Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: About debugging HashTables implementation .

Author: Andrew Williams

Date: 14:31:20 04/10/02

Go up one level in this thread


On April 10, 2002 at 15:28:26, Pat King wrote:

>On April 09, 2002 at 12:11:08, Andrew Williams wrote:
>
>>On April 09, 2002 at 11:56:45, Pat King wrote:
>>
>>>On April 08, 2002 at 15:17:46, Robert Hyatt wrote:
>>>
>>>>How do you define "collision"?  Normally this means two different board
>>>>positions produce the same hash signature.  If you are getting _those_ you
>>>>have a Zobrist hashing bug.
>>>
>>>How do you detect collisions, then? Record the entire position in the hash
>>>table? Even for a debug build, that seems like it would be impractical. >>Although undetected hash bugs are even less practical :)
>>>
>>
>>You check to see if the move stored in the hash entry is legal or not. If
>>the hash keys match but the move isn't legal in the position, it means that
>>you have a hash collision.
>>
>>Andrew
>
>But you rely on the hashed positions being different enough for that move to be
>illegal. Certainly, your technique detects collisions, but not all collisions!?
>
>Pat

Yes. I should have been clearer. I don't have any code in PM to detect
collisions. As you say, what I described wouldn't catch all collisions.

Andrew



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.