Re: [AMBER] Regarding the MPI error in parallel running of 3D-RISM

From: David A Case <david.case.rutgers.edu>
Date: Thu, 28 Jun 2018 14:26:17 -0400

On Thu, Jun 28, 2018, PRITI ROY wrote:
>
> Thank you for your responce. I tried with both of the procedure which you
> have suggested, but getting same error. Is it due to MPI setup or incorrect
> input file?

See if reducing the number of MPI threads (say to 2 or 4) helps, or
changes anything. If not, see if the calculation works in serial mode.
If using a smaller number of threads does help, try to find the largest
number of threads that works.

Even with "bad" grid sizes, you should not get MPI crashes, so I suspect
that something else is wrong; but if not, we need to make the code more
robust.

...dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jun 28 2018 - 11:30:02 PDT
Custom Search