[NLPL Task Force (A)] [uninett.no #199370] Accel Queue

[thierry.toutain via RT]

Hello Jeremy,
we will check with the user how his jobs do use gpus,
   Thierry



On Mon Dec 02 16:53:57 2019, jeremycb at ifi.uio.no wrote:


    Hello,

     

    I wanted to point out that over the last few days it has been very
    difficult to schedule an accel job on Saga. There seems to be a single
    user that has effectively saturated the gpu queue, but their jobs don't
    use the gpus effectively:

     

     

    squeue -p accel > /tmp/accel
    for i in $(squeue -p accel | egrep 'c[0-9]-[0-9]$' | sort -u | awk
    '{print $NF}'); do \
      ssh $i nvidia-smi | grep Default; \
    done > ~/nvidia-smi.log

     

    Would it be possible to let the user know that this situation is
    suboptimal?

     

    Thanks,

     

    Jeremy Barnes

    Language Technology Group

    University of Oslo

    Office 7645

    jeremycb at ifi.uio.no