[slurm-users] Query about Compute + GPUs

[Chris Samuel]

On Thursday, 30 November 2017 9:58:08 PM AEDT Markus Köberl wrote:

> I also saw the wrong Sockets, CPU and Threads. I did not recognize the wrong
> values for RAM. Therefore I did define Sockets, CoresPerSocket,
> ThreadsPerCore and RealMemory.

You can run "slurmd -C" to get it to tell you the config of the node.

> I did hope that slurm somehow tracks the memory so that it gets shared
> between the partitions. I do not like to limit for both because depending
> on the user we need between 2 and 200GB RAM per GPU...

The resources (CPUs, memory, GRES, etc) of a node are tracked at the node 
level, so if a job from one partition is using resources then they should not 
be available to any other job from any other partition.

All the best,
Chris
-- 
 Chris Samuel  :  http://www.csamuel.org/  :  Melbourne, VIC