Effect of Forcefield/Water Model on Simulation Times (specifically with regard to amberff19sb/OPC)

GROMACS version:2022
GROMACS modification: No
I’ve been having an issue with switching GROMACS simulations from the amber99sb-ildn forcefield with TIP3P water to the amber19sb forcefield with OPC water. My understanding is that for the same system, using a 4-point water model would naturally be slower than the 3-point water model due to the extra site, but I’ve noticed that even a simulation with fewer atoms (47544) on amber19sb is taking twice as long as a simulation with a greater number of atoms (52863) on amber99sb-ildn, even when the extra site is accounted for. I was wondering whether this is to be expected or if this indicates some issue with how I’m setting up the simulation (I’ve been using CHARMMGUI to obtain the topology files but have kept to the same mdp settings I’d used for the 99sb-ildn forcefield adapted from the lysozyme tutorial, so far I haven’t done anything more complex than placing a protein in a water box)

Turns out it most likely has to do with the virtual sites not being supported with GPU updates (Virtual sites are not supported)