Announcement

Collapse
No announcement yet.

Future of ATI Linux drivers

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

  • Originally posted by CNCFarraday View Post
    Cheese Man, look you got it all wrong. I mean, literally.

    You didn't understand anything I said. First, some of the 'offensive' stuff wasn't aimed at you, but at the other guy for whom changing access rights on a binary is a 'hack'. Maybe I didn't make myself clear, maybe you got the Angy Internet Man syndrome after reading the first sentences and started posting without really reading what I said.

    The idea is you are totally off the road. For example, Mirv got it: http://phoronix.com/forums/showpost....&postcount=142
    Well, what I have ment to say in the first post is:
    - WINE is not improving linux gaming stack. It is not native and you need to buy windows games and you need nvidia card, because nvidia uses windows drivers on linux.

    - I dont understand the purpose of FGLRX and the whole DRM stuff. If someone likes to watch encrypted DVD/Bluray junk I would understand it. DRM will always be cracked, it highers the production costs and prevents writing software - it is garbage. But I dont buy hollywood movies anyway: they are ugly and with same plot(have a picture of how to "make a hollywood scenario in 1 minute" schema somewhere); all movies that I watched were in theaters, since they are almost all "disposable", like cigarettes or LAW(one use - throw away). Maybe District 9 was single exception from this.

    And if AMD wants to protect their intellectual property, why not to use GPL on drivers? Samsung is already sued for silently using busybox. AMD, look how Toshiba does with its TVs!

    - I have no problem with choices, but the only graphics accelerator manufacturer (of the two), who supports opensource and creates native stack(instead of porting stuff back from other OSes) is AMD. You might work as a software engineer and make money this way; but I see no problem for some serious GAMING shift to linux. Even if I dont game often. Creating games is a job too and being in need to buy windows license every 4 years along with antivirus and support company which HATES opensource, just to play 2-3 games 3-4 times in a month - can see some improvent imho. Currently linux users have to keep and pay windows which uses this money to establish itself over linux. This is wrong.

    Hope you can understand my point without spilling flames, like a dragon. Peace out.

    Comment


    • The need for proprietary drivers from AMD (and nVidia for that matter) stem not from AMD/nVidia themselves, but rather third parties whose technology is licensed. Take s3tc for example - still proprietary, but used everywhere these days. Kinda sucks, but that's the way it is.

      Comment


      • Originally posted by crazycheese View Post
        I dont understand the purpose of FGLRX and the whole DRM stuff.
        There is no DRM in fglrx, although to the extent that we share code with other OSes there are unused elements in data structures and unused code segments related to DRM. Remember that our concerns are not related to protecting DRM in the Linux drivers, we're just trying to make sure we don't do anything with Linux drivers that weakens DRM on other OSes (since the hardware and a good chunk of the code is common).

        Originally posted by crazycheese View Post
        And if AMD wants to protect their intellectual property, why not to use GPL on drivers? Samsung is already sued for silently using busybox. AMD, look how Toshiba does with its TVs!
        GPL only protects the actual implementation of the driver code, not the related IP. It doesn't stop someone from using what they learn from a GPL Linux driver to attack the Windows drivers, for example.

        Comment


        • Don't mess with brigdman and the opensource people!
          RMS will slapp you around bad boy bad boy!
          Hehehe You can't understand anything straight? I'm talking about ATI closed source drivers. I'm using intel's opensource stack on my laptop anyway and I'm quite happy with it. See my post here ATI fanboy fanboy:

          http://phoronix.com/forums/showthrea...t=23068&page=2

          Comment


          • WINE is not improving linux gaming stack. It is not native and you need to buy windows games and you need nvidia card, because nvidia uses windows drivers on linux.
            That is exactly my position too. WINE is only making the ecosystem worse.

