Debug vs Release Memory Allocation Performance Issue

[spiff]

We are migrating from VC++ 6 to VC++ 2003. It is a plain, unmanaged application with both C and C++ source. When running the debug build, even outside the debugger, the memory allocation/deallocation performance appears to be orders of magnitude slower than in VC++ 6. The release build runs fine - no performance problems.

We've rewritten some of the code to do fewer memory allocations/deallocations and that has helped those pieces. However, the application is about 400K lines of code, so we'd obviously prefer not to have to reengineer the entire codebase.

Are there any compiler switches or macro definitions that will speed up the debug memory manager? Basically, what we want is the same performance that we were able to get when building with VC++ 6.

[Balaji J]

First of all what is the problem if the debug version runs slower than the release version, as the release version is the one that going to be shipped. Send the compiler options used for the debug build so that one can see what really is the compiler option that causes the program to run slowly.

Balaji.

"spiff" wrote:

We are migrating from VC++ 6 to VC++ 2003. It is a plain, unmanaged application with both C and C++ source. When running the debug build, even outside the debugger, the memory allocation/deallocation performance appears to be orders of magnitude slower than in VC++ 6. The release build runs fine - no performance problems.

We've rewritten some of the code to do fewer memory allocations/deallocations and that has helped those pieces. However, the application is about 400K lines of code, so we'd obviously prefer not to have to reengineer the entire codebase.

Are there any compiler switches or macro definitions that will speed up the debug memory manager? Basically, what we want is the same performance that we were able to get when building with VC++ 6.