Browse Prior Art Database

Fast Floating Point Operations

IP.com Disclosure Number: IPCOM000077612D
Original Publication Date: 1972-Aug-01
Included in the Prior Art Database: 2005-Feb-25
Document File: 2 page(s) / 32K

Publishing Venue

IBM

Related People

Gardner, PL: AUTHOR

Abstract

In floating-point operations it has been observed that the characteristics of the operands differ by at most one in 728 of cases. An ALU is provided which gives the correct result in these cases. If as a result of comparison of characteristics, which takes place simultaneously with the arithmetic operation, the characteristics are found to differ by more than one, the result is suppressed and the fractions are aligned.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 2

Fast Floating Point Operations

In floating-point operations it has been observed that the characteristics of the operands differ by at most one in 728 of cases. An ALU is provided which gives the correct result in these cases. If as a result of comparison of characteristics, which takes place simultaneously with the arithmetic operation, the characteristics are found to differ by more than one, the result is suppressed and the fractions are aligned.

The ALU comprises an associative store 1 containing tables I to III, identified, respectively, by keys 00, 01 and 10, which give the correct result by table look-up if, respectively, the characteristics A and B are equal, B is one less than A, and A is one less than B. A second associative store 2, operating faster than store 1, uses the two low-order bits of the characteristics to supply to store 1 the appropriate key. The table in store 2 is based on the fact that

(Image Omitted)

1

Page 2 of 2

2

[This page contains 2 pictures or other non-text objects]