Re: 1teraflops cell processor possible?

In article <[EMAIL PROTECTED]>,
Robert Myers  <[EMAIL PROTECTED]> wrote:

>Possibly so, but I haven't engaged in the kind of overblown rhetoric
>the DoE has to justify this machine, whose only real purpose is to
>knock the Earth Simulator off the top spot
>
>http://www.ultrasim.info/esrr_meeting/bland.ppt

I think my major concern around this whole situation is that the
purpose of the Earth Simulator doesn't seem to have been "to knock
ASCI white off the top spot".  I think the really interesting product
of the Earth Simulator so far isn't the machine, or the
fourteen-figure Linpack flops figure, but the annual report

http://www.es.jamstec.go.jp/esc/images/annualreport2002/index.htm

of the kinds of things people are doing with it.  I haven't seen
anything remotely comparable for the ASCI machines; of course the
major application for the ASCI machines does not produce publishable
papers, but I haven't seen even references to a compilation of
"interesting unclassified things done on the ASCI systems".

I may just be ignorant, it may be that the way funding bodies in the
US work mean that interesting calculations done on ASCI would be
dispersed across high-impact-factor journals rather than compiled as a
single annual report; if the "interesting work" compilation exists,
I'd very much appreciate a reference to it.

Tom