Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fruit 1.5 parameter test extended

Author: Jan K.

Date: 11:48:26 08/27/04

Go up one level in this thread



>I personnally am a bit surprised that enabling/disabling the extensions makes
>pretty much no difference, and would be interested in views as to why this would
>be the case.
>

In blitz I'm not surprised by this at all. I usually run 20+ games 5m+2s against
Fruit 1.5 games every time I test new version of my engine and although I would
expect to win 60% of games just because of missing Fruit's king safety
evaluation, the result is usually very close to 50%. In blitz tournaments that I
run from time to time with Crafty,Fritz8,Aristarch,Gothmog,Arasan,... Fruit
usually ends very high, often second behind Fritz. To me it looks that the major
strenght of Fruit is very well tuned search in combination with it's mobility
evaluation. Pawn push extension is useless if there's no place to use it and can
blow up the search if not used carefully, single reply may not be a big win in
blitz as well. Fruit has clear plan-to outplay you positionally with well placed
pieces, which _by the way_ creates pressure on your position and every  mistake
loses the game easily. Thanks to this, there's low branching factor and while
you evaluate some additional stuff and spend time with extensions, Fruit simply
outsearchs you. Given no chance of tactics you lose positionally. That's just my
opinion based on the games I've seen,nothing more. :)



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.