Announcement

Collapse
No announcement yet.

Fedora Project Leader Calls Out NVIDIA Over Their Proprietary Linux Drivers

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

  • Fedora Project Leader Calls Out NVIDIA Over Their Proprietary Linux Drivers

    Phoronix: Fedora Project Leader Calls Out NVIDIA Over Their Proprietary Linux Drivers

    Fedora Project Leader Matthew Miller took to Twitter on Sunday with a long series of tweets of his personal opinion going after NVIDIA's proprietary driver stack and encouraging the company to be more like Intel and AMD with regards to open-source driver support...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Yeah, this is becoming a necessity especially when you see things like these: https://github.com/google/jax/issues...ment-760236170

    It's high time that AMD and Intel come up with competitive hardware, ecosystem will move to the more open software stack in a few years. We're getting first AMD GPU boxes this year just to start this move ...

    Comment


    • #3
      If NVIDIA doesn't want to open up, at least NVIDIA should provide an AMDGPU-PRO-like solution (open-source kernel driver with closed-source user-space driver)...

      Comment


      • #4
        Fuck them!
        Just don't do anything for them.
        Let them do all the hard work and their users who still support this shitty attitude with their wallets.

        Comment


        • #5
          Isn't what Fedora project leader said kind of potentially illegal?

          I mean, most reverse engineered drivers are supposed to be *like* clean room implementation. The moment nvidia employee who saw that specification or code starts working on it, it is no longer clean room implementation.

          Comment


          • #6
            I'd have thought the solution is to detect a Nvidia card and enable a non-accelerated frame buffer. It'll do what you need to use the computer.

            UEFI kernels are signed, so Nvidia can't load kernel modules unless Fedora sign them too (or the user recompiles everything and disables module signing)...

            Comment


            • #7
              Don't buy nVidia hardware. It's that simple.

              Comment


              • #8
                Originally posted by piotrj3 View Post
                Isn't what Fedora project leader said kind of potentially illegal?

                I mean, most reverse engineered drivers are supposed to be *like* clean room implementation. The moment nvidia employee who saw that specification or code starts working on it, it is no longer clean room implementation.
                Remember Nvidia has already contributed to Nouveau before. There is nothing at all legally murky about a vendor choosing to contribute towards a reverse engineering effort directly given the only people with legal standing to sue is Nvidia itself. Microsoft has also worked with Samba developers before for example. What would be questionable is if a rogue employees decides to do it without the company involvement but that is not what is being discussed here at all. The reason why a clean room implementation is a good practice is when you have an uncooperative vendor who is antagonistic towards a reverse engineering effort may lash out at the project legally if it is not cleanly implemented but there is nothing whatsoever stopping a cooperative vendor from participating.
                Last edited by RahulSundaram; 04 April 2022, 07:01 AM.

                Comment


                • #9
                  Originally posted by RahulSundaram View Post

                  Remember Nvidia has already contributed to Nouveau before. There is nothing at all legally murky about a vendor choosing to contribute towards a reverse engineering effort directly given the only people with legal standing to sue is Nvidia itself. Microsoft has also worked with Samba developers before for example. What would be questionable is if a rogue employees decides to do it without the company involvement but that is not what is being discussed here at all. The reason why a clean room implementation is a good practice is when you have an uncooperative vendor who is antagonistic towards a reverse engineering effort may lash out at the project legally if it is not cleanly implemented but there is ever stopping a cooperative vendor from participating.
                  Ah ok, that clarifies a lot, i forgot nvidia did contribute to Nouveau before. Thanks for answearing!

                  Comment


                  • #10
                    Originally posted by piotrj3 View Post

                    Ah ok, that clarifies a lot, i forgot nvidia did contribute to Nouveau before. Thanks for answearing!
                    You are welcome. One quick note to that, you are correct that once a vendor participates in a project directly, it cannot be called a clean room implementation anymore but it is not necessary to have a clean room implementation at that point. Once a vendor participates, they cannot typically then claim any legal issues after the fact in court later because it would be unfair and this is called estoppel.

                    Comment

                    Working...
                    X