Author: Uri Blass
Date: 02:21:51 11/14/02
Go up one level in this thread
On November 13, 2002 at 21:13:32, Jim Monaghan wrote: >>>Latest Movei on AMD1000 Mhz scored better than Crafty on 450 Mhz based on the >>>results. >>> >>>Movei solved 224 out of 360 in 10 seconds per move and I am responsible for some >>>noise by using the internet and writing something in word at part of the time > >Hi Uri, > >That's an excellent result (62.22% = IQ elo of 2311)! Little Movei is a tactical >monster. Thanks. I still know that there is a big room for improvement but when I look at the positions that movei failed I find at least one wrong position(see the end of the post). > >>>summery of movei's results >>> >>>Sec Solved Total >>>---- ------ ------ >>> 1 152 152 >>> 2 25 177 >>> 3 15 192 >>> 4 11 203 >>> 5 7 210 >>> 6 5 215 >>> 7 3 218 >>> 8 0 218 >>> 9 5 223 >>> 10 1 224 > >>>Id: IQ.956 >>>[D]2qr1r2/p2n1ppk/1p2p2p/5b2/2PP2NQ/1B4R1/P4PPP/R5K1 w - - 0 1 >>>Bm: Bc2 >>> >>>Hard one >>>More than 7 minutes only to fail high when it does not know if the fail high is >>>correct fail high. >>> >>>This move is a move that I am going to prune in a game but I guess that movei is >>>smarter than me. >>> >>>At the time of writing these lines latest movei still did not solve the fail >>>high. >>> >>>depth=12 +0.28 a1d1 c8c7 h4e7 d8e8 e7b4 f5g4 g3g4 f7f5 b3c2 h7g8 g4g6 d7f6 >>>Nodes: 49462753 NPS: 166244 >>>Time: 00:04:57.53 >>> >>>depth=12 +0.29 b3c2 >>>Nodes: 73988995 NPS: 167866 >>>Time: 00:07:20.76 >>> >>>Uri >> >>I did not look at the analysis but movei solved the fail high few minutes ago >>and analysis later seems to go fast. >>I understand now that the idea of Bc2 is to let Rh3 later and the way to think >>about it for humans should be first to think about Rh3 and after seeing that the >>bishop can take the rook to think how to get the bishop out of f5. > >Exactly right. Here is Shredder 6.0 > > 11.01 0:17 +0.33 1.Rc1 Qc7 2.Ne3 Bg6 3.Bc2 Bxc2 4.Rxc2 Nf6 5.f3 Kh8 6.Re2 >(4.595.334) 262.6 > 11.02 0:18 +0.34++ 1.Re1 Qc6 (4.775.282) 261.9 > 11.02 0:37 +0.34 1.Re1 Qc7 2.Nxh6 gxh6 3.Qxh6+ Kxh6 4.f3 Qc6 5.Rh3+ Bxh3 >6.Rxe6+ fxe6 (9.771.612) 264.0 > 11.17 1:16 +0.35++ 1.Rh3 Bxg4 2.Qxg4 Nf6 3.Qf4 Qc6 4.Re3 Nh5 5.Qe5 f6 6.Bc2+ >(21.170.616) 276.7 > 11.17 1:23 +0.42 1.Rh3 Bxg4 2.Qxg4 Nf6 3.Qf4 Qc6 4.Re3 Kg8 5.f3 b5 6.c5 a5 >(23.063.661) 276.6 > 12.01 1:54 +0.42 1.Rh3 Bxg4 