Assuming that there are N computations to be done, each taking one clock cycle in a processing unit (PU), the N computations can be done in T = N clock cycles with one PU. It is also possible to share the computations among M PUs so that the computation can be done in a shorter time; that is, T = N/M clock cycles. If M = N PUs are designed, then all the computation can be done in one clock cycle. This simple resource-time trading-off principle is probably one of the most useful methods of resource usage control in digital circuit design.