Error
Could not identify NUMA node of platform GPU ID 0, defaulting to 0.
Your kernel may not have been built with NUMA support.
Reproduce
>>> tf.convert_to_tensor([1,2,3])
2022-11-04 14:40:50.905905: I tensorflow/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.
2022-11-04 14:40:50.905986: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:272] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)
<tf.Tensor: shape=(3,), dtype=int32, numpy=array([1, 2, 3], dtype=int32)>
Solution
Tried this
conda install -c apple tensorflow-deps
as suggested here
That introduced yet more errors. I could not import tensorflow as tf anymore.
Jeff Heaton's suggestion (and YAML file) did work.
After creating the environment using his YAML file, the Apple M1 Max was selected as the GPU.
>>> tf.convert_to_tensor([12,3])
Metal device set to: Apple M1 Max
systemMemory: 64.00 GB
maxCacheSize: 24.00 GB
2022-11-04 14:59:41.275014: I tensorflow/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.
2022-11-04 14:59:41.275444: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:272] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)
<tf.Tensor: shape=(2,), dtype=int32, numpy=array([12, 3], dtype=int32)>