From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Tue Feb 25 2003 - 07:25:26 MST
It looks like the march of technology (used loosely as a
combination of hardware and software evolution) is driving
us forward ever faster.
Harder Working Transistors Through Automated FPGA Compiling
http://www.sciencedaily.com/releases/2003/02/030225070139.htm
I went to a conference roughly 4-5 years ago in Paris where
these techniques were just beginning to be explored. It appears
a fair amount of progress has been made.
The bottom line for people unfamiliar with the technology
is that it may allow your computer to "adapt" its hardware
to the application you are running in real time. This is
somewhat different from the Transmeta approach. The Transmeta
optimizations, are I believe, are based on statistical observations
and adaptations of the rather inefficient 8086 architecture
to a more optimal architecture. The FPGA approach actually
involves designing (and manufacturing on the fly) a better
architecture for the application.
Also of interest is "One such system has actually been realized
by an ISI team including Hall and Diniz. That chip, a "Processor
in Memory" chip called DIVA, is now being investigated under a grant
from DARPA for integration into a HP system."
So PIM hasn't "died" -- it has just been hibernating for a while.
This one of the key computer architectural innovations that was
necessary for IBM's "Blue Gene" project. The article suggests
that it may be migrating from supercomputer class machines to
desktop class machines.
The net of the comments is that they suggest that distributed
computing projects, e.g. folding@home, lets-evolve-a-rogue-AI@home,
etc. will become much more capable during this decade.
Robert
This archive was generated by hypermail 2.1.5 : Tue Feb 25 2003 - 07:27:41 MST