Announcement

Collapse
No announcement yet.

NVIDIA Wants Feedback On Its Device Memory Allocator Project

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

  • #31
    Originally posted by DanL View Post
    If it's an Optimus laptop, the builtin screen uses the integrated Intel GPU while the HDMI output is probably connected to the Nvidia GPU, so the problem is more likely with Intel driver.
    Optimus doesn't work like this. In Optimus all outputs are attached to Intel GPU and NVIDIA card doesn't have any physical output. Thanks to this solution, NVIDIA GPU is enabled only when it's really needed.

    Comment


    • #32
      You'd think after a year of everyone ignoring it they'd get the hint.

      Comment


      • #33
        Originally posted by computerquip View Post
        Intel has always been on top of their Linux game.
        Yeah, especially their BayTrail and CherryTrail drivers /s

        Comment


        • #34
          Originally posted by computerquip View Post
          I hate saying this... but we've been saying that about AMD for almost a decade... and Intel has always been on top of their Linux game.
          Saying what about AMD? That they're a better option? Because only the Stallmanites felt that way going back that far, and frankly, their perspective is blinded by their principles. But AMD is already a very good option for Linux - most of the hardware works great out-of-the-box. The performance level is almost on-par with Windows (and sometimes exceeds it), radeonsi is almost 100% up to OpenGL spec, Vulkan support (depending how you look at it) is almost 100% up to spec, hybrid graphics is mostly functional (from what I heard anyway), etc. There legitimately isn't a whole lot left to complete the Linux AMD experience. OpenCL has some catching up to do, but at least it works. The biggest shortcomings are AMD-specific technologies, like Crossfire, ReLive, and Freesync (though has that been added yet?), and, having a user-friendly interface to configure your hardware.

          So all that being said, AMD is not quite better than Nvidia for Linux (when you account for everything the product can/should be able to do), but, it is undoubtedly very close, and my point is I think Nvidia realizes this. I know for a fact I did not think this way a year ago. Back then I felt AMD's progress was fantastic, and I was perfectly fine using their hardware, but I still felt Nvidia had a more complete and better overall experience.

          As for Intel, they're a mixed bag. They're usually fantastic about supporting their products as soon as they're released, and they are a huge asset to the development of Mesa. But, once the next generation of GPUs comes up, they tend to get very negligent of previous generations. The only exception is their SoCs, which remain very poorly supported in Linux throughout their lifespan.
          Last edited by schmidtbag; 22 November 2017, 12:06 PM.

          Comment


          • #35
            Originally posted by Sevard View Post
            Optimus doesn't work like this. In Optimus all outputs are attached to Intel GPU and NVIDIA card doesn't have any physical output.
            I've never owned a hybrid laptop, but I was thinking of reverse prime:

            Comment


            • #36
              Originally posted by c117152 View Post
              You'd think after a year of everyone ignoring it they'd get the hint.
              Your view of the matter is extremely distorted/biased. It's not, "GBM is awesome! F Nvidia!"
              The goal was to bake the ideas instead of just throwing shit against the wall and hoping it stuck.

              Read it again:
              Again, I understand some people are growing concerned that I've been doing this off on the side in a github project that has primarily NVIDIA contributors. My goal was only to avoid wasting everyone's time with unproven ideas. The intent was never to dump the prototype code as-is on the community and presume acceptance. It's just a public research project. Now the prototyping is nearing completion, and I'd like to renew discussion on whether and how the new mechanisms can be integrated with the Linux graphics stack.
              -- https://lists.freedesktop.org/archiv...er/177632.html

              Comment


              • #37
                Originally posted by DanL View Post
                Your view of the matter is extremely distorted/biased.
                If by bias you mean 20 years of nVidia ignoring the linux kernel's development cycle (and really, any open working group development cycles) by repeatedly not taking part in discussions over API designs and then introducing their own, broken, APIs after the fact, then yes. I'm biased.

                fbdev, kms, cuda... nVidia been deliberately breaking APIs as long as I can remember. Now that Vega is looking promising they suddenly wake up? No thanks. I had my fill thank you very much.

                Comment


                • #38
                  Originally posted by c117152 View Post
                  If by bias you mean 20 years of nVidia ignoring the linux kernel's development cycle (and really, any open working group development cycles) by repeatedly not taking part in discussions over API designs and then introducing their own, broken, APIs after the fact, then yes. I'm biased.
                  Hey, you can knock nvidia for a lot of crap, past and present. This isn't one of them. Nvidia made EGLstreams and then admitted it wasn't ideal. The GBM devs admit their solution isn't ideal. So the two camps look to make something better. There's no guarantee it doesn't end up in squabbling and NIH fragmentation, but based on what we've seen so far, that's a big conclusion to jump to.

                  Comment


                  • #39
                    Originally posted by schmidtbag View Post
                    Interesting to me how Nvidia suddenly cares about the opinions of Linux users, and even Nouveau. Back in Linus' "fuck you Nvidia" days, the available technology and drivers at the time gave Nvidia enough of an edge where they could be their usual selves and just brush it off like "I don't care what any of you think, you're doing it my way because you don't have a better option". But now, AMD+Intel are almost fully caught-up with Nvidia (in terms of performance and features), and Wayland is starting to be more widely adopted. Ignoring hardware limitations, we're very close to a point where AMD will be a resounding better option for Linux (regardless of your principles) and where Nvidia will be holding back development to the point where they're resented rather than "begrudgingly accepted".
                    They don't "suddenly care about Linux users". They care about the enterprise market and AI. This has been apparent for several years. What happened is Wayland isn't going away and their EGL strong arm isn't going to happen. Also AMD woke up and starting getting their stuff together. AMD hasn't had competitive CPUs since before they acquired ATI and really got into the graphics business.

                    AMD's new angle is delivering the entire solution of graphics and compute. This is becoming an actual compelling value proposition and it's only going to get worse for Nvidia with the way it looks like things are going. These Intel laptops with integrated AMD graphics are directly to attack Nvidia.

                    Comment


                    • #40
                      Originally posted by schmidtbag View Post
                      Interesting to me how Nvidia suddenly cares about the opinions of Linux users, and even Nouveau. Back in Linus' "fuck you Nvidia" days, the available technology and drivers at the time gave Nvidia enough of an edge where they could be their usual selves and just brush it off like "I don't care what any of you think, you're doing it my way because you don't have a better option". But now, AMD+Intel are almost fully caught-up with Nvidia (in terms of performance and features), and Wayland is starting to be more widely adopted. Ignoring hardware limitations, we're very close to a point where AMD will be a resounding better option for Linux (regardless of your principles) and where Nvidia will be holding back development to the point where they're resented rather than "begrudgingly accepted".
                      Who does that remind me of... oh yeah, Shuttleworth and Mir. NVidia is bigger so it'll be harder for them to fail as spectacularly as Mir did, but they will feel the $$$ sting this time.

                      Comment

                      Working...
                      X