User avatar
Two Hollywood Phonies @autumn@cafe.autumn.town
1mo
llama.cpp ROCm works better than ollama confirmed !
1
0
1
0
User avatar
Two Hollywood Phonies @autumn@cafe.autumn.town
1mo
not doing my formal tests and haven't tried vulkan but the larger models are noticeably faster, i think i get 30 t/s on qwen3.6 mtp? need to record it but won't today
0
0
0
0