Announcement

Collapse
No announcement yet.

Ubuntu's Unity/Compiz Gets Even Slower

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

  • #46
    Originally posted by ninez View Post
    you do also realize that the compiz that ships (or is updated into Ubuntu/Unity) doesn't always correspond to upstream compiz development, right?
    No, I don't because it's not true! According to Compiz’ own development page Compiz’ release cycle was aligned to Ubuntu’s. It makes absolutely no sense that Ubuntu ships with additional patches on top.

    Said said, I didn't even ask you, so STFU.

    Comment


    • #47
      FYI if you want to follow progress on the regression bugs, see the following (at least):

      https://bugs.launchpad.net/compiz/+bug/1037411 "
      [regression][GLES] Benchmark results are 15-40% lower with the gles2 code (compiz 0.9.8.0)"

      and regarding full screen games performance as hinted in the above report:

      https://bugs.launchpad.net/ubuntu/+s...iz/+bug/980663 "Compiz won't start if "unredirect fullscreen windows" is enabled"

      The latter problem with unredirect fullscreen windows settings is apparently already fixed in the staging PPA (https://launchpad.net/~unity-team/+archive/staging).

      Not sure if there are some unity specific fixes brewing as well, but at least for 12.04 the next Unity stable update should bring regional damage fixes among other things.

      Comment


      • #48
        The problem with compiz is that some games like Trine 2 or Bastion use a peculiar way of going to fullscreen, and unredirect fullscreen windows won't suspend compiz effects, and performance is the same as running the game in a window. The only way to get proper performance is to remove compiz completely which is not possible because Unity is a compiz plugin.

        Comment


        • #49
          Originally posted by narciso View Post
          The problem with compiz is that some games like Trine 2 or Bastion use a peculiar way of going to fullscreen, and unredirect fullscreen windows won't suspend compiz effects, and performance is the same as running the game in a window. The only way to get proper performance is to remove compiz completely which is not possible because Unity is a compiz plugin.
          Hopefully also a bug like that could be analyzed far enough to fix it. Various other redirection related bugs were also very recently fixed (also in staging):

          LP #1041047 - "Unredirect Fullscreen Windows" stay on top (unredirected) even when they're not on top any more (or the output is transformed)"

          LP #1041066 - "Unredirect Fullscreen Windows sometimes fails to unredirect fullscreen windows at all"

          I didn't find a bug describing the Trine 2 / Bastion full screen problem, though, unless it was fixed by the gtk-window-decorator related fixes of the latter bug's branch.

          Comment


          • #50
            Originally posted by Awesomeness View Post
            No, I don't because it's not true! According to Compiz’ own development page Compiz’ release cycle was aligned to Ubuntu’s. It makes absolutely no sense that Ubuntu ships with additional patches on top.
            except that it is true. compiz-bzr is always slightly ahead of 'official' releases / updates. Generally, new code will go through testing / staging before it is officially released / pushed through as an update. (i guess your not familiar with this concept either). I never said they add additional patches to compiz, i don't know where you got that idea.

            Originally posted by Awesomeness View Post
            Said said, I didn't even ask you, so STFU.
            It must be your time of the month - here are some tips on how to lesson your agitation; http://www.wikihow.com/Reduce-Menstrual-Cramps .... maybe some of these will help you out

            I'd also like to point out, since apparently you're not familiar with how forums work - in pretty much any forum; people post and sometimes other people will post a comment/reply to a post - even if the original comment wasn't directed at them. That is generally common place in any forum. ~ apparently you think otherwise...

            Comment


            • #51
              Originally posted by ninez View Post
              GTK stands for 'Gimp Toolkit' - it was not originally developed by Gnome nor was it a Gnome Project - it was developed to replace Motif as Gimp's toolkit.

              lots of DEs/WMs use it and that doesn't make them gnome-powered, it makes them GTK+ powered. an example of a 'partly gnome-powered' DE - would be Unity or Elementary OS's DE or Cinnamon ~ those are examples of DEs that are powered by gnome, since they are all essentially gnome, just with different shells, modified/forked software and other slight (mostly superficial) modifications/alterations.
              I know that, BUT I would guess that many many commits for gtk come from gnome developers, so yes you can use it seperatly and gnome depends on gtk and not the other way around, but again I would guess that more people use gnome than gimp.

              So maybe I am wrong but I guess that more devs who work on gnome-components also send patches or directly work on gtk than any other projekt... but if someone knows more... so if that is true I think it would be valid to say that if lets say 99% of the people would use kde and and nobody would want any gnome stuff (incl. unity) and the gnome-projekt would not exist, gtk would maybe died or would be way more behind in progress than it is today.

              Comment


              • #52
                Originally posted by blackiwid View Post
                I know that, BUT I would guess that many many commits for gtk come from gnome developers, so yes you can use it seperatly and gnome depends on gtk and not the other way around, but again I would guess that more people use gnome than gimp.
                Again, a little bit iffy there. Gnome does have a large user base, but the Gimp does as well. I would not be so quick to call.

                Comment


                • #53
                  Ubuntu's Unity/Compiz Gets Even Slower

                  I appreciate the article but Ubuntu 12.10 just entered Beta a week ago?

