Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About history and aging it

Author: Vasik Rajlich

Date: 14:36:07 03/18/04

Go up one level in this thread


On March 18, 2004 at 05:26:51, Tord Romstad wrote:

>On March 18, 2004 at 03:17:10, Mikael Bäckman wrote:
>
>>I search losing non-captures before losing captures. To fix this, I would need
>>to do an 'opponent-SEE' for all non-captures, which would cost too much, I
>>think.
>
>This depends on where you do it.  There is no reason why your move ordering
>techniques should be the same everywhere in the tree.  It makes sense to do
>more expensive calculations in moves far from the leaves.  For instance, you
>could try to use your SEE for non-captures at all nodes where the remaining
>depth is two plies or more.  I doubt that the cost will be noticable at all.
>
>Tord

Interesting. Offhand I would expect that this effect will be diluted by the use
of internal iterative deepening, ie. the main goal of static move ordering is to
optimize the shallow searches & sub-searches, before good moves get into the
hash table.

To put it another way, if remaining depth>2 and the hash table move failed low,
you're probably failing low.

Probably worth testing though.

Cheers,
Vas



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.