Skip to content
Latchkey

test-ipu workflow (datamol-io/graphium)

The test-ipu workflow from datamol-io/graphium, explained and optimized by Latchkey.

B

CI health: B - good

Run this on Latchkey for self-healing, caching, and up to 58% lower cost.

Grade your own workflow free or run it on Latchkey →
Source: datamol-io/graphium.github/workflows/test_ipu.ymlLicense Apache-2.0View source

What it does

This is the test-ipu workflow from the datamol-io/graphium repository, a real project running GitHub Actions. It is shown here with attribution under its Apache-2.0 license.

Below, Latchkey shows a faster, safer version produced by its optimization engine.

The workflow

workflow (.yml)
name: test-ipu

on:
  push:
    branches: ["main"]
    tags: ["*"]
  pull_request:
    branches:
      - "*"
      - "!gh-pages"
  schedule:
    - cron: "0 4 * * *"

jobs:
  test-ipu:
    strategy:
      fail-fast: false
      matrix:
        python-version: ["3.8"]
        pytorch-version: ["2.0"]

    runs-on: "ubuntu-20.04"
    timeout-minutes: 30

    defaults:
      run:
        shell: bash -l {0}

    name: |
        poptorch_env - 
        python=${{ matrix.python-version }} -
        pytorch=${{ matrix.pytorch-version }}

    steps:
      - name: Checkout the code
        uses: actions/checkout@v3

      - name: Activate SDK + Install Requirements
        run: |
          python3 -m pip install --upgrade pip
          wget -q -O 'poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz' 'https://downloads.graphcore.ai/direct?package=poplar-poplar_sdk_ubuntu_20_04_3.3.0_208993bbb7-3.3.0&file=poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz'
          tar -xzf poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz
          python3 -m pip install poplar_sdk-ubuntu_20_04-3.3.0+1403-208993bbb7/poptorch-3.3.0+113432_960e9c294b_ubuntu_20_04-cp38-cp38-linux_x86_64.whl
          # Enable Poplar SDK (including Poplar and PopART)
          source poplar_sdk-ubuntu_20_04-3.3.0+1403-208993bbb7/enable 
          
          python -c "import poptorch"

          # Download the datafiles (Total ~ 10Mb - nothing compared to the libraries)
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/ZINC12k.csv.gz
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/Tox21-7k-12-labels.csv.gz
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/qm9.csv.gz


          # Install the IPU specific and graphium requirements
          pip install -r requirements_ipu.txt
          # Install Graphium in dev mode
          python -m pip install --no-deps -e .
          python3 -m pytest -m 'not skip_ipu'

      - name: Codecov Upload
        uses: codecov/codecov-action@v3
        with:
          files: ./coverage.xml
          flags: unittests
          name: codecov-umbrella
          fail_ci_if_error: false
          verbose: false
          env_vars: ${{ matrix.python-version }},${{ matrix.pytorch-version }}

The same workflow, on Latchkey

Removes redundant runs and caps runaway jobs. Added and changed lines are highlighted.

name: test-ipu
 
on:
  push:
    branches: ["main"]
    tags: ["*"]
  pull_request:
    branches:
      - "*"
      - "!gh-pages"
  schedule:
    - cron: "0 4 * * *"
 
concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true
 
jobs:
  test-ipu:
    strategy:
      fail-fast: false
      matrix:
        python-version: ["3.8"]
        pytorch-version: ["2.0"]
 
    runs-on: "ubuntu-20.04"
    timeout-minutes: 30
 
    defaults:
      run:
        shell: bash -l {0}
 
    name: |
        poptorch_env - 
        python=${{ matrix.python-version }} -
        pytorch=${{ matrix.pytorch-version }}
 
    steps:
      - name: Checkout the code
        uses: actions/checkout@v3
 
      - name: Activate SDK + Install Requirements
        run: |
          python3 -m pip install --upgrade pip
          wget -q -O 'poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz' 'https://downloads.graphcore.ai/direct?package=poplar-poplar_sdk_ubuntu_20_04_3.3.0_208993bbb7-3.3.0&file=poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz'
          tar -xzf poplar_sdk-ubuntu_20_04-3.3.0-208993bbb7.tar.gz
          python3 -m pip install poplar_sdk-ubuntu_20_04-3.3.0+1403-208993bbb7/poptorch-3.3.0+113432_960e9c294b_ubuntu_20_04-cp38-cp38-linux_x86_64.whl
          # Enable Poplar SDK (including Poplar and PopART)
          source poplar_sdk-ubuntu_20_04-3.3.0+1403-208993bbb7/enable 
          
          python -c "import poptorch"
 
          # Download the datafiles (Total ~ 10Mb - nothing compared to the libraries)
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/ZINC12k.csv.gz
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/Tox21-7k-12-labels.csv.gz
          wget https://storage.valencelabs.com/graphium/datasets/neurips_2023/Small-dataset/qm9.csv.gz
 
 
          # Install the IPU specific and graphium requirements
          pip install -r requirements_ipu.txt
          # Install Graphium in dev mode
          python -m pip install --no-deps -e .
          python3 -m pytest -m 'not skip_ipu'
 
      - name: Codecov Upload
        uses: codecov/codecov-action@v3
        with:
          files: ./coverage.xml
          flags: unittests
          name: codecov-umbrella
          fail_ci_if_error: false
          verbose: false
          env_vars: ${{ matrix.python-version }},${{ matrix.pytorch-version }}
 

What changed

1 third-party action is referenced by a movable tag. Pin it to the commit SHA (Latchkey resolves and applies this automatically) so a repointed tag cannot change what runs.

What Latchkey heals here

This workflow has steps that commonly fail on transient issues (network, registries, flaky browsers). On Latchkey managed runners they are detected, retried, and self-healed instead of failing your build:

This workflow runs 1 job per trigger. On Latchkey the same minutes cost up to 58% less than GitHub-hosted, with zero queue time.

Actions used in this workflow