Author: Robert Hyatt
Date: 21:37:14 05/29/02
Go up one level in this thread
On May 29, 2002 at 18:25:05, Olaf Jenkner wrote: >I noticed a big difference between Crafty 18.13 and 18.14 in a >special position and don't understand what happened. >Maybe you can explain it. > >See this position: > >[D]3B4/1r1kp3/r2p1p2/b1pPP1p1/1pP2PPp/pP5P/P2K4/8 w - - 0 1 > >After 1.e6+ K:d8 2.f5+ it is a draw. > >Version 18.13 finds the move after 35 min at depth 16 > and gives the evaluation -12.40 >Version 18.14 finds the move after 150 min at depth 17 gives > and gives the evaluation -12.34 > >xxxxxxxxxxxxxxxxxxx Little Question xxxxxxxxxxxxxxxxxxxxx >What has been changed in version 18.14 to cause this difference? >In main.c I didn't find a clue. There were minor evaluation changes. Any evaluation change can cause the time for an odd position to change significantly, even though the scoring change doesn't affect the final score or best move. alpha/beta is a strange animal. >xxxxxxxxxxxxxxxxxxx Little Question xxxxxxxxxxxxxxxxxxxxx > > >Here is the output of Crafty 18.13: > >12 16.41 -9.97 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Re8 5. Kd3 Re1 6. g5 Ra1 > 7. Bxd6 Rxd6 >12-> 17.18 -9.97 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Re8 5. Kd3 Re1 6. g5 Ra1 > 7. Bxd6 Rxd6 >13 17.89 -- 1. exf6 >13 39.90 -10.45 1. exf6 exf6 2. Bxf6 gxf4 3. Bxh4 Rb8 > 4. Bg5 f3 5. Ke3 Rf8 6. Kf2 Raa8 > 7.Bh6 Rf6 >13-> 43.18 -10.45 1. exf6 exf6 2. Bxf6 gxf4 3. Bxh4 Rb8 > 4. Bg5 f3 5. Ke3 Rf8 6. Kf2 Raa8 > 7.Bh6 Rf6 >14 46.61 -- 1. exf6 >14 2:05 -11.36 1. exf6 Ke8 2. Bxe7 gxf4 3. Kd1 Rxe7 > 4. fxe7 Ra7 5. g5 Rxe7 6. g6 f3 > 7. g7 Rxg7 8. Ke1 Rg2 >14-> 3:15 -11.36 1. exf6 Ke8 2. Bxe7 gxf4 3. Kd1 Rxe7 > 4. fxe7 Ra7 5. g5 Rxe7 6. g6 f3 > 7.g7 Rxg7 8. Ke1 Rg2 >15 3:25 -- 1. exf6 >15 7:35 -12.28 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 Rxa5 > 4. fxg5 e5 5. dxe6+ Kg6 6. Kd1 Re7 > 7. Ke2 Rxe6+ 8. Kf2 Ra8 9. Kf1 Rf8+ > 10. Kg1 Kxg5 >15-> 8:49 -12.28 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 Rxa5 > 4. fxg5 e5 5. dxe6+ Kg6 6. Kd1 Re7 > 7. Ke2 Rxe6+ 8. Kf2 Ra8 9. Kf1 Rf8+ > 10. Kg1 Kxg5 >16 9:01 -- 1. exf6 >16 21:25 -12.84 1. exf6 Ke8 2. f7+ Kxf7 3. f5 Bxd8 > 4. Ke2 Kf6 5. Kf2 Ke5 6. Ke3 Bb6 > 7. Ke2 Kd4 8. Kd2 Ra5 9. Ke2 >16 34:43 -12.40 1. e6+ Kxd8 2. f5 Rb8 3. Kc1 Raa8 > 4. Kd2 Kc7 5. Kc1 Bb6 6. Kd2 Rb7 > 7. Kc1 Re8 8. Kd2 Rd8 9. Ke3 >16-> 35:03 -12.40 1. e6+ Kxd8 2. f5 Rb8 3. Kc1 Raa8 > 4. Kd2 Kc7 5. Kc1 Bb6 