Announcement

Collapse
No announcement yet.

Intel Ivy Bridge: Unity/Compiz Continues To Impair The Linux Desktop

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Intel Ivy Bridge: Unity/Compiz Continues To Impair The Linux Desktop

    Phoronix: Intel Ivy Bridge: Unity/Compiz Continues To Impair The Linux Desktop

    As the latest Intel Ivy Bridge Linux graphics benchmarks to publish, here is a comparison of some of the different desktop environment options of Ubuntu 12.04 LTS when using the Intel Linux graphics driver on the Core i7 3770K with its HD 4000 graphics. The desktop environments being compared include Unity, Unity 2D, KDE, and Xfce. The default Ubuntu Unity desktop with Compiz continues to have problems for the open-source friendly company's drivers.

    http://www.phoronix.com/vr.php?view=17386

  • #2
    /save link for future reference.

    ... Of-course, the lousy performance exhibit by Unity, will not prevent it from being used in the next Linux vs. *BSD / Windows / OSX.
    I can only wonder why... *, **

    - Gilboa
    * And no, being number #1 in Google trends is enough to make Ubuntu the de-facto Linux standard.
    ** ... Cause Google trends is just as reliable / unreliable as Distrowatch (that lists Ubuntu as #2)
    DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB + 2x3TB, GTX780, F21/x86_64, Dell U2711.
    SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F21/x86_64, Dell U2412..
    BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F21/x86-64.
    LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F21/x86_64.

    Comment


    • #3
      Originally posted by Michael
      Compiz has turned out to be a shivery wreck that is unlikely to see dramatic improvements in the near future.
      Wrong. They found a major 3D performance regression within Unity affecting most applications/games. The first fix will be in "unity 5.12-0ubuntu1.1", currently in precise-proposed, but there are some more changes needed for the full fix.

      See: https://bugs.launchpad.net/ubuntu/+s...ty/+bug/987304 & https://bugs.launchpad.net/ubuntu/+s...y/+bug/1005074

      Comment


      • #4
        Title

        I wonder why the article is titled "Intel Ivy Bridge: ...." whereas it doesn't have to do anything with Ivy Bridge at all.

        Comment


        • #5
          I know it isn't finalized yet, but could we see these benchmarks run against the Pantheon Desktop Environment from Elementary OS Luna? A little Googling reveals which PPAs contain it, or you could install a daily build of Luna.

          Comment


          • #6
            Originally posted by Linuxhippy View Post
            I wonder why the article is titled "Intel Ivy Bridge: ...." whereas it doesn't have to do anything with Ivy Bridge at all.
            The entire thing is about these benchmarks which are running on an Ivy Bridge CPU and graphics.

            Comment


            • #7
              Loosing credibility are we here?

              This article seemed intent on promoting KDE even though by my eyes XCFE has done a far better job than KDE overall. So has Phoronix fallen off the unbiased wagon to take a ride on the KDE buggy?

              Comment


              • #8
                Originally posted by wizard69 View Post
                This article seemed intent on promoting KDE even though by my eyes XCFE has done a far better job than KDE overall. So has Phoronix fallen off the unbiased wagon to take a ride on the KDE buggy?
                1. It's "losing" and not "loosing"
                2. WTF?

                Please cite an example of what you're talking about, because I read the article again and didn't see anything like KDE promotion. Here are the mentions of xfce/KDE:

                The desktops leading to the fastest results were KDE with suspended effects and then Xfce 4.8 with its default window manager.... Xfce and KDE (with suspended effects) were 21% faster than Unity 5.10..... Xfce continued to run in line with the suspended KDE effects position.

                Comment


                • #9
                  Wasn't there another benchmark article that basically showed Ubuntu and Windows at comparable performance on the nvidia driver?

                  Yet the conclusion of the article is that compiz is a turd.

                  Comment


                  • #10
                    Running only 120 instead of 140 fps in ioquake3 games is an "impairment to the linux desktop"?? *facepalm*

                    This article sounds more like Michael had a bavaria-related impairment going on...

