On Sat, Apr 19, 2014 at 1:31 PM, Muthukumaran R <kumaran.bicpu.edu.in>wrote:
> Dear Amber users,
> I have been working on amber package for performing md simulations.
> recently i have started using cluster with 2 nodes with 64 cores each.
> I am eager to know whether a job run on 32 cores for few ns and later i
> continue the same job on 64 cores (or vice versa), will show any
> difference in the results in terms of energies or the conformational
> changes.
>
No, there will be no noticeable difference. But keep in mind that
parallel computations (at least in pmemd) are not reproducible due to
round-off differences that change from run to run.
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Apr 19 2014 - 15:30:02 PDT
