To understand the problem, please read What Every Computer Scientist Should Know About Floating-Point Arithmetic and perlfaq4.

Floating point arithmetic is imprecise and you will need to move to fixed-point arithmetic if you want your idea to work as you expect it.