Intel® Fortran Compiler - Default floating point precision differs between Linux*/Mac OS* and Windows*

Submit New Article

November 3, 2008 12:00 AM PST


The default FPU precision on Linux* and Mac OS* is a 64-bit significant. This equates to the compiler switch -pc80. The corresponding setting on Windows* is a 53-bit significant, equivalent compiler switch -Qpc64. This difference may account for a variation in results when comparing execution of identical programs/algorithms on Linux/Mac OS* and Windows.

Note: These refer to internal or intermediate, register to register calculations, which may be of greater precision than the declared precision for the variables used. When results are stored back to memory, they are rounded to the declared precision for the variables as necessary.

These defaults are at the operating system level and outside the control of the compiler.
 


Do you need more help?


This article applies to: Intel® C++ Compiler for Linux* Knowledge Base,   Intel® C++ Compiler for Mac OS X* Knowledge Base,   Intel® C++ Compiler for Windows* Knowledge Base,   Intel® Fortran Compiler for Linux* Knowledge Base,   Intel® Fortran Compiler for Mac OS X* Knowledge Base,   Intel® Visual Fortran Compiler for Windows* Knowledge Base