Build and publish nightly monarch wheels #686
This file contains hidden or 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 and publish nightly monarch wheels | |
| on: | |
| workflow_dispatch: | |
| # For testing changes to this workflow. | |
| pull_request: | |
| branches: | |
| - main | |
| - gh/** | |
| paths: | |
| - .github/workflows/wheels.yml | |
| - .github/workflows/build-dist.yml | |
| - .github/workflows/build-macos.yml | |
| - Dockerfile.nightly | |
| push: | |
| branches: | |
| - main | |
| # Run as soon as any changes to .github actions land | |
| paths: | |
| - .github/** | |
| schedule: | |
| - cron: '0 0 * * *' # Daily at midnight UTC | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }} | |
| cancel-in-progress: true | |
| jobs: | |
| build-x86_64: | |
| uses: ./.github/workflows/build-dist.yml | |
| with: | |
| python-versions: '["3.10", "3.11", "3.12", "3.13"]' | |
| # Nightly wheels ship cu132 even though PR CI tests at cu130 (see | |
| # set-matrix.yaml): pytorch core does the same split — publish cu132 | |
| # binaries, GPU-test only at cu130 — because no g4dn-fleet AMI yet | |
| # carries an NVIDIA driver that satisfies nvidia-container-cli's | |
| # cuda>=13.2 check. | |
| # | |
| # Building a wheel needs the CUDA toolkit (nvcc, headers, libs) but not | |
| # host-GPU access. The cuda13.2 builder image supplies the toolkit; we | |
| # set gpu-arch-type=cpu so test-infra's linux_job_v2.yml skips the | |
| # `setup-nvidia` step that mounts `--gpus all` and triggers | |
| # nvidia-container-cli's host-driver-vs-CUDA version check. Passing an | |
| # already-tagged docker-image makes test-infra use it verbatim instead | |
| # of auto-suffixing it from gpu-arch-type/version. | |
| runner: linux.c7i.2xlarge | |
| docker-image: 'pytorch/manylinux2_28-builder:cuda13.2' | |
| torch-spec: '--pre torch --extra-index-url https://download.pytorch.org/whl/nightly/cu132' | |
| gpu-arch-type: cpu | |
| gpu-arch-version: '' | |
| arch-label: cuda13.2 | |
| platform: x86_64 | |
| build-arm64: | |
| uses: ./.github/workflows/build-dist.yml | |
| with: | |
| python-versions: '["3.10", "3.11", "3.12", "3.13"]' | |
| runner: linux.arm64.2xlarge | |
| docker-image: 'pytorch/manylinuxaarch64-builder:cuda13.2' | |
| torch-spec: '--pre torch --extra-index-url https://download.pytorch.org/whl/nightly/cu132' | |
| gpu-arch-type: cuda | |
| gpu-arch-version: '13.2' | |
| platform: aarch64 | |
| build-macos: | |
| uses: ./.github/workflows/build-macos.yml | |
| with: | |
| python-versions: '["3.10", "3.11", "3.12", "3.13"]' | |
| runner: macos-latest | |
| torch-spec: '--pre torch --extra-index-url https://download.pytorch.org/whl/nightly/cpu' | |
| publish-to-pypi: | |
| name: Publish to PyPI | |
| needs: [build-x86_64, build-arm64, build-macos] | |
| runs-on: ubuntu-latest | |
| environment: nightly | |
| if: ${{ github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' }} | |
| permissions: | |
| id-token: write # Required for PyPI trusted publishing | |
| contents: read | |
| steps: | |
| - name: Download all artifacts | |
| uses: actions/download-artifact@v4 | |
| with: | |
| path: dist | |
| merge-multiple: true | |
| - name: Display structure of downloaded files | |
| run: ls -R dist/ | |
| - name: Publish to PyPI | |
| uses: pypa/gh-action-pypi-publish@release/v1 | |
| with: | |
| verbose: true | |
| publish-to-container-registry: | |
| name: Publish to Github Container Registry | |
| needs: [build-x86_64, build-arm64] | |
| runs-on: ubuntu-latest | |
| environment: nightly | |
| if: ${{ github.event_name == 'schedule' || github.event_name == 'workflow_dispatch' }} | |
| permissions: | |
| contents: read | |
| packages: write # Grant write permission for packages (GHCR) | |
| steps: | |
| # This step is required to avoid the "no space left on device" error from | |
| # the large pytorch docker image. | |
| - name: Free Disk Space (Ubuntu) | |
| uses: jlumbroso/free-disk-space@v1.3.1 | |
| - name: Checkout repository | |
| uses: actions/checkout@v6 | |
| # Download wheels outside the repo so they don't interfere with the | |
| # Docker build context (dist/ is in .dockerignore). | |
| - name: Download all artifacts | |
| uses: actions/download-artifact@v4 | |
| with: | |
| # Docker container only has python 3.12, so we need to use the x86_64 3.12 wheel. | |
| name: monarch-py3.12-cuda13.2-x86_64 | |
| path: ${{ runner.temp }}/monarch_wheels | |
| merge-multiple: true | |
| - name: Log in to the Container registry | |
| uses: docker/login-action@v3 | |
| with: | |
| registry: ghcr.io | |
| username: ${{ github.actor }} | |
| password: ${{ secrets.GITHUB_TOKEN }} | |
| - name: Set up Docker Buildx | |
| uses: docker/setup-buildx-action@v3 | |
| - name: Get tag for publishing | |
| run: echo "DOCKER_TAG=0.6.0.dev$(date +'%Y%m%d')-cuda13.2" >> $GITHUB_ENV | |
| - name: Build and push Docker image | |
| uses: docker/build-push-action@v6 | |
| with: | |
| context: . # Build context is the root directory | |
| file: ./Dockerfile.nightly | |
| push: true # Push the image to the registry | |
| # Push to monarch-nightly package instead of monarch. | |
| tags: | | |
| ghcr.io/${{ github.repository }}-nightly:${{ env.DOCKER_TAG }} | |
| ghcr.io/${{ github.repository }}-nightly:latest | |
| # TODO: find docker tag that gets updated automatically. | |
| build-args: | | |
| PYTORCH_TAG=2.13.0.dev20260513-cuda13.2-cudnn9-runtime | |
| build-contexts: | | |
| monarch-wheels=${{ runner.temp }}/monarch_wheels |