hii,
I am working on a metalloprotein HCAII, which contains zinc. In one
configuration zinc is bonded with three nytrogen atoms and a oxygen atom of
water molecule and in another it is bonded with three nytrogen atoms and a
oxygen atom of hydroxide ion. I am doing simulation by following steps. 1)
minimization, 2) heating, 3) NVT equilibration . 4) NPT equilibration 5)
NVT production. In first case I am getting error during NPT equilibration
but in second system I got in heating step. (minimization has been done by
pmemd.cuda only). In both case I got same error message but in first case
one message extra which is coming during NPT equilibration.
* ERROR: PMEMD does not support intermolecular PRFs*!
I wrote other error message in my earlier mail. So, is the error coming due
to my systems which contain four metal-nonmetal bond? Or, is there any
other reason? I am trying to trying to compile MPI version of pmemd.cuda by
other MPI rather than OpenMPI.
Thanking you,
Sanjib
On Tue, May 20, 2014 at 7:53 PM, Daniel Roe <daniel.r.roe.gmail.com> wrote:
> On Tue, May 20, 2014 at 7:26 AM, Sanjib Paul <sanjib88paul.gmail.com>
> wrote:
> > *Warning: Conflicting CPU frequencies detected, using:
> 2600.000000.Warning:
> > Conflicting CPU frequencies detected, using: 2600.000000.*
> > In NPT it stopped after giving some message -
>
> This warning isn't from Amber - a quick google search appears to
> indicate that it's related to OpenMPI. From what I gather it probably
> won't affect your results, except maybe for timing data, though you
> should probably run some basic tests of your MPI setup. My
> recommendation would be to switch to a different MPI like mpich - I've
> never had great luck with OpenMPI.
>
> > ERROR: PMEMD does not support intermolecular
> >
> PRFs!--------------------------------------------------------------------------mpirun
>
> If your system has bonding between different molecules and has extra
> points, this is not supported for NPT runs.
>
> > *gpu_download_partial_forces: download failed unspecified launch
> >
> failure--------------------------------------------------------------------------mpirun
>
> The GPU code is more sensitive to bad initial configurations than the
> CPU code. I recommend trying a short run with the CPU code to better
> diagnose what is happening with the system that fails.
>
> Note that unless you have multiple GPUs in a single box both connected
> to PCI-E 2.0 x 16 or better, you probably aren't going to gain much
> running pmemd.cuda.MPI for a single MD job. Make sure you've read up
> on all the information here:
>
> http://ambermd.org/gpus/
>
> particularly the section 'Running GPU Accelerated Simulations'.
>
> Hope this helps,
>
> -Dan
>
> --
> -------------------------
> Daniel R. Roe, PhD
> Department of Medicinal Chemistry
> University of Utah
> 30 South 2000 East, Room 201
> Salt Lake City, UT 84112-5820
> http://home.chpc.utah.edu/~cheatham/
> (801) 587-9652
> (801) 585-6208 (Fax)
>
> _______________________________________________
> 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 Tue May 20 2014 - 08:30:07 PDT
