Floating Point Unit Design for Reduced Gate Count
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital circuit implementations for floating point operations require a large number of gates and are not optimized for speed, making them inefficient in terms of resource usage and processing speed.
Innovation Solution
A floating point multiply-add-subtract implementation that simplifies computation by dropping the least significant bits of operands, uses additional exponent bits for error handling, and employs pipelining and multiplexers to handle operations efficiently, reducing the need for complex error detection circuitry and allowing simultaneous calculation of mantissa and exponent results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard floating point unit implementation is used, then operational accuracy is maintained, but gate count increases to approximately 12,800 gates
Solution Approach 1:
The patent extracts and removes the least significant bits from the mantissa operands before computation. This truncation approach reduces the precision requirements of intermediate calculations, allowing the use of fewer gates while maintaining sufficient final result accuracy for the intended application domain.
Solution Approach 2:
The patent modifies the exponent representation by using two additional bits in the exponent field to encode error condition information (underflow, overflow, NaN). This parameter change allows the circuit to detect errors without requiring separate complex detection circuitry, thereby reducing overall gate count while maintaining operational integrity.
2Reliability
If complex error detection circuitry is implemented, then underflow, overflow, and NaN detection capability is improved, but device complexity increases
Solution Approach 1:
The patent merges the error detection function into the existing exponent calculation circuitry. By using two additional bits in the exponent field to represent error conditions, the design combines multiple functions (exponent calculation and error detection) into a single integrated structure, eliminating the need for separate error detection circuits and reducing overall complexity.
3Device complexity
If sequential processing is used, then circuit complexity is reduced, but processing speed decreases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and preparing operand data in pipelined stages before the final computation. The least significant bits are truncated and exponent adjustments are made in advance, allowing the main computation unit to operate more efficiently and enabling higher clock frequencies without increasing critical path complexity.
Data Source
AI summary
A floating point multiply and addition/subtraction implementation is provided. Two operands are received in a standard floating point format with a code selecting a mathematic operation from addition, subtraction, and multiplication. Result mantissas and exponents are calculated simultaneously for all operations. The implementation simplifies computation of a result mantissa by dropping the least significant bits of the operands before computing the result. Underflow and overflow errors are shown by two extra bits in the exponent portion of the result. The mantissa result and the exponent result are selected by providing the operation code to a mantissa multiplexer and an exponent multiplexer. The selected mantissa and exponent are combined as output.