Author: Mike Byrne
Date: 18:44:32 12/28/02
Go up one level in this thread
On December 28, 2002 at 21:18:27, John Merlino wrote: >On December 28, 2002 at 17:52:39, Dana Turnmire wrote: > >>On December 28, 2002 at 17:39:09, Dana Turnmire wrote: >> >>>[Event ""] >>>[Site ""] >>>[Date "2002.12.28"] >>>[Round ""] >>>[White "CM 9000"] >>>[Black "Genius 7"] >>>[TimeControl "5400"] >>>[Result "*"] >>>[Setup "1"] >>[d]rbnkbnrq/pppppppp/8/8/8/8/PPPPPPPP/RBNKBNRQ w - - 0 1 >> >>>1.g3 c6 2.d4 f6 3.c4 Bf7 4.Ne3 Nb6 5.c5 Nd5 6.Nf5 e6 7.Nd6 Bxd6 8.cxd6 >>>Ke8 9.Nd3 Nb6 10.Bb4 Bg6 11.e4 a5 12.Bc5 Nc8 13.a4 b6 14.Ba3 h6 15.Qf3 >>>Qh7 16.Re1 Rh8 17.h4 Kd8 18.h5 Bf7 19.b4 Qg8 20.e5 axb4 21.Bxb4 Nh7 >>>22.a5 Qe8 23.axb6 Nxb6 24.Rxa8+ Nxa8 25.Nf4 Bxh5 26.Nxh5 fxe5 27.Nxg7 >>>Qg8 28.Qa3 Nb6 29.Qa7 * >>> >>>In this Fischer Random Game CM9000 announced mate in 16. I was surprised >>>because each side only has 90 minutes for the whole game. Here is the final >>>position. Fischer Random may be a good way to test how well programs calculate >>>since none of the positions are stored in an opening book. >>> >>>[d]3k2qr/Q2p2Nn/1npPp2p/4p3/1B1P4/6P1/5P2/1B1KR3 b - - 0 29 > >Two things: > >1) My version of CM9000, on a P3-733, announces Mate in 15 in 5:13. Perhaps you >saw the Mate in 16 announcement before 29.Qa7? > >Time Depth Score Positions Moves >0:00 1/3 8.03 1493 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qxe6+ > Qf7 35.Qxf7+ Kxf7 36.dxe5 >0:00 1/4 8.31 4038 29...Qxg7 30.Qb8+ Nc8 31.Ba5+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qxe6+ > Qf7 35.Qxf7+ Kxf7 36.dxe5 >0:00 1/5 8.69 14440 29...Qxg7 30.Qb8+ Nc8 31.Ba5+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qxe6+ > Qf7 35.Ba2 Qxe6 36.Bxe6+ Kg7 37.dxe5 >0:00 1/6 9.72 44920 29...Qxg7 30.Qb8+ Nc8 31.Ba5+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qxe6+ > Qf7 35.Qg4+ Ng5 36.d7 Kg7 37.d8=Q > Rxd8 38.Bxd8 >0:01 1/7 12.89 124016 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qe8+ > Nf8 35.d7 c5 36.d8=Q cxb4 37.Qxe6+ > Qf7 38.Qxf7+ Kxf7 39.dxe5 >0:02 1/8 15.75 278402 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qe8+ > Nf8 35.d7 c5 36.d8=Q Qg4+ 37.Kc1 > Qg5+ 38.Qxg5+ hxg5 39.dxc5 >0:06 1/9 16.29 762182 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qe8+ > Nf8 35.d7 c5 36.d8=Q Rh7 37.Bxh7+ > Kxh7 38.dxc5 >0:22 1/10 25.18 2675042 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kg8 34.Qe8+ > Nf8 35.d7 c5 36.d8=Q e4 37.Qxe6+ > Qf7 38.Ba2 Rh7 39.Qg6+ Kh8 40.Bxf7 >5:13 2/11 Mate15 38563606 29...Qxg7 30.Qb8+ Nc8 31.Qc7+ Ke8 > 32.Qxc8+ Kf7 33.Qxd7+ Kf8 34.Qd8+ > Kf7 35.Qe7+ Kg8 36.Qe8+ Nf8 37.d7 > c5 38.d8=Q e4 39.Qxe6+ Qf7 40.Bxc5 > Kg7 41.Qxf7+ Kxf7 42.Qe7+ Kg8 43.Qxf8+ > Kh7 44.Bxe4# > >2) Fischer Random chess, IMO, only eliminates opening theory. It does not >eliminate positional considerations or tactical possibilities. Given that chess >engines are generally considered weakest in the opening, FR chess could >certainly be used to test engine knowledge. But the problem is that it is >difficult to say with any certainty what the best move is in a particular >opening, since there IS no opening theory. >Additionally, I suspect that, all else being equal, White has more of an >advantage in FR chess than in regular chess, as Black does not have access to >opening theory which allows for more chances for equalization and/or >complications. This is just a thought, though.... > >jm jm, I backed the problem one move prior to Qa7 and standard CM9K out of the box saw mate-in-16 in 58 seconds on my machine (Xeon P4 1.8 Ghz Win 2K). I tested many of the big name engines I have and not one found it in less than a minute. I also tried it with the Utzinger settings and that did not find in on my machine i>1". He is running pretty fast machine. Michael
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.