Middle schooler hand-writes Metal RMSNorm kernel for own inference engine on Apple Silicon
dosco · x · 2026-10-26
一位中学生发布了自研推理引擎 Artisan（GitHub），用 Metal 手写了 RMSnorm 内核，用于在 Apple Silicon 上运行 DeepSeek-R1-Distill-Qwen-1.5B，目前仍在做 benchmark 和优化。
dosco 感叹：连车都不能合法开的中学生已经在写裸机推理内核了，「算力正在渗透到年轻一代」。
More from Infra
- $21 of DeepSeek API wrote Molten: a Rust+Metal engine serving 300+ tok/s on M5 Macs — GabGarrett · 2026-10-26
- "90% of tokens local, 10% frontier": HF engineer's model allocation take — TheZachMueller · 2026-10-26
- Musk: SpaceX closes $40B GPU debt deal, warns 200+ neoclouds face die-off — XFreeze · 2026-10-26
- Strata ported to Jetson AGX Orin 64GB: Qwen3.6-35B decoding pushed to ~41 tok/s — No-Setting8461 · 2026-10-26
- Sam Altman: robots building robots and data centers building data centers form another AI flywheel — haider1 · 2026-10-26
- NVIDIA Vera CPU packs 1.5TB LPDDR5X as agentic AI drives DRAM demand wave — firstadopter · 2026-10-26