Why do I have different results by casting values

Question

I'm really wasting a lot of time solving a problem in my code. Now it is working, but I do not understand what is going wrong.

Using QT5 / gcc

int pos;
uint m_endZeit, m_startZeit;

float secPerPx = (pos * (m_endZeit - m_startZeit)) / static_cast<float>(width());
uint a = secPerPx + m_startZeit;
uint b = static_cast<uint>(secPerPx) + m_startZeit;
std::cout << a << " vs. " << b << std::endl;

Output:

1404809856 vs. 1404809893

Can anyone explain this to me? Why is this not the same?

Maybe because `(uint)(secPerPx + m_startZeit) != ((uint)secPerPx + m_startZeit)`? — vahancho, Jul 08 '14 at 09:46
Because floats does not gurantee absolute results. Float computation is always rounded to some bounds. You may read some wiki: http://en.wikipedia.org/wiki/Floating_point — Dmitry Sazonov, Jul 08 '14 at 09:48

score 3 · Accepted Answer · edited May 23 '17 at 11:50

What range of numbers can be represented in a 16-, 32- and 64-bit IEEE-754 systems?

The Precision of a float is roughly 8-9 digits and you already have 10.

uint a = secPerPx + m_startZeit;

The above code first casts m_startZeit to a float, adding the 2 values and then casts it back to an Integer.

uint b = static_cast<uint>(secPerPx) + m_startZeit;

This is casting the float to an integer BEFORE they both get added together.

Why do I have different results by casting values

1 Answers1