Does this leverage GPU on M1 Max / Apple GPU platforms?

[cris-pin]

Hi, I'd like to get this running on my M1 Max but am wondering whether it has been supported or tested under Docker on this platform and whether only CUDA / Nvidea cards are supported? Or is there a Apple GPU supported release at all / in the pipeline?

Many Thanks

[aaronabebe]

Hi @cris-pin!

The basic library management functionality is tested on M1 and works.
Everything else regarding the tools is currently still very experimental and doesn't utilize GPU acceleration on ARM Mac's yet.

Since we're offering integrations with a lot of different open source ML models through nendo's plugins and most of them only support CUDA for GPU acceleration, we can't really offer support for Apple's GPU platform yet.

However we know that a lot of people in our community are using ARM Mac's and a supported release is definitely planned!

[itsbrex]

@aaronabebe is there a public backlog for Mac OS? Would like to contribute.

[aaronabebe]

Hi @itsbrex!

I created a backlog item that outlines the inital step for MacOS support here: https://github.com/orgs/okio-ai/projects/6/views/1?pane=issue&itemId=54602024.
Please take a look at that and let me know if you think you can contribute there or if you have any more questions regarding this outlined topic!

Please note, that at the current point there's really no other issues that we can tackle easily on our side, since most of the ARM Mac support has to come from upstream changes to one of the libs we use, e.g. transformers, audiocraft or pytorch.