Qwen3.6 35B-A3B-MTP / UD-Q4_K_M / fork v0.6.10 / PP16384 TG256 at prepared depth131072
on AMD Strix Halo 128GB · Nathanw1014 Strix Halo llama.cpp fork v0.6.10 / Vulkan RADV / 2586f6e
Summary
User reports Qwen3.6 at 35 t/s generation and 356.8 t/s prompt processing on an AMD Strix Halo 128GB machine.
Setup is the Nathanw1014 Strix Halo llama.cpp v0.6.10 fork with UD-Q4_K_M, f16/f16 KV cache, flash attention on, 16 CPU threads, ngl999 and Vulkan0, at a depth of 131072 tokens.
The figures are native llama-bench pp16384/tg256 results, N=3, with MTP disabled and reasoning off; deeper cells gave 238.8/28.8 at 200000 and 167.4/24.9 at 262144.