You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
benchmarks/typedpython/ holds five programs: nbody, spectral-norm, fannkuch, binary-trees, and one dict-heavy workload. Each is written as checked TypedPython and run as interpreted CPython, as @compiled, and against equivalents in Node, Java and Rust on one machine. The targets in design §6 / the performance note: beat V8, numeric loops within 1.5× of Rust, object-heavy code within about 2× of Java.
Done when
A results table has been committed, measured alone (AGENTS.md §8), with the machine and the date. Rerunning it is one command.
Activity
Bench: Record the TypedPython baseline (idle machine, load 1.6)
Bench: Run the TypedPython @compiled column through the pipeline
b-re-w commentedon Oct 3, 2026
Landed in develop via #28.