Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Short chess programs

Author: David Hanley

Date: 13:30:24 09/19/02

Go up one level in this thread


On September 19, 2002 at 16:05:22, Uri Blass wrote:

>My guess based on the descreption that you gave is that tscp is better at short
>time control when your program is better at long time control.
>
>I think that tscp only make the illusion of knowledge but it does not know the
>important things so it is possible to have a simpler evaluation that has the
>same quality.

My eval is really stupid, yet i'm surprised by how well the program does.  Just
material + central control + harebrained king safety plays ok once you get a 8
ply seach, but it needs a little more endgame smarts.  The eval is definitely
where i need to do some work.

>
>I think that the fact that your program has better branching factor should be
>decisive at long time control but the fact that tscp searches more nodes per
>seocnd should be decisive at blitz.

I think you're right...  I did a little more testing.. TSCP gets to 4 ply a lot
faster..  They seem to even out between 5-6 ply..  But my program gets to 8 ply
somewhat more quickly.  From the opening position my program searches 200,202
nodes to get 8 ply, while TSCP hits 7,070,387 nodes to get the same thing.  So i
think you're probably right as to where each would have an edge..

dave



This page took 0.01 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.