Author: John Merlino
Date: 09:45:47 07/29/04
Go up one level in this thread
On July 29, 2004 at 08:20:52, Uri Blass wrote: >On July 29, 2004 at 01:46:54, Uri Blass wrote: > >>On July 28, 2004 at 17:39:10, John Merlino wrote: >> >>>On July 28, 2004 at 16:03:53, Kurt Utzinger wrote: >>> >>>>How fast does your engine find the winning move 24.Bxg6 >>>>but showing a positive score? Below the output of Crafty 19.15 >>>> >>>>24.Bxg6!! >>>>[D]2r2rk1/p3bb1p/2n1Q1p1/q2pP3/3P1P2/p1NB1NR1/1P4P1/1K1R4 w - - 0 24 >>>> >>>>[Event "New York"] >>>>[Site "New York"] >>>>[Date "1947.??.??"] >>>>[Round "?"] >>>>[White "Evans, L."] >>>>[Black "Pilnick, C."] >>>>[Result "1-0"] >>>>[ECO "C13"] >>>>[PlyCount "57"] >>>>[EventDate "1947.??.??"] >>>> >>>>1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. Bg5 Be7 5. Bxf6 Bxf6 6. e5 Be7 7. Qg4 O-O 8. >>>>O-O-O c5 9. h4 cxd4 10. Nce2 Nc6 11. f4 Qa5 12. Kb1 d3 13. cxd3 Bd7 14. Rh3 >>>>Rac8 15. Rg3 g6 16. d4 b5 17. h5 Nb4 18. a3 Nc6 19. hxg6 fxg6 20. Nc3 b4 21. >>>>Bd3 Be8 22. Nf3 bxa3 23. Qxe6+ Bf7 24. Bxg6 axb2 25. Bxh7+ Kh8 26. Qh6 Qa1+ 27. >>>>Kc2 b1=Q+ 28. Rxb1 Nb4+ 29. Kd1 1-0 >>> >>>This position was posted several weeks ago, and resulted in a small article by >>>GM Evans in Chess Life. >>> >>>The original test position was to find 23.Qxe6+ with a winning score. That was >>>VERY difficult, and I don't know if any engine will be able to do it in under 10 >>>minutes. >>> >>>jm >> >>Finding Qe6 with positive score in less than 10 minutes does not seem to me >>extremely hard task for the top engines >> >>Even weak movei on slow hardware almost can do it in analysis mode. >> >>depth=12 +0.08 g4e6 e8f7 d3g6 f7e6 g6h7 g8h8 d1h1 e7h4 h7d3 e6h3 f3h4 c6e5 d4e5 >>a3a2 b1a1 >>Nodes: 80220868 NPS: 125266 >>Time: 00:10:40.40 >> >>Note that movei usually search near 500 Knodes on A3000 >>I am not sure if I can reproduce this result because movei analyzed another >>position earlier so it may have initial different order of pieces that can give >>different number of nodes but I do not expect a big difference. >> >>I will try later on fast hardware without previous analysis after I stop the >>match between movei and Jonny. >> >>Uri > >I remember the words positive score in the first post but I see now that >positive is not enough and you asked for winning score. > >Movei needs one ply more in order to find more convincing positive score of more >than 1 pawn for itself. > >Uri CM9_SKR, on an AMD 2500 (@1.8GHz), needs 8:29 to choose Qxe6+, and 15:23 to give that move a positive score. So, perhaps, on today's top end hardware, it could give the move a winning score in under 10 minutes. Time Depth Score Positions Moves 0:01 1/6 -3.48 220998 23.Qh5 axb2 24.Bxg6 Bxg6+ 25.Rxg6+ Kh8 26.Rh1 Qa1+ 27.Kc2 Nxd4+ 28.Kd2 b1=N+ 29.Nxb1 0:03 1/6 -3.23 475043 23.Qxe6+ Bf7 24.Qf5 axb2 25.e6 Be8 26.Qxd5 Qxc3 0:04 1/6 -2.87 633468 23.f5 axb2 24.fxg6 Qa1+ 25.Kc2 Nxd4+ 26.Kd2 b1=N+ 27.Nxb1 Nxf3+ 28.Rxf3 Qb2+ 29.Ke1 Qxe5+ 30.Kf1 0:04 1/6 -2.82 794152 23.Bc2 Qb4 24.Na4 Nd8 25.Rd2 a2+ 26.Kxa2 Bxa4 27.Bxg6 0:05 1/7 -2.82 997750 23.Bc2 Qb4 24.Na4 Nd8 25.Rd2 a2+ 26.Kxa2 Bxa4 27.Bxg6 0:12 1/8 -2.92 2356505 23.Bc2 Qb4 24.Na4 Nd8 25.Rd2 Rxc2 26.Rxc2 Bxa4 27.Rf2 axb2 28.Rxb2 0:22 1/8 -2.82 4691659 23.f5 axb2 24.fxg6 Qa1+ 25.Kc2 Nxd4+ 26.Kd2 b1=N+ 27.Nxb1 Nxf3+ 28.Rxf3 Qb2+ 29.Ke1 Qxe5+ 30.Kf1 Rxf3+ 31.gxf3 hxg6 32.Bxg6 0:48 1/9 -2.82 10215879 23.f5 axb2 24.fxg6 Qa1+ 25.Kc2 Nxd4+ 26.Kd2 b1=N+ 27.Nxb1 Nxf3+ 28.Rxf3 Qb2+ 29.Ke1 Qxe5+ 30.Kf1 hxg6 31.Bxg6 Rxf3+ 32.gxf3 2:28 2/10 -2.62 31832802 23.f5 Nxe5 24.dxe5 axb2 25.Na2 Rc1+ 26.Rxc1 Qxa2+ 27.Kxa2 bxc1=Q 28.Nd4 h5 6:39 2/11 -2.77 88332852 23.f5 exf5 24.Bxf5 Rb8 25.Ne1 Kh8 26.Bc2 axb2 27.Ba4 Rb4 28.Nd3 Rxa4 29.Nxa4 Qxa4 30.Nxb2 8:29 2/11 -1.92 114990905 23.Qxe6+ Bf7 24.Bxg6 Bxe6 25.Bxh7+ Kh8 26.Rh1 a2+ 27.Nxa2 Bh4 28.Nxh4 Nxd4 29.Ng6+ Kg7 30.Nxf8+ Kxf8 31.Bd3 15:23 2/12 1.45 213656701 23.Qxe6+ Bf7 24.Bxg6 Bxe6 25.Bxh7+ Kh8 26.Rh1 a2+ 27.Ka1 Bh4 28.Rxh4 Rxf4 29.Rxf4 Nb4 30.Be4 Qd8 It would be very interesting to test this on the latest version of The King that will ship with CM 10th Edition.... Johan? jm
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.