Build flash-attention Wheels for Windows #16
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Build flash-attention Wheels for Windows | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| version: | |
| description: 'Version tag of flash-attention to build: (format: v2.3.4/v2.3.4.post1)' | |
| default: 'v2.3.4' | |
| required: true | |
| type: string | |
| workflow_call: | |
| inputs: | |
| version: | |
| description: 'Version tag of flash-attention to build (format: v2.3.4/v2.3.4.post1)' | |
| default: 'v2.3.4' | |
| required: true | |
| type: string | |
| permissions: | |
| contents: write | |
| jobs: | |
| build_wheels: | |
| name: Build wheels for Python ${{ matrix.pyver }}, CUDA ${{ matrix.cuda }}, and Torch ${{ matrix.torchver }} | |
| runs-on: windows-2022 | |
| strategy: | |
| matrix: | |
| pyver: ["3.10", "3.11"] | |
| cuda: ["12.2.2"] | |
| torchver: ["2.2.2", "2.3.1"] | |
| defaults: | |
| run: | |
| shell: pwsh | |
| env: | |
| CUDAVER: ${{ matrix.cuda }} | |
| PCKGVER: ${{ inputs.version }} | |
| steps: | |
| - uses: actions/checkout@v4 | |
| with: | |
| submodules: 'recursive' | |
| - uses: actions/setup-python@v4 | |
| with: | |
| python-version: ${{ matrix.pyver }} | |
| - name: Install VS2022 BuildTools 17.9.7 | |
| run: choco install -y visualstudio2022buildtools --version=117.9.7.0 --params "--add Microsoft.VisualStudio.Component.VC.Tools.x86.x64 --installChannelUri https://aka.ms/vs/17/release/180911598_-255012421/channel" | |
| if: runner.os == 'Windows' | |
| - name: Setup Mamba | |
| uses: conda-incubator/[email protected] | |
| with: | |
| activate-environment: "build" | |
| python-version: ${{ matrix.pyver }} | |
| miniforge-variant: Mambaforge | |
| miniforge-version: latest | |
| use-mamba: true | |
| add-pip-as-python-dependency: true | |
| auto-activate-base: false | |
| - name: Install Dependencies | |
| run: | | |
| $cudaVersion = '${{ matrix.cuda }}' | |
| $cudaVersionPytorch = '${{ matrix.cuda }}'.Remove('${{ matrix.cuda }}'.LastIndexOf('.')).Replace('.','') | |
| mamba install -y -c nvidia/label/cuda-$cudaVersion cuda-toolkit cuda-runtime | |
| if (!(mamba list cuda)[-1].contains('cuda')) {sleep -s 10; mamba install -y 'cuda' $cudaChannels.TrimEnd().Split()} | |
| if (!(mamba list cuda)[-1].contains('cuda')) {throw 'CUDA Toolkit failed to install!'} | |
| python -m pip install --upgrade build setuptools wheel packaging ninja torch==${{ matrix.torchver }} psutil --extra-index-url "https://download.pytorch.org/whl/cu121" | |
| - name: Build Wheel | |
| id: build-wheel | |
| run: | | |
| # --- Spawn the VS shell | |
| if ($IsWindows) { | |
| Import-Module 'C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\Common7\Tools\Microsoft.VisualStudio.DevShell.dll' | |
| Enter-VsDevShell -VsInstallPath 'C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools' -DevCmdArguments '-arch=x64 -host_arch=x64' | |
| $env:DISTUTILS_USE_SDK=1 | |
| } | |
| $cudaVersion = $env:CUDAVER.Remove($env:CUDAVER.LastIndexOf('.')).Replace('.','') | |
| $packageVersion = $env:PCKGVER.TrimStart('v') | |
| $env:CUDA_PATH = $env:CONDA_PREFIX | |
| $env:CUDA_HOME = $env:CONDA_PREFIX | |
| $env:MAX_JOBS = '1' | |
| $env:FLASH_ATTENTION_FORCE_BUILD = 'TRUE' | |
| $env:FLASH_ATTENTION_FORCE_SINGLE_THREAD = 'TRUE' | |
| python -m build -n --wheel | |
| $wheel = (gi '.\dist\*.whl')[0] | |
| $wheelname = $wheel.name.replace("flash_attn-$packageVersion-","flash_attn-$packageVersion+cu$cudaVersion"+"torch${{ matrix.torchver }}cxx11abiFALSE-") | |
| Move-Item $wheel.fullname ".\dist\$wheelname" | |
| - uses: actions/upload-artifact@v3 | |
| with: | |
| name: 'windows-wheels' | |
| path: ./dist/*.whl | |
| - name: Upload files to a GitHub release | |
| uses: svenstaro/[email protected] | |
| continue-on-error: true | |
| with: | |
| file: ./dist/*.whl | |
| tag: ${{ inputs.version }} | |
| file_glob: true | |
| overwrite: true | |
| release_name: ${{ inputs.version }} |