Shredder wins computer tournament – yet again

by ChessBase
3/3/2002 – It does this with clockwork regularity. At the 11th International Paderborn Computer Chess Championship, which finished today, it was the German program Shredder which finished at the top of the list, with six wins and one loss. Second was Fritz, which beat Shredder in their individual encounter, but was not so ruthlessly efficient against the other programs. More

ChessBase 17 - Mega package ChessBase 17 - Mega package

ChessBase is a personal, stand-alone chess database that has become the standard throughout the world. Everyone uses ChessBase, from the World Champion to the amateur next door. It is the program of choice for anyone who loves the game and wants to know more about it.


11th International Paderborn Computer Chess Championship

Paderborn, Germany, 27.2.2002 – 3.3.2002

Every year this international computer chess event is held in the city of Paderborn, which over the past 30 years has become one of the leading German centers for Computer Technology. This development is closely connected with the name of Heinz Nixdorf. The organiser of the tournament is Ulf Lorenz of the University of Paderborn, who is also the author of the program P.ConNerS (see below).

Organiser Ulf Lorenz

The 16 participants of this year's seven-round tournament are listed below. The winner was the German program Shredder, which scored 6 points. Shredder won all its games except for one, which it lost to Fritz. The Dutch-German program Fritz came second with 5.5 points, followed by Comet (4.5). All three are available in the ChessBase shop.

Winner Stefan Meyer-Kahlen, author of Shredder

Final standings

Click to enlarge

Here are all the games in PGN notation (16 KB zipped)

The official web site


The participants

Design and implementation in 1991 originally guided by GNU chess 3.0. search algorithm based on alpha-beta search, triggered by a home-made modification of the MTD driver. uses 3 hash tables (transpositions, evaluation & table-base-cache). (recursive) Null-Move-Heuristic with reduction=2 and 1. 2 knowledge based cuts a depth 1, positional learning function, compatible to Ch. Donningers Auto232 and Chess232, reads and writes standard formats (EPD and PGN), supports Nalimov's 5-man endgame table bases. Available as freeware (DOS & winboard) as well as a native Chessbase analysis engine.

Started winter 1994 with DIEP. But now i'm busy with this experimental parallel program, it's called DIEP. Still using the same huge evaluation, from which as far as i know it's the most extensive chess evaluation that any chessprogram contains (although mainly middlegame/opening heuristics). Diep is now using depth limited alpha-beta with very little extensions and no other pruning mechanisms than double nullmove R=3, running under linux at a quad xeon from University of Alabama, Washington.

Fritz is build around a selective search technique known as the null-move search. As part of its search, Fritz allows one side to move twice (the other side does a null-move). This allows the program to detect weak moves before they are searched to their full depth. Move generators, evaluation functions and data structures have been designed to maximize the effectiveness of the null-move search. Fritz is the winner of the previous computerchess world championship in Hong Kong 1995. 1993 Fritz tied for 1st place in a Blitz tournament in Munich with the complete world elite. It scored the best computer result in the 1996 man-computer Aegon tournament. In 1998 Fritz was leading the prestigious Swedish rating list. It won an active chess tournament Frankfurt 1998 with a full point ahead of 36 grandmasters. Kind regards Mathias Feist ChessBase GmbH.

Gandalf was started around 1985 by Steen Suurballe. The program was a rule-based selective program, which was very slow, but did surprisingly well. In 1993 Dan Wulff joined in the work, and has been doing the opening library ever since. In 1995 Steen decided to skip the selective search, and concentrate on the evaluation function. The program got much stronger after this change, and although it has become a lot faster than the prior version, it is still rater slow, when compared with other programmes. The search was changed to a standard alpha/beta search, with null-move reductions, and a lot of extensions. The latest version of Gandalf is a WinBoard-compatible engine.

