| 摘要 |
A method and apparatus used for performing a floating pointing operation has the ability to calculate a square root of a number. An approximation to the inverse of the square root of the number is provided via a step (18). Steps (20 and 22) are used to improve the precision of the inverse of the square root until a predetermined precision is attained. The inverse of the square root, which has a predetermined precision, is used along with the number to generate both an exact floating point value and a small floating point value via steps (24 and 26). The exact and small floating point values are added together in a sum and manipulated to fit into the floating point representation available to the apparatus. The sum is a substantially close approximation to the square root of the number and is either output directly or slightly modified numerically to more accurately represent the square root of the number.
|
