Browse Prior Art Database

Biased and Non Biased Binary Conversion and Exponent Under Over Flow Detection

IP.com Disclosure Number: IPCOM000055261D
Original Publication Date: 1980-Jun-01
Included in the Prior Art Database: 2005-Feb-13

Publishing Venue

IBM

Related People

Authors:
Hale, JL Letteney, RC Shen, DT Tan, KG [+details]

Abstract

In IBM System/370 processors, the exponent, bits 1-7, of a floating-point number, is expressed in excess-64 notation. The excess-64 notation is formed by adding 64 to the actual binary value of the exponent. Therefore: EXPONENT/64 = EXPONENT/0 + 64 (*1) or in binary form: EXPONENT/64 = EXPONENT/0 + 1000000 Let EXPONENT/0 = 1XXXXXX, where X = 1 or 0 then, EXPONENT/64 = EXPONENT/0 + 1000000 = 1XXXXXX + 1000000 = Carry out of Bit-1 + 0XXXXXX.