Author: Christophe Theron
Date: 00:50:21 04/15/01
Go up one level in this thread
On April 14, 2001 at 19:31:48, Bertil Eklund wrote:
>On April 14, 2001 at 17:04:39, Christophe Theron wrote:
>
(snip)
>>I have already given my point of view on this issue.
>>
>>In short:
>>1) I still do not understand the arbitrary choice of Deep Fritz, Deep Junior and
>>Deep Shredder
>
>I have been into this since 1978 with Chess Challenger10. I have been asked as a
>chessconsultant or chess-computer expert, for what it is worth, to recommend a
>challenger against Kramnik to recommend the best programs that can challenge him
>on a SMP-machine, in this case with 8 cpus and I can only recommend three or
>four programs Deep-Fritz, Deep-Shredder and Deep-Junior and of course Deeper
>Blue. Do you have any other serious candidates for this challenge? This is no
>lottery, the challenge is played in "long" matches in difference to the Computer
>World Champion-ships where only nine (or so) rounds are played.
>
>All the above programs are invited because of their status on the SSDF-list or
>their results in tournaments. Yes it is true that one or two programs are
>missing in this case Tiger(s) and Fritz6. It is possible that Tiger or Gambit is
>stronger then the above mentioned programs on a single processor. I don't
>believe that Tiger or Gambit has any serious chances against any of these
>programs on one cpu vs 8 cpus. You know very well that you had been invited if
>your program could use 8 cpus.
No, I did not know. I had never heard about this event until I learned about it
incidentally here, on CCC.
> You know very well that I love the style of
>Gambit and I critisized you for the style of Tiger 12 (a new Genius, very good
>but extremely boring)
>
>>2) Why is the winner of such a closed tournament supposed to perform the best
>>against a player of a totally different kind?
>
>Any suggestions? How can you decide this, if you don't play against humans for
>several months?
There is one company that has invested in challenging strong human players, and
this company is Schröder BV. The results of Rebel against strong humans
indicates that it is a good candidate.
But the problem is that Rebel is stronger against human players than against
computers. So of course it does not have good chances against 8 processors DF,
DS or DJ...
>>Please note that I do NOT question the integrity of the person chosen to do the
>>qualifying tournament (Enrique), and that I do NOT question the fact that these
>>programs are very good.
>
>Good! Yes most people including me, still miss you, but you have told in
>different foras that you are not going SMP at least for the near future.
But against humans style matters more than speed.
How many additional plies are the Deeps going to suck out of their 8 processors,
compared to Tiger running on a single CPU? Probably 1 to 1.5 ply.
How many ply depths do these programs reach typically in the middlegame (at long
time controls on current fast processors): approximately 13 to 15 plies.
Do you think 1 to 1.5 ply is going to make a difference against a guy like
Kramnik? I don't think it will make a difference. That's my point. The
difference can only come from the way the program plays. STYLE is the keyword.
That's the reason why I believe that Gambit Tiger and Rebel should not have been
thrown away so easily.
I believe there are even arguments in favor of Crafty. Because:
a) it can use 8 processors
b) it can run on a 8x Alpha, which according to Bob is better than x86
processors
c) Bob plays a lot on the chess servers and says that Crafty has been adapted to
play against strong humans
But I guess that arguing is futile and the decision has already been taken
behind the curtains.
Christophe
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.