Announcement

Collapse
No announcement yet.

R600 Open-Source Driver WIth GLSL, OpenGL 2.0

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

  • #46
    Originally posted by barbarbaron View Post
    Sooo.. just to sum it up: R600+ can choose between radeon and fglrx if they want (useable) 3d accel (--> GLSL) and <=R500 users have to use an old fglrx? Is this correct? *THUMBS UP* to ATi.. definately my last ati/amd board..
    Who will be ahead of who? Aaah yes in the open source stack... We r100-r500 users are VERY familiar with it...
    Barbarbaron, if you are going to quote opinions like that as fact you should really quote the responses as well.

    Otherwise you will just end up misleading people and making yourself unhappy in the process.

    If you own 3xx-5xx hardware (you have X1950 ?) you should get familiar with building and running the Gallium3D driver - it's not ready for general use yet but it's making good progress and already has GLSL and GL 2.1 enabled. At minimum you should be monitoring the progress but it wouldn't hurt to try it out periodically. You'll need a new kernel with KMS enabled, don't remember if you are already running KMS.
    Last edited by bridgman; 12-22-2009, 12:01 PM.

    Comment


    • #47
      Originally posted by barbarbaron View Post
      Let me remind you that r600-r700 opengl 2.0 and GLSL support in radeon driver has been pushed by an AMD employee; Richard Li. With the release of Ubuntu 10.04 or F13 we will all see r600-r700 users playing ET Quake Wars on their computers with good performance while r500 performance will remain GARBAGE.
      This makes it clear that AMD urges us users to upgrade to newer GPUs by supporting whatever feature they see fit. Well I will definitely upgrade to a new GPU as you guys suggest, but an NVIDIA ONE.
      You should try researching/googling before you post, it's refreshing. Alex Deucher (agd5f) has been in AMD's employ for about two years now and he has been working on r500 a lot of that time (and still is). Why should the work on r300g stop?

      Apart from that: Of course AMD urges its users to upgrade. They make money that way, as every business does. Try running nvidias drivers for older hardware these days, it sucks (doesn't work with latest X.org or kernels). Why do you think Nouveau is such an interesting option for many users. Intel developers concentrate on newer hardware too (i815 support is more or less broken, and nice stuff is only coming to i965 or later).

      Comment


      • #48
        What this forum need are rules like those on Doom9.

        http://forum.doom9.org/forum-rules.htm

        That way people like barbarbaron will never ruin it for others.

        Comment


        • #49
          He says this shit because he doesn't have an old nVidia GPU himself. I have here a new Intel notebook, an AMD r600 GPU, a geforce 4 and a geforce 6. My experience so far have been great with Intel GPU. Decent with AMD. GARBAGE with nVidia. By GARBAGE I mean, not even compiz working! And with nVidia you don't have any open-source alternative and no perspective in the future. That's how they respect their customers from 3+ years back.

          Compared to Intel, AMD came late in the OSS. But don't you dare to compare them to **** nVidia.

          Comment


          • #50
            Originally posted by barbarbaron View Post
            This makes it clear that AMD urges us users to upgrade to newer GPUs by supporting whatever feature they see fit. Well I will definitely upgrade to a new GPU as you guys suggest, but an NVIDIA ONE.
            I know exactly how you feel. The other day I bought myself an ATI Rage 128 at the flea market for 1 dollar. Turns out it can't even run Unigine Tropics Demo . I mean it's nice that all this work is going on newer cards, but what about me? I also want to play the latest games with my ATI Rage 128.

            Does ATI really care about owners of ATI Rage 128 cards? I think not. Next time I will spend my 1 dollar wisely on an NVidia card. That will teach them!

            Comment


            • #51
              Originally posted by jntesteves View Post
              Compared to Intel, AMD came late in the OSS.
              Not exactly true. In the r1xx-r4xx days (and even mach32, mach64, and rage 128 before that) ATI was probably the best supported display hardware in OSS. ATI actively contributed code and documentation to developers at the time while intel support was pretty limited. ATI OSS support faded during the r3xx-r5xx 3D period due to a lack of resources internally (which was when intel support started to pick up) and then picked up again after the AMD acquisition.

              Comment


              • #52
                Apart from that: Of course AMD urges its users to upgrade. They make money that way, as every business does.
                The civilized way to do that is AFTER giving the user proper support, convincing the user about the advantages of the new hardware, not like this.

