Floating Point Multiply Adder with Extended Exponent Range
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems face inefficiencies in performing multiply add operations with floating point numbers, particularly due to the need to shift product values to fit within a normal exponent range, which increases processing time and complexity.
Innovation Solution
The system employs a multiplier that generates a product with an extended exponent range, allowing the adder to handle subnormal values without additional shifts, and includes count-leading-zero circuitry and shifters to align mantissas, enabling faster multiply add operations by avoiding the need for intermediate exponent normalization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the multiplier generates product exponent within the normal exponent value range, then the result fits standard floating point format, but additional shifting operations are required which increase processing time
Solution Approach 1:
The multiplier performs preliminary action by generating the product exponent with extended range capability before the addition operation. This allows the exponent to accommodate subnormal values without requiring subsequent shifting operations, thereby reducing processing time while maintaining format compliance through final normalization in the adder.
Solution Approach 2:
The patent changes the parameter of exponent value range from the standard fixed range to an extended range that includes lower values for subnormal numbers. This parameter change allows the product exponent to represent a broader set of values without immediate normalization, eliminating the need for time-consuming shift operations.
2Productivity
If the system uses extended exponent value range for product, then subnormal values can be handled without shifting, but the adder must handle extended exponent values which increases circuit complexity
Solution Approach 1:
The adder is designed with multi-functionality to handle both normal and extended exponent value ranges using the same circuitry. By making the adder universal in its exponent handling capability, the patent avoids the need for separate circuits for different exponent ranges, thus increasing productivity without proportionally increasing device complexity.
Solution Approach 2:
The adder employs dynamic adjustment mechanisms that adapt its operation based on the input exponent values. When extended exponent values are detected, the adder dynamically adjusts its alignment and addition logic, allowing it to handle subnormal values efficiently without requiring static additional circuitry for all cases.
3Device complexity
If standard exponent range is used, then circuit design is simpler, but subnormal operands require additional shift operations that reduce processing speed
Solution Approach 1:
The multiplier performs preliminary action by generating product exponents with extended range capability before the addition operation. This preliminary extension of the exponent range eliminates the need for subsequent shift operations on subnormal values, thereby increasing processing speed without significantly complicating the overall circuit design.
Solution Approach 2:
The patent changes the exponent value range parameter to include extended lower values for subnormal numbers. This parameter change allows the circuit to handle a broader range of inputs directly without additional shift operations, improving speed while maintaining relatively simple circuit architecture through unified handling logic.
Data Source
AI summary
A floating point multiply add circuit 24 includes a multiplier 26 and an adder 28. The input operands A, B and C together with the result value all have a normal exponent value range, such as a range consistent with the IEEE Standard 754. The product value which is passed from the multiplier 26 to the adder 28 as an extended exponent value range that extents lower than the normal exponent value range. Shifters 48, 50 within the adder can take account of the extended exponent value range of the product as necessary in order to bring the result value back into the normal exponent value range.