Demystifying Page Load Performance with WProf
summary
Built a lightweight in‑browser profiler called WProf (WebKit extension in Chrome/Safari, C++) that extracts activity timings and a detailed dependency graph and runs critical‑path analysis on 350 popular pages. Key findings: 35% computation of critical‑path time for the median page; synchronous JavaScript blocks parsing on ~60% of top pages; caching removes 80% of objects but only 40% of critical‑path work so speedups are smaller; SPDY helps mainly at high RTTs and mod_pagespeed yields little average benefit.
AI-written summary · about
chapters (13)
- 0:00Demystifying Page Load Performance with WProf
- 0:21Web is the critical part of the Internet
- 1:05Many techniques aim to optimize page load time • Optimization techniques
- 1:45Difficult to understand page load
- 4:37How a page is loaded
- 6:21How to infer dependencies
- 6:58Reverse engineer page loads with test pages
- 9:13Dependency policy categories • Flow dependency
- 10:35Output dependency
- 12:13Dependency policies across browsers
- 13:01WProf architecture Browser Stack
- 15:16Critical path analysis
- 19:08Conclusion
related talks
48mThink Fast, First: Empowering a Performance Culture
29mHigh Performance in the Critical Rendering Path
55mI Feel the Need for Web Page Speed
18mWebsite performance from user perspective
59mProfiling JavaScript Like A Pro
33mRedefining Web Performance
33mDemystifying Web Performance Tooling
39mUnderstanding Page Load