2.Qxg4 Nf6 3.Qf4 Kg8 4.Re1 Rfe8 5.Re5 Nd7 6.Reh5 >e5 (30.951.536) 269.1 > 12.03 2:35 +0.43++ 1.Rc1 Qc7 2.Ne3 Bg6 3.Bc2 Bxc2 4.Rxc2 Nf6 5.f3 Kh8 6.Re2 >a6 (42.119.642) 270.8 > 12.03 3:03 +0.44 1.Rc1 Qc6 2.Ne3 Bg6 3.Bc2 Bxc2 4.Rxc2 Nf6 5.f3 a6 6.Rxg7+ >(49.526.088) 270.0 > 12.20 5:59 +0.45++ 1.Bc2 Qxc4 2.Nxh6 gxh6 3.Rh3 Kg8 4.Qxh6 Qxd4 5.Rg3+ Bg6 >6.Bxg6 Qxa1+ 7.Bb1+ (101.945.636) 283.4 > 12.20 6:57 +0.80++ 1.Bc2 Qxc4 2.Nxh6 gxh6 3.Rh3 Kg8 4.Qxh6 Qxd4 5.Rg3+ Bg6 >6.Bxg6 Qxf2+ 7.Kxf2 fxg6+ 8.Qxf8+ Rxf8+ 9.Rf3 Nb8 (120.325.265) 288.0 >best move: Bb3-c2 time: 18:43.576 min n/s: 311.490 CPU 96.4% nodes: >349.974.086 > > >>depth=13 +2.24 b3c2 f8h8 g4h6 g7h6 h4h5 d7e5 d4e5 d8f8 g3h3 h7g8 c2f5 e6f5 h3g3 >>g8h7 a1d1 >>Nodes: 286405465 NPS: 165674 >>Time: 00:28:48.72 >> >>depth=14 +2.46 b3c2 f8h8 g4h6 g7h6 h4h5 d7e5 d4e5 d8f8 g3h3 h7g8 c2f5 e6f5 h3g3 >>g8h7 a1d1 f5f4 >>Nodes: 332814871 NPS: 165382 >>Time: 00:33:32.39 >> >>I also checked if I can blame the new pruning for not finding >>Bc2 >> >>Here is analysis of previous version(without the special pruning) >> >>depth=10 +0.34 a1c1 c8c6 c1e1 f8e8 g4e3 d7f6 e3f5 e6f5 g3e3 e8e3 f2e3 >>Nodes: 35128776 NPS: 160368 >>Time: 00:03:39.05 >> >>depth=11 +0.31 a1d1 >>Nodes: 65801863 NPS: 159508 >>Time: 00:06:52.53 >> >>depth=11 +0.36 a1d1 f7f6 h4h5 f5g6 h5b5 e6e5 d4e5 a7a6 b5b4 f6e5 g4e3 >>Nodes: 79342912 NPS: 159294 >>Time: 00:08:18.09 >> >>depth=11 +0.37 b3c2 >>Nodes: 91050402 NPS: 160015 >>Time: 00:09:29.01 >> >>depth=11 +0.50 b3c2 f8h8 h4e7 h8f8 c2f5 e6f5 g4e5 f5f4 g3c3 h7g8 e5d7 d8d7 e7e4 >>Nodes: 109364352 NPS: 160924 >>Time: 00:11:19.60 >> >> >>The old version can find Bc2 at depth 11 but it needs more time to see the fail >>high. >> >>On the other hand it needs less time to solve the fail high. > >Hmmm ... the older version seems better on this position. > >>Next position that was not solved. >> >>[D]1k1r4/ppq2p2/8/2pPb1r1/2P1Q1B1/6P1/PP3PK1/R4R2 b - - 0 1 >> >>This seems to be also a positional problem because after f5 Bxf5 Bxg3 Bg4 there >>is no forced win that I can see. > >Engine: Shredder 6 > 13.01 4:51 +0.94 1...f5 2.Bxf5 Bxg3 3.Bg4 Bd6 4.f3 Bh2 5.Rh1 Qg3+ 6.Kf1 Rf8 >7.Re1 a6 8.Re2 Rxg4 (72.166.790) 247.2 >best move: f7-f5 time: 12:42.706 min n/s: 199.794 CPU 63.7% nodes: >152.384.451 > >Even though Black is down two pawns, he has a lot of pressure on the dark >squares (opposite coloured bishops always