diff --git a/.github/workflows/_ascend_npu_torchtitan.yml b/.github/workflows/_ascend_npu_torchtitan.yml new file mode 100644 index 0000000..5ae9e21 --- /dev/null +++ b/.github/workflows/_ascend_npu_torchtitan.yml @@ -0,0 +1,126 @@ +name: "_ascend_npu_torchtitan" + +on: + workflow_call: + inputs: + runner: + required: true + type: string + description: "The runner selected to run on" + image: + required: true + type: string + description: "The docker image which will be loaded" + torch-artifact: + required: false + type: string + description: "The distribution artifact name of torch" + torch-npu-artifact: + required: true + type: string + description: "The distribution artifact name of torch_npu" + +# Bash shells do not use ~/.profile or ~/.bashrc so these shells need to be explicitly +# declared as "shell: bash -el {0}" on steps that need to be properly activated. +# It's used to activate ascend-toolkit environment variables. + +jobs: + setup_environment: + name: run torchtitan tests + runs-on: ${{ inputs.runner }} + container: + image: ${{ inputs.image }} + env: + HF_ENDPOINT: https://hf-mirror.com + outputs: + torch_version: ${{ steps.get_torch_version.outputs.torch-version }} + npu_info: ${{ steps.check_npu.outputs.npu_info }} + steps: + - name: Show NPU info + run: | + npu-smi info + + - name: Config mirrors + run: | + sed -i 's|ports.ubuntu.com|mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list + pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple + + - name: Install system dependencies + run: | + apt-get update + apt-get install -y \ + git gcc g++ make cmake ninja-build curl \ + libgl1 libglib2.0-0 libsndfile1 + + - name: Config git + run: | + git config --global --add safe.directory "$GITHUB_WORKSPACE" + git config --global url."https://gh-proxy.test.osinfra.cn/https://github.com/".insteadOf https://github.com/ + + - name: Checkout + uses: actions/checkout@v4 + + - name: Checkout benchmark + uses: actions/checkout@v4 + with: + repository: pytorch/torchtitan + path: torchtitan + + - name: Download torch artifact + if: ${{ inputs.torch-artifact }} + uses: actions/download-artifact@v4 + with: + name: ${{ inputs.torch-artifact }} + + - name: Install torch + if: ${{ inputs.torch-artifact }} + run: | + pip install ${{ inputs.torch-artifact }} + + - name: Install torch_npu dependencies + if: ${{ !inputs.torch-artifact }} + run: | + pip install -r https://raw.githubusercontent.com/Ascend/pytorch/refs/heads/master/requirements.txt + + - name: List torch version + id: list-torch-version + shell: bash + run: | + torch_version=$(python -c "import torch; print(torch.__version__)") + echo "torch-version=${torch_version}" >> $GITHUB_OUTPUT + + - name: Download torch_npu artifact + uses: actions/download-artifact@v4 + with: + name: ${{ inputs.torch-npu-artifact }} + path: ascend_npu + + - name: Install torch_npu + working-directory: ascend_npu + run: | + pip install ${{ inputs.torch-npu-artifact }} + + - name: Install project dependencies + run: | + pip install pytest pytest-cov tyro tabulate + + - name: Show environment info + id: check_npu + run: | + npu_is_available=$(python -c "import torch; print(torch.npu.is_available())") + npu_count=$(python -c "import torch; print(torch.npu.device_count())") + echo "npu_count=${npu_count}" >> $GITHUB_OUTPUT + echo "NPU is available: ${npu_is_available}" + echo "NPU count: ${npu_count}" + pip list | grep -E 'torch|numpy' + + - name: Run torchtitan integration_test + working-directory: torchtitan + run: | + mkdir artifacts-to-be-uploaded + python -m tests.integration_tests.run_tests --test_name artifacts-to-be-uploaded --ngpu ${{ steps.check_npu.outputs.npu_count }} || true + + - name: Run torchtitan unittest + working-directory: torchtitan· + run: | + pytest tests/unit_tests --cov=. --cov-report=xml --durations=20 -vv diff --git a/.github/workflows/ascend_npu_test.yml b/.github/workflows/ascend_npu_test.yml index c9d5af2..732214a 100644 --- a/.github/workflows/ascend_npu_test.yml +++ b/.github/workflows/ascend_npu_test.yml @@ -168,3 +168,19 @@ jobs: torch-npu-artifact: ${{ needs.build.outputs.torch-npu-artifact }} secrets: hf-token: ${{ secrets.HF_TOKEN }} + + torchtitan: + name: Run torchtitan tests + needs: + - prepare + - build-torch + - build + if: | + !cancelled() && github.event_name != 'repository_dispatch' && + (success() || (needs.build-torch.result == 'skipped' && needs.build.result == 'success')) + uses: ./.github/workflows/_ascend_npu_torchtitan.yml + with: + runner: ${{ needs.prepare.outputs.runner }} + image: ${{ needs.prepare.outputs.image }} + torch-artifact: ${{ needs.build-torch.outputs.torch-artifact }} + torch-npu-artifact: ${{ needs.build.outputs.torch-npu-artifact }}