A search memory subsystem for a general-purpose computer

Abstract

The search memory provides an extremely powerful tool for data-processing applications in which files of data are manipulated---applications such as data correlation, information retrieval and data analysis. For many such problems, a moderate size search memory can increase the performance of a general-purpose computer by several orders of magnitude. This paper describes the system design of a search memory subsystem integral to a general-purpose computer. The unit will operate in much the same manner as a conventional arithmetic unit, receiving data directly from the main memory, and placing the results of a search in the accumulator. This generalized design is independent of the particular magnetic memory element used and is not constrained by any particular computer organization. However, values have been assigned to the various parameters in the system so that the concepts can be concretely described. The techniques and concepts presented in this paper are the result of a company-sponsored program. The following sections present an assumed general-purpose computer structure, the design of a search memory subsystem, and possible implementation. Appendix A presents the basic principles of the search memory organization.