Compile Warning with IQmath Libarary

Hello Jigar!

Look at http://e2e.ti.com/support/microcontrollers/c2000/f/171/p/267132/933769.aspx#933769. Perhaps it will be useful in your specific case.

Regards,

Igor

I am using IQMath V160. "IQmath_Quickstart" pdf page 25 mention that I can use 5.0.2 or later, I am using 6.1.  Even if I suppress this warning, it will not affect operation of IQSin function.

I think IQsin function is not working because of this warning, but I am not sure about this. Can you please suggest possible cause for IQsin function not working.

Jigar Patel1 said:

Can you please suggest possible cause for IQsin function not working.

Not working how?

Have you tried the example in C:\ti\controlSUITE\libs\math\IQmath\v160\examples_ccsv4\C\2833x

-Lori

Hello Lori!

Also could you please give some comments about this post  http://e2e.ti.com/support/microcontrollers/c2000/f/171/p/267132/933769.aspx#933769. And what about type "long" in IQmath (In standard "C" it is an integer type).

Regards,

Igor 

Igor Gorbachev said:

Hello Lori!

Also could you please give some comments about this post  http://e2e.ti.com/support/microcontrollers/c2000/f/171/p/267132/933769.aspx#933769. And what about type "long" in IQmath (In standard "C" it is an integer type).

Regards,

Igor 

Igor,

I commented.  Not sure if it is what you wanted me to comment on.

Using IQmath doesn't change any of the normal C types.

Regards

Lori

Example is working fine. But example does not use IQsin function.

I am using following code. Value of out1 is constant (zero) regardless of input to IQsin function, I checked value of out1 variable in simulator mode (with Q24 format). Value of in1 in watch window is 0.7853900194, which is expected. (Refer screen in first post)

_iq in1,out1;
#define PI 3.14156

int main(void)

{

in1 = _IQ(0.25*PI);
out1 = _IQsin(in1);

}

Jigar Patel1 said:

Example is working fine. But example does not use IQsin function.

It uses IQsin() on line 126:         Dlog.Xwaveform[i] = tempX = _IQmpy(_IQsin(Step.X), Step.GainX);

Jigar Patel1 said:

I am using following code. Value of out1 is constant (zero) regardless of input to IQsin function, I checked value of out1 variable in simulator mode (with Q24 format). Value of in1 in watch window is 0.7853900194, which is expected. (Refer screen in first post)

It is possible that the compiler is tossing out the code that sets out1 since you never do anything with the variable.  You can either assign it to something else or make it volatile to see if this is the case.

Lori

Hi Lori!

Thanks.

Well...I do not want to seem importunate and I'm not afraid to seem stupid. For example (MS VS C++, console application Win32):

using namespace std;

long a=1, b=2, c=0;

float d;

int main( )

{

c = a/b;

cout<<c;// at the screen there is "0"

d = float(c);

cout<<d; //at the screen there is "0"

d = float(a)/b;

cout << d; //now at the screen there is "0.5"

return 0;

} 

It's all right with that.

Why for IQmath C28xx?: 

long helpvar=0;

 long A=1;

 long B=2;

 float solution=0;

helpvar=_IQ11div(A,B); // result is "0"!!! or NOT and I didn't understand someting

solution=_IQ11toF(helpvar); //Convert iq value to float - why result becames "0.5"????? How it works?

Regards,

Igor

Igor Gorbachev said:

Why for IQmath C28xx?: 

long helpvar=0;

 long A=1;

 long B=2;

 float solution=0;

helpvar=_IQ11div(A,B); // result is "0"!!! or NOT and I didn't understand someting

solution=_IQ11toF(helpvar); //Convert iq value to float - why result becames "0.5"????? How it works?

Igor,

Using your example I see helpvar is 0x400 which is the IQ11 representation of .5.

I think setting A and B to hex 0x1 and 0x2 is a little confusing.  But if I set them to the IQ representation of 1 and 2 I still see .5 since it is just a ratio.

Hi Lori!

Thank you very much for reply.

But from your explanation I understood the IQmath has some features of the data representation, after all. IQmath-C28xx  works with data not entirely so as other compilers are doing it. At least C-compilers with which I have dealt earlier (Borland C, GCC, Watcom, MS VS).

Best Regards,

Igor

Igor Gorbachev said:

IQmath has some features of the data representation,

An IQ number is simply a way to represent a floating point number using integer math. 

You previously asked if you use IQmath does the compiler change what a "long" means.  It does not.  When I pass two longs to IQ11div it (meaning the IQ11div function) assumes those two longs are an IQ11 representation of a floating point number. It (IQ11div) is simply a library that knows how to handle the data represented in that format. 

I suggest going through this training, although it is a bit dated.  It describes the representation of an IQ number. 

http://training.ti.com/courses/CourseDescription.asp?iCSID=46234

Regards

Lori

Hello Lori!

Many thanks once again. Everything has become more or less clear. However, everyone can have own interpretation, which ultimately does not affect the result of the calculation, but this interpretation is more understandable from a individual point of view. Thus your explanations "When I pass two longs to IQ11div it (meaning the IQ11div function) assumes those two longs are an IQ11 representation of a floating point number" I can understand only at the following interpretation: "IQmath has some features of the data representation". But this is sophistry which does not have any practical value. Certainly the practical usefulness consists in the knowledge of the rules of the application of IQmath-library. Actually this applies to any tool. Maybe I forgot that during the familiarization with something new (in our case MCU C2000) there is a need to break some old stereotypes sometimes.

Best Regards,

Igor