In recent years researchers have come to believe that in spite of the continued advances in integrated circuit technology, there will be an upper limit to the speed of sequential processors, i.e. computers that process one instruction at a time. An alternate strategy to increase computational speed is to use parallel processing, the use of simultaneous processing of separate streams of data. Thus a most promising line of research would be the development of algorithms for multiprocessors. This equipment grant will provide partial support for the purchase of a 32-node hypercube system by an established group of engineers working in computational mechanics. They have extensive experience with sequential computations but now wish to explore parallel systems. The institutional support is adequate, I recommend support.
