[AMBER] GPU error in the heating step of PMEMD.CUDA

From: <rosaester.forgione.unina.it>
Date: Mon, 22 Jun 2020 13:05:51 +0200


I am facing problems in running MD dynamics by using PMEMD.CUDA

During the 300 K heating step (NPT) the following error appears:

ERROR: Calculation halted. Periodic box dimensions have changed too
much from their initial values.
   Your system density has likely changed by a large amount, probably from
   starting the simulation from a structure a long way from equilibrium.

   [Although this error can also occur if the simulation has blown up
for some reason]

   The GPU code does not automatically reorganize grid cells and thus you
   will need to restart the calculation from the previous restart file.
   This will generate new grid cells and allow the calculation to continue.
   It may be necessary to repeat this restarting multiple times if your system
   is a long way from an equilibrated density.

   Alternatively you can run with the CPU code until the density has converged
   and then switch back to the GPU code.

However, the MD simulations goes on passing to the production steps.

I tried to restart the simulation from the previous restart file, as
suggested by the error message, but it doesn't work since the same
error appears.

I used the same equilibration protocol that I routinely use to carry
on MDs on ligands alone (glycans) and protein-ligand complex.

I tried to run a MD simulation on an "old" system, already analyzed
with amber without any issue, and the same error appears. Thus, it
seems that the problem is not strictly related to the parameters of
the system.
I also run the simulation on GPU 0 and then on 1 and the error is still there.

Anyone can suggest how to fix this problem?

Thanks in advance,


AMBER mailing list
Received on Mon Jun 22 2020 - 04:30:03 PDT
Custom Search