Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Move ordering at root

Author: Robert Hyatt

Date: 11:11:34 04/05/04

Go up one level in this thread


On April 05, 2004 at 12:27:50, rasjid chan wrote:

>On April 05, 2004 at 11:05:30, Robert Hyatt wrote:
>
>Thanks, but I don't need to test yet as you have tested !
>
>I think using nodes searched should be good.
>If a move fails low with very few nodes searched, it probably
>mean being rejected easily, so sort low.
>
>Rasjid

Correct most of the time.  Sometimes a move will have a high node count just
because there are lots of checks and extensions.  But I've been sorting by node
counts for 15+ years with good results, of course forcing the PV move to always
be at the top regardless of its node count...


>
>
>
>
>>On April 04, 2004 at 22:20:35, rasjid chan wrote:
>>
>>>On April 04, 2004 at 14:10:43, Robert Hyatt wrote:
>>>
>>>I do what Renze is doing as I have fails outside window bounds, but I have
>>>not monitored how often it fails low at root below alpha.
>>
>>fails outside window do not give good scores.  To see this clearly, run an N-ply
>>search and print the moves out with their "fail-soft scores".  Then rerun to the
>>same depth, but after searching each move, set alpha and beta back to +/-
>>infinity so that you get a true score for each move.  Compare the ordering.  It
>>will be _significantly_ different...
>>
>>
>>>
>>>Is it found that often there is no scores to sort on?
>>
>>Just wrong scores....
>>
>>alpha/beta fails high on a reasonable move, not always the best move...
>>
>>>
>>>Rasjid
>>>
>>>>On April 04, 2004 at 08:12:50, Renze Steenhuisen wrote:
>>>>
>>>>>
>>>>>Lectori salutem!
>>>>>
>>>>>I was wondering how others are doing the re-ordering at the root-node after an
>>>>>iteration has completed? Because there are some different option actually...
>>>>>
>>>>>I was doing a very simple way of re-ordering, by just ordering them according to
>>>>>the returned scores in descending order (Best Score First, Worst Score Last).
>>>>
>>>>How do you get a score for any move other than the first move?
>>>>
>>>>>
>>>>>The first move to be searched is the best move found on the previous level,
>>>>>sounds logical to me, but what next? I would say that it may pay-off to first
>>>>>search the moves that were best in earlier iterations, before searching the rest
>>>>>of the never-found-to-be-best moves. Finally to have the moves ordered by the
>>>>>returned score.
>>>>>
>>>>>Is this correct? Can I get some good advice on this point? Thanks!
>>>>>
>>>>
>>>>
>>>>
>>>>I don't see where you can get scores to sort on.  I use node counts for each
>>>>sub-tree produced by each root move...
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>Cheers!
>>>>>   Renze



This page took 0.01 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.