Optimized FFT for 32k complex numbers

Hi all,

  1. I am looking for a highly optimized FFT code for size 32k.

I already know Apples OpenCL FFT implementation, but I thought there must be something better, since I don’t need a general purpose algorithm, but only size 32k.

  1. I am also looking for an FFT algorithm that can take advantage of the fact that the input data is either -1 or +1 (well, to be more precise, since it’s complex numbers, it’s {{-1,-1}, {-1,+1}, {+1,-1}, {+1,+1}} ).

Any ideas?