Hacker News Reader

Comment by ignoramous | Hacker News Reader

> the newer Apple Silicon Macs with unified memory

Mixtral has been MLXd already? Write ups, if any?

Not to my knowledge. But because the unified memory doubles as VRAM for the onboard GPU, normal GPU acceleration can access the entire model even if it's 50+ GB. That's why ASi Macs are currently the holy grail for at-home inferencing, and also why projects like llama.cpp focus so much on ASi above all else, and why so many UIs release for macOS first before other operating systems. Certain Mac models offer up to 192GB of unified memory.

But that's not a Macbook. And a Macbook M3Max with 128GB of RAM is almost 8000€.

Considering how inaccessible and expensive 128GB of pro-level cards is, that is believe it or not, a good price.

Not to mention that's 128GB for a single GPU. No need to shard or split.

Yes it has, actually: https://github.com/ml-explore/mlx-examples. It's right in the main repo. NB, I haven't tried this, I'm using llama.cpp with a non-K-quant quantization on my MBP.

I have and don't consider MLX to be production ready. I've tested it on M1Max and M1Ultra (128) machines. It's completely non-deterministic in its resource consumption, sometimes using the GPU fully, sometimes getting seemingly stuck while processing, sometimes the GPU throttles.

However, there's one curious thing: llama.cpp _always_ leads to GPU throttling on Apple Silicon (e.g. the M1Max GPU will go from 1200MHz to around 700MHz), and then fully saturates it. In the rare cases I could get MLX to stay on the GPU, it was able to keep it at the maximum clock rate. However the unpredictable pauses and seemingly unoptimized prompt processing makes it hard to pick a winner in end-to-end tokens/s

Many options for running Mistral models in your terminal using LLM:

https://simonwillison.net/2023/Dec/18/mistral/

I liked "Using Llamafile’s OpenAI API endpoint" described there, using Justine Tunney's llamafiles for Mixtral, but the article link is out of date, as the models have been replaced with newer: https://huggingface.co/jartine

Reply on news.ycombinator.com