Writing GPU kernels in Rust: how I made speakrs 60× faster than pyannote while dropping…
This is a dev post classified by AGK Intelligence as Languages & runtimes (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
Writing GPU kernels in Rust: how I made speakrs 60× faster than pyannote while dropping ONNX Runtime, cuDNN and cuBLAS > speakrs, a Rust diarization library, replaced ONNX/cuDNN with custom cuda-oxide kernels—now ~60× faster than pyannote at same accuracy, mostly agent-written.
Posted by Rust Bytes 🦀 (9k followers) 1 days ago · 123 likes · 3.5k views · view the original post on X. Kept by the Dev Radar as Languages & runtimes.
More dev work like this
- Roundhouse turns Rails apps into standalone binaries or services in Rust, Swift,… — @rustaceans_rs
- We're starting work on Pydantic V3. — @pydantic
- zuvloop: a drop-in asyncio event loop, powered by libuv and written in Zig ⚡ — @marcelotryle
- Small Crates Behind Big Rust Systems — @rustaceans_rs
- js tip. infinity — @xah_lee
- massive tech geeking on javascript. — @xah_lee
- Infinity is a lightweight Rust framework for massively concurrent agents—75k fit on a… — @rustaceans_rs
- Need synchronous work NOW, but your C++ API expects a std::future? 🧠⚡ — @VisualStudio
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.2k posts from 5k X accounts over the last 21 days, 8.8k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-10-30 20:27 UTC. Full method.