Announcement

Collapse
No announcement yet.

AMD Catalyst 11.5 Linux Driver Released

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

  • #21
    Originally posted by BlackStar View Post
    What's the business plan for that?
    For what? DX-on-Linux? Well, for hardware vendor any drivers are abuse, they don't get direct income from this kind of work. But if they will release DX API (that will require some additional effort, maybe some wine code-digging and incorporating), ATI proprietary drivers for Linux will get some superiority against Nvidia. Plus, in addition, that will make Linux more attractive platform for game developers (but that's not an argument for AMD, of course)

    Currently, AMD cards are off-chart for linux gamers, as even OSS drivers are working better and faster than proprietary ones. But OSS drivers are lacking (yet) some major features and modern OpenGL support, so many games are just not supported yet. Plus, many of newly incorporated features like float textures or S3TC support require git kernel, recompiled mesa (with enable patented) or some bindist-unavailable software (libtxc_dxtn), which makes OSS drivers badly available for many of binary-distributions users.

    Just few hot-topic examples:
    1) Starcraft 2: on NVidia working perfectly and after drivers update FPS are even better then on Windows. Fglrx suffering from multiply graphics bugs and require some extensive magic to be even playable. FPS are far from Windows ones. OSS not tested.
    2) Bulletstorm: on NVidia working, but FPS are low. Playable, but without saving (no GfW). On fglrx it's starting only from 11.4 driver version and suffers from terrible graphics miscolor. FPS are terrible. OSS drivers are working mostly ok (on 39_rc* kernel and git mesa with --enable-patented and libtxc_dxtn), but with some black bounding-boxes bug. FPS are better than of fglrx, it would be even playable if not black-bounding-boxes.
    3) Portal 2: on NVidia working perfectly and with good FPS. On fglrx it's runnable only with 11.4 driver version (which was released few weeks after P2 release), but suffers from multiply crashes on high settings and low FPS.
    4) Dead Space 1/2 (same engine): On NVidia working mostly perfectly and with good FPS. On fglrx it crashes in many cases, suffers from multiply graphics bugs, missing effects and so on.

    And such situation is for many games: nvidia - ok, fglrx - fail, oss - not yet ready. Therefore - most of wine games are on NVidia. And as heavy discrete graphics systems like HD5/6 are required mostly for games (or some professional software, unavailable on Linux too) - this means lost hi-end market for AMD. And this situation could not be changed by just somehow suddenly releasing ideal drivers - AMD's reputation is already bad and many people don't believe that OSS drivers may change it (mostly because they don't see their's fast progression due to bindists' inertia). AMD should do something really new and progressive, which was never made by NVidia, if they want this market. DX API support is just that thing. But, imo, AMD just don't want this market =).

    Originally posted by Djhg2000
    Because they would get Microsoft up their trying to get back their lost toy sales.
    I personally doubt such conspiracy theories like "Microsoft paying AMD to make bad drivers for Linux to make everyone go to Windows". On Microsoft's place I will try to pay AMD to make bad drivers for Mac OS X, which is much more dangerous for them on desktop market. I may even believe in some theory that Microsoft founded recent attack on PSN - that will make some sense, as this improved Microsoft's position on consoles market.

    But Linux... Well, it would be loss of money, by now, IMO.

    Comment


    • #22
      Wine allows some games, designed and programmed for windows, to run under Linux. Leaving aside any legal reasons, trying to get native directx on Linux just so that wine can run better is a bad idea.
      Wine also historically runs better with nvidia drivers simply because more effort was put into working around any nvidia problems (the fglrx drivers weren't available back then).
      Native games are what really matter, and there I have no issues with the fglrx drivers. ETQW, Quake4, Doom3, UT2004, Prey, humble bundles, all work without any problems.

      Comment


      • #23
        Originally posted by mirv View Post
        trying to get native directx on Linux just so that wine can run better is a bad idea.
        From users point of view: Having native directx support on linux (currently dx10/11 is implemented on Gallium achitecture as proof-of-concept, so I doubt there is any legal issues - it's not protected anyhow, like --enable-patented) will reduce costs for both porting/development and support on linux. In most cases, games are using some cross-platform solutions for their low-level needs anyway, so DX vs. OpenGL is most hard part of porting (you see - most of games, which are using OpenGL as their primary API were ported to Linux) _and_ require additional support team. This is too much for small companies even for Mac OS X market, which is bigger then desktop Linux market.

        But such arguments are for users. For AMD additional API on Linux will mean additional costs on testing and having more native games on Linux will make the situation even worse (more users == more bugs == more requirements to features == more development effort == more costs). So they need some other argument to do this and better-wine-support (and superiority against NVidia) is just fine here, IMO. But, again, desktop Linux market is too small, so that's hardly possible. We shall wait untill Gallium3D will get all required features and than DX10/11 support in mesa will really rock (but without NVidia supporting OSS drivers or some huge effort on nouveau, developers will not port games to Linux anyway as their games wouldn't work for NVidia users).

        Comment


        • #24
          No mouse cursor

          With 11.5 I am no longer able to see the mouse cursor on my external screen, I can see it's effect but not the cursor.

          I'm using slack64 13.37 on acer aspire 5542-1462 with hd4200.

          on a side note, is it possible to disable hardware accel with aticonfig ? it's disabled when using 2 head settup but then the mouse cursor is buggy on external screen.

          with hardware accel I am experiencing sporadic hardlocks, although it is getting better, and I would have liked to give 11.5 a try.

          thank you

          Comment


          • #25
            Originally posted by Night Nord View Post
            From users point of view: Having native directx support on linux (currently dx10/11 is implemented on Gallium achitecture as proof-of-concept, so I doubt there is any legal issues - it's not protected anyhow, like --enable-patented) will reduce costs for both porting/development and support on linux. In most cases, games are using some cross-platform solutions for their low-level needs anyway, so DX vs. OpenGL is most hard part of porting (you see - most of games, which are using OpenGL as their primary API were ported to Linux) _and_ require additional support team. This is too much for small companies even for Mac OS X market, which is bigger then desktop Linux market.

