ParentFull threaditake·Yep, but isn’t there an integrated ML chip that makes it faster than cpu? Or does llama.cpp not use that?View on HN