Aravind Ravikumar
69 строки
2.3 KiB
YAML
69 строки
2.3 KiB
YAML
name: TheRock Test Packages multi-node
|
|
|
|
on:
|
|
workflow_call:
|
|
inputs:
|
|
amdgpu_families:
|
|
type: string
|
|
test_runs_on:
|
|
type: string
|
|
artifact_run_id:
|
|
type: string
|
|
workflow_dispatch:
|
|
inputs:
|
|
amdgpu_families:
|
|
type: string
|
|
test_runs_on:
|
|
type: string
|
|
artifact_run_id:
|
|
type: string
|
|
|
|
permissions:
|
|
contents: read
|
|
|
|
jobs:
|
|
test_rccl_multi_node:
|
|
name: 'Test multi-node'
|
|
runs-on: ${{ inputs.test_runs_on }}
|
|
defaults:
|
|
run:
|
|
shell: bash
|
|
env:
|
|
VENV_DIR: ${{ github.workspace }}/.venv
|
|
ARTIFACT_RUN_ID: "${{ inputs.artifact_run_id }}"
|
|
OUTPUT_ARTIFACTS_DIR: /home/arravikum/dist_new/dist/rocm
|
|
THEROCK_BIN_DIR: "./build/bin"
|
|
steps:
|
|
- name: Checkout Repository
|
|
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
|
|
with:
|
|
repository: "ROCm/TheRock"
|
|
ref: 890c856134d955441790c8ed2d60ad4fb027f4e5
|
|
|
|
- name: Run setup test environment workflow
|
|
uses: './.github/actions/setup_test_environment'
|
|
with:
|
|
ARTIFACT_RUN_ID: ${{ env.ARTIFACT_RUN_ID }}
|
|
AMDGPU_FAMILIES: ${{ inputs.amdgpu_families }}
|
|
OUTPUT_ARTIFACTS_DIR: ${{ env.OUTPUT_ARTIFACTS_DIR }}
|
|
VENV_DIR: ${{ env.VENV_DIR }}
|
|
FETCH_ARTIFACT_ARGS: "--rccl"
|
|
IS_PR_FROM_FORK: ${{ github.event.pull_request.head.repo.fork }}
|
|
|
|
# The following step leverages slurm to run multi node rccl tests on the slurm mi350x cluster.
|
|
# salloc will hold 4 nodes while the commands inside the block run. After the block completes, salloc automatically releases the nodes.
|
|
- name: Test gfx950
|
|
if: ${{ inputs.amdgpu_families == 'gfx950-dcgpu' }}
|
|
run: |
|
|
salloc -N 4 -p meta64 -t 04:00:00 --exclusive bash -c "
|
|
source /home/arravikum/TheRock/.venv/bin/activate &&
|
|
cd /home/arravikum/cvs &&
|
|
python input/setup.py &&
|
|
pytest -vvv -s ./tests/rccl/rccl_multinode_cvs.py \
|
|
--cluster_file ./input/cluster.json \
|
|
--config_file ./input/mi350_config.json \
|
|
--log-file=/tmp/rccl_log.log \
|
|
--html=/home/arravikum/cvs/ci_test_report.html \
|
|
--capture=tee-sys \
|
|
--self-contained-html"
|