On 2 Aug 2005 11:58:35 -0000, [EMAIL PROTECTED]
<[EMAIL PROTECTED]> wrote:
> 
> > Investing in Linux (mostly), plus a few other open source technologies
> > is highly profitable to IBM, because it helps them sell hardware.  But
> > would they consider it to be profitable to invest in another graphics
> > chip?  They already have some of their own.
> >
> 
> That's a good question. Is a FPGA more expensive to produce than a dedicated 
> chip?

In volume, yes, by a huge margin.  

> If IBM where to use the OGP in e.g. their servers, the production prize would 
> properly be an importnat factor.

For a variety of reasons.

> 
> > Also, how would the community feel about the degree of corporate
> > control that IBM would inevitably try to wield over it?  If I were
> > them, I would want to make sure my investment didn't get away from me.
> >  But what they want and what everyone else wants may not be the same.
> > Yes, it could be a great opportunity to them, but would they see it
> > that way?
> 
> I guess it depends on, if they have a need for it... Do they make other 
> hardware than servers?

I don't know.  Besides the CPUs and some other support hardware for
z-series and ppc, I'm not sure what else they make.  They do have
research labs, however.

> > At full production, Traversal could make millions/year from
> > OGP-related products.  Is IBM going to care about a few million/year,
> > when they usually try to invest in things that make billions?
> 
> I think they would. The PR that the hardware itsself is Open Source have a 
> value too.

Yes, but to some, that value may be negative.  Less control.

> I could see a banner saying "IBM going all the way." =)

I suppose.  But PR only needs to be so high.  If PR ranges from -10
(SCO) to +10 (Apple, perhaps), IBM should feel good anywhere from +2
to +5.  OGP would only give them, say, a +7, and in only some circles.
 What's more, PR that high is hard to maintain, and if you lose the
hype, it can cost you in PR for a while.  For instance if IBM dropped
OGP at some point, it might bring them down to zero for a while
(although still only in some circles).  PR has an underdamped waveform
and does not have proper signal termination.

> > That's a charity amount, although even a small amount could be a great
> > PR move for them.  PR is a good thing, especially when it comes to
> > OSS.  :)
> 
> So asking for $x with no strings attached could be of great interest at this 
> time?

Sure, why not.  But now, that puts the burden on me to prove that I'm
not misappropriating the money.  It's normal to be skeptical about
what people do with money, which is why I've been hesitant to take
donations up to this point.  I'd REALLY like to have a law firm to
manage it, but that would cost way more than it's worth.  A cost of
$1000 to manage $1 million is good.  $1000 to manage $6000 is not. 
Daniel Phillips has attempted to get us some pro bono legal help, but
he has apparently not been successful.  Even Lawrence Lessig doesn't
want to work for free.

> 
> > Again, large volumes to us are small volumes to them.  However, I
> > wouldn't be surprised if they gave us a deep discount on production.
> > So say it cost only 1 million to produce the 100k quantity.  We still
> > have to get what, to us, is a huge amount of money.
> 
> Is $x or producing the proto type/final equally good?

I don't understand the question.

> I have come to think of an article I once read about AMD, where they told 
> that setting up the prodution for a new chip is the most expensive part.

Yup.  It's like $1 million for the masks and machine setup, but the
chips themselves are dirt cheap.  The packaging is the costliest part
of the chip itself.  :)

> Does anyone have experince on costs of manufacturing FPGA's and PCB's?

My partners do, especially Howard.  $15+ years of ASIC design,
management of hardware design projects, boards, etc.  He even does
that evil black magic that they call "analog".  :) I'm probably
selling him short.

> 
> > IBM, like any good business, believes in what's good for business.
> > Trust me.  They're not into open source on principle.
> [snip]
> > Businesses are often slow to adapt, huge ones especially.  IBM wasn't.
> >
> 
> I tend to forget/(didn't knew) that it always ends down to money.

I'll be perfectly honest with you.  While I don't expect to get rich
from OGP, if I thought it was going to cost me more than I brought in,
I wouldn't be interested.  This is one of those things that SHOULD be
a full-time job, because hardware is much harder to manufacture and
test, so we'd like to become successful enough to make a fair profit
from it.  It's give-and-take, as I see it.  We work to provide the
FOSS community with something they need, and they return the favor by
giving us jobs so that we can continue to make more such products.  It
IS incidental though, because really, we only have jobs if they buy
our products, and they only buy our products if they have value.

> 
> >>What if IBM offered to help on the design instead of giving money? Would 
> >>that be of interest?
> >
> > Maybe.  What would they help with?
> 
> Could it be, that they hold patents for  algorithms for the GPU that could be 
> of interst if they licenced them in the way Linus demands licenses from IBM?

Excellent point.

> Could it be of interest to get their source codes for their graphics drivers?

Unlikely.  Anything we have in common with their design will be the
result of us both working from the same OpenGL spec.

> Or perhaps the diagrams for their graphics cards?

No, because our PCB will be very different from theirs also, and the
DESIGN of the PCB really isn't technically hard (although it is very
tedious).

_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to