favour the attacker) and it's >difficult to see how White can improve his position. Does Black steadily make >progress? > >1... f5 2. Bxf5 Bxg3 3. Bg4 Bd6 4. f3 Bh2 5. Qe1 Rdg8 6. Rd1 Qh7 -+ looks good >for Black. > >Hmmm ... I can see why you would want to call this a "positional" problem. It's >not 100% clear. > >In the game White blundered: 1... f5 2. Bxf5 Bxg3 3. Rfe1? Bxf2+! 4. Kxf2 Qh2+ >and it was over quickly. Yes The next position that movei failed is another case when there is no forced win. IQ.959 [D]5r1k/p5b1/1qp4p/1r2p3/4N1P1/2Pn2B1/QP2R1PP/R6K b - - 0 1 Bm: Ra5 It needs slightly more than 10 seconds to fail high and finds Ra5 at depth 9. The score is better at bigger depthes but I do not see a tactical win for black. depth=9 -0.33 b5a5 a2b1 a5a1 b1a1 b6a6 a1g1 d3c5 e2e1 c5e4 e1e4 Nodes: 1762955 NPS: 162036 Time: 00:00:10.88 depth=13 +0.39 b5a5 a2b1 a5a1 b1a1 b6a6 e2e1 a6c4 e1e3 c4a4 a1g1 d3b2 g1b1 a4c2 b1g1 Nodes: 8317461 NPS: 166950 Time: 00:00:49.82 Here pruning is clearly productive and without pruning find Ra5 takes 20 seconds. With the next position it is the opposite and the move is found twice slower because of pruning IQ160 [D]6k1/pp2q2r/2r1p3/1RPpPpp1/3P4/P5P1/3Q1PK1/7R w - - 0 1 Next position IQ965 [D]r6r/p1kn4/2p1b3/q3p1Qp/4P3/2PBB2P/P5P1/3R1RK1 w - - 0 1 Bm: Bc4 depth=9 +0.71 d3c4 Nodes: 5538426 NPS: 165079 Time: 00:00:33.55 depth=9 +1.84 d3c4 e6c4 d1d7 c7d7 f1d1 c4d5 e4d5 a8f8 g5e5 a5a4 d5c6 d7c8 e5e6 c8b8 e6e5 b8a8 Nodes: 8080525 NPS: 162684 Time: 00:00:49.67 depth=9 +1.84 d3c4 e6c4 d1d7 c7d7 f1d1 c4d5 e4d5 a8f8 g5e5 a5a4 d5c6 d7c8 e5e6 c8b8 e6e5 b8a8 Nodes: 8620003 NPS: 162549 Time: 00:00:53.03 depth=10 +1.54 d3c4 Nodes: 12727142 NPS: 162253 Time: 00:01:18.44 This happens only when movei has a fail low and later fail high so it did not solve the fail low and has no pv) depth=10 +1.54 d3c4 Nodes: 26511594 NPS: 164402 Time: 00:02:41.26 depth=11 +2.72 d3c4 e6c4 d1d7 c7d7 g5g7 d7c8 g7h8 c8b7 f1b1 c4b5 h8h7 a5c7 h7h5 c7d6 c3c4 d6d3 h5f7 b7c8 f7f8 c8b7 Nodes: 73342349 NPS: 166740 Time: 00:07:19.86 depth=12 +2.58 d3c4 e6c4 d1d7 c7d7 g5g7 d7c8 g7h8 c8b7 f1b1 c4b5 h8h7 a5c7 h7h5 b7a6 b1d1 a6b7 h5f5 c6c5 d1b1 Nodes: 93588658 NPS: 167619 Time: 00:09:18.34 without pruning depth=9 +0.81 d3c4 Nodes: 12333565 NPS: 157940 Time: 00:01:18.09 depth=9 +1.84 d3c4 e6c4 d1d7 c7d7 f1d1 c4d5 e4d5 a8f8 g5e5 a5a4 d5c6 d7c8 e5e6 c8b8 e6e5 