Power & Memory Usage Of GNOME, KDE, LXDE & Xfce

[BenderBRodriguez]

Then you should change the title of that post/article whatever, Ubuntu != Linux, it is not an oracle that says "KDE is a memory hog because it is a memory hog on Ubuntu", i can testify as my KDE eats 89MB of memory despite your article (gentoo).

[mtippett]

Then you should change the title of that post/article whatever, Ubuntu != Linux, it is not an oracle that says "KDE is a memory hog because it is a memory hog on Ubuntu", i can testify as my KDE eats 89MB of memory despite your article (gentoo).

What does gnome, xfce and so on use on your system? How are you measuring the 89MB?

[kraftman]

What does gnome, xfce and so on use on your system? How are you measuring the 89MB?

The obvious thing is Phoronix didn't measure Gnome, KDE, XFCE memory and power usage, but Ubuntu, Kubuntu and Xubuntu. What's strange Phoronix is claming they actualy did! If you're asking what gnome and xfce use on somebody's system then you should agree it's rather impossible (or very hard) to measure DEs power and memory usage (what apps make Gnome to be Gnome, KDE to be KDE etc.). I don't understand why don't you agree with the obvious fact - the title of the article is WRONG and it's also very misleading.

[mugginz]

Some numbers for your consideration.

All tests performed with a virtual machine courtesy of KVM.
1024M RAM
8GB Virtual Drive
1 CPU
Sound Card Emulated

Kubuntu Karmic Koala
308M....Karmic Release no Updates and after clean boot
413M....Karmic Release with Updates and after clean boot
269M....Karmic Release with Updates and after second clean boot
265M....Karmic Release with Updates and after third clean boot
265M....Karmic Release with Updates and after removal of desktop plasmoids and fourth clean boot
216M....Karmic Release with Updates and no desktop plasmoids after fifth clean boot

Kubuntu Lucid Alpha3
278M....Kubuntu Lucid Alpha3 with Updates and after clean boot
270M....Kubuntu Lucid Alpha3 with Updates after removal of desktop plasmoids and clean boot
266M....Kubuntu Lucid Alpha3 with Updates after removal of desktop plasmoids and second clean boot

Ubuntu Lucid Alpha3
195M....Lucid Alpha3 with Updates and after clean boot at GNOME desktop
(Following numbers at KDE desktop)
394M....Lucid Alpha3 with Updates and added KDE goodness after first boot
388M....Lucid Alpha3 with Updates and added KDE goodness after second boot
233M....Lucid Alpha3 with Updates and added KDE goodness, after removal of desktop plasmoids and third clean boot
252M....Lucid Alpha3 with Updates and added KDE goodness, after desktop plasmoids put back without clean boot
234M....Lucid Alpha3 with Updates and added KDE goodness, after fourth clean boot with desktop plasmoids
242M....Lucid Alpha3 with Updates and added KDE goodness, after fifth clean boot and with desktop plasmoids
232M....Lucid Alpha3 with Updates and added KDE goodness, after sixth clean boot with desktop plasmoids
228M....Lucid Alpha3 with Updates and added KDE goodness, after removal of desktop plasmoids and seventh clean boot

Clearly Gnome in Aplha3 has a better memory footprint than does KDE no matter whether KDE is sitting on Ubuntu or Kubuntu at the moment.

There is considerable fluctuation between boots. This could be due to various reasons.

Assuming that a particular install is allowed to "settle down" there seems to be no reason for the 100M to 150M delta between Gnome and KDE in the Phoronix tests unless the test suite is using a GUI and therefore loading up the KDE session with both KDE and Gnome dependencies.

My tests were carried out by simply starting a Konsole session for KDE and a Gnome terminal for Gnome, then executing the following command.

free -m

[kraftman]

Well, that's over 300MB RAM difference, isn't it? Do you really think, that using different versions in Ubuntu (compared to Arch Linux) make that huge difference? More than twice the usage?

I'm questioning Phoronix' test results, not (K)Ubuntu.

My friend's Gnome consumes about 70MB - Gentoo.

[kraftman]

beside that, michael likes to test out of the box performance and figures for popular distributions. not alot of people like to build their own distro with vanilla packages. also, phoronix is free guys. chill out.

Distros yes, but not DEs :>

[kraftman]

@Mugginz

You could also remove printer applet and then kill two python proceses - about 20MB each.

[mugginz]

I should've added that the Ubuntu Lucid Alpha3 at a Gnome desktop test was performed after a few reboots so would likely reflect the same sort of settling effects as did the other tests.

I'm re-installing now to re-run the tests Gnome though.

[mugginz]

@Mugginz

You could also remove printer applet and then kill two python proceses - about 20MB each.

The printer applet doesn't seem to be running on either Ubuntu Gnome or KDE for the test installs. This could be because no printer is connected.

[mtippett]

The obvious thing is Phoronix didn't measure Gnome, KDE, XFCE memory and power usage, but Ubuntu, Kubuntu and Xubuntu. What's strange Phoronix is claming they actualy did! If you're asking what gnome and xfce use on somebody's system then you should agree it's rather impossible (or very hard) to measure DEs power and memory usage (what apps make Gnome to be Gnome, KDE to be KDE etc.). I don't understand why don't you agree with the obvious fact - the title of the article is WRONG and it's also very misleading.

As per mugginz numbers - the gnome vs kde (as either ubuntu+kde or kubuntu) - the numbers still match. I would be interested in say gentoo or arch with gnome/kde too). Obviously screen size, video played, etc are all relevant to the final numbers.

Regarding the title..

The variable that is changing is the desktop environment. The primary measures that were taken were power and performance. The test was consistent.

I guess the title could have been "Power and Memory Usage with bundled GNOME/KDE/LXDE/XFCE on Ubuntu Karmic while running Video" - but I think that would just look silly.

Matthew