It's dangerous to GC alone. Take this!
summary
IBM open-sources the OMR runtime components and plugs OMR's garbage collector into MRI Ruby, giving a working C-level integration with OMR buffers (variable-sized on-heap buffers replacing malloc), thread-local heaps for lock-free bump allocation, and a parallel mark-sweep implementation that preserves C-API compatibility. The result ships as a tech preview, shows about 8% higher Rails throughput and reduces GC overhead from ~8% to ~3%, and includes GC visualization and live profiling tools.
AI-written summary · about
chapters (7)
- 0:00Introduction and speakers
- 3:39Why Ruby for OMR GC
- 7:38Integrating OMR GC into Ruby
- 12:19Speeding up Ruby with OMR
- 20:12Performance results and benchmarks
- 23:22Future work and next steps
- 25:27Q&A session
related talks
51mCompacting GC for MRI
30mBuilding the Next-Generation Garbage Collector in Ruby
42mTo Know A Garbage Collector
34mUnder the Hood of Ruby's Generational Garbage Collector
48mMethods of Memory Management in MRI
51mCompacting GC in MRI
54mKeynote: Exploring Memory in Ruby - Building a Compacting GC
30mBuilding the Next-Generation Garbage Collector in Ruby