            The purpose of FGLRX is to provide a driver for linux. The ugly truth is that the entire 'multimedia' stack in Linux is badly designed. Eveything that we have today was made by people and organizations that needed a framebuffer to blit pixels into. You need an overall architecture built into the operating system for proper graphics, especially for gaming, multi-monitor output, stereo-3D, video decoding and the like. Part of the reason ATI and nVidia buildtheir drivers the way they are is because there isn't much 'underneath' to support proper graphics/video/3d handling. Also the Linux community cannot dictate terms to graphics hardware manufacturers, like M$ does.

            No matter how much M$ stuff sucks and no matter how closed it is, it has one rewarding merit for game developers: there is ONE API. It is insanely well documented, it provides everything you need to get up and going and all the extra goodies. Game developers want to make GAMES not to see which version of library x, when used with static library y which was compiled with gcc version z doesn't work and crashes. People with the talent and imagination to create good games, in the gameplay/artistic sense, utterly lack the technical knowledge to develop an engine and the tools and to tweak and debug it on the linux ecosystem.

            Look, I love Linux and all it represents, I use it daily at home and at work and I've been using it for >10 years (prior to that i used UNIX :P).

            However, I do understand that the idea of Linux and FOSS come at a price. The ecosystem is VERY fragmented, heterogeneous and dynamic. This has many good parts, but also some bad parts, the worst being that Big Business doesn't like it for a myriad of reasons that I can lay out if you want. Developing a commercial product 'for the masses' (not for a niche where you can afford or circumvent the diversity wildcard) for linux, especially something like a game - an entertainment product that a kid is suppose to pop into the hardware and play with no fuss - is very hard to do, even harder to support and worse to justify to the people with money.

            If the community really wants to boost Linux gaming it should make something comparable to Flash ( only in capabilities, not in performance ). A complete solution for people with poor technical skills but great artistic ones and lots of imagination to make games. The most imaginative games today are indie games and they thrive on Flash because it offers them a complete solution - development, integration and distribution/availability - to make GAMES. Most of the people who can make the next X-COM, the next Fallout ( not that crap from Bethsada ) the next Half-Life and so on, lack the technical skills to build up the infrastructure (engine, tools, testing and distribution support) to make Linux Games.

            Give them the tools and they will thrive. Let the architect make a building, don't force him to use the hammer-o-lamp-o-sprocket. Because most of what Linux is, including the foss ecosystem was and still is made by professionals and 'hobbyists' it is very contrived and limiting for people not having and not wanting and not needing the skills. Most of the concepts and meta-concepts of the Linux world are from the singular point of view of the computer science / programming perspective.

            Note that I am definitely NOT arguing for 'dumbing down' Linux. I am arguing FOR better tools, designed not only by programming gurus, but also by specialists in other domain fields and for horizontal integration of the various mega-projects of the FOSS world.

            I'd like to see something like Flash or JAVA FX ( yeah poor choice, but still ) made 100% FOSS based on GTK, Cairo for graphics and Mozilla-tech tools for clients ( gui tools, editors etc ), and Python for scripting.

            Comment


            • Originally posted by CNCFarraday View Post
              That is exactly my position too. WINE is only making the ecosystem worse.

              The purpose of FGLRX is to provide a driver for linux. The ugly truth is that the entire 'multimedia' stack in Linux is badly designed. Eveything that we have today was made by people and organizations that needed a framebuffer to blit pixels into. You need an overall architecture built into the operating system for proper graphics, especially for gaming, multi-monitor output, stereo-3D, video decoding and the like. Part of the reason ATI and nVidia buildtheir drivers the way they are is because there isn't much 'underneath' to support proper graphics/video/3d handling. Also the Linux community cannot dictate terms to graphics hardware manufacturers, like M$ does.

              No matter how much M$ stuff sucks and no matter how closed it is, it has one rewarding merit for game developers: there is ONE API. It is insanely well documented, it provides everything you need to get up and going and all the extra goodies. Game developers want to make GAMES not to see which version of library x, when used with static library y which was compiled with gcc version z doesn't work and crashes. People with the talent and imagination to create good games, in the gameplay/artistic sense, utterly lack the technical knowledge to develop an engine and the tools and to tweak and debug it on the linux ecosystem.

