Field arithmetic performance across six Rust implementations. Same hardware, same conditions.
Wall-clock time for core field operations across six libraries. Lower is better. All benchmarks run on identical hardware under identical conditions.
Reproducible benchmarks you can run yourself.
Apple M2 Pro, 32 GB unified memory. Single-threaded. Results will vary by platform — run locally for your hardware.
Rust nightly (2025-11-09), criterion.rs v0.5. 200 samples per benchmark, reporting mean wall-clock time.
Clone the repo, run cargo bench. Raw data and benchmark harness in /benches.
All implementations compiled with identical flags (-C target-cpu=native). No implementation-specific tuning. Latest stable release of each library.