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.

Fastmath with optimization on 6740

Hi,

I've a problem with the fastmath67x.lib on C6747.
If I use the fastmath library in combination with optimization (File level), the gen_twiddle_fft16x32.c from TI doesnt work anymore. It hangs up when I call it.
Does anyone know, what happens?

 

cheers,
Joe

  • If you halt the processor what function is executing?  It would be good to know a little more about what's going on.

    Are you using DSP/BIOS?  If so, please open the DSP/BIOS -> Message Log in CCS and check to see if there are any messages such as an exception having occurred.  You might also open DSP/BIOS -> Kernel Object Viewer to see if you have any issues there.

    What codegen version are you using?  There was a fairly serious optimization issue in 6.1.7 and earlier.  So you should make sure you are on 6.1.8 or later.

    Brad

  • Hi Brad,

    I use Codegen 6.1.9.

    The message log has the following entries:

    13851   EXC_exceptionHandler: EFR=0x2
    13852     NRP=0xc0046ca0
    13853     mode=supervisor
    13854   Internal exception: IERR=0x10
    13855     Resource conflict exception
    13856   SYS abort called with message 'Run-time exception detected, aborting ...'

    And the processor is executing:

    C004F6C8          C$L1:
    C004F6C8 0002A120            BNOP.S1       C$L1 (PC+8 = 0xc004f6c8),5

     

    Joe

  • Hmmm, I don't see what the resource conflict is there.  Can you copy/paste more instructions from the disassembly?  Which function is executing?  Also, try doing a reset before loading your optimized code.

  • Joe, Was your problem resolved? If not, it will be nice if you can provide a recipe to recreate the problem or an example to demonstrate the problem and we can take a look at it.

    Gagan

  • Hi Gagan,

    Sorry for the delay. The problem was just before my holydays and afterward I forgot to reply. Btw I changed entire calculation since then, so I don't have the problem anymore and cannot reproduce it for you, because I don't use the DSP LIB at the moment.
    Joe