Flutter Performance: No Silver Bullet, Only Smart Engineering
summary
Claims that const gives a universal speedup get tested against two benchmarks (one shows 15% boost, the other shows none) and against microexperiments using the DevTools memory profiler that track object counts (examples go four→three instances, then to five and six when moved to top-level or exposed via constructors). A deep nested-rebuild test shows const can stop subtree rebuilds and save frame budget (60/120 fps); a UI case shows an empty button and hidden spinning indicators waste frames until fixed with repaint boundaries — but adding layers trades rasterization cost and Impeller breaks the old tool so measurement and targeted fixes win, not rules.
AI-written summary · about
chapters (11)
- 0:00Welcome
- 1:42Performance matters
- 5:19Reliable performance advice
- 8:12Const widget claims
- 12:34Measuring const behavior
- 23:58Const optimizations impact
- 27:57Framework experiments
- 35:31Debugging framework issues
- 38:45Repaint optimization tradeoffs
- 41:35Proactive measurement
- 43:44Key takeaways
related talks
31mBuilding Consistently Fast Flutter Apps
40mFluttering at 60 FPS
22mYou can go EVERYWHERE but I can go FAST - performance case study
24mPurely Fast
24mSwiftUI Performance Tips
46mFlutter: 60 Frames of AOT Badassery
49mDemystifying performance tune up in Flutter and Dart
30mFast but not furious: debugging user interaction performance issues