Hacker News new | ask | show | jobs
by Ambix 1100 days ago
Is it possible to write llama.cpp in Futhark? Like do effective math manipulations on 4-bit vectors within GPU?