question is about XMOS DSP library v3.0.0 and FFT function.

I was comparing the old library which uses two complex arrays to store FFT results and the new one (v3.0.0), which uses only one array, since I want to use the new one, because of the less memory usage.

As an example I used two real signals sig1(x) = 0.75*sin(x) and sig2(x) = 0.75 * cos(x).

- When old library was used, the function lib_dsp_fft_forward_tworeals transformed two real signals in frequency domain and results were :

FFT output spectrum of signal 1:

re, im

0.0000000000, 0.0000000000

0.0000000000, -0.3749999991

0.0000000000, 0.0000000000

0.0000000000, 0.0000000000

0.0000000000, 0.0000000000

0.0000000000, 0.0000000000

0.0000000000, 0.0000000000

0.0000000000, 0.3749999991

FFT output spectrum of signal 2:

re, im

-0.0000000009, 0.0000000000

0.3749999991, 0.0000000000

-0.0000000009, 0.0000000000

0.0000000000, 0.0000000000

-0.0000000009, 0.0000000000

0.0000000000, 0.0000000000

-0.0000000009, 0.0000000000

0.3749999991, 0.0000000000

- When the new version of the library was used, functions dsp_fft_bit_reverse and dsp_fft_forward were used to save FFT results in one complex array:

FFT output spectrum of signals

re, im

0.0000000000, -0.0000000009

0.0000000000, 0.0000000000

0.0000000000, -0.0000000009

0.0000000000, 0.0000000000

0.0000000000, -0.0000000009

0.0000000000, 0.0000000000

0.0000000000, -0.0000000009

0.0000000000, 0.7499999981

When inverse FFT is done, in both cases (in old library by using lib_dsp_fft_inverse_tworeals, in the new library by using functions dsp_fft_bit_reverse and dsp_fft_inverse), real time signals are correct.

What's bugging me is the following (what I would expect): Every signal after FFT holds N samples of information. FFT results of the first signal should be in [0, N/2 -1] and of the second signal in [N/2, N-1] samples and in this case it doesn't look so or am I missing something ?

So the question is how (in what order) are stored FFT results, when doing FFT on two real signals simultaneously ?

Thank you