Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Storing values for searches which will be interrupted

Author: Harald Lüßen

Date: 12:05:07 03/27/04

Go up one level in this thread


On March 27, 2004 at 11:35:17, Andrew Wagner wrote:

>So I was going through some stuff using the CCC search engine, searching for a
>better method to debug my move generator. Anyway, I came across this post:
>
>http://chessprogramming.org/cccsearch/ccc.php?art_id=307634
>
>Wow! That sounds *exactly* like what happens to Trueno sometimes! How do people
>get around saving hash values for searches that are going to be interrupted?

Where is the problem? I only store moves and values in the
hash table after searching the move. It may have been a fail
high or all moves in a position were examined. Then I don't
lose any information in case of an interrupt. Another point
ist that the search can only be interruptet at special points.
In my case at the beginning of the search function every few
thousend nodes I look for an interrupt. When it occurs I throw
an exception that leads me to the appropiate main function
down the stack where it is catched (C++). There is no store
action.

Harald



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.