No announcement yet.

How next-gen OpenGL [5] will kill DirectX 12 (i.e. AMD's Mantle on Linux & Android) !

  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by kaprikawn View Post
    No, game developers will target the widest range of hardware available, which means the lowest watermark, not the high watermark as you seem to suggest. Mobile devs will be on OpenGL ES 2.0 for a long time to come, because that's what just about every mobile chipset supports.

    And if game devs focus on OpenGL ES, chipset vendors will carry on supporting that, perpetuating the current 'chicken and egg' situation. Sure, they'll support OpenGL ES 3, but until smartphone generations churn to the point where most people are carrying ES 3 hardware, devs will carry on targeting ES 2.

    OpenGL ES is the standard on mobile, Mantle and/or OpenGL 5 won't even figure.
    As for OpenGL ES, I disagree. "on paper" support is already here. Real support, will require somebody using those code paths, to fail & report it to gpu vendors, so it will take time. (Esp. without direct Google involvement. Khronos test suite ... hard to see that working as well as WHQL..)

    But OpenGL ES 3.0 is already here (on bigger number of devices then iDevieces combined!). OpenGL ES 3.1 is behind the hill top (but we will have to see if any updates to current drivers land.. As to date I can not find any updates on Khronos conformance list. :| )

    So its not so gloomy for OpenGL ES 3.0/3.1 especially for apps that will benefit (anything but simple 2D over 3D).

    That being said, hardcore gaming on ibm pc clones is still big business (and it can benefit from consoles, more then ever).
    So Mantle/DX12/OpenGL "AZDO" are important here. (OpenGL "AZDO" -> OpenGL 4.2 + extensions, or any driver that currently support OpenGL 4.4, so its AVAILABLE RIGHT NOW)

    Biggest show stopper for OGL AZDO is OSX (if writing OpenGL code then OSX is important target most of the time). Apple need to step up their game... which they may not want to, as Metal 2.0 would suite them better (but that would also "release" Linux/Windows OpenGL from reliance on Apple whim).

    And Mantle and DX12 share many basic concepts. So if game engines are being designed for it, they will work well with DX12 when its available, because all system-wide decisions will be made beforehand.


    • #12
      Originally posted by johnc View Post
      I also think that people have extraordinary over-the-boundaries opinions of Mantle, and I'm not sure where this even comes from. Apparently marketing is really effective.
      1. I had a Radeon x800, that was the last grafics card that u seen a differnce between this card and newer ones, because it supported not fully dx 9 I think (shader model 3.0) So u at best had no (dont remember the name a better version of bloom). some games just did not run at all.

      Since then, there were exactly NO improvements in visuals in real games. yes u could draw more stuff because dx10 or dx11 are more efficient, so can draw cheap maybe 20 objects instead of 5, or u can draw higher textures. All other stuff were at best vendorspecific stuff, like the nvidia physic effects, or amds better faces technology (dont remember the names).

      But even on that front microsoft failed, even planetside 2 has a dx11 render path, still a good quadcore cpu is not enough to play it fluently it has to be at least a Core i7 to have some fun.

      So now in the past it was normal that u can buy a 100 euro cpu and a good grafic card to play games at least mid to high level if not better. Now this seemed to be not true anymore. It looked like future games need a Intel processor and amd cpus are just to bad even a midrange intel cpu would not be good enough it had to be a 250 euro core i7.

      So Mantle solved that problem, sadly not for every title because not every game supports it but it shows that every game can run wihtout problems with a 100 euro cpu. If it does not its because the developers made it explizitly by design choice to not run on your hardware (because they did not use mantle).

      Thats huge. And especialy as amd hardware fan, thats the difference between amd is obsolete complete for gaming or not.

      And they delivered bigger improvements in speed than dx10 dx10.1 dx11.0 dx11.1 brought together.

      Sorry thats the biggest event happend in gaming since 10 years. Even if mantle itself really would fail, its as example the death of directx, yes there is a dx12, but its not really dx12 its a more or less complete rewrite... its another thing, they call it again dx, cause of marketing and using the brand. But in reality its a complete rewrite.
      Last edited by blackiwid; 08-06-2014, 12:31 PM.


      • #13
        its even bigger than that, it removes nearly all advantages of consoles.

        one of the biggest argument for consoles were, u can buy a console once and then u can play games for several years and they keep getting looking better because the develoopers learn to use the hardware more efficiant.

        Thats with mantle true for computers true, or at least it has the potential to be that way. If u mix that with some kind of gaming-pc versioning like steammachines that have some same hardware like in 1st gen steammachines minimum 8gb ram etc. u can define 3-4 default pc layouts for every 2 years or so u support and tune your games to this specs.

        yes thats more profiles than u have on consoles but its more managable than supporting everything have no specification groups that many have the same.

        Hope that makes sense, but it changes the hole gaming market very much, mantle, in respect to consoles and to intel amd and so on...

        So thats why people are excited I think.


        • #14
          I have to admit that I'm quite a bit more pessimistic. Graphics quality on consoles is way sub-par compared to what you can have on the PC, and console gamers just came to adjust themselves to lower resolutions, no AA, lower frame rates, etc. Developers target a much lower standard.

