[torqueusers] Cputime >> walltime

[Glen Beane]

How many processors are you using?  It is normal for CPU time to be
higher than walltime in a parallel job (unless you do not use a TM job
launcher an TORQUE can not track the cpu time of all processes).


On Mon, Jun 29, 2009 at 5:26 AM, ChrisJob.fr<chrisjob.fr at gmail.com> wrote:
> Hi,
>
>  We have  a PBS queue with this configuration :
>  set queue X resources_max.walltime = 72:00:00
>
> For  job 5555 , we have these informations :
>  resources_used.walltime = 29:56:51
>   resources_used.cput = 99:03:31
>  The walltime is inferior to the cputime !
>
>  Here is the PBS script 5555 (a lot of mpirun commands for onlye  one
> script):
>
> #PBS ....
> mpirun toto
> mpirun tata
> mpirun titi
> ...
>
>   Have you got any idea of how we can do to avoid this probleme ?
>
> Thank you in advance
> Chris
>
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers
>