Ternary Optical Computer Integer Multiplication Routine

Multiplication is the important functions required for ternary optical computer when it is appliedin the numerical computing, it is the key operation to do matrix transformation, linear equationssolving, digital signal processing and graphic image processing procedure, etc. However, there isno hardware unit to achieve multiplication function yet in the SD11, the latest generationexperimental system of ternary optical computer. Therefore it is essential to find a feasiblesolution to realize the multiplication function on SD11. Multiplication function can be achievedvia hardware or software, while the way through software to achieve is relatively easy as thecurrent research situation of ternary optical computer. Multiplication routine is a certain softwareprogram to achieve multiplication function. It is packaged in the form of the foundation library,and calling the three-valued logic processors and the MSD adder existed in SD11to completeinteger data multiplication function. It is a feasible and effective way in project implementation.According to the advantages of ternary optical computer, such as reconfigurable arithmetic unit,large number of data bits and no-carry-delay MSD adder, a40-bits multiplication routine isdesigned and developed. In the routine, the numerical value is expressed by the MSD numbersystem like in the ternary optical computer, the partial products are generated via a three-valuedlogic transform (M), and the product is obtained by summing all partial products through a MSDadder with the iterative method. A fast-calculation method is applied in the M transform and apipeline technology in the MSD adder to accumulate the partial products. Meanwhile, in thepaper, the operation steps and simulation experiments of the routine are given in detail, and theperformance comparing with electronic computer is analyzed.In this paper, the multiplication routine is designed and implemented according to structuralcharacteristics of SD11system closely; it has a high practical applicability. On the another hand,the speed of this routine is faster than electronic multiplier when calculating large value dataaccording to the multiplication principle revealed in this paper.