Analyzers
Talk to fellow users of Intel Analyzer tools (Intel VTune™ Profiler, Intel Advisor)
4975 Discussions

Analyzing an application at the loop level

ncubed01
Beginner
323 Views
I am currently evaluating Vtune to see if it can help me get some performance data about my application. I would really like to find out which loops in my functions are contributing the most time to the total execution time. I have been able to access the source view and it shows the self time of a function, but does not give any information about the lines of code or loops inside the function. Is it possible to analyze at this level? How do I do it?
0 Kudos
1 Solution
TimP
Honored Contributor III
323 Views
If you have built with debug symbols (-g for linux or gcc, -Zi for Windows other than gcc) you should see the event sampling counts displayed on source lines associated with the loops. You would specify explicit optimization levels, as the debug options default to no optimization. You may wish to disable in-lining or interprocedural optimization, so that the events get associated with original source lines, rather than the site of in-lining.

View solution in original post

0 Kudos
2 Replies
TimP
Honored Contributor III
324 Views
If you have built with debug symbols (-g for linux or gcc, -Zi for Windows other than gcc) you should see the event sampling counts displayed on source lines associated with the loops. You would specify explicit optimization levels, as the debug options default to no optimization. You may wish to disable in-lining or interprocedural optimization, so that the events get associated with original source lines, rather than the site of in-lining.
0 Kudos
ncubed01
Beginner
323 Views
Quoting - tim18
If you have built with debug symbols (-g for linux or gcc, -Zi for Windows other than gcc) you should see the event sampling counts displayed on source lines associated with the loops. You would specify explicit optimization levels, as the debug options default to no optimization. You may wish to disable in-lining or interprocedural optimization, so that the events get associated with original source lines, rather than the site of in-lining.

You are awesome. Thank you.
0 Kudos
Reply