Add nightly benchmarking on Triton pytorch and triton-main versions #102
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
name: TritonBench PR Test | |
on: | |
pull_request: | |
push: | |
branches: | |
- main | |
jobs: | |
h100-pytorch-test: | |
# Don't run on forked repos | |
if: github.repository_owner == 'pytorch-labs' | |
runs-on: [gcp-h100-runner] | |
timeout-minutes: 240 | |
environment: docker-s3-upload | |
env: | |
CONDA_ENV: "pytorch" | |
SETUP_SCRIPT: "/workspace/setup_instance.sh" | |
steps: | |
- name: Checkout Tritonbench | |
uses: actions/checkout@v3 | |
with: | |
# no need to checkout submodules recursively | |
submodules: true | |
- name: Tune Nvidia GPU | |
run: | | |
sudo nvidia-smi -pm 1 | |
sudo ldconfig | |
nvidia-smi | |
- name: Test Tritonbench operators on H100 GPU | |
run: | | |
bash ./.ci/tritonbench/test-gpu.sh | |
concurrency: | |
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}-${{ github.event_name == 'workflow_dispatch' }} | |
cancel-in-progress: true |