Announcement

Collapse
No announcement yet.

NVIDIA's Looking To Expand Its Linux Team

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

  • #11
    Originally posted by TheBlackCat View Post
    " Heck, any sort of working user-level GUI for configuring monitors would be nice. Their current tools do not even remember their settings between sessions, so they are essentially useless without editing your xorg.conf file, which requires root privileges and most Linux distros nowadays don't even use xorg.conf.
    Add
    Code:
    nvidia-settings ?load-config-only &
    to your ~/.xinitrc file, or in ~/.kderc (KDE) or in ~/.gnomerc, etc?

    How about optimus support? Many, if not most, upcoming laptops ship with graphics cards that cannot be used under Linux because they rely on optimus, while Nvidia has said they have no plans to support optimus under Linux.
    Licensing clash, doubt you will ever see it.

    How about kernel mode setting?
    See above.

    Or anything else needed to get wayland working.
    Doable but not a high priority until wayland starts proving itself that it will be the replacement for X.
    On the software side a lot of work is going into getting ready for it, but Nvidia seems to be totally ignoring it.
    Don't be so sure about that. Nvidia doesn't typically announce anything until they have a solution ready for the public. They "didn't have any plans" for accelerated video decode for example and then "Bam!" there was vdpau.

    Comment


    • #12
      Support for optimus and profiles for games would be great.

      I seriously doubt nVidia is going to release an open source driver anytime soon. If they are going to release an open driver in the very distant future, the best way to do this would be to open source their stable branch of their drivers. In order to do this they would probably need to strip this driver from any support for future products, and then add it when the new products arrive.

      Comment


      • #13
        For KDE users:
        Code:
        mkdir -p ~/.kde/Autostart
        printf '"#!/bin/sh\nnvidia-settings -l\n' > ~/.kde/Autostart/nvidia.sh
        chmod +x ~/.kde/Autostart/nvidia.sh
        is usually enough for some settings (no "&" needed, it terminates directly). But not every setting is stored in ~/.nvidia-settings-rc - if you need other options then you need to use the direct commands. For special Twinview settings those are stored in xorg.conf in order to get active, that's not really optimal. Even fglrx is more flexible there.

        Comment


        • #14
          Originally posted by glxextxexlg View Post
          You know, if it to be AMD to announce this job offer (ie: we're looking for guys to develop binary blobs), Q wouldn't make fantastic statements like this. It would all be justified for him.

          But it is not. AMD gives way to linux developers while nvidia employs them and you know why? They are successful in linux workstations, handheld linux devices, scientific visualization etc. and amd is not.
          AMD does more for Open Source and linux than Nvidia. It's not even close.

          It's not just the graphics stuff, also CPUs, chipset stuff, etc.

          Comment


          • #15
            Originally posted by deanjo View Post
            Licensing clash, doubt you will ever see it.
            So most recent nvidia laptops will not work with Linux.

            This is clearly a terrible situation. Every AMD chipset ever released works under Linux, even using OSS drivers.

            Comment


            • #16
              Originally posted by efikkan View Post
              If they are going to release an open driver in the very distant future, the best way to do this would be to open source their stable branch of their drivers.
              I don't think that this would be a good idea, for several reasons:

              - 3rd party code
              - patented software
              - it would be an island separate from everything else in Linux land like it is now. Much of the nouveau core is well designed and efficient, it's just missing the countless optimisations across the board. A HUGE driver which reimplements all of Mesa, most of X, and the entire linux kernel would be impossible for most OSS hackers to understand, let alone improve on.

              The best way would be to support nouveau with documentation (most important) and developers.

              Comment


              • #17
                The open source drivers lack features and performance, and is useless for high end games. They way these drivers are made they will never get close to the quality of nVidia's proprietary drivers.

                It would actually be better to improve a good driver architecture than a mess consisting of mesa and gallium3d.

                Comment


                • #18
                  The nvidia driver is millions of lines of code, much of it dirty hacks made for specific games.

                  Releasing that would likely not be very useful. The AMD OSS developers have access to the Catalyst code (which is similar in terms of complexity and performance), and I don't think that any of it gets used directly.

                  It's simply too complicated. Documentation and support are more important.

                  Comment


                  • #19
                    Also, please note that r300g doesn't really lack either features or performance.

                    It's simply a matter of documentation and manpower. There is nothing fundamentally wrong with Mesa or Gallium3d.

                    Writing and maintaining new Mesa and Gallium3d architecture for each card (one for nvidia, one for AMD, one for Intel...), that's a dumb suggestion.

                    Comment


                    • #20
                      Originally posted by deanjo View Post
                      Add
                      Code:
                      nvidia-settings ?load-config-only &
                      to your ~/.xinitrc file, or in ~/.kderc (KDE) or in ~/.gnomerc, etc?
                      As I explained already, this will not restore your monitor settings, things like resolution, multi-monitor layout, and so on.

                      Besides, editing shell scripts is not a gui-based approach anyway, so even if it did work it wouldn't actually solve the problem. On windows, you just set the settings and you are done. Same with anyone using randr. But for Nvidia there is no purely GUI-based user-level solution.

                      Of course, even if it did, that still leaves the question of why Nvidia can't just support randr 1.2 like everyone else. It also leaves the question of why they said years ago that it would be out soon, why they they said years ago it was a "top priority", then later admitted it was never a priority and probably never will be.

                      Originally posted by deanjo View Post
                      Licensing clash, doubt you will ever see it.
                      We simply aren't going to be able use Nvidia on laptops, then.

                      Originally posted by deanjo View Post
                      Doable but not a high priority until wayland starts proving itself that it will be the replacement for X.
                      Everyone else seems pretty convinced.

                      This is part of the problem with Nvidia's approach. They don't even begin thinking of trying to work on support for something until a stable version is out. That means you are left with months, if not years, of it working with every other driver until Nvidia finally gets around to working on it.

                      It took something like 6 months after KDE 4.0 was released before Nvidia finally got around to fixing crippling bugs in their drivers, bugs no other drivers. Heck, it took a month or two until they even admitted the bugs were their fault. The bugs had been known for at least that long beforehand during the beta cycle, but Nvidia said they refused to even look at the problem since it was still beta software.

                      Originally posted by deanjo View Post
                      Don't be so sure about that. Nvidia doesn't typically announce anything until they have a solution ready for the public. They "didn't have any plans" for accelerated video decode for example and then "Bam!" there was vdpau.
                      Yes, that is true when implementing their own internal stuff. But when supporting Linux software or standard components of Linux or Xorg, in the past even when they said they would work on it that doesn't actually mean they will.

                      Comment

                      Working...
                      X