              Look, I love Linux and all it represents, I use it daily at home and at work and I've been using it for >10 years (prior to that i used UNIX :P).

              However, I do understand that the idea of Linux and FOSS come at a price. The ecosystem is VERY fragmented, heterogeneous and dynamic. This has many good parts, but also some bad parts, the worst being that Big Business doesn't like it for a myriad of reasons that I can lay out if you want. Developing a commercial product 'for the masses' (not for a niche where you can afford or circumvent the diversity wildcard) for linux, especially something like a game - an entertainment product that a kid is suppose to pop into the hardware and play with no fuss - is very hard to do, even harder to support and worse to justify to the people with money.

              If the community really wants to boost Linux gaming it should make something comparable to Flash ( only in capabilities, not in performance ). A complete solution for people with poor technical skills but great artistic ones and lots of imagination to make games. The most imaginative games today are indie games and they thrive on Flash because it offers them a complete solution - development, integration and distribution/availability - to make GAMES. Most of the people who can make the next X-COM, the next Fallout ( not that crap from Bethsada ) the next Half-Life and so on, lack the technical skills to build up the infrastructure (engine, tools, testing and distribution support) to make Linux Games.

              Give them the tools and they will thrive. Let the architect make a building, don't force him to use the hammer-o-lamp-o-sprocket. Because most of what Linux is, including the foss ecosystem was and still is made by professionals and 'hobbyists' it is very contrived and limiting for people not having and not wanting and not needing the skills. Most of the concepts and meta-concepts of the Linux world are from the singular point of view of the computer science / programming perspective.

              Note that I am definitely NOT arguing for 'dumbing down' Linux. I am arguing FOR better tools, designed not only by programming gurus, but also by specialists in other domain fields and for horizontal integration of the various mega-projects of the FOSS world.

              I'd like to see something like Flash or JAVA FX ( yeah poor choice, but still ) made 100% FOSS based on GTK, Cairo for graphics and Mozilla-tech tools for clients ( gui tools, editors etc ), and Python for scripting.
              Let me tell you what we have at present:
              OpenGL - well, the only bad side is speed of improvement, since all members have to accept the changes. But the payback is thats acknowledged by all. And documentation. Although I dont really think documentation is that bad. It is open, consistent, independant to platform, hardware and OS. It is also easier to program, because its a system, rather than an interface. But I want to ask you to leave it as is, because it is the only and the best counterpart to Direct3D. Unless MS desides to open DX(which it will never will). Ask yourself why does MS keep ignoring OpenGL to large extend and instead puts money in DX, not even a fork, but complete different system. Its called lobbing.
              Gstreamer - an equivalent of DirectShow, makes installing and using codecs a fuzz. Play, record stuff easy. Very mature.
              Bullet - the open physics engine is present.
              Network stack is one of the best in the world.
              OpenCL - is here already.
              Udev and evdev - are here for input hotplug detection and xorg supports various input devices.
              SDL - for writing crossplatform code, including engines. Takes care of dependencies to OpenGL and sound systems well(OpenAL,Gstreamer,Alsa,Oss).
              HTML5 is coming, willing to replace flash(as much as adobe fears it). I think it is the right way. Flash-like technology should really be open, because its enriches www to great extent. Adobe should at least open the specs and provide own implementation if they want. Flash-nonfree is present in repos as well as gnash and swfdec. But HTML5 is really the correct way.
              I really dont know with java, maybe oracle will give some chance..


              There are several great opensource engines and even game-makers present for linux.

              Compiling against next version isnt bad, if you know what "DLL hell" is. Additionally you dont have to recompile every time, you just call for somelib.so, which is a link to somelib.so.4 or so.6 and so on.
              It isnt a issue for opensource projects, proprietary have their own ways(why protect the code and not the title(the purchase) in the first place?) Im not the fan of proprietary anyway, because chinese earn somehow different amount as swiss. And security is crap. And others cannot contribute to the same extend as vs GPL. Forcing to pay is not a right option, making people invest for the options they want is much better IMHO(in the case they aint programmers and dont want to implement them themselves). But thats offtopic. UrbanTerror for example uses a precompiled binary that run (for me) on several different Ubuntu versions(8.04-10.04) and both x86/x86_64 without issues.

