Hi,

I've tried Ubuntu 10.10 (GNOME) and 11.04 with Unity. I've also tried Fedora 14 (GNOME) and Fedora 15 with GNOME 3. My findings: RAM usage is excessive in all cases. I have 1GB and upon startup about 400MB are used. A few tasks down the line I'm using 600 to 800MB; responsiveness starts to deteriorate and SWAP starts being used.

On a laptop with 512MB Debian (GNOME) uses less than 200MB with Firefox running and with an uptime of 27 hours (many, many tasks performed/software used) compared to 30 minutes with one of the others.

Is Debian just better "tuned"?

Different distros install a different amount of packages and services with the default, unconfigured installation. If you just click through the 'next' buttons during installation with fedora, you're going to end up with a lot of junk installed that you probably don't need. I have a server running fedora with a database system on it and it's only consuming 500mb of ram or so when not being queried.

When you are going through the installer, do the minimal installation. You can further customize what is installed by clicking 'customize now' during the repo configuration. Lastly, you can configure what services run at boot-time with the chkconfig utility once the system has been installed and booted up.

Basically, to answer your question, I would say no: debian does not have better memory management than fedora.

Linux uses less RAM on your laptop because less RAM is available. You need to repeat the experiment on the identical hardware to have a meaningful comparison. The more RAM you have, the more Linux will use. Here is an excellent resource on the topic: http://www.linuxatemyram.com/

That being said, there are many users (myself included) who find that Debian does, indeed, offer excellent performance. If you take Debian for a test drive on your Ubuntu/Fedora machine and get good results, I will not be surprised.

In case you are misunderstanding ram use statistics, please read
http://www.linuxatemyram.com/

None of that is necessarily wrong or bad. Even some use of swap may be normal and best (as compared to using less cache).

That obviously is bad. But how did you measure that? Is it real or is it psychosomatic?

You didn't say which version of Debian. If there is a difference, it is unlikely to be directly related to the distribution itself or to any tuning done by the distribution.

There might be some significant difference in the version of Gnome or the version of X causing some significant difference in ram use. But more likely, the difference is some service that is turned on in the systems with greater ram use and turned off in your laptop.

Possibly the Debian you installed has the services that differ off by default vs. the other distributions have such service on. Or maybe the installer is tweaking some default choices because of the difference between a 1GB desktop and a half GB laptop.

Possibly the service that differs is really making demands on ram that do measurably reduce performance. But possibly, it is just doing a lot of file I/O and the lack of real demand for ram is the only reason that file I/O increases cache usage.

I don't actually know much about Linux services, so I can't make a good guess at what optional services might be running. A service similar to Windows indexing service would really fit your symptoms. On a quick google search, I couldn't even figure out if Linux has such a service, much less which distributions might turn it on by default.

So you should find out what services are running rather than wait for someone to guess which one might be causing your symptoms.

I wont say anything about Fedora, I dont have enough experience with it.

Debian, well, you could say 'better tuned' if you want, though that is debatable. Last I used debian in comparison to ubuntu, there was less packages installed. Ubuntu really doesnt give a damn about RAM usage, or sensible package dependencies. Check out the dependencies for 'mountall' if you dont believe me-

http://packages.ubuntu.com/maverick/mountall

What the hell is with making mountall dependent on plymouth? That pretty much makes plymouth unremovable. Its a bootsplash.....Last I heard though the ubuntu people were redefining plymouth, but I forget the exact term they were using.

In my experience Debian is frugal with RAM compared to Ubuntu. I don't know about Fedora.

On a fresh install my laptop with 512MB of RAM will use 150MB idling after boot with Ubuntu and 110MB with Debian. Clean both up removing packages that I don't need or that I know eat resources like there is no tomorrow and I can get Ubuntu down to 130MB and Debian under 100MB RAM usage just idling at start up and not using any SWAP. During normal usage Debian consistently uses 20-30MB RAM less than Ubuntu with SWAP being about the same usage.

Thanks for all the comments. I'll try and answer your questions; please ask again if I've missed anything.

I am reading free -m as follows:
-/+ buffers/cache: 445 (used) 543 (free)
That is on Ubuntu 10.10 with a clean boot and just Firefox open.

By the performance deteriorating I mean it starts to behave slowly (I'm physically noticing it). It's not psychological; it's noticeable enough that I'm almost pulling my hair out.

In terms of better battery life, is it better to have less RAM so less is used? Am I right in thinking that the more RAM used the more battery will be consumed (in and outside of suspend)? I recall resuming Ubuntu from suspend after a full charge just 12 hours later to find about 20% of my battery gone. Compared to Arch (LXDE) that usually just drops about 4% in that time (in suspend of course) on the same netbook.

Is this at start up or the whole time?

If it is at start up ureadahead may be something to look at. From personal experience, and it was your "almost pulling my hair out" comment that struck a cord with me, ureadahead could have a negative impact on some machines. So much so that after a couple of days of research, and also waiting for over 2 hours for my laptop to actually be usable on a couple of occasions, that I found ureadahead was the issue with my, and a few other, machines. Now I wouldn't ordinarily suggest removal of standard system parts but ureadahead is another matter altogether and when I do Ubuntu installs on anything less than 2MB RAM I remove ureadahead totally. Debian doesn't have it so I see no need for it.

It is after some time that is starts to grind to a halt... when my computer uses 60% to 90% of its memory. I found Nautilus and Firefox to be the major RAM hogs at these times, but then there are many things that use between 10 and 40MB that all add up.

If I just stick to one task at a time I'm usually OK, but my concern about RAM usage over battery performance would be nice to clear up; if someone can confirm this.

I'd use a better distro without gnome if I was worried about memory. Slitaz is pretty low resourced if all you need is web.

I find it kind of odd but you may have a driver issue or some other app that is causing some memory leak.

Is is better at ram. Well, not really but what version of debian are you talking about?

Debian + GNOME is much lighter than Ubuntu in my experience. Vanilla Debian starts with 192 MBs of RAM, Ubuntu uses double.

Info: Processes 120 Uptime 4:50 Memory 184.7/248.3MB

IBM Thinkpad R40 LXDE with guake+shell-fm, wicd, okular, firefox, pcmanfm all running

System: Host debian Kernel 2.6.32-5-686 i686 (32 bit) Distro Debian GNU/Linux wheezy/sid
CPU: Single core Intel Pentium M (-UP-) cache 1024 KB flags (sse sse2) bmips 1196.21 clocked at 600.00 MHz
Graphics: Card ATI Radeon Mobility M7 LW [Radeon Mobility 7500] X.Org 1.9.4.901 Res: 1024x768@60.0hz

I'd agree with jefro: something odd is happening.

I'm in Opera, so I've started Firefox, OpenOffice, Nautilus, and the system monitor. Free shows 285448 used, 78628 buffers, 431300 cached: total 795376. You shouldn't be needing swap with 1GB.