[gmx-users] pme_loadbal problem

himanshu khandelia hkhandelia at gmail.com
Fri Nov 27 11:43:08 CET 2015

Dear All

I am having this error on our local cluster, which I did not have before,
with v. 5.1.

Program gmx mdrun, VERSION 5.1
Source code file:
ng.cpp, line: 934

Software inconsistency error:
pme_loadbal_do called at an interval != nstlist
For more information and tips for troubleshooting, please check the GROMACS
website at http://www.gromacs.org/Documentation/Errors

* The value for nstlist is 20

* The hardware is either pure CPU nodes with 24-cores on each node, or the
same nodes with 2 GPUs

* I have tried to run this with and without GPUs.

* The settings for the GPU are:

#SBATCH --nodes 4
#SBATCH --gres=gpu:2

### MPI ranks per node:

#SBATCH --ntasks-per-node 8 #

gmx_gpu_mpi mdrun"

gmx_gpu_mpi mdrun -v -deffnm molname -cpi molname

* The settings without GPUs is:
#SBATCH --nodes 8
#SBATCH --ntasks-per-node 24

gmx_mpi mdrun -v -deffnm molname -cpi molname

Himanshu Khandelia, PhD
Associate Professor
MEMPHYS, Center for BioMembrane Physics: www.memphys.dk/Himanshu-Khandelia
University of Southern Denmark (SDU)
Campusvej 55, Odense M 5230, Denmark
Phone: +45 6550 3510, +45 2912 5467
In India: +917893140837
email: hkhandel at sdu.dk <hkhandel at memphys.sdu.dk>, hkhandelia at gmail.com

More information about the gromacs.org_gmx-users mailing list