No announcement yet.

Why The Open-Source Graphics Card Failed

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    As far as I can tell theosib's and Michael's reports are not contradictory, so in my view both POVs are justified. But even when assuming that Michael got the whole thing wrong because he was a "late comer", it is clear that communication inside the project was pretty bad. Or else these things would have been clear to everybody without misunderstandings. You might try to argue that core people knew about this, but keep in mind this was supposed to be an open source project. Neglecting to inform the rest of the community about the motivations/background of the project just because they joined late is bad enough alone, and given this, it is no wonder that the OGP failed.

    Anyway, I am a hardware designer myself (with the eggshell still on my back, but I think I'm capable), and I've played with the idea of an open graphics card myself. Sadly it involves too much rare knowledge from multiple people to be able to produce such a card compliant with today's standards. A minimum is a good team that knows what it's doing and it is hard enough to gather such people with deep understanding of hardware board and chip design. Especially for no payment. You'd have to have really rare luck to be able to find such people, and then also to be able to gather money for the prototypes. Though in the beginning you can get away with simulations and cheap prototype boards (the tools are free!), then when you can show something that the project has merit and a chance to survive, you could surely find investors. But in the end, when approaching the state where you want to have something commonly usable in the hand, it is going to need a crazy amount of money.


    • #17
      OGP succeeded

      Originally posted by ultimA View Post
      it is no wonder that the OGP failed.
      Arguably, the project didn't fail. We developed a 100% open source FPGA-based graphics card development platform, and then we completed a manufacturing run of them. Also, they work too. You can pop a fully-programmed OGD1 into a PC and boot it up as the console.


      • #18

        Just to clarify a bit, I have been on the Open-Graphics mailing lists since pretty close to the beginning. I can tell you that Traversals role in this whole endeavor was well known and agreed upon by "The List". What was not common knowledge, but no huge secret, was "the secret sellout". There was no "sellout"! Any of the old time list members can tell you that Tim and company were a good bunch of guys to hang out with on the lists.



        • #19
          I read about OGP at the time it was active, that was quite a while ago but I recalled reading at the time that OGP would be used for other purposes besides a VGA card. That said.... with OGP people wanted a fully open card design, the design was implemented, it was fabbed, sold, and worked. The FAQ for OGP (dated 2002-2006) mentions Traversal multiple times, including that OGP is dual license (GPL & commercial) and that Traversal would sell commercial licenses.

          Playing devil's advocate, though, as a GPL'ed design, a company could end up using it without paying a nickel or contributing anything to development. Although actual modifications to the design would have to have source available (per the GPL), it seems to me a on-chip or on-board bus would permit mixture of proprietary and GPL components.

          One thing I can say, although both open source graphics projects did for the most part fizzle, they did both bring awareness to people wanting things to be more open than nvidia and ati were having things at the time.