GromitChess is a C++-program, developed in a Linux-environment (Emacs, gcc). It searches about 25000 to 50000 nodes per second on a K6/200 and tries to be intelligent rather than fast. Attacktables are the primary datastructure (16 bit for every square and player; bit n is set if piece n attacks the square). The search uses iterative deepening, PVS, transpositiontables, killer- and historyheuristic, nullmove (R=2), about 10 chess-specific extensions and some pruning heuristics. The quiescence uses a static exchange evaluator and includes some checks and other threatening moves. Parts of the evaluation are initialized at the root but most of the work is done at the leafnodes. You can find more information and executables in the WWW:

Development of Ikarus started in January 1997 when our previous program, named "BasicChess", reached the 64kb memory limit of Borland Pascal 7.0 and its source code had grown completely cryptic. The 32-bit language Borland Delphi 2.0 allowed us to finally use hash tables and the next year or so saw us implement a graphical user interface and most of the usual standard search heuristics (null move pruning, history heuristic, search extensions etc.) as well as some advanced data structures such as a pawn- king hash table. From March 1998 on a Winboard-compatible version has been autoplaying a variety of computer opponents. Ikarus also got a new hand-crafted opening book. Over Christmas 1998 we added support for the endgame databases created by Eugene Nalimov; so our program contains a port of the probing code provided by the author.

In 1991 I started to write my first C++ Project, a Class-Lib for a DOS-Window-Manager-Interface. Inspired from David Levy's "Computer Chess Compendum" (specially the Article about Chess4) i started to write a chess-algorithm in bottom-up manner (beginning with data structures like piece-sets and bitboards and fast assembler routines to modify them). Two incremential updated redundant sets PIECESET _ControlledBy[64] for each square and BITBOARD _ControllTo[32] for each piece are used for move generation and evaluation purposes. The Search is a standard alpha-beta Nullwindow search with Iterative Deepening and several thread extensions and Nullmove. Standard Heuristics like Killer and History are used. The Leave-Evaluation performs several tasks like extension-detection (Kingdanger, passed pawns) and sevaral Mate in one detections. With my own C++ Class-library an implementation of a graphical user interface for the chessprogram was a quite easy task - IsiChess was born. Special Feature is the abilty to play simultaniously with up to ten chessboards in separate windows.

Brutus-Experimental is the new chess engine written by Chrilly Donninger.

Patzer uses the standard alpha-beta PVS search, enhanced by hashtables (4 retries replacement scheme), recursive nullmove (R=2) with verification if only one piece present, special pruning heuristic for ALL-nodes, various extensions. It also uses a special material hash table to adjust the material balance values for certain endgames where the "usual" values do not apply. It values king safety and passed pawns rather high (too high?). It is a incremental bitboard program with attack tables that are also used during move generation and sorting.

P.ConNerS stands for 'Parallel Controlled Conspiracy Number Search'. It has been written by Ulf Lorenz, who is a member of Prof. Dr. Burkhard Monien's research group at the University of Paderborn. U. Lorenz mainly works on the research fields of domain independent selective search in game trees, and on the field of efficient parallel algorithms for optimization problems. P.ConNerS uses a variant of the so called 'Controlled Conspiracy Number Search' algorithm. As a result it examines highly selective and irregular game trees. Evaluations are done by the help of depth 2 alphabeta searches. When it runs on a parallel machine with 60 Pentium 300 MHz processors, P.ConNerS reaches a rate of about 1.2 million nodes per second. In February 1999 it won the 8th International Paderborn Computer Chess Championship.

Winner of the 9th WMCCC and the 9th IPCCC 2000 in Paderborn!
Shredder has started in 1995 as a project at university. Good tournament and test results encouraged me to spend more work in it and lead to the winning of the 1996 WMCCC in Jakarta. Shredder has been commercially available since then and continued to perform very well in computer chess championships. It was 3rd in Paris 1997 and managed to finish as the runner up in the blitz championship there. Shredder is written in ANSI-C and therefore it can easily compiled on various hardware platforms. I think Shredder has good chances in Paderborn this year because it is one of the strongest computer chess programs running on an micro around.

SOS is an amateur program which was started in 1993 and has since then competed in a number of tournaments. The newest version runs on multiprocessor systems with a parallelized version of mtd(f) as its minimax search algorithm. SOS used to be a relatively fast searcher and relied on outsearching the opponent. This has changed now and more knowledge and special cases have been implemented which slow it down. Little effort is spent on the opening book. It plays a very broad range of openings. However it learns to avoid unsuccessful lines and tries not to repeat lost games. It uses publicly available endgame databases.

Reports about chess: tournaments, championships, portraits, interviews, World Championships, product launches and more.


Rules for reader comments


Not registered yet? Register