Exploring multi-level parallelism in cellular automata networks

Abstract

Usually physical systems are characterized by different cou-pled parameters accounting for the interaction of their different compo-nents. The Cellular Automata Network (CAN) model [1] allows to repre-sent each component of a physical system in terms of Cellular Automata (CA) [9], and the interaction among these components in terms of CA networks. In this paper we report our experimentations in exploiting two different kinds of parallelism offered by the CAN model using policies for network restructuring and thread assignment. At this purpose we used a prototype graphic tool (CANviz) designed to let the user experimenting heuristics to efficiently exploit two-level parallelism in CAN applications.