Modeling and analysis for reconfigurable cell based on stochastic Petri net

Abstract

Reconfigurable architecture is an important trend of current high-performance parallel computing. Reconfigurable cell is the critical part of a prototype system of the new-concept, high-efficiency and new-type computer. Currently, such prototype system is at the stage of feasibility analysis and experimental research. This paper proposes a method to model the reconfigurable cell based on stochastic Petri net. For its structural properties, such as accessibility, boundedness and safety, we present the logical correctness verification. As to its performance properties, such as stable state probability, transition rate and throughput, we present the quantification analysis. Especially, this paper discusses the effects between dynamic reconfiguration time and task response time under three different conditions. The work of this paper may offer some reference value for the research and analysis of the coming new-type computer architecture. © 2011 Springer-Verlag.