Towards Ruby 3x3 performance
summary
Project pursues a 3x performance target by adding an RTL VM instruction set and a JIT integrated into MRI (repo: vnmakarov/ruby). Early builds execute bytecode-generated native code and beat pure interpreter on many microbenchmarks, but show high warm‑up sensitivity and increased memory use; compilation produces correct fast paths but misses optimizations (inlining, scalar evolution) and needs more work to close gaps versus JRuby, Truffle/Graal and OMR Ruby.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:11Ruby 3x3 goals
- 3:39Optimization strategies
- 6:40JIT compilation approach
- 10:27Instruction handling
- 13:44Type checks and optimizations
- 17:02Implementation challenges
- 20:28Performance benchmarks
- 23:25LLVM and long-running programs
- 26:53Warm-up and compilation
- 30:15Execution models
- 33:14Optimization results
- 35:05Benchmarking methods
- 36:58Real-world performance
- 39:12Memory usage
- 41:04Ruby on Rails impact
- 42:19Future directions
- 43:28Conclusion and Q&A
