Understanding and Using Floating Point Numbers

Most programmers have heard or observed one strange thing or another about floating point numbers. For example, we often discover that floating point numbers that look the same do not necessarily satisfy C's "==" test. New programmers are usually taught never to use == for floating point numbers for this reason. Occasionally we run into other exceptional cases, for instance mathematically sound formulae which, when implemented using floating point, produce seemingly random or disappointingly inaccurate results.

http://www.cprogramming.com/tutorial/floating_point/understanding_floating_point.html