Computers
...hort, then 10 of these may be able to produce 10X FLOPS. In the case of huge parallel processing, a thousand processors may produce 1000X FLOPS. The many problems of obvious interest for parallel processing because of their computational intensity are matrix inversion, artificial vision, weather forecasting, finite element analysis, real-time analysis of large data bases, computational fluid dynamics, computer animation, nuclear energy research, petroleum exploration, simulation, 3-D modeling, optical ray trace, signal processing, and optimization. Parallel processing, also known as concurrent computing, is not completely new in terms of its concept. For as long as there have been jobs that can be separated into multiple tasks that in turn can be handed out to individual workers for simultaneous performance, team projects have been an effective way to keep up with its tight and speedy schedule. In the world of computation, one recalls the depression era U.S. government’s Works Progress Administration (WPA) projects of the 1930s to generate trigonometric and logarithmic table that employed hundreds of mathematicians, each calculating a small portion of the total work. Lenses were designed the same way, with each optical engineer tracking one ray through a candidate design. The recent excitement for parallel computer architectures results from the fast rising demand for supercomputer performance and the simultaneous maturing of constituent computer technologies that make parallel processing supercomputers a real possibility. A supercomputer is the fastest among computers and is designed for one purpose; to perform specialized applications that require immense amounts of mathematical computations. It is the class of computers that share the features of high speed and large capacity compared with the average of what is available at any given time. Supercomputers are used for computationally intensive applications such as the examples ...