DeepSeek Writes Custom Metal Kernel for Kimi in 50 Minutes
technaturalism · reddit · 2026-08-09
While running a quantized DeepSeek model on a Mac Studio (512GB), a developer realized there was no existing Metal kernel on GitHub for Unsloth's ultra-low bit quantization of Kimi K2/K3. They decided to let the model write one itself.
The model successfully wrote a custom Metal kernel in about 50 minutes. Although the performance is mediocre, it is significantly better than pure CPU compute, achieving around 4 t/s decode and 20 t/s prefill on the K3 Q10 quantization. The author expressed amazement at the coding capabilities demonstrated by the model at such a low bit quant level.
More from coding & agent
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- Claude Code and Codex now autocomplete the user's next action — tokenbender · 2026-10-16
- Hermes Agent ships official Reverse Engineer Anything plugin with one-command setup — Teknium · 2026-10-16
- MLExpert: 35B MoE Agent Trained to Autonomously Handle Machine Learning Engineering — Shangheng Du · 2026-10-16
- Cowork-Gym: Alibaba Synthesizes 64.9K Agent Work Environments, Boosting Qwen by 8.7 Points — Anni Zou · 2026-10-16
- MMSearch-OPD: Joint RL and On-Policy Distillation for Long-Horizon Multimodal Search Agents — PekingUniversity · 2026-10-16