Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Killer / History

Author: Daniel Clausen

Date: 15:13:01 04/16/00

Go up one level in this thread


Hi

On April 16, 2000 at 15:19:32, Georg v. Zimmermann wrote:

>the program I'm working on uses a simple concept :
>KillerMoves are the moves which are best in the same ply most often.
>History is how often one move has been best in the whole search.
>Is this the most common way to use History/Killers ?

As far as I know, killer moves are the best moves (leading to a cutoff)
on the same ply, which are used most frequently. That's not exactly the
same as "most often". Quite often people use 2 killer moves.

Note that most people only store non-capture moves in the history/killer
table, since the good captures were already tried before.


>I read in some paper ( I forgot the author sorry ) about SSS and others
>that the idea of Killermoves is outdated and almost everyone only uses
>History today. Is that correct ?

That's new to me. I thought most programs still use history AND killer.
I don't see how that can be outdated. I mean if you made a test with
killer-moves and the engine was ~10% faster than without the killers,
how can this be outdated now? (given that the rest of the move-ordering
didn't change much)


>Why do I get so few responses on my post while very qualified programers
>jump on posts like "Whats the optimal hash table size ...?"

I wonder too, but then, it's their choice to answer to whatever question
they want.


>Am I asking too simple questions ? Sorry I don't know better.

Hardly.

Kind regards,
 -sargon



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.