitYUV:;
+		return VIDEO_FORMAT_UYVY;
 	}
-
-	return VIDEO_FORMAT_UYVY;
 }
 
 static inline int ConvertChannelFormat(speaker_layout format)
@@ -168,21 +168,28 @@
 		packets->Release();
 	}
 
-	IDeckLinkVideoConversion *frameConverter =
-		CreateVideoConversionInstance();
+	IDeckLinkVideoFrame *frame;
+	if (videoFrame->GetPixelFormat() != convertFrame->GetPixelFormat()) {
+		IDeckLinkVideoConversion *frameConverter =
+			CreateVideoConversionInstance();
+
+		frameConverter->ConvertFrame(videoFrame, convertFrame);
 
-	frameConverter->ConvertFrame(videoFrame, convertFrame);
+		frame = convertFrame;
+	} else {
+		frame = videoFrame;
+	}
 
 	void *bytes;
-	if (convertFrame->GetBytes(&bytes) != S_OK) {
+	if (frame->GetBytes(&bytes) != S_OK) {
 		LOG(LOG_WARNING, "Failed to get video frame data");
 		return;
 	}
 
 	currentFrame.data[0] = (uint8_t *)bytes;
-	currentFrame.linesize[0] = (uint32_t)convertFrame->GetRowBytes();
-	currentFrame.width = (uint32_t)convertFrame->GetWidth();
-	currentFrame.height = (uint32_t)convertFrame->GetHeight();
+	currentFrame.linesize[0] = (uint32_t)frame->GetRowBytes();
+	currentFrame.width = (uint32_t)frame->GetWidth();
+	currentFrame.height = (uint32_t)frame->GetHeight();
 	currentFrame.timestamp = timestamp;
 
 	obs_source_output_video2(
@@ -326,10 +333,22 @@
 				    currentFrame.color_range_min,
 				    currentFrame.color_range_max);
 
-	if (convertFrame) {
-		delete convertFrame;
+	delete convertFrame;
+
+	BMDPixelFormat convertFormat;
+	switch (pixelFormat) {
+	case bmdFormat8BitBGRA:
+		convertFormat = bmdFormat8BitBGRA;
+		break;
+	default:
+	case bmdFormat10BitYUV:
+	case bmdFormat8BitYUV:;
+		convertFormat = bmdFormat8BitYUV;
+		break;
 	}
-	convertFrame = new OBSVideoFrame(mode_->GetWidth(), mode_->GetHeight());
+
+	convertFrame = new OBSVideoFrame(mode_->GetWidth(), mode_->GetHeight(),
+					 convertFormat);
 
 #ifdef LOG_SETUP_VIDEO_FORMAT
 	LOG(LOG_INFO, "Setup video format: %s, %s, %s",
@@ -340,6 +359,7 @@
 }
 
 bool DeckLinkDeviceInstance::StartCapture(DeckLinkDeviceMode *mode_,
+					  bool allow10Bit_,
 					  BMDVideoConnection bmdVideoConnection,
 					  BMDAudioConnection bmdAudioConnection)
 {
@@ -392,7 +412,11 @@
 	bool isauto = mode_->GetName() == "Auto";
 	if (isauto) {
 		displayMode = bmdModeNTSC;
-		pixelFormat = bmdFormat10BitYUV;
+		if (allow10Bit) {
+			pixelFormat = bmdFormat10BitYUV;
+		} else {
+			pixelFormat = bmdFormat8BitYUV;
+		}
 		flags = bmdVideoInputEnableFormatDetection;
 	} else {
 		displayMode = mode_->GetDisplayMode();
@@ -401,6 +425,8 @@
 		flags = bmdVideoInputFlagDefault;
 	}
 
