Announcement

Collapse
No announcement yet.

Open-Source R500 Driver Can Compete With Legacy Catalyst

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

  • #31
    Originally posted by YaPeL View Post
    Instead of fighting with Q you could share some info, like, future plans for r500, power managment, 2d acceleration, video decoding etc etc.
    Done multiple times already, but here's a summary :

    Don't think we have any specific plans for r500. The reported power management issues seem to be largely on newer hardware so far.

    We are working on improvements to power management but hard to give ETAs because it depends on what info we end up being able to release and when.

    No specific plans for 2D acceleration AFAIK but periodically new DEs and apps use the accel in new ways and bottlenecks occur, and in those cases we usually try to pitch in and help speed up whatever has become a problem. Are you talking specifically about r3xx-5xx (which has dedicated 2D acceleration hardware) or r6xx and higher (which does not, and therefore uses the 3D engine for everything) ?

    Video decoding has been discussed at length in the last week or so -- Christian has been working on it or a while, but that work is speculative in the sense that we don't know if we will be able to release it. The day we know for sure, you will know as well.
    Last edited by bridgman; 06-08-2012, 01:21 PM.

    Comment


    • #32
      Originally posted by bridgman View Post
      Originally posted by Qaridarium
      irrelevent blather about what OS our admin systems use and the fact that I have a BluRay player in my home system
      Best post summary ever.

      Comment


      • #33
        Originally posted by GreatEmerald View Post
        Best post summary ever.
        sure its questionable but i think this is interesting because this proved the opinion that open-source operating systems "Linux" is not for watching movies ---> because all people should buy a blueray player if they want to watch movies.

        For me this is not acceptable.

        or is this the new universal answer if someone ask about video acceleration ? buy a blueray player and buy your videos in the store ?

        Comment


        • #34
          Originally posted by bridgman View Post
          No specific plans for 2D acceleration AFAIK but periodically new DEs and apps use the accel in new ways and bottlenecks occur, and in those cases we usually try to pitch in and help speed up whatever has become a problem. Are you talking specifically about r3xx-5xx (which has dedicated 2D acceleration hardware) or r6xx and higher (which does not, and therefore uses the 3D engine for everything) ?
          Has any further thought gone into 2D for SI?

          Will AMD be working on improving the xa state tracker, or going to some other solution?

          Comment


          • #35
            Some discussions, but I think the overall plan is more like :

            1. Figure out why the 3D is still hanging
            2. Celebrate; take the rest of the weekend off
            3. Look at 2D some more

            Comment


            • #36
              Originally posted by bridgman View Post
              Some discussions, but I think the overall plan is more like :

              1. Figure out why the 3D is still hanging
              2. Celebrate; take the rest of the weekend off
              3. Look at 2D some more
              a good plan

              maybe i wana buy a hd7750 passiv cooled for my system to replace the hd4770
              but i really need the opensource driver even if i use the catalyst i need the opensource driver as a fallback.

              right now I'm unable to install the catalyst at all... so i really need the opensource driver.

              a card like this:

              Comment


              • #37
                Fantastic stuff.

                This same sort of thing happened with the R200 generation of cards. The OSS driver eventually got to the point were it was on the same performance level and had much higher levels of compatibility then the original hacked proprietary drivers that people managed to get running in LInux.

                It was to the point were some hard-core people still preferred R200 series cards up until a couple years ago when it became impossible to get new hardware with AGP slots.

                It seemed certain alliances with Microsoft over Xbox 360 and the rise of the popularity of DRM systems really put a kabosh on any sort of proper Linux support by ATI.

                The sign that Linux and X devs finally got R500 series caught up is a fantastic thing.


                Oh and Q,
                I hear that http://en.wikipedia.org/wiki/Lithium_pharmacology is helpful.
                you may have nice or insightful things to say, but it all lost when you act like a maniac in every single post.

                Comment


                • #38
                  Reality Check

                  Q is certainly a bit hard to understand at times.

                  But one thing is for sure, AMD and their apologists are just clowns.

                  Every SINGLE thing I ever said about the radeon driver has come to pass. It is going backwards. And no amount of lies and stupidity will ever fix it.

                  Comment


                  • #39
                    hey bridgeman, can you tell the hardware designers and whoever is responsible for designing the end user video cards to stop taking up 2 whole expansion slots. like, i could understand for the dual chips cards and maybe even the top of the line flagship cards but cmon, even mid grade these days are dual slot. i use nvidia parts so maybe its just worse with them, but i remember when i did my first computer build and i put a geforce 6600gt in it and it was sigle slot, that was a upper mid range card then, now, even lower mid range cards are 2 slots, wtf.

                    now, why even design a motherboard with the pcie1x slots under and 16x slot since anyone using a pcie slot for a video card is prolly gunna stick a mid to high end card in there since the low end cards are outpaced by the cpu+gpu combos that are now manditory since nobody wants to make just a cpu anymore.

                    oh, and whats your views on openMAX as a cross platform video acceleration api and will anyone ever care to use it?

