[gmx-users] Load imbalance and Y size error
Mark.Abraham at anu.edu.au
Thu Oct 8 14:46:03 CEST 2009
Berk Hess wrote:
> Do not "randomly" change the fourierspacing.
> Are you really using PME in vacuum?
> The box size can only change if you have pressure coupling turned on,
> which should, obviously, should not do in vacuum.
This suggests adding a warning about pressure coupling when the atomic
distribution is strongly non-homogeneous... but what's an easy test for
that? Is a 3x3x3 grid search in grompp out of the question? More than
50% variation in atom number density per grid volume would probably pick
up vacuum simulations but not show false-positive membrane simulations.
More information about the gromacs.org_gmx-users