I am using a cluster computer system which has a queue that allows 2 nodes per job with a maximum limit for number of jobs that a user can submit. So I wanted to use the two nodes with a single job submission, as I am currently executing single node jobs only. Earlier in a Cray system, I could use the following script to submit two jobs in two different nodes from a single job file
#PBS -l select=2:ncpus=40:mpiprocs=1:ompthreads=40
#PBS -l place=scatter
(cd FOLDER_1 ; aprun -n 1 -d 40 ./a.out ) &
(cd FOLDER_2 ; aprun -n 1 -d 40 ./a.out) &
wait
But in this present cluster, this runs both the jobs in a single node. [Note: unlike Cray, this cluster doesn’t use aprun
to submit jobs]. Is there a way to achieve this with some tweak in the job script?