burnt-synapse

#llama.cpp

every note tagged llama.cpp

datenotetags

22 Aug 2026 qwen3.8 27b on $500 of scrap qwen3.8 27b — q8 weights, f16 kv cache — doing 200 t/s of prompt processing and 30 t/s of generation on one box of $500 scrap. no gpu in sight. #qwen#llama.cpp#cpu#hardware 22 Aug 2026 one million tokens, ten hours, three cves opencode, pointed at a local qwen, left alone for about ten hours. it came back with roughly a million output tokens, three promising cves, and proof-of-concept scripts that actually worked. clean run — no restarts, no… #qwen#llama.cpp#opencode#gpu#cve#security 22 Aug 2026 mtp: 60 t/s when the text is predictable the box does 35 t/s of generation. with mtp — multi-token prediction — it does 60+ when the next few tokens are easy to guess. #qwen#llama.cpp#mtp