Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hash codes

Author: Antonio Dieguez

Date: 05:35:31 11/15/01

Go up one level in this thread


On November 14, 2001 at 15:53:18, Koundinya Veluri wrote:

...
>Hi Bob,

Am sorry I am not Bob :)

>I have a question. If the hash table index is computed from the hash of the
>position then wouldn't collisions happen a lot more often than if the index is
>computed independantly of the hash key?

Yes. And you are even so right, that the people wich use 64 bits as total hash
signature and calculate the hash index with & and store the FULL key in the
hashtable, is wasting more ram for nothing!. I'm impressed of this because even
Robert Hyatt denied that the other day.

>If two positions have an identical hash key, then the indeces generated from
>this key (say by taking the lowest 8 bits) will also be identical and therefore
>the positions hash into the same index in the hash table, causing a collision.

Yes, in the case of a full key for everything, if a position comes up with a
hashkey that is on the hashtable already then it will collide. So the bigger the
hashtable, the bigger the possibility of collisions as there are more keys
already there waiting to be repeated.

Be well.



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.