Hi,
On Thu, Sep 30, 2010 at 2:25 AM, Vijay Manickam Achari
<vjrajamany.yahoo.com> wrote:
> I found no difference in the output files by commands sander.MPI and pmemd using 16 nodes. Both showing same temperature, total energy, kinectic energy, potential energy and so on.
>
> BUT there is a difference in the output files when I use 32 nodes using both commands. I noticed in the output file generated by command sander.MPI the temperature of NSTEP = 1 is 349.14K and Etot = 34557.6758 EKtot = 27485.1983 EPtot = 7072.4775.
So just to be clear, when you run with 16 nodes do things behave well
(like the first SANDER run) or do they crash out (like the second
PMEMD run)?
>
> NSTEP = 1 TIME(PS) = 1950.001 TEMP(K) = 349.14 PRESS = -272.7
> Etot = 34557.6758 EKtot = 27485.1983 EPtot = 7072.4775
>
> NSTEP = 1 TIME(PS) = 1950.001 TEMP(K) = 367.51 PRESS = -94.0
> Etot = 36003.2544 EKtot = 28930.7768 EPtot = 7072.4775
The same PE is being calculated indicating the coordinates and energy
functions are not the issue. The problem seems to be with the
velocities, but those should be the same as well since they should be
read in from the input coords. I wonder if the issue could be the
Anderson thermostat? As a test could you rerun those two short
simulations using the Berendsen (ntt=1) or Langevin (ntt=3) temp.
regulation instead and see if that changes things?
Also, did you apply all the bugfixes and run all the test cases for
PMEMD? If for some reason PMEMD didn't compile correctly that might be
a problem. Could you provide the topology and coordinate files you are
using to see if this error is reproducible on multiple compilations?
Good luck,
-Dan
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 30 2010 - 06:30:08 PDT
