[torqueusers] PBS_NODEFILE order question
Jackie Scoggins
jscoggins at lbl.gov
Tue Feb 17 17:43:04 MST 2009
torque - version 2.1.8
maui version 3.2.6p19
Is there a way to make the PBS_NODEFILE order the nodes as follows:
n0000
n0001
n0000
n0001
n0000
n0001
instead of in order like:
n0000
n0000
n0000
n0000
n0001
n0001
n0001
n0001
based on your request of
-l nodes=2:ppn=4
The reason for this request is so that we would like users to be able
to specify that they want to run on less processors than are available
per node for memory intensive jobs. If you are allocated 8 processors
(on two nodes as above) and you naively request mpirun -np 4 then you
will end up running only on the first node, this defeats the purpose.
The current situation does permit ppn=1 to be specified, however, this
still allows another job to run on the same nodes, so that too is
undesirable.
I have tried chopping up the PBS_NODEFILE and using a different
hostfile for mpirun but I get an orted error from mpirun because it
does not look like it is selecting the nodes in the order in which I
specify in my hostfile.
Any help on this will be appreciated.
Thanks
Jackie
More information about the torqueusers
mailing list