Improving game performance with Android Dynamic Performance Framework
summary
ADPF claims measurable reductions in overheating and battery use while stabilizing CPU work by using the performance hint session API to report target/actual CPU durations and the thermal headroom metric (1.0 = severe throttling) to trigger workload scaling. Demos—Dr Arm and Candy Clash—run on Pixel 6/7 with Perfetto power traces showing lower big-core/GPU watts and maintained FPS versus off, and post‑process and particle scalers remove effects transparently. APIs: thermal from Android 11, hint from Android 12, Game Mode/State from Android 13+; Unity support lands in 2022.1–2023.1 via the Android provider.
AI-written summary · about
chapters (18)
- 0:00Android adaptability tools
- 13:22Unity adaptive performance
- 17:54Implementing new providers
- 20:17Technical feature integration
- 27:48The scaler system
- 29:04Scalers and profiles
- 31:38ARM case study overview
- 32:59Testing methodologies
- 35:31Dr. ARM performance data
- 36:36Visual impact analysis
- 37:24Scaler data results
- 38:01Frame rate improvements
- 38:52Power savings analysis
- 40:23Candy Clash study
- 41:39Candy Clash metrics
- 42:12Generational device testing
- 43:43Key conclusions
- 44:41Q&A session
related talks
32mOptimizing Android games performance
41mA Deep Look Inside Games: GameSDK and Unity Adaptive Performance
53mGraphics Something Android Something Performance
53mPractical Performance Tuning on Android
40mAndroid Graphics Performance
40mAndroid high performance baby
39mDeep-Diving Android Performance with Perfetto
13m12 game mode APIs