          The only real advantage consoles have is that it's a single target platform so developers can target the one piece of known hardware.

          It's not really evident that game developers even care about PC gaming, and even with Mantle. Their first target is the console market and then they "port" it over to PC, targeting the least common denominator (so DX9 and Intel integrated graphics) and call it a day.

          Mantle isn't really going to change anything. It helps in a handful of CPU-limited scenarios... but so would OGL 4.x and DX12... and developers aren't going to use those because it takes them at least a decade to embrace new tech, so who cares?


          • #15
            Here's a typical example:

            Kepler has a nice feature set which we'll call "bindless multidraw indirect". Now Kepler has been out for what... three years now? And how many have spent hundreds on these video cards?

            Can somebody count for me the number of games out there that take advantage of BMDI? Yeah... it's zero.

            So people paid for a feature that isn't even used by game developers, and paid for GPU efficiency that's completely unused.

            Look at how long the industry dragged its feet to adopt the new things in DX 11, or 11.1 or 11.2. OGL 4.x has been out for YEARS and there's not a single game using it.

            The problem isn't the API here. It's not the GPU vendors either.


            • #16
              This is totally wrong

              This is all based on poor assumptions. To take anything like this seriously you need to link something for proof, instead of taking up space with conjecture..

              An example, -- Directx 12 is NOT based on Mantle, they are different api's and was mentioned since Mantle was announced..


              The Mantle Angle

              We?ve spoken to several sources with additional information on the topic who have told us that Microsoft?s interest in developing a new API is a recent phenomenon, and that the new DirectX (likely DirectX 12) will substantially duplicate the capabilities of AMD?s Mantle. The two APIs won?t be identical ? Microsoft is doing its own implementation ? but the end result, for consumers, should be the same: lower CPU overhead and better scaling in modern titles.


              • #17
                Originally posted by johnc View Post
                Look at how long the industry dragged its feet to adopt the new things in DX 11, or 11.1 or 11.2. OGL 4.x has been out for YEARS and there's not a single game using it.
                The problem isn't the API here. It's not the GPU vendors either.
                that even shows the success of Mantle more, OGL 4.x is out for years still no game, but Mantle is kind of beta out for a few months and there are some games using it, and many that will include it anounced.

                And that its not a API problem we have to see that, OGL 5.0 or dx12 has to proove that it can at least nearly as fast as mantle. They are both more highlevel apis still, so it has to be proved that it can still be that good.

                We see that multiplattform EVEN Multi-Vendor support doesnt matter taht much, else OGL would be very successful. So theese arguments are not taht true.

                Except maybe with steamos we see more opengl ports?

                Its pretty simple, most advantages opengl has gaming develoopers dont care at all, they only use it now for linux because its more or less the only way to port a game to linux, they have no choice if they had one they would choose always not to use it, like we seen before linux was a target plattform for them.

                They just see they have more work to do to get it running with opengl and till now that did not lead in big performance advantages.

                Btw a higher level api also means often more abstraction, what means more difficult to develop or maintain and to test. So this "advantage" is for many developers a disadvantage at least thats what I guess, thats maybe the reason why the gaming comapnies have no big problem with a low level engine, because they just target a few grafic cards and dont have to implement 100 different coding paths for different generations. At least not now... later on yes with 5 hardware generations and 3-10 hardware vendors it would change into more work, but in this constellation its most likely way easier to support than opengl.


                • #18
                  Gaming development is the last year production line producment, u take one engine unreal engine 3 and produce 1000 games which are all very similar change a few textures a few mechanics and u reduce product live time to a minium. And it seems opengl failed in the past onto that.

                  Maybe thats the fault of some engine developers but even the best opengl-gaming companies quited using opengl id software.

                  I just dont see it come back. I am not so deep into it, but I seen that most gaming opengl developers were pretty much pissed that opengl v4 was just a minor improvement instead of brave stepp, the problem is kronos dont prioritise gaming, its just a small part they care, they care mostly not about gaming but everything else, and that cant compete with a api that primary targets gaming development like directx. (or direct3d) even that it is all included in one package instaed of direct3d + 20 different sound and input engines helps DX. Ok we now have sdl but how long did we have that now since 15 years so companies had time before to migrate to that, why would they do that now, to migrate some old games to linux yes maybe... but I dont see majore AAA titels using it... maybe I am wrong we will see...


                  • #19
                    There are games out there that have paths for both 3.x and 4.x ... Or 3.x with some extensions from 4.x

                    And whole point of "AZDO" is to spread good PR about specific ways of solving problems that are fast, meaning that more software using those is coming.


                    • #20
                      Originally posted by johnc View Post
                      Look at how long the industry dragged its feet to adopt the new things in DX 11, or 11.1 or 11.2. OGL 4.x has been out for YEARS and there's not a single game using it.

                      The problem isn't the API here. It's not the GPU vendors either.
                      Try to think from the devs' perspective. Say 10% of your market has GL4 cards - installed base is king. How many extra sales would you make if you spent the effort to port your engine to gl4 techniques? Would those extra sales offset the cost of making the port?

                      The answers to those questions tend to be "something between 0 and 2" and "no, hell no".

                      If you tried to go hardcode and did GL4 only, you just cut your sales by 90%, possibly more.