On Monday 14 August 2006 15:40, James Richard Tyrer wrote: > Lourens Veen wrote: > > > > Yes, but my point was that the economics are completely different. > > If they make their CPU docs proprietary then they have to write > > their own compiler and probably their own OS kernel. They'd lose > > their entire market, which wants to run Windows, Linux, or perhaps > > BSD. The cost would be enormous. The risk of relying on external > > suppliers is negligible; external implementations have proven their > > worth. Also, the CPU provides a generic abstraction; the coupling > > between the CPU and the OS is not at all specific (hence Linux runs > > on a whole lot of CPU architectures, and GCC can compile for them), > > so it is technically easy to separate the two. > > > > In contrast, the coupling between a video card and its driver is > > much tighter, the driver is essentially one half of the product. > > They already write drivers, and will have to keep doing so because > > the board producers aren't going to buy half a product, and open > > source developers aren't going to write Windows drivers. Opening up > > the drivers, or properly supporting various OSes other than > > Windows, only costs more money, increases (perceived) risk, and > > yields hardly any rewards, because anything other than Windows is a > > niche market for 3D. > > Sorry but your argument doesn't make sense.
That's quite possible, I've never been good at business stuff. I like playing Devil's Advocate now and then though... > Compare the cost of > writing a driver (which they do) with the cost of converting the > internal documentation to a book (which they would sell at a profit). Compare the benefits of writing a driver (being able to sell millions of items of your hardware for use with Windows, making enough money to keep the business afloat) to the benefits of converting the internal documentation to a book (by the way, wouldn't a book be outdated by the time it came off the press?). It's not really going to be a best seller, and the extra hardware sales due to better driver support on open OSes are negligible compared to the Windows sales. You open yourself up to the risk of being badmouthed in the press by the competition if you publicise the faults in your silicon, and they might take some of your tricks and use them in their own products. Not to mention the risk of patent lawsuits. Even if you do make a small monetary profit on the documentation itself it still seems like a losing proposition to me. > Currently the documentation isn't published because it would contain > trade secrets; it really is that simple. Whose trade secrets? Surely ATI (or whomever) decides whether something only they know should be kept a trade secret or not. As for outside trade secrets, what are they? Macrovision copy protection on TV-out is a well-known example, but what else? What functionality is in that binary blob the Intel drivers have? Lourens
pgpTNlI3nrI2P.pgp
Description: PGP signature
_______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
