This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
I've been trying a VLFFT project from China TI training forum to test the FFT performance of 6678. But following the user guide, I just got same results no matter how many cores I used to calculate.
in the Vlfft ok.rar is the project I'm using.(make sure you use edma3_2_11_2_4 to build the project, higher version will cause cmd failures.)
I
To configure the number of DSP cores to compute, open the file vlfftconfig.h under ...\demo\vlfft\vlfftInc and
change the constant definition NUM_CORES_FOR_FFT_COMPUTE to one of the following numbers
• 8-core device: 1, 2, 4, 8
But no matter which one I used, I just got 6.6ms result, and it was a 8 cores result.
Can any one tell me why? and Really appreciate if you can give me a hint.
Thanks
Hi,
It should work. Could you please try the following?
1) Compile the code in Debug mode
2) On core0, set the breakpoint right after the function call vlfftParamsInit in vlfftApps.c.
3) Check the following in the structure VLFFTparams.
VLFFTparams->log2NumCoresForFftCompute
VLFFTparams->numCoresForFftCompute
VLFFTparams->N
VLFFTparams->N1
VLFFTparams->N2
Xiaohui
Hi,
Xiaohui Li, I am working on your very large FFT code , I would like to know how you have measured the DSP cycles for FFT performance table of your paper "Very large FFT for TMS320C6678 processors" .
Regards