Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Minor tablebase experiment

Author: Pete Galati

Date: 23:17:00 03/29/00

Go up one level in this thread


On March 30, 2000 at 02:06:10, Jouni Uski wrote:

>On March 30, 2000 at 01:53:33, Pete Galati wrote:
>
>>On March 30, 2000 at 00:44:30, Jouni Uski wrote:
>>
>>>I run my 43 position endgame test with three different TB settings:
>>>1) no tablebases
>>>2) 3+4 piece
>>>3) 2,8 GB from Chessbase
>>>
>>>   Crafty 17.7      Fritz6a
>>>1)     35/43           39
>>>2)     39              41
>>>3)     39              41
>>>
>>>As You see Fritz6a is really stunning as it equals Crafty even without TBs.
>>>It seems to have a lot knowledge in engine - isn't it unnecessary as tablebases
>>>solve most situations??
>>>
>>>Jouni
>>
>>My opinion is that no, it's not unnecessary, because the game is won or lost in
>>the midgame to a very large extent.
>>
>>Pete
>
>I am meaning endgame knowledge. I wonder, if it make Fritz slower. May be the
>best solution is this: if You have 3+4 piece tablebases, then same knowledge
>is removed in evaluation function. Isn't his just the way Nimzo works?!
>
>Jouni

I don't know about Nimzo.  I think your program has to have endgame knowledge
built into it because it would be imposible (these days) to cover every position
in the tablebases.  Then there's a matter of being practical too, in order to
have the 3 & 4 man tablebases in their compressed form, you only need about 30mb
of spare space to store them, but I'm hearing that that figure goes up to about
6Gb if you want to have all the 5pc tablebases in compressed form.  Allot of us
can't spare that much harddrive space.

Pete



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.