KDE Memory consumption

[nolim]

Hey everyone,
I keep seeing information circulating online that KDE Plasma 6 is incredibly lightweight these days, often compared to XFCE and frequently praised as being significantly lighter than GNOME.
I recently set up a new laptop (Intel Core Ultra 7, 16GB RAM) and decided to test this out.

Here is what I found:I installed Fedora 43 KDE, and out of the box, the memory footprint was higher than expected. Even after doing some under-the-hood optimization, specifically disabling packagekitd and DiscoveryNotifier (which reclaimed about ~400MB of idle RAM, removal of korganizer that consume ram), my system idles at ~1.7-1.8 GB. That is definitely optimized, but not exactly "XFCE territory".
I did some configuration like moving taskbar on top, but I don't think something that suppose to increase memory by much. Unfortunately haven't tested it on completed vanilla install.
I can post htop output sorted by MRES if needed.

For comparison, I checked my highly optimized Ubuntu GNOME setup. After applying similar tweaks (like disabling gnome-software from autostarting), that machine idles smoothly at just ~1.5-1.6 GB.

One thing I can think of is that the Ubuntu Gnome has dGPU vs laptop vs iGPU. Is there a better way to benchmark this?
EDIT: BTW, on an older laptop Kubuntu 24.04 the RAM is low as 1.1GB!

My initial impression has always been that KDE is inherently a bit heavier, so I'm surprised by the prevailing narrative that it's noticeably lighter than modern GNOME.
Is it distribution configuration Ubuntu VS Fedora?

What are your idle RAM numbers looking like on Plasma 6 vs. GNOME these days?

 

[deb_user]

Don't know, don't care. The differences between a few megabytes (or even hundreds of megabytes) of RAM use are insignificant. If that's actually going to affect you, you need to install more RAM. I haven't used Gnome in many years, because i don't like it, not because of memory use. I run KDE Plasma because I like it, not because of RAM use. I also like XFCE, but not as much as Plasma. But they all run acceptably well on my systems, which are far behind leading edge.

 

[kc1di]

Hello @nolim
Welcome to the Linux.org forum, enjoy!
I'm with @deb_user I use KDE most of the time not because it's more efficient than another DE but because I like it can customize it the way I like and it runs on my systems well. If I need a light weight DE would most likely go with XFCE or LXQT. Gnome is just not my cup of tea for many reasons.
Which ever you use, make sure you enjoy it and it runs well on you box.

 

[nolim]

Don't know, don't care. The differences between a few megabytes (or even hundreds of megabytes) of RAM use are insignificant. If that's actually going to affect you, you need to install more RAM. I haven't used Gnome in many years, because i don't like it, not because of memory use. I run KDE Plasma because I like it, not because of RAM use. I also like XFCE, but not as much as Plasma. But they all run acceptably well on my systems, which are far behind leading edge.

This might be true for my desktop computer with large RAM, but when you have old laptop with 8GB of RAM or even a low tier laptop with 8GB of RAM a 1.1GB of RAM vs >2Gb of RAM is not neglect for vanilla startup.

 

[Mike-BTU]

This might be true for my desktop computer with large RAM, but when you have old laptop with 8GB of RAM or even a low tier laptop with 8GB of RAM ...

I can testify that KDE Plasma runs just fine with "only" 8GB of RAM, with most of the RAM being free.

This is on an older Asus VivoPC i5 and an i3 mini computer, both with 8GB.

 

[kibasnowpaw]

it really depend on what KDE you use and what you have run on the PC my PC idle is 7-8 GB im on Ubuntu server 25.10 with KDE but i have stuff running on it at all time so it never idle under 7 GB ram

 

[Rocketing-warp9]

My setup (Unoptimized) about 5 gigs with Firefox and schoolwork open. This includes 4 panels, 2 top, 2 bottom. I used to have gnome but I don't have the numbers. When I first launch- About 19% of 16 gigs. (about 3gigs). I run Kubuntu 25.10 with a bunch of GUI tweaks. I7 11th Gen.

 

[dos2unix]

KDE 6.6, (Wayland, and Chrome with 2 tabs, puts me at about 4GB. KDE is not light weight by any means.
Which is a main reason I use it. Why have tons of RAM, if you never use it?

 

[nolim]

I recently did some memory footprint testing across four different computers I own, and the results completely surprised me. It really highlights how modern distributions scale their memory usage based on the hardware they are installed on, rather than just the desktop environment itself.

Here is the breakdown of my idle RAM footprints:

1. Desktop (96GB RAM, dGPU) - Ubuntu 24.04 GNOME: Idles at 1.5 - 1.6 GB of system RAM (with another ~200MB offloaded to the dGPU's VRAM). Note: This is after disabling gnome-software from autostarting.
2. New Laptop (16GB RAM, iGPU) - Fedora 43 KDE: Idles at ~1.7 GB of RAM. (After optimizing and removing some background updaters).
3. Old Laptop (8GB RAM, iGPU) - Kubuntu 24.04: Idles at a very light 1.1 GB of RAM.
4. Old Laptop (8GB RAM, iGPU) - Ubuntu 24.04 GNOME: Idles at an incredibly low 0.9 GB of RAM. (Vanilla settings, completely out of the box!)

My Takeaway:It seems clear that distributions have built-in configurations to dynamically activate, disable, or scale background caching depending on the total hardware available.