              To my knowledge only graphic pipeline is a problem.
              Currently the most complete OpenGL stack is provided by nvidia(which is the only single positive point in their drivers, sadly).
              But because its restricted, barely any distribution ships with it, even gentoo has it hard and keyword masked.
              If AMD provides OpenGL 3.2 opensource stack with at least 50% of closed source driver perfomance together with optional videoacceleration, that case would be closed.


              And I share your points on XCOM and Fallout, XCOM1 has is really in need for soldier stat overfollow patch(ie TU 254>>255>>0). But no one can patch it since its closed and license is hold by Take2(to my knowledge). And someone should really be both skilled at programming as well as overall logic and be able to keep whole game consequent to write something like XCOM. This is HARD even if you(as in "any man") have rights on title. Technology develops, people knowledge regresses.

              Comment


              • Ubuntu unrolls very easy. It was my first linux distro, installing it (8.04) was like fuzz compared to XP. I really liked linux since then. But then come the closed VGA drivers. "Do you want to install them?" This is the single disaster point, and linux IS the OS with opensource spirit. If this flattens out - installing system, installing games will be very very easy (if staying focused only on gaming).

                Comment


                • Originally posted by barbarbaron View Post
                  Hehehe You can't understand anything straight? I'm talking about ATI closed source drivers. I'm using intel's opensource stack on my laptop anyway and I'm quite happy with it. See my post here ATI fanboy fanboy:

                  http://phoronix.com/forums/showthrea...t=23068&page=2
                  Intel lacks 3D acceleration, in terms of hardware power. They are OK for film playback and 2D and the show stops here. And, to my knowledge, their input to implementing opensource stack is way lower than AMDs. And they cost much more than AMD, esp. motherboard(chipset?) prices.

                  Comment


                  • In the interests of fairness I have to point out that Intel's open source graphics team is actually larger than ours, and includes developers who have made significant contributions to the open source graphics stack (outside of just the HW driver) both before and after joining Intel.

                    Thptth ! I hate having to say things like that

                    Comment


                    • @bridgman: that's why you're in management, and not PR

                      Comment


                      • Originally posted by bridgman View Post
                        In the interests of fairness I have to point out that Intel's open source graphics team is actually larger than ours, and includes developers who have made significant contributions to the open source graphics stack (outside of just the HW driver) both before and after joining Intel.

                        Thptth ! I hate having to say things like that
                        Didn't know it, but still. I had E5300 on uATX Asrock P43ME for half-a-year as my main pc at home; whilst also built similar system on base of Athlon II x2 240 with uATX Asus M3N(nvidia 8300 chipset). The reason taking nvidia was 3D performance under linux. I was also forced to buy some 3D card for my P43ME build(which was 9800gt green). Note that this was half-a-year ago, and now watching phoronix and seeing what amd and nvídia do, I have switched 9800 for hd4770 to support opensource. Back on the topic, for same money, Athlon II has better box cooler and was running generally cooler(I had to buy scythe big shuriken to replace box intel cooler(65C for E5300), where athlon II box runs COOL(40C) and QUIET); AII has coolnquiet able to drop to 800Mhz and drop voltage, whilst E5300 can go from 2.6 to 1.6Ghz without voltage drop. Athlon mobo has all-solid-caps for 50€ price, whilst I was unable to find any solid-cap board for under 80€ with ICH10(uses least power). Socket 775 is out of fab, you will have to sell ddr2, mobo, cpu on next upgrade.
                        With athlon, you just sell AM2+ board with ddr2 and you done. Or you can buy CPU and set in into the old board with old ram.

                        So, for me, intel is already outdated in low-price segment, with prices higher for the same build and will force you to sell everything should you deside to upgrade cpu or board.

