By default, the LF solver use a single core of the CPU, so the total CPU usage can appear quite low if you have several cores.
In the "Solver" settings of the simulation, you can choose to use more than 1 core. This relies on the MPI library to parallelize the workload between different CPU cores.
The speedup you gain with MPI depends on many parameters, including discretization of the computational domain (e.g. number of grid cells) and hardware specs (e.g. available RAM). The only reliable way to know is to try on a few representative examples, with different number of CPU cores.
Note that the MPI parallelization requires specific license tokens. If you try and it fails with licensing errors, you can always request a trial license for this feature.