Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rybka - Unconfirmed Analysis Bug

Author: Uri Blass

Date: 11:56:42 12/05/05

Go up one level in this thread


On December 05, 2005 at 12:41:58, Steve Lim wrote:

>On December 05, 2005 at 11:56:05, Uri Blass wrote:
>
>>On December 05, 2005 at 11:45:19, Steve Lim wrote:
>>
>>>Hi All,
>>>
>>>I have just started testing the new engine. I can't quite put my finger on it
>>>right now. However I am quite sure that there is a bug when using infinite
>>>analysis under the chessbase GUI.
>>>
>>>When compared to playing a game/match, the infinite analysis mode seems to give
>>>unreliable / weak pvs.
>>>
>>>Please check if you have the same findings.
>>>
>>>In match mode, it seems to be a very interesting engine.
>>>
>>>Steve.
>>
>>please post an example(I do not see it but I did not analyze many positions with
>>it)
>>
>>Uri
>
>[D]r1b1kn1r/2q2pb1/p1p3p1/2P1p2n/Pp2P2p/2N1BP2/1P1QBNPP/R4RK1 w kq - 0 17
>
>From Thomas' example posted below...
>
>Its hard to explain.. seems the hash is inconsistent.
> Try the above mentioned position a few times with cleared hash on infinite
>analysis vs game in 2/hours time control forcing Rybka to move...
>
>I find the 2 are very different.

I only find that Rybka does not find the move with 512 mbytes hash but the
behaviour is the same at 2 hours/40 moves and at analysis mode.

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.