On Tuesday, June 26, 2018, the gpu@wagap-pro, gpu@arien-pro, and gpu_long@arien-pro queues setting has been changed:
Due to the limitation of non-GPU jobs access to GPU machines, we have set the gpu and gpu_long queues on both PBS servers only for jobs explicitly requesting at least one GPU card:
If the GPU card is not required in the qsub, the following message is displayed and the job is not accepted by the PBS server:
'qsub: Job violates queue and/or server resource limits'
At the same time, we set up the gpu queue sharing between the two PBS servers (jobs from arien-pro can be run at wagap-pro and vice versa). The gpu_long queue is managed only by the arien-pro PBS server, so the change does not apply.
More information about GPU machines can be found at https://wiki.metacentrum.cz/wiki/GPU_clusters
Thank you for your understanding,
MetaCentre users support