Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hash Collision

Author: David Rasmussen

Date: 02:32:37 12/08/02

Go up one level in this thread


On December 07, 2002 at 18:30:23, Bruce Moreland wrote:

>
>It's no better to exit than to crash.  You are no less dead if you shoot
>yourself before the impact.
>
>The thing to do in this case is to find some non-awful way to punt.
>
>In my programs, I store a move in the hash table, and give it higher precedence
>when the move generator generates it.  If the move generator never generates it,
>it's as if it didn't exist.
>
>There are other obvious ways to handle this.
>

There is an obvious way of handling it: If the move isn't pseudo-legal, don't
use it at all. There is no hash move. If it happens sometimes (very very rarely)
that some move is pseudo-legal in a different position with the same signature,
we just use it, and have slightly worse move ordering. It doesn't matter. All of
this is trivial.
When I chose to quit rather than just beeping or writing to a log file, it was
just because I didn't come up with the very abstract and brilliant idea of not
exiting, it was because I wanted to know what was happening and how much. Now I
know that it took almost two years before this happened, with many many games on
the chess servers. I wouldn't necesarily have known that otherwise. Also, it was
of course not a problem that I quit, since I just reconnected and the game
continued. I might change it in the future, but now I know something about how
often it happens.

/David



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.