Performance Tuning Twitter Services With Machine Learning, Part 2 JIT and GC
summary
Twitter replaces C2 with Graal JIT and uses Bayesian optimization to tune three inlining parameters, cutting allocated bytes per tweet about 7% less memory allocated, lowering p99 latency roughly 28% (20% from Graal, +8% from ML) and reducing CPU use ~18%, which equates to 18% fewer machines. Google uses ML to reconfigure cooling and reports ~40% cooling energy reduction and ~50% overall energy savings in pilots.
AI-written summary · about
chapters (12)
- 0:00Intro
- 2:00Who am I
- 4:40Sprints
- 7:16Compiler Optimization
- 8:45Electricity Usage
- 10:37The Good
- 14:27The Bad
- 15:55Machine Learning
- 17:35Twitter Example
- 20:33Bitcoin Mining
- 24:47Bitcoin History
- 29:49Twitter Light
related talks
54mPerformance tuning Twitter services with Graal and Machine Learning
56mPerformance tuning Twitter services with Graal and Machine Learning
53mTwitter's quest for a wholly Graal runtime
50mTwitter's quest for a wholly Graal runtime
54mTwitter's quest for a wholly Graal runtime
41mTwitter’s Quest for a Wholly Graal Runtime
40mScala ♥ Graal
41mGraal: How to use the new JVM JIT compiler in real life