An algebraic theory for parallel processor design
Josef Jelínek · The Computer Journal · 1979
One specific feature of parallel processors is the possibility of transferring data among cell processors even during some instructions execution. This data transfer determines the nature of many instructions and prevents a reasonable description of parallel processors behaviour by means of Boolean algebra. This paper presents an algebraic theory that may serve as a tool for parallel processor design. Its power is demonstrated by analysing the complete instruction set of the CLIP3 parallel image processor. The investigation is completed by a discussion of how an image parallel processor should look.