Adaptive Bit Shifting for FFT Radix Kernel Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Fast Fourier Transform (FFT) implementations using fixed shifting do not perform computations at optimum resolution due to varying bit requirements based on the magnitude of signed binary integers, leading to inefficiencies in radix kernel operations.
Innovation Solution
Adaptive bit shifting of signed binary integers is employed before and after radix kernel operations, using an adaptive left bit shifter and adaptive right bit shifter based on the leading bit count, accessed through lookup tables, to optimize the resolution of computations and preserve or insert bits as necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed shifting is used in radix kernel operations, then device complexity is reduced, but manufacturing precision of computation resolution deteriorates
Solution Approach 1:
The patent applies dynamic bit shifting by determining the number of leading bits in the input data and adjusting the shift amount accordingly. Instead of using a fixed shift value, the system dynamically calculates the optimal shift amount based on the actual data characteristics, thereby achieving optimal computation resolution without excessive complexity increase.
Solution Approach 2:
The patent changes the parameter of bit shift amount from a fixed constant to a variable determined by the leading bit count of the input data. This parameter change allows the system to adapt the computation resolution to the actual data magnitude, improving precision while maintaining reasonable complexity through efficient leading bit detection.
2Manufacturing precision
If adaptive bit shifting is implemented, then manufacturing precision of computation resolution is improved, but device complexity increases
Solution Approach 1:
The patent segments the bit shifting operation into distinct phases: leading bit detection, shift amount calculation, and actual shifting execution. This segmentation allows each component to be optimized independently, reducing overall complexity while maintaining high computation resolution through adaptive shifting.
Solution Approach 2:
The system performs self-service by automatically determining the appropriate bit shift amount based on the leading bit count of the input data without requiring external control or complex configuration. The leading bit detection mechanism enables the system to self-adjust the computation resolution to match the actual data characteristics.
3Ease of operation
If fixed shifting is used, then ease of operation is improved, but loss of information increases
Solution Approach 1:
The patent implements dynamic bit shifting that adapts to the input data characteristics. By determining the leading bit count and adjusting the shift amount accordingly, the system preserves maximum information from the input data while maintaining operational simplicity through automated adaptation rather than manual configuration.
Solution Approach 2:
The system changes the shift parameter from a fixed value to a data-dependent variable. This parameter change enables the system to preserve more information by avoiding excessive shifting that would occur with fixed methods, while the automated determination process keeps the operation simple and easeful.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Performing a Fast Fourier Transformation (FFT) with increased resolution by applying an adaptive left shift to signed binary integers of an input of a radix kernel and adaptive right shift to signed binary integers of an output of a butterfly of the radix kernel which is based on a leading bit count of the input. The adaptive left shift increases a resolution of the radix kernel computation and the adaptive right shift determines a number of bits of the increased resolution preserved in an output of the radix kernel.