gilmap: parallel map runtime for Python by Pratik Bhadanegilmap: parallel map runtime for Python by Pratik Bhadane

gilmap: parallel map runtime for Python

Portfolio project
Pratik Bhadane
gilmap is an open-source Python and Rust library I created. It runs CPU-bound numeric functions in parallel without the usual limits of Python's global interpreter lock.
It picks the fastest route for each function automatically: native compilation with numba, Arrow compute kernels, a Cranelift JIT with SIMD, or a Rust worker pool where each worker has its own Python sub-interpreter. Data moves between them as Apache Arrow, and Rayon handles the parallelism. On the repository benchmark suite (Apple M3 Max) it is up to 464x faster than built-in map and 47x faster than multiprocessing.Pool. Version 0.1.0 is on PyPI under the MIT licence, with prebuilt wheels for Linux, macOS and Windows.

0 likes · 0 views · Nov 22, 2026

Open-source Python and Rust library that runs CPU-bound numeric functions in parallel. Up to 464x faster than built-in map on its benchmark suite. I am the crea

Tags