Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Chess program similarity experiment (Results)

Author: Robert Hyatt

Date: 19:41:50 01/29/99

Go up one level in this thread


On January 29, 1999 at 11:52:05, KarinsDad wrote:

>On January 29, 1999 at 10:45:51, Albrecht Heeffer wrote:
>
>[snip]
>>
>>After the tournament I burned a CD-R with the complete directory image
>>of all sources, executables and log files. The version on the website
>>'bionic41.exe' is ftp'd directly from the CD onto the web site. It _is_
>>the same version as used during both weekends of the tournament, I
>>can assure you. The SMP code must be very indeterministic if we can't
>>reproduce the moves somehow.
>>
>>Albrecht Heeffer
>
>Actually, it may be extremely deterministic, just not deterministic with the
>environments you are using. To reproduce the results, you may need to run on the
>exact same hardware and have the "opposing player" respond on each move within
>the exact same time frame.

I can guarantee you it is highly non-deterministic, and it was designed that
way for performance.  Practically no places where one processor is 'waiting'.
No 'synchronization' points, etc.  The hash table causes huge non-determinism.
random processor timing resulting in different split points every time you run
the same position.  It is a debugging nightmare, but a barn-burner in terms of
performance...



>
>The number of variables in a chess program is huge. The number of variables in
>an SMP version of a chess program is larger. I realize that when testing most
>non-SMP versions of programs, that the program will often come up with the exact
>same move within 1 minute as it does in 5 minutes (and hence Bruce's test
>suite), however, an SMP version of a program may have different results.
>
>Have you examined the moves that are not corresponding with the tournament
>results to see if the move considered at a lesser time during the tournament is
>different than the actual one made and identical to the one made in the test? I
>am not saying that this is where the problem lies, it's just a possibility for
>some of the mis-matches.
>
>KarinsDad



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.