FIX: Incorrect Code Generated When Using Floating-Point and /Og

• Microsoft Visual C++, 32-bit Editions, version 5.0

SYMPTOMS

Incorrect results from floating point operations may be obtained when building with /Og and /Oi or a switch that includes /Og and /Oi (such as /Ox, /O1, or /O2.)

STATUS

Microsoft has confirmed this to be a bug in the Microsoft products listed at the beginning of this article. This bug has been corrected in Visual Studio 97, Service Pack 1.

For additional information about the Visual Studio 97 Service Pack 1, please see the following article in the Microsoft Knowledge Base:

Q170365 INFO: Visual Studio 97 Service Packs - What, Where, and Why

MORE INFORMATION

The following sample demonstrates one form of this bug. As you step through this code in a debugger, you will see that inconsistent results are obtained.

Sample Code

   // compile options /Ogi
   #include <math.h>
   float e1(float x)
   {
      float value;
      if (x < 0.0)
         value = exp(x);
      else
         value = 1.0;
      return value;
   }
   #pragma ("g", off )
   void f()
   {
      float f1 = e1( -1.2f); // 1.00000
      float f2 = exp(-1.2f); // 0.301194
   } 

Additional query words:

Keywords : kbVS97sp1fix kbVS97sp2fix
Version : 5.0
Platform : NT WINDOWS
Issue type : kbbug