logxio/picchio
Catch local LLMs spilling into the CPU. One run shows GPU placement, prefill, decode, memory, power and a verdict.
19 best points · 2 comments · 3 Hacker News posts · 18 GitHub stars · View on GitHub
Hacker News discussions
- Same model, same Q4_K_M label: 5.02, 5.07 and 5.27 bits per weight (19 points, 2 comments, by logickkk1, 2026-07-14)
- I found why Qwen3.8-27B was 14× slower on my 4070 Super (8 points, 0 comments, by logickkk1, 2026-08-23)
- Show HN: Catch your local LLM falling back to CPU (2 points, 0 comments, by logickkk1, 2026-07-11)