Can unified FFT IP accurately calculate double precision FFT？

Hi,

Recently, I used the single precision option in Unified FFT IP.

It works well. (256point ，sampling frequency is 12800hz. The real input is 100sin2pi50t, the imag input is 0, and the output of FFT IP is -12800 at the frequency of 50hz , which equals to 100*(256/2).)

But when I change single precision to double precision in FFT IP and change the input data into double precision floating-point numbers of the same size, it is not -12800 at the frequency of 50hz.

Do I need to add other designs to realize Double precision fft or Double precision FFT
cannot be realized with Unifed FFT IP?

If you have any idea, can you please tell me? I've been bothering about it for a long time.

Thanks!

Regards!