[gmx-users] GMX 2020 - COMM Removal Issue

[Daniel Kozuch]

Hello,

I am experimenting with GROMACS 2020. I have compiled the mpi threaded
version and am using the new settings
(GMX_GPU_DD_COMMS, GMX_GPU_PME_PP_COMMS, GMX_FORCE_UPDATE_DEFAULT_GPU) as
suggested on at the following link:
https://devblogs.nvidia.com/creating-faster-molecular-dynamics-simulations-with-gromacs-2020/.
I am running mdrun  with the suggested options:
"-pin on -nb gpu -bonded gpu -pme gpu -npme 1 -nstlist 400" on 4 GPUs and
28 CPUs with "-ntmpi 4 -ntomp 7".

I am currently running a membrane system with a transmembrane protein in
water solvent. I am using the following settings for COM removal:

comm_mode  = linear
comm_grps  = PROT_MEMB SOL_ION

Here I choose to couple the the protein and the membrane from the advice in
this thread:
https://mailman-1.sys.kth.se/pipermail/gromacs.org_gmx-users/2016-September/108584.html

Unfortunately, I still see a large drift in the z-dimension for the entire
membrane/protein group. Currently I have nstcalcenergy/nstcomm set to 100,
as decreasing them leads to poor performance. (Hopefully it is unnecessary
to set them to 1)

Does anyone have suggestions for avoiding the COM drift? I know this issue
has been discussed before (https://redmine.gromacs.org/issues/2867) but it
looks like it was resolved in earlier GROMACS versions. As a note, I am
using a CHARMM force field with CMAP dihedrals.


Here are some other potentially relevant mdp options (from CHARMM) in case
they help:

integrator              = md
dt                      = 0.002
nstcalcenergy           = 100
;
cutoff-scheme           = Verlet
nstlist                 = 20
rlist                   = 1.2
coulombtype             = pme
rcoulomb                = 1.2
vdwtype                 = Cut-off
vdw-modifier            = Force-switch
rvdw_switch             = 1.0
rvdw                    = 1.2
;
tcoupl                  = v-rescale
tc_grps                 = PROT_MEMB   SOL_ION
tau_t                   = 1.0    1.0
ref_t                   = 303.15 303.15
;
pcoupl                  = Berendsen
pcoupltype              = semiisotropic
tau_p                   = 5.0
compressibility         = 4.5e-5  4.5e-5
ref_p                   = 1.0     1.0
;
constraints             = h-bonds
constraint_algorithm    = LINCS
continuation            = yes

Best,
Dan