Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hashing idea

Author: Bruce Moreland

Date: 12:23:03 02/26/99

Go up one level in this thread



On February 26, 1999 at 15:18:07, Dann Corbit wrote:

>I would have a hundred memory areas per cache or so.
>If you counted all the positions your game has ever played from the first game
>it ever attempted up to now, do you think it would be more than one million
>distinct positions?  If you did have that many, and had one hundred million
>nodes for each of those million positions, that is only 1e14 nodes, if none of
>the surrounding nodes overlapped.  I suspect that most of them do.

Ah, you are talking about machine learning now.  Sorry, I thought you were
talking about all the positions ever *searched*, not *played*.

A secondary table of root positions might be a good idea.  I'm not sure it is
worth using 100x the amount of memory that you use on main hash though :-)
Probably best to use a conventional hash table.

bruce



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.