The fact that the exact same OS (Ubuntu 24.04 GNOME) pulls ~1.6GB on a 96GB workstation (was ~2GB in vanilla settings), but comfortably squeezes itself down to 0.9GB on an older 8GB laptop without any tweaks, makes perfect sense but is still wild to see side-by-side. Might be just linux not releasing memory when the computer has lots of RAM and the "real" footprint on all those is similar.
It definitely changes the "KDE vs. GNOME" RAM debate when the OS base and total hardware change the baseline.

Anyway, I find this interesting

 

[Rocketing-warp9]

That indeed is interesting! Makes me wonder what fluxbox would do...

 

[osprey]

I recently did some memory footprint testing across four different computers I own, and the results completely surprised me. It really highlights how modern distributions scale their memory usage based on the hardware they are installed on, rather than just the desktop environment itself.

Here is the breakdown of my idle RAM footprints:

1. Desktop (96GB RAM, dGPU) - Ubuntu 24.04 GNOME: Idles at 1.5 - 1.6 GB of system RAM (with another ~200MB offloaded to the dGPU's VRAM). Note: This is after disabling gnome-software from autostarting.
2. New Laptop (16GB RAM, iGPU) - Fedora 43 KDE: Idles at ~1.7 GB of RAM. (After optimizing and removing some background updaters).
3. Old Laptop (8GB RAM, iGPU) - Kubuntu 24.04: Idles at a very light 1.1 GB of RAM.
4. Old Laptop (8GB RAM, iGPU) - Ubuntu 24.04 GNOME: Idles at an incredibly low 0.9 GB of RAM. (Vanilla settings, completely out of the box!)

My Takeaway:It seems clear that distributions have built-in configurations to dynamically activate, disable, or scale background caching depending on the total hardware available.

The fact that the exact same OS (Ubuntu 24.04 GNOME) pulls ~1.6GB on a 96GB workstation (was ~2GB in vanilla settings), but comfortably squeezes itself down to 0.9GB on an older 8GB laptop without any tweaks, makes perfect sense but is still wild to see side-by-side. Might be just linux not releasing memory when the computer has lots of RAM and the "real" footprint on all those is similar.
It definitely changes the "KDE vs. GNOME" RAM debate when the OS base and total hardware change the baseline.

Anyway, I find this interesting

There may be some methodological issues involved here. For example, on each tested machine, do the same results appear when the same user applied conditions are set over ten or more repetitions for each machine? In other words, can the reliability of the results be corroborated or established to give one a degree of confidence in them. It may also be useful to measure the total RAM used in smaller units, like kibibytes to detect differences in a more fine grained manner. Memory in linux is complex and the kernel is not always doing the same thing at the same time each time an installation is booted up, and certainly doesn't do the same thing with different hardware because different drivers are involved. This approach for more precision and understanding may not affect the overall results reported so far, but would add a perspective of interest I guess.

In general, if the goal is to use as little memory as possible, the logical approach would be to use applications that use the least memory to achieve any particular desired functionality. In the case of Desktop Environments, there are lesser memory sapping applications to be found in the window managers. Lesser again in terms of memory usage is working in the framebuffer on the console. There are other means of memory control too. If a user compiles their own applications before installation, as is the case in Gentoo, and also in the Linux From Scratch approach, then applications may be able to be configured to be more memory efficient by not enabling functionalities that may be deemed unnecessary but which are usually included by default in the precompiled binaries in major distros.

 

[deb_user]

The Linux kernel is pretty good at managing memory, and it has steadily improved over time. Older kernels may not be as efficient as newer ones, especially on newer hardware. Different desktop environments, as well as different distros, will usually have different programs autoloaded into memory, and use memory somewhat differently. It may be possible to make generalized predictions about which distro/DE combination will use the least or the most memory, but it's a moving target. I think that two users can each install and configure a computer with similar specs and have different amounts of RAM in use at any random time. This just isn't something I can convince myself to spend time worrying about.

 

[rado84]

Hey everyone,
I keep seeing information circulating online that KDE Plasma 6 is incredibly lightweight these days,

Keep tellin' yourself that! Nothing could be further from the truth! As you can see in the screenshot below - only one process runnin' and the difference between used VRAM is huge. When I open all of the 6 programs, in Cinnamon the used VRAM jumps to 600-660 MiB. In Plasma 6 it jumps to 2000+ MiB, often even a lot more than that. Good luck with your Plasma! You're gonna need a video card with 12 GiB VRAM as a bare minimum to run this hungry monster named Plasma...

 

[johnblackwood1]

dont trust anyone who says KDE is light. its not. one time i installed it and it bricked the network manager and drivers. had to reinstall the distro, if you want the LIGHTEST de then choose lxde. lightest one. if you want just pure linux delete all your desktop enviorments and type in startx. that should start X11 window manager.

 

[Mike-BTU]

one time i installed it and it bricked the network manager and drivers.

When was that?

KDE was buggy in the past.

 

[johnblackwood1]

When was that?

today. fresh installation still exept i installed tux math and tux typing

 

[johnblackwood1]

When was that?

KDE was buggy in the past.

i think its because KDE likes to include stuff like their own drivers and network managers and it failed to install one for some reason and tried to delete the one i use

 

[bob466]

If this was the old days of 32bit I'd be concerned but it isn't.

We all know modern computers use more Ram...I'm using 2.6GB of Ram right now but have 16GB of Ram installed so who cares.