                    Comment


                    • #11
                      I agree Compiz is a turd, but I think there is other problem that is causing it to become a turd. I have had to resort back to using Openbox with Ubuntu 12.04, Unity and Gnome3 are both unstable for me. I'm running AMD Phenom II N620 Duel Core + AMD 4200 on-board graphics (laptop). With Unity and Gnome3, I have Xorg and Compiz processes taking CPU to full load. I have tried both the open driver and AMD drivers from the repository and installed from the AMD web site. With the same results. I did not see this in Ubuntu 11.10 and even in the 12.04 beta. I have looked around the web and have ran across some bug reports with users experiencing the same problem all with different hardware. Some of the point to Compiz other point the finger at the PAE kernel. I have tried most of the easy suggestions to fix this, but none of theses work. I'm hoping this will be addressed in further updates.

                      TLDR: Compiz,Unity and or Ubuntu's Kernel is broken And it needs to be fixed.

                      Comment


                      • #12
                        Originally posted by not.sure View Post
                        Running only 120 instead of 140 fps in ioquake3 games is an "impairment to the linux desktop"?? *facepalm*

                        This article sounds more like Michael had a bavaria-related impairment going on...
                        Not really: 10-15% makes a huge difference once the FPS drops (E.g. 25fps vs 30+fps) *

                        - Gilboa
                        * Generally speaking.
                        DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB + 2x3TB, GTX780, F21/x86_64, Dell U2711.
                        SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F21/x86_64, Dell U2412..
                        BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F21/x86-64.
                        LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F21/x86_64.

                        Comment


                        • #13
                          Originally posted by johnc View Post
                          Wasn't there another benchmark article that basically showed Ubuntu and Windows at comparable performance on the nvidia driver?

                          Yet the conclusion of the article is that compiz is a turd.
                          There was, and you are correct.

                          Ivy bridge doesn't have the best support in linux, and provides a fairly crappy experience (compared to Nvidia/AMD). to quote Michael regarding win7 vs. linux on Ivy bridge;

                          Originally posted by Michael
                          As this Windows 7 vs. Ubuntu 12.04 graphics comparison for Intel Sandy/Ivy Bridge hardware was getting underway, I certainly did not expect to see the Intel Linux graphics driver be slaughtered so badly... /... the Windows 7 driver led to frame-rates that were commonly 50~80% higher than what was provided by the open-source Intel Linux driver. /... Beyond the Intel Windows driver being faster, it has better OpenGL support too.
                          regarding win7 vs. linux using Nvidia gtx 680;

                          Originally posted by Michael
                          Overall, these results are not too surprising. For the most part the NVIDIA GeForce GTX 680 performance between the two operating systems was more or less the same, which is to be expected when the driver code-base is largely shared on NVIDIA's end between Windows and Linux (and BSD and Solaris).
                          So in reality, this article should be called "Intel Ivy Bridge Continues To Be Garbage on The Ubuntu Desktop" ...

                          To blame compiz at the end of it all, when you aren't even benchmarking it - seems a bit silly.

                          As d2kx pointed out;

                          Originally posted by d2kx View Post
                          Wrong. They found a major 3D performance regression within Unity affecting most applications/games. The first fix will be in "unity 5.12-0ubuntu1.1", currently in precise-proposed, but there are some more changes needed for the full fix.

                          See: https://bugs.launchpad.net/ubuntu/+s...ty/+bug/987304 & https://bugs.launchpad.net/ubuntu/+s...y/+bug/1005074
                          I was told abut this a while ago, by one of the main compiz-devs. We were discussing this very issue. He stated that Unity needed some changes because it had some perofrmance issues, which compiz didn't have. He even said something along the lines of but i'm sure it doesn't matter to you (me), since you aren't using Ubuntu. ~ to which i added that Unity runs on Archlinux, so i know some people who it does matter to
                          Last edited by ninez; 05-28-2012, 03:41 PM.

                          Comment


                          • #14
                            Ohm

                            Originally posted by coder543 View Post
                            The entire thing is about these benchmarks which are running on an Ivy Bridge CPU and graphics.
                            The article was about LLVMPipe, and nothing else. Ivy Bridge was in the title simply to draw attention.

                            Comment


                            • #15
                              Originally posted by Linuxhippy View Post
                              The article was about LLVMPipe, and nothing else. Ivy Bridge was in the title simply to draw attention.
                              Are you sure you're not confusing this article (about Ivy Bridge performance) with the other one posted today, which was indeed about LLVMPipe?

                              Comment

                              Working...
                              X