Announcement

Collapse
No announcement yet.

What Will Happen To xf86-video-nv In 2010?

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

  • #31
    Originally posted by [Knuckles] View Post
    I would like to introduce a case study: the forcedeth driver.

    I know it is a network driver, but it was originally a reverse engineered network driver for nvidia nforce2 network hardware (using it right now on this pc). Nvidia also had its own closed-source blob, but at one point just stopped maintaining it and started contributing to forcedeth.

    A link to a bigger version of the story: http://liquidat.wordpress.com/2007/0...rcedeth-story/ .
    This won't happen for graphic, network driver are "simple" things, they pretty much all do the same things, their might be a bunch of clever tricks but i am not expecting that anyone have a network hw that provide an advantage over others hw.

    Comment


    • #32
      Originally posted by Remco View Post
      I think they will just stop any opensource effort. NVIDIA only cares about their proprietary driver. The opensource driver is only there so that Linux boxes with NVIDIA hardware will at least get to the login prompt. When Nouveau does that (and more), there's no need to waste resources.
      first comment I came across that pretty much said what I was thinking.

      Right now I'm trying to wrap my head around what Nvidia's actual... goal... is. Nvidia has exhibited several brain-dead streaks of logic over the years, and quite frankly, they are in a situation even more awkward than 3DFX once was. For reference, 3DFX once claimed that nobody cared about gaming in 24/32bit color, so their Voodoo 3 cards only accelerated 16bit color... while competitors such as ATi, Nvidia, and even Matrox if memory serves were pumping out cards that could accelerate 24/32bit color. The marketing mis-steps by 3DFX took them out of the market. Products were late or never completed, they tossed off the user-base, and they cut exclusive promotional deals.

      Nvidia's now doing pretty much the same thing. They're still relying on a mega-architecture design that's overly complex, so barring other issues, the products late. Nvidia's cutting deals with developers like Rocksteady and Gearbox to sabotage support on non Nvidia graphics cards.

      From a certain perspective, Nvidia's doing exactly what 3DFX did... and um... does anybody remember just how well that went over for 3DFX? Oh yeah, they went bankrupt.

      ***

      Looking at Nvidia's current, trajectory, I think it might be their aim to leave the gaming graphics card scene completely. The Fermi whitepaper pretty much emphasizes everything else a complex processor can do... but accelerate graphics. Nvidia seems to be aiming themselves into the niche, but profitable, super-computer market where the parallel designs of a GPU can be put to work on things like weather modeling or physics calculations. Nvidia also seems to be trying to set themselves up as a purchasable target for Apple, pushing the Tegra platform which, to my eyes, seems to be Nvidia's attempt to create a next generation Ipod platform.

      If this is Nvidia's goal, to get out of the consumer graphics card market and focus solely on platform deals and systems contracts, then NV can just die. Desktop features won't matter at all to a boffin trying to simulate rain drops flowing over the chassis of the next Ford Focus, or to the boffin trying to model a storm-front moving across a mountain range.

      In this light, I expect that Nvidia will simply abandon all pretense to be a supporter of open-source software. I, personally, have long maintained that Nvidia is anti-open source. I don't think that Nvidia has ever taken Desktop Linux seriously. I don't think they've ever actually approached it as something that could be a profitable market venture. It's always just been something that the 3D rendering houses who buy Nvidia hardware needed worked on as an afterthought. Gaming and desktop performance has also been something that Nvidia's had to work on when they got the contract for the PS3.

      I don't think Desktop use has ever really been a focus of Nvidia's. As I saw NV's purpose, it was basically an olive branch so that Nvidia could claim Open-Source support, and it only served as a bootloader. Since Nvidia wasn't, in my view, interested in the desktop market, there was no reason to improve the driver or unobscure it.

      Now, Nouveau is giving Nvidia an exit strategy. Nvidia can remove itself from Open-Source support, know that it's cards will still somewhat work, and just go on from there.

      Comment


      • #33
        Originally posted by glisse View Post
        This won't happen for graphic, network driver are "simple" things, they pretty much all do the same things, their might be a bunch of clever tricks but i am not expecting that anyone have a network hw that provide an advantage over others hw.
        Agreed that graphics and network are different, but they DID develop a closed-source driver for the network card, instead of an open one, and they never opensourced it, they switched their work to something open-source that appeared in the meantime, which is what I was trying to highlight.

