Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Dann questions

Author: Dann Corbit

Date: 10:01:49 11/25/03

Go up one level in this thread


On November 25, 2003 at 04:26:27, K. Burcham wrote:
>
>If Brutus gets 15 mnps
>900,000,000 npm
>3,600,000,000 in four minutes
>
>If Shredder is getting 375 kns (2166 mhz)
>22,500,00 npm
>3,600,000,000 in 160 minutes
>
>is this math correct?
>are there too many variables to make an estimate like this?
>Is a node a node, node matter whose program is searching?

The nodes of one program are not comparable to another.

The depth of one program is not comparable to another.

There will be some correspondence of some sort.  And the more nodes and the
deeper a program gets the better it is going to do on average.

But one program might call a search 14 plies and another program that searched
just as deeply might call it 12 or 16.  The definitions of these things is kind
of fuzzy, especially when you consider extensions and pruning.

If you just count the wood on the board, you could probably get several million
nodes per second.  But it would play like a piece of crap with sunglasses on.

Or you can add so much knowledge that a five ply search takes an hour.  So you
have to find the happy medium and search smart.

Nodes and plies don't tell the whole story.  But obviously, they are important.



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.