> Am 05.12.2015 um 16:47 schrieb Dj Merrill <[email protected]>: > > On 12/5/2015 9:55 AM, Rajil Saraswat wrote: >> Each node has 32 cpus and 1 gpu. Why is pe_slots being limited to 2 >> even when 64 cpus are available? > > I have ours setup in a similar manner, and have the same result. My > interpretation is that requesting "-l gpu=1" requires a gpu for each slot, so > you can only use 1 of the 32 cpu slots on each machine, thus only giving you > 2 total pe_slots since you only have 2 gpus.
Is gpu set to FORCED on your case? -- Reuti > > I haven't yet figured out a way to request a multi-cpu job per gpu, but I'm > guessing someone on the list here has already figured this out and might be > able to share with us. > > -Dj > > _______________________________________________ > users mailing list > [email protected] > https://gridengine.org/mailman/listinfo/users _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
