one_sharing for a specific gres.
*From:* slurm-users on behalf of
Matthias Leopold
*Sent:* Thursday, January 27, 2022 16:27
*To:* Slurm User Community List
*Subject:* [slurm-users] addressing NVIDIA MIG + non MIG devi
I looked at option
> 2.2.3 using partial "AutoDetect=nvml"
again and saw that the reason for failure was indeed the sanity check,
but it was my fault because I set an invalid "Links" value for the
"hardcoded" GPUs. So this variant of gres.conf setup works and gives me
everything I want, sorry f
I should have mentioned more prominently that my problem is about
addressing NVIDIA MIG + non MIG devices in Slurm WITHIN ONE NODE
(although this is obvious from reading my report). That also gives me
the last resort solution of having one node with all MIG devices and one
node with no MIG devi
ve this flag implied. This flag is not combatible with
one_sharing for a specific gres.
From: slurm-users on behalf of Matthias
Leopold
Sent: Thursday, January 27, 2022 16:27
To: Slurm User Community List
Subject: [slurm-users] addressing NVIDIA MIG + non MIG
Hi,
we have 2 DGX A100 systems which we would like to use with Slurm. We
want to use the MIG feature for _some_ of the GPUs. As I somehow
suspected I couldn't find a working setup for this in Slurm yet. I'll
describe the configuration variants I tried after creating the MIG
instances, it migh
