How could I implement minimization step with openmpi

GROMACS version: 2020.2
GROMACS modification: Yes/No
Hi GROMACS users.
I’ve tried to energy minimization simulation with my system. But when I prompt command ‘top’, my machine used only one thread’ with mdrun -ntomp 4 option as from the user manual.
I want to implement energy minimization step more faster.
In my machine, openmpi-4.0.2 was installed and cpu with 12 thread and one gpu . How can I implement more faster minimization step with openmpi?

Thank you :)