KDE vs XFCE speed

English Install/Boot/Login
1

I’m on Leap 15.4 and soon I will upgrade to the soon to be released 15.5. I currently use xfce.

With all the plasma/kde improvements, is now kde as fast as xfce or at least comparable in speed?

I do a lot of compiling and in 15.1 (the last time I used kde), kde used to take slightly over 16 seconds for the compile then I switched to xfce and the same compile went to just under 8.5 seconds.

Big difference!

Anybody with experience with both?

Thanks

2

16 vs. 8.5 may be big in percentage, but what’s 7.5 seconds really if you like the other features better? 16 vs. 8.5 minutes I could better understand.

There are multiple online tests comparing XFCE to Plasma. Going back as much as 3 years using Google, there’s no clear winner. Read them for yourself. Plasma can be faster, but it may or may not happen, based on what your usage habits are, and how much tweaking you’re willing to try. Several years ago KDE put a lot of effort into breaking its heavyweight reputation by speeding it up, but reputations are hard to overcome.

You can install both at once in Leap. Compare for yourself and let us know. Until you are sure which you prefer, create an extra user for one of the two, to ensure you encounter no personal settings entanglements. You shouldn’t need to, but one never knows.

3

On host 3400G KDE running in the session slice eats a tiny fraction of total cpu time:

3400G:~ # systemd-cgtop user.slice --cpu=time --depth 3 | cat
user.slice                                                       1458 5h 22min 45.156968s    12.7G        -        -
user.slice/user-1000.slice                                       1030  5h 20min 1.551372s    11.1G        -        -
user.slice/user-1000.slice/user@1000.service                     1022  5h 20min 257.262ms    11.1G        -        -
user.slice/user-1000.slice/user@1000.service/app.slice            850  5h 9min 15.693468s     8.7G        -        -
user.slice/user-1000.slice/user@1000.service/session.slice        132    10min 20.156010s   751.6M        -        -
3400G:~ #

KDE uses 10min / 5h 20min = 3% of total user cpu time.

System slice:

3400G:~ # systemd-cgtop system.slice --cpu=time --depth 3 | cat
system.slice                                            67 12min 28.863461s     4.0G        -        -
system.slice/display-manager.service                    17  9min 28.245883s        -        -        -
system.slice/fetchmail.service                           1       18.558822s        -        -        -
system.slice/systemd-journald.service                    1       10.883859s        -        -        -
system.slice/udisks2.service                             6        3.365551s        -        -        -
system.slice/dbus.service                                1        3.337230s        -        -        -
system.slice/systemd-udevd.service                       1        3.207682s        -        -        -
system.slice/irqbalance.service                          2        1.710360s        -        -        -
system.slice/NetworkManager.service                      4        1.476031s        -        -        -
system.slice/apache2.service                             6        1.119170s        -        -        -
system.slice/polkit.service                              4        970.265ms        -        -        -
system.slice/postfix.service                             3        700.399ms        -        -        -
system.slice/minidlna.service                            2        638.447ms        -        -        -
system.slice/rtkit-daemon.service                        3        495.307ms        -        -        -
system.slice/systemd-logind.service                      1        433.809ms        -        -        -
system.slice/chronyd.service                             1        209.850ms        -        -        -
system.slice/cups.service                                1        186.728ms        -        -        -
system.slice/avahi-daemon.service                        1        172.023ms        -        -        -
system.slice/hd-idle.service                             1        168.822ms        -        -        -
system.slice/ModemManager.service                        4         85.080ms        -        -        -
system.slice/upower.service                              4         50.051ms        -        -        -
system.slice/sshd.service                                1         31.457ms        -        -        -
system.slice/mcelog.service                              1          9.825ms        -        -        -
system.slice/system-getty.slice                          1          4.262ms        -        -        -
system.slice/system-getty.slice/getty@tty1.service       1          4.262ms        -        -        -
3400G:~ #

Presumably you are misusing your resources. You may want to check your system in more detail.

1 Like
4

Yes, 16 s versus 8.5 s is a big difference.

What kind of compile did you do? I presume you did run it on the console and if so I do not see how a DE can make suc a big difference.

1 Like
5

??? – compiling usually involves invoking a Make/CMake file from the command line –

  • except, if, you’re using a Development Environment – such as:
    KDevelop
    Eclipse

And, the compiler doesn’t run on the Desktop Environment as such – it runs as a Command Line.

Which can only mean that, for whatever reason, your Desktop Environment as such, isn’t passing enough compute resources to the (not a GUI) compiler executable.

Normal compiler issues:

  • User Quotas – disk and memory.
  • User disk cache issues.
  • User temporary directory free space and location – ‘tmp’ on disk or in memory?
6

To me it’s a big difference. On some days, I’ll do more than 500 compiles testing + debugging… It’s big difference.

7

cmake + python

ctrl-alt-t to open the terminal and run my compiles. I could do the compile from other environment, like atom…, but i finally settled on gvim and the terminal to do my compiles.

Thanks

8

Okay, so compile for both is using the (graphical) console.

It would be interesting to see the same compile on both DE’s with “/usr/bin/time -v” in front of compile command that you are executing and compare the output that gives.

9

Are you converting Python scripts to a binary executable with either “py2exe” or, “PyInstaller” or, something else?

10

Just my 2 cents. If you want fast then use icewm or enlightenment. You will see a big difference.
Back to xfce and kde, I noticed if i do a simple human rig in blender and render, xfce is faster by one minute
like 3 minutes for kde and 2 minutes for xfce.

11

Desktop environment is not usually required to build software as build scripts don’t interface directly with the DE. It’s mostly I/O and terminal output. That said:

  1. If there’s lots out terminal output, terminal performance will impact compile times as it blocks the process emitting those logs. You can tweak terminal settings and/or build script to minimize logs.
  2. Plasma does by default indexing of files on home directory. Make sure build target directory is excluded from indexing.
  3. Are you building from the exact same sources? Project might have been optimized since the last time you tried on Plasma.
12

KDE in TW is a pain the last months, constantly loosing functionality (desktop icons, automatic taskbar show-up of plunged in USB-devices etc etc etc). In Leap KDE is not that much better any more imho. It simply has become a pain and peak-KDE has been in my experience with opensuse 13.2.

I think I will switch completely to xfce in the future…

13

Can people please stick to the subject: “KDE vs XCFE speed” with a focus on compiler speed?

All frustrations about the one or other DE experience now or in the past are off topic.

1 Like