On Thu, Jan 11, 2024, Dulal Mondal via AMBER wrote:
>
>I am trying to run cuda.MPI in amber22 for remd (constant pH). I have 4
>replica. I do the following command in script.
>
>mpirun -n $SLURM_NTASKS $exe -ng 8 -groupfile
>
>WARNING: There was an error initializing an OpenFabrics device.
>
>*ocal host: gpu021 Local device:
>mlx5_0--------------------------------------------------------------------------setup_groups:
>unexpected end of file in
>groupfile--------------------------------------------------------------------------MPI_ABORT
(a) Can you run any parallel Amber jobs, e.g. just a short run with
pmemd.MPI?
(b) if other parallel jobs run OK, do the same with a simple run using
pmemd.cuda.MPI.
(c) Finally, look carefully at your groupfile -- note that the message above
reports an unexpected end of file there. Post that file if you can't figure
out what the error is.
...good luck...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 11 2024 - 11:30:04 PST
