Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Speed vs board representation statistic

Author: Dan Newman

Date: 17:28:22 02/12/00

Go up one level in this thread


On February 12, 2000 at 20:02:26, José Carlos wrote:

>  Many things have been said about board representation (paticularly about
>bitboards) and speed. I think it could be a nice idea (just for fun) to make an
>statistic about what program use wich board representation and its aprox speed.
>Could you post what b.r. do you use in your programs and aprox nps in midgame
>(and the hardware used)?
>  I know it strongly depends on the amount of knowledge implemented, but it
>could be just an aproximation. And it's just for fun... :)
>
>  José C.

It also depends heavily on how good your move ordering is too.  As move
ordering is improved it tends to bump the node rate down.  This is mainly
due (I think) to defraying the cost of move generation over fewer usable
moves.  There is also the extra cost of the move ordering code too.

Anyway, that being said, Shrike (non-rotated bitboard) gets about 490 knps
on WAC on a PIII/500 (solves 267/300 at 1s/posn, 297/300 at 1min/posn).  I
wouldn't call Shrike very knowledge intensive yet.  What I've added so far
has only bumped the node rate down about 8%, and it also hasn't been tuned
much either.

-Dan.



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.