Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question about using Crafty to debug my program's perft

Author: Keith Evans

Date: 17:24:42 07/16/03

Go up one level in this thread


On July 16, 2003 at 18:32:40, Anthony Cozzie wrote:

>On July 16, 2003 at 17:45:14, Keith Evans wrote:
>
>>I'm not sure if this problem is with my understanding of Crafty, or maybe I have
>>a bad build of Crafty 19.3.
>>
>>On a Linux box I compiled crafty for linux-elf and made sure to define TRACE.
>>(It looks to me like I want that based on looking at the code for OptionPerft.)
>>
>>Then I run crafty (leaving the normal starting position) and type the following
>>commands:
>>
>>trace 4
>>perft 4
>>
>>I captured the output to a file, and edited out everything wasn't a move. I
>>ended up with 197,742 lines even though perft found the correct number (197,281)
>>of leaves. (Note if I try perft 1, 2, or 3 then everything comes out equal.)
>>
>>I'm not sure why there are extra lines in there. (I piped it from sort into uniq
>>and didn't catch any duplicates.) My goal was to sort this list and then diff it
>>against the list of moves that my program generates.
>>
>>Any ideas?
>
>
>Why bother?  If you generate fewer moves, it stands to reason that you will get
>smaller trees and outsearch other engines.  This technique has always worked for
>Zappa :-)
>
>anthony


Right now I'm getting more moves...



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.