Monday, January 7, 2008
2007 Gothic Chess Computer World Championship - Final Standings
Here are the final standings for the 2007 Gothic Chess Computer World Championship:
1-2, Gothic Vortex, 13.0-1.0
1-2, Variant Pulverizer, 13.0-1.0
3, Tornado, 9.0-5.0
4, SMIRF, 8.0-6.0
5, TSCP Gothic 64, 6.0-8.0
6, fmax4, 4.5-9.5
7, ChessV, 2.0-10.0
8, Zillions 2, 0.5-13.5
It was a more exciting event this year featuring a "surprise element" with the new programs Variant Pulverizer and Tornado making strong showings. Tornado had the unfortunate experience of being paired against the strongest programs early on, so it fought from deep within a hole to emerge into a solid third place showing.
Pulverizer has the best time management code I have ever seen. Literally in ever strong position it had, it would move within a few seconds. When the opponent would seem to be regaining ground, it would make wise use of its time and send the other program back into much more difficult positions. Once it had a solid win, it would never use more than 5 minutes remaining to play out the rest of its game.
fmax was another welcomed edition to the event. Programmed by H.G. Muller, this program contains hardly 2000 characters of source code and was deliberately designed to the be the smallest Gothic Chess program ever! I would like to remind everyone that it's draw against Zillions 2 early on was the result of a very bad bug, and fmax could have finished with 5 points just 1 point behind TSCP Gothic which was the second place finisher in several Gothic Chess Championships! This year, the event featured its strongest competitors yet, so fmax had a tough time dealing with the more elaborate programs out there.
TSCP Gothic is still the fastest searching Gothic Chess program with observed speeds of over 1.1 million nodes per second at times on the 2.4 GHz AMD machine on which all programs ran. Its lower finish this year seems to indicate the pendulum has swung to favor slower, better evaluations rather than faster, brute force methods. Recall even in 2004, it seemed "faster dumber" programs in Gothic were outperforming the slower "more intelligent" ones, but this seems to have reveresed, for the time being.
The SMIRF program had submitted two changes during the event, with unknown overall effectiveness. Had these changes been made earlier, SMIRF may have done better, but time will tell. A 4th place finish in this event is still very respectable. We look forward to the future and wonder if SMIRF will support the Winboard_F protocol, which will no doubt be the future of this annual event.
The ChessV program continues to finish at the "bottom of the barrel" when you consider that Zillions 2 is not a dedicated engine but a scripted, compiled, general-purpose interpretter. While ChessV does allow for the play of many so-called variants, it must play them all poorly based on its nearly dead-last finish in 2007. It is safe to say that ChessV is in need of a complete redesign.