Job submission on HPC using pbs script

User discussions
,
1

GROMACS version:2022.1
GROMACS modification: Yes
Respected expert,

IMG_20231102_094222
IMG_20231102_0942221920×1920 566 KB

I am using our institute HPC for mdrun. I can utilise one node but cannot utilise multiple nodes.
I have learnt that PBS.sh script(image attached)is used for job submission on our institute’s HPC
Even though i have tried modifications like varying number of nodes, process per node, but no relief, job is running only on one node .
Please help

2

Hi,

As far as I know, pmemd.MPI is an Amber executable, so it shouldn’t be used in front of gmx_mpi. On most clusters, one of the following should work:

  • mpirun gmx_mpi mdrun ...
  • mpiexec gmx_mpi mdrun ...
  • srun gmx_mpi mdrun ...

using all the CPUs that are requested via the PBS direcrtrives. Right now, you’re requesting 5*20=100 CPUs but trying to use -np 24 which would anyway waste most of the resources.

1 Like