pytorch/.github/workflows/torchtitan.yml at develop · ROCm/pytorch · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
# Owner(s): ["oncall: distributed"]
name: torchtitan-test

on:
  push:
    branches:
      - main
      - release/*
    tags:
      - ciflow/torchtitan/*
  workflow_dispatch:

concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
  cancel-in-progress: true

permissions:
  id-token: write
  contents: read
  actions: read

jobs:
  get-label-type:
    name: get-label-type
    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
    if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
    with:
      triggering_actor: ${{ github.triggering_actor }}
      issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
      curr_branch: ${{ github.head_ref || github.ref_name }}
      curr_ref_type: ${{ github.ref_type }}
      check_experiments: arc,lf

  # Resolve the stable CUDA version from generate_binary_build_matrix.py
  # (CUDA_STABLE) so the build/test environment tracks it without hardcoding.
  get-cuda-version:
    name: get-cuda-version
    runs-on: ubuntu-latest
    if: github.repository_owner == 'pytorch'
    outputs:
      stable-cuda: ${{ steps.get.outputs.stable-cuda }}
    steps:
      - name: Checkout PyTorch
        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
        with:
          submodules: false
          fetch-depth: 1
      - name: Get stable CUDA version
        id: get
        run: |
          set -eou pipefail
          echo "stable-cuda=$(python3 .github/scripts/get_ci_variable.py --cuda-stable-version)" >> "${GITHUB_OUTPUT}"

  build:
    name: torchtitan-x-pytorch-build
    if: github.repository_owner == 'pytorch'
    uses: ./.github/workflows/_linux-build.yml
    needs:
      - get-label-type
      - get-cuda-version
    with:
      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
      build-environment: linux-jammy-cuda${{ needs.get-cuda-version.outputs.stable-cuda }}-py3-gcc11
      docker-image-name: ci-image:pytorch-linux-jammy-cuda${{ needs.get-cuda-version.outputs.stable-cuda }}-cudnn9-py3-gcc11
      ci-docker-hash: ${{ needs.get-label-type.outputs.ci-docker-hash }}
      cuda-arch-list: '8.0 8.9 9.0 10.0'
      runner: linux.24xlarge.memory
      # torchtitan_h100_integration disabled pending runner availability
      test-matrix: |
        { include: [
          { config: "torchtitan_features_integration", shard: 1, num_shards: 1, runner: "linux.g5.48xlarge.nvidia.gpu" },
          { config: "torchtitan_models_integration", shard: 1, num_shards: 1, runner: "linux.g5.48xlarge.nvidia.gpu" }
        ]}
      use-arc: true
      python-version: "3.10"
      compiler: gcc11
      cuda-version: ${{ needs.get-cuda-version.outputs.stable-cuda }}
    secrets: inherit

  test:
    name: torchtitan-x-pytorch-test
    uses: ./.github/workflows/_linux-test.yml
    needs:
      - build
      - get-label-type
      - get-cuda-version
    with:
      build-environment: linux-jammy-cuda${{ needs.get-cuda-version.outputs.stable-cuda }}-py3-gcc11
      docker-image: ${{ needs.build.outputs.docker-image }}
      test-matrix: ${{ needs.build.outputs.test-matrix }}
      use-arc: true
      python-version: "3.10"
      compiler: gcc11
      cuda-version: ${{ needs.get-cuda-version.outputs.stable-cuda }}
    secrets: inherit