GoKawiilYk JIT toolkit adapted for CPython, outperforms Python's experimental JIT on some benchmarks
Researchers applied the yk interpreter JIT-compilation toolkit to CPython, producing a variant called ykcpython, after earlier testing had focused on the much smaller PUC Lua interpreter. The work took roughly 3-4 days of direct effort on top of a week and a half of generic bug fixing in yk, and benchmark results using the AWFY suite show ykcpython beating both standard CPython and CPython's own experimental JIT on most tests, though not all.
GoKawiil's interpretation of the reporting above, not reported fact.
The result suggests yk's approach to automatically generating JIT compilers may scale beyond small interpreters to large, widely used ones like CPython, addressing a specific challenge raised by a critic about whether the technique generalizes. If confirmed with further testing, this could offer an alternative path to speeding up Python execution without a fully hand-built JIT, though the benchmarks here are limited in scope and the work is described as preliminary.
- Yk, a toolkit for adding JIT compilers to C interpreters, has been adapted to work with CPython as 'ykcpython'.
- Benchmark results show ykcpython often outperforming both standard CPython and its experimental JIT, though it loses on a few individual tests.
- The project took only days of targeted work after prior infrastructure fixes, suggesting yk could extend to other large interpreters beyond its original small-interpreter testing.
Source: tratt.net, 2026-11-12
Published there as: “Adapting CPython for Yk”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.