Dear all

I'm a LAMMPS user, dealing with large-scale simulation, over 100,000 particles and >100 ns.

Inevitably, I must utilize the GPU accelerating power to speed up my running performance.

Firstly, I'd like to know: Which programs is faster in GPU calculation, GROMACS or LAMMPS?

More importantly, I've realized that LAMMPS cannot implement some specific pair styles and also hybrid ones in GPU. Does GROMACS incur such inability?

Also I'd like to use GPU for TIP4P water model, but LAMMPS cannot. What about GROMACS?

Any hint is welcome.