            But such arguments are for users. For AMD additional API on Linux will mean additional costs on testing and having more native games on Linux will make the situation even worse (more users == more bugs == more requirements to features == more development effort == more costs). So they need some other argument to do this and better-wine-support (and superiority against NVidia) is just fine here, IMO. But, again, desktop Linux market is too small, so that's hardly possible. We shall wait untill Gallium3D will get all required features and than DX10/11 support in mesa will really rock (but without NVidia supporting OSS drivers or some huge effort on nouveau, developers will not port games to Linux anyway as their games wouldn't work for NVidia users).
            I'm sure microsoft would find a way to create a legal issue, or introduce some way of making it incompatible. The directX stuff is pretty tightly integrated into windows too, and I did read that not all of it is documented, so I'm not convinced personally of any viability of a Gallium3D state tracker for DX10/11 beyond "ok, cool, it can do that". That's more the fault of microsoft than anyone else.
            All of which is moot really - games can be made using OpenGL, and would be if developers saw linux as a money-making market. Time will tell I guess!

            Comment


            • #26
              Originally posted by acer View Post
              With 11.5 I am no longer able to see the mouse cursor on my external screen, I can see it's effect but not the cursor.

              I'm using slack64 13.37 on acer aspire 5542-1462 with hd4200.

              on a side note, is it possible to disable hardware accel with aticonfig ? it's disabled when using 2 head settup but then the mouse cursor is buggy on external screen.

              with hardware accel I am experiencing sporadic hardlocks, although it is getting better, and I would have liked to give 11.5 a try.

              thank you
              Ah, that issue of a screwy cursor still around is it? I've had that since....11.2 I think it was. I've switched back to 1 monitor these days (of all things, new video card, fan spins quieter with one monitor), but the fix I had was to use software cursors (manually edit xorg.conf for that).
              Section "Device"
              ...
              Option "SWCursor" "true"
              EndSection

              I did that for both monitors, but you should really only need it for the external (or secondary) one.

              Comment


              • #27
                Originally posted by mirv View Post
                I'm sure microsoft would find a way to create a legal issue, or introduce some way of making it incompatible. The directX stuff is pretty tightly integrated into windows too, and I did read that not all of it is documented, so I'm not convinced personally of any viability of a Gallium3D state tracker for DX10/11 beyond "ok, cool, it can do that". That's more the fault of microsoft than anyone else.
                All of which is moot really - games can be made using OpenGL, and would be if developers saw linux as a money-making market. Time will tell I guess!
                Well, you see, not only "geeks" are interested in projects like wine or mesa. Wine d3dx9 code is used in VMware and VirtualBox and mesa code is directly copyrighted by VMware now. So it wouldn't be so easy for MS to legally block further development of DX-compatible API (anyway, mesa is "library with API very similar to OpenGL", nothing prevents it to be "library with APIs very similar to OpenGL and DirectX"). Of course, they may do some tricky "improvements" and obfuscate their documentation even more, but that will only slowdown development a bit (wine developers are working under non-documented APIs all the time, so it's possible too).

                Wine state on DX9 is very robust by now, AFAIK. It misses some rare features and rare formats. Antialiasing not working, for example. Maybe some performance issues with shaders and so on are still open, but this is just a matter of time and polishing. More problems are coming from input subsystem and New-Cool-MS-Technologies? like .NET. If you're Nvidia happy user, you may be sure for 90% that your DX9 game will have good graphics, if it will work at all. And if it's not working at all, it means that some other system failed (DRM, or .NET client, or whatever).

                But existence of this problems doesn't help AMD hardware owners, when they see that NVidia users got everything working, but they are still waiting for new fglrx release..

                Comment


                • #28
                  Originally posted by mirv View Post
                  Ah, that issue of a screwy cursor still around is it? I've had that since....11.2 I think it was. I've switched back to 1 monitor these days (of all things, new video card, fan spins quieter with one monitor), but the fix I had was to use software cursors (manually edit xorg.conf for that).
                  Section "Device"
                  ...
                  Option "SWCursor" "true"
                  EndSection

                  I did that for both monitors, but you should really only need it for the external (or secondary) one.
                  Thank you mirv!
                  I can now see the cursor on the external screen , fingers crossed for it to run forever and ever :/

                  Comment


                  • #29
                    Originally posted by Night Nord View Post
                    I personally doubt such conspiracy theories like "Microsoft paying AMD to make bad drivers for Linux to make everyone go to Windows". On Microsoft's place I will try to pay AMD to make bad drivers for Mac OS X, which is much more dangerous for them on desktop market. I may even believe in some theory that Microsoft founded recent attack on PSN - that will make some sense, as this improved Microsoft's position on consoles market.

                    But Linux... Well, it would be loss of money, by now, IMO.
                    Actually that was intended to be a joke, but while we're on the topic I think Linux would be a bigger threat since hardcore gamers wouldn't buy an Apple computer for games but might well install some Linux distro in a dual boot configuration.

                    And AMD has a lot to gain from supporting DirectX on Linux since their customers will have more money to spend on graphics cards if they don't have to buy Windows.

                    Comment


                    • #30
                      This is amazing, I've had lock up problems since I bought this laptop and now with software cursor it is perfect, I've abused it anyway I can and it is rock solid thus far.
                      I'm starting to get my confidence back and starting to trust the thing, it's plain weird that something so "simple" as a mouse cursor is getting in ati's way.

                      Comment

                      Working...
                      X