Author: Enrique Irazoqui
Date: 03:29:56 05/02/01
Brain Games Network, the organizers of the Man vs Machine match scheduled for October 2001 in Bahrain, asked me to play a qualification event among the strongest chess programs to determine which will play Mr Kramnik. One important condition specified by BGN was that the programs should be able to run on an eight processor machine which will be used in the match in Bahrain; another was that the Qualifier must be concluded before the second week of May, in order to meet contractual agreements with the BGN world champion Vladimir Kramnik. After careful consideration and consultation with the sponsors and organizers, it was decided that the two top programs on the International computer rating lists, as well as the winner of the ICCA computer chess world championship 1999, should be invited. All three programs have SMP capabilities and have played in competition running on multiple processors. The programs Tiger and Rebel, distributed by the Rebel company of Mr Schröder, have never been available in SMP versions, nor have they played in competition on multi-processor machines. Rebel is a DOS program and as far from multiprocessing capabilities as imaginable, while the author of Chess Tiger, Mr Christophe Théron, had explicitly stated in public forums that he did not intend to develop an SMP version in the near future. For this reason they were not considered in the BGN invitation list. When the invitations were made public, Mr Schröder and Mr Théron were understandably disappointed. Out of an old friendship with Ed Schröder, on April 8th I wrote him an email explaining the SMP requirement. I told him this message was personal. On April 20th, only four days before the starting of the Qualifier, Ed Schröder suddenly stated in a group email that both Rebel and Gambit Tiger had SMP implementations. This sounded extremely implausible, especially since a few days previously Ed Schröder had stated in a public forum: "to make a good SMP chess program you need to rewrite major parts of a chess program from scratch. Before you have a reliable multi-processor chess program it will cost you a full year. Why else do you think that nowadays SMP programs do NOT dominate?" (at the time he was arguing that single processor programs should be allowed in the BGN qualifier). And in another public forum, Christophe Théron had described all the computers he owned, and did not include a multi-processor machine. In spite of my doubts, I agreed to examine a copy of the SMP version of Chess Tiger. However, when I asked Mr Théron to submit an SMP version of Tiger by April 23d, 24 hours before the starting of the tournament, Mr Schröder replied that "Christophe as I already suspected is not able to produce the FINAL SMP version to your suggested dead-line." This was, of course, quite unacceptable to the sponsors, organizers and Mr Kramnik, who had fixed all dates. It was also impossible to include a program that was not only unknown to everyone until April 20th, but not even completed and had not yet played a single game in any public event of any kind. GM Raymond Keene and Byron Jacobs of BGN, as well as Thoralf Karlsson and Bertil Eklund of the SSDF, were on the CC line of the above exchanges. As a personal matter I am convinced that neither an SMP Tiger nor an SMP Rebel were ready for the Qualifiers. However, Mr Schröder and Mr Théron have a chance to refute my conviction by playing with their SMP programs in the International CSVN computer tournament in Holland in the second half of May 2001. Enrique Irazoqui
This page took 0.02 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.