Re: Loop profiling
- From: "user923005" <dcorbit@xxxxxxxxx>
- Date: 28 Mar 2007 15:40:06 -0700
On Mar 28, 11:50 am, Chrisitan Christmann <plfr...@xxxxxxxx> wrote:
On Wed, 28 Mar 2007 16:43:34 +0000, Randy Howard wrote:
Any ideas? Right now, I help myself out with C macros but the
instrumentation is still very tedious.
Well, easy answer would be to write the loop iteration count statistics
to a log file, in a well-defined format that included the function
name, a line number, etc., then have another tool (or even
a shell script using grep intelligently) to pull the data back out and
present it in a more nicely formatted way.
Thank you for your ideas. But, as I understood your ideas, they all
expect the user to add some individual code for each loop under test.
That's what I'm currently doing, but for more complex applications
this becomes a tedious job. This was the reason why I was looking for
an automatic approach, something in the direction of gcov...
There are some tools that might be helpful. For example:
http://covtool.sourceforge.net/
I did a port of this to WIN32 if you want it. It's C++ specific.
It's also aimed at code coverage rather than profile per se, but it
would be a good starting point. It is pretty clever in that it parses
the source code files and does the instrumentation for you.
There are lots of similar things on SourceForge.
.
- References:
- Loop profiling
- From: Christian Christmann
- Re: Loop profiling
- From: Ed Prochak
- Re: Loop profiling
- From: Christian Christmann
- Re: Loop profiling
- From: Randy Howard
- Re: Loop profiling
- From: Christian Christmann
- Re: Loop profiling
- From: Randy Howard
- Re: Loop profiling
- From: Chrisitan Christmann
- Loop profiling
- Prev by Date: Re: implementation note for scapegoat tree
- Next by Date: Re: implementation note for scapegoat tree
- Previous by thread: Re: Loop profiling
- Next by thread: Re: Loop profiling
- Index(es):
Relevant Pages
|