+	allow10Bit = allow10Bit_;
+
 	const HRESULT videoResult =
 		input->EnableVideoInput(displayMode, pixelFormat, flags);
 	if (videoResult != S_OK) {
@@ -631,15 +657,22 @@
 {
 
 	if (events & bmdVideoInputColorspaceChanged) {
-		switch (detectedSignalFlags) {
-		case bmdDetectedVideoInputRGB444:
+		if (detectedSignalFlags & bmdDetectedVideoInputRGB444) {
 			pixelFormat = bmdFormat8BitBGRA;
-			break;
-
-		default:
-		case bmdDetectedVideoInputYCbCr422:
-			pixelFormat = bmdFormat10BitYUV;
-			break;
+		}
+		if (detectedSignalFlags & bmdDetectedVideoInputYCbCr422) {
+			if (detectedSignalFlags &
+			    bmdDetectedVideoInput10BitDepth) {
+				if (allow10Bit) {
+					pixelFormat = bmdFormat10BitYUV;
+				} else {
+					pixelFormat = bmdFormat8BitYUV;
+				}
+			}
+			if (detectedSignalFlags &
+			    bmdDetectedVideoInput8BitDepth) {
+				pixelFormat = bmdFormat8BitYUV;
+			}
 		}
 	}

obs-studio-26.1.0.tar.xz/plugins/decklink/decklink-device-instance.hpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/decklink-device-instance.hpp Changed

obs-studio-26.1.0.tar.xz/plugins/decklink/decklink-source.cpp -> obs-studio-26.1.1.tar.xz/plugins/decklink/decklink-source.cpp Changed

@@ -80,6 +80,7 @@
 	decklink->SetChannelFormat(channelFormat);
 	decklink->hash = std::string(hash);
 	decklink->swap = obs_data_get_bool(settings, SWAP);
+	decklink->allow10Bit = obs_data_get_bool(settings, ALLOW_10_BIT);
 	decklink->Activate(device, id, videoConnection, audioConnection);
 }
 
@@ -247,6 +248,9 @@
 	list = obs_properties_get(props, PIXEL_FORMAT);
 	obs_property_set_visible(list, id != MODE_ID_AUTO);
 
+	auto allow10BitProp = obs_properties_get(props, ALLOW_10_BIT);
+	obs_property_set_visible(allow10BitProp, id == MODE_ID_AUTO);
+
 	return true;
 }
 
@@ -277,6 +281,7 @@
 				       OBS_COMBO_FORMAT_INT);
 
 	obs_property_list_add_int(list, "8-bit YUV", bmdFormat8BitYUV);
+	obs_property_list_add_int(list, "10-bit YUV", bmdFormat10BitYUV);
 	obs_property_list_add_int(list, "8-bit BGRA", bmdFormat8BitBGRA);
 
 	list = obs_properties_add_list(props, COLOR_SPACE, TEXT_COLOR_SPACE,
@@ -322,6 +327,8 @@
 
 	obs_properties_add_bool(props, DEACTIVATE_WNS, TEXT_DWNS);
 
+	obs_properties_add_bool(props, ALLOW_10_BIT, TEXT_ALLOW_10_BIT);
+
 	UNUSED_PARAMETER(data);
 	return props;
 }

obs-studio-26.1.0.tar.xz/plugins/image-source/image-source.c -> obs-studio-26.1.1.tar.xz/plugins/image-source/image-source.c Changed

obs-studio-26.1.0.tar.xz/plugins/image-source/obs-slideshow.c -> obs-studio-26.1.1.tar.xz/plugins/image-source/obs-slideshow.c Changed

obs-studio-26.1.0.tar.xz/plugins/linux-jack/jack-wrapper.c -> obs-studio-26.1.1.tar.xz/plugins/linux-jack/jack-wrapper.c Changed

@@ -61,11 +61,15 @@
 int jack_process_callback(jack_nframes_t nframes, void *arg)
 {
 	struct jack_data *data = (struct jack_data *)arg;
+	jack_nframes_t current_frames;
+	jack_time_t current_usecs, next_usecs;
+	float period_usecs;
+
+	uint64_t now = os_gettime_ns();
+
 	if (data == 0)
 		return 0;
 
-	pthread_mutex_lock(&data->jack_mutex);
-
 	struct obs_source_audio out;
 	out.speakers = jack_channels_to_obs_speakers(data->channels);
 	out.samples_per_sec = jack_get_sample_rate(data->jack_client);
@@ -80,11 +84,19 @@
 	}
 
 	out.frames = nframes;
-	out.timestamp = os_gettime_ns() -
-			jack_frames_to_time(data->jack_client, nframes);
+	if (!jack_get_cycle_times(data->jack_client, &current_frames,
+				 &current_usecs, &next_usecs, &period_usecs)) {
+		out.timestamp = now - (int64_t)(period_usecs * 1000);
+	} else {
+		out.timestamp = now - util_mul_div64(nframes, 1000000000ULL,
+						 data->samples_per_sec);
+		blog(LOG_WARNING,
+		 "jack_get_cycle_times error: guessing timestamp");
+	}
 
