Announcement

Collapse
No announcement yet.

NVIDIA Doesn't Expect To Have Linux 5.9 Driver Support For Another Month

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

  • #31
    Originally posted by coder View Post
    Same old tired trolling.

    Does anyone force you to use Linux? You have freedom to go and use proprietary shit, so why not do it and leave us alone?

    Linux' wealth of hardware support is one of its superpowers. Having open source drivers means that you can use devices on platforms other than what the HW vendor originally intended. Not to mention security and other benefits. It's an utterly logical and conventional position that we want opensource drivers. Nvidia's position is really the exception, here.

    Raging against open source is really messing with your head. You cannot even think rationally, any longer.
    Your train of thought is really messed up. Linux is not in a position to dictate the rules of the game unless you want the OS to be reserved for geeks and IT elite. Is this what you want? Really? Most commenters here really love a semi-broken OS but "IT IS OPEN SOURCE I'M SO HAPPY".

    Security benefits? Which ones? I don't remember NVIDIA drivers having been used for a single successful exploitation attempt.

    Originally posted by V1tol View Post
    Just a simple question - why novidia was waiting for a kernel release to announce missing support? What prevented them to use RC to test their blob and if failed - announce that there may be some problems?
    Have you written a single serious application in your life? No? Drivers for GPUs are currently the most complicated piece of software in the software world. Have you ever heard of QA/QC? No? Well, NVIDIA has it and they cannot release something fast which might or might not work. Do you know how much effort is required to work around the newfound kernel limitations?

    novidia? How many GPUs have you helped design in your short worthless life? Zero? I thought so.
    Last edited by birdie; 17 October 2020, 06:37 PM.

    Comment


    • #32
      Originally posted by _r00t- View Post
      RDNA2? What about RDNA1? There's no support yet!
      It works partially. OpenCL does, MIOpen (e.g. for tensorflow support) doesn't work, yet. There seems to be a branch with some activity.

      Comment


      • #33
        'birdie'-darling

        Right or wrong, your self aggrandising soapboxing layered on top of your histrionic delivery has become incredibly boring. Pissing on people's lives over a licensing issue just cement's your shoe's in public opinion.

        Time for a change of scenery for a bit mate, and let loose. Get some perspective.
        Hi

        Comment


        • #34
          Sometimes I don't understand some answers ... Linux and Gnu are born with the aim of free software, there are people who dedicate their precious time to developing and maintaining free software code, free software means first of all sharing their own work and this has allowed the development of the Gnu / linux ecosystem.
          No one condemns a company that instead follows other policies, but always in compliance with the license used in Gnu / linux.
          Contamination of proprietary software in the kernel is unacceptable and we have been too patient and Nvidia has known about this for a long time.

          Comment


          • #35
            Originally posted by stiiixy View Post
            'birdie'-darling

            Right or wrong, your self aggrandising soapboxing layered on top of your histrionic delivery has become incredibly boring. Pissing on people's lives over a licensing issue just cement's your shoe's in public opinion.

            Time for a change of scenery for a bit mate, and let loose. Get some perspective.
            You've confused me with someone else here. I vote for modesty and humbleness - absolute most Linux/Open Source fans here are self-entitled assholes who absolutely love to demand and at the same time they do absolutely nothing for Open Source. Me, on the other hand? I've been helping various Open Source projects for over two decades now. I've helped resolve multiple critical bugs in the Linux kernel and multiple other Open Source projects. I've been using Linux longer than some of the commentators here have lived.

            So, please, kindly fuck off.

            Comment


            • #36
              Originally posted by birdie View Post

              Another utterly asinine comment full of hatred, bigotry and fanaticism.
              You have to treat cancerous ideas like proprietary software like that or they spread their piss everywhere and rob you hard earned money for what you could do for significantly less. I mean, what paid software used daily hasn't been replaced by a much better alternative? Basically, nothing.

              How much code do you have in the kernel? I have slightly above 0. If you have 0, work your fat fingers more than your far mouth before you criticize what is made for you on behalf of others. Not only that, I've contributed to user space stuff I use that had issues/needed maintenance, also.

              Also, Linux is in a position to dictate the rules of the game, that's why they do. In fact, they're so powerful, they're more akin to gods. They don't answer to you, me, or corporations, governments, or anyone who doesn't can't back their license. They invulnerable to anything, even time. As they can die, but the code still lives on.

              Your simping for proprietary shit is retarded considering the forum you're on. I'll never reply to your dumb posts ever again, not worth the time to entertain your dumb posts.

              Comment


              • #37
                Originally posted by abott View Post

                You have to treat cancerous ideas like proprietary software like that or they spread their piss everywhere and rob you hard earned money for what you could do for significantly less. I mean, what paid software used daily hasn't been replaced by a much better alternative? Basically, nothing.

