Announcement

Collapse
No announcement yet.

Readers Think It's A Toss Up Whether Mesa Will Hit OpenGL 4.5 Next Year

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

  • Readers Think It's A Toss Up Whether Mesa Will Hit OpenGL 4.5 Next Year

    Phoronix: Readers Think It's A Toss Up Whether Mesa Will Hit OpenGL 4.5 Next Year

    Our poll on Twitter this weekend was: Do you think Mesa will catch up to supporting the latest Khronos OpenGL specification (OpenGL 4.5+) next year?..

    http://www.phoronix.com/scan.php?pag...g-Up-2016-Poll

  • #2
    Those twitter followers are crazy optimistic. If you look at Mesa's history, from 3.0 (which was early 2012 and should be thought of as when Mesa got any major backing) it took 8 months to get 3.1, 13 months after that to get 3.3 (skipping 3.2), and then 22 months to get 4.2 in core. Its obviously taking longer every jump, and while we will certainly get 4.3 this year, enhanced layouts represent a ton of shader expansion in 4.4 and are barely started. I expect those to hold up Mesa development for another year or two, and that is considering it will probably take most of 2016 just to get compute shaders.

    2016 will end with probably close to 4.3 support across the board, but after that there is very little pressure from Intel's benefactors to implement 4.4 / 4.5, and for gaming OpenGL games are targeting 4.1 due to OSX anyway. We are already seeing how slowly Intel is at getting tessellation shaders since they weren't a part of GLES 3.1, and they might only care about them now because they are in GLES 3.2.

    Comment


    • #3
      Originally posted by zanny View Post
      … it will probably take most of 2016 just to get compute shaders. …
      Compute shaders are done and even partly reviewed: http://lists.freedesktop.org/archive...er/100602.html

      I don't think 4.4 in core mesa in 2016 is too optimistic, a GPU supporting it probably is, though not completely unrealistic.

      Comment


      • #4
        Originally posted by zanny View Post
        Those twitter followers are crazy optimistic. If you look at Mesa's history, from 3.0 (which was early 2012 and should be thought of as when Mesa got any major backing) it took 8 months to get 3.1, 13 months after that to get 3.3 (skipping 3.2), and then 22 months to get 4.2 in core. Its obviously taking longer every jump, and while we will certainly get 4.3 this year, enhanced layouts represent a ton of shader expansion in 4.4 and are barely started. I expect those to hold up Mesa development for another year or two, and that is considering it will probably take most of 2016 just to get compute shaders.

        2016 will end with probably close to 4.3 support across the board, but after that there is very little pressure from Intel's benefactors to implement 4.4 / 4.5, and for gaming OpenGL games are targeting 4.1 due to OSX anyway. We are already seeing how slowly Intel is at getting tessellation shaders since they weren't a part of GLES 3.1, and they might only care about them now because they are in GLES 3.2.

        New AAA games in Linux are targeting OpenGL4.3. But I agree with the rest.

        Comment


        • #5
          Originally posted by article
          Readers Think blah blah
          Nope, it's most certainly not the readers that think that way. Because I'm a reader, I read phoronix every day, even several times a day. Yet I'm still never asked the questions, I'm just told about the answers afterwards.

          Comment


          • #6
            Originally posted by zanny View Post
            Those twitter followers are crazy optimistic. If you look at Mesa's history, from 3.0 (which was early 2012 and should be thought of as when Mesa got any major backing) it took 8 months to get 3.1, 13 months after that to get 3.3 (skipping 3.2), and then 22 months to get 4.2 in core. Its obviously taking longer every jump
            I don't think you can really use those figures to predict when 4.5 will be finished. Extensions are finished out of order and a large number of them are already done and almost all are now started.

            Originally posted by zanny View Post
            and while we will certainly get 4.3 this year, enhanced layouts represent a ton of shader expansion in 4.4 and are barely started. I expect those to hold up Mesa development for another year or two,
            As the developer working on ARB_enhanced_layouts I can tell you that is not the case. There are 6 pieces to this extension that can be completed independently but none of them represent a huge amount of work.

            Originally posted by zanny View Post
            and that is considering it will probably take most of 2016 just to get compute shaders.
            Compute shaders is done for i965 its just needs review.

            Originally posted by zanny View Post
            2016 will end with probably close to 4.3 support across the board, but after that there is very little pressure from Intel's benefactors to implement 4.4 / 4.5, and for gaming OpenGL games are targeting 4.1 due to OSX anyway. We are already seeing how slowly Intel is at getting tessellation shaders since they weren't a part of GLES 3.1, and they might only care about them now because they are in GLES 3.2.
            Intel is funding 4.4/4.5 development, and they funded a large part of the core work on tessellation shaders.

            Comment


            • #7
              Originally posted by from the article
              The poll this weekend asking readers whether they feel that Mesa will catch-up in 2016 is a mixed bag: 51% yes, 49% no. (There was also a dormant forum thread.)
              Well I don't know about the rest of your readership Michael, but very rarely do I enter the forums outside of an article and when I do, I can't say I browse the general discussion section. On the other hand if you post an article (doesn't have to be long) like you have for former polls then things will actually work out.

              Comment


              • #8
                If you prefer instead of an article you can put a poll box somewhere in the homepage.
                ## VGA ##
                AMD: X1950XTX, HD3870, HD5870
                Intel: GMA45, HD3000 (Core i5 2500K)

                Comment


                • #9
                  Originally posted by zanny View Post
                  Those twitter followers are crazy optimistic. If you look at Mesa's history, from 3.0 (which was early 2012 and should be thought of as when Mesa got any major backing) it took 8 months to get 3.1, 13 months after that to get 3.3 (skipping 3.2), and then 22 months to get 4.2 in core. Its obviously taking longer every jump, and while we will certainly get 4.3 this year, enhanced layouts represent a ton of shader expansion in 4.4 and are barely started. I expect those to hold up Mesa development for another year or two, and that is considering it will probably take most of 2016 just to get compute shaders.

                  2016 will end with probably close to 4.3 support across the board, but after that there is very little pressure from Intel's benefactors to implement 4.4 / 4.5, and for gaming OpenGL games are targeting 4.1 due to OSX anyway. We are already seeing how slowly Intel is at getting tessellation shaders since they weren't a part of GLES 3.1, and they might only care about them now because they are in GLES 3.2.
                  Fixed that for you, breaking that down to all jumps that were made. You cannot skip a GL version number just because Mesa skipped it:

                  3.0 to 3.1 - 8 months
                  3.1 to 3.2 - 6.5 months (approx.)
                  3.2 to 3.3 - 6.5 months (approx.)
                  3.3 to 4.0 - 7.4 months (approx.)
                  4.0 to 4.1 - 7.4 months (approx.)
                  4.1 to 4.2 - 7.4 months (approx.)

                  Fixed conclusion:
                  "Its obviously taking approximately the same amount of time every jump, and thus we will get 4.3 2016. Also enhanced layouts represent not too much work of shader expansion in 4.4 thus should also be ready in 2016."

                  If Mesa 11.2 will get GL 4.3 core completed (thus Mesa 12), and if we then stay in this rythm, the following numbers would look like this:

                  4.2 to 4.3 - 6 months (March 2016)
                  4.3 to 4.4 - 6 months
                  4.4 to 4.5 - 6 months (March 2017)

                  Thats all for Mesa Core. Not too bad I think.

                  Comment


                  • #10
                    Do we still talk about OpenGL? I thought Vulkan was all the rage these days.

                    Comment

                    Working...
                    X