parallel pool for deep learning

Dear all,

I am training a network for deeplearning.

In order to speed up the training, I created a parpool (in this case of 2 (instead of one).

However the process time for the training remains the same, so it looks like the parpool does not wotk for training a deeplearning network.

I do have the parallel toolbox 2019b (ver 7.1) and deep learning toolbox 2019b (ver 13.0)

Can anyone help me?

Thx

Answer 1

[Maksym Tymchenko]

This example demonstrated how you can train a single deep learning network using multiple GPUs in parallel:

https://www.mathworks.com/help/deeplearning/ug/deep-learning-with-matlab-on-multiple-gpus.html

If you are interested in training multiple deep learning experiments in parallel instead, check out this example:

https://www.mathworks.com/help/deeplearning/ug/train-deep-learning-networks-in-parallel.html