lightning/.azure/hpu-tests.yml

# Pipeline to run the HPU tests in DL1 Instance

trigger:
  tags:
    include:
      - '*'
  branches:
    include:
      - "master"
      - "release/*"
      - "refs/tags/*"
  paths:
    include:
      - ".azure/hpu-tests.yml"
      - "examples/pl_hpu/mnist_sample.py"
      - "requirements/pytorch/**"
      - "src/pytorch_lightning/**"
      - "tests/tests_pytorch/**"

pr:
  branches:
    include:
      - "master"
      - "release/*"
  paths:
    include:
      - ".azure/hpu-tests.yml"
      - "examples/pl_hpu/mnist_sample.py"
      - "requirements/pytorch/**"
      - "src/pytorch_lightning/**"
      - "tests/tests_pytorch/**"

jobs:
  - job: testing
    # how long to run the job before automatically cancelling
    timeoutInMinutes: "10"
    # how much time to give 'run always even if cancelled tasks' before stopping them
    cancelTimeoutInMinutes: "2"
    pool: habana-gaudi-hpus
    container:
      image: "vault.habana.ai/gaudi-docker/1.5.0/ubuntu20.04/habanalabs/pytorch-installer-1.11.0:latest"
      options: "--runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host --shm-size=4g --name cd-container -v /usr/bin/docker:/tmp/docker:ro"
    workspace:
      clean: all

    steps:
    - script: |
        /tmp/docker exec -t -u 0 cd-container \
        sh -c "apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -o Dpkg::Options::="--force-confold" -y install sudo"
      displayName: 'Install Sudo in container (thanks Microsoft!)'

    - bash: |
        sudo apt-get install -y hwinfo
        hwinfo --short
        python --version
        sudo pip install pip -U
      displayName: 'Instance HW info'

    - bash: |
        set -e
        pip --version
        sudo pip uninstall -y lightning pytorch-lightning
        pip install -q -r .actions/requirements.txt
        python .actions/assistant.py requirements-prune-pkgs torch,torchvision
        pip install ".[extra,test]"
        pip list
      env:
        PACKAGE_NAME: pytorch
        FREEZE_REQUIREMENTS: 1
      displayName: 'Install dependencies'

    - bash: |
        hl-smi -L
        lsmod | grep habanalabs
      displayName: 'Check the driver status'

    - bash: |
        python -m pytest -sv accelerators/test_hpu.py --forked --junitxml=hpu1_test-results.xml
      workingDirectory: tests/tests_pytorch
      displayName: 'Single card HPU test'

    - bash: |
        python -m pytest -sv accelerators/test_hpu.py --forked --hpus 8 --junitxml=hpu8_test-results.xml
      workingDirectory: tests/tests_pytorch
      displayName: 'Multi card(8) HPU test'

    - bash: |
        python -m pytest -sv plugins/precision/hpu/test_hpu.py --hmp-bf16 \
          'plugins/precision/hpu/ops_bf16.txt' --hmp-fp32 \
          'plugins/precision/hpu/ops_fp32.txt' --forked \
          --junitxml=hpu1_precision_test-results.xml
      workingDirectory: tests/tests_pytorch
      displayName: 'HPU precision test'

    - bash: |
        export PYTHONPATH="${PYTHONPATH}:$(pwd)"
        python "pl_hpu/mnist_sample.py"
      workingDirectory: examples
      displayName: 'Testing: HPU examples'

    - task: PublishTestResults@2
      inputs:
        testResultsFiles: 'tests/tests_pytorch/hpu*_test-results.xml'
        testRunTitle: '$(Agent.OS) - $(Build.DefinitionName) - Python $(python.version)'
      condition: succeededOrFailed()
      displayName: 'Publish test results'