Re: [Boost-users] [Boost.Graph] Profiling using the graph library

From: Ewgenij Sokolovski (ewgenijkkg_at_[hidden])
Date: 2007-06-30 10:31:11

> The problem is that the BGL data structures and algorithms use many
> small, inlined functions. These functions need to be inlined away for
> performance reasons, but that means that you don't get profiling
> information about them. If the compiler *did* put profiling information
> in them, it would be useless: the code of profiling these small routines
> exceeds the cost of the routines themselves, producing very misleading
> profiling information.
> 
>   - Doug

Ah, gradually I can see the light at the end of the tunnel:) But if you read the tutorial on gprof, then you will see the following:

If you compile only some of the modules of the program with `-pg', you can still profile the program, but you won't get complete information about the modules that were compiled without `-pg'. The only information you get for the functions in those modules is the total time spent in them; there is no record of how many times they were called, or from where. This will not affect the flat profile (except that the calls field for the functions will be blank), but will greatly reduce the usefulness of the call graph.
http://www.cs.utah.edu/dept/old/texinfo/as/gprof.html#SEC2

So the names of the boost functions as well as the time spent inside of them should be shown in the profiler output. Although the corresponding libraries were not compiled with profiling options. Merely the amount of calls and caller names will be not available.
But this is apparently not the case here. So what's wrong?

And why these other two functions were in the profiler output? They were not called by my own function directly, so they should be somewhere lower in the hierarchy.

Ewgenij
-- 
GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS.
Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail