[geeks] Opinions on the HP C3x00?
Charles Shannon Hendrix
shannon at widomaker.com
Wed Mar 29 18:35:22 CST 2006
Wed, 29 Mar 2006 @ 03:14 -0500, der Mouse said:
> Well, sure, but that's like saying that next to a two-year-old, most
> North Americans are fluent in English.
...or worse, that the English are fluent in English.
> The technology exists, but for reasons I do not understand, those who
> make it refuse to release documentation, so good speed is available
> only to those running canned software, like those games.
Fear, patents, and short term thinking.
I've talked to people at nVidia, and I'm pretty sure they are honest
about some of the issues they have with releasing full information.
The biggest problem is that none of the chipset makers own all the technology
they are using. 3D graphics is one of the most heavily patented areas in
technology. They say that releasing hardware specs or just the drivers would
require getting the permission of a number of patent holders, and they simply
won't give it.
One of the primary problem children mentioned by both ATI and nVidia
people is SGI. Evidently they "own" key bits of almost all graphics
chipsets on the market, and won't agree to release details.
The other issue is fear. Most companies are afraid to publish
information because other companies would use it to build competing
products.
They are probably right, but I don't really have much sympathy for them. If
they have to work harder, so be it. I don't think it would hurt them as much
as they say anyway.
It sucks, but that's where we are right now.
The PC graphics market basically has two players now, and some benchwarmers,
and its not a good situation.
The other graphics markets are either priced out of reach, or the older stuff
*still* isn't fully documented to the public at large.
> A cg6 is actually pretty good; it apparently is supposed to do things
> like blits fast enough that the bus to the CPU is the limiting factor.
> (I've never measured it, but an article by the designers said so.)
> Certainly in my experience they've been about as snappy as anything
> gets on an SBus SPARC. Of course, it doesn't satisfy your depth
> demands. A cg14 does, but is insufficiently documented.
If a cg6 is that good, then why didn't anyone ever write fast drivers
for it?
I've never seen one that was very fast, and most of the time they were
painfully slow.
--
shannon "AT" widomaker.com -- ["The determined programmer can write a
FORTRAN program in any language." ]
More information about the geeks
mailing list