Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Root Move Ordering

Author: Ulrich Tuerke

Date: 13:41:39 01/15/02

Go up one level in this thread


On January 15, 2002 at 14:04:24, David Rasmussen wrote:

>On January 15, 2002 at 13:54:56, Frank Schneider wrote:
>
>>
>>thats right. If your timecontrol-mechanism can stop a search before
>>finishing an iteration the opposite seems to be clever.
>>Suppose your first move fails low and your short on time. Then you
>>search the "difficult" moves first (moves which took a lot of time
>>in previous searches) and your timecontrol mechanism stops the search
>>after only searching a few moves. This reduces the chance to find
>>a better move. And even if you finish the iteration you probably search
>>all the expensive moves with a bad search-window.
>>
>>Frank
>>
>
>Any serious program can stop at will whether in the middle of an iteration or
>not. I don't think the advantage of a better move ordering scheme is for time
>control. If one root move ordering leads to half the tree compared to another,
>you will simply search deeper and find better moves, given the same time.
>
>/David

Of course you can stop the search at will. The problem is when you stop before
completing a fail low verification search, then the move failing low is your
best one.
This is a bad move if you are lucky; but possibly a move which throws away the
game.

Uli



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.