[gmx-users] restarting run on GPU

Mark Abraham mark.j.abraham at gmail.com
Fri Mar 28 11:51:54 CET 2014

Changing the cutoff scheme does not permit a logically continuous
simulation; the trajectory is not continuous, and the ensemble is not
necessarily the same one. You should conduct an equilibration and view the
trajectory as a new one.

The are numerous design features of the group cutoff scheme that cannot /
should not / will not be implemented (efficiently) with the Verlet scheme
on grounds of correctness or performance. Your problem could relate to .mdp
option choices, simulation size, hardware, drivers, or compilation issues,
so we'll need a lot more information before we might be able to help. A
description of your simulation, .mdp files, the output of mdrun -version
and the group and verlet scheme performance tables at the end of the .log
files are the minimum we might need from you.


On Fri, Mar 28, 2014 at 11:28 AM, pratibha <kapoorpratibha7 at gmail.com>wrote:

> I have generated my md_new.tpr using:
> grompp -f new.mdp -c npt.gro -t npt.cpt -p topol.top -o md_new.tpr
> where my new.mdp file has everything same except cutoff-scheme as verlet
> I am getting very bad performance even worse than only CPU simulations.
> My GPU is well tested before and was giving good performance. Is there
> something wrong in the procedure I have followed?
> --
> View this message in context:
> http://gromacs.5086.x6.nabble.com/restarting-run-on-GPU-tp5015444p5015451.html
> Sent from the GROMACS Users Forum mailing list archive at Nabble.com.
> --
> Gromacs Users mailing list
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.

More information about the gromacs.org_gmx-users mailing list