My assumption is that containers run with rund would be able to use M1 hardware natively, so containers would be able to use the M1 GPU and Apple Neural Engine (ANE) cores? It'd be good to understand how to limit their use for particular containers too, similarly to this example given for NVIDIA GPU-utilising containers: docker run -it --rm --gpus '"device=0,2"' tensorflow/tensorflow:latest-gpu nvidia-smi
(Thank you so much for your work on this project - please let us know if there are ways we can help/contribute, I am sure if you outlined tasks that others could take on they'd pitch in)
My assumption is that containers run with rund would be able to use M1 hardware natively, so containers would be able to use the M1 GPU and Apple Neural Engine (ANE) cores? It'd be good to understand how to limit their use for particular containers too, similarly to this example given for NVIDIA GPU-utilising containers:
docker run -it --rm --gpus '"device=0,2"' tensorflow/tensorflow:latest-gpu nvidia-smi(Thank you so much for your work on this project - please let us know if there are ways we can help/contribute, I am sure if you outlined tasks that others could take on they'd pitch in)