Announcement

Collapse
No announcement yet.

Why The Open-Source Graphics Card Failed

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

  • Why The Open-Source Graphics Card Failed

    Phoronix: Why The Open-Source Graphics Card Failed

    Last week I wrote that the open-source graphics card is dead. The developer behind Project VGA has now written a lengthy email to me to explain why the open-source graphics card is no more...

    http://www.phoronix.com/vr.php?view=MTEwODY

  • #2
    open hardware is very interesting, but very challenging. The entry costs to make anything are huge, so you need big backing before you can even start. I think an open graphics card is probably the most ambitious open hardware project that there has been (even more than an open phone).

    However there are projects that make a good success, for example the arduino, and all the open addon boards. the beagleboard (though it has big backing from TI), rasberry pi, isostick, reprap etc.

    With luck things will get easier and cheaper. ARM chips mean you can do a lot in small space for not too much money. I wonder if someone could put an ARM with a Mali GPU (or several) on a PCI-E card, and make a basic graphics card for 100. It wont impress gamers, but it would be interesting for some.

    Comment


    • #3
      well that and it would have been a much more interesting project if they had the backing to build a full opensource HTPC mobo based around something like the OpenSPARC spec and what the http://www.pchdtv.com/ project was doing.

      Also, what happened to the PCHDTV project? It looks like they're still selling the same PCI HD-5500 card they where like 6 years ago, these days most mobos come with only 1 PCI slot, if that, would hope they would make a PCIe 1x card to stay relevant.

      Comment


      • #4
        Awww, sad.
        But an honest email.

        And yes, even some years ago people said that it was nearly impossible to get a real HW house up and running producing CPUs or GPUs since there is way too high leveled know how etc. involved that you can't just reach within a few months. Plus you need a lot of money to invest first and that takes time until you reach ROI.
        Though that shouldn't mean there wouldn't be an option for simpler chips. Or design a board from ready components (OpenPhoenux and the like)

        Anyway IMO any company should open specs to a degree that people can write a fully working driver for it. Yes.

        Comment


        • #5
          I'm extremely curious as to what that third points company name is. Name and shame? Hell yes. People who know and care about opensource etc. should at least know how to avoid them.

          Comment


          • #6
            Nothing wrong with anonymous contributions to FOSS projects

            I'm the founder the Open Graphics Project.

            I don't fully understand Mr. Meeuwisse's objection. A respectable company was kind enough to contribute substantially to the OGP, but they requested that their contribution remain anonymous. I don't see anything wrong with that. We were very grateful for their help, and we could not have succeeded without it. But we did succeed, and the FOSS community is better off for it, with full design specs for a 100% purely open source graphics card development platform.

            Making an anonymous contribution or code or designs to a FOSS project is no different from making an anonymous, no-strings monetary contribution. But Mr. Meeuwisse seems to disagree for some reason and took it upon himself to "expose" them.

            Comment


            • #7
              Originally posted by oliver View Post
              I'm extremely curious as to what that third points company name is. Name and shame? Hell yes. People who know and care about opensource etc. should at least know how to avoid them.
              No company should be shamed for contributing honestly (anonymously or otherwise) to a FOSS project.

              Comment


              • #8
                How about six words, greed and lack of enough talent

                How about six words, greed and lack of enough talent?

                Those stubborn enough to do it without market leader support don't get enough support from others with good enough talent; there are only a limited number of people who can actually improve on the current state-of-the-art.

                The greed? Well, some of those talented guys will probably be sucked up into the biig machinery. Ok, not greed. Reality.

                Comment


                • #9
                  Greed?

                  Greed? I fail to see greed here. OGD1 is 100% open, that was one of the original goals. That goal has been met. What greed..?

                  Comment


                  • #10
                    wen i read this "Why The Open-Source Graphics Card Failed" i imagine a article about Bridgmans Failure in manage the AMD opensource driver team...

                    Why the HDMI code release failed and why the reverse engineered HDMI support is only stereo sound.

                    Why the Powermanagment failed and why the SPEC is so incomplete that even in the lowest power profile the graphiccard burn more energy than with the catalyst driver.

                    and the APU performance scandal you only get 10% of the performance of the catalyst because the powermanagment for the APUs isn't to save energy its a dynamic Overclocking feature to hit the "maximum performance" no powermanagment means shit performance for the APUs.

                    and it is Bridgman's pure intentional to let the community do everything themselves because they don't write the open source drivers and they manage to release such a incomplete spec to make sure the opensource driver will LOSE the battle against catalyst.

