Re: [AMBER] Error in running AMBER 12 GPU simulations

From: David A Case <david.case.rutgers.edu>
Date: Thu, 31 Dec 2015 11:22:59 -0500

On Thu, Dec 31, 2015, Yogeeshwar Ajjugal wrote:
>
> Iam trying to run production step in GPU (AMBER 12). At the end
> of the job i can see some information in output file but there is 0 size
> in rst and crd file. i think its not running MD step. Iam using CUDA-5.5
> version, Here iam attaching my Pbs script and output file. please kindly
> help me, any suggestions can be appreciated.

Usual debugging instructions: try a short simulation (say nstlim=100),
printing every step (ntpr=1). Do this first in serial on a CPU, then (if that
works), try on a GPU (also in serial).

Try to avoid complex submission scripts, because you may loose error outputs.

There are very few circumstances where using the MPI version of pmemd.cuda
is worthwhile: certainly avoid this until you are sure that everything is
working. Even after you get everything working, you will almost certainly
want to avoid parallel GPU jobs.

...good luck....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 31 2015 - 08:30:03 PST
Custom Search