Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: hanging pieces in rebel

Author: rasjid chan

Date: 23:59:12 01/16/04

Go up one level in this thread


On January 16, 2004 at 04:23:48, Volker Böhm wrote:

>On January 15, 2004 at 18:18:17, rasjid chan wrote:
>
>>The idea of this attack table is only to record whether a piece( type)
>>is attacked by other types and this is usually the most important.
>>It cannot even distinguish if knight/bishop is attacking; this is what
>>8 bits can be can give. The number of attacks must only be a total.
>>
>>Is'nt it very important to know if your queen is threathened by a pawn
>>or a piece is hung - has attacker and zero defender.
>
>There is a good Web-Page with informations about rebell:
>http://members.home.nl/matador/chess840.htm
>
>I use a bit different data-structure:
>1. 1 Bit for the King
>2. 1 Bit for the Queen
>3. 2 Bits for Rooks
>4. 2 Bits for Bishops and Knights
>5. 2 Bits for Pawns
>
>total 8 Bits, but don´t cover if 2 queens are attacking the same sqare. They are
>counted as 1 queen. Worst if 4 bishops and knights are attacking the same sqare
>(verry rare isn´t it). They are counted as one rook (overflow).
>
>I´am not using exact attack-tables to calulating them. When building these
>"eval-attack" tables I am able to "walk-through" pieces. For Example a queen is
>able to walk through a rook. Only point I am missing is i don´t know witch goes
>first.
>
>And yes it is time consuming (haven´t got bit-boards).
>
>Greetings Mangar

I have suspicion that there is no need for other type of attack tables.

reason :-
Ed has been deeply involved with chess programming since day 1 and must
have considered most situations. I seem to find his design
"the most efficient" probably for the use he wanted. I only find minor
deficiency which has a simple solution.

I have no idea of your design nor its purpose. I have doubts about
2 bits for knights/bishop/rooks etc. There should be simple solution
using Ed's attacks if you need to have finer details about
multiple attacks.

I have now fully implemented Ed's attack tables and calls eval()
at every interior nodes thus giving "full attacks" all the time.
That brought down nps critically doing 35000 nps versus 150000 of TSCP
on a P4 1.4Ghz but it has not yet brought down its ELO. So I think
the attack tables actually is paying off in my way of use.




























This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.