There is lots of explanations and resources in internet. some of them is mumbo-jumbo, some of them is somewhat clear saying that it's because of FPU works differently and e.t.c. but no clear and short answer is shown though.. can anybody explain in a popular language terms Why and How Division operation is not giving accurate and precise result when it comes to Float numbers?
Asked
Active
Viewed 310 times
0
Giorgi Tsiklauri
- 9,715
- 8
- 45
- 66
-
2Is the suggested [related question](http://stackoverflow.com/questions/588004/is-floating-point-math-broken?rq=1) unsatisfactory for some reason? – Damien_The_Unbeliever Jul 29 '15 at 07:05
-
3Long story short - it works fine, but peoples *decimal* expectations don't match the *binary* reality. – Damien_The_Unbeliever Jul 29 '15 at 07:06
1 Answers
2
Division is one of IEEE 754 basic operations for which the result is, in the default rounding mode, the nearest representable floating-point number to the real result. Here is one deck of slides, of which “IEEE-754: specification of the arithmetic operations” lists division as one of the operations having this property.
What other result would you expect for floating-point division that would be more accurate?
Pascal Cuoq
- 79,187
- 7
- 161
- 281