Only explanation I could find on inverse FFT with accelarators was on page 6-15 of ADSP-214xx SHARC Processor Hardware Reference, Rev. 0.3 which wasn't very clarifying for me. Here is stated: "... IFFTs by setting up a coefficient TCB with change of sign for the sine twiddles (FFT uses twiddles cosine, sine, -sine, cosine, the IFFT uses cosine, -sine, sine, cosine)..." What I want to learn is, do we supply the acc. with these coefs. or do we inform the acc. about the nature of fft (inverse or not). If that is the case, how do we let acc. know that it is inverse?