x86 avx512: add more reduction functions for 32-bit and 64-bit intege… #42

Workflow file for this run

.github/workflows/gcc-snapshot.yml at c3e0eb1

	name: GCC-16

	on:
	push:
	branches:
	- 'master'
	workflow_dispatch: {}
	repository_dispatch:
	types: [on-demand-gcc-16]

	concurrency:
	group: gcc-snapshot-${{ github.event.pull_request.number \|\| github.ref }}
	cancel-in-progress: true

	jobs:
	x86:
	runs-on: ubuntu-24.04
	strategy:
	fail-fast: false
	matrix:
	isax:
	- -DSIMDE_NATURAL_VECTOR_SIZE=256 -march=x86-64-v3 -mavx512bw -mavx512vl
	# https://en.wikipedia.org/wiki/X86-64#Microarchitecture_levels
	- -march=x86-64 -maes -mpclmul # the x86-64 baseline is CMOV, CX8, FPU, FXSR, MMX, OSFXSR, SCE, SSE, SSE2
	- -march=x86-64-v2 # CMPXCHG16B, LAHF-SAHF, POPCNT, SSE3, SSE4_1, SSE4_2, SSSE3
	- -march=x86-64-v3 # AVX, AVX2, BMI1, BMI2, F16C, FMA, LZCNT, MOVBE, OSXSAVE
	- -march=x86-64-v4 # AVX512F, AVX512BW, AVX512CD, AVX512DQ, AVX512VL a.k.a. skylake+ (including zen4)
	include:
	- target: tgl
	- isax: -march=x86-64-v4 -mcx16 -mxsave -mpclmul -mfsgsbase -mrdrnd -mhle -mrdseed -maes -mclflushopt -mxsavec -mxsaves -msgx -mpku -msha -mrdpid -mavx512vpopcntdq -mavx512ifma -mavx512vbmi -mavx512vnni -mavx512vbmi2 -mavx512bitalg -mvpclmulqdq -mgfni -mvaes # icelake
	target: icl
	- isax: -march=x86-64-v4 -mcx16 -mxsave -mpclmul -mfsgsbase -mrdrnd -mhle -mrdseed -maes -mclflushopt -mxsavec -mxsaves -msgx -mpku -msha -mrdpid -mavx512vpopcntdq -mavx512ifma -mavx512vbmi -mavx512vnni -mavx512vbmi2 -mavx512bitalg -mvpclmulqdq -mgfni -mvaes -mpconfig -mwbnoinvd -mclwb -mmovdiri -mmovdir64b -menqcmd -mcldemote -mptwrite -mwaitpkg -mserialize -mtsxldtrk -muintr -mavxvnni -mavx512fp16 # sapphire rapids without bf16
	# See https://github.com/simd-everywhere/simde/issues/1095
	target: spr
	env:
	CFLAGS: -Wall -Wextra -Werror ${{ matrix.isax }}
	CXXFLAGS: -Wall -Wextra -Werror ${{ matrix.isax }}
	INTEL_TARGET: ${{ matrix.target }}
	steps:
	- uses: actions/checkout@v6
	if: ${{ github.event_name != 'repository_dispatch' }}
	with:
	submodules: recursive
	- uses: actions/checkout@v6
	if: ${{ github.event_name == 'repository_dispatch' }}
	with:
	submodules: recursive
	repository: ${{ github.event.client_payload.repository }}
	ref: ${{ github.event.client_payload.ref }}
	- name: CPU Information
	run: cat /proc/cpuinfo
	- name: Install APT Dependencies
	run: \|
	sudo add-apt-repository ppa:misterc/gcc-snapshot-latest-lts
	# sudo add-apt-repository ppa:stephanosio/ccache
	sudo apt-get install -y --no-install-recommends ninja-build pipx
	sudo apt-get install -y ${{ github.event.client_payload.extra }} gcc-16 g++-16 # ccache
	sudo apt-get purge -y gcc g++
	sudo ln -s /usr/bin/gcc-16 /usr/bin/gcc
	sudo ln -s /usr/bin/g++-16 /usr/bin/g++
	pipx install meson==1.3.2
	# - name: ccache
	# uses: hendrikmuhs/[email protected]
	# with:
	# key: gcc-16-${{ github.job }}-${{ matrix.isax }}
	# verbose: 2
	# - name: add ccache to the build path
	# run: \|
	# export PATH="/usr/lib/ccache:/usr/local/opt/ccache/libexec:$PATH"
	- name: Configure
	run: meson setup build \|\| (cat build/meson-logs/meson-log.txt ; false)
	- name: Test run native?
	run: \|
	test/check-flags.sh query && echo Tests with "$CFLAGS" will be run natively
	test/check-flags.sh query \|\| echo Tests with "$CFLAGS" will be run using SDE
	- name: Build
	run: ninja -C build -v
	- name: Test
	run: \|
	# shellcheck disable=SC2046
	meson test -C build --print-errorlogs --wrapper "${GITHUB_WORKSPACE}/test/check-flags.sh sde" # $(meson test -C build --list \| grep -v emul)

