RC Macaulay wrote:
Stephen Lawrence wrote
>I'm over at http://www.physicsinsights.org/
The Glue Function is a fascinating study. Did you compose these works
Stephen?
Thanks; I'm glad you like it! Yes, the words are all mine.
The concepts, glue function included, are standard math & physics, but
the point of view is somewhat different from the way these things are
generally presented.
Having mentioned in prior posts my interest in "quadratic computing" (
use of 4 computers rather than 2 ( parallel computing),
4 computers? I don't understand.
Linux clusters commonly use ~ 100 computers in parallel, and the biggest
ones use tens of thousands. Individual processor boards are build with
anywhere from 1 to 16 CPUs depending on the manufacturer (Intel boards
typically have 2 per board; IBM/RS6000 and Dec-Tru64 boards often run
around 8). The processor cores themselves (the individual CPU chips)
have from 1 to 4 or even 8 ALU's depending on the design; I don't know
how many ALUs the current Intel chips use. There are usualy fewer
floating point units, since they're larger and usually not as heavily used.
For ~ $10,000 you can build a Beowolf cluster in your basement with at
least a dozen computers running in parallel (but you'll need some pretty
serious air conditioning). (For $100,000 you can build something
amazingly powerful, capable of simulating spherical implosions to high
accuracy using open-source software downloaded from offshore sites which
DOD can't control ... this is the dark side of tabletop supercomputing.)
