Support for 3D/nD FFTs

[jornada]

Is there any way to perform 3D or nD FFTs with mkl_fft, even if in a hacky way? I need 3D FFTs for my application, and mkl_fft just so much faster than pyfftw or the mkl-optimized anaconda distribution.

Great job, by the way!

[fakufaku]

Hi @jornada ,

Thanks for the feedback.

The MKL allows for nD FFTs but we have not implemented any wrapper at this point. It would be very nice to add this functionality to the library, but we don't have a lot of time on our hands at this point.

If you have some time, we'd be super grateful if you want to help implement this. The library is a light wrapper around the MKL FFT library (we are calling C functions from python with a few checks around). You can check the code for fft2 and add the code for nD. The MKL doc is here. We are trying to replicate the api from numpy fft for compatibility, so you could check out the options from numpy.fftn.

Cheers