Those constants are integers, so the math is done with integer math. Try
float p1 = (6.0 / 100.0);
edit — @Stephen Canon wisely points out that since "p1" is a float
, there's no reason not to do the whole computation as float
:
float p1 = (6.0f / 100.0f);
Now, since those things are both constants, I'd imagine there's a really good chance that the work is going to be done by the compiler anyway. It's also true that because on some modern machines (ie, Intel architecture), the floating-point processor instruction set is sufficiently weird that something that seems like an obvious "optimization" may or may not work out that way. Finally I suppose it might be the case that doing the operation with float
constants could (in some cases) give a different result that doing the operation with double
values and then casting to float
, which if true would probably be the best argument for deciding one way or the other.