Objective-C Programming: The Big Nerd Ranch Guide

Floating-point numbers

If you need a number with a decimal point, like 3.2, you use a floating-point number. Most programmers think of a floating-point number as a mantissa multiplied by 10 to an integer exponent. For example, 345.32 is thought of as 3.4532 x 10². And this is essentially how they are stored: a 32-bit floating number has 8 bits dedicated to holding the exponent (a signed integer) and 23 bits dedicated to holding the mantissa, with the remaining 1 bit used to hold the sign.

Like integers, floating-point numbers come in several sizes. Unlike integers, floating-point numbers are always signed:

float g;       // 32 bits
double h;      // 64 bits
long double i; // 128 bits

Tokens for displaying floating-point numbers