CCS/TMS320F28027F: Rounding floating pointing number

Thanks for your suggestion .
But right now i am not using IQmath library , instead of that i am using math library . I know it is not the efficient way to do floating point arithmetic on Fixed point device without IQmath . In my case it is not time-critical system and also i don't think any advantages of using IQmath library for my project.
Thank you.

Indrajit,

Not possible, sorry. Any digital machine has finite numerical precision, so you can't necessarily represent a number with trailing zeros exactly. For example, if you did this...

#include <math.h>

volatile float x = 3.584743647;
volatile float y, z;

main()
{
y = round(x * 10000);
z = y / 10000;

...you would be rounding y to the nearest integer then dividing back to restore the decimal point. With infinite precision it would work, but on a finite precision machine you may not see a number with only four decimal places. With 32-bit floating point you can represent 3.58469987 or 3.58470011, but not 3.58470000. With more bits you'll get better precision, but the issue will still be there. This is true whether the floating point number is emulated in the RTS or not.

If truncating the decimal places is important you'll need to use an integral data type such as IQ.

Regards,

Richard

Thanks for clearing my doubts.
I will use IQmath.
Thanks.