Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Gambit Fruit Beta 4bx available.

Author: Ryan B.

Date: 13:36:19 12/05/05

Go up one level in this thread


Yes this is possible.  The eval is still small enough that it is questionable if
LE helps or hurts.  Gambit fruit is also a bit slower now because it does eval
stating at horizon depth 1 now.  I plan to used this extra data for something
productive soon.

Ryan

On December 05, 2005 at 16:28:44, Peter Eizenhammer wrote:

>On December 05, 2005 at 15:58:30, Ryan B. wrote:
>
>>Fixed a bug that hurt play in mid to late games in Beta 4b and added some king
>>safety tuning options.
>>
>>Ryan
>
>Hello Ryan,
>just a little question:
>For some games at 1 min/game + 1sec/move I set chessknowledge to 500,
>so that lazy eval should be off, right?
>The results were surprisingly good, GF4bx was ahead of Fritz9 and Rybka,
>and face to face with Sh9CB, in 20 game matches.
>(few games, short time control, of course ...)
>
>I dont really think to have found something, but nevertheless the question:
>Is it possible that lazy eval does hurt for the new version(s), with the new
>knowledge implemented?
>(I expected the chess knowledge 500 to work clearly worse, but play looked
>really good for bad human watcher.
>I wanted to ask, because it happens much more often that a new "setting" hurts
>strength.)
>
>It is of course possible that results were lucky and my impression completely
>wrong,
>but maybe you have an opinion about this lazy eval issue?
>
>Peter



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.