        I'm not saying that nvidia will start putting a lot of resources behind nouveau, I'm just saying that maybe they could just contribute the same as they do with nv: pretty much basic modesetting and pci id's.

        It's not much but it would be a start.

        Comment


        • #34
          Originally posted by airlied View Post
          I'm not sure which X.org devs have screamed for the blob to be opened I can certainly say I don't know of any, you might want to provide some sort of citation to back your "facts"

          My impression is you have made a basic "because I value something it means everyone must think its important" mistake in your logic. Xrandr support wasn't provided for a desktop user to set their PC up once and leave it alone, it was developed for laptop users who plug their laptops into projectors a lot and want to show stuff.

          The story is that some Intel exec was doing a Linux demo and realised he couldn't hotplug his monitor on laptop will all Intel parts and he kicked heads due to its unacceptability. (may not be true but it sounds good)

          Things happen because ppl want them to happen, if we don't have tearfree video on open source drivers its not because nobody knows how to do it, its because nobody considers it worth doing. For example at Red Hat we can't ship a DVD player with our OS, so why the hell would we invest money in tearfree movie playing? Most Linux users playing movie are playing dodgy legal rips of content from other sources, its not something we get much paying customer demand for at all.

          Dave.
          Hi Dave

          I'm sorry but I feel compelled to disagree. With the advent of youtube, the likes of hulu and netflix in the US and iPlayer and 4OD (and the rest) in the UK more and more people are watching legal video on their computers

          Granted this is mostly all via flash but tear free movie playing is very important. Not just to end users but to vendors that want to ship linux based devices as front ends to these services

          Regards

          Mike

          Comment


          • #35
            Originally posted by MostAwesomeDude View Post
            Which video codec do you use to encode your video? If it's anything but MJPEG or Theora, it's probably patented in a way that makes it impossible to ship in free distros. Which audio codec? Better hope it's Vorbis, or maybe Speex.

            This has nothing to do with copyright and everything to do with software patents.
            You really are reaching for straws here. You guys have already accepted vdpau and va-api into the fold. You guys are providing access to the capabilities, what xyz decoder group does with it really isn't of your concern nor holds you liable. It does not matter if you do it through propriatary engines or open standards and solutions such as openCL (no matter if they are less efficient then addressing hw specific solutions as long as they are more efficient then current raw power cpu) providing a desirable end user experience as long as it's a feature that truly can the "make or break" difference. If you want, absolutely go ahead and accelerate things like Theora and Dirac. At least then your providing users with a reasonable alternative but alas no real such effort has been made (except again on nvidia hardware in the case of dirac). If your going to worry about potential illegal uses you might as well get rid of every debugger, every compiler, every piece of hardware support as pretty much every aspect of computing can be used in "illegal" practices.

            Comment


            • #36
              Originally posted by MostAwesomeDude View Post
              You may want some context to go with your assertion. Since you likely don't know it, I'll provide it.
              I know the context very well, this is not the first time that items like this has happened either, just one of many over the years of subscribing to x86free and xorg mailing lists. On the subject of the patch I did and still do find it absolutely ludicrous that "maintaining" a condition in a 6 line patch was a burden as if it broke pretty much every other condition would follow as well regardless of it's inclusion or not.

              Comment


              • #37
                Originally posted by deanjo View Post
                So you hope ATI gets a monopoly. That's real smart isn't it. But saying peace and freedom comes from the extermination of others is not a new concept to your country is it?
                no i only hope Intel lose his graphic chip monopoly of 60% of the market share.

                i know you deanjo you work for an despicable company with the name "Apple"

                this toy company Apple supports the Monopoly of Intel by using this intel-shit cpus!

                Comment


                • #38
                  Originally posted by birdie View Post
                  Dude, you need to grow up and become sober.
                  the Bio-chemical process of grow up and learning is only 1 function in our brain!

                  bain-cells die a baby do have much more brain-cells than a older-human.

                  so you only tell me to kill some Brain-cells our of my head.



                  o well my head is not so full of cocaine and alcoholic brainwashed to being like you wana me to be.

                  Comment


                  • #39
                    Originally posted by RahulSundaram View Post
                    In short, reward good behaviour with your money.
                    Totally agree. Its so sad we've had to drag these hardware manufacturers and proprietary software conglomerates kicking and screaming into the 21st century. It will only happen if users continue demand support for free software.

