Fast Application Launch on Personal Computing/Communication Devices
summary
Paralfetch is a Linux kernel prefetcher implemented on desktop/laptop, Raspberry Pi 3 and an Android smartphone that collects accurate tracing of launch-related reads, pre-schedules them with range merge and metadata shift to raise SSD QD and merges IO, and overlaps app execution with prefetching. Evaluation shows significant app-launch reductions versus GSoC Prefetch and FAST (FAST 2011), with negligible CPU/memory cost and results approaching warm-start performance—within about three percentage points on a laptop (Paralfetch 54% vs warm 51%).
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 2:20Launch scenarios and goals
- 3:32Parafetch architecture
- 5:42Implementation details
- 7:31Tracing and comparison
- 10:48Pre-scheduling techniques
- 12:55Optimizing pre-scheduling
- 15:08Performance evaluation
- 17:29Conclusion
related talks
41mA Solution to High Latencies Caused by I/O
44mPredictive Prefetching
11mMaking Performance Personal at FT Labs
12mHow to speed up Largest Contentful Paint (LCP) with cross-site prefetching
39mDeep-Diving Android Performance with Perfetto
27mpreconnect, prefetch, preload, pre what? An intro to Resource Hints
39mAndroid App Performance in a Nutshell
29mWhat's new in app performance