fast-dispatch-full-regressions-and-models.yaml

name: Nightly fast dispatch tests

on:
  workflow_dispatch:
  workflow_call:
  schedule:
    - cron: "0 */2 * * *"

jobs:
  build-artifact:
    uses: ./.github/workflows/build-artifact.yaml
    secrets: inherit
  fd-nightly:
    needs: build-artifact
    strategy:
      # Do not fail-fast because we need to ensure all tests go to completion
      # so we try not to get hanging machines
      fail-fast: false
      matrix:
        test-group:
          [
            { name: "Common models GS", arch: grayskull, cmd: tests/scripts/single_card/nightly/run_common_models.sh, timeout: 40 },
            { name: "Common models N300 WH B0", arch: wormhole_b0, cmd: tests/scripts/single_card/nightly/run_common_models.sh, timeout: 40 },
            { name: "GS-only ttnn nightly", arch: grayskull, cmd: tests/scripts/single_card/nightly/run_ttnn.sh, timeout: 40 },
            { name: "GS-only models", arch: grayskull, cmd: tests/scripts/single_card/nightly/run_gs_only.sh, timeout: 40 },
            { name: "N300 WH-only models", arch: wormhole_b0, cmd: tests/scripts/single_card/nightly/run_wh_b0_only.sh, timeout: 50 },
            { name: "API tests GS", arch: grayskull, cmd: ./tests/scripts/run_tests.sh --tt-arch grayskull --pipeline-type frequent_api --dispatch-mode fast, timeout: 40 },
            { name: "API tests N300 WH B0", arch: wormhole_b0, cmd: ./tests/scripts/run_tests.sh --tt-arch wormhole_b0 --pipeline-type frequent_api --dispatch-mode fast, timeout: 40 },
            # #9945: Skip SD for now
            # { name: "[Unstable] N300 models", arch: wormhole_b0, cmd: tests/scripts/single_card/nightly/run_wh_b0_unstable.sh, timeout: 45 },
          ]
    name: FD ${{ matrix.test-group.name }} ${{ matrix.test-group.arch }}
    env:
      TT_METAL_ENV: ${{ vars.TT_METAL_ENV }}
      ARCH_NAME: ${{ matrix.test-group.arch }}
      LOGURU_LEVEL: INFO
      LD_LIBRARY_PATH: ${{ github.workspace }}/build/lib
    runs-on: model-runner-${{ matrix.test-group.arch }}
    steps:
      - uses: tenstorrent-metal/metal-workflows/.github/actions/checkout-with-submodule-lfs@v2.0.0
      - name: Ensure weka mount is active
        timeout-minutes: 3
        run: |
          sudo systemctl restart mnt-MLPerf.mount
          sudo /etc/rc.local
          ls -al /mnt/MLPerf/bit_error_tests
      - name: Set up dyanmic env vars for build
        run: |
          echo "TT_METAL_HOME=$(pwd)" >> $GITHUB_ENV
      - uses: actions/download-artifact@v4
        with:
          name: TTMetal_build_${{ matrix.test-group.arch }}
      - name: Extract files
        run: tar -xvf ttm_${{ matrix.test-group.arch }}.tar
      - uses: ./.github/actions/install-python-deps
      - name: Run frequent reg tests scripts
        timeout-minutes: ${{ matrix.test-group.timeout }}
        run: |
          source ${{ github.workspace }}/python_env/bin/activate
          cd $TT_METAL_HOME
          export PYTHONPATH=$TT_METAL_HOME
          ${{ matrix.test-group.cmd }}