At 12:00 -0500 8/7/12, Fred wrote:
What IS a "supercomputer"?
(some people define it in terms of speed!)
...the obvious question, and one we have been over multiple times.
I'm envisioning, but have nothing like the skills to create,
a 3 dimensional plot. X-axis is year of introduction, Y-axis is
available RAM (or equivalent quick-access memory), Z-axis is MFLOPS
or suitable measure of computation throughput [1]. Dots might be
color-coded by relative price, adjusted for inflation, to essentially
encode a 4th dimension, but one normally correlated to Z-axis. Or one
might use color for one of the other possible axes (bit width, I/O
bandwidth, etc.).
I suspect a surface could be constructed in that plot that
would allow one to reasonably define anything above the surface to be
a supercomputer, while anything below it not a supercomputer. I'm
sure exceptions could be made (Distributed.net? The PlayStation 3
array?) but maybe it would cut down some on the inevitable argument.
Does this already exist somewhere? I note
http://en.wikipedia.org/wiki/File:PPTMooresLawai.jpg
which is on the way toward what I'm thinking of.
[1] Yeah, that's a whole other discussion on its own. And I have
totally ignored the question of I/O capability, so there could be
lots more axes on the plot. I know, it's not an easy question to
answer, no brickbats please!
--
- Mark 210-379-4635
-----------------------------------------------------------------------
Large Asteroids headed toward planets
inhabited by beings that don't have
technology adequate to stop them:
Think of it as Evolution in Fast-Forward.