>Oh does llama.cpp use MLX or whatever?
replyNo. It runs on MacOS but uses Metal instead of MLX.
ANE-powered inference (at least for prefill, which is a key bottleneck on pre-M5 platforms) is also in the works, per https://github.com/ggml-org/llama.cpp/issues/10453#issuecomm...
replyIs that better or worse?
replyDepends.
replyMLX is faster because it has better integration with Apple hardware. On the other hand GGUF is a far more popular format so there will be more programs and model variety.
So its kinda like having a very specific diet that you swear is better for you but you can only order food from a few restaurants.
But you can always fall back to GGUF while waiting for the world to build a few more MLX restaurants. Or something like that; the analogy is a bit stretched.
replyllama.cpp uses GGML which uses Metal directly.
reply