3 ms·

Based on George Hotz's testing it is very broken. It's possible it has improved since then, I guess but he streamed this a few weeks ago.

dagmx 4y ago
It supports a subset of the operators (as mentioned in the release notes). I don’t think it’s broken for the ones that it does support though.
- norgie 4y ago
  Yes, this is my experience. Many off the shelf models still don't work, but several of my own models work great as long as they don't use unsupported operators.
  - glial 4y ago
    Where can I find a list of the supported operators?
    - PaulMest 4y ago
      This is the master tracking list for MPS operator support: https://github.com/pytorch/pytorch/issues/77764 https://github.com/pytorch/pytorch/issues/77764
- mochomocha 4y ago
  That's been my experience. However when fallback to CPU happens, it sometimes end up making a specific graph execution slower. But that's explicitly mentioned by the warning and pretty much expected.
- brucethemoose2 4y ago
  You'd think it would fall back to GPU/CPU for unsupported operations instead of failing, but I guess thats easier said than done.
  - rfw300 4y ago
    In fact there is an environment variable that enables exactly that. There are obviously performance issues associated with that, but it does work.
bigbillheck 4y ago
[flagged]
- jeron 4y ago
  so, you would bet he was holding it wrong?
  - remexre 4y ago
    More likely, no bet would be placed.
microtonal 4y ago
We tested inference for all spaCy transformer models and they work: https://explosion.ai/blog/metal-performance-shaders https://explosion.ai/blog/metal-performance-shaders It depends very much on the ops that your model is using.
laichzeit0 4y ago
I could fine-tune a Detectron2 model a few months ago using PyTorch and MPS backend [1]. I'd be interested if it's working yet. https://github.com/facebookresearch/detectron2/issues/4342 https://github.com/facebookresearch/detectron2/issues/4342

dagmx 4y ago

It supports a subset of the operators (as mentioned in the release notes). I don’t think it’s broken for the ones that it does support though.

norgie 4y ago
Yes, this is my experience. Many off the shelf models still don't work, but several of my own models work great as long as they don't use unsupported operators.
- glial 4y ago
  Where can I find a list of the supported operators?
  - PaulMest 4y ago
    This is the master tracking list for MPS operator support: https://github.com/pytorch/pytorch/issues/77764 https://github.com/pytorch/pytorch/issues/77764
mochomocha 4y ago
That's been my experience. However when fallback to CPU happens, it sometimes end up making a specific graph execution slower. But that's explicitly mentioned by the warning and pretty much expected.
brucethemoose2 4y ago
You'd think it would fall back to GPU/CPU for unsupported operations instead of failing, but I guess thats easier said than done.
- rfw300 4y ago
  In fact there is an environment variable that enables exactly that. There are obviously performance issues associated with that, but it does work.

bigbillheck 4y ago

[flagged]

jeron 4y ago
so, you would bet he was holding it wrong?
- remexre 4y ago
  More likely, no bet would be placed.

microtonal 4y ago

We tested inference for all spaCy transformer models and they work: https://explosion.ai/blog/metal-performance-shaders https://explosion.ai/blog/metal-performance-shaders It depends very much on the ops that your model is using.

laichzeit0 4y ago

I could fine-tune a Detectron2 model a few months ago using PyTorch and MPS backend [1]. I'd be interested if it's working yet. https://github.com/facebookresearch/detectron2/issues/4342 https://github.com/facebookresearch/detectron2/issues/4342