                How much code do you have in the kernel? I have slightly above 0. If you have 0, work your fat fingers more than your far mouth before you criticize what is made for you on behalf of others. Not only that, I've contributed to user space stuff I use that had issues/needed maintenance, also.

                Also, Linux is in a position to dictate the rules of the game, that's why they do. In fact, they're so powerful, they're more akin to gods. They don't answer to you, me, or corporations, governments, or anyone who doesn't can't back their license. They invulnerable to anything, even time. As they can die, but the code still lives on.

                Your simping for proprietary shit is retarded considering the forum you're on. I'll never reply to your dumb posts ever again, not worth the time to entertain your dumb posts.
                Sorry but it is not 0. He has contributed to the kernel, and even reported on some major problems. Heck, he even got an article on Phoronix.
                Come on, he is being honest. Linux is not ready for the desktop, and we have to agree.

                Comment


                • #38
                  Originally posted by V1tol View Post
                  Just a simple question - why novidia was waiting for a kernel release to announce missing support? What prevented them to use RC to test their blob and if failed - announce that there may be some problems?
                  I don't think they were waiting, they just didn't finish the required work in time. If they were just starting on changing the code, they wouldn't have a due date already.

                  Originally posted by Charlie68 View Post
                  Sometimes I don't understand some answers ... Linux and Gnu are born with the aim of free software, there are people who dedicate their precious time to developing and maintaining free software code, free software means first of all sharing their own work and this has allowed the development of the Gnu / linux ecosystem.
                  The conterargument being it's not that free if it places a(ny) burden on you the minute you start using it.
                  I don't think you should include free software in proprietary code, but you should be able to layer proprietary stuff on top of free software. (Just my point of view, let's not try to establish here where the line should be drawn )

                  Originally posted by Charlie68 View Post
                  No one condemns a company that instead follows other policies, but always in compliance with the license used in Gnu / linux.
                  Contamination of proprietary software in the kernel is unacceptable and we have been too patient and Nvidia has known about this for a long time.
                  I don't believe contamination was what was happening here, the driver wasn't in the kernel (how could it?), it just used a kernel module.

                  What escapes me (and I'm sure there's a good reason for this, I just don't know it), is why is the kernel basically exposing two interfaces to the outside world, one for GPL code and one for everybody else? I don't recall seeing this mo anywhere else.

                  Comment


                  • #39
                    Originally posted by V1tol View Post
                    Just a simple question - why novidia was waiting for a kernel release to announce missing support? What prevented them to use RC to test their blob and if failed - announce that there may be some problems?
                    Almost certainly they (nvidia, vmware, etc.) had started the processes internally (all the vendors doing closed source drivers/modules track such things), but in addition to wanting to make sure the work they did would apply to the release (there have been reverts that have occurred at the last cycle), they have typically have a (very) long review and QA process, since they are re-implementing, or finding alternative solutions to, sometimes very low level functionality (memory mapping, etc.), and that always takes longer than a minor parameter change that sometimes happens and can be done quickly. And until the actual release they likely cannot be sure about how many weeks later things will take (if the release had been pushed off yet again (and Linus admitted there were more fixes than he would have preferred), it might have been only three weeks, and then two weeks, etc.). My suspicion is that this announcement is likely about expectation setting for their large corporate customers using CUDA/OpenCL on datacenter hardware who were looking forward to other features in Linux 5.9 (like improved performance). Of course, those same customers have other options to stay 100% in compliance to the licenses too, but some want a more turnkey solution. Last I knew nvidia is constantly looking to expand their linux kernel team, so if you want to contribute, and have the talent, you should apply, so that others might see a shorter cycle.

                    Comment


                    • #40
                      Originally posted by QwertyChouskie View Post

                      Couldn't they just do what AMD did with AMDGPU? One kernel driver, then do whatever they want with the userspace.
                      AMD used to have a proprietary driver which they inherited from their acquisition of ATI (called FGLRX). The acquisition of ATI was terrible on a lot of fronts, but one of them was the inherited driver from ATI was so terrible that AMD did the decision to just recode it from scratch.

                      NVidia had the privilege (which they still had) of having a well designed and functioning driver, I highly doubt they are going to re code it. Furthermore, the scope of NVidia's driver is much more nowadays. Its not just a GPU driver, there are thins like DLSS, Geforce Streaming, CUDA, G-Sync etc etc etc. Recoding all of this is just not going to happen and parts of their driver (i.e. the neural models used for DLSS) definitely fall into the bucket of trade secrets.

                      Comment

                      Working...
                      X