Announcement

Collapse
No announcement yet.

Mir Now Allows Multi-Threaded Compositing

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

  • Mir Now Allows Multi-Threaded Compositing

    Phoronix: Mir Now Allows Multi-Threaded Compositing

    Canonical's Mir Display Server for Ubuntu has received some interesting commits in the past day...

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

  • #2
    Originally posted by BO$$ View Post
    So apparently mir isn't so bad now is it?
    Has anyone really complained about the lack of support for multi-threaded compositing or a fully synchronous client API in Mir?

    Comment


    • #3
      Originally posted by Teho View Post
      Has anyone really complained about the lack of support for multi-threaded compositing or a fully synchronous client API in Mir?
      I wonder if Wayland compositors are going to be multi-threaded in the same way.

      Comment


      • #4
        Originally posted by renox View Post
        I wonder if Wayland compositors are going to be multi-threaded in the same way.
        There should not be much problem with that, as its back-end implementation details more than protocol issues (and even then you can easily add extensions to wayland).

        Though I have no idea if Wayland/Weston will want such actual architecture.

        Comment


        • #5
          Originally posted by BO$$ View Post
          So apparently mir isn't so bad now is it?
          Wayland doesn't need this at all because wayland is not a server nor is a compositor.

          i'll try to put it as easy to understand as i can, ok?

          1.) Wayland is a library that allows you to send information to the GPU in a standard fashion [like opengl library or openvg or GPU ASM] to realize an specific task(in this case handling buffers/FB/input and surfaces).
          2.) LOL, i can do that with simple Opengl and Glut or even ASM!!!, yes is true but if everyone handle the GPU in his own accord you will get 50 different incompatible render systems and 50 compositor totally different.
          3.) then OMG why not write a common library that do that, so we are all in the same page?? welcome to Wayland, Wayland is exactly that
          4.) Ok, so wayland is not a server/compositor/renderer/etc but is more like a new GPU language designed to handle UI? i mean like OpenVG for SVG graphics?, yes, but more exactly is to handle the minimal coordination and allocation needed for GUI like surfaces/input/overlays/layers and buffers
          5.) So wayland actually don't draw anything? nope it doesn't have to, wayland is the foundation barebone minimally needed to handle Gui in the GPU, once wayland deliver you a functional surface you can do whatever the hell you want with it as long as is valid for the GPU ofc.
          6.) So i can paint a wayland surface with Directx and OpenVG or using my pro skills in AMD GPU ASM? hey, what about TGSI? yes in all as long as you have proper driver support
          7.) What is a compositor needed then? well the compositor is the control center that manages the global "Desktop" concept and how all those surface react and interact with each other[pretty much anything that kwin/shell/compiz/etc do]
          8.) Cool i wanna write my own compositor, where i can see an example of how to do it?, well you can checkout weston which is exactly that a runnable example of how to write a compositor for Wayland
          9.) Weston was not part of wayland? no, those are 2 different projects
          10.) so i don't need Weston? no, sure weston can be used and expanded if you consider is good enough for your needs but is only an example remember
          11.) I wanna write a compositor, can i do it with threads/SIMD and C++11 cool stuffs? sure, is up to you which tool use to make the best of your compositor
          12.) Im a game dev and i know many cool tricks to render more efficiently in Opengl, can i use them? sure as long as is a GPU complaint and the hardware supports it, you can composite or render using any technique or language you feel is best for the task.
          13.) wayland is a power hog, i kinda heard? nope, in fact is probably the most power efficient protocol ever since it barely use the GPU to start with
          14.) ohh Yeah, wayland is not a server/compositor, so my compositor and the toolkits are the actuals responsible for power efficiency, right? yes, exactly since both the Toolkit and the compositor handle the GPU operations and is ovbious if you boost the eyecandy and your monitor size the GPU need more power to keep up, just like happens with games.
          15.) so Mir is the same stuff? no, Mir is an actual server/compositor like X11 was but with less overhead and backtrips that happens to use EGL too and is in a much earlier stage of development compared to wayland.
          16.) so this threaded compositing stuff for Mir should be compared to wayland? i think is with other compositors, right? yes, you should ask if kwin have threaded compositor or Mutter or Compiz, this has nothing to do with Wayland

          as always anyone is free if point out if a correction is needed and i hope it explain a bit better what is wayland and stop spreading FUD[prolly hell will frooze first but hell i can try]

          Comment


          • #6
            Thank you jrch.. cause I got tired of being the only one here who knew what they were talking about / issuing corrections in Wayland threads..

