Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: tablebases Fritz problem

Author: blass uri

Date: 22:03:00 04/04/00

Go up one level in this thread


On April 04, 2000 at 17:32:39, Bertil Eklund wrote:

>On April 04, 2000 at 15:34:59, Jeremiah Penery wrote:
>
>>On April 04, 2000 at 15:15:27, Bertil Eklund wrote:
>>
>>>On April 04, 2000 at 05:18:53, blass uri wrote:
>>>
>>>>On April 04, 2000 at 03:36:48, Bertil Eklund wrote:
>>>>
>>>>>On April 04, 2000 at 02:37:26, blass uri wrote:
>>>>>
>>>>>>On April 04, 2000 at 02:17:14, Bertil Eklund wrote:
>>>>>>
>>>>>>
>>>>>>>>
>>>>>>>>You see the result: worthless.
>>>>>>>>
>>>>>>>>Suggestions:
>>>>>>>>1) only use all 3-man and 4-man TBs and leave the few 5-man from the
>>>>>>>>EG-Turbo-CDs out.
>>>>>>>>2) get/buy/generate the 5-man with queen(s) because of pawn promotion
>>>>>>>>3) get/buy/generate ALL 5-man so you get REAL and MAXIMUM performance.
>>>>>>>>
>>>>>>>>The EG-Turbo-CDs aren't worth a penny. "...contain the most important
>>>>>>>>endgames..." is marketing rubbish as you can see yourself with such games.
>>>>>>>
>>>>>>>Hi!
>>>>>>>
>>>>>>>Totally nonsense, I have followed houndreds of games with the Turbo-CDs and
>>>>>>>usually it works excellent. The CDs contain almost all important endings and if
>>>>>>>you want to earn say a ½ elo or support Seagate or IBM you can fill them with
>>>>>>>the other mostly obscure endgames.
>>>>>>>
>>>>>>>Bertil
>>>>>>
>>>>>>I think that the main reason that it work excellent is that the opponent usually
>>>>>>resigns before going to endgame like KRQ vs KR
>>>>>>
>>>>>>Uri
>>>>>
>>>>>Hi!
>>>>>
>>>>>I haven't checked the above game but I haven't seen a program that can not win
>>>>>with queen and rook against rook.
>>>>>
>>>>>Bertil
>>>>
>>>>The problem is that Fritz does not go to this endgame because this endgame is
>>>>not in the tablebases and it prefers a to do a draw by avoiding this endgame.
>>>>
>>>>Fritz does not know to stop using the tablebases when they tell it that the
>>>>distance to mate does not decrease and because of this reason it needs the
>>>>KRQ vs KR tablebases.
>>>>
>>>>If you do not give it this tablebases then programs that never resign can get
>>>>better result and I do not like it.
>>>>
>>>>I do not think that customers are interested more in a program only because they
>>>>never resign.
>>>>
>>>>Uri
>>>Hi!
>>>
>>>I have checked the game now and the draw has nothing to do with missing
>>>tablebases, Fritz6 draw the game because some error in the tablebase handling,
>>>as you can see the number of moves to mate goes up and down.
>>>I checked the position with Shredder4 (with exactly the same TBs from
>>>Chessbase), and S4 wins without any problems.
>>
>>The mate numbers go up and down because when Fritz could promote the pawn into a
>>won KRQKR ending it didn't, because it did not have the appropriate TB.
>>Instead, it preferred to stay in the TB, even though the distance to mate was
>>not decreasing.  What it was doing was finding the shortest mate it could find
>>(in the TBs it had), and sometimes this was a longer one than on the previous
>>move.
>Hi!
>
>I have checked this with Junior6 and it wins easily, in example move 95 Junior6
>plays Rc7 with aforced win. Same tablebases.
>
>Bertil

I suspect that only fritz needs the KRQ vs KR  tablebases and not Junior.

I generated the KRQ vs KR tablebases and my Fritz6light has no problem to play
different moves and I guess that Fritz6a has also no problem to win if you give
it also the KQR vs KR tablebases.

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.