CAM-Based Array Converter for URR Floating-Point Arithmetic

Kuei-Ming LU  Keikichi TAMARU  

IEICE TRANSACTIONS on Information and Systems   Vol.E81-D   No.10   pp.1120-1130
Publication Date: 1998/10/25
Online ISSN: 
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Computer Applications
floating-point arithmetic,  content addressable memory,  iterative array,  URR format conversion,  

Full Text: PDF>>
Buy this Article

In order to lessen overflow or underflow problem in numerical computation, several new floating-point arithmetics have been proposed. The significant advantage of these new arithmetics is that a number can be represented in a wider range since the fields of exponent and mantissa are changed depending on the magnitude of number. The main issues of these arithmetics are how to find the boundary between exponent and mantissa as well as to convert the formats between new floating-point arithmetic and fixed-point arithmetic quickly. In this paper, a CAM-based array converter based on the Universal Representation of Real number (URR) floating-point arithmetic is described. Using match retrieval device CAM, the detection of the boundary can be accomplished faster than conventional circuits. Arranging the basic cells into iterative array structure, the fast separation/connection operation is achieved. The speed, area and power consumption of the converter is estimated.