+	/* FIXME: this function is not realtime-safe, we should do something
+	 * about this */
 	obs_source_output_audio(data->source, &out);
-	pthread_mutex_unlock(&data->jack_mutex);
 	return 0;
 }
 
@@ -115,7 +127,7 @@
 
 		data->jack_ports[i] = jack_port_register(
 			data->jack_client, port_name, JACK_DEFAULT_AUDIO_TYPE,
-			JackPortIsInput, 0);
+			JackPortIsInput | JackPortIsTerminal, 0);
 		if (data->jack_ports[i] == NULL) {
 			blog(LOG_ERROR,
 			 "jack_port_register Error:"
@@ -151,17 +163,11 @@
 	pthread_mutex_lock(&data->jack_mutex);
 
 	if (data->jack_client) {
+		jack_client_close(data->jack_client);
 		if (data->jack_ports != NULL) {
-			for (int i = 0; i < data->channels; ++i) {
-				if (data->jack_ports[i] != NULL)
-					jack_port_unregister(
-						data->jack_client,
-						data->jack_ports[i]);
-			}
 			bfree(data->jack_ports);
 			data->jack_ports = NULL;
 		}
-		jack_client_close(data->jack_client);
 		data->jack_client = NULL;
 	}
 	pthread_mutex_unlock(&data->jack_mutex);

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/CMSampleBufferUtils.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/CMSampleBufferUtils.mm Changed

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/Logging.h -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/Logging.h Changed

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALDevice.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALDevice.mm Changed

@@ -86,9 +86,7 @@
 	case kCMIODevicePropertyDeviceMaster:
 		return sizeof(pid_t);
 	default:
-		DLog(@"Device unhandled getPropertyDataSizeWithAddress for %@",
-		 [OBSDALObjectStore
-			 StringFromPropertySelector:address.mSelector]);
+		break;
 	};
 
 	return 0;
@@ -191,10 +189,6 @@
 		*dataUsed = sizeof(pid_t);
 		break;
 	default:
-		DLog(@"Device unhandled getPropertyDataWithAddress for %@",
-		 [OBSDALObjectStore
-			 StringFromPropertySelector:address.mSelector]);
-		*dataUsed = 0;
 		break;
 	};
 }
@@ -227,9 +221,6 @@
 	case kCMIODevicePropertyLinkedCoreAudioDeviceUID:
 		return false;
 	default:
-		DLog(@"Device unhandled hasPropertyWithAddress for %@",
-		 [OBSDALObjectStore
-			 StringFromPropertySelector:address.mSelector]);
 		return false;
 	};
 }
@@ -262,9 +253,6 @@
 	case kCMIODevicePropertyDeviceMaster:
 		return true;
 	default:
-		DLog(@"Device unhandled isPropertySettableWithAddress for %@",
-		 [OBSDALObjectStore
-			 StringFromPropertySelector:address.mSelector]);
 		return false;
 	};
 }
@@ -285,9 +273,6 @@
 		self.masterPid = *static_cast<const pid_t *>(data);
 		break;
 	default:
-		DLog(@"Device unhandled setPropertyDataWithAddress for %@",
-		 [OBSDALObjectStore
-			 StringFromPropertySelector:address.mSelector]);
 		break;
 	};
 }

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALPlugInInterface.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALPlugInInterface.mm Changed

@@ -34,6 +34,8 @@
 
 ULONG HardwarePlugIn_AddRef(CMIOHardwarePlugInRef self)
 {
+	UNUSED_PARAMETER(self);
+
 	sRefCount += 1;
 	DLogFunc(@"sRefCount now = %d", sRefCount);
 	return sRefCount;
@@ -41,6 +43,8 @@
 
