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.

SPRC265: DSPLib C674х Matrix functions

Part Number: SPRC265

Hi team,

My customer realized that DSPLib for the C674х core have significantly reduced matrix functionality in comparison with the C66x DSPLib.

Do we have a DSPLib version for the C674х core with extended matrix functionality?

  • Can you please indicate what functionality is your customer looking for on C674x that is missing compared to C66x. C66x is the next generation DSP architecture to C674x that added complex data type functionality and instructions and added enhanced support for double precision floating data type.  IF you compare the C674x and C66x, only functions with complex data type and double precision functions should be missing.

    Other than that if there is any other functions missing, please let us know. If the customer wants to leverage complex matrix or matrix with double precision on C674x, they have the option to use the natural C implementation of the function by cross compiling it on the C674x DSP. 

    (Every function in DSPLIB has a C implementation and a optimized callable implementation that is optimzed for the DSP using C intrinsics or assembly that is only supported for the DSP architecture. Natural C code can be compiled on any core)

    Regards,

    Rahul 

  • Hi Rahul,

    The customer wondering about dspf_sp_lud_inv_cmplx function.

    Do we have it for C674x lib?

  • No, the optimized DSP implementation of that function only exists for C66x DSP and not on C674x DSP. As I indicated, the complex instruction were introduced with the C66x architecture and don`t apply to C674x. The customer will need to use the C implementation of the function provided int he file:

    they can compile that code with -o3 and use #PRAGMA supported in compiler to optmize the loop on C674x.