Arithmetic Processing Unit Saturation Detection Circuit
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Arithmetic processing units face performance impairment due to complex and time-consuming saturation detection in multiply-add operations, especially when representing results within limited binary bit ranges, leading to overflow or underflow conditions.
Innovation Solution
Incorporating a bias generator and output selection circuitry to simplify saturation detection by adding a bias value to the intermediate product and addend, allowing for quick determination of overflow or underflow conditions through multiplexing operations, reducing the complexity of saturation handling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If saturation detection is performed using conventional methods in multiply-add operations, then accuracy of saturation detection is improved, but computational performance deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing saturation threshold values in lookup tables before the multiply-add operation executes. During the operation, the pre-computed thresholds are directly retrieved and compared with the intermediate result, eliminating the need for complex real-time saturation detection calculations and thereby maintaining both accuracy and high computational performance
Solution Approach 2:
The saturation detection process is segmented into separate independent steps: first computing the intermediate result of the multiply-add operation, then separately comparing it against pre-stored saturation thresholds. This segmentation allows the detection to be performed efficiently without interfering with the main computational flow, resolving the contradiction between detection accuracy and computational performance
2Productivity
If saturation detection complexity is reduced to improve computational performance, then productivity is improved, but measurement precision deteriorates
Solution Approach 1:
The patent changes the parameter representation by pre-computing saturation thresholds in advance and storing them in standardized formats. This parameter transformation allows the detection process to use simple comparison operations rather than complex calculations, maintaining high computational performance while ensuring detection accuracy through the use of precisely pre-calculated threshold values
Data Source
AI summary
Sum and carry signals are formed representing a product of a first and a second operand. A bias signal is formed having a value determined by a sign of a product of the first and the second operand. An output signal is provided based on an addition of the sum signal, the carry signal, a sign-extended addend, and the bias signal. A portion of the output signal, a saturated minimum value, or a saturated maximum value, is selected as a final result based on the sign of the product and a sign of the output signal.