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.

how to select C64+ fixed point DSP from CCS5.2 on OMAPL138 device?

Other Parts Discussed in Thread: OMAPL138, OMAP-L138

After installed the CCS5.2, when I tried to create a project, I did not find the option that allow me select C64+ fixed-point DSP instead of C674x floating point DSP.

The only seting that looks correct to me is: Device Family: C6000; Variant: OMAP-L1x : OMAPL138.

Can some one answer my questions?

  • Adam,

    The OMAPL138 device contains a C674x DSP and ARM9 processor on it, it does not contain a C64x+ DSP. so what you are seeing in CCS 5.2 is indeed accurate. C674x DSP is a next generation of the C64x+ DSP so all the C64x+ code is compatible on C674x. Connecting to OMAPL using CCS has been described here:

    http://processors.wiki.ti.com/index.php/How_to_connect_to_the_OMAP-L138/C6748/AM1808_EVM_board_using_CCS%3F

    Hope this answers your questions.

    Regards,

    Rahul

  • Hi Rahul,

     

    You are right. The OMAP-L138 only has one DSP - C674x.

     

    From "OMAP-L138 C6-Integra DSP+ARM Processor (Rev. D)", I also found the following information:

    C674x is a Fixed/Floating Point CPU.

    The C674x CPU combines the performance of the C64x+ core with the floating-point capabilities of the

    C67x+ core.

     

    My question is: how do I configure C674x CPU into a Fixed Point CPU mode?

     

    Here is the scenario that I have:

    Assume I have a variable of 0.5, if I use floating point DPS, I can directly assign var = 0.5;

    but if I use fixed point DPS, I must assign var = 0.5*2^15 (if it is "short");

     

    So I want to explicitly configure C674x CPU into a fixed point DSP for my whole project, how do I configure it from CCS?

     

    Regards,

    Adam

  • Adam,

    You do not need to configure the DSP as fixed point processor to run fixed point algorithms. A fixed point DSP also has the ability recognize a float variable and doesn`t really need the conversion, what you are refering to is defining a variable in Q point format(fixed point arithmetic) which can be done in software and doesn`t require hardware configuration. For writing algorithms using Q point arithmetic, we provide the IQMATH library. You can also use all the C64x+ fixed point DSP libraries compute fixed point arithmetic optimally on the C674x core. Please look at the following wiki to view all fixed point DSP libraries:

    http://processors.wiki.ti.com/index.php/Software_libraries

    Regards,

    Rahul