Bitte wählen Sie ihr Lieferland und ihre Kundengruppe
The authors described a parallel computing method for high speed calculation, in order to obtain knowledge of the limit of high-speed operation of the parallel computing by MIMD-type parallel processors. Generally, high-speed parallel computing can be realized by improving the paralleling ability and by shortening the operation interval. It is known that in order to improve paralleling ability, it is effective to provide satisfactory number of arithmetic operation units and start switching with the completion of data, but there are difficulties in shortening the operation interval. The authors showed that a method of shortening this operation interval exists if the necessary number of parallel units for given problems and a switching matrix with a similar number of processing units and incoming/outgoing lines is used.