Announcement

Collapse
No announcement yet.

Nouveau Makes Its Own NV40 Firmware Replacement

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

  • #11
    Originally posted by Louise View Post
    But those that use Linux are aware of the driver situation, and if one vendor "just works out of the box"... That will effect OEM's and workstation sells.

    It is like XP. People use at work what they use home and visa versa.

    If the guy that orders the hardware and software, are used to AMD on Linux, that just works, with no work done on his side, he is going to order AMD machines at work.
    Most folk that use linux or have contact with linux don't even know it - it's mostly embedded or servers. The desktop market for linux is still tiny - even with netbooks

    In this case the Nvidia binary blob still works better than any other offering for linux graphics. It has support for OpenGL 3.2 OpenCL and CUDA, mesa is playing catch up big time. Even flgrx isn't as advanced as the Nvidia blob

    I'm not saying I like it - I use AMD and Intel, IcedTea and Chromium. The only blobs on my PC is firmware and Flash (and I'll switch to gnash or swfdec as soon as it works properly with chromium)

    Comment


    • #12
      Originally posted by Louise View Post
      But those that use Linux are aware of the driver situation, and if one vendor "just works out of the box"... That will effect OEM's and workstation sells.

      It is like XP. People use at work what they use home and visa versa.

      If the guy that orders the hardware and software, are used to AMD on Linux, that just works, with no work done on his side, he is going to order AMD machines at work.
      Until foss drivers can match them in features and performance that won't happen. Even things like accelerated video acceleration foss X devs are reluctant to incorporate because of "possible patent" issues and that won't change in the foreseeable future. A corporate purchaser is not going to buy card that is more expensive to achieve less performance. Even if the foss drivers are eventually able to achieve 70% performance of their blob counterparts when it comes to purchasing the buyer is going to see that they get more bang for their buck (and more then likely cheaper) going with a blob solution.

      Comment


      • #13
        It would be pretty amusing if Nvidia hardware ends up being better supported than ATI in Gnewsense because of this development.

        Comment


        • #14
          Originally posted by bulletxt View Post
          ...
          I'm starting to believe NVIDIA will have an hard time with linux at the end of next year and starting from 2011.
          ...
          Maybe in some professional areas, but are we sure on the long term a binary blob will win over open source drivers? ...
          Phoronix did an interview with Andy Ritger fairly recently where his statements suggested that the "professional areas" are the big portions of their business.

          I Quote:

          "There has been, and continues to be, significant Linux workstation interest from a variety of workstation segments (e.g., Oil & Gas, Automotive, Film and Broadcast, etc)."

          "I don't know many concrete percentages. Highend workstation visualization is roughly half Linux, and Digital Content Creation (DCC) is largely Linux. NVIDIA Linux graphics powers a respectable portion of the 3D workstations. Our CUDA user base also has a large Linux contingent."

          I think nvidia is betting on their proprietary drivers because that's where they think most of their consumers are.

          I think AMD previously thought the same at one point. I don't have any evidence for this but the name "fglrx" really does emphasize their firegl workstation gfx cards.

          Thankfully AMD has done a fantastic job at opening up their specs so better opensource drivers can be written.

          Comment


          • #15
            The difference between these blobs, and the AMD/ATI blobs, is in what they do. These blobs are definitions of the context-switching layout, used to provide hardware support for tracking state. ATI blobs are used to start up the CP/DMA engine, one level lower. Completely different things.

            Also, these are non-trivial in size and not required for card operation, unlike the ATI blobs. Additionally, they're actually reversible; ATI's blobs are for custom, single-purpose chips which only use ucode for cost reasons, and it would not be useful at all to have them reverse-engineered since there are no additional features to be gleaned from them. (IIRC there have only been two updates to the ATI blobs, ever, and those were both for older Radeons, not current-gen stuff.)

            Originally posted by Ex-Cyber View Post
            It would be pretty amusing if Nvidia hardware ends up being better supported than ATI in Gnewsense because of this development.
            Well, don't forget Matrox, either. But sure, you can enjoy nVidia cards. I'll be over here enjoying WiFi, and VLC, and 3D. (Although I guess you can have 3D too, since GLX is now free enough for FSF to not throw fits over.)

            Comment


            • #16
              Originally posted by MostAwesomeDude View Post
              Well, don't forget Matrox, either. But sure, you can enjoy nVidia cards. I'll be over here enjoying WiFi, and VLC, and 3D. (Although I guess you can have 3D too, since GLX is now free enough for FSF to not throw fits over.)
              There's a reason I said it would be "amusing" and not "a huge blow to ATI". I might have to make some popcorn and open up the mailing list.

              Anyway, thanks for clarifying the differences between these blobs.

              Comment


              • #17
                Originally posted by FireBurn View Post
                mesa is playing catch up big time.
                Hardly playing anything, at least ever since gallium came out.

                Comment


                • #18
                  Originally posted by Melcar View Post
                  Incredibly sexy news.
                  Impossible to say it better

                  Comment


                  • #19
                    Originally posted by hax0r View Post
                    Hardly playing anything, at least ever since gallium came out.
                    Are you saying you think gallium is feature complete or on some sort of feature parity with the blobs?

                    It's still being worked on heavily, it's the future and will hopefully be comparable to the blobs, but it's still catching up with mesa just now never mind the blobs

                    Comment


                    • #20
                      which means the NVIDIA ctx_voodoo right now will only be needed for the GeForce 8 series graphics cards and newer.
                      which means > 80% of NVIDIA users will need it.

                      [fixed for you]

                      Comment

                      Working...
                      X