DSP library FFT Usage

Hi,

after using the library DSPF_sp_fftSPxSP. The i/p and o/p values obtaned in CCS are as follows

the values obtained in matlab are as follows,

Gopal,

The DSP library expects Real and Imaginary input format. Is that you are providing as input?

Regards

Asheesh

The 2 below threads might helpo to get a better understanding on how to use DSPF_sp_fftSPxSP:
http://e2e.ti.com/support/dsp/tms320c6000_high_performance_dsps/f/115/t/265955.aspx

http://e2e.ti.com/support/dsp/tms320c6000_high_performance_dsps/f/115/t/265955.aspx

A.

Hi Asheesh,

thanks for your reply. Yes i'm providing ,i'm copying real and imaginary values as follows, my input values are in code_0 which are complex values.

for(i=1; i<=12; i++)
{
/* real part */
x_ref[ 2*i] = creal(code_0[i]);
/* img part */
x_ref[ 2*i + 1] =cimag(code_0[i]);
}

After computing FFT i'm getting first value as correct value but remaining are not matching with the matlab values.

The DSPlib FFT library supports Radix-4 and Mixed Radix-2 implementation. 12 is not a power of 2.

There is another similar post which you can refer for the example

http://e2e.ti.com/support/embedded/tirtos/f/355/p/305811/1082832.aspx#1082832

Regards,

Asheesh

Hi Asheesh,

According to your suggestion, i appended 4 zeros to input values and computed 16 point FFT but still the same problem.

If possible can you provide me some example of FFT by using DSP library other than inputs as sine wave, the inputs should be some discrete complex values.

Thanks & Regards,

Gopal.

Michael,

The forum post I have mentioned earlier has the example for the floating point FFT library function on C674x with the bit reversing concept and how to match the Matlab and DSP library outputs.

The DSP library floating point function has a bit reverse algorithm in the function. You can bit reverse the Matlab output and then match with the DSP lib output. The C66x code has the test example for the library functions which you can follow for your development.

Regards,

Asheesh

Hi michael,

Yes i'm passing x_ref directly as the input array to FFT function as follows,

 DSPF_sp_fftSPxSP(N,x_ref, w_sp, y_sp, brev, 2, 0, N). I have declared x_ref as floating point variable as follows.

float   x_ref [2*N];

Hi everyone,

Thanks a lot guys FFT is working fine.

Regards,

Gopal