microbenchmarks · refreshed on every commit to main

How fast does it lift?

Wrenlift compared head-to-head with standard Wren and a handful of other scripting languages on the same GitHub Actions runner. Best of 10 runs, executed code only; interpreter startup is excluded. Source for every benchmark lives in bench/.

Tiered JIT OSR · hot loops switch mid-iteration Best of 10 · per language Same runner · ubuntu-latest

Loading benchmark results…

Shorter bars are better. Each benchmark is run 10 times and the best wall-clock time is kept. Only the benchmarked code is timed; interpreter startup is excluded. Output is correctness-checked against standard Wren before timing; any divergence fails the workflow.