[NTLUG:Discuss] CPU History

[Steve Baker]

David wrote:

> No, NVidia's profits and survival depends on their ability to get me
> to buy their hardware.  Whether or not I can freely develop and
> distribute NVidia drivers, anyone wanting NVidia HARDWARE on which to
> run those drivers would have to pay NVidia.  
> 
> NVidia doesn't make drivers because they want to sell them.  They make
> drivers because they must do so, in order to sell hardware.  That's
> where their money is.  GPL drivers don't threaten the hardware
> business.

nVidia carefully explained their position on this on the OpenGL
game developer's mailing list about two weeks ago.

Their view (as expressed in that email) is that the driver *IS* a
large proportion of the product you buy when purchase one of their
cards.  If they released the hardware specs then a competitor could
develop a register-compatible card and do so for less because they
wouldn't need to develop their own drivers.  Much of what makes their
cards better than the opposition is in the quality and performance of
their drivers.

I'm not sure I buy *that* argument - but I do truly believe they
have good business reasons for not releasing their specs...otherwise
they'd have done so already.  They aren't withholding this stuff out
of malice - or because Microsoft are strong-arming them into not releasing
them...if that were the case then they wouldn't be releasing binary
drivers for Linux either.

> Last year I bought three new systems for my company.  They originally
> were spec'd for NVidia graphics boards.  When I discovered that NVidia
> withheld the technical information necessary to make a driver, I
> changed the order.  My systems now carry Matrox G-450 video cards, and
> use the GPL drivers.  Yes, the NVidia's would have performed better
> than the Matrox cards.  I don't care.  I'm willing to sacrifice that
> performance margin to keep my freedom.

It may be that YOU can get away with the performance degradation - but:

1) The Matrox G450's are probably 20 times slower than the GeForce-4
    ti4600's.  It's not a small differential.

2) Matrox have only released *some* of the register specs - and only
    for the previous generation of hardware.  Hence, the OpenSource
    drivers are unable to expose the neat things these cards can do.

If you are trying to do state of the art 3D graphics - the Matrox
cards are not even close to being contenders.

Of course if you are only going to run simple stuff on them - then
the Matrox solution may work for you.

> To accept the NVidia cards is to live as a bird in a cage furnished by
> NVidia.  It's a very nice cage, with some pretty graphics, but it's
> still a cage.  I refuse to surrender the ability to fly for myself.

It would be nice to have that choice - but realistically, that's not
much of a choice...and for the NEXT generation of hardware with
'Shader' technology - you are doomed.  Not one 3D graphics company
has offered to release the specs of hardware released later than
January 2002.

Give it a year and there won't be any graphics cards on the store
shelves that can possibly have OpenSourced 3D drivers written for
them.

Now what?

   * Boycotting *all* 3D graphics cards isn't an option...not if
     you actually want reasonably modern games - or other 3D apps.

   * There is no way to pressure the vendors into releasing
     the specs.

   * Reverse engineering something as complex as a 3D card seems
     unlikely.

...so - we're screwed.

I *really* wish there was an alternative - but binary OpenGL drivers
are here to stay.

---------------------------- Steve Baker -------------------------
HomeEmail: <sjbaker1 at airmail.net>    WorkEmail: <sjbaker at link.com>
HomePage : http://www.sjbaker.org
Projects : http://plib.sf.net    http://tuxaqfh.sf.net
            http://tuxkart.sf.net http://prettypoly.sf.net