The Linuxed – Exploring Linux Distros website has over 170 reviews of distros and common Linux programs by Arindam Sen. What makes it special is that Mr. Sen puts his experiences together in useful charts, such as this one that compares RAM usage for twenty different GUI versions, or this one that compares CPU and RAM usage for 20 KDE distros (page down to see the chart). While the site says it’s merely “A non-techie’s view of the Linux world,” many techies will find it useful as well.
I wonder what use these tables are.
CPU and RAM usage will change depending on what you do and what you install. Also after installing updates things may or may not change significantly.
True, however, they do help you to get a feel for the general ballpark and direction given desktops are going, resource-wise.
Exactly. It’s not all useless. I wonder where window managers like IceWM and JWM would place on that chart. He’s got the classic mid-end (GNOME 2, MATE…) and modern high-end (GNOME 3, KDE4…) desktop environments pretty well covered though. KDE3 is also missing… understandably, but it would still be interesting to see where it would be. openSUSE still provides it in their repositories. And then there’s Trinity, which I wish would gain more traction as an alternative to the typical KDE4 and million GNOME-based desktops in the increasingly-fragmenting world of GNOME.
Edited 2013-06-25 06:01 UTC
I just wish they’d figure out why I’ve never had a system where it can be installed alongside KDE 4 components without rendering the Trinity control panel empty.
(I run an LXDE desktop with K3b and FileLight from KDE 4 for the updates but BasKet for KDE 4 is too buggy to use and, last I checked, KAudioCreator hadn’t been ported.)
No, they don’t. One might be skewed to cache more of its components into memory ( file browser, common apps, etc), while the other does not. So starting up all the apps in lxde that one might use, might use more ram than KDE for the equivalent apps( as well as being slower as it would have to load them from disk).
With the table, all you can say is that lxde is more efficient at … doing nothing. Measuring efficiency, bloat, and ‘lightweightness’ is too subjective to objectively determine for all use cases. It depends so heavily on what is important to you.
It would also be nice to know exactly how “RAM usage” is measured. Maybe it’s explained somewhere on the site but I didn’t look that much (shouldn’t have to).
Indeed. Not to rain on his parade, but his numbers are pretty meaningless. Measuring memory usage is a very complicated matter, especially on systems that use shared memory.
Here is an interesting article on that:
http://virtualthreads.blogspot.be/2006/02/understanding-memory-usag…
As a rule of thumb, I use resident – shared, but even that is misleading as some programs (such as Firefox) may allocate more memory when it’s available. It’s such a waste to have 80% of your RAM unused while it could be used for caching and speeding up algorithms.
These lightweight desktops may use less memory, but as soon as you open some applications, they will load their own libraries rather than using shared ones. The memory footprint may rise quickly.
Edited 2013-06-25 10:30 UTC
The memory usage of the desktop environments is interesting as you can measure them on their own. But I agree the distro-wide figures have to be taken with a huge pinch of salt.
Still, not really, some DE (KDE and Gnome), use memory on purpose. KDE, for example, have many shared caches (Like the icon theme) and shared process to prevent loading multiple copies of common thing. That being said, it leave many of these items in memory even if nobody use them. This is done on purpose and there is nothing wrong with this. All OS do that, including bare-bone Linux kernel. Caching when you have enough memory is a good thing, not an evil.
I’m surprised to see the new builds of KDE4 use less RAM than the newer builds of GNOME3 and Unity. Given the amount of “bloat” in KDE4, I’m actually rather impressed.
What definition of “techie” can you possibly be using if posting tables of memory usage of Linux desktop environments doesn’t qualify you?
As for the “these tables are useless” argument, I’m going to go with “it’s better to be done than perfect.” Sure, they aren’t the whole story, just like you can’t say that Manute Bol was a better basketball player because he was 7’7″ to Michael Jordan’s 6’6″. That doesn’t mean it’s useless to know how tall they are.
Only if they’re not misleading. And given the complexities of measuring memory use, it’s almost impossible for them not to be misleading…
I’m going to go with “if you’re going to do it wrong and with misleading results it’s better not to do it”.
Which is exactly why it’s important to know the method used to measure things. Without knowing the method the numbers ARE meaningless.
I questioned whether or not all this “analysis” serves a real purpose or does actual good to the reader. For example, I can’t honestly say I’ve known anyone who factors in how much ram usage there is. Just the sheer number of distros is enough to turn people away, much less throwing all these mostly useless comparisons at them.
If a user is looking for a linux distro to try and they don’t know exactly what they want, that person should just go to linux.org and pick one off their ‘most popular distros’ list.