                  Comment


                  • #54
                    Originally posted by SmSpillaz View Post
                    The regional updates code is currently not active while we figure out how to make it work with changes in upstream compiz. This is why the slowness exists at the moment.
                    Thanks. Do you expect to have this fixed in time for Quantal final? And where is upstream compiz located these days?

                    Comment


                    • #55
                      I do not give ETAs on how long things will take. Needless to say, we are working on it, and some good ideas are being thrown around as to how to deal with it. That being

                      1. Using glBlitBuffer instead of painting a texture backed framebuffer
                      2. Ensuring the contents of vertex buffers are more static (improves video)
                      3. Fullscreen window unredirection.

                      Thats not a promise that any of this will land *this cycle* per-se, but it is a note that we are aware there was a performance hit when we merged the work to make compiz work on arm platforms and we're actively dealing with it.

                      Comment


                      • #56
                        Originally posted by SmSpillaz View Post
                        ...
                        3. Fullscreen window unredirection.

                        Thats not a promise that any of this will land *this cycle* per-se, but it is a note that we are aware there was a performance hit when we merged the work to make compiz work on arm platforms and we're actively dealing with it.
                        Thanks Sam. I really appreciate your reply. Keep up the good work and drink lots of caffeine because we are waiting anxiously

                        Am I correct in interpreting the compiz 0.9.8.2 release notes as saying that the fullscreen window unredirection problems are now mostly fixed? From the release notes:

                        * Gamers rejoice! CCSM > Composite > "Unredirect Fullscreen Windows" now works more reliably, with a few bugs remaining. If you turn it on then fullscreen windows like games will get direct rendering access and won't be slowed by Compiz (or Unity) any more.

                        They go on to mention the following bugs amongst the list of bugs fixed:

                        980663 - Compiz won't start if "unredirect fullscreen windows" is enabled
                        1041047 - Unredirect Fullscreen Windows stay on top (unredirected) even when they're not on top any more (or the output is transformed)
                        1041066 - Unredirect Fullscreen Windows sometimes fails to unredirect fullscreen windows at all
                        1046661 - Unredirect Fullscreen Windows stay on top (unredirected) even when an RGBA window is stacked above it

                        Comment


                        • #57
                          Originally posted by madbiologist View Post
                          Thanks Sam. I really appreciate your reply. Keep up the good work and drink lots of caffeine because we are waiting anxiously

                          Am I correct in interpreting the compiz 0.9.8.2 release notes as saying that the fullscreen window unredirection problems are now mostly fixed? From the release notes:

                          * Gamers rejoice! CCSM > Composite > "Unredirect Fullscreen Windows" now works more reliably, with a few bugs remaining. If you turn it on then fullscreen windows like games will get direct rendering access and won't be slowed by Compiz (or Unity) any more.

                          They go on to mention the following bugs amongst the list of bugs fixed:

                          980663 - Compiz won't start if "unredirect fullscreen windows" is enabled
                          1041047 - Unredirect Fullscreen Windows stay on top (unredirected) even when they're not on top any more (or the output is transformed)
                          1041066 - Unredirect Fullscreen Windows sometimes fails to unredirect fullscreen windows at all
                          1046661 - Unredirect Fullscreen Windows stay on top (unredirected) even when an RGBA window is stacked above it
                          Yes, however it isn't guarunteed to work with unity at the moment (for some fairly nontrivial reasons).

                          Comment


                          • #58
                            Oh I totally forgot to mention. Upstream compiz is developed on launchpad.

                            lp:compiz

                            Comment


                            • #59
                              Control Test

                              What I'd really like to see is one of the things Martin suggested in his benchmarking article that should be easy for Michael to do in future benchmarks of compositors; include a control test. That is, include a test where there is no compositor / desktop environment running, just plain X server.

                              Here's the quote from his site:

                              "Now I think back of the hypothesis we have and I notice that while we have quite some data sets on the influence of desktop environments on the game rendering performance, one important data set is missing: the control. For the given hypothesis only one control can be thought of: running just an X-Server without any desktop environment and run the test there. This would be a very good control as it ensures that there is no overhead introduced by any desktop environment. "

                              Comment


                              • #60
                                Originally posted by sheldonl View Post
                                What I'd really like to see is one of the things Martin suggested in his benchmarking article that should be easy for Michael to do in future benchmarks of compositors; include a control test. That is, include a test where there is no compositor / desktop environment running, just plain X server.

                                Here's the quote from his site:

                                "Now I think back of the hypothesis we have and I notice that while we have quite some data sets on the influence of desktop environments on the game rendering performance, one important data set is missing: the control. For the given hypothesis only one control can be thought of: running just an X-Server without any desktop environment and run the test there. This would be a very good control as it ensures that there is no overhead introduced by any desktop environment. "

                                A number of us suggested that BMarks also be run without compositor (use xinit twm or something) or without a WM altogether (use xinit someBenchmark). Unfortunately, a number of the BMarks in PTS probably won't run without a WM. Does anyone have a moment to try Heaven or some other benchmark without a WM?

                                F

                                Comment

                                Working...
                                X