(Single-card) Device perf regressions #2288
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: "Device perf regressions and output report" | |
| on: | |
| workflow_dispatch: | |
| schedule: | |
| - cron: "0 2,7,10,14,17,20,23 * * *" | |
| workflow_call: | |
| jobs: | |
| build-artifact-profiler: | |
| uses: ./.github/workflows/build-artifact.yaml | |
| with: | |
| tracy: true | |
| secrets: inherit | |
| device-perf: | |
| needs: build-artifact-profiler | |
| strategy: | |
| # Do not fail-fast because we need to ensure all tests go to completion | |
| # so we try not to get hanging machines | |
| fail-fast: false | |
| matrix: | |
| test-info: [ | |
| {name: "GS", arch: grayskull, runs-on: ["perf-no-reset-grayskull", "self-reset"], machine-type: "bare_metal", timeout: 40}, | |
| {name: "N300 WH B0", arch: wormhole_b0, runs-on: ["perf-wormhole_b0", "self-reset"], machine-type: "bare_metal", timeout: 30}, | |
| ] | |
| name: "${{ matrix.test-info.name }} device perf" | |
| env: | |
| TT_METAL_ENV: ${{ vars.TT_METAL_ENV }} | |
| ARCH_NAME: ${{ matrix.test-info.arch }} | |
| LOGURU_LEVEL: INFO | |
| LD_LIBRARY_PATH: ${{ github.workspace }}/build/lib | |
| environment: dev | |
| runs-on: ${{ matrix.test-info.runs-on }} | |
| steps: | |
| - uses: tenstorrent-metal/metal-workflows/.github/actions/[email protected] | |
| - name: Ensure weka mount is active | |
| run: | | |
| sudo systemctl restart mnt-MLPerf.mount | |
| sudo /etc/rc.local | |
| ls -al /mnt/MLPerf/bit_error_tests | |
| - name: Set up dynamic env vars for build | |
| run: | | |
| echo "TT_METAL_HOME=$(pwd)" >> $GITHUB_ENV | |
| - uses: actions/download-artifact@v4 | |
| with: | |
| name: TTMetal_build_${{ matrix.test-info.arch }}_profiler | |
| - name: Extract files | |
| run: tar -xvf ttm_${{ matrix.test-info.arch }}.tar | |
| - uses: ./.github/actions/install-python-deps | |
| - name: Run device performance regressions | |
| timeout-minutes: ${{ matrix.test-info.timeout }} | |
| run: | | |
| source python_env/bin/activate | |
| ./tests/scripts/run_tests.sh --tt-arch $ARCH_NAME --pipeline-type models_device_performance_${{ matrix.test-info.machine-type }} | |
| - name: Check device perf report exists | |
| id: check-device-perf-report | |
| if: ${{ !cancelled() }} | |
| run: | | |
| ls -hal | |
| export DEVICE_PERF_REPORT_FILENAME=Models_Device_Perf_$(date +%Y_%m_%d).csv | |
| ls -hal $DEVICE_PERF_REPORT_FILENAME | |
| echo "device_perf_report_filename=$DEVICE_PERF_REPORT_FILENAME" >> "$GITHUB_OUTPUT" | |
| - name: Upload device perf report | |
| if: ${{ !cancelled() && steps.check-device-perf-report.conclusion == 'success' }} | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: device-perf-report-csv-${{ matrix.test-info.arch }}-${{ matrix.test-info.machine-type }} | |
| path: "${{ steps.check-device-perf-report.outputs.device_perf_report_filename }}" |