On 22 August 2012 23:53, Joseph Farran <jfar...@uci.edu> wrote: > Hi. > > I have a GPU node with 24 Intel cores and 4 GPU Tesla cards. > > My configuration has: > > The node: > # qconf -se compute-1-14 | egrep "gpu|processors" > complex_values gpu=4 > processors 24 > > > The complex setup: > qconf -sc | grep gpu > gpu gpu INT <= YES YES 0 0 > > The queue: > # qconf -sq gpu | egrep "complex|slots" > slots 24 > complex_values NONE > > The PE setup: > # qconf -sp gpu > pe_name gpu > slots 24 > user_lists NONE > xuser_lists NONE > start_proc_args NONE > stop_proc_args NONE > allocation_rule $pe_slots > control_slaves TRUE > job_is_first_task TRUE > urgency_slots min > accounting_summary TRUE > qsort_args NONE > > > However, I cannot seem to be able to request both the number of Intel cores > and GPU cards with: > > $ qrsh -l gpu=1 -pe gpu 6 > error: no suitable queues > > I can do the "-l gpu=1", or the "-pe gpu 6" option, but not both. > > > What do I have wrong above? How do I request 6 Intel cores with 1 GPU card?
You have consumable set to YES which means the request is multiplied by the number of slots you request 6>4 so you can't allocate them. Assuming you have a recent version of Grid Engine you can set the consumable setting to JOB. This will mean that the gpu resource is allocated for the MASTER slot of the job. Unfortunately no version of grid engine (AFAIK) supports per-host resource consumption but as you are using a $pe_slots PE it shouldn't matter. William > > Joseph > > > > > > _______________________________________________ > users mailing list > users@gridengine.org > https://gridengine.org/mailman/listinfo/users > > _______________________________________________ users mailing list users@gridengine.org https://gridengine.org/mailman/listinfo/users