 ULONG HardwarePlugIn_Release(CMIOHardwarePlugInRef self)
 {
+	UNUSED_PARAMETER(self);
+
 	sRefCount -= 1;
 	DLogFunc(@"sRefCount now = %d", sRefCount);
 	return sRefCount;
@@ -49,6 +53,7 @@
 HRESULT HardwarePlugIn_QueryInterface(CMIOHardwarePlugInRef self, REFIID uuid,
 				 LPVOID *interface)
 {
+	UNUSED_PARAMETER(self);
 	DLogFunc(@"");
 
 	if (!interface) {
@@ -161,6 +166,7 @@
 void HardwarePlugIn_ObjectShow(CMIOHardwarePlugInRef self,
 			 CMIOObjectID objectID)
 {
+	UNUSED_PARAMETER(objectID);
 	DLogFunc(@"self=%p", self);
 }
 
@@ -169,6 +175,7 @@
 				 CMIOObjectID objectID,
 				 const CMIOObjectPropertyAddress *address)
 {
+	UNUSED_PARAMETER(self);
 
 	NSObject<CMIOObject> *object =
 		[OBSDALObjectStore GetObjectWithId:objectID];
@@ -215,6 +222,7 @@
 	const CMIOObjectPropertyAddress *address, UInt32 qualifierDataSize,
 	const void *qualifierData, UInt32 *dataSize)
 {
+	UNUSED_PARAMETER(self);
 
 	NSObject<CMIOObject> *object =
 		[OBSDALObjectStore GetObjectWithId:objectID];
@@ -240,6 +248,7 @@
 	const void *qualifierData, UInt32 dataSize, UInt32 *dataUsed,
 	void *data)
 {
+	UNUSED_PARAMETER(self);
 
 	NSObject<CMIOObject> *object =
 		[OBSDALObjectStore GetObjectWithId:objectID];
@@ -345,6 +354,8 @@
 OSStatus HardwarePlugIn_DeviceSuspend(CMIOHardwarePlugInRef self,
 				 CMIODeviceID deviceID)
 {
+	UNUSED_PARAMETER(deviceID);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareNoError;
 }
@@ -352,6 +363,8 @@
 OSStatus HardwarePlugIn_DeviceResume(CMIOHardwarePlugInRef self,
 				 CMIODeviceID deviceID)
 {
+	UNUSED_PARAMETER(deviceID);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareNoError;
 }
@@ -380,6 +393,9 @@
 				 CMIODeviceID deviceID,
 				 CMIODeviceAVCCommand *ioAVCCommand)
 {
+	UNUSED_PARAMETER(deviceID);
+	UNUSED_PARAMETER(ioAVCCommand);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareNoError;
 }
@@ -389,6 +405,9 @@
 					 CMIODeviceID deviceID,
 					 CMIODeviceRS422Command *ioRS422Command)
 {
+	UNUSED_PARAMETER(deviceID);
+	UNUSED_PARAMETER(ioRS422Command);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareNoError;
 }
@@ -396,6 +415,8 @@
 OSStatus HardwarePlugIn_StreamDeckPlay(CMIOHardwarePlugInRef self,
 				 CMIOStreamID streamID)
 {
+	UNUSED_PARAMETER(streamID);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareIllegalOperationError;
 }
@@ -403,6 +424,8 @@
 OSStatus HardwarePlugIn_StreamDeckStop(CMIOHardwarePlugInRef self,
 				 CMIOStreamID streamID)
 {
+	UNUSED_PARAMETER(streamID);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareIllegalOperationError;
 }
@@ -410,6 +433,9 @@
 OSStatus HardwarePlugIn_StreamDeckJog(CMIOHardwarePlugInRef self,
 				 CMIOStreamID streamID, SInt32 speed)
 {
+	UNUSED_PARAMETER(streamID);
+	UNUSED_PARAMETER(speed);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareIllegalOperationError;
 }
@@ -419,6 +445,10 @@
 					Float64 requestedTimecode,
 					Boolean playOnCue)
 {
+	UNUSED_PARAMETER(streamID);
+	UNUSED_PARAMETER(requestedTimecode);
+	UNUSED_PARAMETER(playOnCue);
+
 	DLogFunc(@"self=%p", self);
 	return kCMIOHardwareIllegalOperationError;
 }

