Author: Terry McCracken
Date: 13:40:28 02/09/06
Go up one level in this thread
On February 09, 2006 at 13:00:39, Andreas St. wrote: >Hi, >yesterday i posted this position: > >[D]2r5/2pR1pk1/p1P3p1/P2K4/5P1p/5P1P/8/8 w - - 0 1 > >and some users said, Rybka finds the solution. > >>Not Rybka 13b, but Rybka 1.0 Beta 32-bit (the first version!!!) >>Best regards, >>Orlando > >>Rybka 1.0 Beta 32-bit, sees Rxc7 quickly. I've no idea how the person testing >>whatever version had this failure. > >>Terry > > >What do you mean??? No Rybka version finds the only winning move Rxc7! >If you analyse long enough, you will see, all Rybka versions prefers f5. >At the latest after one hour no Rxc7, but f5! >And look at the score of rybka. He sees nothing in this position. >Fritz 9 for example has +3.50 after 42sec. >Shredder 9 after 0 sec.! After 17sec. Score +3.05 I didn't let it run and run to see if it changes its tune. It has to be some kind of glich. > >Mephisto Vancouver finds the correct move in 8min 45sec! >Fritz, Junior, Shredder.... have no problems to find the correct move Rxc7 with >a high score about +4: > >Here a winning line (white wins quickly): > >1.Rxc7!! Rxc7 2.Kd6 Rc8 3.Kd7 Rh8 4.c7 Kf6 5. c8Q Rxc8 6.Kxc8 and white wins! I know the winning line, it's very simple. After the Rooks are exchanged White picks up the a6 pawn and Queens on the a-file with it's a5 pawn much faster than Black can, so Black is lost. > > >Here an analysis of Rybka (all Pentium IV, 3,28 GHZ, 256MB hash) >Rybka 1.0 Beta 32-bit: > >1.Kd4 Kf6 2.Kc4 Ke6 3.Kc5 Kf6 4.Kd4 > ² (0.32) Tiefe: 10 00:00:00 11kN >1.Kd4 Kf6 2.Kc4 Ke6 3.Kc5 Kf6 4.Kd4 Ke6 > = (0.15) Tiefe: 11 00:00:00 25kN >1.Kd4 Kf6 2.Kc4 Ke6 3.Kc5 Kf6 4.Kd4 Ke6 5.Kc4 Kf6 > = (0.25) Tiefe: 12 00:00:00 41kN >1.Kd4 Kf6 2.Kc4 Ke6 3.Kc5 Kf6 4.Kd4 Ke6 5.Kc4 Kf6 > = (0.18) Tiefe: 13 00:00:00 63kN >1.Kd4 Kf6 2.Kc4 Ke6 3.Kd3 Kf6 4.Kd4 Ke6 5.Kc4 Kf6 > = (0.18) Tiefe: 14 00:00:00 150kN >1.Kd4 Kf6 2.Kc4 Ke6 3.Kd3 Kf6 4.Kd4 Ke6 5.Kc4 Kf6 > = (0.19) Tiefe: 15 00:00:00 217kN >1.Kd4 Kf6 2.Kc5 Ke6 3.Kc4 Kf6 4.Kd3 Ke6 5.Ke4 Kf6 > = (0.09) Tiefe: 16 00:00:02 594kN >1.Kd4 Kf6 2.Kc5 Ke6 3.Kc4 Kf6 4.Kd3 Ke6 5.Ke4 Kf6 > = (0.09) Tiefe: 17 00:00:02 793kN >1.Txc7 Txc7 2.Kd6 Tc8 3.c7 Kf6 4.Kd7 Th8 5.c8D Txc8 > ± (0.93) Tiefe: 17 00:00:03 1333kN >1.Txc7 Txc7 2.Kd6 Tc8 3.c7 Kf6 4.Kd7 Th8 5.c8D Txc8 > ± (1.00) Tiefe: 18 00:00:03 1580kN >1.Txc7 Txc7 2.Kd6 Tc8 3.c7 Kf6 4.Kd7 Th8 5.c8D Txc8 > ± (0.88) Tiefe: 19 00:00:04 1789kN >1.Txc7 Txc7 2.Kd6 Tc8 3.c7 Kf6 4.Kd7 Txc7+ 5.Kxc7 Ke6 > ± (1.04) Tiefe: 20 00:00:06 2980kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 Ta8 4.c7 Kf8 5.c8D+ Txc8 > ± (1.04) Tiefe: 21 00:00:08 3828kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 Ta8 4.c7 Kf8 5.c8D+ Txc8 > ± (0.97) Tiefe: 22 00:00:12 4906kN >1.f5 g5 2.Txc7 Txc7 3.Kd6 Tc8 4.c7 Kf6 5.Kd7 Th8 > ± (1.28) Tiefe: 22 00:01:18 22266kN >1.f5 g5 2.Txc7 Txc7 3.Kd6 Tc8 4.c7 Kf6 5.Kd7 Th8 > ± (1.39) Tiefe: 23 00:01:45 27645kN >1.f5 g5 2.Ke5 Te8+ 3.Kd4 Tc8 4.Ke3 Te8+ 5.Kf2 Kf6 > ± (1.04) Tiefe: 24 00:03:53 49182kN >1.f5 g5 2.Ke5 Te8+ 3.Kd4 Tc8 4.Ke3 Te8+ 5.Kf2 Kf6 > ± (0.90) Tiefe: 25 00:05:21 64774kN >1.f5 g5 2.Ke5 Te8+ 3.Kd4 Tc8 4.Ke3 Te8+ 5.Kd3 Tc8 > ± (0.87) Tiefe: 26 00:10:04 113829kN >1.f5 g5 2.Ke5 Te8+ 3.Kd4 Tc8 4.Ke3 Te8+ 5.Kf2 Kf6 > ± (0.88) Tiefe: 27 00:23:52 228381kN >1.f5 g5 2.Ke5 Te8+ 3.Kd4 Tc8 4.Ke3 Te8+ 5.Kf2 Kf6 > ± (0.84) Tiefe: 28 00:35:36 318508kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 Ta8 4.c7 Kf8 5.c8D+ Txc8 > ± (0.94) Tiefe: 28 00:41:48 468953kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 Ta8 4.c7 Kf8 5.c8D+ Txc8 > ± (0.88) Tiefe: 29 00:47:24 555387kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 Ta8 4.c7 Kf8 5.c8D+ Txc8 > ± (0.88) Tiefe: 30 01:06:30 757475kN >1.Txc7 Txc7 2.Kd6 Tc8 > ± (1.03) Tiefe: 31 01:31:05 1042592kN >1.Txc7 Txc7 2.Kd6 Tc8 > +- (1.47) Tiefe: 32 02:32:48 1626426kN >1.Txc7 Txc7 2.Kd6 Tc8 3.Kd7 > +- (1.49) Tiefe: 33 04:45:24 2788067kN > >Rybka finds Rxc7 the best after 41 min, but only with a score +0.94. >So he don't see a win! >I aborted with Rybka 13b after one hour, he wants f5. For a human the win is elementary. Computers don't "understand" and have to search to great depths. Terry
This page took 0.01 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.