Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: root search ordening

Author: Robert Hyatt

Date: 09:25:54 05/02/01

Go up one level in this thread


On May 02, 2001 at 11:34:24, Gian-Carlo Pascutto wrote:

>On May 02, 2001 at 11:23:37, Robert Hyatt wrote:
>
>>Depends on how bad you are doing before you fix it.  IE it can be a factor
>>of 100 faster if you are doing a lousy job of ordering the root moves before
>>you make the change...
>
>Ehm, I assumed you are already putting the PV move first of course.
>
>I used to do normal move ordering at the root.
>When I switched to this trick it didn't help very much. Certainly
>not a factor 100.
>
>--
>GCP


As I said, it depends on what you do to start with.  I have seen programs
that did _not_ put the PV move first for reasons unknown.  Or if they are
flip-flopping between two moves each iteration, both didn't appear at the
top of the list for the next search...

The _real_ reason we chose to start doing this in Cray Blitz was to recognize
the fact that some move was threatening to become the best move, but it was
sorted far down in the move list.  If time ran out before we even considered
it the last iteration, we would never know if it might now replace the best
move.  This moved it up near the top of the list where it would almost
certainly be searched in the last iteration...



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.