Trouble while running Tensorflow after installing TensorFlow-metal

[mikkel-larsen]

I followed the instructions, but afterwards I get the errorcode

Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.

every time I run something with Tensorflow. The program runs as otherwise expected, and the programs does not stop but continues to run, only very slow (far slower than using CPU).

Besides that errorcode I also get

Plugin optimizer for device_type GPU is enabled.

every time.

I am using a regular M1, and have followed the instructions, and also tested the "list physical devices" which works just fine.

TensorFlow 2.7

[AnshuGusain]

Im having the same issue when compiling a model on my M2pro. I tensorflow[/core/common_runtime/pluggable_device/pluggable_device_factory.cc:306](https://file+.vscode-resource.vscode-cdn.net/core/common_runtime/pluggable_device/pluggable_device_factory.cc:306)] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.

[michael-learns]

I had the same error as well. But I fixed it through this Apple Developer post. Dropping back to versions: tensorflow-macos==2.9 and tensorflow-metal==0.5.0 fixed the issue for me.

https://developer.apple.com/forums/thread/721619