                    I bought my 4870 6 months ago specifically because I knew that either xf86-video-ati or radeonhd would eventually be the best drivers available. I see having open-source drivers as adding significant value to the product. I probably won't buy Nvidia again unless they have a similarly well supported open-source offering.
                    Last edited by Smorg; 12-12-2009, 10:14 AM.

                    Comment


                    • #40
                      Originally posted by tball View Post
                      Uhm most of us end users are able to run those HD video contents just fine. Recently it works just fine with fglrx also

                      I would choose xrandr over hd video decoding any day.
                      Why would I need hd video decoding, if I weren't able to hotplug my plasma tv to my computer, if I already are able to run HD contents decoded with my cpu?
                      Try to play HD content in linux on a laptop with a non-vdpau capable system and then play it on a vdpau system. I dare you to compare the battery life and playback experience. Same goes with HTPC's where efficient quiet cooling is desired where playback of HD content as well as doing a plethora of other tasks such as realtime recording during HD playback. Such a task is well beyond the capability of non accelerated HD playback but is perfectly capable of doing so with a low end dual core (single core if not encoding to HD) and a vdpau card.

                      Comment


                      • #41
                        Just a quick comment on the "screaming that all we needed were specs" subtopic.

                        If you focus on what the developers said, there was no screaming, and their position could be summarized as "so, like, we are going to keep working on the driver anyways but if you gave us some information it would go a lot faster". The developers have been true to their word - we provided information/support and they have used that to push ahead and make remarkable progress on the drivers.

                        Where it gets tricky is when the developers comments have been processed a few times by the "collective intelligence of the internet", and developer comments like "given some time we'll make a pretty good driver" turn into "in a couple of weeks the open source driver will be far superior to the binary driver and fglrx will die Die DIE !!!". Those kind of comments led to some unrealistic expectations in the user community, and initial disappointment about how long the work was taking.

                        Anyways, bottom line here is that the development community has been very consistent in terms of what they asked for - and if anything their progress has been faster than they originally expected, not slower.
                        Last edited by bridgman; 12-12-2009, 12:29 PM.

                        Comment


                        • #42
                          Originally posted by bridgman View Post
                          Just a quick comment on the "screaming that all we needed were specs" subtopic.

                          If you focus on what the developers said, there was no screaming, and their position could be summarized as "so, like, we are going to keep working on the driver anyways but if you gave us some information it would go a lot faster". The developers have been true to their word - we provided information/support and they have used that to push ahead and make remarkable progress on the drivers.

                          Where it gets tricky is when the developers comments have been processed a few times by the "collective intelligence of the internet", and developer comments like "given some time we'll make a pretty good driver" turn into "in a couple of weeks the open source driver will be far superior to the binary driver and fglrx will die Die DIE !!!". Those kind of comments led to some unrealistic expectations in the user community, and initial disappointment about how long the work was taking.

                          Anyways, bottom line here is that the development community has been very consistent in terms of what they asked for - and if anything their progress has been faster than they originally expected, not slower.
                          Of course it would be. These people are finally getting what they wanted. So they work hard to prove that they can use it. And that's probably why they have pretty much exceeded initial expectations.

                          I wish my computer came with an ATI card, but it didn't. So, its merely a matter of waiting for the time to buy a new ATI graphics card and putting it into my system. I hold no faith towards nVidia. They don't really care too much for Linux, and the only reason we have a driver is because DreamWorks, Pixar, etc. all use nVidia+Linux for render farms.

                          Also, they have always been a little curt and mean to me and a few others I know whenever we have sent emails to them.

                          It's like talking to the people at Epic... They just don't like talking to people.

                          Comment


                          • #43
                            Originally posted by Qaridarium View Post
                            the Bio-chemical process of grow up and learning is only 1 function in our brain!

                            bain-cells die a baby do have much more brain-cells than a older-human.

                            so you only tell me to kill some Brain-cells our of my head.



                            o well my head is not so full of cocaine and alcoholic brainwashed to being like you wana me to be.
                            OK. I really couldn't constrain myself from repleying anymore, so here I go...

                            Dear Qaridarium. Your dick probably can't even get aroused yet. Did know that each time you get an orgasm 10.000 brain cells are grown in your brain? Probably didn't knew that, did you?

                            And well it doesn't even fscking matter because humans only use about 3% of all of their brain cells. Except for those who have a photographic memmory, but still not much more than 3%...

