Computer Chess Club Archives




Subject: Re: A move to search 2nd... Keep in the trans table?

Author: Robert Hyatt

Date: 12:17:22 07/22/04

Go up one level in this thread

On July 22, 2004 at 11:13:55, Tord Romstad wrote:

>On July 22, 2004 at 10:39:41, Robert Hyatt wrote:
>>I'll ask again.  _how_ can you use the transposition table to order moves at the
>>root position?  That is a single position.  How many "best" moves can you store
>>in a single transposition position and how do you sort them to figure out which
>>is best, second best, third best, etc...
>For the first move, I use the transposition table (of course).  For the
>remaining root moves, I use exactly the same techniques as in internal nodes
>(captures ordered by SEE values, history heuristic, and so on).  I am sure it
>is not optimal, but I also doubt that I would notice any big improvement if I
>changed this.

OK.  Then you aren't using the hash table to order moves at the root.  You are
using it to order "a move first" at the root.  That was what I thought but it
sounded like you were doing something else.

>>You don't have to have a separate search for ply=1 if you don't want to, while
>>still treating ply 1 differently as I do...
>Yes, as mentioned elsewhere in the thread, I could also add lots of
>if(Ply==0) ... statements to my search function.  Because my code is already
>unacceptably complicated and messy, I don't want to do this.  My new program
>is a different story, I will probably try ordering the moves differently at
>the root there.

Getting the root order right pays off big of course.  Particularly when you are
going to change your mind at the next iteration.  It is far more efficient to
have the potential new best move high in the list to search it right after the
usual best move.

This page took 0.1 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.