	x86-xop:
	runs-on: ubuntu-24.04
	strategy:
	fail-fast: false
	env:
	CFLAGS: -Wall -Wextra -Werror -march=bdver2
	CXXFLAGS: -Wall -Wextra -Werror -march=bdver2
	steps:
	- uses: actions/checkout@v6
	if: ${{ github.event_name != 'repository_dispatch' }}
	with:
	submodules: recursive
	- uses: actions/checkout@v6
	if: ${{ github.event_name == 'repository_dispatch' }}
	with:
	submodules: recursive
	repository: ${{ github.event.client_payload.repository }}
	ref: ${{ github.event.client_payload.ref }}
	- name: CPU Information
	run: cat /proc/cpuinfo
	- name: Install APT Dependencies
	run: \|
	sudo add-apt-repository ppa:misterc/gcc-snapshot-latest-lts
	# sudo add-apt-repository ppa:stephanosio/ccache
	sudo apt-get install -y --no-install-recommends ninja-build pipx \
	qemu-user-static # ccache
	sudo apt-get install -y ${{ github.event.client_payload.extra }} gcc-16 g++-16
	sudo apt-get purge -y gcc g++
	sudo ln -s /usr/bin/gcc-16 /usr/bin/gcc
	sudo ln -s /usr/bin/g++-16 /usr/bin/g++
	pipx install meson==1.3.2
	# - name: ccache
	# uses: hendrikmuhs/[email protected]
	# with:
	# key: gcc-16-${{ github.job }}
	# verbose: 2
	# - name: add ccache to the build path
	# run: \|
	# export PATH="/usr/lib/ccache:/usr/local/opt/ccache/libexec:$PATH"
	- name: Configure
	run: meson setup build \|\| (cat build/meson-logs/meson-log.txt ; false)
	- name: Build
	run: ninja -C build -v
	# can't test until we find a combination of `gcc -march=` and `qemu -cpu` that both enable XOP and allows qemu to test it
	# - name: Test
	# run: meson test -C build --print-errorlogs --wrapper "qemu-amd64-static -cpu Opteron_G5-v1"

