Dear All,
I am utilizing a GPU with 4 graphics cards to run relatively short amber md
jobs with amber20 pmemd.cuda executable. I have a total of ~ 80,000 such
short amber jobs. Each job takes < 30 seconds to complete on a graphics
card on the GPU.
A (potential) problem with this strategy is that currently I can not run
multiple jobs at the same time on a particular graphics card. As a result,
the queue time is rather long as I am only running four such jobs at a time
(on four graphics cards).
Is there a way around this? I am using the SLURM queuing system to submit
the jobs on the GPU.
Thank you.
P. Nandigrami
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Feb 15 2022 - 07:30:03 PST
