0

I am trying to run rllib algorithms on ray cluster. I get the following message as, "Memory usage on this node: 20.8/64.4 GB"

How shall i make it use the memory fully? How to cross check whether the GPU or CPU utilization is above 90%?

Kindly help....

Deepak Nayak
  • 157
  • 1
  • 1
  • 7
  • You can usually check CPU utilization with `top` and GPU utilization with `nvidia-smi`. You probably don't want memory utilization to be 100% because then things will run out of memory. – Robert Nishihara Feb 21 '19 at 17:31

1 Answers1

0

You can usually check CPU utilization with top and GPU utilization with nvidia-smi. You probably don't want memory utilization to be 100% because then things will run out of memory.

Robert Nishihara
  • 3,276
  • 16
  • 17
  • As i am running the algorithms on Ray cluster. How do i check with "top" command? – Deepak Nayak Feb 22 '19 at 04:19
  • Oh I see, you can ssh to each machine and run `top`. You can also view the information in the dashboard, which was merged in https://github.com/ray-project/ray/pull/4066. However that has not been released yet, but will be in the next release. – Robert Nishihara Feb 22 '19 at 05:53