On Tue, May 24, 2005 at 04:11:17PM -0400, Timothy Miller wrote:
> I do find that I spend a lot of time trying to squash misconceptions.
> Should I not bother? Perhaps I should recruit some list members to
> deal with that for me. But on the other hand, I LIKE communicating on
> the list.
It's your call. Personally, I understand what's driving the project
decisions a lot better after reading your postings. You may have noticed
that I started agreeing after some key facts came up. But it all depends on
whether you have time.
> Even the first ASIC will have "options" in terms of what external
> chips can be connected. There will be a full 32-bit (or 64-bit or
> 128-bit) video bus on the ASIC. It's there as a bit of CYA in case
> something inside doesn't work right (like if the internal DAC sucks),
> and it's there to let us add features that we didn't want to put
> inside.
That's very much the way I like to implement a spec. Plenty of
future-proofing, as long as it doesn't burden the design task and the parts
list too much. A lot of the options I put in a board don't get used, but
there's usually at least one that saves a re-spin.
> The gist of what I'm saying is that OGP 1.0 could result in a lot of
> different card models.
I think it pretty much has to. Putting in as many flexibility hooks
as the lead frame and design time can afford will stretch the production
life and the payback from that very expensive mask set.
Thinking about internal vs. external DACs... I find it hard to
believe that a set of DACs resident on a chip full of logic can operate as
cleanly as an external DAC. For one thing, the lead frame inductance is
pretty much guaranteed to cause pseudo-random bouncing as the various gates
and registers shove charge around. For another thing, a process optimized
for logic probably isn't going to deliver the most accurate and
fastest-settling analog switches, or the lowest-capacitance circuit nodes.
Good analog processes tend to run at higher voltages than the latest and
fastest CMOS can survive, I think. So I tend to think some users will
happily pay for state-of-the-art off-chip video DACs, and others won't.
Maybe when it comes time to design the second-generation boards
implemented with ASICs, the DACs might be placed outboard to eliminate cable
effects. A DVI-to-analog external accessory might even become a product in
its own right, usable with anybody's fast graphics board.
There might even be a board family down the road, with a larger
outline than the regular one, that carries both the ASIC and one or two
FPGAs to extend the logic itself. That, too, could extend the production
life of the first ASIC.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
