Re: [AMBER] segmentation fault using GPUs

From: Marc van der Kamp <marcvanderkamp.gmail.com>
Date: Fri, 16 Dec 2011 12:10:22 +0000

Hi,

Before the experts (which I am not!) need to jump in, the obvious questions
are:

- Does the same input and command run ok with pmemd and/or sander (on one
CPU)?
- Have you applied all the bugfixes (to AT1.5 and amber11) before
compiling pmemd.cuda.MPI ?

It may be useful if you try and run the command with an input that
specifies just a few hundred steps and prints energies etc. every step.

What struck me about what you describe is that you seem to be trying to
restart a simulation after doing a (back)mutation using velocities from a
previous production run. Are you sure that this can work, i.e. how do you
make sure the atoms that stay the same get the correct velocities, and how
are velocities assigned for the atoms that changed through mutation? (I'd
expect that at the least, your residues before and after mutation need to
have the same number of atoms?)

--Marc

On 16 December 2011 11:38, Ouaray Z. <zo1g10.soton.ac.uk> wrote:

> Hi,
>
> I am trying to run a simulation from the last structure of a previous MD
> simulation (100ns), where the mutated struture is mutated back to the
> wild-type (that concern only one residu). i prepared the back mutant via
> tleap (keeping all the parameters from the last simulation: water box, ion,
> zinc parametrization) and i did a minimization only on the residue mutated.
> but once i try to run the simulation (using GPUs) with this command:
>
> time mpirun -np 2 pmemd.cuda.MPI -O -o MutChang_MD.out -i mdrun.in -p
> MutChang.parm7 -c MutChangVMin.rst -ref MutChangVMin.rst -r MutChang_MD.rst
> -x MutChang_MD.mdcrd
>
> i have this error message:
>
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> --------------------------------------------------------------------------
> mpirun has exited due to process rank 1 with PID 7062 on
> node purple007 exiting without calling "finalize". This may
> have caused other processes in the application to be
> terminated by signals sent by mpirun (as reported here).
> --------------------------------------------------------------------------
>
> I attached the output file, tleap, minimization file and the MD file.
> can someone help me with this please?
> Kind regards,
> Zohra.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Dec 16 2011 - 04:30:02 PST
Custom Search