Skip to content

Latest commit

 

History

History
62 lines (48 loc) · 2.33 KB

README.md

File metadata and controls

62 lines (48 loc) · 2.33 KB

Run Benchmarks ⏱️

A Github Action that runs your Android microbenchmarks in Firebase Test Labs and optionally:

See the action in action (ha ha!) at https://github.com/martinbonnin/run-benchmarks-sample

Note: Macrobenchmarks are not supported just yet

Configuration

on:
  schedule:
    # Run every night
    - cron: '0 3 * * *'

jobs:
  benchmarks:
    runs-on: ubuntu-20.04
    steps:
      - uses: actions/checkout@v3
      - uses: actions/setup-java@v3
        with:
          distribution: 'temurin'
          java-version: '11'
      - uses: gradle/[email protected]
      - run: |
          # Build the benchmark apks
          ./gradlew :microbenchmark:packageReleaseAndroidTest :app:assembleRelease 
      - name: microbenchmarks
        uses: martinbonnin/run-benchmarks@main
        with:
          google_services_json: ${{ secrets.GOOGLE_SERVICES_JSON }}
          
          app_apk: 'app/build/outputs/apk/release/app-release.apk'
          test_apk: 'microbenchmark/build/outputs/apk/androidTest/release/microbenchmark-release-androidTest.apk'
          device_model: 'redfin,locale=en,orientation=portrait' 
          
          # Optional, upload to datadog
          dd_api_key: ${{ secrets.DD_API_KEY }}
          dd_metric_prefix: 'android.benchmark'

          # Optional, create a dashboard issue that publishes the latest results
          github_token: ${{ github.token }}
          dd_dashboard_url: 'https://p.datadoghq.com/sb/'

Datadog integration

If you specify dd_api_key, the metrics are uploaded to Datadog automatically. This action publishes 2 metrics:

  • ${dd_metric_prefix}.allocs: the median number of allocations per test
  • ${dd_metric_prefix}.nanos: the median number of nanosecond per test

Each metric is tagged with the class of the test and the name of the test. For an example:

  • class:com.example.BenchmarkTest
  • test:parseSomething