Arithmetic microsystems for the synthesis of function generators

The continuing reduction of the size and cost of integrated circuit logic elements encourages the utilization of more complex logic nets in digital computers. In arithmetic processors, the application of integrated circuits will permit the replacement of sequential logic nets by their combinational equivalents, and the replacement of programmed subroutines by arithmetic fraction generators. At the present time, such function generators are used in the Variable Structure Computer at the University of California, Los Angeles. The potentially low cost of logic elements can be realized if large numbers of elements are contained in one package with a limited number of connections to the outside, and if very few types of packages are used in large quantities. Present-day arithmetic processors cannot be readily subdivided into uniform packages because of irregularities in their internal structure. The paper describes the advantages gained by the application of signed-digit number systems in the design of such packages, called arithmetic microsystems. Three types of combinational microsystems are described and their internal design is illustrated by examples for a radix 16 system. The microsystems can accept conventional binary or signed-digit operands. Combinational reconversion of results to the binary system is also discussed. The application of microsystems is illustrated by radix 16 combinational arrays for the operations of multiplication and division.