OpenMP Amdahl's Efficiency

OpenMP Amdahl's Efficiency indicates serial computation and can be computed as follows:

$\begin{eqnarray*}Amdahl's\; Efficiency=\frac{parallel\; execution\; time}{total\; runtime} \end{eqnarray*}$

where parallel execution time is a time spent in OpenMP parallel regions. Amdahl's Efficiency computed per MPI rank, significant difference between MAX and MIN values indicate that some process have bigger serialisation part than others.