Announcement

Collapse
No announcement yet.

It Doesn't Look Like AMD Mantle Is Coming To Linux, SteamOS

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

  • #16
    It's better to improve and support OpenGL which Linux already has, than to bring over mantle which is brand new and may not even be practical to use.

    Comment


    • #17
      Originally posted by MWisBest View Post
      "others" includes AMD, but unless somebody clicks the link to the article talking about that they'd most likely assume it's just Nvidia and Intel doing that while AMD is doing nothing. I'd appreciate it if that could at least be reworded.
      Well in fairness it was NVIDIA alone that was pushing that OpenGL presentation for a year+ until Intel and AMD decided to join the talks.

      And it was NVIDIA alone that was pursuing D3D11 optimizations in their latest Windows driver.

      Comment


      • #18
        Originally posted by johnc View Post
        Well in fairness it was NVIDIA alone that was pushing that OpenGL presentation for a year+ until Intel and AMD decided to join the talks.

        And it was NVIDIA alone that was pursuing D3D11 optimizations in their latest Windows driver.
        Consider, though, that for AMD and Intel, their principle OpenGL implementation is still at 3.3. For all intents and purposes at least. Very few games use OpenGL on Windows. WebGL might change that a bit, but in practice almost all real OpenGL demand comes from CAD and Linux use cases. Since Apple iron fists their OpenGL anyway.

        So why would Intel and AMD be pushing anything major while the Mesa project still hasn't hit 4.0 yet? I'd worry about the immediate needs to implement the rest of what is out there rather than push the API forward, set themselves further behind, and give Nvidia an edge by sharing their Windows and Linux codebases in an unauditable behemoth driver.

        Comment


        • #19
          Originally posted by zanny View Post
          Consider, though, that for AMD and Intel, their principle OpenGL implementation is still at 3.3. For all intents and purposes at least. Very few games use OpenGL on Windows. WebGL might change that a bit, but in practice almost all real OpenGL demand comes from CAD and Linux use cases. Since Apple iron fists their OpenGL anyway.

          So why would Intel and AMD be pushing anything major while the Mesa project still hasn't hit 4.0 yet? I'd worry about the immediate needs to implement the rest of what is out there rather than push the API forward, set themselves further behind, and give Nvidia an edge by sharing their Windows and Linux codebases in an unauditable behemoth driver.
          Well one clear benefit is that it gives game developers a reason to actually adopt OpenGL. We see just how hesitant they have been to walk away from D3D, and we know that the absolutely awful OpenGL drivers of at least one of the vendors is partly to blame.

          So when you're going up against the best rendering API out there (D3D), you might need to bring something a little bit better than a 5-year old implementation. But I think what you said can basically be reiterated by acknowledging what we already know: AMD and Intel don't really care jack about OpenGL. AMD so much so that they made their own friggin' API instead of just working inside OGL.

          JMO.

          Comment


          • #20
            Originally posted by phoronix View Post
            NVIDIA and others (including AMD) are instead more focused on reducing the OpenGL driver overhead
            That right thing to do.

            Comment


            • #21
              Honestly I never expected to see mantle in Linux... for that matter, I don't expect to see mantle anywhere in the next few years.
              Besides, AMD has given us Freesync(just submitted, and apparently accepted by VESA) so one less place for proprietary solutions.

              Comment


              • #22
                Originally posted by liam View Post
                Honestly I never expected to see mantle in Linux... for that matter, I don't expect to see mantle anywhere in the next few years.
                Besides, AMD has given us Freesync(just submitted, and apparently accepted by VESA) so one less place for proprietary solutions.
                Hmm... I am already using mantle on my gaming computer so yes, it is here. Both Battlefield 4 and Thief are running great with it and I got a few more FPS out of it. More importantly, the gaming experience feels a lot more smooth than with DirectX. (And yes, I have compared with both AMD and Nvidia cards).

                I love it, and I would love to have it come to Linux since both Cryengine and FrostBite do support mantle.

