I built TritonX to solve a common bottleneck: execution latency when running heavy matrix operations in Python without needing a full GPU setup. By writing the core compute engine in Rust using Rayon for parallel processing and connecting it directly to Python via low-overhead C-ABI/FFI bindings, TritonX delivers up to ~1200x speedups over pure Python implementations. Key Highlights: Multi-Threaded CPU Engine: Written in Rust + Rayon Low-Latency Python Bridge: Raw C-ABI integration with NumPy and FastAPI Real-Time Visuals: Interactive Chart.js dashboard tracking execution time & GFLOPS One-Command Setup: Run .\run_tritonx.ps1 to test it locally The project is 100% open-source: https://github.com/pulisherij-rg... I’d love to hear your thoughts on thread allocation, CPU memory layout, or any optimizations you'd like to see!

TritonX

High performance Rust matrix engine with PythonCABI bindings

High performance Rust matrix engine with PythonCABI bindings