Index Dependent Nested Loops Parallelization with an Even Distributed Number of Steps

Abstract

Parallel processing of algorithms is an effective way to achieve higher performance on multiprocessor systems rather. During parallelization, it is critical to minimize the difference between the processing time for threads. It is necessary to choose a method that can efficiently distribute the workload evenly across the threads. This paper deals with a special kind of nested loops where the internal loop iterator depends on the outer loop iterator. In such cases, the process can be represented as an upper (or lower) triangular matrix. This paper introduces a method for partitioning the outer loop according to the indices in an almost optimal manner, so that the partial loops in each thread will take nearly the same number of steps. In addition, we examine the potential of a perfect partition and try to determine the maximum (but still meaningful) partition size.Povzetek: Predstavljena je metoda paralelizma za posebno vrsto vgnezdenih zank.