                        In the middle segment, 1150 boards are too expensive(100€ for all-solid-caps) and entry processor is i3-540, which costs another 100€ and gives 2real and 2 virtual cores. Integrated graphics are still unusable in terms of hardware performance, falling back behind comparable IGP from AMD. 200€ for what?

                        Now for 80€ I get true quad core Athlon II x4, and for 65€ all-solid-cap AM3 board with integrated 4200 IGP and 4x ddr3, firewire.
                        145€ for awesome cfg.

                        This is plain unbeatable, and because of you, guys(AMD) effort in giving opensource 3D stack some love, I went for all-amd including selling my 9800gt green for hd4770.

                        So, for me, if intel starts selling 3D capable hardware along with lowering prices I would consider it as an option(I believe what you say about human resources they put in, Mr.Bridgman).

                        Besides Intel vs AMD boxed CPUs, when looking at booklets show amazing stuff - Intel :" Our Logo is COPYRIGHTED!(on front page)". AMD - "thanks for buying our product, enjoy it".

                        Comment


                        • Originally posted by CNCFarraday View Post
                          I'd like to see something like Flash or JAVA FX ( yeah poor choice, but still ) made 100% FOSS based on GTK, Cairo for graphics and Mozilla-tech tools for clients ( gui tools, editors etc ), and Python for scripting.
                          There are about a hundred game engines that fit this description. Off hand, Delta3d, Panda3d and Blender Game Engine. They give you GUI-based editors, proper Linux support, Python scripting.

                          We are not lacking tools, we are lacking an audience that could make developers' time worthwhile. (We are also lacking in driver stability, but lets ignore that for now). Even with those drawbacks, there's a steady trickle of Linux games being released every year.

                          Maybe with the advent of WebGL the landscape will start to change.

                          Comment


                          • And, to my knowledge, their input to implementing opensource stack is way lower than AMDs.
                            Well yes its true that intel hardware is behind nv or ati in terms of performance, but: Its NOT as expensive as their products and consumes less TDP making battery life longer. And for the opensource support side.

                            1. I can watch HD content without flickering via VA-API while you ATI guys watch crap whether or not using closed or opensource drivers.
                            2. OpenGL extensions supported by the latest intel drivers is much more than other opensource stacks:

