[gmx-users] Getting a high potential energy error when using some specific gpu options. Gromacs 2019.4

[Cardenas, Alfredo E]

Hi,

I am testing a new computer system with single nodes containing four GPU’s, for a membrane containing a peptide with numerous restraints. I have read in the gmx_users list that for this kind of systems maybe the more efficient use of a node is to run several simulations on it so I have tried several input options. Anyway, what I show here is an attempt to run a single trajectory on two specific GPUs of a node and using -pme gpu. The command is:

mdrun -v -deffnm topolshort -nt 8 -pin on -pinoffset 8 -pinstride 1 -ntmpi 8 -ntomp 1 -npme 1 -pme gpu -gpu_id 23

 I obtained the error:



Program:     mdrun, version 2019.4

Source file: src/gromacs/mdlib/sim_util.cpp (line 752)

MPI rank:    6 (out of 8)



Fatal error:

Step 0: The total potential energy is 1.40845e+15, which is extremely high.

The LJ and electrostatic contributions to the energy are 11111.2 and -118338,

respectively. A very high potential energy can be caused by overlapping

interactions in bonded interactions or very large coordinate values. Usually

this is caused by a badly- or non-equilibrated initial configuration,

incorrect interactions or parameters in the topology.



Link to see the complete log file: https://utexas.box.com/s/eofa6ocil6mp947ahbvvtstt9qhnppsg



When I removed “-npme 1 -pme gpu”, the error doesn’t occur and the trajectory finish normally.

What it could be the cause of the error?

Thanks,

Alfredo