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALPluginMain.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALPluginMain.mm Changed

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALStream.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/dal-plugin/OBSDALStream.mm Changed

@@ -299,7 +299,6 @@
 - (void)fillFrame
 {
 	if (CMSimpleQueueGetFullness(self.queue) >= 1.0) {
-		DLog(@"Queue is full, bailing out");
 		return;
 	}
 
@@ -433,9 +432,6 @@
 	case kCMIOStreamPropertyClock:
 		return sizeof(CFTypeRef);
 	default:
-		DLog(@"Stream unhandled getPropertyDataSizeWithAddress for %@",
-		     [OBSDALObjectStore
-			     StringFromPropertySelector:address.mSelector]);
 		return 0;
 	};
 }
@@ -509,9 +505,6 @@
 		*dataUsed = sizeof(CFTypeRef);
 		break;
 	default:
-		DLog(@"Stream unhandled getPropertyDataWithAddress for %@",
-		     [OBSDALObjectStore
-			     StringFromPropertySelector:address.mSelector]);
 		*dataUsed = 0;
 	};
 }
@@ -543,17 +536,12 @@
 			     StringFromPropertySelector:address.mSelector]);
 		return false;
 	default:
-		DLog(@"Stream unhandled hasPropertyWithAddress for %@",
-		     [OBSDALObjectStore
-			     StringFromPropertySelector:address.mSelector]);
 		return false;
 	};
 }
 
 - (BOOL)isPropertySettableWithAddress:(CMIOObjectPropertyAddress)address
 {
-	DLog(@"Stream unhandled isPropertySettableWithAddress for %@",
-	     [OBSDALObjectStore StringFromPropertySelector:address.mSelector]);
 	return false;
 }
 
@@ -563,8 +551,6 @@
 			  dataSize:(UInt32)dataSize
 			      data:(nonnull const void *)data
 {
-	DLog(@"Stream unhandled setPropertyDataWithAddress for %@",
-	     [OBSDALObjectStore StringFromPropertySelector:address.mSelector]);
 }
 
 @end

obs-studio-26.1.0.tar.xz/plugins/mac-virtualcam/src/obs-plugin/plugin-main.mm -> obs-studio-26.1.1.tar.xz/plugins/mac-virtualcam/src/obs-plugin/plugin-main.mm Changed

@@ -117,6 +117,8 @@
 static void *virtualcam_output_create(obs_data_t *settings,
 				      obs_output_t *output)
 {
+	UNUSED_PARAMETER(settings);
+
 	outputRef = output;
 
 	blog(LOG_DEBUG, "output_create");
@@ -126,12 +128,15 @@
 
 static void virtualcam_output_destroy(void *data)
 {
+	UNUSED_PARAMETER(data);
 	blog(LOG_DEBUG, "output_destroy");
 	sMachServer = nil;
 }
 
 static bool virtualcam_output_start(void *data)
 {
+	UNUSED_PARAMETER(data);
+
 	bool hasDalPlugin = check_dal_plugin();
 
 	if (!hasDalPlugin) {
@@ -158,6 +163,9 @@
 
 static void virtualcam_output_stop(void *data, uint64_t ts)
 {
+	UNUSED_PARAMETER(data);
+	UNUSED_PARAMETER(ts);
+
 	blog(LOG_DEBUG, "output_stop");
 	obs_output_end_data_capture(outputRef);
 	[sMachServer stop];
@@ -165,6 +173,8 @@
 
 static void virtualcam_output_raw_video(void *data, struct video_data *frame)
 {
+	UNUSED_PARAMETER(data);
+
 	uint8_t *outData = frame->data[0];
 	if (frame->linesize[0] != (videoInfo.output_width * 2)) {
 		blog(LOG_ERROR,

obs-studio-26.1.0.tar.xz/plugins/obs-browser/obs-browser-plugin.cpp -> obs-studio-26.1.1.tar.xz/plugins/obs-browser/obs-browser-plugin.cpp Changed

obs-studio-26.1.0.tar.xz/plugins/obs-ffmpeg/ffmpeg-mux/ffmpeg-mux.c -> obs-studio-26.1.1.tar.xz/plugins/obs-ffmpeg/ffmpeg-mux/ffmpeg-mux.c Changed

obs-studio-26.1.0.tar.xz/plugins/rtmp-services/data/package.json -> obs-studio-26.1.1.tar.xz/plugins/rtmp-services/data/package.json Changed

obs-studio-26.1.0.tar.xz/plugins/rtmp-services/data/services.json -> obs-studio-26.1.1.tar.xz/plugins/rtmp-services/data/services.json Changed

