Abstract
We will present a new fast binary multiplication scheme based on the use of a nonholographic parallel optical content addressable memory (CAM). The multiplication operation is performed by means of binary logarithmic addition that uses a sign/logarithm number system. Multiplication of two binary numbers a and b begins by converting them into a sign/logarithm number system. Multiplication is accomplished by adding appropriate logarithms. A 3-stage non-holographic CAM is required to implement a sign/logarithm number multiplier. By means of a Quine-McCluskey minimization method, the number of CAM's minterms are reduced. For a 7-bit binary multiplication the first CAM, reduced from 605 to 111 minterms, converts the binary numbers to sign/logarithm numbers. To add the two logarithms, a second CAM, reduced from 2519 to 270 minterms, performs a floating-point binary-carry look-ahead addition. Finally, a third CAM with 329 instead of 891 minterms, does the conversion from sign/logarithm numbers back to binary numbers. In general, the storage capacity needed for each CAM stage depends on the range of input numbers and the calculation accuracy.
© 1990 Optical Society of America
PDF ArticleMore Like This
Andrew Kostrzewski, George Eichmann, Dai Hyun Kim, and Yao Li
MA5 Optical Computing (IP) 1991
Andrew Kostrzewski, Dai Kyun Kim, Yao Li, and George Eichmann
TuUU4 OSA Annual Meeting (FIO) 1990
George Eichmann, Andrew Koslrzewski, Dai Hyun Kim, and Yao Li
TuUU8 OSA Annual Meeting (FIO) 1990