Computer Chess Club Archives




Subject: Re: DB NPS (anyone know the position used)?

Author: Ernst A. Heinz

Date: 14:38:16 01/26/00

Go up one level in this thread

On January 26, 2000 at 16:26:08, Robert Hyatt wrote:
> [...]
>I'm not following this very well.  Are we talking about the _size_ of the
>tree being searched?  Or the size of the tree divided by the time needed to
>search it?

Peter's initial hypothesis was that DB's NPS rate should be almost
constant for any given root position because all the functional units
of the chess chips operate with fixed time complexity regardless of
the given board situation.

I doubted this because of (1) the "fast" and futility-guarded "slow"
parts of the static evaluation and (2) the varying time complexities
for different *compositions* of the basic fixed-time operations at
varying nodes (e.g. node with "stand pat" cutoff is very cheap ==>
the larger the percentage of cheap nodes, the greater the overall
speed in NPS even for DB).

Now, if I understand Peter correctly he thinks that (2) does not
apply because the varieties in time complexity of parent and successor
nodes somehow "level off". I doubt this because I actually cannot
comprehend how this magic balance is supposed to happen.

Still, (1) is for real anyway as Hsu's article in IEEE Micro,
March/April 1999 confirms.

Hence, DB's NPS rate definitely varies for different root positions.


This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.