ABSTRACT
1 INTRODUCTION
Our desire for faster computation seems insatiable. Computer architecture has evolved along several paths, all of which ultimately are guided by economics. In the sequential process model, known today as the von Neumann model, both data and instructions are stored in memory. A bottleneck of this architecture is that caused by the memory access circuits; all activity must pass through this portion of the machine. The sequential nature of program execution also limits speed. John von Neumann, a mathematician, did not create the so-called "von Neumann bottleneck." It was created by the economics of pre-integrated-circuit technology and is being destroyed by the economics of very large scale integration (VLSI). Electronic circuits are nearing their asymptote in switching speed. Speed-ups in uniprocessors are getting smaller, therefore, while the costs of attaining the speed-ups are getting larger. Concurrent process architecture may prove to be a path of less economic resistance. VLSI technology is making feasible the physical construciton of cellular automata, a highly concurrent architectural class. Interestingly, cellular automata were first, investigated by John von Neumann [1] . Technology and economies are just now catching up.
