IBM is providing software for OS/2 WARP to assist customers concerned
with the accuracy of the Pentium floating point operations.  This
software effectively "turns off" the floating point support of the
Pentium processor.  This software detects the specific level of the
Pentium chip and redirects applications to use floating point
emulation.  However, using floating point emulation can degradate the
performance of your application(s). For typical spreadsheet and word
processor applications, this degradation will be minimal.  For
floating point-intensive applications, such as CAD, the degradation
will be more noticeable.

For OS/2 Warp customers, this solution works for all OS/2, DOS, and
Windows applications.  This is provided as a device driver that
directs application floating point instructions to be executed by the
available emulators including OS/2's built-in emulator for OS/2
applications, the emulators either built into DOS or Windows
applications or the floating point emulator shipped as part of WINOS2
(OS/2's Windows support).  Because the application is "informed" that
hardware floating point is not available, applications that require
hardware floating point support will not run.  The solution completely
prevents applications (DOS, Windows, and OS/2) from using the
Pentium's floating point support.

IBM is making this software available on most of the commercial
on-line services, its own bulletin board systems, and across the
Internet.

To Install type DDINSTAL on the OS/2 command prompt and follow directions.  

This install will copy 586npx.sys to the os2 subdir, and add a
BASEDEV=586NPX.SYS line to your config.sys file.

586npx.sys  - will turn on the OS2 floating point emulator if it detects
              a pentium with a bad math co-processor.

If you want to at a later time enable your pentium math co-processor
you can REM out the statement in your config.sys that was added during
the installation, reboot and your Pentium chip will be reenabled for
floating point processes.


