Author: Charles Roberson
Date: 09:51:38 06/17/05
Go up one level in this thread
On June 17, 2005 at 08:25:01, Christopher Conkie wrote:
>Know that the winner of the WCCC 2005 will be the best hardware, not the best
>engine.
>
>
>The WCCC will show no insight whatsoever for chess engines abilities while these
>anomalies exist.
>
>The championships are in effect.....meaningless.
>
First, the winner is not guaranteed to be the best hardware.
The mult-deacades long history of the WCCC clearly reveals that
best HW is not guaranteed to win.
Examples:
In the late 1990's Shredder won the championship with a
single proc Pentium 90Mhz machine. It beat a 1000 node
supper computer and Deep Blue. Also, the P 90 was not the
HW that Stefan meant to use. His HW was damaged in shipping
and the P 90 was an old machine sitting in a closet. The best
spare machine they had.
In 2002, Hydra did not win. Also, Lambchop faired well and
did not have endgame tablebases.
On uniform platform events:
Such events would eliminate people that don't develop or don't have
access to that platform. This is most unfair. Example: I believe Tord's
native platform has been the MAC and he did run into several issues
that he did not expect in porting to the PC. This happens to nearly
everybody in porting to other platforms. Thus, even if a developer
ported to the "event platform", he would be at a considerable
disadvantage.
As far as results being "meaningless", this is not true. The point of
it is not to decide what people should go out and buy for their PC. The
point is to provide a competition that will all researchers in the field
to test there efforts.
Example: In the 1970's, there were people using multiprocessor
super-computers. From that sprung research in the best ways to
paralellize MiniMax search algorithms (which are no limited to
chess or games). Because of this early effort on "Advanced HW",
we have the multithreaded programs of today that can take advantage
of the affordable dual-processor machines of today.
If you doubt the advantage of allowing the use of "Advanced HW" as
I just stated. I suggest you read some papers on the efforts of
scientists to create/find an effective and efficient algorithm
that utilizes multi-proc machines for MiniMax with Alpha/Beta
prunning. You'll find that the "obvious" and "not so obvious"
thoughts were quite inefficient. Some of them took 5 processors
just to get a 2x speedup. With many years of research, the more
modern ideas came to be where we get a 1.9x speedup with 2 procs.
Also, these algorithms are quite applicable to other fields of science.
I do not see why you want to hold back progress in these fields inorder
to see which program is best on the PC at this single point in time.
My guess is that you didn't understand the point of the tournament.
Your statements make sense if the point of the tournament is to find
the best PC program. But, that's not the point.
If you wish to find the best PC program, then there are several home
run uniform platform events with full details on the web.
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.