FFT IP Core: User Guide

ID 683374
Date 11/06/2017
Document Table of Contents

3.1. Fixed Transform FFTs

The buffered, burst, and streaming FFTs use a radix-4 decomposition, which divides the input sequence recursively to form four-point sequences, requires only trivial multiplications in the four-point DFT. Radix-4 gives the highest throughput decomposition, while requiring non-trivial complex multiplications in the post-butterfly twiddle-factor rotations only. In cases where N is an odd power of two, the FFT MegaCore automatically implements a radix-2 pass on the last pass to complete the transform.

To maintain a high signal-to-noise ratio throughout the transform computation, the fixed transform FFTs use a block-floating-point architecture, which is a trade-off point between fixed-point and full-floating-point architectures.