Parallel Computing

This book offers an approach to understanding the principles of parallel computing, derived from the author's experience of building the Texas Reconfigurable Array Computer (TRAC). It provides the principles for the analysis and comparison of parallel computers, and stresses two main ideas, an extension of the time-space product for measuring performance of a system and a graph-theoretical definition of interconnection networks. The book presents these two ideas as the theoretical basis for comparing parallel computers, and then applies them to TRAC and other parallel computers to illustrate the machines' utility. The designs of parallel computers are then analyzed by evaluating the essence of their architecture.