Announcement

Collapse
No announcement yet.

support for first generation UVD blocks (RV6xx, RS780, RS880 and RV790)

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

  • support for first generation UVD blocks (RV6xx, RS780, RS880 and RV790)

    Hi,

    According to the following post by Christian König, UVD support is ready and only awaits reviewer's approval.

    http://lists.freedesktop.org/archive...ry/051584.html

    I run Linux on an Asus M4A89GTD PRO/USB3 motherboard (i.e. with AMD 890GX/RS880D chipset, with UVD 2), and I really would like support for first generation UVD blocks to be validated, and go open source.
    This would allow people with no-brand-new AMD chipset like me to take advantage of their hardware.
    But, as Christian's message says, this could happen soon or never.
    Maybe it could help if people show their interest for this support, maybe not.
    Maybe this "possible" new open source UVD support could also be advertised on Phoronix news ;-)

  • #2
    See it as it is: They already have your money, so you are low priority. For them it is more important to work on drivers for newer hardware, so that they can get more money, instead of not pissing of customers they already have (which in turn possibly would buy more AMD hardware if they would be satisfied with AMD). AMD's marketing is something no sane person can understand.

    Comment


    • #3
      There are a number of threads on this already, but each one starts out pretty much the same. Folks are misinterpreting Christian's comment as "yeah, the code is ready but we might never review it for release", but in fact he is talking about two separate issues :

      1. Yes, code is ready for review but not all the code that gets written is ever able to be released.

      2. Reviews are significant efforts, not just checking for rude comments in the source code, and it's not unusual for them to fail. If that happens, the developers take what they learned from the review feedback and try to come up with a different (and hopefully acceptable) solution and another review is organized.

      An acceptable implementation *was* eventually developed for the later UVD hardware and we were able to release the code after a couple of years of work. We don't know yet if we'll be able to do something similar for earlier UVD hardware.

      EDIT -- now the misinterpretation has its own article. Sigh...
      Last edited by bridgman; 01-29-2014, 12:32 AM.

      Comment


      • #4
        Originally posted by bridgman View Post
        There are a number of threads on this already, but each one starts out pretty much the same. Folks are misinterpreting Christian's comment as "yeah, the code is ready but we might never review it for release", but in fact he is talking about two separate issues :

        1. Yes, code is ready for review but not all the code that gets written is ever able to be released.

        2. Reviews are significant efforts, not just checking for rude comments in the source code, and it's not unusual for them to fail. If that happens, the developers take what they learned from the review feedback and try to come up with a different (and hopefully acceptable) solution and another review is organized.

        An acceptable implementation *was* eventually developed for the later UVD hardware and we were able to release the code after a couple of years of work. We don't know yet if we'll be able to do something similar for earlier UVD hardware.

        EDIT -- now the misinterpretation has its own article. Sigh...
        Indeed you have explained that before. But what is not open to misinterpretation is this quote from Christian König:
        It just doesn't has a high priority for the reviewer because we don't really sell that old hardware any more.
        Which is clearly wrong. Newegg alone lists 18 different mainboards with RS780/880 chipsets that are still sold.

        Comment


        • #5
          Not only that, but who is ever going to buy AMD again, if this is the kind of support they can expect? Not me, for a start......

          --
          Pete

          Comment


          • #6
            It should be noted that AMD never said we would support UVD in open source. We said repeatedly that we would look into it, but if it was deemed too risky we may not be able to do it. It should also be noted that we never said UVD would be supported on Linux at all (closed or open source) when these families of hw were launched. UVD 1.x and 2.0 are much different at a hw level compared to UVD 2.2 and newer. As such the older hardware needed had to be evaluated separately for release in open source. Additionally, this family of UVD hardware is several generations old so the relevant architects that need to review it have to look up the relevant details for the older hardware in order to review it properly. For closed source, you can still use catalyst along with a supported distro. Having a drm breach due to open source UVD and losing our ability to sell into a lot of markets is not going to help anyone. It doesn't matter that many of the various drm technologies have already been hacked. AMD is still legally responsible for it's part of the drm stack regardless.

            Comment


            • #7
              Attention rs880 stopped working dpm with 3.14 kernel (http://kernel.ubuntu.com/~kernel-ppa...-next/current/)
              when using radeon.dpm=1 then
              OpenGL renderer string: Gallium 0.4 on llvmpipe

              Comment


              • #8
                Attention rs880 stopped working dpm with 3.14 kernel(drm-next)
                when using radeon.dpm=1 then
                OpenGL renderer string: Gallium 0.4 on llvmpipe

                Comment


                • #9
                  Originally posted by pchristy View Post
                  Not only that, but who is ever going to buy AMD again, if this is the kind of support they can expect? Not me, for a start......
                  I'll never buy NVIDIA hardware until there are open source drivers that are supported and documentation.