	sleef:
	runs-on: ubuntu-24.04
	env:
	CFLAGS: -march=native -Wall -Wextra -Werror
	CXXFLAGS: -march=native -Wall -Wextra -Werror
	steps:
	- uses: actions/checkout@v6
	if: ${{ github.event_name != 'repository_dispatch' }}
	with:
	submodules: recursive
	- uses: actions/checkout@v6
	if: ${{ github.event_name == 'repository_dispatch' }}
	with:
	submodules: recursive
	repository: ${{ github.event.client_payload.repository }}
	ref: ${{ github.event.client_payload.ref }}
	- id: cpu
	name: CPU Information
	run: \|
	cat /proc/cpuinfo
	- name: Install APT Dependencies
	run: \|
	sudo add-apt-repository ppa:misterc/gcc-snapshot-latest-lts
	# sudo add-apt-repository ppa:stephanosio/ccache
	sudo apt-get install -y --no-install-recommends ninja-build pipx \
	libsleef-dev # ccache
	sudo apt-get install -y ${{ github.event.client_payload.extra }} gcc-16 g++-16
	sudo apt-get purge -y gcc g++
	sudo ln -s /usr/bin/gcc-16 /usr/bin/gcc
	sudo ln -s /usr/bin/g++-16 /usr/bin/g++
	pipx install meson==1.3.2
	# - name: ccache
	# uses: hendrikmuhs/[email protected]
	# with:
	# key: gcc-16-${{ github.job }}
	# verbose: 2
	# - name: add ccache to the build path
	# run: \|
	# export PATH="/usr/lib/ccache:/usr/local/opt/ccache/libexec:$PATH"
	- name: Configure
	run: meson setup build -Dsleef=enabled \|\| (cat build/meson-logs/meson-log.txt ; false)
	- name: Build
	run: ninja -C build -v
	- name: Test
	run: \|
	# shellcheck disable=SC2046
	meson test -C build --print-errorlogs # $(meson test -C build --list \| grep -v emul)

	gcc-16:
	strategy:
	fail-fast: false
	matrix:
	include:
	- distro: ubuntu-24.04
	arch_flags: -ffast-math
	- distro: ubuntu-24.04
	- distro: ubuntu-24.04-arm
	- distro: ubuntu-24.04-arm
	plain: true
	- distro: ubuntu-24.04-arm # plain armhf
	gcc: arm-linux-gnueabihf
	plain: true
	- distro: ubuntu-24.04-arm # armv7
	gcc: arm-linux-gnueabihf
	arch_flags: -march=armv7-a+simd+neon-fp16
	plain: true
	- distro: ubuntu-24.04-arm # armv8 32-bit
	gcc: arm-linux-gnueabihf
	arch_flags: -mcpu=cortex-a32 -mfpu=crypto-neon-fp-armv8
	plain: true
	runs-on: ${{ matrix.distro }}
	env:
	CFLAGS: ${{ matrix.arch_flags }} ${{ case(
	matrix.plain == true, '-Wall -Wextra -Werror',
	'-march=native -Wall -Wextra -Werror') }}
	CXXFLAGS: ${{ matrix.arch_flags }} ${{ case(
	matrix.plain == true, '-Wall -Wextra -Werror',
	'-march=native -Wall -Wextra -Werror') }}
	CC: ${{
	case(
	matrix.gcc == '', 'gcc-16',
	format('{0}-gcc-16', matrix.gcc)
	) }}
	CXX: ${{
	case(
	matrix.gcc == '', 'g++-16',
	format('{0}-g++-16', matrix.gcc)
	) }}

	steps:
	- uses: actions/checkout@v6
	if: ${{ github.event_name != 'repository_dispatch' }}
	with:
	submodules: recursive
	- uses: actions/checkout@v6
	if: ${{ github.event_name == 'repository_dispatch' }}
	with:
	submodules: recursive
	repository: ${{ github.event.client_payload.repository }}
	ref: ${{ github.event.client_payload.ref }}
	- id: cpu
	name: CPU Information
	run: \|
	cat /proc/cpuinfo
	- name: enable armhf
	if: ${{ matrix.gcc == 'arm-linux-gnueabihf' }}
	run: sudo dpkg --add-architecture armhf && sudo apt-get update && sudo apt-get install -y libstdc++6:armhf
	- name: Install APT Dependencies
	run: \|
	sudo add-apt-repository ppa:misterc/gcc-snapshot-latest-lts
	# sudo add-apt-repository ppa:stephanosio/ccache
	sudo apt-get -y install --no-install-recommends pipx ninja-build \
	# ccache
	sudo apt-get install -y ${{ github.event.client_payload.extra }} \
	gcc-${{ case(matrix.gcc == '', '16', format('16-{0}', matrix.gcc)) }} \
	g++-${{ case(matrix.gcc == '', '16', format('16-{0}', matrix.gcc)) }}
	sudo apt-get -y purge g++ gcc
	pipx install meson==1.3.2
	# - name: ccache
	# uses: hendrikmuhs/[email protected]
	# with:
	# key: ${{ github.job }}-${{ matrix.version }}-${{ matrix.distro }}-${{ matrix.arch_flags }}
	# verbose: 2
	# - name: add ccache to the build path
	# run: \|
	# export PATH="/usr/lib/ccache:/usr/local/opt/ccache/libexec:$PATH"
	- name: Configure
	run: meson setup build \|\| (cat build/meson-logs/meson-log.txt ; false)
	- name: Build
	run: meson compile -C build --verbose
	- name: Test
	run: \|
	# shellcheck disable=SC2046
	meson test -C build --print-errorlogs # $(meson test -C build --list \| grep -v emul)

