Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder crushing Chess Tiger.

Author: John Rissler

Date: 19:08:03 12/17/03

Go up one level in this thread


On December 16, 2003 at 01:52:02, Uri Blass wrote:

>On December 15, 2003 at 23:06:35, John Rissler wrote:
>
>>On December 14, 2003 at 18:53:08, Christophe Theron wrote:
>>
>>>On December 14, 2003 at 15:49:38, Richard Sutherland wrote:
>>>
>>>>On December 14, 2003 at 15:39:47, Rajen Gupta wrote:
>>>>
>>>>>On December 14, 2003 at 13:26:33, Richard Sutherland wrote:
>>>>>
>>>>>>Well, I know you all probably know this already, but although Shredder is the
>>>>>>World Champion I am a little surprised at the ease with which it is crushing
>>>>>>Chess Tiger 15 in my mini-match right now. I am playing an auto232 match between
>>>>>>two identical PC's (Intel 2.54Gb) an game in one-hour and Shredder is leading
>>>>>>5-0 with two draws at this time.
>>>>>>
>>>>>>Wonder if there is some tweaking I need to do?
>>>>>
>>>>>although it is along time ago, i think i remember christophe theron saying that
>>>>>due to some kind of bug, chesstiger 15 plays better when moves/min is selected
>>>>>rather than game in/min:
>>>>>
>>>>>the other easo could just be
>>>>>
>>>>>a) a statistical anomaly:
>>>>>
>>>>>b) the effect of he shredder book which is optimised for shredder whereas the
>>>>>tiger book(chessbase)is only a general opening book made form a collection of
>>>>>high quality games:
>>>>>
>>>>>i have found chesstiger 15 to be on par with shreder7
>>>>
>>>>
>>>>
>>>>I've changed the Tiger settings to Suicidal and to use the Shredder book. The
>>>>funny thing is, when you select 128MB of hash, Tiger only loads 96? That's
>>>>consistent on both PC's.
>>>
>>>
>>>
>>>1) I don't recommend the "Gambit suicidal" settings. Use the normal style or the
>>>Gambit style for better results. Please note that if you are using CT15 inside
>>>the ChessBase GUI, due to a bug in the CB stuff you will have to manually edit
>>>the CT.INI file.
>>>
>>>2) Chess Tiger hash table must always be 12 times some power of two, so when you
>>>select 128Mb (which is not 12 times some power of two), Chess Tiger
>>>automatically uses the biggest amount that does not exceed what you asked for.
>>>In this case it selects 96Mb.
>>>
>>>3) If you draw any conclusion after 5-0, my advice is that you should definitely
>>>pick a different hobby. But I'm sure you know that already and that you are not
>>>drawing any real conclusion so early, right? :)
>>
>>I am currently conducting a match in my (limited) spare time pitting Palm Chess
>>Tiger 15 on my Sony Palm Clie 33 MHZ (but overclocked to 54 MHZ with a 1.5 MB
>>Hash table) versus the Saitek RISC 2500 with 128K RAM.  Level is game in 10
>>minutes, normal style for both the CT & SR.  SR has tournament book selected.
>>
>>Anyway, it started out with the Saitek Risc racing out to a 6 1/2 - 1/2 lead at
>>the beginning -- mostly by finding crushing middle-game tactics.  Judging from
>>those first 7 games, it would seemingly indicate that the poor Chess Tiger on my
>>Sony was hopelessly outclassed.
>>
>>After  34 games the Saitek Risc is still leading - but now only by 18-16!
>>
>>I hope to eventually make it a 100 game match.
>
>learning may explain the result.

True, the Palm Chess Tiger has the learning feature, and the Saitek Risc 2500
does not. So, if the learning is effective, than a long match between the two
might be viewed has having an illegitimate result, I suppose.

If I would have stopped the match at 20 games, like the SSDF folks often do, the
Saitek   Risc would

According to the SSDF rating list, however, the CT 15 program on my overclocked
Sony should be slightly stronger.  (At least at 40/2.)

>
>If you draw any conclusion which program is better based only on the final
>result of a match of 10000 games then my advice is that you should definitely
>pick abother hobby.
>
>one program can start by leading 20-0 and 20 draws but after the opponent starts
>to win it can repeat its win every game and win the match 9900-100
>
>Is the winner better when practically people do not play matchs of 10000 games
>and practically the loser performs better against chess programs?
>
>
>Uri



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.