                    Comment


                    • #40
                      Originally posted by gordboy View Post
                      Every SINGLE thing I ever said about the radeon driver has come to pass. It is going backwards. And no amount of lies and stupidity will ever fix it.
                      I'm in a weird mood right now, so I'll indulge in some troll-feeding and let you humor me. So here we go:

                      Show me your posts where you were saying this stuff about radeon, then show me how exactly they came to pass. Then point out the supposed "lies". Come on, let me see it. Humor me.
                      Last edited by Gusar; 06-08-2012, 07:46 PM.

                      Comment


                      • #41
                        Originally posted by benjamin545 View Post
                        hey bridgeman, can you tell the hardware designers and whoever is responsible for designing the end user video cards to stop taking up 2 whole expansion slots.
                        If you don't want the GPU to fry itself, the cooling solution has to be dual-slot. No other way around it. GPUs nowadays have more transistors than CPUs, and look a the coolers needed to prevent those from frying themselves.

                        Originally posted by benjamin545 View Post
                        oh, and whats your views on openMAX as a cross platform video acceleration api and will anyone ever care to use it?
                        According to daniels, openmax is a terrible API: http://phoronix.com/forums/showthrea...191#post266191
                        So I'd rather we just stick to what's already available and working on the desktop: vdpau and vaapi.

                        Comment


                        • #42
                          maybe GPU's are too powerfull nowadays?
                          actualy thats really proboobly true, i bought the geforce 6600gt back then for under 200 usd, but now a geforce gtx 660 will proboboly cost 300-350usd. its kinda like naming of the card doesnt realy mean how poerfull the card is in comparison to demand on the card, but only really puts it in a position reletive to other cards in the family. back then, a 6600gt could be maxed out on then current games. i think when the 660 comes out you will be able to really max it out unless you put some sort of huge display size requirments on it or hinder it with something else using it in the background.

                          and as far as openMAX..... openMAX 2.0??? maybe to fix the problems??? there is 2 parts to openmax, theres a low level api and a high level api and the only one that is comparable to va-api and the like is the low level one, i think the high level one is like a competator to gstreamer. i dunno, but most people would agree its not idea to havea windows video api and a mac video api (which is hidden and only availible internaly for apple from what i remember) and 3-4 linux api's with an adnroid api and whatever consols use and whatnot.

                          Comment


                          • #43
                            Originally posted by benjamin545 View Post
                            and as far as openMAX..... openMAX 2.0??? maybe to fix the problems??? there is 2 parts to openmax, theres a low level api and a high level api and the only one that is comparable to va-api and the like is the low level one, i think the high level one is like a competator to gstreamer. i dunno, but most people would agree its not idea to havea windows video api and a mac video api (which is hidden and only availible internaly for apple from what i remember) and 3-4 linux api's with an adnroid api and whatever consols use and whatnot.
                            There are only 2 linux API's that matter:

                            VAAPI and VDPAU.

                            I don't see anyone ever switching away from those APIs. And really you should only have to support one, because each has a backend outputting into the other one.

                            Comment


                            • #44
                              Originally posted by benjamin545 View Post
                              maybe GPU's are too powerfull nowadays?
                              actualy thats really proboobly true
                              , i bought the geforce 6600gt back then for under 200 usd, but now a geforce gtx 660 will proboboly cost 300-350usd. its kinda like naming of the card doesnt realy mean how poerfull the card is in comparison to demand on the card, but only really puts it in a position reletive to other cards in the family. back then, a 6600gt could be maxed out on then current games. i think when the 660 comes out you will be able to really max it out unless you put some sort of huge display size requirments on it or hinder it with something else using it in the background.

                              and as far as openMAX..... openMAX 2.0??? maybe to fix the problems??? there is 2 parts to openmax, theres a low level api and a high level api and the only one that is comparable to va-api and the like is the low level one, i think the high level one is like a competator to gstreamer. i dunno, but most people would agree its not idea to havea windows video api and a mac video api (which is hidden and only availible internaly for apple from what i remember) and 3-4 linux api's with an adnroid api and whatever consols use and whatnot.
                              regarding "GPU's are too powerfull nowadays?" , not in the slightest even if your just referring to the hardware and not the substandard under performing Software components etc, first acknowledge the problems then address them with real working optimized code (it seems to be ARM and Linaro's stance today).

                              in fact regarding "openMAX" if you want to see how the Pro ARM Linaro collective companies try to deal with these types of problems just read this generic outline and realize they put time limits on their ARM time to market working sample code usually around 3 months max, no x86 vendor seems wiling to meet that these days for a substantial project like this for instance "OpenMax Integration" on ARM (cortex)
                              https://wiki.linaro.org/Events/2011-...MaxIntegration

                              its quite refreshing to see a whole ARM collective industry come together and openly show anyone willing to read and contribute how to make real progress and advance the core upstream Linux code-base as a side benefit to all longer term.
                              Last edited by popper; 06-08-2012, 10:14 PM.

                              Comment


                              • #45
                                Originally posted by oliver View Post
                                bridgeman is my hero

                                I stillcan't belive people bashing AMD for the huge amounts of effort they put in. Sure, more is always better and I'm sure peopl ealways argue there's more needed. All that said and done, it's remarkable how much AMD _does_ do. Remember, AMD is also the saller 'underdog'. It doesn't have the intel budgets for things. It is still for lack of better word, struggling to keepalive.

                                So Bridgeman, the messenger, is my hero.
                                This +1.

                                Comment

                                Working...
                                X