Are 3.5 and 3.5f the same in C++?

Question

Possible Duplicate:
strange output in comparision of float with float literal
Why comparing double and float leads to unexpected result?

In following code I was expecting answer to be "Not equal", because by default the 3.5 should be double in C++, but the result was "equal".

What is difference in declaring float a=3.5f and float a=3.5?

#include<iostream>
using namespace std;

int main()
{
    float a=3.5;
    if(a==3.5)  
        cout<<"equal"<<endl;
    else
        cout<<"Not equal"<<endl;
    return 0;
}

Try creating a double and float which are both equal to 1.0/3.0 and then seee if they are equal. You'll find that they're not. — Benj, Apr 28 '12 at 07:36
Yes Benj I got your point. What you mean to say this those are not equal because of precision. — Abhishek Gupta, Apr 28 '12 at 07:42
when comparing floating point values for equality I always use : fabs(v1-v2) — marcinj, Apr 28 '12 at 07:59

score 14 · Accepted Answer · answered Apr 28 '12 at 07:29

14

No.

The type of3.5 is double whereas the type of 3.5f is float. So they are not guaranteed to be equal in values.

#include<iostream>

void f(double) { std::cout << "it is double" << std::endl; }
void f(float)  { std::cout << "it is float" << std::endl; }

int main()
{
   f(3.5);
   f(3.5f);
}

Output:

it is double
it is float

answered Apr 28 '12 at 07:29

Nawaz

327,095
105
629
812

2

`3.5 == 3.5f` *is* guaranteed by IEEE754, which most "normal" systems use these days. – fredoverflow Apr 28 '12 at 08:10
But 10.0f/3.0f is not guaranteed to be equal to 10.0/3.0. He happened to pick an exact number. – David Hammen Apr 28 '12 at 08:21
@FredOverflow: Yes, *"3.5 == 3.5f is guaranteed by IEEE754"*, but it is not guaranteed by C++, as an implementation can choose *any* floating-point format. – Nawaz Apr 28 '12 at 11:43

score 2 · Answer 2 · answered Apr 28 '12 at 07:29

2

Float are not an exact number. Comparing them with == is never an good idea. Here is why: What Every Computer Scientist Should Know About Floating-Point Arithmetic

answered Apr 28 '12 at 07:29

RvdK

18,577
3
56
100

8

Floats _do_ represent exact numbers; however, there are many numbers which cannot be represented by any float. – Mankarse Apr 28 '12 at 07:48

Are 3.5 and 3.5f the same in C++?

2 Answers2