                    It is AMD's desire to prove that open source is worse than closed source.

                    and in the end the UVD unit to make sure only the catalyst get a advantage in the stars microcontoller to transcode with shaders to eliminate overheat to the CPU. and no this feature has nothing to do with any DRM or copyprotection its just the fact that there isn't another microcontoler to handle this in the open-driver

                    it shows to me that they only do an opensource driver to prove it wrong like this "hey watch the closed source driver: better power managment and better performance and less power drain its GREEN to use closed source and so much more features you can't imagine how wonderfull closed source is "

                    amd its only how bad you are and not how bad opensource are. bridgman trash your blueray player and uninstall your windows vista!

                    Comment


                    • #11
                      A vga "card" with 640x480 framebuffer is nothing really fancy. It was a toy project from the beginning it seems. That others used the design to make something more useful i not bad, but maybe they should have given back a bit more infos/sample cards. I dont think it is a huge loss for the community, the new pi seems to be much more interesting in terms of usefullness. Even the oldest gfx cards work in vesa mode and why to create something "new" with lower specs, thats something i dont get...

                      Comment


                      • #12
                        Originally posted by theosib View Post
                        No company should be shamed for contributing honestly (anonymously or otherwise) to a FOSS project.
                        That's not what I read in the article. Did I read it wrongly?

                        Comment


                        • #13
                          ProjectVGA and OGP were mostly independent

                          Originally posted by Kano View Post
                          A vga "card" with 640x480 framebuffer is nothing really fancy. It was a toy project from the beginning it seems. That others used the design to make something more useful i not bad, but maybe they should have given back a bit more infos/sample cards. I dont think it is a huge loss for the community, the new pi seems to be much more interesting in terms of usefullness. Even the oldest gfx cards work in vesa mode and why to create something "new" with lower specs, thats something i dont get...
                          I'm not sure that anyone derived any related works from ProjectVGA. In any case, the PCBs were completely unrelated, and very little FPGA code, if any, was shared between the projects.

                          Comment


                          • #14
                            Originally posted by theosib View Post
                            No company should be shamed for contributing honestly (anonymously or otherwise) to a FOSS project.
                            Without knowing anything about this situation, I will say I think there is a difference between "anonymously" and "secretly". It seems one person is claiming 1 and another the other, and they are quite different.

                            Comment


                            • #15
                              That's what Traversal was for!

                              Originally posted by smitty3268 View Post
                              Without knowing anything about this situation, I will say I think there is a difference between "anonymously" and "secretly". It seems one person is claiming 1 and another the other, and they are quite different.
                              Traversal Technology, the commercial arm of the OGP, was specifically set up to do business in the commercial and military sector, even under NDA, as long as it was legal and ethical. For instance, if we had a military or aerospace customer who wanted to license our technology but do so secretly, then Traversal could handle this. This was part of Traversal's charter from the outset, and this fact was no secret. PART of Open Hardware Foundation's charter was to be a community-representing board to over-see such activities of Traversal without the need to disclose everything, and as such the OHF was privy to all of Traversal's activities, secret or otherwise.

                              The point is that no one should have been surprised that Traversal would have engaged in the sort of business that they were openly chartered to do. Michael's objections exist in part because he was a late-comer, and he missed out on all of the mailing list discussions where these things were discussed at length and fully organized and agreed upon.

                              Regardless of what Michael may have felt, there was no dark conspiracy here. Indeed, it was often mentioned on the mailing list that we did this or that using this particular company's resources, at no cost to anyone but that company. Anyone who was really PAYING ATTENTION (e.g. most people on the mailing list) wasn't surprised in the least that we had collaborated with a company and that we benefitted greatly form this collaboration. Indeed, this company's name was mentioned quite a bit on the mailing list. My relationship with this company was also well-known, as can be seen in the early slashdot and kerneltrap articles about the OGP. At this company's request, we did not talk about some aspects of the collaboration ON THE WEB, and we didn't leverage it for publicity, but on the open graphics mailing list, it wasn't really much of a secret. Indeed, Michael was not the first to notice this relationship, nor was he the first to contact me about it; he was just the first to make wild assumptions about it and blab his misinformed assumptions publicly. (As usual, the truth is much less interesting than the twisted ravings of tabloid journalists.)

                              I'm not mentioning this company's name here because I don't trust that you too won't try to sully their name in exchange for having benefitted the FOSS community immensely. But if you can use Google, you can find out who they are.

                              Comment

                              Working...
                              X