b8a8 Nodes: 15774551 NPS: 157887 Time: 00:01:39.91 depth=10 +1.54 d3c4 Nodes: 21260287 NPS: 159827 Time: 00:02:13.02 depth=10 +1.54 d3c4 Nodes: 39701150 NPS: 162350 Time: 00:04:04.54 Next hard position r5k1/1p3pbp/2q1p1p1/p1p1P3/n1P5/P1B2N1P/1P2QPP1/3R2K1 w - - 0 1 bm Bxa5 depth=13 +0.26 c3a5 Nodes: 18653563 NPS: 165383 Time: 00:01:52.79 depth=13 +0.44 c3a5 a8a5 d1d8 g7f8 e2d2 c6c7 d8e8 g8g7 b2b3 a5a6 b3a4 a6a4 d2d3 g7g8 Nodes: 26700546 NPS: 166503 Time: 00:02:40.36 depth=14 +0.48 c3a5 a8a5 d1d8 g7f8 e2d2 c6c7 d8e8 g8g7 b2b3 a5a6 b3a4 a6a4 d2d3 g7g8 g2g4 Nodes: 54805927 NPS: 167510 Time: 00:05:27.18 Again my special pruning seem to help. without them I get only depth=13 +0.32 c3e1 a4b6 e1d2 h7h6 e2d3 c6a4 d1c1 a4d7 d3d7 b6d7 c1e1 a5a4 g1f1 Nodes: 23653778 NPS: 157765 Time: 00:02:29.93 depth=13 +0.33 c3a5 Nodes: 31957205 NPS: 159802 Time: 00:03:19.98 depth=13 +0.44 c3a5 a8a5 d1d8 g7f8 e2d2 c6c7 d8e8 g8g7 b2b3 a5a6 b3a4 a6a4 d2d3 g7g8 Nodes: 39043571 NPS: 160362 Time: 00:04:03.47 Next position seems to be wrong [D]4rbk1/1q3ppp/2Rr4/1p1P1B2/2b1PR2/p5P1/5P1P/B1Q3K1 w - - 0 1 bm Bxh7+ Movei finds Bxh7+ but later changes it's mind to Rxd6 Bxd6 Bxh7+ When I continue the analysis the score only improces depth=10 +4.16 c6d6 f8d6 f5h7 g8f8 a1g7 f8e7 e4e5 d6c5 d5d6 e7d8 h7f5 e8e6 f5e6 f7e6 f4f8 d8d7 Nodes: 6053294 NPS: 156821 Time: 00:00:38.60 depth=11 +4.97 c6d6 f8d6 f5h7 g8h8 a1g7 h8g7 f4g4 g7f8 g4g8 f8e7 c1g5 e7d7 h7f5 d7c7 g8e8 c7b6 g5h6 b7c7 e8a8 Nodes: 19352494 NPS: 152489 Time: 00:02:06.91 depth=11 +4.97 c6d6 f8d6 f5h7 g8h8 a1g7 h8g7 f4g4 g7f8 g4g8 f8e7 c1g5 e7d7 h7f5 d7c7 g8e8 c7b6 g5h6 b7c7 e8a8 Nodes: 24086138 NPS: 154428 Time: 00:02:35.97 depth=12 +5.33 c6d6 f8d6 f5h7 g8f8 a1g7 f8e7 e4e5 a3a2 e5d6 e7d6 f4f6 d6c5 g7h6 a2a1q c1a1 c4d5 a1c3 d5c4 h6e3 c5d5 Nodes: 108694351 NPS: 154791 Time: 00:11:42.20 depth=12 +5.33 c6d6 f8d6 f5h7 g8f8 a1g7 f8e7 e4e5 a3a2 e5d6 e7d6 f4f6 d6c5 g7h6 a2a1q c1a1 c4d5 a1c3 d5c4 h6e3 c5d5 Nodes: 112652649 NPS: 155303 Time: 00:12:05.37 After Bxh7+ movei cannot see +5 or even +4 and finds only depth=11 -2.20 g8h7 c6d6 f7f6 d6c6 h7g8 a1d4 a3a2 f2f3 f8b4 f4f5 b7d7 g3g4 Nodes: 11098490 NPS: 155050 Time: 00:01:11.58 depth=12 -2.23 g8h7 c6d6 f7f5 d6e6 e8e6 d5e6 c4e6 e4f5 e6c4 f5f6 a3a2 c1c2 h7g8 c2g6 c4d5 f6g7 f8g7 a1g7 b7g7 Nodes: 29861432 NPS: 157472 Time: 00:03:09.63 Uri
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.