use correct rounding functions this should fix a floating point exception on some systems lets hope these arent missing anywhere, if so another solution must be found