| fixed-point (fĭkst'point') adj. Of, relating to, or being a method of writing numerical quantities with a predetermined number of digits and with the decimal located at a single unchanging position. |
fixed-point programming
A number representation scheme where a number, F is represented by an integer I such that F=I*R^-P, where R is the (assumed) radix of the representation and P is the (fixed) number of digits after the radix point.
On computers with no floating-point unit, fixed-point calculations are significantly faster than floating-point as all the operations are basically integer operations. Fixed-point representation also has the advantage of having uniform density, i.e., the smallest resolvable difference of the representation is R^-P throughout the representable range, in contrast to floating-point representations.
For example, in PL/I, FIXED data has both a precision and a scale-factor (P above). So a number declared as 'FIXED DECIMAL(7,2)' has a precision of seven and a scale-factor of two, indicating five integer and two fractional decimal digits. The smallest difference between numbers will be 0.01.
(2006-11-15)