Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: WCCC: when using elostat ....

Author: George Sobala

Date: 14:36:17 11/30/03

Go up one level in this thread


On November 30, 2003 at 17:16:23, Robert Allgeuer wrote:

>
>1. All games played:
>
>
>    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
>
>  1 Shredder                  : 2889  167 282    13    84.6 %   2593   15.4 %
>  2 Fritz                     : 2810  177 254    13    76.9 %   2601   15.4 %
>  3 Junior                    : 2797  197 277    11    81.8 %   2536   18.2 %
>  4 Brutus                    : 2734  204 311    11    77.3 %   2522    9.1 %
>  5 List                      : 2623  272 272     8    50.0 %   2623   25.0 %
>  6 Green Light               : 2541  247 210    11    54.5 %   2509   18.2 %
>  7 Diep                      : 2510  247 210    11    54.5 %   2478   18.2 %
>  8 Chinito                   : 2463  272 247    11    45.5 %   2495    0.0 %
>  9 Quark                     : 2444  250 237    11    40.9 %   2507    9.1 %
> 10 ParSOS                    : 2422  210 247    11    45.5 %   2454   18.2 %
> 11 Deep Sjeng                : 2375  250 237    11    40.9 %   2439    9.1 %
> 12 Jonny                     : 2370  228 227    11    36.4 %   2467   18.2 %
> 13 Falcon                    : 2368  279 244    10    35.0 %   2475   10.0 %
> 14 Nexus                     : 2306  207 219    11    31.8 %   2439   27.3 %
> 15 Hossa                     : 2087  360 210    10    15.0 %   2388   10.0 %
> 16 Ruy Lopez                 : 2079  529 203    10    10.0 %   2461    0.0 %
>
>
>
>2. With Shredder-Jonny a draw:
>
>    Program                     Elo    +   -   Games   Score   Av.Op.  Draws
>
>  1 Shredder                  : 2845  172 230    13    80.8 %   2595   23.1 %
>  2 Fritz                     : 2802  177 254    13    76.9 %   2593   15.4 %
>  3 Junior                    : 2795  197 277    11    81.8 %   2534   18.2 %
>  4 Brutus                    : 2735  204 311    11    77.3 %   2523    9.1 %
>  5 List                      : 2619  272 272     8    50.0 %   2619   25.0 %
>  6 Green Light               : 2542  247 210    11    54.5 %   2510   18.2 %
>  7 Diep                      : 2512  247 210    11    54.5 %   2480   18.2 %
>  8 Chinito                   : 2464  272 247    11    45.5 %   2496    0.0 %
>  9 Quark                     : 2446  250 237    11    40.9 %   2510    9.1 %
> 10 ParSOS                    : 2424  210 247    11    45.5 %   2456   18.2 %
> 11 Jonny                     : 2404  192 237    11    40.9 %   2468   27.3 %
> 12 Deep Sjeng                : 2381  250 237    11    40.9 %   2445    9.1 %
> 13 Falcon                    : 2369  279 244    10    35.0 %   2477   10.0 %
> 14 Nexus                     : 2312  207 219    11    31.8 %   2444   27.3 %
> 15 Hossa                     : 2094  360 210    10    15.0 %   2395   10.0 %
> 16 Ruy Lopez                 : 2083  529 203    10    10.0 %   2465    0.0 %
>
>
>Robert

These tables show a huge gulf in ELO between the top 3 and the rest - Brutus
huffing and puffing to catch up, the disqualified List bit further behind, and
the rest nowhere! Now most of this difference was due to sheer brute force depth
of search on the multiprocessor rigs beating the single processor machines.

But I bet if you set these engines up to play 16 GMs instead of each other, the
ELO spread between top and bottom would be far smaller.



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.