Linux Gaming Benchmarks With Plasma-Next, KDE Frameworks 5

[phoronix]

Phoronix: Linux Gaming Benchmarks With Plasma-Next, KDE Frameworks 5

Up for your viewing pleasure today were some quick benchmarks done of the next-generation KDE desktop stack compared to the KDE 4.13.0 and Unity 7.2.1 desktops of Ubuntu 14.04 LTS.

http://www.phoronix.com/vr.php?view=20616

[mgraesslin]

Intel Core i7 3517U CPU having HD Graphics 4000 from the Ivy Bridge generation

I think I pointed out for another benchmark that suspend fullscreen is ignored for Intel hardware... Back then you complained that the GUI doesn't state so, luckily it does so in KWin 5 now.

[Kayden]

I think I pointed out for another benchmark that suspend fullscreen is ignored for Intel hardware... Back then you complained that the GUI doesn't state so, luckily it does so in KWin 5 now.

Hey Martin,

Is there a particular reason for that, still? I see this in the KWin code:

    if (GLPlatform::instance()->driver() == Driver_Intel)
        options->setUnredirectFullscreen(false); // bug #252817

but, bug #252817 is a 2010 report originally for i945 hardware. That's not even the same 3D driver as most modern Intel hardware, so I wonder if it's still necessary? If there's still a crash, or broken GLX behavior, we'd definitely like to fix it...

Thanks!

[widardd]

What about an up-to-date comparison of DE/WM regarding memory consumption? I don't really care about the non-existent gaming performance of my old Eeepc 901.

P.S.:Currently sticking to LXDE, but it does feel pretty dated.

[mgraesslin]

but, bug #252817 is a 2010 report originally for i945 hardware. That's not even the same 3D driver as most modern Intel hardware, so I wonder if it's still necessary? If there's still a crash, or broken GLX behavior, we'd definitely like to fix it...

I don't know whether it's still valid. It was valid in 2010 (most often reported KWin bug ever, we spent about a person day just to mark duplicates), than it was fixed, then it came back in 2013. Better safe than sorry and it's not a really useful option anyway. Granted if enabled we will score better in Phoronix benchmarks, but that's not a development aim for us. Unredirect fullscreen doesn't make any sense except for games and for that we have window specific rules and also an X property to disable the compositor all together which would give way better results anyway. As it's by that an option for experienced users we don't see a lot of need for it or to investigate whether the workaround is still needed - better safe than sorry.

[edoantonioco]

So, seems to be ubuntu is the best way to play in linux.

[TheBlackCat]

So, seems to be ubuntu is the best way to play in linux.

Or that Ubuntu's DE works best on Ubuntu.

[dungeon]

So, seems to be ubuntu is the best way to play in linux.

Benchmarks are relative thing, you can not conclude for sure what is smoother or faster out of it, but you can repeat the test for you and see .

Btw i use openbox, but always test xinit when i want to check driver improvments, there are always more or less overhead when testing using DE .

[Kemosabe]

Or that Ubuntu's DE works best on Ubuntu.

Excatly. This is anything but a representative test of desktop environments. On phoronix you see only Ubuntu benchmarks with invalid conclusions regarding to general performance of specific software suits.
People stated again and again they don't want to see so much Ubuntu only stuff but michael doesn't care.

[Vash63]

I already knew that KDE was slow, kinda disappointed that this isn't faster than KDE4 but they still have a few months before the first release to fix it up. Why isn't GNOME in this comparison? Mutter has always felt like the fastest cross-distro composited environment to me. Unity doesn't port well to Arch or Fedora or any distros I run so I haven't used it in a long time.