How I accelerated Python matrix ops by 1200x using Rust & C-ABI

I built TritonX, an open-source high-performance matrix compute engine written in Rust using Rayon for parallel execution, bridged to Python via C-ABI bindings.

Benchmarks

It offloads heavy matrix math to multi-threaded Rust worker pools, achieving up to ~1200x speedups over pure Python loops.

Code & Setup

Check out the repository, run local benchmarks using .\run_tritonx.ps1, or contribute to open issues:
https://github.com/pulisherij-rgb/TritonX

Comments

[TritonX]

I built TritonX to explore low-overhead multithreaded matrix operations in Rust (using Rayon) while maintaining seamless, direct C-ABI bindings for Python execution.

It offloads matrix calculations to parallel worker pools, achieving up to ~1200x speedups over baseline Python matrix loops.

You can view the code and run local benchmarks using .\run_tritonx.ps1:
github.com/pulisherij-rgb/TritonX

Would love feedback on FFI boundary overhead, memory layout, or further parallelization optimizations!