Well, first I want to determine if a job is using the GPUs at all.
Second, I'd like to know for what portion of the job the GPUs were
actually used, either as a percentage of job run time, GPU-hours, or
even total wallclock time.
Does that help clear things up a bit?
Prentice
On 11/14/19 1:48 PM, Ryan Novosielski wrote:
Do you mean akin to what some would consider "CPU efficiency" on a CPU job? "How
much... used" is a little vague.
________________________________________
From: slurm-users <slurm-users-boun...@lists.schedmd.com> on behalf of Prentice
Bisbal <pbis...@pppl.gov>
Sent: Thursday, November 14, 2019 13:41
To: Slurm User Community List
Subject: [slurm-users] Get GPU usage from sacct?
Is there any way to see how much a job used the GPU(s) on a cluster
using sacct or any other slurm command?
--
Prentice
