--===============0542703041==
Content-Type: multipart/signed;
  boundary="nextPart1255253.ske5eDS5LR";
  protocol="application/pgp-signature";
  micalg=pgp-sha1
Content-Transfer-Encoding: 7bit

--nextPart1255253.ske5eDS5LR
Content-Type: Text/Plain;
  charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

On Monday 13 July 2009, Thiago Macieira wrote:
> Our stacks are far too large. That's easily visible in many applications.
> Here's an excerpt from /proc/`pidof plasma-desktop`/smaps:
>
> 096d8000-168ad000 rw-p 096d8000 00:00 0          [heap]

em. our stacks are far too large, so here's the heap measure? :)

aseigo@independence:~/Documents> cat /proc/`pidof plasma-desktop`/smaps | g=
rep=20
stack -A4
bf95e000-bf973000 rw-p bffeb000 00:00 0          [stack]
Size:                 84 kB
Rss:                  80 kB
Pss:                  80 kB
Shared_Clean:          0 kB

anyways:

> Size:             214868 kB
> Rss:              214576 kB
> Pss:              214576 kB
>
> Can anyone tell me why Plasma needed to allocate 209.8 MB of RAM? (binary
> 1024 units)

probably a memory leak somewhere. or maybe something in your configuration=
=20
that's taking an unreasonable amount of ram.

profiling would be helpful. even knowing what your configuration is would b=
e=20
helpful. (there have been reports that suggest there's a leak somewhere in =
the=20
slideshow wallpaper, though it hasn't been identified. iirc you use a=20
slideshow paper over two screens, yes?)

so .. here's my laptop's current uptime (for which time it's been constantl=
y=20
logged into a kde4 session):

 12:42am  up 3 days 22:09,  23 users,  load average: 0.41, 0.35, 0.37

and the [heap] entry from plasma's smaps is:

 08055000-09729000 rw-p 08055000 00:00 0          [heap]
Size:              23376 kB
Rss:               23096 kB
Pss:               22984 kB
Shared_Clean:          0 kB

but then right now on this machine i'm only running one panel, one screen w=
ith=20
a single image wallpaper, tasks, kickoff, device notifier, system tray,=20
battery, digital clock weather and translation widgets.

a good chunk of that 23mb is likely to be the wallpaper i'm using (though=20
maybe that isn't permanently stored in the heap; i haven't looked yet) and =
the=20
rest will be various bits of windows and widgets and data structures.

nearly zero profiling has gone on for memory usage within plasma, and that=
=20
certainly shows. it could also be a lot worse.

> There are definitely memory leaks in some places. In other places, we need
> profiling to find out why we're wasting so much memory.

yep.

> However, I also know that we are using more memory for caching things,
> pre-rendered pixmaps for example. We use more memory to make the system
> faster. So I don't expect KDE 4 to reach KDE 3 levels again.

quite true ... there are probably places we do things unnecessarily.

> No. I expect KDE 4 to excel in other areas, like battery time. For
> example, right now, Plasma wakes up between 20 and 60 times per second due
> to timers. Killing Plasma makes my laptop's power consumption drop by 1 W

bugs get fixed as and when reported. this was probably the taskbar timers=20
issue i fixed before leaving for vancouver.

and really,  it's kind of that simple. most of the things in this thread ar=
e=20
of the "huh, something's broken or not as good as it should be" type to whi=
ch=20
the answer is simple:

* identify the problem specifically ("timers are firing" =3D=3D not specifi=
c=20
enough; "i remove the taskbar and it goes quiet" =3D=3D a great start)

* figure out a solution ("let's look at use of timers and QObject::connect =
in=20
the tasks widget and libtaskmanager")

* fix it

there really are no shortcuts. and as such, unless this kind of discussion =
is=20
accompanied with some of the above steps it's all just an exercise is=20
uselessness.

and getting real numbers, not just garbage from top-alikes, is pretty=20
important, too.

=2D-=20
Aaron J. Seigo
humru othro a kohnu se
GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA  EE75 D6B7 2EB1 A7F1 DB43

KDE core developer sponsored by Qt Software

--nextPart1255253.ske5eDS5LR
Content-Type: application/pgp-signature; name=signature.asc 
Content-Description: This is a digitally signed message part.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.9 (GNU/Linux)

iEYEABECAAYFAkpdfAIACgkQ1rcusafx20PxhwCfSIwwGcCbJ+HI9+9Jz9POYcts
u74AnRh91uAxI3dN7s/1tOyXiYE7sCIZ
=bfJ3
-----END PGP SIGNATURE-----

--nextPart1255253.ske5eDS5LR--

--===============0542703041==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<

--===============0542703041==--