Runtime Scheduling on the Web
summary
Chrome team claims a native scheduler can reduce main thread congestion by giving JS tasks explicit priorities and signals so apps meet latency targets like 10 ms for input and 16 ms per frame instead of long script runs that hit ~600 ms median blocking. Prototypes include inputPending signals, a postTask API with priority queues, and priority hints for resource/script loading. Work with Maps and React validates the approach but microtask behavior and cross‑library coordination remain unresolved and origin trials are ongoing.
AI-written summary · about
chapters (36)
- 0:00Introduction
- 0:55Demo
- 1:23Profile
- 1:55Real
- 2:53Data Analysis
- 3:49Input Latency
- 6:03Script Learning
- 6:26Summary
- 9:47Deadlines
- 10:35Priorities
- 10:48Responsiveness
- 11:25Chunking
- 11:40Frameworks
- 12:27Google Maps Demo
- 13:13Browser Rendering Pipeline
- 14:05Browser Rendering Pipeline Categories
- 15:36Key Insights
- 16:01Existing Priorities
- 17:20Callouts
- 17:47Scheduling
- 19:12Map Scheduler
- 21:17LowLevel APIs
- 22:10Coordination Problem
- 23:17Scheduling Down
- 23:47Benefits of Scheduling
- 24:34Scheduling Priorities
- 26:04Task Queues
- 27:00Sample Code
- 27:25Challenges
- 28:22Priority Hints
- 29:32MicroTasks
- 31:12Continuation Support
- 32:37Layout thrashing
- 33:58Current plan
- 34:57Feedback
- 36:04Credits Shoutouts
related talks
33mA Quest to Guarantee Responsiveness: Scheduling On and Off the Main Thread
24mChrome ❤️ ⚛️ (no, really)
42mElevating the Web Platform with the JavaScript Framework Community
27mDeep diving on Concurrent React
29mDeep Diving on Concurrent React
59mWeb performance APIs you (probably) didn't know existed
39mThe best is yet to come: the Future of React
01h 03mSpeed up your Web Application