How to train on several CPUS


I want to train on all of my CPUs at once.
I noticed that the defualt is 1.

Is there a way to define how many cpus I want to use? How do I define it?


DLS uses all the CPUs for training if you do not have GPU in your system. There is no setting to set the number of CPUs in DLS.

if your system has multiple GPUs then you can select the number of GPUs you want to use for training.

1 Like

How I can see this option of choosing GPU? I dont see this is my DLS UI? I can see GPU detected, but when DLS starts training, it starts with CPU mode.

Status o/p.

“command”: “train”,
“gpus”: 1,
“inuse”: 1,
“project_id”: “1”,
“run_id”: “2”,
“uptime”: “2067.044609315”

go to training tab, reload the page. Do you see number of GPUs dropdown ?

What is the URL of the DLS ? is it running on port 80 or some other port ?

Thanks, issue was with nvidia-docker. Though gpus detected, I think docker container was not able to see it as nvidia-docker driver was not okay.

Now fixed.