Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: you make the call ....

Author: Robert Hyatt

Date: 19:04:51 02/22/06

Go up one level in this thread


On February 22, 2006 at 19:33:08, Mike Byrne wrote:

>Today  I ran this little tournemant -- hoping it would help me decide which
>version of Crafty to run in CCT 8
>
>Standard Crafty 20.3 or Crafty 20.3 SE
>
>Crafty 20.3 SE essentaily has a modified search extension limiter (found in
>chess.h) and a 3 way null mover instead of the standard 2.(R may equal 2, 3 or 4
>-  dependng on search depth and pieces on the board.)
>
>I was hopping for a decisive result - playing a 3 way match between Standard
>Crafty 20.3, Crafty 20.3 SE and Glaurung-AMD64 (very nice program Tord!)
>
>   Engine           Score              Gl             Cr             Cr    S-B
>1: Glaurung-AMD64   14.5/28 ·············· ==00111=1==101 ==00=01100=1==  200.00
>2: Crafty 20.3b     14.0/28 ==11000=0==010 ·············· ==1=00=11=1==1  194.50
>3: Crafty 20.3b ORI 13.5/28 ==11=10011=0== ==0=11=00=0==0 ··············  193.00
>
>For blitz, this is very good Crafty result, Glarung is ranked right between Cm9K
>amd Cm10K default and between Ruffian 2.1 and Ruffian 2.02 and this higher than
>any Crafty currently on the CEFT blitz list.
>
>The dilemma a of course is that A beats B, B beats C and C beats A -- which
>Crafty do you go with for CCT 8 - B or C?


It is actually even more complicated.  I found a bug in the forward pruning and
fixed it today.  I am also currently running some tests to vary the extension
amounts all over the place to find the strongest/fastest combination.  You can
play what you want of course, since you volunteered.  :)

I should have some fairly definitive data by tomorrow on the extension tuning
issues, and will play a bunch of NUNN-type matches against 19.20 to compare...



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.