@@ -260,77 +260,56 @@
             }
         },
         {
-            "name": "VIMM",
+            "name": "Loola.tv",
+            "common": false,
             "servers": [
                 {
-                    "name": "Europe: Frankfurt",
-                    "url": "rtmp://eu.vimm.tv/live"
+                    "name": "US East: Virginia",
+                    "url": "rtmp://rtmp.loola.tv/push"
                 },
                 {
-                    "name": "North America: Montreal",
-                    "url": "rtmp://us.vimm.tv/live"
+                    "name": "EU Central: Germany",
+                    "url": "rtmp://rtmp-eu.loola.tv/push"
+                },
+                {
+                    "name": "South America: Brazil",
+                    "url": "rtmp://rtmp-sa.loola.tv/push"
+                },
+                {
+                    "name": "Asia/Pacific: Singapore",
+                    "url": "rtmp://rtmp-sg.loola.tv/push"
+                },
+                {
+                    "name": "Middle East: Bahrain",
+                    "url": "rtmp://rtmp-me.loola.tv/push"
                 }
             ],
             "recommended": {
                 "keyint": 2,
-                "max video bitrate": 8000,
-                "max audio bitrate": 320,
+                "profile": "high",
+                "max video bitrate": 2500,
+                "max audio bitrate": 160,
+                "bframes": 2,
                 "x264opts": "scenecut=0"
             }
         },
         {
-            "name": "Smashcast",
+            "name": "VIMM",
             "servers": [
                 {
-                    "name": "Default",
-                    "url": "rtmp://live.hitbox.tv/push"
-                },
-                {
-                    "name": "EU-North: Amsterdam, Netherlands",
-                    "url": "rtmp://live.ams.hitbox.tv/push"
-                },
-                {
-                    "name": "EU-West: Paris, France",
-                    "url": "rtmp://live.cdg.hitbox.tv/push"
-                },
-                {
-                    "name": "EU-South: Milan, Italia",
-                    "url": "rtmp://live.mxp.hitbox.tv/push"
-                },
-                {
-                    "name": "Russia: Moscow",
-                    "url": "rtmp://live.dme.hitbox.tv/push"
-                },
-                {
-                    "name": "US-East: New York",
-                    "url": "rtmp://live.jfk.hitbox.tv/push"
-                },
-                {
-                    "name": "US-West: San Francisco",
-                    "url": "rtmp://live.sfo.hitbox.tv/push"
-                },
-                {
-                    "name": "US-West: Los Angeles",
-                    "url": "rtmp://live.lax.hitbox.tv/push"
-                },
-                {
-                    "name": "South America: Sao Paulo, Brazil",
-                    "url": "rtmp://live.gru.hitbox.tv/push"
-                },
-                {
-                    "name": "Asia: Singapore",
-                    "url": "rtmp://live.sin.hitbox.tv/push"
+                    "name": "Europe: Frankfurt",
+                    "url": "rtmp://eu.vimm.tv/live"
                 },
                 {
-                    "name": "Oceania: Sydney, Australia",
-                    "url": "rtmp://live.syd.hitbox.tv/push"
+                    "name": "North America: Montreal",
+                    "url": "rtmp://us.vimm.tv/live"
                 }
             ],
             "recommended": {
                 "keyint": 2,
-                "profile": "high",
-                "max video bitrate": 3500,
-                "max audio bitrate": 320
+                "max video bitrate": 8000,
+                "max audio bitrate": 320,
+                "x264opts": "scenecut=0"
             }
         },
         {
@@ -399,7 +378,7 @@
                 {
                     "name": "US: New York, NY",
                     "url": "rtmp://live-nyc.vaughnsoft.net/live"
-                },		    
+                },
                 {
                     "name": "US: Miami, FL",
                     "url": "rtmp://live-mia.vaughnsoft.net/live"
@@ -441,7 +420,7 @@
                 {
                     "name": "US: New York, NY",
                     "url": "rtmp://live-nyc.vaughnsoft.net/live"
-                },		    
+                },
                 {
                     "name": "US: Miami, FL",
                     "url": "rtmp://live-mia.vaughnsoft.net/live"
@@ -1779,7 +1758,7 @@
                 "max audio bitrate": 160,
                 "x264opts": "tune=zerolatency"
             }
-	},
+        },
         {
             "name": "Mux",
             "servers": [

