DATA PROCESSING IN BULK TRANSFER SYSTEMS.

Digital bulk transfer systems are characterized by a truly massive, parallel data transfer, and by very limited logic modification of the data during the transfer. An examination is made of the characteristics and potential application areas of such systems, based on extrapolated capabilities of one particular embodiment of such a system that might be attached to a general-purpose computer. It is concluded that bulk transfer systems offer a substantial increase in performance over that achieved by conventional digital system organizations, but only for certain applications that 1 operate on a large, prestorable data base, 2 require only limited inputoutput to the store, and 3 can employ algorithms utilizing many parallel, identical operations. Such applications that have been examined include Fourier-transform spectrum analysis, the solution by relaxation methods of partial differential equations, image processing, and function interpolation over a large data base. Certain physical limitations are discussed that restrict the algorithmic flexibility of the system in its present conception, and suggestions are made for further research to avoid or circumvent these limitations. Author