Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Nondeterminsitic behavior

Author: Bruce Moreland

Date: 09:06:55 07/19/99

Go up one level in this thread



On July 18, 1999 at 03:01:33, Dave Gomboc wrote:

>In the past, people have discussed how difficult it can be to debug a chess
>program due to non-deterministic behavior.  Clearly, this can arise from a
>multi-threaded search.  If, however, the search is single-threaded (and
>discounting the play from the start of the game until the end of the opening
>book), what factors, if any, might allow a program to still not play identically
>each time?

Retention of data across searches, and timing inaccuracies.  If you try to play
over a game again, it's unlikely that you'll have the same amount of time in
both cases.  And if you don't clear your hash table between searches, who knows
what you'll find in there and how that will affect your search.

The kind of thing I think you are really referring to is instabilities caused by
hash table, pruning based upon search bounds, etc.  This causes strange
behavior, but it should be the same strange behavior each time you do a search.

If you have a single-threaded program and tell it to search 10 plies, then stop
it and tell it to search the same 10 plies again, it should search the same tree
and produce the same result.

bruce



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.