Computer Chess Club Archives

Search

Terms

Messages

Subject: Old Bessy tournment...rybka is good at short time controls and lower dep

Author: Joseph Ciarrochi

Date: 01:42:33 02/17/06

I think Rybka is quite good at shorter time controls, at least according to my
tests (see table below). This is using 40/4 repeated time controls, like CEGT
blitz, but i am using slower hardware p3 446 mhz. Noomen test sets. 2650 start
value(older version of engine never plays newer version)

as you can see, there was a big jump in strength between rybka 12 and 13b.

Difference between rybka13b and fritz 9

Cegt 40/40                     = 88
cegt 40/4                      =  38
Old Bessy (my computer) 40/4   =  75

So my results are, at present, a bit different from CEG 40/4 and more like the
40/40. (i'll keep running to reduce error, so all conclusions tentative)
However, as you can also see, fruit does not do well at these time controls on
slower hardware.

Another striking thing about the cegt blitz....rybka 13b is not better than
rybka beta 1........

Program                          Elo    +   -   Games   Score   Av.Op.  Draws

  1 Rybka 1.01 Beta 13b 32-bit     : 2793   35  35   343    71.4 %   2634   21.0
%
  2 Rybka 1.01 Beta 12 32-bit      : 2739   26  26   563    66.8 %   2618   22.7
%
  3 Fritz 9                        : 2718   23  23   706    60.5 %   2644   19.8
%
  4 Fruit 2.2.1                    : 2665   22  22   707    52.1 %   2651   29.7
%
  5 Toga II 1.1                    : 2645   25  25   542    50.8 %   2640   29.3
%
  6 Spike 1.1                      : 2585   37  37   257    40.3 %   2653   26.1
%
  7 Spike 1.0a Mainz               : 2558   27  28   500    35.5 %   2662   23.4
%
  8 List 512                       : 2555   42  43   233    32.2 %   2685   19.7
%
  9 Ruffian 2.1.0                  : 2541   24  24   643    31.0 %   2680   25.7
%

On February 17, 2006 at 04:01:47, Bigler wrote:

>Hi folks,
>
>When comparing the CEGT 40/40 to the CEGT blitz rating list, we see that Rybka
>has a bigger margin at longer time control than at short time control.
>Lot of assumption was made saying that shorter the time control is bigger the
>margin is for Rybka.
>
>According to the CEGT lists, it’s the other way round.
>
>Can we make the assumption that at tournament time control the margin would be
>even bigger ?
>
>Fruit 2.2.1, fritz 9 and Hiarcs 10 Hypermodern are the 3 contenders.

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.