3 days benchmarking most llama.cpp flags on my weird 40gb vram laptop + tb4 egpu setup. Got +70% generation, +40% prefill, 60k more context, and filed a bug in llama around MTP. What I learned.

tldr: went from 16~ t/s to 27~ t/s generation. got my usable context up from 220k to the full 262k without sacrificing anything. prefill also increased from 376 to 573 command I ended up with, fwiw: llama-server -m Qwen3.8-27B-UD-Q6_K_XL.gguf -c 262144 -ngl 999 -fa on \ -ctk q8_0 -ctv q8_0 --spec-ty