Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Attack Tables

Author: Vincent Diepeveen

Date: 06:26:14 06/26/01

Go up one level in this thread


On June 25, 2001 at 14:00:31, Dan Andersson wrote:

>This is called AttackBoards and IMO its the best Bitboard implementation for the
>32 bit architectures. It greatly speeds up SEE and can be used to greatly
>improve and speed up move ordering and evaluation. The concept of control can be
>implemented cheaply. One issue is if it is cheaper to implement a indirect
>attacks board or compute them on the fly.
>
>Regards Dan Andersson

How many can you generate a second of these after say 1.e4,e5 2.d4,d5?

I can do a million a second here and that's non-bitboards in C,
not even assembly. Though GCC converts a few things automatically
to IA32 here:

int board[64];
int piecelist[2][16];
and so forth.

Of course i can't use your attackboards, too little info. I also
need the number of attackers at each square and whether a knight
is attacking here, whereas you have this information only indirect,
which is extra instructions all over a 100000 clocks evaluation.
So not going to be rewarding to generate that 'on-the-fly' from
an attackboard.

Note that in my attacktables i also have this info at the upper 16 bits
(only for 1 side of course).

But i'm interested in knowing how fast a bitboarder at a 32 bits
cpu can generate attackboards!

Best regards,
Vincent







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.