FFT computation

Mohamed Sohal

Expert 1170 points

Hi!

I have some doubts regarding the FFT computation algorithms provided in the DSP library of 6678.

Are they all one dimensional?

I have gone through the code of DSP_fft16*16. It seems to be one dimensional. If so why is it named as 16*16

Regards

Sohal

over 13 years ago

0 DanRinkes over 13 years ago

TI__Expert 8055 points

Mohamad,

They are all one dimensional. The 16s refer to the number of bits in the coefficients.

You can use these to compute pieces of a multidimensional FFT, but we don't have any multidimensional DSPLib algorithms. The CPU computes the multidimensional FFT in only a single direction at a time, so there's no advantage to having specific optimized algorithms for N-dimensions. You should be able to achieve the same thing by using these to generate your own multidimensional FFT.

Regards,

Dan

0 Mohamed Sohal over 13 years ago in reply to DanRinkes

Expert 1170 points

Thank you Dan!

but still dsp_fft16 will do.

why is it named as 16*16?

regards,

Sohal

0 DanRinkes over 13 years ago in reply to Mohamed Sohal

TI__Expert 8055 points

I suspect that's to remind the user that the input and output are both 16-bits.

Processors

Processors forum

FFT computation