Timothy Miller wrote:
On 3/4/06, James Richard Tyrer <[EMAIL PROTECTED]> wrote:
Attila Kinali wrote:
2) Before any card is sold, we _NEED_ a working, at
least correctly simulating and preliminary tested
PCI interface. Noone is going to buy a PCI card for that
he needs to first write/buy a PCI interface before
he even could think about using it. Also we need
a good way to upgrade the PCI interface in case
a bug is found.
Would it be better to use a PCI chip?
Why are you thinking so short-term?
Perhaps I am actually thinking longer term.
OGC needs a PCI core that has a cost lower than and specs better than
anything we could license on the market. Oh, and OGC needs a PCI core
we could license under GPL. OGC needs its own internal PCI core to
keep costs down and efficiency up.
(And since you're wondering, yes, we would have to charge more for the
PCIe version due to the additional external chip.)
I presumed that we would want AGP-2x/4x for the final product, and that
PCI was just for the development board. Some (i.e. new) motherboards
are going to need AGP-8x or PCI-express so a separate chip would
probably be the way to go for the final design(s). NVida made the
mistake of building AGP into their graphics chips and is now having a
problem with that.
With a separate bus interface chip, we only pay the NRE charge once for
the graphics processor and can change the interface using a separate
chip even it it is a CPLD.
--
JRT
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
