Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ETC vs. incremental move generator

Author: Stuart Cracraft

Date: 09:40:30 01/16/06

Go up one level in this thread


On January 15, 2006 at 17:25:55, Tord Romstad wrote:

>On January 15, 2006 at 14:18:46, Maurizio Monge wrote:
>
>>BTW, doesn't incremental move generator have a high cost
>>too, as you can't order the generated moves with SEE?
>>If not, why (as IIRC many engines seem to use it)?
>
>I think most programmers who claim to use incremental
>move generators don't really have *completely* incremental
>move generators, but rather what I would call semi-incremental
>move generators.  They don't generate one move at a time, nor
>all moves at once, but use an intermediate approach:  Moves
>are generated in chunks.  One possibility would be to first
>generate and search just the hash table move (if there is any),
>then generate, order and search all captures and promotions,
>proceed to generate and search the killer moves, and finally
>generate, order and search the remaining moves.  I think Crafty
>and Fruit do something similar to this, for instance.
>
>In Glaurung, I don't use incremental move generation at all.  I
>always generate all moves.
>
>Tord

Even when in quiescence you generate all moves?



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.