Author: Christopher Morgan
Date: 08:29:01 09/08/04
Go up one level in this thread
On September 08, 2004 at 01:32:48, Kurt Utzinger wrote: >On September 07, 2004 at 20:22:13, Christopher Morgan wrote: > >>On September 07, 2004 at 13:39:07, Kurt Utzinger wrote: >> >>>On September 07, 2004 at 12:32:20, David Dahlem wrote: >>> >>>>On September 07, 2004 at 12:06:03, Christopher Morgan wrote: >>>> >>>>>Eduard, >>>>> >>>>>Interesting results so far. Please post the settings. >>>>> >>>>>In recent tournaments posted here Aristarch 4.50 has had very strong results. >>>>>On my AMD 64 3000+, 200MB hash, at 40' + 30", no tablebases (Pro Deo can’t use >>>>>TBs), I have a 30 game match in Shredder 8 CB GUI, alternating colors, ponder >>>>>off, Pro Deo 1.0 (Rebel 2 setting) v Aristarch 4.50 using the 15 Fischer Random >>>>>Chess openings where the king and rooks (only) are in their classical starting >>>>>chess positions so that normal rules of castling apply. CB/engines not set up >>>>>to play true FRC, however, by using the 15 selected positions only you do get >>>>>true FRC games. After ten games it’s 8-2 in favor of Pro Deo: Pro Deo 6 wins, 4 >>>>>draws, and no losses. Average length of the games is 64 moves which adds an >>>>>additional 32 minutes to each engine clock making max game time for each >>>>>engine’s thinking time about 72 minutes on average. I understand that both Pro >>>>>Deo and Aristarch play better at longer time controls. >>>>> >>>>>This match should be a good test of pure engine strength. >>>> >>>>I'm not so sure about this last statement. I think engines are programmed to >>>>play from the standard start position. For example, they are rewarded for >>>>advancing the center pawns, controlling the center, etc. which might not be best >>>>in FRC!! >>>> >>>>Regards >>>>Dave >>> >>> Hi Dave >>> I fully agree with your statement. FRC is different from >>> classical chess and may not be the correct way to compare >>> the playing strength in matches with classical chess. >>> Kurt >> >>Kurt, >> >>At least in the 15 FRC test positions in blitz games we see the usual suspects >>at the top of the list. See http://www.beepworld.de/members53/frc-list/ This >>leads me to believe that the ability of an engine to analyze is, at least, >>substantially independent of what position it is analyzing for its best move. >>With the typically huge opening books and tablebases we are looking at tactics/ >>middlegames for the most part, not at all unlike FRC. Controlling the center, >>developing your pieces, advancing pawns, are the same in FRC in the opening as >>in classical chess, as well as checkmate, the object of the game. Fischer >>wanted to eliminate the effect of rote memorization of long opening sequences >>and he was successful. He did not disturb the “theory” of chess. >> >>Chris > > Hi Chris > You are perhaps right and from my posting you can see > that I used the word "may" -:) > Kurt Kurt, Thanks. I am in no way suggesting that FRC testing at non-blitz time controls is the definitive test method, however I believe it will add to, and supplement, the other test results published here. Insofar as I can tell, there has been no testing of the 15 FRC opening positions (where normal castling rules apply) at decent, non-blitz, time controls. I intend to continue such testing and will publish results here and make the games available to all. Thanks for the outstanding work you are doing and your GREAT site! Chris
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.