fft

Implementation

OpenCL

Usage

output = FFT_large_ATI(input); for 2k to 16m points fft

output = FFT_small_ATI(input); for 2 to 1024 points fft

Class Support

input and output are complex number

Algorithm

We implement different small size FFT, such as 8, 16, 32 points FFT. For large size FFT, we invoke the small size FFT multiple times. We also implement 64 points by using share memory so that we can invoke 64 points to do some steps for the large size FFT.