Part Number: AM5728
Tool/software: TI-RTOS
There are set of FFT functions available in DSPLIB but there is no clear description of which FFT function shall be used in which scenario.
In my case I have an array of 16 bit real valued samples. I want to have 32 bit packed (16bit real , 16bit imag) complex array as output but I am not sure which of the library function does this most efficiently. I am using c66x DSP but want to implement FFT in fixed point only.
I believe if I am utilizing the info that my input is real then there shall be considerable optimization achieved from the given benchmark in test reports shared with dsplib.
Also if there are any comparisons available for following fixed point FFT functions then that will help a lot.
| DSP_fft16x16 | |
| DSP_fft16x16r | |
| DSP_fft16x16_imre | |
| DSP_fft16x32 | |
| DSP_fft32x32 | |
| DSP_fft32x32s | |
| DSP_ifft16x16 | |
| DSP_ifft16x16_imre | |
| DSP_ifft16x32 | |
| DSP_ifft32x32 |