Hi, I just wanted to note that the GPU queues don't always contain gpu in the name. Currently on Sanger's AI clusters, there are queues such as training, training-parallel, inference, etc. which are all GPU queues that would not fit the pattern.
This will pose problems for the snippet shown:
|
# set gpu parameters for gpu queues |
|
gpu='' |
|
if [[ $queue == *"gpu"* ]] |
|
then |
|
gpu="-gpu \"mode=shared:j_exclusive=no:gmem=${gmem}:num=1\"" |
|
fi |
Hi, I just wanted to note that the GPU queues don't always contain
gpuin the name. Currently on Sanger's AI clusters, there are queues such astraining,training-parallel,inference, etc. which are all GPU queues that would not fit the pattern.This will pose problems for the snippet shown:
vs-code-lsf/bsub.vscode.tunnel.sh
Lines 56 to 61 in 1d2d0f4