                Comment


                • #23
                  Originally posted by AJSB View Post
                  Like others, I'm not worried at all with Mantle on Linux and rather prefer that AMD solve their problems with Catalyst like the stuttering in (some) Source Engine games....and Catalist A.I. option should be by default disabled (when enabled it makes graphical glitches in several games) so users will have a better experience out-of-the-box.
                  Isn't that train of thought a bit weird? Disabling Catalyst AI by default is seriously just a "flip of the switch" type of effort from AMD's side. Wouldn't it make more sense if they, y'know, fixed the bugs that cause these graphical glitches in the first place? That's a bit more comparable to working on Mantle, effort-wise.

                  Comment


                  • #24
                    Originally posted by liam View Post
                    Honestly I never expected to see mantle in Linux... for that matter, I don't expect to see mantle anywhere in the next few years.
                    Besides, AMD has given us Freesync(just submitted, and apparently accepted by VESA) so one less place for proprietary solutions.
                    Variable VBlank refresh is an eDisplayPort feature, AMD just called it something else and showed laptops using it. They did probably put enough pressure on VESA to get it in DP 1.2a, though.

                    AMD so much so that they made their own friggin' API instead of just working inside OGL.
                    I'd counter argue this saying "OpenGL is a massive bloated 25 year old mess" and that a really well architected modern stateless API for GPUs would be great, but only if it were an open standard, and Mantle isn't, so it can still piss off. GLES is kind of an attempt to remove some cruft, but it still uses an imperative C derived style in functions and shading that makes me feel like I'm in film grain in the 1980s.

                    Comment


                    • #25
                      Originally posted by zanny View Post
                      I'd counter argue this saying "OpenGL is a massive bloated 25 year old mess" and that a really well architected modern stateless API for GPUs would be great, but only if it were an open standard, and Mantle isn't, so it can still piss off. GLES is kind of an attempt to remove some cruft, but it still uses an imperative C derived style in functions and shading that makes me feel like I'm in film grain in the 1980s.
                      You also can look at OpenGL/OpenGLES with all its versions and extremely different performance behaviors (I call it the performance breaking minefield) on different vendor implementations as an enormous amount of different APIs that just use the same function names to raise the confusion.

                      Why does Khronos even have Microsoft as board member? The worlds most determined enemy of OpenGL I can think of...

                      DSA functions are still not a part of the core versions and the EXT version is not implemented by intel. It seems the us few users of such esoteric languages like C++ just don't matter that much.

                      Oh and look at ARB_buffer_storage, the little cripple that got a hint stuck in its knee. Even the ARB specification has some wry humor about that... So you still stuck without knowing if you got VRAM, pinned memory or non pinned memory including all the random buffer moving by moody drivers.

                      AMD continues with there multiple personality disorder that makes them change there mind every few months. Right now it looks like they want mantle to be a pre DX12 exclusively for there GCN GPUs and after DX12 let it die without ever open it to the public. But hey! Maybe in one month AMD tells us again about the all open for everyone Mantle API... who knows!?

                      Comment


                      • #26
                        Isn't AMD currently adopting OGL4.4 and Nvidia's new OGL extensions?
                        And aren't most big game engines moving to OGL?

                        If so isn't Mantle a moot point at this point?

                        Comment


                        • #27
                          AMD talked about upcoming Mantle games, and they had this interesting statement about Linux (emphasis mine): http://community.amd.com/community/a...are-on-the-way
                          Originally posted by AMD
                          On March 20 Microsoft announced DirectX® 12, the next major evolution of its own game API. This is terrific news because it really draws attention to the value of low-level programming and Mantle’s leading contribution. With DirectX 12 games still over 18 months away and no alternatives in sight for Linux gamers, Mantle’s future looks bright.
                          If my interpretation is correct, this means that either Mantle or some other form of low-level programming will come to Linux.

                          Comment


                          • #28
                            But why is mantle not supported in the linux catalyst since 14.1? Why is it STILL not supported by linux catalyst? Why isn't there even a public specification? Not even public documentation. Not even a single line of example code?


                            Is it cynical to joke that we will probably have Direct3D 12 support in Mesa on linux before any mantle support?

                            Comment


                            • #29
                              what to make of this?

                              http://www.pcworld.com/article/23647...tle-tools.html

                              Does Richard Huddy have the clout to make this happen? Seems he recently rejoined AMD.

                              Comment

                              Working...
                              X