Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: An Experiment that disproves Hyatt's 1000X NPS Theory

Author: Chan Rasjid

Date: 11:12:11 09/18/05

Go up one level in this thread


I do not know about your experiment, if they could be erroneous by some chance
you are not aware, nps reported out by 200 x etc..

Quote:-
"A lot of STRENGTH is EVALUATION FUNCTION".
Seemingly,logic don't show evaluation to be important as long as you outsearch
others. Just consider how many plies ahead is an eval() good for.

I tested in the past this :-
program A - pure fixed depth search that uses only material and all check-mate.
program B - A + some basic common evaluation (non controversial)

Result( most probable ?)- A wins B (all the time ?)as long as it searches 4-5
plies deeper.

This means there is NO-MATCH whatsoever with a 1 - 1000x nps difference.
Fruit's seems to sacrifice complicated (undue) evaluation in favour of high nps
and makes the top.

So likely, your experiment is not measuring what it is supposed to proof.

Chan Rasjid.



This page took 0.01 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.