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.

RTOS/TMDSEVM6678: The diffent performace between function in DSP lib and function in optimized source code in folder of examples of DSPlib?

Part Number: TMDSEVM6678

Tool/software: TI-RTOS

Hi every one, 

I am trying to learn optimizing in C6000 by trying some examples in DSP lib in my 6678 board, currently is Cholesky examples

I wonder about : 

 If I use function of DSP lib, the performance of this situation is higer 2 times when compares with your optimized function in source code in foder of examples?

(I saw in C6000 optimation workshop and I saw in your optimized function in source code in foder of examples and i don't know how can I continue to optimize it ?

Can you give me some suggestions? 

Thank you for your help!

  • I've forwarded your query to the software experts. Their feedback should be posted here.

    BR
    Tsvetolin Shulev
  • Thanks for your help, waif for your answer !
  • Please note that the number reported in DSPLIB use a C66x simulator which assumes flat memory on a real system to get optimized numbers, you should run code from L2 memory and with Data also in L2 or MSMC.

    If you run code and have data in DDR memory then you need to turn on DSP cache or use EDMA to move data to L2 to get optimal numbers. The additional performance optimization is upto the system integrator to perform in the system.

    The DSPLIB has a linker command file(.cmd), check for placement of code and data sections in that file and make sure, you are runnign everything of DSP internal memory or MSMC . Also, when you connect to the DSP, make sure you are using GEL file to setup system clocks and PLL and DDR else the DSP willl be running in PLL bypass mode.

    Regards,
    Rahul