obs-studio-26.1.0.tar.xz/plugins/win-dshow/libdshowcapture/dshowcapture.hpp -> obs-studio-26.1.1.tar.xz/plugins/win-dshow/libdshowcapture/dshowcapture.hpp Changed

obs-studio-26.1.0.tar.xz/plugins/win-dshow/libdshowcapture/source/dshow-base.cpp -> obs-studio-26.1.1.tar.xz/plugins/win-dshow/libdshowcapture/source/dshow-base.cpp Changed

@@ -780,9 +780,105 @@
 	return hr;
 }
 
+static HRESULT GetFriendlyName(REFCLSID deviceClass, const wchar_t *devPath,
+			 wchar_t *name, int nameSize)
+{
+	/* Sanity checks */
+	if (!devPath)
+		return E_POINTER;
+	if (!name)
+		return E_POINTER;
+
+	/* Create device enumerator */
+	ComPtr<ICreateDevEnum> createDevEnum;
+	HRESULT hr = CoCreateInstance(CLSID_SystemDeviceEnum, NULL,
+				 CLSCTX_INPROC_SERVER, IID_ICreateDevEnum,
+				 (void **)&createDevEnum);
+
+	/* Enumerate filters */
+	ComPtr<IEnumMoniker> enumMoniker;
+	if (SUCCEEDED(hr)) {
+		/* returns S_FALSE if no devices are installed */
+		hr = createDevEnum->CreateClassEnumerator(deviceClass,
+							 &enumMoniker, 0);
+		if (!enumMoniker)
+			hr = E_FAIL;
+	}
+
+	/* Cycle through the enumeration */
+	if (SUCCEEDED(hr)) {
+		ULONG fetched = 0;
+		ComPtr<IMoniker> moniker;
+
+		enumMoniker->Reset();
+
+		while (enumMoniker->Next(1, &moniker, &fetched) == S_OK) {
+
+			/* Get device path from moniker */
+			wchar_t monikerDevPath[512];
+			hr = ReadProperty(moniker, L"DevicePath",
+					 monikerDevPath,
+					 _ARRAYSIZE(monikerDevPath));
+
+			/* Find desired filter */
+			if (wcscmp(devPath, monikerDevPath) == 0) {
+
+				/* Get friendly name */
+				hr = ReadProperty(moniker, L"FriendlyName",
+						 name, nameSize);
+				return hr;
+			}
+		}
+	}
+
+	return E_FAIL;
+}
+
+static bool MatchFriendlyNames(const wchar_t *vidName, const wchar_t *audName)
+{
+	/* Sanity checks */
+	if (!vidName)
+		return false;
+	if (!audName)
+		return false;
+
+	/* Convert strings to lower case */
+	wstring strVidName = vidName;
+	for (wchar_t &c : strVidName)
+		c = (wchar_t)tolower(c);
+	wstring strAudName = audName;
+	for (wchar_t &c : strAudName)
+		c = (wchar_t)tolower(c);
+
+	/* Remove 'video' from friendly name */
+	size_t posVid;
+	wstring searchVid[] = {L"(video) ", L"(video)", L"video ", L"video"};
+	for (int i = 0; i < _ARRAYSIZE(searchVid); i++) {
+		wstring &search = searchVid[i];
+		while ((posVid = strVidName.find(search)) !=
+		 std::string::npos) {
+			strVidName.replace(posVid, search.length(), L"");
+		}
+	}
+
+	/* Remove 'audio' from friendly name */
+	size_t posAud;
+	wstring searchAud[] = {L"(audio) ", L"(audio)", L"audio ", L"audio"};
+	for (int i = 0; i < _ARRAYSIZE(searchAud); i++) {
+		wstring &search = searchAud[i];
+		while ((posAud = strAudName.find(search)) !=
+		 std::string::npos) {
+			strAudName.replace(posAud, search.length(), L"");
+		}
+	}
+
+	return strVidName == strAudName;
+}
+
 static bool GetDeviceAudioFilterInternal(REFCLSID deviceClass,
 					 const wchar_t *vidDevPath,
-					 IBaseFilter **audioCaptureFilter)
+					 IBaseFilter **audioCaptureFilter,
+					 bool matchFilterName = false)
 {
 	/* Get video device instance path */
 	wchar_t vidDevInstPath[512];
@@ -797,6 +893,15 @@
 		return false;
 #endif
 
