> > > Until recently, it's been hard to get graphics cards that work well
> > > with Free Software.
> >
> > You mean there *are* some?
>
> I think Matrox has FOSS drivers for their latest stuff (G550?),
> although not 100% support. And Intel is working on stuff. Also, AMD
> has been talking about it. Also, you can STILL get FOSS drivers for
> older Radeon chips.
>
> I don't want to say there aren't any and get caught in a lie, but it's
> also true that what support there is sucks.
I hadn't heard about Matrox having FLOSS drivers for their latest stuff.
IIUC, the Intel chips are only available on mainboards, not expansion cards.
Given their reputation for buggy chips and poor customer support I'll pass.
I have read about ATI/AMD's talking about better support for FLOSS,
but I haven't found actual quotes. I suspect what they mean is better
binary-only drivers. Probably only for Linux, and surely not for any
CPU other than x86/AMD64.
Maybe something like
We couldn't find any chip newer than the Radeon 9200 with
sufficient documentation to write a fully functional
device driver.
And last time I looked, 9200 are only available as PCI and AGP,
no PCIe cards. In this case, maybe you do mention the year the
Radeon 9200 came out, assuming it was long enough ago to sound
obsolete.
> > > Our design is largely a derivative of 1990's-style fixed-function
> >
> > Given the rate of tech progress, "1990's" makes it sound obsolete,
> > practically prehistoric.
>
> I know. But any other description uses words that many audience
> members won't understand, like "rasterizer and fixed-function fragment
> shader".
>
> How about this:
>
> To figure out how to develop something feasible, we started by
> examining GPU designs from the 1990s and have worked forward from
> there.
>
>
> The problem with this wording is it sounds like I'm equivocating.
An unknown tech word or two probably isn't the end of the world, as
long as you don't overdo it. I assume explaining them would take
too much time.
You don't want to discourage the audience by saying you are
building an open source stone axe. The trick is to find a way
to put a positive spin on something that isn't the latest bleeding
edge, without lying/misleading.
Do you need to include this level of detail in a short general talk?
Maybe spend that time listing things it will do, especially things
that brand X doesn't.
> > > We based our design on the OpenGL 2.0 spec but
> > > restricted the features mostly to that of OpenGL 1.3.
> >
> > Would it be accurate to say "OpenGL 1.3 plus some of OpenGL 2.0" ?
> > Hearing that you are getting "more than 1.3" sounds nicer than hearing
> > "less than 2.0".
>
> Ok. How about this:
>
> We decided that the OpenGL 1.3 feature set would be a good minimum for
> a useful desktop GPU. However, we got all of our information and math
> from the OpenGL 2.0 spec.
>
>
> This sucks too. I'm not sure how to explain that I used the OpenGL
> 2.0 spec as a reference but ignored all of the ARB (programmable
> shader) stuff. We also decidec on a few other restrictions, like
> cutting the number of texture units to 2 from 4. There's also some
> other ambient environment thing we dropped.
Could you say just "OpenGL", without providing the version number?
I'm thinking that the details aren't needed and would be distracting
in a short general talk. You might want to have a prepared answer
ready, in case someone asks in the Q&A.
> > > The next problem is fabrication. Even if we could design
> > > and finish testing OGA1 infinitely fast, it would still take 6 months
> > > to get chips in our hands.
> >
> > Does it really take 6 months to get a chip fabbed? (once it is layed out
> > and ready to go) This is just everyday vanilla silicon, right? Nothing
> > exotic like bipolar ECL, or Gaas, or bleeding edge feature sizes, ...
>
> I think this isn't JUST fabrication of silicon. I'll verify with Howard.
Maybe 6 months for layout plus fab? If you're fabbing in the far east
and talking production quantities, include a month for shipping by boat.
I assume prototypes would come via air.
----
Maybe include something to get the audience thinking about how the
chip/board could solve *their* problem. Invite them to ask if
OGP does/could solve their particular problem. Maybe they don't
care about FLOSS, but do care about purchase price, or power/heat,
or whatever.
Assuming you're still looking to raise money via OGD sales, maybe list
a few things that OGD might be used for. PCI bus sniffer, etc.
You might get questions about the price (OGD, OGC, TRV10).
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
