I am trying to run a program on cluster. The program can distribute each of the calculation with an corresponding parameter, say k1, k2,...k4, to run on one core. I want to use 4 nodes with 1 processor per node to run the job. This is the script I wrote:
#!/bin/tcsh
#PBS -S /bin/bash
#PBS -l nodes=4:ppn=1
#PBS -l walltime=48:00:00
#PBS -q cluster1
cd $PBS_O_WORKDIR
(..command running the program...)
However, it turns out that the number of process would be greater than one in some nodes. For examples, node1 would be running the calculation of k1, k2, k3 simultaneously and node2 with k4. As the calculation of each of the k's would require a large virtual memory, I would prefer for each of the node there is only one k calculation. Are there any ways to restrict the no. of process on one node?
Any help would be appreciated. Thanks in advance.
Aucun commentaire:
Enregistrer un commentaire