Hard and Soft Page Faults
I am using AQTime Pro 7.10 to profile an unmanaged MSVC 6.0 application for page faults performance. This is a repeatable application that does the same thing each time it is run.
If I run AQTime, I get a total Hard Page Fault count in the 50 range. Changing the capture parameter in AQTime and running again, I get a total Soft Page Fault count in the 110,000 range.
With both runs, the Windows Task Manager Page Faults value (which should include HPF and SPF together) is in the 6,000 range right before I terminate the application to get the above results from AQTime.
Then I run Windows Performance Monitor on the same application, monitoring Page Faults/sec and Transition Faults/sec. When the two deviate away from zero, I observe peaks in the PF/s with corresponding low TF/s values (e.g. 100:1 ratio). According to Microsoft documentation, TF corresponds to soft page faults, which then says that my application is exhibiting higher HPFs than SPFs.
Any ideas on how to set up a "tie-breaker" experiment?
If I run AQTime, I get a total Hard Page Fault count in the 50 range. Changing the capture parameter in AQTime and running again, I get a total Soft Page Fault count in the 110,000 range.
With both runs, the Windows Task Manager Page Faults value (which should include HPF and SPF together) is in the 6,000 range right before I terminate the application to get the above results from AQTime.
Then I run Windows Performance Monitor on the same application, monitoring Page Faults/sec and Transition Faults/sec. When the two deviate away from zero, I observe peaks in the PF/s with corresponding low TF/s values (e.g. 100:1 ratio). According to Microsoft documentation, TF corresponds to soft page faults, which then says that my application is exhibiting higher HPFs than SPFs.
Any ideas on how to set up a "tie-breaker" experiment?
