← Latest products

SimSIMD vs SciPy

How AVX-512 and SVE make SIMD nicer and ML 10x faster. AVX-512 and Arm SVE bring masked loads and native f16 math to vector distances. SimSIMD exploits both, running up to 200x faster than SciPy on modern CPUs.

Visit ashvardanian.comexternal site

Reactions (1)

Similar products

by description similarity

  1. StringZilla v3 with C++

    github.com

    Rust, and Swift bindings, and AVX-512 and NEON

    Hacker News15 pts

  2. Faking SIMD to Search and Sort Strings 5x Faster

    ashvardanian.com

    StringZilla uses SIMD tricks to hit 16 GB/s substring search, beating standard libraries by 5-10x and parsing multi-terabyte files Python couldn't handle.

    Hacker News29 pts

  3. Faster FastAPI with simdjson and io_uring on Linux 5.19

    github.com

    Web Serving and Remote Procedure Calls at 50x lower latency and 70x higher bandwidth than FastAPI, implementing JSON-RPC & REST over io_uring ☎️

    Hacker News290 pts

  4. Mmdr

    github.com

    1000x faster Mermaid rendering in pure Rust (no browser)

    Hacker News47 pts

  5. TurboQuant-WASM

    github.com

    Google's vector quantization in the browser

    Hacker News165 pts