[prev in list] [next in list] [prev in thread] [next in thread] 

List:       oprofile-list
Subject:    Re: Announce:  Sysprof 1.0 -- a sampling, systemwide Linux profiler
From:       Greg Banks <gnb () melbourne ! sgi ! com>
Date:       2005-09-19 10:10:07
Message-ID: 1127124607.27973.192.camel () hole ! melbourne ! sgi ! com
[Download RAW message or body]

G'day,

Distribution changed to the oprofile list.  Here's the
announcement in full, for those that don't read LKML:

On Sun, 2005-09-18 at 07:05, Soeren Sandmann wrote:
> * What is it?
> --------------------------
> 
> Sysprof is a sampling system-wide CPU profiler for Linux.
> 
> Sysprof uses a Linux kernel module to profile the entire system, not
> just an individual application. 
> 
> Of course, sysprof handles threads and shared libraries, and
> applications do not have to be recompiled or instrumented. In fact
> they don't even have to be restarted. 
> 
> Just insert the kernel module and start sysprof.
>
> * Features:
> ----------------------
> 
>     - Profiles all running processes, not just a single application
> 
>     - Call graph support showing time spent in each branch of the call tree
> 
>     - Has a simple graphical interface
> 
>     - Profiles can be loaded and saved
> 
>     - Easy to use: Just insert the kernel module and start sysprof
> 
>     - Supports Fedora debuginfo packages
> 
> 
> * Sysprof 1.0
> --------------------------
> 
> This is the first release of sysprof with a 1.x version number. I am
> very excited about that.
> 
> New features since Sysprof 0.91:
> 
> - support for x86-64, thanks to Mike Frysinger
> - better reporting of time spent in the kernel
> 
> Please report bugs as well as success or failure stories to 
> 
>     sandmann@daimi au dk
> 
> 
> * Where can I get it?
> --------------------------------------------
> 
> Home page:
> 
>        http://www.daimi.au.dk/~sandmann/sysprof/
> 
> Source code:
> 
>        http://www.daimi.au.dk/~sandmann/sysprof/sysprof-1.0.tar.gz
> 
> Sysprof requires
> 
>         - Linux 2.6.11 or later, compiled with profiling support
>         - GTK+ 2.6
> 	- libglade 2.5
> 
> It is known to work out the box on Fedora Core 4.
> 
> 
> Søren
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

As several people have pointed out on LKML, your GUI could be
useful on top of the oprofile kernel code.  So I took a look at
the GUI screenshots on your website and I have some questions...

1.  Does the GUI support choosing trigger events?  Doing
    this horribly detailed and platform-specific task in
    a sensible and easy way would be real value for a GUI.

2.  Why is the Samples: field greyed out?

3.  What units are the Self and Cumulative/Total columns,
    seconds or percent?  Why isn't this obvious?

4.  Is there a way to get a processes-only view, i.e.
    how many cycles the X server is consuming vs the window
    manager, ignoring finer level details?

5.  I'm confused by the Callers and Descendants panes.  Why
    are these asymmetric, in both terminology and presentation?
    I would have thought that if you present a tree of all the
    descendants of a function, you'd also present a tree
    (albeit inverted) of its ancestors.  For my usage, the
    ancestor information is often more interesting.

Greg.
-- 
Greg Banks, R&D Software Engineer, SGI Australian Software Group.
I don't speak for SGI.




-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
oprofile-list mailing list
oprofile-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oprofile-list

[prev in list] [next in list] [prev in thread] [next in thread]