                            The amount of brain cells doesn't equal how good it functions. It doesn't correlate with intelligence either.

                            Now that we are talking about brains... did you know that around the age of twelve, the frontal brain part starts to develop. Yup... not any earlyer than around that age. So that means at about twelve years of age, insight starts to develop. That's not yet the case with your brain as we can clearly notice...

                            There is this website called digg.com. You might want to check it out.

                            Kind regards,
                            V!NCENT
                            Last edited by V!NCENT; 12-12-2009, 02:19 PM.

                            Comment


                            • #44
                              Originally posted by airlied
                              Things happen because ppl want them to happen, if we don't have tearfree video on open source drivers its not because nobody knows how to do it, its because nobody considers it worth doing.
                              Except, of course, that you are wrong, unless what you meant was "nobody who would be able to do it considers it worth doing".



                              Originally posted by airlied
                              For example at Red Hat we can't ship a DVD player with our OS, so why the hell would we invest money in tearfree movie playing? Most Linux users playing movie are playing dodgy legal rips of content from other sources, its not something we get much paying customer demand for at all.
                              As a linux user I find that last comment completely out of line. This is the kind of thing I would expect to read from the worst trolls in places like the linuxhater blog, not from an X developer. Interestingly, 'most linux users' is not a category you consider yourself included in. On the other hand, I appreciate the honesty in the first part of the paragraph, and I don't mean it in an ironic way. It's always refreshing to hear from an insider how things work and what drives development; if this was done more often many people would have a more realistic view of what this is about.

                              Originally posted by MostAwesomeDude
                              Which video codec do you use to encode your video? If it's anything but MJPEG or Theora, it's probably patented in a way that makes it impossible to ship in free distros. Which audio codec? Better hope it's Vorbis, or maybe Speex.

                              This has nothing to do with copyright and everything to do with software patents.
                              Should we then interpret under a new light bridgman's comments about how GPU offloading of some stages of video playback will possibly be implemented in shaders under Gallium? A Speex-accelerated box is not what was in my mind.

                              Originally posted by Saist
                              I don't think Desktop use has ever really been a focus of Nvidia's.
                              So how do you read the fact that desktop oriented features like video acceleration were implemented by Nvidia in the linux driver? (not only that, but they were also the first ones, and doing a good job at it) Or what about the fact that the most used cards in linux seem to be Nvidia? Or that Nvidia support for newer kernels and X servers is usually implemented rather quickly? (mind you, business users are not the ones constantly upgrading their systems)


                              Originally posted by deanjo
                              So you hope ATI gets a monopoly. That's real smart isn't it. But saying peace and freedom comes from the extermination of others is not a new concept to your country is it?
                              The guy is a pain, but that wasn't the most fortunate of your almost always sharp and interesting comments, Mr Deanjo.

                              Comment


                              • #45
                                Originally posted by deanjo View Post
                                You really are reaching for straws here. You guys have already accepted vdpau and va-api into the fold. You guys are providing access to the capabilities, what xyz decoder group does with it really isn't of your concern nor holds you liable. It does not matter if you do it through propriatary engines or open standards and solutions such as openCL (no matter if they are less efficient then addressing hw specific solutions as long as they are more efficient then current raw power cpu) providing a desirable end user experience as long as it's a feature that truly can the "make or break" difference. If you want, absolutely go ahead and accelerate things like Theora and Dirac. At least then your providing users with a reasonable alternative but alas no real such effort has been made (except again on nvidia hardware in the case of dirac). If your going to worry about potential illegal uses you might as well get rid of every debugger, every compiler, every piece of hardware support as pretty much every aspect of computing can be used in "illegal" practices.
                                Who are "you guys" here? Dave was pointing out that Fedora (alongside Debian and other large distros) doesn't ship non-free codecs. What actually gets accelerated in Gallium doesn't matter, although the distro maintainers have already emphasized that we need to provide compile-time flags to let them turn off the patented stuff.

                                I, personally, as an amateur, would not mind implementing all kinds of video codecs, but I have better (read: paid!) things to do, and so does everybody else working on this code, with the possible exception of Younes.

                                So, really, if you want this feature, just like with all other features, you can either donate code that makes it happen, or you can wait for somebody to get paid to do it. Until those goddamn patents go away, you can bet on the former.

                                Comment

                                Working...
                                X