> On Apr 8, 2015, at 11:50 PM, Abhishek TYAGI <atyagiaa.connect.ust.hk> wrote:
>
> Hi,
>
>
> 1. Dan: "It seems like somehow your cpptraj was built with MPI but ended up not
> being named cpptraj.MPI. What were the exact steps you used to
> configure and build amber?"
>
>
> I have two files named as cpptraj in the amberhome/bin/
>
>
> [atyagiaa.login-0 bin]$ $AMBERHOME/bin/cpptraj.MPI
>
> CPPTRAJ: Trajectory Analysis. V14.25 MPI
> ___ ___ ___ ___
> | \/ | \/ | \/ |
> _|_/\_|_/\_|_/\_|_
> Running on 1 threads
>> exit
> TIME: Total execution time: 84634.6979 seconds.
>
>
> [atyagiaa.login-0 bin]$ $AMBERHOME/bin/cpptraj
>
> CPPTRAJ: Trajectory Analysis. V14.25 MPI
> ___ ___ ___ ___
> | \/ | \/ | \/ |
> _|_/\_|_/\_|_/\_|_
> Running on 1 threads
This is a problem. I was interpreting your error message backwards -- Dan’s advice was the one to listen to.
This output shows that serial cpptraj is actually the MPI-parallel version. This should never happen (and indeed, I’ve never seen it happen), so something you did must have made cpptraj point to the parallel version (did you copy cpptraj.MPI into cpptraj?).
The solution is to reinstall Amber in *serial*.
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 Thu Apr 09 2015 - 05:00:02 PDT
