Author: Majd Al-Ansari
Date: 09:48:49 01/30/06
Just finished a 504 game tourney using Fritz 9 GUI and Fritz 9 book. Time was 10 min +3 sec per move. Tourney engines were Fritz 9, Shredder 9.1 UCI, Fruit 2.2.1, Rybka Beta 10d. Rybka 10d was used because when I started this tournament 2 weeks ago that was the latest version of Rybka. Results are quite amazing ... especially if you factor in the fact that when analyzing some of the games, I noticed a lot of games that Rybka missed wins in the endgame would be won when using Beta 13. So I would imagine if the tourney was done again with Rybka 13 the results would be even more lopsided. Also Rybka 10d is a huge improvement over the other Rybka's and it wins a lot of positions that were drawn in my first Tourney. Still there are some gaps in the endgame, but they are getting less and less. Mostly glaring would be that Rybka still doesn't understand how to evaluate positions with opposite bishops, and can misplay many such positions. Although they might not add to many ELO points, I would say that in this tourney there were about 20 games for Rybka where it could have done better had it understood opposite bishops. I am analyzing the games now and will post highlights. One other note ... Fruit 2.2.1 is really strong and Fritz 9 is by far the most mature endgame player. Regardless of the results, Fritz seems to understand much better than the other engines when a position is drawn. Here are the tabulated results. I hope the copy paste from Fritz table works !!! 10_3 tournament 2006 1 2 3 4 1 Rybka 1.01 Beta 10d 32-bit 53.0 - 31.056.5 - 27.551.5 - 32.5** 161.0/252 2 Fruit 2.2.1 31.0 - 53.042.5 - 38.542.0 - 38.0 ** 115.5/245 3 Fritz 9 27.5 - 56.538.5 - 42.543.5 - 35.5 ** 109.5/244 4 Shredder 9.1 UCI 32.5 - 51.538.0 - 42.035.5 - 43.5 ** 106.0/243
This page took 0.01 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.