So I am running a dynamics simulation using plane wave density functional theory using the NWPW directive, and I am submitting it as a job to a cluster which uses the PBSpro has it's job controller. I have tested my nwchem input script using a single core and it runs fine but I've been running into an issue with running on multiple cores.
I submit the job with 'ncpus=8,mpiprocs=8' on the assumption that each MPI task will use a single core and my command to use mpirun is 'mpirun -np 8', but soon after the job starts it returns an error that PBS killed the job because 'ncpus 22.99 exceeded limit 8' I assume then that each non-controller MPI process requests 3 cpus each which then exceeds the CPU limit of the job causing PBS to kill.
Is there a command in NWChem which controls the number of cores that an MPI process can request or is this something I have to fix on the job submission script side?