Retropie ??? Amibian ???

[sulu]

It is the very same 3 year old chart pulled from the very same page you linked before...?

It is obvioulsy not, as you could have easily seen, had you actually looked at the charts.
Now, where is your comparison between a current Xfce and a 6 years old Gnome 2, showing, that Xfce is more lightweight than Mate?

 

[Letalis Sonus]

It is obvioulsy not, as you could have easily seen, had you actually looked at the charts.

It actually linked it as its source, smartass. Did you quote the wrong post?

Now, where is your comparison between a current Xfce and a 6 years old Gnome 2, showing, that Xfce is more lightweight than Mate?

Here's a 6 year old comparison:

The big issue with those comparison is that many tested systems don't ship the default desktop, e.g. current Xubuntu uses the much more bloated Whiskers main menu. Those little changed applets have a huge influence in the memory footprint, e.g. Mate's clock applet already takes up over 22 MiB on my system. This post tried to filter out the base system and ended up with a memory footprint of about 287 MiB for Xfce 4.12 - Mate has a lot more processes running that make filtering our the actual footprint rather difficult.

To make the comparison even more confusing, here's the same graph but after running for 2 minutes instead of measuring right after the start:

 

[bzar]

Simple memory plots such as these don't really tell the whole story though. You'd need separate plots for resident memory, shared memory and total memory after some use, not just idling.

After a while the mostly unused memory regions get swapped out. Running applications using the same shared libraries mitigates the shared memory footprint effect, sometimes potentially even making it cheaper to run a DE on some popular toolkit instead of opting for a smaller minimum footprint using a homegrown toolkit. What's left in memory after stuff has been swapped out and routinely shared stuff has been discounted for is closer to the actual overhead of the DE, ie. the amount of memory not available for other applications to use or share.

If someone finds a benchmark taking all these into account in some good way, I'd be interested in the results.

 

[sulu]

It actually linked it as its source, smartass. Did you quote the wrong post?

No. Did you refer to the wrong one?
I linked to this one. [1] bzar linked to this. [2] (which links here: [3]) Those are obviously two very different statistics based on different systems but inherently consistent. What they have in common is that they show Mate using less RAM than Xfce.

Here's a 6 year old comparison:

That does (obviously) neither contain statistics for Mate, nor for a CURRENT Xfce.
And that's the point here: You current claim that Xfce uses less RAM than Mate is based on 6 years old statistics comparing 6 years old Xfce with Gnome 2. When I showed you statistics that say something else you blame me for digging up 3 years old data.
When someone else shows you a different comparison confirming my claim, you fail to see that it's two different charts. And when I point you to that failure you call me something I'd rather not repeat. I'm done here.

Simple memory plots such as these don't really tell the whole story though. You'd need separate plots for resident memory, shared memory and total memory after some use, not just idling.

That's true. But I remember that I made my own comparisons when I first saw the graph you linked. I made a minimal Debian installation in VBox, cloned it and installed Xfce in one and Mate in the other. The outcome was the same. Mate used slightly less RAM than Xfce. This was consistent at different stages. I checked right after login, after some minutes of letting the desktop sit idle and after starting some very basic programs like Iceweasel or the same music player. When starting more heavy applications like Libreoffice, the difference soon drowned in the total RAM usage, so I wasn't confident anymore that this was still significant.
Since both Mate and Xfce in Jessie are (mainly) based on Gtk2, I believe that there was no or little cross-toolkit bias.


[1] https://l3net.wordpress.com/2013/05/07/ubuntu-desktop-memory-comparison/
[2] http://dwaves.de/2016/01/04/comparison-of-ram-usage-of-lightweight-linux-desktops/
[3] https://l3net.wordpress.com/2013/03/17/a-memory-comparison-of-light-linux-desktops/

 

[bzar]

That's true. But I remember that I made my own comparisons when I first saw the graph you linked. I made a minimal Debian installation in VBox, cloned it and installed Xfce in one and Mate in the other. The outcome was the same. Mate used slightly less RAM than Xfce. This was consistent at different stages. I checked right after login, after some minutes of letting the desktop sit idle and after starting some very basic programs like Iceweasel or the same music player. When starting more heavy applications like Libreoffice, the difference soon drowned in the total RAM usage, so I wasn't confident anymore that this was still significant.
Since both Mate and Xfce in Jessie are (mainly) based on Gtk2, I believe that there was no or little cross-toolkit bias.

Yes, I was talking more in general, when comparing a large number of desktops big and small, with different toolkits and feature sets.