player999

Huge error in floating point computation

Discussion created by player999 on Dec 13, 2011
Latest reply on Dec 15, 2011 by player999

Comrads, I'm geting frustrated about math accuracy in OpenCL

Here is the problem, simple multiplication:

0.647059 * -1.148511 = -1.040416

WHY IT IS SO?

I suppose it is some compiler configs, but i don't know which o_O

Outcomes