Click above to get retro games delivered to your door ever month!

(16-bit only) The compiler will generate in-line 80x87 numeric data
processor instructions into the object code for floating-point operations.
 Depending on which library the code is linked against, these instructions
will be left as is or they will be replaced by special interrupt
instructions.  In the latter case, floating-point will be emulated if an
80x87 is not present.  This is the default floating-point option if none is
specified.
(32-bit only) The compiler will generate in-line 387-compatible numeric data
processor instructions into the object code for floating-point operations.
 When any module containing floating-point operations is compiled with the
"fpi" option, coprocessor emulation software will be included in the
application when it is linked.

For 32-bit Watcom Windows-extender applications or 32-bit applications run
in Windows 3.1 DOS boxes, you must also include the WEMU387.386 file in the
[386enh] section of the SYSTEM.INI file.

Example:

     device=C:\WATCOM\binw\wemu387.386

Note that the WDEBUG.386 file which is installed by the Watcom Installation
software contains the emulation support found in the WEMU387.386 file.

Thus, a math coprocessor need not be present at run-time.  This is the
default floating-point option if none is specified.  The macros __FPI__ and
__SW_FPI will be predefined if "fpi" is selected.

+--------------------------------------------------------------------------+
|   Note:  When any module in an application is compiled with a particular |
| "floating-point" option, then all modules must be compiled with the same |
| option.                                                                  |
  If you wish to have floating-point emulation software included in the    |
| application, you should select the "fpi" option.  A math coprocessor     |
| need not be present at run-time.                                         |
|                                                                          |
  Different math libraries are provided for applications which have been   |
| compiled with a particular floating-point option.  See the section       |
| entitled Watcom C/C++ Math Libraries.                                    |
|                                                                          |
+--------------------------------------------------------------------------+
See the section entitled The NO87 Environment Variable for information on
testing the math coprocessor emulation code on personal computers equipped
with a coprocessor.