
<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">Hello,<div class=""><br class=""></div><div class="">yesterday we upgrade our cluster from Slurm 20.02.2 to 20.02.5 and recognized some problems with the usage of gpus and more than one cpu per task.</div><div class="">I could reproduce that problem in a little Docker container, which description you could find on the following link.</div><div class=""><a href="https://github.com/bikerdanny/docker-centos-slurm/tree/gres-bug" class="">https://github.com/bikerdanny/docker-centos-slurm/tree/gres-bug</a></div><div class=""><br class=""></div><div class="">I created a separate branch (gres-bug) for reproducing that problem, please checkout the README.md.</div><div class=""><br class=""></div><div class="">Could anybody of you tell me, what do we wrong, how can we solve that problem?</div><div class="">We also found out that using „--cpus-per-gpu“ instead of „--cpus-per-task“ works with more than 1.</div><div class=""><br class=""></div><div class="">Kind regards and stay healthy</div><div class="">Danny Rotscher</div><div class=""><div class=""><br class="Apple-interchange-newline">

</div>

<br class=""></div></body></html>