In order to compute the FFT with Accelerate framework, I see a lot of people using (the old way I guess) fft_zrip instead of the recommended one by Apple in the vDSP documentation DSP_DFT_Execute.
Is there any reason why DSP_DFT_Execute is so under used?
The DFT API is newer, so it is not as well known. The FFT API was around for many years, so knowledge of it has spread, along with examples around the net.
For lengths that are powers of two, they will use the same implementation underneath the API.