No announcement yet.

NVIDIA Starts Publishing GPU Hardware Documentation To Help Open-Source Drivers

  • Filter
  • Time
  • Show
Clear All
new posts

  • #51
    Originally posted by entropy View Post

    I'm also looking for a RyZen notebook.

    Last time I checked there was not a single notebook that featured a display better than FHD.

    Acer Predator Helios 500

    Basically a portable desktop.


    • #52
      Since 2001, i plan to buy NVidia card (for now only in laptop)


      • #53
        Maybe I'm just ignorant, but honest question: what is so special about this documentation? Is it any different than what has always lived at ?


        • #54
          Originally posted by Neuro-Chef View Post
          But if Nvidia should one day decide to actively work on open source linux kernel drivers, they would have the resources to get first results out a lot faster than AMD.
          Maybe, but until they actually prove it that is just a claim without any foundation whatsoever.

          There are mainly two reasons why it took AMD so long:
          1. IP idiocy (they had to go through all of the tech docs and all of the code to ensure they didn't violate any IP when releasing that)
          2. Untangling the whole thing so it can be integrated into the Linux kernel, which in completeness only started to happen last year


          • #55
            Originally posted by dragon321 View Post

            They can't. Nvidia develop drivers not only for Linux, but also for FreeBSD and Solaris. They share a lot of code between these platforms. Nouveau is Linux only (Yeah, I know about ports for BSD but they are just ports, Nouveau is developed for Linux). AMD officially supports only Linux (and Windows of course).
            I don't see what the problem would be there.
            The whole KMS/DRM stack is explicitly licenced such that it could be integrated into *BSD.


            • #56
              Originally posted by sarmad View Post

              Hmm... is that the fastest Radeon mobile chipset out there? Coz that's still much slower than the GTX 980m on the Oryx Pro that I bought three years ago.
              No, it's Tonga based, there were ones released after based on Polaris, there weren't any Mobile discreet cards for Vega - just APUs. Fingers crossed for Navi


              • #57
                That's a FHD notebook.

                I'm not looking for such a gaming notebook.
                Rather something I can also use for work and, hence, HiDPI would make totally sense for me.
                Last edited by entropy; 08 August 2019, 05:47 AM.


                • #58
                  Originally posted by cb88 View Post

                  You mean like 1440P or something? Well until now AMD hasn't had a GPU architecture that could push 1440P for non trivial tasks... it would probalby make some sense to release som 17in laptops with a mobile HBM based Navi GPU... though. The 5700 itself doesn't make a ton of sense there due to how much power is wasted on GDDR6.
                  Hmmm. Do you really think that is the reason?
                  I mean the Vega mobile architecture should be capable for that.
                  In particular for business notebooks like the Thinkpad T-model a HiDPI display would make totally sense.
                  Not so much for a high-end gaming notebook. Agreed.


                  • #59
                    Originally posted by schmidtbag View Post
                    So does this mean reclo-


                    But seriously... that's the bare minimum we ask of Nvidia. The nouveau devs are smart people and have done a fantastic job considering how little they have to work with. But they can't do anything about the firmware.
                    I consider it to be a quixotic effort , really. Too much effort, too little results.

                    Nvidia hardware should be totally banned in Linux until Nvidia Corporation really collaborates in the Linux ecosystem in a sane way.

                    What about the vaporware effort of a new Unix device memory allocation API that could succeed GBM, EGLStreams and potentially Android's Gralloc? Where's the published stuff? Any news or it's just another Nvidia's bad joke again?

                    Last edited by timofonic; 08 August 2019, 05:28 AM.


                    • #60
                      Originally posted by KillYourFM View Post
                      Maybe I'm just ignorant, but honest question: what is so special about this documentation? Is it any different than what has always lived at
                      It is explained in that repo’s README:
                      This content was originally kept at:

                      ...but has been moved to github, mainly for easier version tracking. Some of these documents get updates and errata, and the above site was forced to create separate subdirectories for any updates.
                      The old location has been deprecated and is no longer updated, and so doesn’t contain the new MMU manuals that were added in the past month on GitHub for example.