                And about nvidia not supporting old hardware take a look at ati. My gpu (X1950XT) is built in late 2006 and ati dropped support for it in early 2009. Same class hardware (geforce 7 series) are still perfectly supported in their drivers and they have proper VDPAU video acceleration too.

                @bridgman: Thanks for your suggestion to test gallium3d drivers but as Edgar Allan Poe once said: "never more". Gallium3d has a long way ahead of it to be in mainstream usage.

                And I should add that I have no objection to the work done by people on the free drivers and I appreciate it much.
                Last edited by barbarbaron; 12-22-2009, 02:28 PM.

                Comment


                • #53
                  Originally posted by barbarbaron View Post
                  ...GARBAGE...
                  Um, yeah. Okay. Then just go and buy your nvidia card. Be happy with it and then please just leave us alone.
                  I have a comparison of nvidia cards, ATI chips and cards as well as via chips. And I must say: Nvidia is no better in anything. They only really care for their binary blob and this blob knows the word "legacy" very well. One is completely at their mercy.
                  Noveau team is reverse engineering for maybe 5 to 7 years now and it's just getting aflot over the water surface and not drowning.
                  I heard of noveau when I had a nv20 as my main chip. That one is sooo legacy now. Still it is not supported.
                  With VIA it is a horror most times, you can be happy to have modesetting and basic 2d acceleration.
                  With AMD-ATI I'm best off at the moment. There is a partially good partially mediocre fglrx and we have 2 free drivers. And I can see and experience the progress in them. And AMD is giving specs so there is a real chance for anything to happen. Older chip, recent chips, all of them.

                  I do not see any "proper support" for my nv11 and nv20 anywhere. And no, I am not convinced by empty shells that are presented at booths and I cannot be convinced with Wattage monsters that need their own fission plant and cooling towers.
                  Nvidia may have a decent binary only driver for their current stuff but that's it.

                  If one want's to have something on the long term support side then it is neccessary to have specs or a completely working free driver. And I only see this happening on intel's (let out poulsbo) and AMD-ATI's side.

                  Comment


                  • #54
                    Originally posted by barbarbaron View Post
                    And about nvidia not supporting old hardware take a look at ati. My gpu (X1950XT) is built in late 2006 and ati dropped support for it in early 2009.
                    And the Card stopped sorking? My Radeon 9500 (r300) work without problems on my Ubuntu 8.10 with the fglrx.

                    Originally posted by barbarbaron View Post
                    Same class hardware (geforce 7 series) are still perfectly supported in their drivers and they have proper VDPAU video acceleration too.
                    Geforce 7 has no VDPAU support.

                    Comment


                    • #55
                      to me it simply seems as though the honeymoon period with the open source drivers is gradually coming to an end, as more and more people forget about the caveats of fglrx and keep running into the limitations of the radeon drivers.

                      i have had my laptop for about 2 years now, and it's still quite capable in my opinion, and i don't think my budget will allow me to get a new one and get my money's worth for perhaps another 2 years. i think my r500 card is more than enough for most of my needs at the moment, however i feel cheated to not be able to use it to it's total potential.

                      i don't think anyone is really impressed anymore by the open source driver's hability to run compiz or do accelerated video .. being able to use our hardware up to it's specificated capacities really isn't too much to ask for.

                      so anyway, keep up the good work, but never forget that actual people paid actual money for your hardware.

                      Comment


                      • #56
                        Originally posted by pedepy View Post
                        to me it simply seems as though the honeymoon period with the open source drivers is gradually coming to an end, as more and more people forget about the caveats of fglrx and keep running into the limitations of the radeon drivers.
                        Sure, but (a) those limitations are continuing to be reduced/removed, and (b) if you are one of the users who needs all the capabilties of fglrx then you probably should still be using fglrx and a supported distro release. I know it seems like a lifetime but we only dropped support for 3xx-5xx from Catalyst 8 months ago.

                        Originally posted by pedepy View Post
                        i have had my laptop for about 2 years now, and it's still quite capable in my opinion, and i don't think my budget will allow me to get a new one and get my money's worth for perhaps another 2 years. i think my r500 card is more than enough for most of my needs at the moment, however i feel cheated to not be able to use it to it's total potential.
                        With respect, we only stopped adding support for new OS releases; it's not like we drove to your house and took your working driver away.

