Author: Christophe Theron
Date: 20:07:21 04/08/02
Go up one level in this thread
On April 08, 2002 at 21:24:18, pavel wrote:
>On April 08, 2002 at 21:17:46, Harpo wrote:
>
>>Just installed Fritz 7 on my P3 1200. With 190 mgs hash it gets just over 600
>>KN/s (Fritzmark=851), while Junior gets around 1000 KN/s. Is this normal?
>>
>>Harpo
>
>Junior's KN/s is differant from Fritz's KN/s.
>Junior's "depth" is differant from Fritz's "depth".
>
>Little Goliath is the fastest chess program I have seen so far, it searches
>(ussually) an average about 1million nps to 1.5million nps in the middle game,
>and yet its not the strongest chess program out there, I suspect its defination
>of nps is differant from other programs.
>
>bigger nps doesnt necessary mean stronger program.
>
>pavs
The definition of nps is quite standardized by now.
Ply depth is not, you are right, but nps means the same thing generally.
Some programmers have chosen to look at more nodes (position) and to extract
less information from each position.
It is a design choice that has its advantages. Those programs reach very high
NPS, but as you pointed out they are not necessarily stronger that "slower"
programs.
There are also more selective programs that spend more time on each position and
generate only the most relevant moves. Each move takes more time to generate,
but less moves have to be examined. Such a program is "slower" but not
necessarily "weaker".
That's why NPS is a relatively well standardized term, but has little relation
with playing strength.
Christophe
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.