I would like to test multiple parameters on hetero refinement to detect small changes in a structure. If I queue multiple hetero refinements, two will run simultaneously and I run out of GPU memory. I cannot decrease the box size, because the differences I want to see are quite small. Is there a setting to prevent multiple refinements from running at the same time?
If I’m understanding your post correctly, it seems like you’re running two separate multi-class heterogeneous refinement jobs at the same time. If thats the case, each job will run on separate GPUs. Can you post the errors you get when queuing both jobs at the same time?. What box size are you using? Also, what GPU’s are you using?
Hi @stephan, I apologize I think the error was due to the large box size and high particle number, rather than queuing simultaneously. Sorry!