GROMACS version:
GROMACS modification: Yes/No
Here post your question
Replacing old mdp entry ‘verlet-buffer-drift’ by ‘verlet-buffer-tolerance’
Replacing old mdp entry ‘nstxtcout’ by ‘nstxout-compressed’
Replacing old mdp entry ‘xtc_precision’ by ‘compressed-x-precision’
NOTE 1 [file grompp.mdp]:
With Verlet lists the optimal nstlist is >= 10, with GPUs >= 20. Note
that with the Verlet scheme, nstlist has no effect on the accuracy of
your simulation.
WARNING 1 [file grompp.mdp]:
The Berendsen thermostat does not generate the correct kinetic energy
distribution, and should not be used for new production simulations (in
our opinion). We would recommend the V-rescale thermostat.
The replacing notes are just minor details regarding the change of wording of some keywords in the mdp file, e.g., back in the day it was xtc_precision and now it is compressed-x-precision. Probably you are using an old GROMACS mdp file, per se it’s not a problem as long as GROMACS will handle these internally.
The first note is telling you that nstlist can be larger as you are running on GPUs. Usually this value is overwritten to something > 100 by GROMACS with a corresponding rescaling of the rlist, you can see this in the output on the screen and on the .log file of the simulation.
The warning is importan: never use berendsen neither as thermostat nor as barostat. Back in the day it was okay to switch between berendsen and nose-hoover for thermostats and berendsen and parrinello-rahman for barostats when going from equilibration to production, but nowadays it is generally better to use the newer v-rescale and c-rescale which are very robust for equilibration and produce the correct ensembles for production.