PMR: Fast Application Response via Parallel Memory Reclaim on Mobile Devices
summary
PMR is a kernel patch that decouples page shrinking from writeback and implements proactive page shrinking (PPS) plus storage-friendly page writeback (SPW) to batch unmaps and drive bulk I/O. Implemented and measured on Pixel 6 Pro, Pixel 5 and Redmi Note 11 with 36 real apps, it doubles reclaim throughput (previous peaks ~120 MB/s), cuts app response latency up to 43.6%, and further reduces latency 38% when combined with FLEET.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:01Talk structure and background
- 2:33Android memory reclaim methods
- 4:22Low memory killer problems
- 6:12Bottlenecks in memory reclaim
- 8:02Page shrinking inefficiencies
- 9:26Memory reclaim challenges
- 9:48PMR parallel reclaim scheme
- 10:32PMR design details
- 11:33Experimental results
- 12:59Sensitivity study
- 13:19Conclusion
