Announcement

Collapse
No announcement yet.

KDE & Xfce Don't Lead To Performance Wins Over Windows 8

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

  • #46
    SS Setting

    Originally posted by Pawlerson View Post
    Where I need to to put this?
    You put it in the console (use the ~ key to drop it down).. Enter each line one at a time. It made a huge difference for me.

    Comment


    • #47
      If Dell would tweak a notebook for absolute performace under Ubuntu, it would take:
      1. Intel writing a superb closed source driver with all the propriatary algorithms
      2. Realtek writing the best audio driver for Linux in gernel
      3. Atheros delievering essential an up-to-date Wifi driver
      4. Dell tweaking the kernel and quality testing the entire thing for 6 months.

      Because, that is what happens when companies get advanced copies of Windows 8 to prep for their next product.

      Microsoft is all but leasing Windows to their customers. I wouldn't be surprised if they didn't start selling their own brand laptop for rent.

      Linux and distributions in general, are too busy preping for the next product release other than cleaning up problems that have existed for years.

      Excuses:
      1. Upstream problem
      2. Won't fix see #1
      3. Angst against closed source drivers
      4. Opinions
      5. etc.....

      Comment


      • #48
        I'm surprised too to see kde to perform worse than other DEs.
        Now, given the fact that kwin has desktop effects disabled for fullscreen apps,
        two things come to mind:

        a- could it be that compositing active still has an impact?
        b- Or maybe some background process is stealing cpu cycles?

        So i would suggest to:
        1- manually disable effects in the whole kde (Alt-shift-F12) prior to start the benchmark
        2- having a report about background processes cpu use during the bench.

        Point 2 could be implemented in PTS by starting a monitoring "top -b" or something similar...
        I think something like that would be a nice addition to PTS what do you think, michael?

        Comment


        • #49
          Thanks for the article. I hope in the future it will be possible to benchmark more “mainstream” games, games from the Humble Bundles maybe (yes, some of them are slow), Valve/steam games, or popular Windows games under Wine.

          Comment


          • #50
            Originally posted by kokoko3k View Post
            I'm surprised too to see kde to perform worse than other DEs.
            Now, given the fact that kwin has desktop effects disabled for fullscreen apps,
            two things come to mind:

            a- could it be that compositing active still has an impact?
            b- Or maybe some background process is stealing cpu cycles?
            Well, check this out http://jaysonrowe.blogspot.com/2013/...kde-users.html . Basically it's a bad default setting in KDE. KDE should change the default. Micheal's tests shows that when the default KDE is used, it under-performs compared to the other desktops with their defaults.

            Comment


            • #51
              Originally posted by Adarion View Post
              WTF?

              1. The test conditions / setup box is as always unreadably small and must be scrolled.
              2. From what I am reading there it is using an intel chip.
              3. What the hell is this "test"? It is comparing the driver implementation for intel chips on Windows and Linux. Doesn't really have that much to do with the desktop environment.
              Please try to read and comprehend the article.

              Comment


              • #52
                Originally posted by Calinou View Post
                Don't forget the article is talking about Intel graphics, which are known for being slower on Linux than on Windows. Do the same with a NVIDIA graphics card and the performance will pretty much be the same. 8)
                Indeed -

                http://www.phoronix.com/scan.php?pag..._nvidia1&num=5

                Comment


                • #53
                  Originally posted by stqn View Post
                  Thanks for the article. I hope in the future it will be possible to benchmark more “mainstream” games, games from the Humble Bundles maybe (yes, some of them are slow), Valve/steam games, or popular Windows games under Wine.
                  Name recognition is not the problem here - it is about verifiability. The games here are known to be solid and use the same back-ends (for the most part) on both Linux and Windows, and therefore are better for benchmarking the driver. Benchmarking a game under Wine would be a huge joke as the fact that it is not native would put overhead on Linux simply because it is not being ran natively. This is not about showing what games you can run on your Intel cards, but rather the state of the drivers.

                  Comment


                  • #54
                    Originally posted by Hamish Wilson View Post
                    Name recognition is not the problem here - it is about verifiability. The games here are known to be solid and use the same back-ends (for the most part) on both Linux and Windows, and therefore are better for benchmarking the driver. Benchmarking a game under Wine would be a huge joke as the fact that it is not native would put overhead on Linux simply because it is not being ran natively. This is not about showing what games you can run on your Intel cards, but rather the state of the drivers.
                    You state your opinions as if they were the ultimate truth, but they aren’t.

                    Comment


                    • #55
                      Originally posted by stqn View Post
                      You state your opinions as if they were the ultimate truth, but they aren’t.
                      Care to actually respond to my "opinions" then?

                      Comment


                      • #56
                        It doesnt have enough benchmarks to really show the state of the drivers. It doesnt show how it performs with wine. It doesnt show how it performs with Unigine engine. It doesnt show how it performs with any steam game.

                        All it shows is on the benchmarks run the linux Intel driver is sometimes slower than the windows Intel driver. But that isnt what the article title says. That is not what the conclusion says.

                        Comment


                        • #57
                          Originally posted by ua=42 View Post
                          Well, check this out http://jaysonrowe.blogspot.com/2013/...kde-users.html . Basically it's a bad default setting in KDE. KDE should change the default. Micheal's tests shows that when the default KDE is used, it under-performs compared to the other desktops with their defaults.
                          These tests were with the suspended desktop effects for full-screen windows.
                          Michael Larabel
                          http://www.michaellarabel.com/

                          Comment


                          • #58
                            Originally posted by stqn View Post
                            Thanks for the article. I hope in the future it will be possible to benchmark more “mainstream” games, games from the Humble Bundles maybe (yes, some of them are slow), Valve/steam games, or popular Windows games under Wine.
                            New PTS/OpenBenchmarking.org test profiles are happily accepted.
                            Michael Larabel
                            http://www.michaellarabel.com/

                            Comment


                            • #59
                              Originally posted by Michael View Post
                              These tests were with the suspended desktop effects for full-screen windows.
                              Hi Michael, if you'll see the top of my blog post, Martin Gräßlin who is the Kwin and Compositor maintainer for KDE told me that simply suspending desktop effects for full-screen windows doesn't have that much effect. You should try either doing alt+shift+f12 to disable compositing, or Alt+F3 -> More Actions -> Special Window Settigns -> Appearance & Fixes and enable block compositing/force/yes which will basically create a "rule" that whenever that game is started, compositing will be suspended completely automatically (basically the same thing as alt+shift_f12, but you don't have to remember to do it each time).

                              Once I did this, I really noticed a difference.

                              Comment


                              • #60
                                Here I tested 3 cards with all Killing Floor renderes on Highest. Pretty interesting I would say...

                                http://steamcommunity.com/app/221410...4089658431267/

                                Of course I used KDE 4.8 with Kanotix.

                                Comment

                                Working...
                                X