Author: Albert Silver
Date: 19:21:36 10/17/98
On October 16, 1998 at 17:13:14, Steven Schwartz wrote:
>Just in from Thorsten...
>- Steve
>
>beside my tournament i do proudly present the first tournament game
>..
>
>the machines were both HIGH-END machines =
>
>Pentium2 / 450 Mhz 128 MB hash for the new 32-Bit Junior5
>and the kyrotech AMD k6/2 450 Mhz.
I realize this may be off-topic, but it does relate to computer chess
indirectly. Of course the above machines are the high end machines of the day
(not including multiprocessor machines or the Digital machines), but I was
wondering just where and we are going with all this. Intel has had a vice on the
micro-processor market for quite a while and I just read an interesting article
at Tom's Hardware site (http://www.tomshardware.com) that seems to indicate that
this may change in the very near future with the advent of AMD's K7. Anyone seen
it or know anything about it?
That's next year though. What about later? It seems unbelievable somehow to be
talking about the physical limitations of miniaturization, yet that's already a
topic nowadays. We haven't reached it yet, but if one had begun talking
seriously about this some 15 years back there would no doubt have been a lot of
chortling. I wonder what lies in store once we can't make the damn things any
smaller. Thus far that has been the pivot of technological advance: cramming
more into less, but what about after? Will we slowly (it will certainly take
time) revert to bigger is better? After all, when you can't make things any
smaller, the only way left to increase computing power significantly will be to
just keep on adding more. The problem with this is that while adding more will
indeed render bigger brutes, the efficiency of their growth will be impaired
because it would be the equivalent of merely creating some mastodon with an
infinite amount of microprocessors (take your pick), processors that were meant
to take the most advantage of their individually small size but not of their new
giant collage size. Thus we will have to begin designing gigantic super
processors. Another question will then arise: Now that I have this computer the
size of a planet, I need a program to take advantage of it. Will super power be
enough to reach the mythical (for the moment) true AI? No, of course not. Yet,
the complexity of programs that such computers could run will eventually be
beyond the scope of any single human or team of humans. It will be a bit like
the quandry Hollywood faces. As they are having trouble keeping the degree of
creativity up (a direct consequence of the inability to process the overflow of
information of today), they just keep on adding more and more side-effects and
bigger and bigger explosions. More spectacular certainly, but hardly a
significant improvement. We will be able to push their graphics and simulations
to the limits of reality but what will we do with this, and what about smarter?
So, the computer will have to somehow program itself. We will have to program
ourselves the lowest echelon of true AI, or heuristics, and thus allow it to be
able to push the envelope, an envelope that we will find ourselves incapable of
pushing efficiently ourselves. Perhaps genetics will have a few answers for that
though.
Well, now that I have indeed gone WAY off-topic, I'll just leave it at that.
Won't bother apologizing. I'll just let those interested digest it and comment
if they so desire.
Albert Silver
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.