Look-Up Table Transform for Hardware Area Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The implementation of fixed-function hardware logic for processors is hindered by the large area requirements of look-up tables (LUTs) needed to represent complex functions, especially when higher precision is demanded, leading to infeasible hardware sizes in space-constrained devices.
Innovation Solution
The method involves generating a single look-up table representing a function, applying a transform such as the Haar wavelet to divide it into multiple derived LUTs, and implementing these in fixed-function hardware logic along with an inverse transform unit, allowing for reduced hardware area while maintaining low latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single large look-up table is used to represent complex functions with high precision, then measurement precision is improved, but area of stationary object increases significantly
Solution Approach 1:
The patent divides a single large look-up table into multiple smaller derived look-up tables by applying a transform (e.g., Haar wavelet transform). Each derived LUT stores a portion of the function data, and their combined outputs reconstruct the original function with minimal error. This segmentation reduces the area required for each individual LUT while maintaining overall function precision.
2Area of stationary object
If multiple derived LUTs are implemented to reduce hardware area, then area of stationary object is reduced, but device complexity increases
Solution Approach 1:
The patent introduces a transform unit as an intermediary component that processes inputs to select and combine outputs from multiple derived LUTs. This mediator enables the system to use smaller LUTs while reconstructing the original function, balancing area reduction against the added complexity of the transform logic.
3Productivity
If numerous instances of a function are implemented to satisfy performance metrics, then productivity is improved, but area of stationary object becomes very large
Solution Approach 1:
The patent creates a universal function implementation using multiple derived LUTs that can serve multiple function instances. By sharing the transformed LUT structure across numerous function instances, the system achieves high throughput while avoiding the area multiplication that would occur with identical separate implementations for each instance.
Data Source
AI summary
Methods for implementing or synthesizing functions in hardware and fixed-function hardware include generating a look-up table, LUT, representing the function and then applying a transform to the LUT to transform the LUT into a plurality of derived LUTs. The transform may be applied recursively. A hardware design implementing each of the derived LUTs in fixed-function hardware logic, along with a logic unit that performs the inverse transform, is then created.