Computer Chess Club Archives




Subject: Re: Hash table efficiency

Author: David Rasmussen

Date: 00:39:22 12/09/00

Go up one level in this thread

On December 08, 2000 at 23:04:42, Robert Hyatt wrote:

>On December 07, 2000 at 13:20:29, Miguel A. Ballicora wrote:
>>On December 06, 2000 at 16:17:22, Robert Hyatt wrote:
>>>Repetition is a known hashing "issue".  But in fact, it is there even when
>>>you aren't thinking about a repetition... because what says that you can search
>>>from the node where you get the hash hit, to the terminal node that follows
>>>that deeper in the tree, without repeating something?
>>>It is messy.  I simply ignore it.
>>If I understand what you mean, you ignore the issue altogether right? you don't
>>mean you ignore the repetition. In other words, you find a repetition, return
>>DRAW from search() and also store the previous postion with score=DRAW in the
>>hash table. Correct? If that is true, this is not the origin of my problem...
>This is what I do, yes.  Draw scores are a problem.  Some don't store them.
>Dave Slate was a proponent of that approach.  I _always_ stored draw scores,
>and still do.  Either way leads to errors.  My way leads to a faster search,

How can not storing the draws, lead to errors, and not only ineffeiciency? A
node that isnt hashed but searched, is still "correct", I gather. Whereas, a
node that was stored as a draw because of repetition, and later read in a node
where it isnt a repetition because of different history, will be "incorrect".


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.