CPU vs GPU

Dear Organizers,

I have a question regarding GPU vs CPU allocation. It seems like in order to enable GPU one needs to change: FROM ubuntu:20.04 to FROM nvidia/cuda:11.1-runtime-ubuntu20.04 in the docker file. That seems like to be not working as the public key is not available. I guess nvidia has rotated the keys recently. That's what I have found on forums. If anyone else also encountered the issue I appreciate letting me know.

My current algorithms are running on CPU, and I get successful results both for seg/det and survival. Does it imply that the algorithm successfully run on all the hidden test images? Or because of CPU allocation it might have only run on some of them, maybe due to the time limits?

I just wanted to make sure regardless of using GPU or CPU, if one succeeds getting results, it implies that the job is run on all the hidden test images.

Thanks a lot.

Best, Arian

Dear Arian,

Yes, I also encounter this now. A possible fix that worked for me was to add this the Dockerfile

solution

Yes, if your job succeeded, all images have been processed.

Best wishes, Mart

Hi Mart,

Thanks for the hint. Yes, that resolved the issue.

Best, Arian