Execution Time t(np) is the time needed to complete execution when running on n processors. Speedup s(np) is defined as t(1)/t(np) , which means how much faster will the model perform a simulation when using np processors rather than just only one (serial version); valid values are within 1 to np.
Another performance parameter to be considered is the Utilization (B.1), representing the average fraction of time a processor is doing useful work, as compared to executing on one processor. We will always try to have as much utilization as possible, although 100% utilization is almost impossible to obtain due to the delays needed at task synchronization points.
