User Profile

Post not yet marked as solved

Click to stop watching this thread.

You have stopped watching this post. Click to start watching again.

9 Replies

0 Views

Post marked as unsolved with 9 replies, 0 views

Replied In Why is it so slow?

Small batch sizes (default is 32) run much faster on CPU. Thats normal. GPU takes over once you have big workload.

Post marked as solved

Click to stop watching this thread.

You have stopped watching this post. Click to start watching again.

13 Replies

0 Views

Post marked as solved with 13 replies, 0 views

Replied In How to install Tensorflow on the Apple M1 Notebook

Compile from Source guideline: https://github.com/tensorflow/tensorflow/issues/51506#issuecomment-901460541

Post not yet marked as solved

Click to stop watching this thread.

You have stopped watching this post. Click to start watching again.

1 Replies

0 Views

Post marked as unsolved with 1 replies, 0 views

Replied In M1 MBP not accelerating gpu for TensorFlow

You can switch between CPU and GPU, both work fine with TF 2.5 and MBP M1. Depending on your batch size, the CPU can be much faster. from tensorflow.python.framework.ops import disable_eager_execution disable_eager_execution() with tf.device('/CPU:0'): # GPU or CPU ... TF 2.5 installation steps: https://developer.apple.com/metal/tensorflow-plugin/