release_wheel.yml

# Adapted from https://github.com/punica-ai/punica/blob/591b59899f0a20760821785d06b331c8a2e5cb86/.github/workflows/release_wheel.yml
name: Release
on:
  workflow_dispatch:
    inputs:
      tag_name:
        required: true
        type: string
  workflow_call:
    inputs:
      tag_name:
        required: true
        type: string
    secrets:
      WHL_TOKEN:
        required: true
      # PYPI_TEST_TOKEN:
      #   required: true

env:
  TORCH_CUDA_ARCH_LIST: "8.0 8.9 9.0+PTX"

jobs:
  build:
    strategy:
      fail-fast: false
      matrix:
        python: ["3.8", "3.9", "3.10", "3.11"]
        cuda: ["11.8", "12.1", "12.4"]
        torch: ["2.2", "2.3", "2.4"]
        exclude: # for cuda 12.4, we only support torch 2.4+
          - cuda: "12.4"
            torch: "2.2"
          - cuda: "12.4"
            torch: "2.3"

    runs-on: [self-hosted]
    steps:
      - uses: actions/checkout@v4
        with:
          submodules: true

      - name: Build wheel
        run: |
          chown -R $CI_UID:$CI_GID "$GITHUB_WORKSPACE"
          version="$(cat version.txt)"
          docker run --rm -t \
              -v "$CI_RUNNER_CACHE_DIR":/ci-cache \
              -v "$GITHUB_WORKSPACE":/app \
              -e FLASHINFER_CI_PYTHON_VERSION=${{ matrix.python }} \
              -e FLASHINFER_CI_CUDA_VERSION=${{ matrix.cuda }} \
              -e FLASHINFER_CI_TORCH_VERSION=${{ matrix.torch }} \
              -e FLASHINFER_BUILD_VERSION=$version \
              -e TORCH_CUDA_ARCH_LIST="$TORCH_CUDA_ARCH_LIST" \
              --user $CI_UID:$CI_GID \
              pytorch/manylinux-builder:cuda${{ matrix.cuda }} \
              bash /app/scripts/run-ci-build-wheel.sh
        timeout-minutes: 120
      - run: du -h python/dist/*

      - uses: actions/upload-artifact@v4
        with:
          name: wheel-cuda${{ matrix.cuda }}-torch${{ matrix.torch }}-python${{ matrix.python }}
          path: python/dist/*

  release:
    needs: build
    runs-on: [self-hosted]
    steps:
      - uses: actions/download-artifact@v4
        with:
          path: python/dist/
          merge-multiple: true
          pattern: wheel-*

      - run: ls -lah python/dist/

      - uses: softprops/action-gh-release@v1
        with:
          tag_name: ${{ inputs.tag_name }}
          files: |
            python/dist/flashinfer*cp38*.whl

      - uses: softprops/action-gh-release@v1
        with:
          tag_name: ${{ inputs.tag_name }}
          files: |
            python/dist/flashinfer*cp39*.whl

      - uses: softprops/action-gh-release@v1
        with:
          tag_name: ${{ inputs.tag_name }}
          files: |
            python/dist/flashinfer*cp310*.whl

      - uses: softprops/action-gh-release@v1
        with:
          tag_name: ${{ inputs.tag_name }}
          files: |
            python/dist/flashinfer*cp311*.whl

      - uses: softprops/action-gh-release@v1
        with:
          tag_name: ${{ inputs.tag_name }}
          files: |
            python/dist/flashinfer-*.tar.gz

      - name: Clone wheel index
        run: git clone https://oauth2:${WHL_TOKEN}@github.com/flashinfer-ai/whl.git flashinfer-whl
        env:
          WHL_TOKEN: ${{ secrets.WHL_TOKEN }}

      - name: Update wheel index
        run: python3 scripts/update_whl_index.py

      - name: Push wheel index
        run: |
          cd flashinfer-whl
          git config --local user.name "github-actions[bot]"
          git config --local user.email "41898282+github-actions[bot]@users.noreply.github.com"
          git add -A
          git commit -m "update whl"
          git push

      # - name: Upload sdist to pypi
      #   run: |
      #     pip install twine
      #     python -m twine upload --repository testpypi --username=__token__ dist/*.tar.gz
      #   env:
      #     TWINE_PASSWORD: ${{ secrets.PYPI_TEST_TOKEN }}