Synchronous Nets for Single Instruction Stream - Multiple Data Stream Computers

摘要

Synchronous Nets, or S Nets, are developed as a modeling tool particularized for describing processes on Single Instruction Stream - Multiple Data Stream (SIMD) computers. S Nets are a modification of Petri Nets, using transitions and places to model events and conditions. However S Nets introduce vector-mask places to model the conditions of the array resources of SIMD machines. These places are distinguished from scalar places which model the scalar resources. S Nets also introduce a new kind of transition. One type correlates with the Petri Net transition, but the mask firing transition is particularized to the SIMD environment, modeling the inherent capability of a computation executing on a SIMD machine to alter the participation of the vector aggregates in successor events. (Author)