Android Benchmarking and other stories

on YouTube ↗33m

" summary

Covers using Android Profiler plus Google libraries microbenchmark and macrobenchmark to measure CPU, frames and startup times (examples: sorting 10,000 items, RecyclerView scrolling), and the new in‑app junkstats to flag slow frames; benchmarks emit JSON with min/median/max (examples show runs around 651–749 ms variance) and require real devices or Firebase Test Lab. CI workflow installs APKs, runs benchmarks, stores build→time pairs, and detects regressions with a step‑fitting algorithm (sliding window) instead of naïve thresholds.

AI-written summary · about

" chapters (11)

1. 0:00Introduction and speakers
2. 1:17What is benchmarking
3. 3:00Android Profiler overview
4. 5:49Microbenchmark framework
5. 9:21Macrobenchmark framework
6. 13:18JankStats framework
7. 16:48Detecting regressions in CI
8. 20:01App startup time example
9. 25:46Step fitting algorithm
10. 29:18Recap and resources
11. 31:03Q&A and closing

" related talks

• 33mAndroid Benchmarking and other storiesEnrique López Mañas, Iury Souza · droidcon San Francisco 2022
• 40mMonitor App performance using Jetpack BenchmarkSagar Das · droidcon New York 2023
• 20mMeasure What Matters with Jetpack Benchmark (Lightning)JB Lorenzo · droidCon Berlin 2019
• 37mImproving app performance with benchmarkingDustin Lam, Chris Craik · Google I/O 2019
• 34mPerformance in a Kotlin worldSean McQuillan, Dan Galpin · droidcon NYC 2019
• 43mExperiments for your Android builds driven by Gradle ProfilerUbiratan Soares · Android Summit 2020
• 29mWhat's new in app performanceBen Weiss, Lokesh Gidra · Google I/O 2022
• 40mMy Journey: Adding Baseline Profiles to a Production-Based Android App: Success or Failure?Ahmed Tikiwa · droidcon Berlin 2023