Hey List!

I wrote a GUI for visualizing Massif log data, and thought it would be a good 
idea to announce it here:

http://kde-apps.org/content/show.php/Massif+Visualizer?content=122409
http://gitorious.org/massif-visualizer

Here the description from the project website:
~~~~
Massif Visualizer is a tool that - *who would guess that* - visualizes massif 
data. You run your application in Valgrind with `--tool=massif` and the open 
the generated `massif.out.%pid` in this application. You can also compress the 
log with Gzip or Bzip2 and open it transparently with the visualizer.

The application consists of three parts:

# The Overview Chart

The first thing you will notice is a nice chart that displays the same as e.g. 
`ms_print` does in Ascii-Art: total memory consumption over time.

What Massif-Visualizer goes further is by additionally showing the top ten 
most cost-intensive locations in your code as a stacked graph below the total 
cost. The graph also reacts on user-interaction.

This view you can use for

- checking whether your application has memory leaks
- finding too expensive peaks
- finding locations that significantly contribute to the overall memory 
consumption of your application

# The Snapshot Data Tree

Directly next to the above chart, you will see a tree with all of the massif 
data. The tree items are colored depending on their cost, with red opaque 
being the most interesting (peak) elements. Green/transparent items are 
negligible and do not add significant cost to your application.

You can also search the tree and when you select something in it, the snapshot 
gets highlighted in the overview chart and the call graph gets updated.

# The Call Graph for Detailed Snapshots

Massif generates a few detailed snapshots that essentially make up the tree. 
If you want to get an overview in a more comfortable way than the simple tree 
view, switch over to the detailed snapshot tab and see
the tree visualized as a call graph. Zoom in, zoom out, use the birds eye view 
and see what contributes to a given snapshot. Note that function calls with 
the same memory cost are grouped to easily find the interesting parts.
~~~

I'd appreciate any kind of feedback, esp. in regard to whether that tool helps 
or whether my choices here and there are good. I think mostly of the (simple) 
algorithm that gets the interesting data points in the stacked chart below the 
total mem consumption. Currently I decided to select the first item that forks 
in the Massif data tree. Anyways, you should probably use it and tell me 
whether the graphs and charts are OK or could be done better.

It's depending on kdelibs from 4.X (tested with 4.3 and above). Furthermore 
there is currently a compile and runtime dependency on KGraphViewer from 
Trunk: http://extragear.kde.org/apps/kgraphviewer/
I plan to make the latter optional.

Have a nice holiday where applicable or at least a nice weekend.

To the valgrind devels out there: Thanks for the amazing work, I love it!
-- 
Milian Wolff
[email protected]
http://milianw.de

Attachment: signature.asc
Description: This is a digitally signed message part.

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Valgrind-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/valgrind-users

Reply via email to