C64x+BEDSPLIB - Big Endian DSP Library - FFT and IFFT scaling issues

Alphan Karacaer56

Hello,

I am using Big Endian DSP Library.

FFT16 and IFFT16 functions works only very limited ranges, like -1000 +1000, and never managed to get correct results from FFT32 and IFFT32 functions.

Anybody knows what are the optimum ranges for FFT16 and IFFT16. And can anybody use FFT32 anf IFFT32 functions correctly, if yes can you send me example source codes for it.

Thanks a lot.

Alphan

over 14 years ago

0 Alphan Karacaer56 over 14 years ago

Intellectual 835 points

Are there anyone who uses FFT from dsplib, whether little endian or big endian.

0 Alphan Karacaer56 over 14 years ago in reply to Alphan Karacaer56

Intellectual 835 points

Since no one has answered my question, I guess nobody uses Big Endian libraries. Even though I want to report a mistake in the Big Endian DSP Library. There is a function gen_twiddle_fft32x32_be for generating twiddle factors to be used by FFT32 and IFFT32, but when you use that function, results are never true for FFT32 and IFFT32 operations, instead of that, if you use gen_twiddle_fft32x32 function from little endian library you can get correct results for Big Endian FFT32 and IFFT32 operations.

Processors

Processors forum

C64x+BEDSPLIB - Big Endian DSP Library - FFT and IFFT scaling issues