Announcement

Collapse
No announcement yet.

NVIDIA Confirms It's Working On Optimus Linux Support

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

  • #16
    Originally posted by Michael View Post
    Uhhh I mentioned it back on launch day that the Catalyst driver supported it, but still from as far as I know it's not the best support possible.
    Yes. Catalyst work. But seems it's more a LUCKY behaviour from the catalyst drivers than an official support.
    Doesn't matter what distro. You need to do some tricks for intel opengl work 100%.
    And NOR ALL cards/computers have the lucky of own a bios with the option for default video output or that is coded with AMD being the primary.
    Example, AMD 6850 card. I still didn't know ANY notebook with catalyst working. Look around just a minority cards/manufacturer work with this "STRANGE CATALYST BEHAVIOUR"

    Seems is more like a weird behaviour from catalyst than something "purposeful".

    Comment


    • #17
      Actually, for some GPUs nouveau power managment works wonderfully. For my GF8600M I actually switched to nouveau because I couldn't take that POS blob "power management" any more. The latter switches power states too late/in the wrong moments, and you end up with a choppy desktop AND huge power consumption; the worst of both worlds.

      With nouveau, switching isn't automatic by default, but at least I can put the GPU in a power state I want, and it'll stay there! And even for the lowest state I get a fluid desktop; something the blob can't seem to achieve.

      Comment


      • #18
        Dont forget that if it was Intel instead of nvidia that had the responsibility to support optimus, then intel would have pushed whatever patches they needed to get the job done. nvidia just decided from the beginning they werent going to help until other developers did most of the work (bumblebee, xrandr, etc.) to get nvidia's own technology working on linux.

        IMO, too little, too late nvidia.... And that also goes for AMD as well. Intel isnt even a graphics card company and they're ahead of AMD's open source drivers when it' comes to OpenGL 3.X support and that is just sad.. I've ended up buying a lot of Ivy Bridge hardware this year. I'm not too happy with AMD dropping support for hardware, long before the open source drivers are complete... AMD really needs to get more manpower on their graphics drivers, ASAP. If it wasnt for people like Marek, AMD would have practically nothing... I was expecting to buy AMD Trinity this year, but it looks like the open source drivers wont even work for Trinity until well into next year. Meanwhile, Intel has their Haswell graphics drivers out a year before the hardware is even released?And on top of that, Intel heavily supports wayland and Meego/Tizen with entirely dedicated engineering teams? Vote with your wallet.

        Comment


        • #19
          Originally posted by fritzls View Post
          From the AMD side.
          Catalyst already WORK with muxless cards. And well.
          BUT!!!!
          If you have how to change in bios the default output card or if the builder code the bios with the amd and not intel being the primary output ( not speaking about muxed card bios choice, but the trigger that turn the primary card using the hardware output).
          If you have luck of own a good coded bios manufacturer one. will work like a charm.
          But if you are a owner from a low end brand like all HP notebooks ( I'm a owner of a shit envy second generation 3d). then you are in troubles.
          BAD ACPI/BIOS coded, No way of change nothing in bios. Nor the card is well coded and the amd being the primary output.

          Seems all the problem are not coming just from BAD support or drivers.

          But from BAD vendors like HPshit.
          Nor just in linux, since in windows they abandon their product after release the product.
          As an owner of one of those "shit" Envy's you mention I assure you that you don't need to go into the bios to get this working - you will however need to restart X if you'd like to switch between the main GPUs

          Comment


          • #20
            Originally posted by Gusar View Post
            I've been saying for a long long time, Optimus isn't just up to Nvidia. It was the graphics stack itself (from the kernel to X) that wasn't capable and needed to catch up. With xrandr1.4 and dma-buf, the graphics stack has caught up. *This* allowed Nvidia to start working on Optimus.
            Since you're talking about open source projects (Kernel and X), there was nothing stopping Nvidia from doing the needed work themselves was there?
            So they opted for "let's wait until someone else does it", and that's fine, but that doesn't mean that Nvidia had no choice like you are implying.

            Comment


            • #21
              Originally posted by aliasbody View Post
              Hey there..

              I don't want to be rude... it is not my type at all... but this happens everytime with me... so please... F**** You Nvidia...
              Am I glad I didn't catch you on a day you wanted to be rude...

              Comment


              • #22
                Originally posted by devius View Post
                Since you're talking about open source projects (Kernel and X), there was nothing stopping Nvidia from doing the needed work themselves was there?
                So they opted for "let's wait until someone else does it", and that's fine, but that doesn't mean that Nvidia had no choice like you are implying.
                We should all send letters to nvidia to fire this "Aaron Plattner" guy. Lazy worthless employee!

                I don't know why all companies don't go opensource. The community does the work for free, and you get all the profit.

                Comment


                • #23
                  Originally posted by boast View Post
                  The community does the work for free, and you get all the profit.
                  Oh if that were only true.

