Abstract

t. Hence, an algorithm that takes time t on a P-processor machine performs work W = P t. In either case, the work roughly captures the actual cost to perform the computation, assuming that the cost of a parallel machine is proportional to the number of processors in the machine. We call an algorithm work-efficient (or just efficient) if it performs the same amount of work, to within a constant factor, as the fastest known sequential algorithm. For example, a parallel algorithm that sorts n keys in O( p n log n) time using p n processors is efficient since the work, O(n log n), is as good as any (comparison-based) sequential algorithm. However, a sorting algorithm that runs in O(log n) time using n 2 processors is

Citations