CPU and GPU usage

[sanketgujar]

I am using one GPU to train the model, but it is still using around 10-16G CPU memory. Would you please tell what operations are using so much memory on the CPU.

[philip-huang]

Hmm interesting, I have never noticed this before. My guess would be the input preprocessing stuff is using too many threads/memory. This part is handled by the CPU because GPU training is not really the bottleneck (data preprocessing is).

My suggestion would be to take a look at the input preprocessing stuff and see if you can come up with a better way for your specific application.

[sanketgujar]

Thanks for the reply, I did and the preprocessing was the issue. Closing the issue.