                  Comment


                  • #24
                    Let's keep it in perspective.

                    Consider how many people out there actually bought an Optimus laptop to run Linux. A very, very tiny amount.

                    Take the amount of man-hours required to get something like this working, multiply it by the hourly compensation rate, and divide it by the number of people purchasing the laptops for Linux use and compare it to the revenue gained from each sale. If it doesn't make profit for the company, that decision could cost someone his job. That's just sound business.

                    As Linux fans, we really can't make demands on companies, devs, etc. until we have the market share. That's where we need to focus our energy.

                    Comment


                    • #25
                      Originally posted by boast View Post
                      We should all send letters to nvidia to fire this "Aaron Plattner" guy. Lazy worthless employee!

                      Comment


                      • #26
                        would be nice to also get HDMI support for optimus by default:

                        http://www.webupd8.org/2012/08/get-h...ptimus-on.html

                        Comment


                        • #27
                          Originally posted by devius View Post
                          As an owner of one of those "shit" Envy's you mention I assure you that you don't need to go into the bios to get this working - you will however need to restart X if you'd like to switch between the main GPUs.
                          If you are a owner from the SHITENVY first generation. You are lucky you can unlock your bios ( by hacks) and then change the default output (some bios are not double signed like second and 3 generations) and other ones are sold with the primary output card being the amd.
                          If you are owner from 14/15"envy 1, 2 generation you already have a advanced option in bios to change the default output card. Then just switch to amd and. OK

                          If you have a second generation or 3 generation Envy 17 or 17 3d. good luck. ( seems the 3 generation. Nor in windows you can switch between cards. Same if HP sell this being possible).

                          There is no WAY. You can disable the AMd card with vgaswitcheroo. but not use the AMD one.
                          does't matter what trick you do. boot with intel, later copy xorg and boot with amd and restart x. Change framebuffer in boot kernel.
                          NOTHING. The driver load PERFECT. BUT BLACK SCREEN. Just there is no output for the default monitor. If you use an external monitor. then. sometimes work. But not default one.
                          If you have an second generation HP envy. Or anyone here with an amd 6850 card working with catalyst and switching the cards. Please, let me know how.

                          ho. About the not need of BIOS.
                          Do a test. go in bios and change the primary output in a working catalyst linux one. And change to INTEL the primary output card. Then give us the result after the boot. I already do this test in so many notebooks I have a crystal ball for know what will happens.
                          Last edited by fritzls; 09-01-2012, 04:21 PM.

                          Comment


                          • #28
                            Everything except xrandr 1.4 (the tool), the Intel ddx, and Nouveau ddx is PRIME-enabled and present in the Ubuntu 12.10 repos.

                            But PRIME still really sucks. I could only get mplayer2, WebGL, and glxgears to render with it. OpenArena and glxspheres loaded but wouldn't render due to PGRAPH errors.

                            eric@gensokyo:~$ DRI_PRIME=1 glxinfo | grep OpenGL
                            OpenGL vendor string: nouveau
                            OpenGL renderer string: Gallium 0.4 on NVC3
                            OpenGL version string: 3.0 Mesa 8.1-devel
                            OpenGL shading language version string: 1.30
                            OpenGL extensions:

                            eric@gensokyo:~$ xrandr --listproviders | grep Provider
                            Providers: number : 2
                            Provider 0: id: 141 cap: b nc: 2 no: 4 nap 1 name:Intel
                            Provider 1: id: 98 cap: 5 nc: 2 no: 1 nap 1 name:nouveau

                            Comment


                            • #29
                              Originally posted by Gusar View Post
                              I've been saying for a long long time, Optimus isn't just up to Nvidia. It was the graphics stack itself (from the kernel to X) that wasn't capable and needed to catch up. With xrandr1.4 and dma-buf, the graphics stack has caught up. *This* allowed Nvidia to start working on Optimus. Not the middle finger.
                              Nvidia is using the graphics stack, so they could have helped a lot by providing the Xorg developers with details of what new features they needed from the graphics stack years ago (preferably even before the actual hardware was commercially available), instead of doing this as an afterthought…

                              Comment


                              • #30
                                Originally posted by JanC View Post
                                Nvidia is using the graphics stack, so they could have helped a lot by providing the Xorg developers with details of what new features they needed from the graphics stack years ago (preferably even before the actual hardware was commercially available), instead of doing this as an afterthought…
                                You say X.org developers like we have a few spare sitting around for companies to tell us what to do.

                                We've known we wanted X to do this stuff for years, it doesn't change that unless you commit developers it doesn't happen. NVIDIA could have helped work on this they had the expertise but the company didn't see optimus support on Linux as important enough to dedicate resources to. Thats just a commerical reality.

                                The only reason X.org got this functionality ever is because I decided I had to have it and my boss accepted that as a worthy task to spend a few years on.

                                Dave.

                                Comment

                                Working...
                                X