Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hash Table Collisions

Author: KarinsDad

Date: 08:51:02 04/13/00

Go up one level in this thread


On April 13, 2000 at 02:46:33, Tony Werten wrote:

[snip]
>
>I though about this. Normally this would go ok. But what about search 65 ? Your
>counter gets set back to zero, which is always smaller than the entries in your
>table, so you never get permission to store the new entry.
>

I do not know about the rest of your questions, but this one is easy. You don't
care.

If current counter is not equal to hash counter, then you overwrite the node.
You do not care if the reason that they are not equal is that current > hash or
hash > current.

ANY counter which is not equal to current counter must be an old counter and
hence can be gotten rid of.

If the counters are the same, then it means that either the hash node was
created 64 ply ago (or 128 ply ago, or ...), or that it was just created in this
current search. In either case, you again don't care. You assume it was created
in this search and if you fail to overwrite an old node, oh well. You left an
old node in the table and did not put in a new one. No great loss. The chances
of it happening are so slim that you do not bother to attempt to put in special
code to try to handle it.

KarinsDad :)



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.