This is the companion repository for my blog post here. You'll find plenty of details there.
This directory contains several trials of implementing efficiently the Fast Fourier Transform (FFT) on Arduino. As a comparison, I took this implementation.
If you want to understand the basis of how to implement the FFT, I wrote a small tutorial here (English version is here). Currently, here are the benchmarks on the following test signal, with a varying number of points.
Below is a list of the different implementation.
This is a direct translation to C++ of the implementation in my tutorial.
This is an implementation of the FFT by user abhilash_patel on Instructables.
It is there for comparison purpose only, as it is not very space-efficient.
A floating-point implementation of the FFT, using some rewriting of
multiplications in ExactFFT, and some bit magic to accelerate float operations.
Featuring, but not limited to :
- A table of sines to remove any call to
sin(); - A bit-magic implementation of the square root;
- A homemade implementation of the square operation for floats that I detailed here (in French);
- Some rewritings of complex multiplications of ExactFFT using this method;
A 16-bits fixed-point implementation of the FFT on AVR.
An 8-bits fixed-point implementation of the FFT on AVR.
