Floating point calculation

[lucckygeek]

if i multiply a number by float....and want get the integral part of the product
i used

#include<iostream>
#include<cmath>
using namespace std;
int main()
{
	int y=floor( 8700 * 0.94 );
	cout<<y<<endl;
}

o/p=8177
but it should be 8178
please tell why is it differing ...and how to correct it

[system]

int y = ( 8700*0.94 + 1e-8 )

[lucckygeek]

but why we have to add that ...
and how will we know when to add and when not to add.
what is the reason why previously it was giving incorrect ans....

[system]

Double is stored as a recurring base 2 value in memory. Truncation to int can cause loss of precision. You should always add some extra precision value before such truncation.

[lucckygeek]

how do we know how much extra precision should we add......??
like in this case u said to add 1e-8 will this be same always or depends ..

[system]

Its platform + compiler dependent.