Intel has developed a set of recommended algorithms for performing non-IEEE floating-point division, square root, reciprocal, and reciprocal square root operations on the Intel® Itanium® architecture. All the Intel-provided algorithms have been carefully designed to provide results that have an error of no more than 1 ulp (unit-in-the-last-place) when rounding to nearest mode is used (but most often less than 0.5 ulp), and no more than 1.5 ulp when other rounding modes are used. The document provides an overview of the scope of the algorithm set, describes the mathematics of each algorithm, and includes each in the form of a function coded in assembly language for the Intel® Itanium™ processors. The intention is that these can be inlined by compilers, used as the core of mathematical libraries, or called on as macros by assembly language programmers.
| File Name/Size: |
| non_ieee_divsqrt.pdf
656 Kbytes |
| Download From: |
| U.S. FTP Server (Download within North America) |
