[Powerlord]

Originally Posted by Black Rose I think it's 6 decimals precision if i interpret this Wiki page correctly: http://en.wikipedia.org/wiki/Single-...g-point_format

It can (generally) be anywhere from 6 to 9 digits. Remember, floats are an approximation and are actually stored as an equation in scientific notation in the form x*2^y.

Edit:

Because it's in scientific notation, x is always in the form a.bcdefghijklmno (this number is a 23-bit positive number; the sign bit is stored in a different location).
y is a number between -128 and 127 which has had 128 added to it.

Note that these rules are for signed floats, unsigned floats use a 24-bit positive number and a 0 to 255 exponent range.