Optimizing GROMACS for md simulation

GROMACS version: 2024.2
GROMACS modification: Yes (CUDA Support)
I am running GROMACS with NVIDIA gpu support compiled using CUDA. Is there a way of distributing tasks between the gpu and cpu for best performance?

Yes! Take a look at all the mdrun flags.

please have a look here too