                        Originally posted by pedepy View Post
                        i don't think anyone is really impressed anymore by the open source driver's hability to run compiz or do accelerated video ..
                        Nor should they be - that was the big deal for 2008. In 2009 the drivers picked up kernel modesetting, GEM/TTM memory management, GL 1.5, and a lot of progress on a Gallium3D-based driver with higher levels of GL support. I know that the architectural rework required for higher levels of GL support and overall user experience doesn't give you the kind of immediate feel-good that adding a new feature to an existing code base would, but for the features you want this work has to be done first.

                        Originally posted by pedepy View Post
                        being able to use our hardware up to it's specificated capacities really isn't too much to ask for.
                        The "specified capabilities" are only specified for Windows (read the box), so you can do that today.

                        What I think you're asking for is "the ability to use your hardware up to its specified capabilities on a different operating system from what the specs say, and not just any version of that operating system but any new distro version of your choice at any time in the future", which is a somewhat bigger thing to ask for.

                        If you make heavy use of 3D then today you should be using fglrx and a supported distro version such as Ubuntu 8.10. Going forward, the open source drivers are making huge progress and should be able to offer a similar experience on newer distro versions -- better in some ways, not quite as good in others.
                        Last edited by bridgman; 12-22-2009, 03:42 PM.

                        Comment


                        • #57
                          btw, Bridgman, Alex, do you have plans for improving performance of OSS driver?
                          cuz it's not great right now.

                          maybe after finishing gl2.0 support(and fixnig bugs) ?

                          or only after moving to gallium3d driver ?

                          Comment


                          • #58
                            I really think you should be more concerned about the possibility that AMD will announce in the not too distant future that they've done what they promised...delivered as much documentation as they think they get away with and still be clean with regard to their other obligations, delivered a working example of a more or less fully functioned driver for each of their reasonably current hardware families, and are now turning over responsibility for the continuing development of the open source driver to the "community" so they can redirect their resources to development of the workstation market, where they actually make some money.

                            Be afraid. Be very afraid.
                            Last edited by rbmorse; 12-22-2009, 08:08 PM.

                            Comment


                            • #59
                              Originally posted by netkas View Post
                              btw, Bridgman, Alex, do you have plans for improving performance of OSS driver?
                              cuz it's not great right now.

                              maybe after finishing gl2.0 support(and fixnig bugs) ?

                              or only after moving to gallium3d driver ?
                              why not after moving to Gallium3D ??

                              Old mesa style only brings 50% of the fglrx speed.
                              the opensource driver needs newstyle.....
                              Galium3D+OpenGl3stage+LLVM compiler brings full speed!
                              without galium3D+LLVM you'll never get the full speed.


                              exampel:

                              your game is programmed for 320SPs witout LLVM
                              your HD5870 only use 320SPs and only clock/ram speed brings fps...
                              LLVM can recompile the 1SPs openGL code split into 2 or 4 SPs code....
                              after that you can use 1600SPs much more speed.

                              just a simpel exampel.

                              http://en.wikipedia.org/wiki/Low_Level_Virtual_Machine


                              http://en.wikipedia.org/wiki/Gallium3D

                              "Gallium3D represents each shader program using an extensible binary intermediate representation called Tungsten Graphics Shader Infrastructure (TGSI). When Gallium targets LLVM the TGSI code is converted to the LLVM instruction set."

                              Comment


                              • #60
                                Originally posted by netkas View Post
                                btw, Bridgman, Alex, do you have plans for improving performance of OSS driver? cuz it's not great right now.

                                maybe after finishing gl2.0 support(and fixnig bugs) ? or only after moving to gallium3d driver ?
                                I don't think there are any specific plans to improve performance but that's primarily because the focus right now is entirely on adding functionality and completing all of the architectural rework. The DRI2 stack is not real fast right now, for a variety of reasons -- a brand new, unoptimized memory manager, a new lockless DRI protocol whose first implementation involves sending a lot of redundant state, and probably a few other things along the way.

                                Best guess is that performance work will happen after moving to Gallium3D, at least for 3xx-5xx -- partly because it makes more sense to look at the final implementation, partly because the workload running GL 2.x apps is likely to be different from running older apps, and partly because there are only so many things the dev community can work on at the same time.

                                It's probably safe to say that any performance work will be done on the DRI2 paths, not DRI1, and that the first step will be figuring out where the time is going

                                Comment

                                Working...
                                X