FIX: /Og May Cause Incorrect Floating Point Calculation Result

• Microsoft Visual C++, 32-bit Editions, version 6.0

SYMPTOMS

Code compiled with the /Og option can cause incorrect calculation for doubles. In some cases, it loads only half of a double's bytes into memory causing incorrect results in later calculations.

RESOLUTION

Disable global optimizations.

Global optimizations can be disabled for the entire project, or for a particular source file by removing /Og from either the project's release build compiler settings or an individual file's release build compiler settings.

NOTE: /O1 (minimize size), and /O2 (maximize speed) are composite switches that include /Og. If you are using /O1 or /O2 disable global optimizations by adding /Og- to the appropriate compiler settings.

Global optimization can also be disabled on a function by function basis through the use of #pragma optimize as follows:

#ifndef _DEBUG  // current build is not a debug build<BR/>
#pragma optimize("g", off) // disable global optimization<BR/>
#endif // _DEBUG<BR/><BR/>
void SomeFunction()<BR/>

*** Body of SomeFunction Here ***
return;

#ifndef _DEBUG // current build is not a debug build<BR/>
#pragma optimize("g", on) // re-enable global optimization<BR/>
#endif // _DEBUG<BR/> 

STATUS

Microsoft has confirmed this to be a bug in the Microsoft products listed at the beginning of this article.

This bug was corrected in Visual Studio 6.0 Service Pack 3. For more information about Visual Studio service packs, please see the following articles in the Microsoft Knowledge Base:

Q194022 INFO: Visual Studio 6.0 Service Packs, What, Where, Why

Q194295 HOWTO: Tell That Visual Studio 6.0 Service Packs Are Installed

Additional query words:

Keywords : kbservicepack kbCompiler kbVC600bug kbVS600sp2 kbVS600SP1 kbVS600sp3fix
Version : winnt:6.0
Platform : winnt
Issue type : kbbug