Python 3.15's JIT is now back on track

7 hours ago

CPython JIT performance goals achieved early: 11-12% faster on macOS AArch64 and 5-6% faster on x86_64 Linux compared to interpreters.
Initial struggles included the JIT project being slower than the interpreter and losing main sponsor funding, putting its future in doubt.
Success attributed to luck, right timing, and key contributors including Savannah Ostrowski, Mark Shannon, Diego Russo, Brandt Bucher, and others.
Community stewardship and breaking down complex problems into manageable tasks helped attract new contributors.
Key improvements included a tracing JIT frontend, dual dispatch mechanism, and reference count elimination, significantly boosting performance.
Daily JIT performance runs and infrastructure support were crucial for catching regressions and validating optimizations.
Collaboration and knowledge sharing with compiler experts like CF Bolz-Tereick and Max Bernstein played a significant role in the project's success.

Hasty Briefsbeta