            Comment


            • #7
              Only one? Hey, I've been telling anyone who listens that Wayland is a standard, over and over... but the Ubuntrolls don't listen. They just go on and on about "lolz wayland still isn't finished we want a new display server now!!!!111" which kind of makes me wonder why they think it's a good idea to start an entirely new one, instead of using all the effort already put in to wayland...

              Comment


              • #8
                Holy crap. "asynchronous" is spelled right. But there's an extra "a" floating just before it in that sentence. So close.

                Comment


                • #9
                  Though I have no idea if Wayland/Weston will want such actual architecture.

                  Comment


                  • #10
                    Originally posted by dee. View Post
                    Only one? Hey, I've been telling anyone who listens that Wayland is a standard, over and over... but the Ubuntrolls don't listen. They just go on and on about "lolz wayland still isn't finished we want a new display server now!!!!111" which kind of makes me wonder why they think it's a good idea to start an entirely new one, instead of using all the effort already put in to wayland...
                    Remember that standard is what is actually used. Current standard is X.Org and the next one could easily be Mir.

                    Comment


                    • #11
                      Originally posted by BO$$ View Post
                      So apparently mir isn't so bad now is it?
                      Big oops for Ubuntu haters.

                      Comment


                      • #12
                        Originally posted by ворот93 View Post
                        Remember that standard is what is actually used. Current standard is X.Org and the next one could easily be Mir.
                        No, a "standard" is something created or adopted by a recognized standards-settings body that is intended to allow interoperability between different projects.

                        Mir is neither of these, it was created by one group, is intended to primarily serve the needs of that group, and is not accepted by any recognized standards-setting organization. Wayland, on the other hand, is a standard, specifically a standard accepted by freedesktop.org (the recognized standards-setting body for FLOSS software), and was created with the goal of being cross-distribution.

                        What you are thinking of is "de facto standard", which is not the same thing.

                        Originally posted by ворот93 View Post
                        Big oops for Ubuntu haters.
                        As others have pointed out, this was never one of the criticisms of MIR, so I don't see how it really affects the discussion in any meaningful way.

                        Comment


                        • #13
                          Originally posted by TheBlackCat View Post
                          Wayland, on the other hand, is a standard, specifically a standard accepted by freedesktop.org (the recognized standards-setting body for FLOSS software), and was created with the goal of being cross-distribution
                          True for the second part but for the first part since when freedesktop.org is a standard setting body for FLOSS software?

                          A standard setting body is something like the IETF, W3C or the ISO, I don't think that freedesktop.org is like this..

                          Comment


                          • #14
                            Originally posted by TheBlackCat View Post
                            No, a "standard" is something created or adopted by a recognized standards-settings body that is intended to allow interoperability between different projects.

                            Mir is neither of these, it was created by one group, is intended to primarily serve the needs of that group, and is not accepted by any recognized standards-setting organization.
                            And the "standards" body needs to be accepted by everyone in the first place. Otherwise I can claim to be a Linux king who sets THE standards
                            Originally posted by TheBlackCat View Post
                            Wayland, on the other hand, is a standard, specifically a standard accepted by freedesktop.org (the recognized standards-setting body for FLOSS software), and was created with the goal of being cross-distribution.
                            Freedesktop.org rubber stamps everything by Red Hat / GNOME.

                            Comment


                            • #15
                              Originally posted by renox View Post
                              True for the second part but for the first part since when freedesktop.org is a standard setting body for FLOSS software?

                              A standard setting body is something like the IETF, W3C or the ISO, I don't think that freedesktop.org is like this..
                              Sorry, I should say "FLOSS desktop environments". And yes, the whole point of freedesktop.org is that it is recognized by all FLOSS desktop environments as the body responsible for settings the standards for those desktop environements. In fact that is its one and only purpose in existing.

                              In this regard it is more like the W3C rather than the ISO, in that it establishes standards that apply to a certain class of software, rather than to anything and everything.

                              Comment

                              Working...
                              X