Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: perft results (how accurate is accurate enough ?)

Author: Uri Blass

Date: 14:02:40 09/23/04

Go up one level in this thread


On September 23, 2004 at 16:58:33, Roman Hartmann wrote:

>hello all,
>
>while testing my move generator (10x12 array, only legal moves) and comparing
>the perft result from Crafty and my own results I noticed that my results differ
>for some positions a bit. As the results start to differ only at some depth it's
>not that easy to track this bug. Although I'm quite certain that the problem is
>somehow related to EP-moves I wasn't able to solve this riddle yet.
>
>By searching the web I found some websites with perft results from other chess
>engines with numbers even more off than the results from my move generator.
>
>some results -total nodes- from my move generator (proper Results from Crafty):
>rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
>perft 4: 20603 ok
>perft 5: 5072212 ok
>perft 6: 124132506 (124132536) differ
>
>r3k2r/p1ppqpb1/bn2pnp1/3PN3/1p2P3/2N2Q1p/PPPBBPPP/R3K2R w KQkq -
>perft 3: 99949 ok
>perft 4: 4185800 (4185552) differ
>perft 5: 197909832 (197876242) differ
>
>8/2p5/3p4/KP5r/1R3p1k/8/4P1P1/8 w - -
>perft 3: 3017 ok
>perft 4: 46264 (46255) differ
>perft 5: 720964 (720879) differ
>perft 6: 11761790 (11750962) differ
>
>8/PPP4k/8/8/8/8/4Kppp/8 w - -
>perft 5: 1840260 ok
>perft 6: 36177037 ok
>
>Well, as searching for this bug has allready consumed much more time than I ever
>expected it would I'm not sure -after having seen this sites- that the the move
>generator has to be perfect in order to create a working chess engine. On the
>other hand my move generating process is quite complicated and I'm not sure if I
>will still understand all the parts of it in a few months like I do understand
>them now.
>
>What should I do about that? Try to track down this bug at any price or just go
>on an do some work on other parts of the engine and fix this bug later? Or
>should I just forget about the bug at all?
>Thanks for all suggestions in advance.
>
>Best regards
>Roman

I think that you should try to track the bug.

if perft 4 is different than perft 3 is also sifferent in some position and you
may find it and continue to go down until you find position when perft 1 is
different.

Uri



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.