The information in this article applies to:
- Microsoft Windows Software Development Kit (SDK) for Windows
version 3.1
SUMMARY
WinFloat: Floating-Point Samples for Windows(TM)
WinFloat consists of five programs and libraries that provide sample code
and tools to explore the floating-point mechanisms of Microsoft(R) Windows:
- FLOATAPP.EXE: Windows-based application used as a shell for testing
application and dynamic-link library (DLL) floating-point issues.
This application tests floating-point speed, generates exceptions,
installs signal handler, masks and unmasks exceptions, and
communicates with WIN87EM.DLL. Some capabilities depend on the math
package chosen and/or the presence of a math coprocessor.
- FLOATDLL.DLL: Floating-point Windows DLL. FLOATDLL is a subset of
FLOATAPP, to which it dynamically links. FLOATDLL can generate and
handle floating-point exceptions, and can be configured to install
and maintain its own signal handler or use the default (or
application) handler.
- FLOATDOS.EXE: Floating-point MS-DOS(R) application. FLOATDOS is a
subset of FLOATAPP that runs outside of Windows (or in a "DOS
box"), thus allowing you to compare an MS-DOS application with a
Windows-based application for exception handling and (most
importantly) calculation speed.
- HIDE87.COM: MS-DOS terminate-and-stay-resident (TSR) program that
hides the 80x87 from Windows-based applications. HIDE87 must be
installed before you start Windows so you can trick Windows into
thinking there's no math coprocessor.
- SHOW87.EXE: MS-DOS control program that disables the HIDE87 TSR.
SHOW87 provides a way to turn off HIDE87 without rebooting the
machine. As with HIDE87, SHOW87 must be executed before you start
Windows so it can detect the math coprocessor.
See the sample code and the "Floating Point in Windows" technical
article for descriptions of these programs, their coding techniques,
and the Windows floating point architecture. For information on
building these programs and discussions of C version 6.0 versus C/C++
version 7.0 and emulator versus alternate math packages, see the make
file. Note that these sample have major limitations if compiled with C
6.0.
WINFLOAT was built and tested under Microsoft Windows version 3.1
using Microsoft Macro Assembler version 6.0 and Microsoft C/C++
version 7.0.
Download WINFLOAT.EXE, a self-extracting file, from the Microsoft Software
Library (MSL) on the following services:
|