Re: 1teraflops cell processor possible?

In article <[EMAIL PROTECTED]>,
Robert Myers  <[EMAIL PROTECTED]> wrote:
>
>If the DOE, or anyone else, were providing a realistic level of
>funding for basic research in parallel computation, I would not be so
>offended at their throwing however many million at just another big
>machine.

DARPA pumped money into this for years.  The result was, essentially,
that everyone working on anything but tightly-coupled multiprocessors
built to run multiuser Unix workloads faster or special-purpose SIMD
machines threw up their hands, felt embarassed about how much money
they'd wasted, and went home.  Seen a CM2 -- or anything like it --
lately?  What's the last program *you* ran into that was written in
||c?  Do you remember what DADO and NONVON were?  The people working
in the area were no dummies (look what David Shaw's done since) and
they certainly didn't lack funding, but ultimately they beat
themselves up against some very hard problems for long enough, and
that was that.

The ubiquity of clusters, which amusingly share some of the same
constraints as the early massively-parallel machines funded by DARPA
in the 1980s -- a small fraction of the total processing power and
memory at each node, a relatively slow/high latency interconnect --
has led to a resurgence of interest in efficient parallel algorithms,
efficient techniques for programming parallel machines, etc. etc.,
but the problems are still just as hard, and the last time a huge
amount of money was pumped into this, the results were not all that
impressive.  Having been around one of these projects the last time
through, I would not be so sure as you seem to be that the real
problem in parallel computing is that nobody is throwing money at
the software problems.

-- 
 Thor Lancelot Simon                                          [EMAIL PROTECTED]
   But as he knew no bad language, he had called him all the names of common
 objects that he could think of, and had screamed: "You lamp!  You towel!  You
 plate!" and so on.              --Sigmund Freud