You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Given that pytorch supports GPUs (via Metal/MPS) on Mac M1 machines (since version pytorch 1.12, should it not be possible to enable this in the pytorch engine?
How much work would this be now that the JNI bindings are now ARM (aarch64)
Currently we are already using the offical torch 1.12.1 MacOS Arm64 wheel to build the JNI. So it has capabitlity to extend to GPU, but we didn't implement the MPS device yet. We are welcome for any contribution to this device addition. You can try to get it work by adding MPS device in the code
Given that pytorch supports GPUs (via Metal/MPS) on Mac M1 machines (since version pytorch 1.12, should it not be possible to enable this in the pytorch engine?
How much work would this be now that the JNI bindings are now ARM (aarch64)
The speed up is significant for Mac M1 users.
References
https://pytorch.org/blog/introducing-accelerated-pytorch-training-on-mac/
The text was updated successfully, but these errors were encountered: