Why is 'afxDump << etc' slower than TRACE(); in Visual C++ .net? Is it just me?

I've knocked up a quick demo app and added the following button handler:

void CafxDumpTimeTestDlg::OnBnClickedBtnStart()
{
     // TODO: Add your control notification handler code here
     DWORD dwStart = 0L;
     DWORD dwEnd = 0L;
     DWORD dwLapse = 0L;
     DWORD dwLoop = 0L;

     dwStart = ::GetTickCount();

     for(dwLoop = 0 ; dwLoop < 20 ; dwLoop++){
          afxDump << "Loop " << dwLoop << "\n";
     }

     dwEnd = ::GetTickCount();
     dwLapse = dwEnd - dwStart;

     afxDump << "==== afxDump Lapsed " << dwLapse << "\n";

     dwStart = ::GetTickCount();

     for(dwLoop = 0 ; dwLoop < 20 ; dwLoop++){
          TRACE("Loop %i\n", dwLoop);
     }

     dwEnd = ::GetTickCount();
     dwLapse = dwEnd - dwStart;

     afxDump << "==== TRACE Lapsed " << dwLapse << "\n";
}

and I get the results in the debug output window:
Loop 0
Loop 1
Loop 2
Loop 3
Loop 4
Loop 5
Loop 6
Loop 7
Loop 8
Loop 9
Loop 10
Loop 11
Loop 12
Loop 13
Loop 14
Loop 15
Loop 16
Loop 17
Loop 18
Loop 19
==== afxDump Lapsed 3575
Loop 0
Loop 1
Loop 2
Loop 3
Loop 4
Loop 5
Loop 6
Loop 7
Loop 8
Loop 9
Loop 10
Loop 11
Loop 12
Loop 13
Loop 14
Loop 15
Loop 16
Loop 17
Loop 18
Loop 19
==== TRACE Lapsed 1202


which makes 'TRACE()' maybe three times quicker than afxDump !