Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: about bug in fruit.2.2(I send bug report to fabien few minutes ago)

Author: William Penn

Date: 19:53:30 12/01/05

Go up one level in this thread


I've used Fruit 2.2.1 to do quite a bit of analysis and have noticed something
similar. It may produce different candidate moves depending on conditions of the
analysis. For example if I reset the GUI or my computer, occasionally a
different candidate move may be produced in the second analysis. I have wondered
if it is a bug, but I can't say for sure.

I compared your two analyses fairly closely and they produce the same candidate
move (1.gxh5) after about 17 ply, or 3 minutes. It seems to me the two analyses
are fairly similar. After 17 ply the second number is diverging (17/51 versus
17/69) which presumably relates to the condition of the memory in your computer
- which involves hash, paging, virtual memory, your hard drive, etc. Because our
computers use virtual memory and paging in a complicated memory swapping system,
it is almost impossible to reproduce conditions exactly.

So I'm not sure it is a bug. It may just be a highly sensitive engine doing its
normal thing, where a little apparent randomness creeps in.

Five years ago with an engine like Chessmaster, I could get two analyses to
agree exactly, but that seems to be a thing of the past. The newer engines are
more unpredictable.

Or possibly Fruit has a little bit of learning built in somehow(?). It wouldn't
take much to make two analyses diverge after several minutes.
WP



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.