+	/* Get friendly name */
+	wchar_t vidName[512];
+	if (matchFilterName) {
+		hr = GetFriendlyName(CLSID_VideoInputDeviceCategory, vidDevPath,
+				 vidName, _ARRAYSIZE(vidName));
+		if (FAILED(hr))
+			return false;
+	}
+
 	/* Create device enumerator */
 	ComPtr<ICreateDevEnum> createDevEnum;
 	if (SUCCEEDED(hr))
@@ -823,12 +928,6 @@
 
 		while (enumMoniker->Next(1, &moniker, &fetched) == S_OK) {
 			bool samePath = false;
-#if 0
-			/* Get friendly name (helpful for debugging) */
-			wchar_t friendlyName[512];
-			ReadProperty(moniker, L"FriendlyName", friendlyName,
-					_ARRAYSIZE(friendlyName));
-#endif
 
 			/* Get device path */
 			wchar_t audDevPath[512];
@@ -848,11 +947,29 @@
 
 			/* Get audio capture filter */
 			if (samePath) {
-				hr = moniker->BindToObject(
-					0, 0, IID_IBaseFilter,
-					(void **)audioCaptureFilter);
-				if (SUCCEEDED(hr))
-					return true;
+				/* Match video and audio filter names */
+				bool isSameFilterName = false;
+				if (matchFilterName) {
+					wchar_t audName[512];
+					hr = ReadProperty(moniker,
+							 L"FriendlyName",
+							 audName,
+							 _ARRAYSIZE(audName));
+					if (SUCCEEDED(hr)) {
+						isSameFilterName =
+							MatchFriendlyNames(
+								vidName,
+								audName);
+					}
+				}
+
+				if (!matchFilterName || isSameFilterName) {
+					hr = moniker->BindToObject(
+						0, 0, IID_IBaseFilter,
+						(void **)audioCaptureFilter);
+					if (SUCCEEDED(hr))
+						return true;
+				}
 			}
 		}
 	}
@@ -863,9 +980,23 @@
 bool GetDeviceAudioFilter(const wchar_t *vidDevPath,
 			 IBaseFilter **audioCaptureFilter)
 {
-	/* Search in "Audio capture sources" */
+	/* Search in "Audio capture sources" and match filter name */
 	bool success = GetDeviceAudioFilterInternal(
-		CLSID_AudioInputDeviceCategory, vidDevPath, audioCaptureFilter);
+		CLSID_AudioInputDeviceCategory, vidDevPath, audioCaptureFilter,
+		true);
+
+	/* Search in "WDM Streaming Capture Devices" and match filter name */
+	if (!success)
+		success = GetDeviceAudioFilterInternal(KSCATEGORY_CAPTURE,
+						 vidDevPath,
+						 audioCaptureFilter,
+						 true);
+
+	/* Search in "Audio capture sources" */
+	if (!success)
+		success = GetDeviceAudioFilterInternal(
+			CLSID_AudioInputDeviceCategory, vidDevPath,
+			audioCaptureFilter);
 
 	/* Search in "WDM Streaming Capture Devices" */
 	if (!success)

obs-studio-26.1.0.tar.xz/plugins/win-dshow/libdshowcapture/source/output-filter.cpp -> obs-studio-26.1.1.tar.xz/plugins/win-dshow/libdshowcapture/source/output-filter.cpp Changed

Build Results
RPM Lint

Refresh

Comments for request 5292 0

Request History

boombatower created request almost 4 years ago

boombatower accepted request almost 4 years ago