Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Branching factor, etc

Author: Omid David

Date: 11:31:33 08/30/02

Go up one level in this thread


On August 30, 2002 at 11:24:50, David Hanley wrote:

>Dan, i really appreciate your help with this.
>
>I should have been more clear.  My sorted search from the opening position was
>an eight ply search.  Here's a six ply search:
>
>Score: -55
>Searched 28744 nodes
>Got 8063 hash hits (28 percent hits)
>There were 16100 quiesce nodes
>The pawn hash contains 2833 entries
>Branching factor is 4.827881
>
>So, that's 28744 total nodes, and 12644 non-quiesce nodes.
>
>According to your list, a six ply search from the opening for the other programs
>reports node counts of:
>
>yace : 34,240
>glc-218 : 12,970
>gnu: 17,175
>phalanx: 9,829
>genesis 40,582
>
>I'm not sure what to think.  Perhaps yace and genesis are counting all nodes,
>and glc,gnu,and phalanx are reporting non-quiesce nodes.  If that is the case,
>then things are going quite all right for me, and i am in the thick of things.
>That makes sense to me, since i think yace is one of the strongest programs, yet
>here it looks like phalanx is beating it by nearly a factor of four in search
>efficency.
>
>But this helps me a lot, i appreciate your taking the time to assist me.
>
>dave

Yes, Genesis counts all nodes (quiescence, leaf, cut, hash, etc). I don't know
whether there are programs that disclude quiescence nodes, but I've come across
many academic articles that count only the "normal" nodes.

Omid.



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.