Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Parallel search article RBF

Author: Gian-Carlo Pascutto

Date: 12:10:49 09/11/02

Go up one level in this thread


On September 11, 2002 at 14:58:29, Dann Corbit wrote:


>I saw a serious flaw in the algorithm -- when they search for the opponent
>response.
>
>They examine *only* the best current opponent response to see if they can reduce
>it.  Therefore, if it stays about the same, they will never see a better
>opponent response.  Hence, the algorithm (as printed) has an enormous blind
>side.

I briefly skimmed the paper, but IIRC they also randomly select a move
every once in a while. That will fix the biggest blindnesses, but of course,
but the fundamental problem doesn't go away.

>The think that I thought was interesting was:
>1.  That it works at all (I wonder how that comes about?)
>2.  That there is definitely a linear improvement in new CPUs and 64 CPUs is
>almost exactly twice as good as 32 CPUs.  I think that is pretty astonishing,
>and so however it is that communication happens between nodes should be tried
>in other systems.

It's a natural consequence of the fact that they have (almost) no
synchronisation costs.

--
GCP



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.