use openmp cannot reach 100% cpu usage

I call a mex function that using openmp for parallel computing. My CPU has 8 physical cores, so I set the openmp parallel thread number to 8. But when I run the mex function, the CPU usage is just 50%. I change the thread number to 4, run the mex function from 2 different matlab processes simultaneously or use parfor with 2 workers, then the CPU usage reaches 100%. Is there any way to make the CPU reach 100% usage in one process?

[James Tursa]

Can you post a small subset of your source code that reproduces the problem?