July 2013 Archives

Fixed point arithmetic is perhaps the simplest alternative to floating point. Fixed point numbers maintain a fixed number of digits after the point, rather than a fixed number of digits of precision. Typically they are represented by an integer with the assumption that some constant number of the least significant base 10 digits fall below the decimal point. For example, assuming 2 decimal places, we would represent π with 314.

Full text...

submit to reddit