post new topic

kn/s comparisons with different chess engines

Related Forum Topics:
Rybka 1.1 Vs Lokasoft Engines
Fritz 7 and opening moves
Bug found in updated version of Fritz 7: k...
Unorthodox chess engines for Fritz?
Position with the greatest number of le...
Position with the greatest number of le...


kn/s comparisons with different chess engines - 2008/04/25 23:22 Why does some chess engines go slower than others? For example, fritz 11 seems to run at 1800 or even 2200 kn/s on my quad core system while rybka 2.3.2 MP only runs at 150-200 kn/s.

I have read that some engines spend lots of time analyzing the quality/strength of a position or line rather than analyzing a bunch of bad moves... but even still, Fritz's depth ends up being 2 entire moves more (which is quite massive when you think about the exponential time involved in calculating 2 moves deeper than another engine).

So what is Rybka's problem?



  Popular posts by egervari
Chessbase problems
  | | | post reply
kn/s comparisons with different chess engines - 2008/04/26 07:41 Rybka's true node count is disguised.
Relative node counts can't really be used to compare engines.

Regards, Graham.



  Popular posts by GrahamBanks
CCRL update (30th June 2007)
CCRL update (1st February 2008)
CCRL update (4th April 2008)
  | | | post reply
kn/s comparisons with different chess engines - 2008/04/26 10:25 Ahh.. perhaps you are right.

I did a test. Fritz 11 vs. Rybka 2.3.2 for 15 games. I gave each game 15 minutes. Basically it was even.

Fritz: 2 wins
Rybka: 3 wins

They drew the other 10 games together.

Another question. What is the difference between the LK and the normal versions of the engine? Any real strength difference, or just a style difference? (I notice that some move suggestions are different).



  Popular posts by egervari
Chessbase problems
  | | | post reply
Re:kn/s comparisons with different chess engines - 2008/04/26 20:27 egervari wrote:
What is the difference between the LK and the normal versions of the engine? Any real strength difference, or just a style difference? (I notice that some move suggestions are different).

Not 100% sure to be honest.
LK stands for Larry Kaufmann, so I suspect that there might have been some experimentation with some of his ideas in that version.
The Rybka forum would be the best place to ask this question.

Regards, Graham.



  Popular posts by GrahamBanks
CCRL update (30th June 2007)
CCRL update (1st February 2008)
CCRL update (4th April 2008)
  | | | post reply



© 2008 ChessCircle
Joomla! is Free Software released under the GNU/GPL License.