Lookup Table Input Offsetting for Precision Memory Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing circuit implementations that use floating-point format input values to produce floating-point format outputs require significant memory and processing resources, especially when dealing with wide dynamic ranges, leading to high error rates due to resolution variations.
Innovation Solution
A computation unit that adds an offset to the input in a higher precision format to generate an offset input in a lower precision format, allowing for a lookup table or interpolation circuit to produce an output with reduced memory and processing requirements while maintaining high resolution in critical regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a lookup table is used to implement function f(I) with full precision for all possible input values, then measurement precision is improved, but device complexity and memory requirements worsen
Solution Approach 1:
The input domain is segmented into multiple regions, with different precision levels applied to different segments. Critical regions receive higher precision representation while non-critical regions use lower precision, reducing overall memory requirements while maintaining output precision where needed
Solution Approach 2:
Different regions of the input domain are assigned different qualities of representation. Critical regions (where the function has high rate of change) use higher precision encoding, while non-critical regions use lower precision, optimizing the balance between memory usage and output accuracy
2Adaptability or versatility
If floating-point format with wide dynamic range is used for input variables, then adaptability is improved, but measurement precision worsens due to resolution variations
Solution Approach 1:
The precision parameter is dynamically adjusted based on the input value's magnitude and the criticality of the region. By changing the resolution parameter locally across different input domains, the system maintains high precision where needed while accepting lower precision elsewhere, thus preserving adaptability across wide dynamic ranges
3Manufacturing precision
If combinational logic is implemented based on large lookup tables, then manufacturing precision is improved, but use of energy and area consumption worsen
Solution Approach 1:
The patent extracts only the essential portions of the lookup table that are needed for accurate computation in critical regions. By taking out and implementing only the necessary high-precision entries in combinational logic, the system achieves manufacturing precision where needed while reducing the overall circuit size and power consumption
Data Source
AI summary
A computation unit computes a function f(I). The function f(I) has a target output range over a first domain of an input I encoded using a first format. A first circuit receives the encoded input I in the first format including X bits, to add an offset C to the encoded input I to generate an offset input SI=I+C, in a second format including fewer than X bits. The offset C is equal to a difference between the first domain in f(I) and a higher precision domain of the second format for the offset input SI. A second circuit is operatively coupled to receive the offset input SI in the second format, to output a value equal to a function f(SI) to provide an encoded output value f(I).