Announcement

Collapse
No announcement yet.

Adobe Rants Over Linux Video Acceleration APIs

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

  • #41
    Originally posted by admax88 View Post
    nVIDIA's open source efforts? Haha yeah right. Just because the API is documented does not make it open source.
    Ah, good old NIH syndrome. We totally can't use that Wheel thing because those jerks over there invented it! Let's stick with the Square even though it's technically inferior!

    Originally posted by admax88 View Post
    Also if we just pick vdpau then we're stuck with nVIDIA hardware. In my opinion VA-API is a better choice at this point since it has VDPAU and XvBA backends, plus direct support on intel hardware. This way we can support hardware accelerated video on nVIDIA/ATI/Intel GPUs, rather than just nVIDIA.
    Nothing is preventing those other vendors from implementing VDPAU drivers.

    Comment


    • #42
      Originally posted by md1032 View Post
      Nothing is preventing those other vendors from implementing VDPAU drivers.
      Isn't that a bad idea for people that aren't Nvidia? Since NV controls the API, they would have a competitive advantage. It should be better for each other company to have an open abstraction layer, like what VA-API is doing with a VDPAU backend, right?

      Comment


      • #43
        Re Flash GL accel - last I checked this had to be forced on for everyone running FOSS drivers. Not to mention GL accel is on older cards worse than x11, not to even mention xv.

        Comment


        • #44
          Originally posted by garytr24 View Post
          Isn't that a bad idea for people that aren't Nvidia? Since NV controls the API, they would have a competitive advantage. It should be better for each other company to have an open abstraction layer, like what VA-API is doing with a VDPAU backend, right?
          freedesktop.org controls the API.

          Comment


          • #45
            Originally posted by garytr24 View Post
            Isn't that a bad idea for people that aren't Nvidia? Since NV controls the API, they would have a competitive advantage. It should be better for each other company to have an open abstraction layer, like what VA-API is doing with a VDPAU backend, right?
            If nv controls the API then you might as well say intel controls VA-API (you know the company that has been getting hammered on in court with anticompetitive practices). Nvidia made it painfully clear since the very beginning that vendors are free to implement their own vdpau support but it seems some people like to append their own baseless conclusions about it's license.

            Comment


            • #46
              Originally posted by admax88 View Post
              nVIDIA's open source efforts? Haha yeah right. Just because the API is documented does not make it open source.

              Also if we just pick vdpau then we're stuck with nVIDIA hardware. In my opinion VA-API is a better choice at this point since it has VDPAU and XvBA backends, plus direct support on intel hardware. This way we can support hardware accelerated video on nVIDIA/ATI/Intel GPUs, rather than just nVIDIA.
              the api is documented and is open source i though.t

              Originally posted by md1032 View Post
              Nothing is preventing those other vendors from implementing VDPAU drivers.
              that's exactly what i was trying to say! i didn't mean that everyone should develop for nvidia hardware. that's just dumb. i meant that intel and ati should take up vdpau.

              Comment


              • #47
                The follow-up post is available now

                Hi guys, the follow-up post mentioned in that blog entry is now available, and explains the thought process better, and answers the questions posed in the previous blog's comments, plus some other common questions about Flash, CPU's, and GPU's.

                Comment


                • #48
                  Originally posted by md1032 View Post
                  Ah, good old NIH syndrome. We totally can't use that Wheel thing because those jerks over there invented it!
                  Yeah, why did NVIDIA create VDPAU when good old VAAPI already existed?

                  I don't necessarily see anything wrong with companies choosing to implement vdpau acceleration in their drivers, it's certainly possible. But the real question is why should they? VAAPI already supports Intel + ATI (sort of) + Nvidia, so it doesn't make much sense from their perspective to suddenly change their development to focus on a different API that nvidia is pushing for.

                  I mean, sure it's a little buggier for now, but if they all suddenly switched to vdpau their vdpau implementations would all be even buggier. Hopefully they'll get all the bugs ironed out soon.

                  Comment


                  • #49
                    Originally posted by smitty3268 View Post
                    Yeah, why did NVIDIA create VDPAU when good old VAAPI already existed?
                    VA-API was an a slow train to nowhere, was featureless ( and somewhat still is lagging ) and really was only useful at the time on the GMA500 (and we all know how well it's supported). Nvidia got the ball rolling and was extremely well received and adopted for various projects.

                    You might want to check out http://video.linuxfoundation.org/video/1575 as it also covers some yet untapped potential of vdpau.

                    Comment


                    • #50
                      Originally posted by deanjo View Post
                      VA-API was an a slow train to nowhere, was featureless ( and somewhat still is lagging ) and really was only useful at the time on the GMA500 (and we all know how well it's supported). Nvidia got the ball rolling and was extremely well received and adopted for various projects.

                      You might want to check out http://video.linuxfoundation.org/video/1575 as it also covers some yet untapped potential of vdpau.
                      Sure, but NVidia could have improved VA-API rather than implementing their own completely from scratch. Maybe it was easier for them to start from a clean base, but it does mean that attacks on VA-API being NIH-syndrome are pretty misplaced.

                      Comment

                      Working...
                      X