Work in progress!
The fftLite class is a very simple ("lite") FFT implementation.
This is mainly an effort to learn how the FFT can be computed (not just a naive DFT). Another goals is to produce a very small implementation that can be used where performance is not critical and without worrying about code bloat or large third-party libraries. It's not terribly fast, especially for non-power-of-two lengths! There is much that still needs to be implemented, such as different radix sizes, but it's a start and the output matches MATLAB's fft (close to machine precision).
TODO: Compare with MKL, etc. Experiment with SSE intrinsics as suggested by Intel's examples.