                  Comment


                  • #10
                    Originally posted by agd5f View Post
                    For closed source, you can still use catalyst along with a supported distro.
                    For sure, using an ancient kernel.
                    Interested in the latest F2FS improvements? You don't get it!
                    Improvements made to BTRFS? Nope, not for you!
                    Any improvements in the kernel at all since 3.4? AMd says: No, use this old version!

                    So we have the choice between using an ancient kernel or using new stuff that doesn't support our hardware fully? Great, really great.

                    Comment


                    • #11
                      Originally posted by agd5f View Post
                      It should be noted that AMD never said we would support UVD in open source. We said repeatedly that we would look into it, but if it was deemed too risky we may not be able to do it. It should also be noted that we never said UVD would be supported on Linux at all (closed or open source) when these families of hw were launched. UVD 1.x and 2.0 are much different at a hw level compared to UVD 2.2 and newer. As such the older hardware needed had to be evaluated separately for release in open source. Additionally, this family of UVD hardware is several generations old so the relevant architects that need to review it have to look up the relevant details for the older hardware in order to review it properly. For closed source, you can still use catalyst along with a supported distro. Having a drm breach due to open source UVD and losing our ability to sell into a lot of markets is not going to help anyone. It doesn't matter that many of the various drm technologies have already been hacked. AMD is still legally responsible for it's part of the drm stack regardless.
                      And what guarantee is there that if I buy another AMD laptop, I won't find that, once again, support is withdrawn five minutes after I take delivery? Sorry, AMD has just become too risky for me to invest my hard-earned cash in!

                      Frankly, I don't care if the support is open-source or proprietary. I don't care about games. All I want to do is to be able to display HD video (1080p) via HDMI in real time. Surely this is simple enough. And I know the hardware can do it under Linux, because it used to until AMD dropped support. The last proprietary driver was extremely flakey even in its day, and the only way to continue with it is to use obsolete kernels and Xorg.

                      This simply isn't good enough!

                      Intel can do it. All their drivers seem to be open-source. If I can find a non-touch screen 12" laptop with Intel inside, that will be my next purchase.

                      Good-bye, AMD.

                      --
                      Pete

                      Comment


                      • #12
                        Originally posted by agd5f View Post
                        For closed source, you can still use catalyst along with a supported distro.
                        Does catalyst legacy support any kind of hw decoding on old hardware? Isnt xvba supported only on uvd 2 and newer?

                        Comment


                        • #13
                          Originally posted by gradinaruvasile View Post
                          Does catalyst legacy support any kind of hw decoding on old hardware? Isnt xvba supported only on uvd 2 and newer?
                          Well, it worked on my HD4000 laptop using MPlayer-vaapi. That is, when the 13.1 driver didn't cause the machine to freeze while booting in the first place! I never used xvba - only vaapi. Ffmpeg with vaapi enabled also worked reasonably well, despite the low-powered processor. However, I never managed to get VLC to work with hardware acceleration. But then, it doesn't under windoze, either! Only Micro$oft's media player seemed to work with hardware acceleration under windoze.....

                          Comment


                          • #14
                            Originally posted by gradinaruvasile View Post
                            Does catalyst legacy support any kind of hw decoding on old hardware? Isnt xvba supported only on uvd 2 and newer?
                            Yes, just UVD 2.0 and newer.

                            Comment


                            • #15
                              Originally posted by pchristy View Post
                              And what guarantee is there that if I buy another AMD laptop, I won't find that, once again, support is withdrawn five minutes after I take delivery? Sorry, AMD has just become too risky for me to invest my hard-earned cash in!

                              Frankly, I don't care if the support is open-source or proprietary. I don't care about games. All I want to do is to be able to display HD video (1080p) via HDMI in real time. Surely this is simple enough. And I know the hardware can do it under Linux, because it used to until AMD dropped support. The last proprietary driver was extremely flakey even in its day, and the only way to continue with it is to use obsolete kernels and Xorg.

                              This simply isn't good enough!

                              Intel can do it. All their drivers seem to be open-source. If I can find a non-touch screen 12" laptop with Intel inside, that will be my next purchase.
                              I would hardly call the possibility that we may not be able to support UVD on a few older asics a lack of commitment to open source. We support just about every feature of the hw with the exception of UVD on a few older parts and even then we may be able to release older UVD support once the review is done. There are features on older asics that Intel has chosen not to support as well.

                              Comment

                              Working...
                              X