Bandwidth analysis of m–Level HMPMB interconnections for multiprocessors

Abstract

Translate Article

The interconnection network (IN) is one of the major factors which determine the performance of a multiprocessor system. It is imperative to have a model by which the efficiency of the IN can be correctly evaluated. This paper presents an m–level HMPMB multiprocessor system and analyses its bandwidth performance. Two models are developed for the hierarchically equally likely case and favourite case, respectively. They show that the m–level HMPMB multiprocessor system performs well and is fairly close to a typical memory–oriented partial multiple bus system (MPMB) for high rates of favourite requests.