Tensorflow-metal for M2 Max chip

TensorFlow General Discussion
,
1

Hi,

I have installed TensorFlow-metal in my MacBook Pro (38 GPU/32GB ram), and the time speed is acceptable when I run my experiment (transformer model using only encoder block):

  • Batches: 4
  • Data X training: (32, 1024. 512) EEG data of 1 subject
  • Epochs: 200
  • Time? It takes around 40 minutes to finish

But when I run my experiment:

  • Batches: 72
  • Data X training: (1024, 1024, 512) EEG data of 32 subjects
  • Epochs: 1000
  • Time? After 6 hrs, it is still on epoch 1!

Is it normal?
Tensorflow-metal plugging is not taking the 38 GPU of the computer?

3

@Rick_Licona,

Welcome to the Tensorflow Forum!

Tensorflow-metal plugging is not taking the 38 GPU of the computer?

This is related to Tensorflow-metal. Could you please post this issue in the Apple developer forum Apple Developer Forums for quick assistance?

Thank you!

5

Hi @Rick_Licona ,

Could you ever use your Mac GPU for model training or inference ?

I am desperately looking for information all over the internet and the Apple Developer forum for tensorflow-metal looks dead i.e. no official support whatsoever.

TIA