Author: Michael Conway
Date: 18:14:34 05/24/01
I noticed the following two posts on rec.games.chess.computer POST #1 I finally received Chess Tiger (text version) via email and proceeded to play several games against ChessGenius for the Palm at various clock settings with permanent brain turned off. These are my observations regarding their respective playing strengths: Tiger has a more comprehensive opening book and is therefore a bit stronger in the opening. This advantage can potentially grow since Tiger's opening book is a separate file which can be augmented in the future. Genius seems to be a little stronger in the middlegame. In my games, Genius typically gained a one- to two-pawn advantage at some point. I wonder if Tiger's hash tables work against it in the middlegame. Perhaps someone can comment on this. Tiger has a significant advantage in the endgame, most likely thanks to its hash tables. In the seven games played, Tiger won six and the seventh was drawn. In one game Tiger gained an advantage in the middlegame and won it easily. In the remainder, Genius had the advantage, but tiger was able to come back in the endgame. One extreme example of Genius's weakness in the endgame occurred in a five-minute match where Genius had an overwhelming two-pawn advantage according to both engines' evaluations with over two minutes remaining. It then blundered into a mate-in-three. The other endgames featured Tiger gaining its advantage slowly but steadily. Looking beyond playing strength, I expect the graphical (commercial) version of Tiger to be more feature-rich than Genius especially when it comes to analysis. On the other hand, Tiger (the text version) requires three-and-a-half times the memory as Genius (270 KB vs. 80 KB). If anyone else has had any experience with the new Chess Tiger for the Palm, please post your observations. Thanks. Robert POST #2 (from the same author) Well, I've played some more Tiger vs. Genius games (yeah, I gotta get a life) and I've concluded that Tiger is the stronger engine. In a 60-minute match, Tiger outplays Genius throughout and beats it rather soundly. In short (five-minute) games, however, Genius tends to get an advantage during the middlegame, but invariably, Tiger comes back in the late middlegame and endgame, as its hash tables take greater effect. Genius also seems more prone to blunders than Tiger. I figure that in the middlegame of a short match, since there are so many candidate moves, the processor time used by the hash table routines is mostly wasted as it is less likely for positions to repeat during the search. This would give Genius the advantage since it is able to see and evaluate more positions. This advantage, however, is lost in circumstances where positions are more likely to repeat during the search, such as in the endgame or at longer time controls. Anyway, that's my take on it. Tiger is the stronger engine, but Genius may have a slight advantage during the middlegame of matches with short time controls. Perhaps someone with more knowledge of chess engine programming can let me know if this analysis makes any sense. Thanks. Robert p.s. I've been playing these games with a 16MHz Palm overclocked to 32MHz with Afterburner. Glad I got these NiMH rechargables! END OF POSTS
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.