Looking Into Peephole Optimizations
summary
Shows that MRI runs a peephole pass after compilation (implemented in compile.c calling iseq_peep_optimize) over YARV bytecode produced by RubyVM::InstructionSequence, with about 20+ cases of pattern replacement. Examples include shrinking range construction to one instruction (measured 5.4× faster), removing redundant splat/dup (1.3× and 1.26× gains, sometimes within noise), and replacing literal-array allocations; benchmarks use yjit_bench and report modest instruction and memory reductions.
AI-written summary · about
chapters (4)
- 0:00Start
- 4:01Peephole optimization is used in other layers of the stack and other languages.
- 7:16How does Ruby implement peephole optimization?
- 23:12Ruby is a dynamic language, so it's hard to get information at compile time.
