Re: 1teraflops cell processor possible?

"del cecchi" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
>
> "Patrick Schaaf" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]
> > Benjamin Ylvisaker <[EMAIL PROTECTED]> writes:
> >
> > >On Tue, 18 Nov 2003 18:48:39 -0500
> > >Robert Myers <[EMAIL PROTECTED]> wrote:
> >
> > >> Q. Will streaming architectures be used other than in GPU's?
> >
> > >Yes.  As we make computers out of smaller and smaller devices, the
> cost
> > >of communication (in terms of time and energy) grows too much to
> > >continue building completely centralized architectures.  These trends
> > >will only accelerate as we get into funky sub-lithographic nano-wire/
> > >tube technologies.  In fact, I believe that soon we will see the
> fastest
> > >single-threaded, strictly maintain the illusion of one instruction
> > >executing after another-type processor that will ever be built.
> >
> > >How's that for a stake in the ground from a nobody on the internet?
> >
> > For what it's worth: it's on par with other big truths like "in 30
> years
> > there will be no more oil, so we'll have to use new energy forms", and
> > "we have enough weapons of mass destruction to blow the whole planet
> up,
> > we have to get rid of them". Obvious, true, and almost nobody seems to
> > give a damn, as long as the old ways appear to be working
> economically,
> > and as long as the economically successfull are those who make the
> money.
> >
> > Apologies to comp.arch readers for straying from the newsgroup topic.
> > I'll shut up again, now.
> >
> > best regards
> >   Patrick
>
> Like those other truths, "a third the population of the US will starve
> by 1990", "we are heading into an ice age", "we can't do optical
> lithography below the wavelength of light so we better build a cyclotron
> for x-ray lithography"  What would any one want a computer in their
> house for?  There is a market for maybe five computers in the country.
> I could throw in "what'll we do with all those mips" or some others.
>
> We've been going to run out of oil in 30 years for 50 years.
>
> del cecchi

The current rate of laying down new hydrocarbon stocks is quite low. The
current oil stocks were laid down in a much warmer climate. Oil extraction
and usage has not declined, so its hard to see how we won't run out of oil
sometime, although it may be more than 30 years.

Peter