Announcement

Collapse
No announcement yet.

NVIDIA Contributes Much Less To The Linux Kernel Than Intel Or AMD

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

  • #11
    And don't forget Nvidia does everything they can to prevent consumer GPUs from running via passthrough on KVM. You have to jump through hoops to do it, if it can be done at all.

    Nvidia, like Intel, is just too darned greedy.

    Comment


    • #12
      Nobody wants anything to do with NVIDIA crap except themselves.

      Comment


      • #13
        This is unsurprising, AMD and Intel both have CPUs and GPUs (or at the very least iGPUs) that require code, meanwhile Nvidia only has GPUs and maintains it's main driver code outside of the kernel in the first place...

        So... Tell us something we don't know? If, to be fair to nvidia, you remove all CPU and GPU (or at least GPU Driver) related code, will it still be contributing 'much less' than both amd and intel? I strongly doubt it, this is just meaningless bashing of nvidia here. Nvidia only does dedicated GPUs. AMD only does CPUs and GPUs, intel does a shitload of things (and maintains clearlinux as some icing on that cake) other than CPUs and GPUs, so again, if you ignore intel's contributions except directly related to CPU and GPU, is intel still as far ahead of amd and nvidia too? I doubt this too.

        Comment


        • #14
          Originally posted by hax0r View Post
          It's OK, nvidia doesn't have to contribute if they don't wan to or have no interest (except if they're pressured by their clients e.g. what happened with Tegra).
          We should look at this matter from other perspective, Enduser contributions vs Big companies. When you think about it you experience mixed feelings, and end up in NetBSD land.
          There is nothing wrong per say with contributions from big companies. The problem is when big companies are completely hostile to open systems. As for some companies not getting open source well many individuals don’t either.

          Comment


          • #15
            nVidia doesn't have to contribute a lot of code because they do everything through their shim layer to the binary driver. So, you may hate their binary blob stance, but it works just fine. I ran nVidia for 8+ years while ATI / AMD GPU were absolutely awful driver wise and never had a major issue with Arch or FreeBSD. Is everyone here just forgetting that AMD until recently (last 2 years or so) was also all closed source shit and the only the reason they switched is because the open source driver was actually better than their own? I don't remember how open or closed ATI / AMD was about getting documentation out about their GPUs vs. how reverse engineered the open source driver was, but their closed sourced driver was terrible even on Windows. nVidia ate their lunch, dinner, and midnight snack as far as quality goes.

            In the last 2 years? Yeah, I'm all Ryzen and AMD GPUs in my house running Arch and FreeBSD because their drivers have stepped up and Ryzens just rock, but if they turn to shit again then "shit open source" vs "working binary blob" is really not a hard decision for me.

            Comment


            • #16

              Comment


              • #17
                Originally posted by rhavenn View Post
                nVidia doesn't have to contribute a lot of code because they do everything through their shim layer to the binary driver. So, you may hate their binary blob stance, but it works just fine.
                No, it does not work just fine:
                - It's buggy so you get system stability issues
                - it doesn't follow the standards. You still can't run Wayland on it.
                - It's not well integrated into the overall system. What's up with the driver providing it's own randr configuration when the DE already provides one?
                - it's incomplete. Will we ever get Prime working as well as it does on Windows or as good as Amd on Linux?

                Comment


                • #18
                  Originally posted by muncrief View Post
                  And don't forget Nvidia does everything they can to prevent consumer GPUs from running via passthrough on KVM. You have to jump through hoops to do it, if it can be done at all.

                  Nvidia, like Intel, is just too darned greedy.
                  I also hate how you can't use NVFBC or NVIFR to capture the screen unless you waste your money on a Quadro/Tesla card.

                  Comment


                  • #19
                    Yeah, right because NVIDIA doesn't have its own x86 CPU, like AMD and Intel do, then Intel has a lot of other hardware which is supported by Linux not limited to ACPI (mostly developed by Intel), UEFI (likewise), NIC (they have a ton of them, WiFi and LAN), BlueTooth, etc. etc. AMD has a CPU and a ton of GPUs (fully open sourced drivers).

                    On the other hand, NVIDIA has very little hardware which requires direct kernel support.

                    But that won't stop open source fanboys from hating NVIDIA.

                    Comment


                    • #20
                      Originally posted by r08z View Post
                      I remember when Linus gave a very insightful critique on NVIDIA. With the way things are going, I think Linus' insight will be applicable for generations to come.
                      There was nothing insightful about it.

                      NVIDIA supports Linux as much as it needs and wants to. It does not owe anyone anything. People here somehow believe otherwise. Since Linux exists, you must support it or you're "bad". That's so childish and stupid I've got no words.

                      Comment

                      Working...
                      X