Performance tuning Twitter services with Graal and Machine Learning
summary
Running Graal as a drop-in JIT for HotSpot cuts CPU on a heavy Scala finagle tweet service by about 11% vs C2, then automated Bayesian tuning with Autotune (Bayesian optimization) finds another ~8% (verification shows up to 18% total instance reduction), with a social-graph service showing ~5.5% from Graal and ~7.8% extra from Autotune (≈13% total). Autotune is a wetlab/spearmint-based service that runs short dedicated-machine evaluations (30 min slices) and also tunes C2 where it finds ~5% best gains; Graal JIT replacing C2.
AI-written summary · about
chapters (12)
- 0:00Start
- 5:54PERFORMANCE OPTIMIZATION
- 7:33PERFORMANCE TUNING
- 9:07OPTIMIZATION OF A BLACK BOX FUNCTION
- 13:21WE LIKE BAYESIAN OPTIMIZATION...
- 14:55WHAT IS AUTOTUNE?
- 16:48WHAT IS GRAAL?
- 18:24WHICH PARAMETERS?
- 25:34TEST SETUP
- 29:16TWEET SERVICE: OBJECTIVE
- 30:00TWEET SERVICE: CONSTRAINTS
- 39:32SOCIAL GRAPH: CONSTRAINTS
related talks
46mPerformance-Tuning Twitter Services with Graal and Machine Learning
34mPerformance tuning Twitter services with Graal and Machine Learning
54mPerformance tuning Twitter services with Graal and Machine Learning
49mPerformance tuning Twitter services with Graal and Machine Learning
34mPerformance Tuning Twitter Services With Machine Learning, Part 2 JIT and GC
54mTwitter's quest for a wholly Graal runtime
53mTwitter's quest for a wholly Graal runtime
50mTwitter's quest for a wholly Graal runtime