Controlling number of precisions

[Sarath]

Dear All,

I've a float variable with value float f = 43.125623; // some number
I need to consider only 3 digits (43.125) from the precision because
the on doing further calculations the further precision digits should
not be affected.

Converting to string with desired precesing and converting back again
to float is a solution, but is there any better way to manage this?

 

[Dennis Jones]

Dear All,

I've a float variable with value float f = 43.125623; // some number
I need to consider only 3 digits (43.125) from the precision because
the on doing further calculations the further precision digits should
not be affected.

Converting to string with desired precesing and converting back again
to float is a solution, but is there any better way to manage this?

If you don't care about rounding, an easy solution is to simply multiply by
1000, cast to 'int', then divide by 1000. For example:

number = int(number * 1000)/1000.0;

If you want rounding, try this simple rounding method:

double round( double value, unsigned int decimalplaces )
{
double pow10factor = pow10(decimalplaces);
return floor(value * pow10factor + 0.5) / pow10factor;
}

- Dennis

 

[gw7rib]

Dear All,

I've a float variable with value float f = 43.125623; // some number
I need to consider only 3 digits (43.125) from the precision because
the on doing further calculations the further precision digits should
not be affected.

Converting to string with desired precesing and converting back again
to float is a solution, but is there any better way to manage this?

Do you really gain anything by doing this? Unlike a person, the
computer can probably manipulate the real number just as easily as it
can the rounded off one.

Incidentally, as humans use base ten but computers generally use
binary, it may not be storing the rounded off number accurately
anyway.

Hope that helps.
Paul.

 

[Knockr]

Do you really gain anything by doing this? Unlike a person, the
computer can probably manipulate the real number just as easily as it
can the rounded off one.

Incidentally, as humans use base ten but computers generally use
binary, it may not be storing the rounded off number accurately
anyway.

Hope that helps.
Paul.

Hello Paul,

Thanks for your reply. Actually I'm dealing with an OpenGL program
where my range of drawing is -1 to +1. So minor precision matters.

 

[Sarath]

Thanks Victor.

Check out "fixed point arithmetic" on the Web.

V