• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 100 subscribers
  • Views 915 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Problem with IQmath library

Vitalii Kukush
Intellectual 830 points

Other Parts Discussed in Thread: TMS320C6713B

Dear e2e Community!

Can't compile the C - project with the function _IQsqrt(in1) from IQmath library. During linking CCS writes the following message:

 

<Linking>

 undefined     first referenced   symbol                          in file                                                            

                                           _IQisqrtTable D:\\kukush.v\\ProjCCS\\OctFRectifierDecimRxy\\IQmath_c64x .lib<IQNsqrt.obj>

error: unresolved symbols remain error: errors encountered during linking; "./Debug/OctFRectifierDecimRxy.out"    not built

 

Remark: OctFRectifierDecimRxy is the name of my project.

Files "IQmath.h" and "typedef.h" are already included to the project. IQmath_c64x .lib is also linked to the progect. All as described in SPRUGG9.

Note, please, that linker can't find _IQisqrtTable for _IQsqrt, not a _IQsqrt.

Same thing happens, when I want to use IQNsin (it can't find _IQsinTable...)

I work with C6747 EVM (OMAP L137 EVM). Use CCS v3.3 Build options: Target C64x  (mv6400 ), Little endian.

 

Help me please! Where can I find *.lib with the necessary tables... ?

With best regards, Vitalii

  • 0
    Intellectual 830 points

    Deare e2e community!!!

    I have found solution!!!

    For linking IQmath library on a little-endian device you MUST link to a project TWO *.lib files: 1. IQmath_c64x+.lib; 2. IQmath_RAM_c64x+.lib.

    Attention!!! Linking only IQmath_c64x+.lib is insufficient !!!!

    Thank You very much!!! Vitalii

  • 0 in reply to Vitalii Kukush
    Intellectual 580 points

    Hi

    I am using TMS320C6713b and code composer studio v5.5. Its a floating point dsp. I am not very sure whether I specifically need floating point dsp or fixed point dsp can fill my purpose. Is there anyway i can compare the results for using the two dsps? currently In order to check that I am making a project in CCSv5.5 using IQmath lib. But every where I search I find IQmath for C2000 and C64X. I wanted do know is IQmath lib available for C6713b? If yes which version? Kindly help in this regard.

    Thankyou.

    Maheen

  • 0 in reply to Maheen Islam
    Intellectual 830 points

    Hi Maheen!

    As I understand, IQmath lib is a library for FIXED point computations. 'IQ' acronym means, that number has fixed bits for integer (I) and fractional (Q) parts of number. So IQmath.lib is aimed to fixed-point processors (for example, C64, C64+ ). As I understand C67x and C64x+ DSP's aren't code compatible. (They are code compatible with C674x, OMAPL137, L138, but this isn't your case. Please see the Introduction-Presentation in the 'Basic' chapter of C6000 Teaching ROM (http://e2e.ti.com/group/universityprogram/educators/w/wiki/2041.c6000-teaching-rom.aspx).

    So i think, that IQmath-libs are avalible only for fixed-point DSPs... But I'm not sure. I'm beginner with TI DSPs. Ask please TI Employees about such compability!

    With best regards, Vitalii

  • 0 in reply to Vitalii Kukush
    Intellectual 580 points

    Thank you Vitalii for your response.

     But I compiled the same code for both C64x(fixed point) and C67x(floating-point) without adding any special libraries. And I got exactly same results. That is why I don't understand the purpose of IQmathLib.

      Thanks.

    Maheen
     

  • 0 in reply to Maheen Islam
    Intellectual 830 points

    Maheen! It is very Interesting.

    As I understand, Result can be same/ But what can You say about a speed of calculations? For example You can use ' * ' or _IQNmpy from IQmath lib for multiplication. Results can be same but the speeds of execution will be different.

    When the compiler sees ' * ' it uses general-purpose algoritm. This general-purpose algoritm is universal for all DSPs and don't use any hardware accelerators (I mean hardware fixed or floating point multiplier etc.)

    Have You estimated the speed of computation?