[gmx-users] V-rescale thermostat, PME, Estimate for the relative computational load of the PME mesh part: 0.97

Justin A. Lemkul jalemkul at vt.edu
Tue Jan 25 14:05:44 CET 2011

gromacs wrote:
> HI Friends,
> I get the following note,
> The Berendsen thermostat does not generate the correct kinetic energy
>   distribution. You might want to consider using the V-rescale thermostat.
> I want to keep the T at 300K, so does it matter to select any thermostat 
> method?

The choice of thermostat certainly does matter, otherwise you wouldn't get this 
note.  Refer to the numerous discussions in the list archive as to why one would 
or would not (usually) use the Berendsen thermostat, as well as:


> Another note when i use PME:
> Estimate for the relative computational load of the PME mesh part: 0.97
> NOTE 1 [file aminoacids.dat, line 1]:
>   The optimal PME mesh load for parallel simulations is below 0.5
>   and for highly parallel simulations between 0.25 and 0.33,
>   for higher performance, increase the cut-off and the PME grid spacing
> So what is the reason? I use type=PME

Your combination of settings (rcoulomb, fourierspacing, and perhaps a few 
others) indicate that your simulation is going to spend an inordinate amount of 
time doing PME calculations, so your performance will suffer.  Seeing your 
entire .mdp file would be necessary if you want further guidance.


> Is my setting proper?
> Thanks


Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list