Re: [AMBER] MPI_ABORT errorcode 1 for sander.MPI

From: David A Case <>
Date: Wed, 15 Jun 2011 15:46:36 -0400

On Wed, Jun 15, 2011, Erin Kelly wrote:

> Sorry again---he correct output is simply what is pasted in the main text of
> the *email*.

I think maybe you had better start over...the amber web page is down, and I
and others may not have any access to earlier emails.

My reason for suggesting a short, interactive, serial run is just to really
see what the error is, without many, many lines of MPI stuff, and the
possibility that error messages are being lost or misfiled. Plus, it rules
out possible problems that could show up with the 32-processor parallel run.

If you indeed get the "should not be here" error in your mdout file, then
please try the suggestions in *my* earlier email:

> >> peek_ewald_inpcrd: SHOULD NOT BE HERE
> >>
> >> This indicates that the program could not process your inpcrd file
> >> (3OGP_equil.rst) correctly. Are you sure that this is really a
> >> restart file from a previous MD (not minimization) run? Is there any
> >> chance that your file naming scheme got mixed up, or something like
> >> that?
> >>
> >> Look at the inpcrd file, especially the last few lines (you might want to
> >> post those). Compare to a restart file from some run that works.
> >>


AMBER mailing list
Received on Wed Jun 15 2011 - 13:00:04 PDT
Custom Search