                            GL_ARB_copy_buffer, GL_ARB_depth_texture, GL_ARB_depth_clamp,
                            GL_ARB_draw_buffers, GL_ARB_draw_elements_base_vertex,
                            GL_ARB_fragment_coord_conventions, GL_ARB_fragment_program,
                            GL_ARB_fragment_program_shadow, GL_ARB_fragment_shader,
                            GL_ARB_framebuffer_object, GL_ARB_half_float_pixel,
                            GL_ARB_half_float_vertex, GL_ARB_map_buffer_range, GL_ARB_multisample, GL_ARB_multitexture, GL_ARB_occlusion_query, GL_ARB_pixel_buffer_object,
                            GL_ARB_point_parameters, GL_ARB_point_sprite, GL_ARB_provoking_vertex,
                            GL_ARB_seamless_cube_map, GL_ARB_shader_objects,
                            GL_ARB_shading_language_100, GL_ARB_shading_language_120, GL_ARB_shadow,
                            GL_ARB_sync, GL_ARB_texture_border_clamp, GL_ARB_texture_compression,
                            GL_ARB_texture_cube_map, GL_ARB_texture_env_add,
                            GL_ARB_texture_env_combine, GL_ARB_texture_env_crossbar,
                            GL_ARB_texture_env_dot3, GL_ARB_texture_mirrored_repeat,
                            GL_ARB_texture_non_power_of_two, GL_ARB_texture_rectangle,
                            GL_ARB_transpose_matrix, GL_ARB_vertex_array_bgra,
                            GL_ARB_vertex_array_object, GL_ARB_vertex_buffer_object,
                            GL_ARB_vertex_program, GL_ARB_vertex_shader, GL_ARB_window_pos,
                            GL_EXT_abgr, GL_EXT_bgra, GL_EXT_blend_color,
                            GL_EXT_blend_equation_separate, GL_EXT_blend_func_separate,
                            GL_EXT_blend_logic_op, GL_EXT_blend_minmax, GL_EXT_blend_subtract,
                            GL_EXT_cull_vertex, GL_EXT_compiled_vertex_array, GL_EXT_copy_texture,
                            GL_EXT_draw_buffers2, GL_EXT_draw_range_elements, GL_EXT_framebuffer_blit,
                            GL_EXT_framebuffer_object, GL_EXT_fog_coord,
                            GL_EXT_gpu_program_parameters, GL_EXT_multi_draw_arrays,
                            GL_EXT_packed_depth_stencil, GL_EXT_packed_pixels,
                            GL_EXT_pixel_buffer_object, GL_EXT_point_parameters,
                            GL_EXT_polygon_offset, GL_EXT_provoking_vertex, GL_EXT_rescale_normal,
                            GL_EXT_secondary_color, GL_EXT_separate_specular_color,
                            GL_EXT_shadow_funcs, GL_EXT_stencil_two_side, GL_EXT_stencil_wrap,
                            GL_EXT_subtexture, GL_EXT_texture, GL_EXT_texture3D,
                            GL_EXT_texture_compression_s3tc, GL_EXT_texture_cube_map,
                            GL_EXT_texture_edge_clamp, GL_EXT_texture_env_add,
                            GL_EXT_texture_env_combine, GL_EXT_texture_env_dot3,
                            GL_EXT_texture_filter_anisotropic, GL_EXT_texture_lod_bias,
                            GL_EXT_texture_object, GL_EXT_texture_rectangle, GL_EXT_texture_sRGB,
                            GL_EXT_texture_swizzle, GL_EXT_vertex_array, GL_EXT_vertex_array_bgra,
                            GL_3DFX_texture_compression_FXT1, GL_APPLE_client_storage,
                            GL_APPLE_packed_pixels, GL_APPLE_vertex_array_object,
                            GL_APPLE_object_purgeable, GL_ATI_blend_equation_separate,
                            GL_ATI_envmap_bumpmap, GL_ATI_texture_env_combine3,
                            GL_ATI_separate_stencil, GL_IBM_multimode_draw_arrays,
                            GL_IBM_rasterpos_clip, GL_IBM_texture_mirrored_repeat,
                            GL_INGR_blend_func_separate, GL_MESA_pack_invert,
                            GL_MESA_texture_signed_rgba, GL_MESA_ycbcr_texture, GL_MESA_window_pos,
                            GL_NV_blend_square, GL_NV_depth_clamp, GL_NV_light_max_exponent,
                            GL_NV_packed_depth_stencil, GL_NV_texture_env_combine4,
                            GL_NV_texture_rectangle, GL_NV_texgen_reflection, GL_NV_vertex_program,
                            GL_NV_vertex_program1_1, GL_OES_read_format, GL_SGIS_generate_mipmap,
                            GL_SGIS_texture_border_clamp, GL_SGIS_texture_edge_clamp,
                            GL_SGIS_texture_lod, GL_SUN_multi_draw_arrays, GL_OES_EGL_image

                            Just compare it...

                            Comment


                            • Hm, then Intel has to do lots in hardware and AMD lots in software.

                              Comment


                              • Originally posted by BlackStar View Post
                                We are not lacking tools, we are lacking an audience that could make developers' time worthwhile.
                                It's the old question of the cicken and the egg. Back in 1940 the CEO (or something like that) of IBM said that there would be a global market for some 5 computers at most.

                                "If you build it they will come"

                                BTW, the other day I installed kernel 2.6.33 and libdrm and all that stuff that makes 3D acceleration work with the opensource radeon drivers on my RS880, and I finally felt like I actually made a good decision of going the AMD way on my last PC upgrade. Oh, and I'm amazed by the fact that I installed a few libraries that aren't in the main repositories of my distro and still the system didn't break. Things have evolved a lot in the past 4 years.

                                Comment

                                Working...
                                X