Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Usage of general.ctg book+CB learner by ChessMaster in SSDF testing

Author: Chessfun

Date: 21:55:00 11/01/03

Go up one level in this thread


On November 01, 2003 at 02:49:37, Mike S. wrote:

>On October 31, 2003 at 22:35:22, Chessfun wrote:
>
>>On October 31, 2003 at 21:08:32, Mike S. wrote:
>>
>>>On October 31, 2003 at 20:21:30, Chessfun wrote:
>>>
>>>>(...)
>
>>"(...) So, if you want a possible translations
>>within the GUI's, go always from the Chessmaster-Setting to Malin and not the
>>other way around :-)"
>
>Yes; that last remark is the key sentence regarding this. IOW, the proper way to
>design alternative settings is, create the settings in the CM8000/9000 GUI
>first. For the conversion, usually engine log files are used to see, which
>values are really sent to the engine by the CM GUI. These values are written in
>the Wb2Uci.eng then.
>
>Too bad that Wb2Uci doesn't work for you... I had some difficulties too,
>expecially in *Win98* which I use still on the other computer. I suspect there
>may be something wrong with King's hash size in this OS using Wb2Uci (not in
>WinME). But I didn't test that yet (it's on my todo list... :-)).


Currently I still use InBetween which works fine though it does of course mean
other engines won't run. And also that it's more of a long winded setup.


>Brian Kostick and others of the CCC people had done some efforts to create a
>standard format for the Wb2Uci King settings in Fritz, using the load/save
>function of the F3 dialogue. Unfortunately this didn't raise much interest as it
>seems. But I use it. My Fritz .param file for CM9_SKR looks like this:
>
>[TheKing323]
>My Pawn (myp)=100
>My Knight (myn)=100
>My Bishop (myb)=100
>My Rook (myr)=100
>My Queen (myq)=101
>My Control of Center (mycc)=105
>My Mobility (mymob)=110
>My King Safety (myks)=155
>My Passed Pawns (mypp)=100
>My Pawns Weakness (mypw)=100
>Contempt for Draw (cfd)=0
>Strength of Play (sop)=100
>Attacker/Defender (avd)=-2
>Randomness (rnd)=0
>Opposing Pawn (opp)=100
>Opposing Knight (opn)=100
>Opposing Bishop (opb)=100
>Opposing Rook (opr)=100
>Opposing Queen (opq)=100
>Opposing Control of Center (opcc)=105
>Opposing Mobility (opmob)=110
>Opposing King Safety (opks)=155
>Opposing Passed Pawns (oppp)=100
>Opposing Pawns Weakness (oppw)=100
>Selective Search (sel)=12
>Max Search Depth (md)=99
>Logfile=1
>
>The CM GUI itself is a good testing environment for WinBoard-Eniges IMO, because
>it's possible to assign a CM book to engines which have none (i.e. play a match
>or tournament where all use CM9000.obk). With King there is a top engine
>available to test against, Ruffian 1.0.1 etc. too, for WB. engine testing and
>comparison. Runs great, with the exception that there's no reversed color option
>for the opening variants used. I prefer that over Wb2Uci in general (or simply
>use WinBoard even), although the variety of opponents is limited then and some
>of the strongest unavailable. I guess the new Win-Rebel 12 should also run in
>CM8000/9000 (?), being a WB. engine. There were no reports so far about that
>AFAIK.
>
>I use "CM9K_modified", this is simply 32 MB hash, Sel.=12 and king safety
>settings 150/150. Rest default. This allows easy documentation :-)) but at the
>same time, I believe that I get ~90% of the improvement potential by that. My
>earlier experiments with CM_Brain were not very successfull. Often, only larger
>hash and Sel. 12 is used in ratings lists, but as you know most strong settings
>have the increased king safety values in common, which I consider to be
>important.
>
>An example:
>
>Drei_Chessmeister_9K.CMT  2003
>
>                    1 2 3 4 5 6 7 8
>1   CM9_8777        * 1 0 1 1 1 0 1   5.0/7  16.50
>2   Delfi 4.1       0 * ½ ½ 1 1 1 1   5.0/7  13.75
>3   CM9K_modified   1 ½ * 0 0 1 1 1   4.5/7
>4   CM9 SKR         0 ½ 1 * 1 0 ½ 1   4.0/7
>5   Ruffian 1.0.1   0 0 1 0 * 1 1 0   3.0/7  9.50
>6   Crafty 18.15    0 0 0 1 0 * 1 1   3.0/7  7.50
>7   WBNimzo 2000b   1 0 0 ½ 0 0 * ½   2.0/7
>8   Yace Paderborn  0 0 0 0 1 0 ½ *   1.5/7
>
>40/20:00 on P3/700, own books (Crafty with custom book), but here CM9K_mod. used
>128 MB hash IIRC, although this probably made not much difference.
>
>Btw. unlike previous Chessmaster versions, there was no Chessmaster 9000 sold in
>the normal computer & electronics shops in germany and german speaking
>countries, for the first time! I have my own theory what the reason for this may
>have been :-))
>
>Let's wait and see.

In Canada it was available as always at EB and I'd expect no changes here till
CB open an office ;-)

Thanks for your detailed post Mike.

Sarah.




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.