Dear Dan:
I do not think it is problem of nodes because I ran sander.MPI with 20 and
28 nodes with different mdin.
It was running fine. It is something else I am not able to figure out
presently.
On Wed, Jul 28, 2010 at 10:22 AM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
> Hi Shubhra,
>
> >From your mdout file:
>
> | Running AMBER/MPI version on 20 nodes
>
> | WARNING: The number of processors is not a power of 2
> | this may be inefficient on some systems.
>
> Try running a shorter version of the job with fewer processors (start
> at 2 or 4 and increase accordingly) and see if that fixes the issue.
> It's highly likely there are parts of the parallel code that aren't
> tested for a number of processors that are not a power of 2.
>
> -Dan
>
> _______________________________________________
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jul 28 2010 - 00:30:05 PDT
