Electrical computers: arithmetic processing and calculating – Electrical digital calculating computer – Particular function performed
Reexamination Certificate
1998-08-25
2001-01-23
Mai, Tan V. (Department: 2787)
Electrical computers: arithmetic processing and calculating
Electrical digital calculating computer
Particular function performed
C708S205000
Reexamination Certificate
active
06178437
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention relates to a method and apparatus for data processing in general, and in particular to a method and apparatus for performing result normalization in a floating-point processor. Still more particularly, the present invention relates to a method and apparatus for anticipating leading zeros/ones in a floating-point processor.
2. Description of the Prior Art
Normalization is the removal of leading zeros or leading ones from a respective positive or negative output of a floating-point adder. Full-precision leading-zero anticipators (LZA) (or leading-zero predictors) are commonly utilized to improve the speed of the normalization process. An LZA can be the most critical path of a floating-point adder because it is not obvious a priori whether the result from the adder will be positive or negative. Thus, it is necessary to perform both leading-zero and leading-one analysis, and select the proper normalization shift amount based upon the sign of the result when the result is finally available.
According to the IEEE 754 standard, floating-point numbers are represented by three elements, namely, a binary sign bit, a binary encoded exponent, and a binary encoded mantissa. In a normalized floating-point number, the exponent is that which ensures the first digit of the mantissa is a logical one (except for special cases such as zero, infinities, and unrepresentable numbers). During a normalized floating-point addition, one of the mantissas of the addend and adder is shifted and the exponent is incremented or decremented until the exponents for both the addend and adder are equal. This shifting process is known as alignment. Once aligned, the mantissas of the addend and adder are added or subtracted depending upon the signs of the addend and adder as well as the type of operation (either addition or substraction) to be performed. Once the result (either sum or difference) is formed, depending upon the operation, the sign of the resulting mantissa is examined. If the sign of the result is negative, the boolean complement of the result is initially formed and the sign is then complemented. In order to convert the result to a normalized form, the exponent of the result is decremented and the mantissa of the result is left-shifted until the leading digit of the mantissa is a logical one (in absence of exceptional conditions such as those mentioned supra).
The determination of how many leading zeros or leading ones need to be removed (i.e., the amount of left-shifting) is preferably accomplished in parallel with the arithmetical operations within the floating-point adder. Such concurrency is important because even a few clock cycles may have a major performance impact on the composite speed of the floating-point processor. The present disclosure describes an improved LZA architecture that provides a significant speed improvements over its predecessors.
SUMMARY OF THE INVENTION
In view of the foregoing, it is therefore an object of the present invention to provide an improved method and apparatus for data processing.
It is another object of the present invention to provide an improved method and apparatus performing result normalization in a floating-point processor.
It is yet another object of the present invention to provide an improved method and apparatus for anticipating leading zeros/ones in a floating-point processor.
In accordance with the method and system of the present invention, a leading zeros string and a leading ones string is generated by examining carry propagates, generates, and kills of two adjacent bits of two input operands to an adder within a floating-point processor. The leading zeros string is for a positive sum, and the leading ones string is for a negative sum. A normalization shift amount is calculated directly and concurrently from the leading zeros string and the leading ones strings prior to a determination of a sign of an output of the positive sum and the negative sum.
All objects, features, and advantages of the present invention will become apparent in the following detailed written description.
REFERENCES:
patent: 5204825 (1993-04-01), Ng
patent: 5282156 (1994-01-01), Miyoshi et al.
patent: 5317527 (1994-05-01), Britton et al.
patent: 5493520 (1996-02-01), Schmookler et al.
patent: 5798952 (1998-08-01), Miller, Jr. etr al.
patent: 5831884 (1998-11-01), Suzuki
Dhong Sang Hoo
Ngo Hung Cai
Nowka Kevin John
Felsman Bradley Vaden Gunter & Dillon, LLP
International Business Machines - Corporation
Mai Tan V.
Slays Casimer K.
LandOfFree
Method and apparatus for anticipating leading digits and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for anticipating leading digits and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for anticipating leading digits and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2556881