Confusing and possibly incomplete update on gpu error message in gmx2022.2

User discussions
1

GROMACS version: 2022.2
GROMACS modification: No

I got this error message when attempting gmx mdrun -gpu_id 01 -nb gpu -pme gpu -bonded cpu -update gpu -npme 1 -ntmpi 8 -nt 16 -s this.tpr

Inconsistency in user input:
Update task on the GPU was required,
but the following condition(s) were not satisfied:
Domain decomposition is only supported with constraints when update groups are
used. This means constraining all bonds is not supporters, except for small
molecules, and box sizes close to half the pair-list cutoff are not supported.

Q1: Does the final line really mean that box sizes close to double the cutoff are not supported? Otherwise, can somebody please help me understand this message? I thought the box had to be bigger than the pair-list cutoff always.

Q2: I am using constraints=hbond and the box is 10x10x12 nm with a 1.4 nm cutoff, so it’s not clear to me how either of the above conditions apply. Are there also other (unlisted) conditions that would cause this to fail like this?

PS: for what it’s worth, I much preferred the mailing list to this online forum. I suggest consider reverting.

Thank you,
Chris.

2

You’re right, the wording is incorrect. Thanks for bringing it up.

There are a few others, though they should be very rare. Do you have any messages starting with “When checking whether update groups are usable:” in your md.log?

You can leave your feedback in the dedicated sub-forum: Site feedback - GROMACS forums.

3

Thanks for the reply. I do get the following message, which is more than I noticed before but not particularly precise. This system does use constraints within the protein. Could that be it?

When checking whether update groups are usable:
At least one moleculetype does not conform to the requirements for using update groups

4

This system does use constraints within the protein. Could that be it?

Likely. The update group mechanism (which we use for handling the constraints on the GPU) is pretty picky about the allowed configurations of constraints.

5

When I get this same error for the update groups despite only having h-bond constraints I see this:

When checking whether update groups are usable:
  At least one moleculetype does not conform to the requirements for using update groups

I updated the oplsaa forcefield atomtypes to add in small-molecule ligand but didn’t do anything to the moleculetypes. I am using spce water.

This is the mdp file:

integrator              = md
dt                      = 0.002
nsteps                  = 500000
nstxtcout               = 50000
nstvout                 = 50000
nstfout                 = 50000
nstcalcenergy           = 100
nstenergy               = 1000
nstlog                  = 1000
;
cutoff-scheme           = Verlet
nstlist                 = 20
vdwtype                 = Cut-off
vdw-modifier            = Force-switch
rvdw_switch             = 1.0
rvdw                    = 1.2
rlist                   = 1.2
rcoulomb                = 1.2
coulombtype             = pme
;
tcoupl					= V-rescale
tc_grps                 = Protein Non-Protein
tau_t                   = 1.0 1.0
ref_t                   = 300 300
;
pcoupl                  = Parrinello-Rahman
pcoupltype              = isotropic
tau_p                   = 5.0
compressibility         = 4.5e-5
ref_p                   = 1.0
;
constraints             = h-bonds
constraint_algorithm    = LINCS
continuation            = yes
;
nstcomm                 = 100
comm_mode               = linear
comm_grps               = Protein Non-Protein
;
6

This could be caused by topology ordering, when some hydrogen are not directly after the heavy atom they are bound to. If that is the case reordering the topology should fix the issue.

1 Like