Constant Geometry Split Radix FFT Architecture

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional FFT architectures, including those using split radix algorithms, face limitations in throughput due to irregular scheduling and high latency issues caused by 'L-shaped' butterflies, making them unsuitable for high-speed applications.

Innovation Solution

A constant geometry architecture employing symmetrical butterflies with multiplexers and latches to selectively perform addition and subtraction operations, eliminating the need for complex multipliers and reducing latency, while using a reorder matrix and controller to manage data flow efficiently.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If conventional split radix architecture uses L-shaped butterflies, then the algorithm reduces non-trivial complex multiplications, but irregular scheduling and uneven latency between datapaths reduce throughput

Engineering Contradiction:
Improvenumber of complex multiplicationsVSAvoidthroughput
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent transforms the traditional L-shaped butterfly into a symmetrical butterfly structure where all four datapaths have equal length and symmetry. This symmetrization eliminates the irregular scheduling and uneven latency problems while preserving the reduced multiplication count advantage of split radix algorithms.

Inventive Principle:
Principle #4Asymmetry

Solution Approach 2:

The patent introduces control signals that dynamically switch the operation mode of the symmetrical butterfly between different computational modes (e.g., radix-2 and radix-4 modes). This dynamic switching allows the architecture to adapt to different computational requirements while maintaining regular scheduling and uniform latency.

Inventive Principle:
Principle #15Dynamics

2Adaptability or versatility

If variable geometry architecture uses multiplexers to achieve Cooley-Tukey algorithm, then flexibility in computation is improved, but switching overhead reduces throughput

Engineering Contradiction:
Improvecomputation flexibilityVSAvoidthroughput
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments the butterfly computation into separate real and imaginary datapaths with dedicated summing circuits for each. This segmentation eliminates the need for multiplexers to switch between different computational modes, as each datapath is dedicated to specific operations, thereby removing switching overhead while maintaining computational flexibility through control signals.

Inventive Principle:
Principle #1Segmentation

3Productivity

If constant geometry architecture uses fixed wires to avoid multiplexers, then switching overhead is reduced, but the architecture cannot efficiently implement split radix algorithm with irregular scheduling

Engineering Contradiction:
ImprovethroughputVSAvoidalgorithm implementation efficiency
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent designs a universal symmetrical butterfly structure that can efficiently implement both radix-2 and radix-4 operations within the split radix algorithm. The constant geometry architecture achieves this universality through control signals that direct the fixed datapaths to perform different operations, eliminating the need for variable geometry while maintaining algorithm implementation efficiency.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS8819097B2Constant geometry split radix FFT
Publication Date: 2014.08.26 TEXAS INSTRUMENTS INC
  • US8819097B2 patent drawing
  • US8819097B2 patent drawing
  • US8819097B2 patent drawing

AI summary

An apparatus for performing a Fast Fourier Transform (FFT) is provided. The apparatus comprises a reorder matrix, symmetrical butterflies, and a memory. The reorder matrix is configured to have a constant geometry, and the butterflies are coupled in parallel to the reorder matrix. The memory is also coupled to the reorder matrix and each butterfly. The reorder matrix, the butterflies, and the memory can then execute a split radix algorithm.