Project Leyden Update
summary
Project Leyden ships an Early Access AOT pipeline that records training runs, assembles an AOT cache of pre-linked classes and profiles, and uses it at deployment to cut startup and warmup work. The prototype implements AOT linked classes, eager resolution for constant pool entries, cached invokeDynamic/Lambda linkage and ahead‑of‑time method profiles and code; results show ~3x speedups on "hello world" frameworks and ~2.7x on a 20k‑class Spring Petclinic. Constraints are minimal (no pre‑load classfile hooks).
AI-written summary · about
chapters (14)
- 0:00Project Leyden overview
- 0:58Early access and features
- 2:38Benchmarking results
- 6:54Shifting work safely
- 10:19Java's dynamic nature
- 16:10Startup and warmup issues
- 19:22Leveraging training runs
- 22:35CDS and class loading
- 26:14AOT class loading and linking
- 30:56Shifting resolution work
- 35:35Three-phase workflow
- 38:03AOT method profiling
- 40:37AOT code compilation
- 44:06Leyden future roadmap
related talks
23mAhead of Time Computation
46mProject Leyden's AOT - Shifting Java Startup into High Gear
41mSupercharge your JVM performance with Project Leyden
42mSupercharge your JVM performance with Project Leyden and Spring Boot
46mAssembling Project Leyden
48mProject Leyden - Capturing Lightning in a Bottle
51mSupercharge your JVM performance with Project Leyden
45mProject Leyden: Improving Java’s Startup Time