floats

[sans-hubris]

I'm trying to remember how floats are stored in memory. I remember that a certain number of bytes are stored for the significant digits in the float, and then there are others that are stored for the power of ten.

I don't remember the details of which bits are used for what though. It's probably architecture dependent, right?

[Strike]

IEEE 754 standard, is usually what is used. So many bites for the "characteristic" and so many bites for the "mantissa" and whatnot. I haven't studied it in a long time, but that standard name and number always stuck with me. Google should help now.

[Strike]

http://http.cs.berkeley.edu/~wkahan/ieee754status/ieee754.ps

found it :)