	gcc-qemu:
	strategy:
	fail-fast: false
	matrix:
	include:
	# - version: 16
	# cross: armel
	# arch_gnu_abi: eabi
	# arch_deb: armel
	# arch_gnu: arm
	# distro: ubuntu-24.04
	- version: 16
	cross: riscv64
	arch_gnu: riscv64
	arch_deb: riscv64
	distro: ubuntu-24.04
	- extra: -O3
	version: 16
	cross: riscv64
	arch_gnu: riscv64
	arch_deb: riscv64
	distro: ubuntu-24.04
	- version: 16
	cross: s390x
	arch_gnu: s390x
	arch_deb: s390x
	distro: ubuntu-24.04
	# - version: 16
	# cross: power9
	# arch_gnu: powerpc64le
	# arch_deb: ppc64el
	# distro: ubuntu-24.04
	- version: 16
	cross: loongarch64
	arch_gnu: loongarch64
	arch_deb: loong64
	distro: ubuntu-24.04
	runs-on: ${{ matrix.distro }}
	steps:
	- uses: actions/checkout@v6
	if: ${{ github.event_name != 'repository_dispatch' }}
	with:
	submodules: recursive
	- uses: actions/checkout@v6
	if: ${{ github.event_name == 'repository_dispatch' }}
	with:
	submodules: recursive
	repository: ${{ github.event.client_payload.repository }}
	ref: ${{ github.event.client_payload.ref }}
	- name: CPU Information
	run: cat /proc/cpuinfo
	- name: Install APT Dependencies
	run: \|
	sudo add-apt-repository ppa:misterc/gcc-snapshot-latest-lts
	sudo apt-get -y --no-install-recommends install ccache ninja-build \
	binfmt-support qemu-user-static pipx libc6-${{ matrix.arch_deb }}-cross
	sudo apt-get install -y ${{ github.event.client_payload.extra }} \
	gcc-${{ matrix.version }}-${{ matrix.arch_gnu }}-linux-gnu \
	g++-${{ matrix.version }}-${{ matrix.arch_gnu }}-linux-gnu \
	libstdc++-${{ matrix.version }}-dev-${{ matrix.arch_deb }}-cross
	pipx install meson==1.3.2
	# - name: ccache
	# uses: hendrikmuhs/[email protected]
	# with:
	# key: ${{ github.job }}-${{ matrix.version}}${{ matrix.extra }}-${{ matrix.distro }}-${{ matrix.cross }}
	# verbose: 2
	- name: Configure
	run: meson setup --cross-file=docker/cross-files/${{ matrix.cross }}-gcc-${{ matrix.version }}${{ matrix.extra}}-ccache.cross build \|\| (cat build/meson-logs/meson-log.txt ; false)
	- name: Build
	run: ninja -C build -v
	- name: Test
	run: \|
	# shellcheck disable=SC2046
	meson test -C build --print-errorlogs

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

x86 avx512: add more reduction functions for 32-bit and 64-bit intege… #42

Workflow file

x86 avx512: add more reduction functions for 32-bit and 64-bit intege… #42

Uh oh!

Workflow file for this run