chore: Move to the Ingonyama MSM #438
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
| # Run regression check only when attempting to merge, shown as skipped status check beforehand | |
| name: GPU benchmark regression test | |
| on: | |
| pull_request: | |
| types: [opened, synchronize, reopened, ready_for_review] | |
| branches: [dev] | |
| merge_group: | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }} | |
| cancel-in-progress: true | |
| jobs: | |
| # Run comparative benchmark against dev, open issue on regression | |
| gpu-benchmark: | |
| if: github.event_name != 'pull_request' || github.event.action == 'enqueued' | |
| name: Run benchmarks on GPU | |
| runs-on: [self-hosted, gpu-bench] | |
| steps: | |
| - uses: actions/checkout@v4 | |
| with: | |
| repository: lurk-lab/ci-workflows | |
| - uses: ./.github/actions/gpu-setup | |
| with: | |
| gpu-framework: 'cuda' | |
| - uses: ./.github/actions/ci-env | |
| - uses: actions/checkout@v4 | |
| # Install dependencies | |
| - uses: dtolnay/rust-toolchain@stable | |
| - uses: Swatinem/rust-cache@v2 | |
| - uses: taiki-e/install-action@v2 | |
| with: | |
| tool: [email protected] | |
| - name: Install criterion | |
| run: | | |
| cargo install cargo-criterion | |
| cargo install criterion-table | |
| - name: Set bench output format and base SHA | |
| run: | | |
| echo "BENCH_OUTPUT=commit-comment" | tee -a $GITHUB_ENV | |
| echo "BENCH_NUM_CONS=16384,1038732" | tee -a $GITHUB_ENV | |
| echo "BASE_COMMIT=${{ github.event.merge_group.base_sha }}" | tee -a $GITHUB_ENV | |
| GPU_NAME=$(nvidia-smi --query-gpu=gpu_name --format=csv,noheader,nounits | tail -n1) | |
| echo "GPU_ID=$(echo $GPU_NAME | awk '{ print $NF }')" | tee -a $GITHUB_ENV | |
| echo "GPU_NAME=$GPU_NAME" | tee -a $GITHUB_ENV | |
| # Checkout base branch for comparative bench | |
| - uses: actions/checkout@v4 | |
| with: | |
| ref: dev | |
| path: dev | |
| # Copy the script so the base can bench with the same parameters | |
| - name: Run GPU bench on base branch | |
| run: | | |
| # Copy justfile to dev, overwriting existing config with that of PR branch | |
| cp ../benches/justfile . | |
| # Run benchmark | |
| just gpu-bench-ci supernova-ci | |
| # Copy bench output to PR branch | |
| cp supernova-ci-${{ env.BASE_COMMIT }}.json .. | |
| working-directory: ${{ github.workspace }}/dev | |
| - name: Run GPU bench on PR branch | |
| run: | | |
| just gpu-bench-ci supernova-ci | |
| cp supernova-ci-${{ github.sha }}.json .. | |
| working-directory: ${{ github.workspace }}/benches | |
| - name: copy the benchmark template and prepare it with data | |
| run: | | |
| cp .github/tables.toml . | |
| # Get CPU model | |
| CPU_MODEL=$(grep '^model name' /proc/cpuinfo | head -1 | awk -F ': ' '{ print $2 }') | |
| # Get vCPU count | |
| NUM_VCPUS=$(nproc --all) | |
| # Get total RAM in GB | |
| TOTAL_RAM=$(grep MemTotal /proc/meminfo | awk '{$2=$2/(1024^2); print int($2), "GB RAM";}') | |
| WORKFLOW_URL="https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}" | |
| # Use conditionals to ensure that only non-empty variables are inserted | |
| [[ ! -z "${{ env.GPU_NAME }}" ]] && sed -i "/^\"\"\"$/i ${{ env.GPU_NAME }}" tables.toml | |
| [[ ! -z "$CPU_MODEL" ]] && sed -i "/^\"\"\"$/i $CPU_MODEL" tables.toml | |
| [[ ! -z "$NUM_VCPUS" ]] && sed -i "/^\"\"\"$/i $NUM_VCPUS vCPUs" tables.toml | |
| [[ ! -z "$TOTAL_RAM" ]] && sed -i "/^\"\"\"$/i $TOTAL_RAM" tables.toml | |
| sed -i "/^\"\"\"$/i Workflow run: $WORKFLOW_URL" tables.toml | |
| echo "WORKFLOW_URL=$WORKFLOW_URL" | tee -a $GITHUB_ENV | |
| working-directory: ${{ github.workspace }} | |
| # Create a `criterion-table` and write in commit comment | |
| - name: Run `criterion-table` | |
| run: | | |
| cat supernova-ci-${{ env.BASE_COMMIT }}.json supernova-ci-${{ github.sha }}.json | criterion-table > BENCHMARKS.md | |
| - name: Write bench on commit comment | |
| uses: peter-evans/commit-comment@v3 | |
| with: | |
| body-path: BENCHMARKS.md | |
| # TODO: Set `$BENCH_NOISE_THRESHOLD` via `cardinalby/export-env-action` or hardcode to 1.3 | |
| # Check for a slowdown >= `$BENCH_NOISE_THRESHOLD` (fallback is 30%/1.3x). If so, open an issue but don't block merge | |
| # Since we are parsing for slowdowns, we simply add 1 to the noise threshold decimal to get the regression factor | |
| - name: Check for perf regression | |
| id: regression-check | |
| run: | | |
| REGRESSIONS=$(grep -o '[0-9.]*x slower' BENCHMARKS.md | cut -d 'x' -f1) | |
| echo $REGRESSIONS | |
| if [ ! -z "${{ env.BENCH_NOISE_THRESHOLD}}" ]; then | |
| REGRESSION_FACTOR=$(echo "${{ env.BENCH_NOISE_THRESHOLD }}+1" | bc) | |
| else | |
| REGRESSION_FACTOR=1.3 | |
| fi | |
| echo "NOISE_THRESHOLD=$(echo "($REGRESSION_FACTOR-1)*100" | bc)" | tee -a $GITHUB_ENV | |
| for r in $REGRESSIONS | |
| do | |
| if (( $(echo "$r >= $REGRESSION_FACTOR" | bc -l) )) | |
| then | |
| echo "regression=true" | tee -a $GITHUB_OUTPUT | |
| fi | |
| done | |
| # Not possible to use ${{ github.event.number }} with the `merge_group` trigger | |
| - name: Get PR number from merge branch | |
| if: steps.regression-check.outputs.regression == 'true' | |
| run: | | |
| echo "PR_NUMBER=$(echo ${{ github.event.merge_group.head_ref }} | sed -e 's/.*pr-\(.*\)-.*/\1/')" | tee -a $GITHUB_ENV | |
| - uses: JasonEtco/create-an-issue@v2 | |
| if: steps.regression-check.outputs.regression == 'true' | |
| env: | |
| GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} | |
| PR_NUMBER: ${{ env.PR_NUMBER }} | |
| GIT_SHA: ${{ github.sha }} | |
| WORKFLOW_URL: ${{ env.WORKFLOW_URL }} | |
| NOISE_THRESHOLD: ${{ env.NOISE_THRESHOLD }} | |
| with: | |
| update_existing: true | |
| filename: .github/PERF_REGRESSION.md |