How to utilize multiple CPUs for training of YOLO?

Question

I have access to a large CPU cluster that does not have GPUs. Is it possible to speed up YOLO training by parallelizing between multiple CPU nodes?
The docs say that device parameter specifies the computational device(s) for training: a single GPU (device=0), multiple GPUs (device=0,1), CPU (device=cpu), or MPS for Apple silicon (device=mps). What about multiple CPUs?

Karl · Accepted Answer · 2025-01-19 17:50:23Z

0

You can use torch.set_num_threads(int) (docs) to control how many CPU processes pytorch uses to execute operations.

answered Jan 19 at 17:50

Karl

5,9661 gold badge11 silver badges19 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Artem Lebedev Jan 22 at 20:45

Did not work for me. Just tried. It runs on 16 threads no matter what I set, and of these 16 only 8 actually compute.

Collectives™ on Stack Overflow

How to utilize multiple CPUs for training of YOLO?

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related