6. Kd2 Rb7 > 7. Kc1 Re8 8. Kd2 Rd8 9. Ke3 >17 38:02 -12.42 1. e6+ Kxd8 2. f5 Rb8 3. Kc1 Kc7 > 4. Kd2 Bb6 5. Kc1 Ra5 6. Kd2 Rh8 > 7. Kd3 Rh6 8. Ke4 Kd8 9. Ke3 Kc7 >17-> 42:28 -12.42 1. e6+ Kxd8 2. f5 Rb8 3. Kc1 Kc7 > 4. Kd2 Bb6 5. Kc1 Ra5 6. Kd2 Rh8 > 7. Kd3 Rh6 8. Ke4 Kd8 9. Ke3 Kc7 > > >Here is the output of Crafty 18.14: > >12 21.08 -9.94 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Re8 5. g5 Ke7 6. Ke3 Kf7+ > 7. Kf3 Kg6 >12-> 21.92 -9.94 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Re8 5. g5 Ke7 6. Ke3 Kf7+ > 7. Kf3 Kg6 >13 22.67 -- 1. exf6 >13 1:03 -10.16 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Rf8 5. Ke3 Raa8 6. g5 Rae8+ > 7. Kf3 Rg8 8. Kg4 >13-> 1:06 -10.16 1. exf6 exf6 2. Bxf6 gxf4 3. Bg5 Rb8 > 4. Bxf4 Rf8 5. Ke3 Raa8 6. g5 Rae8+ > 7. Kf3 Rg8 8. Kg4 >14 1:07 -- 1. exf6 >14 3:06 -11.36 1. exf6 Ke8 2. Bxe7 gxf4 3. Kd1 Rxe7 > 4. fxe7 Ra7 5. g5 Rxe7 6. g6 f3 > 7. g7 Rxg7 8. Ke1 Rg2 >14-> 4:09 -11.36 1. exf6 Ke8 2. Bxe7 gxf4 3. Kd1 Rxe7 > 4. fxe7 Ra7 5. g5 Rxe7 6. g6 f3 > 7.g7 Rxg7 8. Ke1 Rg2 >15 4:18 -- 1. exf6 >15 11:19 -12.17 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 gxf4 > 4. Bd8 Kf6 5. Ke2 Ra8 6. Kf3 Kg5 > 7. Bxe7+ Rxe7 8. Kf2 Re3 9. Kg2 >15-> 12:51 -12.17 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 gxf4 > 4. Bd8 Kf6 5. Ke2 Ra8 6. Kf3 Kg5 > 7. Bxe7+ Rxe7 8. Kf2 Re3 9. Kg2 >16 30:07 -12.37 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 gxf4 > 4. Bd8 Ra8 5. Ke2 Rxd8 6. Kf3 e5 > 7. dxe6+ Kxe6 8. Kxf4 Rf7+ 9. Ke3 d5 > 10. cxd5+ Kxd5 11. g5 c4 >16-> 34:40 -12.37 1. exf6 Ke8 2. f7+ Kxf7 3. Bxa5 gxf4 > 4. Bd8 Ra8 5. Ke2 Rxd8 6. Kf3 e5 > 7.dxe6+Kxe6 8. Kxf4 Rf7+ 9. Ke3 d5 > 10. cxd5+ Kxd5 11. g5 c4 >17 35:43 -- 1. exf6 >17 108:54 -12.94 1. exf6 Ke8 2. f5 Bxd8 3. f7+ Kxf7 > 4. Ke1 Kf6 5. Kd2 Rb8 6. Ke3 Ke5 > 7.Kf3 Kd4 <HT> >17 150:54 -12.34 1. e6+ Kxd8 2. f5 Ra8 3. Kd3 Rc8 > 4.Kd2 Bb6 5. Ke3 Ke8 6. Ke4 Rbc7 > 7. Kf3 Rd8 8. Ke4 Kf8 9. Ke3 Kg7 >17-> 150:59 -12.34 1. e6+ Kxd8 2. f5 Ra8 3. Kd3 Rc8 > 4. Kd2 Bb6 5. Ke3 Ke8 6. Ke4 Rbc7 > 7. Kf3 Rd8 8. Ke4 Kf8 9. Ke3 Kg7 >18 164:52 -12.35 1. e6+ Kxd8 2. f5 Ra8 3. Kd3 Kc7 > 4. Ke3 Rba7 5. Ke4 Bb6 6. Kd3 Ra5 > 7. Ke2 Rh8 8. Kd3 Rh6 9. Ke4 Kb7 10. Ke3 >18-> 180:54 -12.35 1. e6+ Kxd8 2. f5 Ra8 3. Kd3 Kc7 > 4. Ke3 Rba7 5. Ke4 Bb6 6. Kd3 Ra5 > 7. Ke2 Rh8 8. Kd3 Rh6 9. Ke4 Kb7 10. Ke3 > > >Before Crafty finds 1.e6+ it considers 1.exf6 Ke8. >I tried to find out why Crafty 18.14 needs one ply more. > >See the diagram: > >[D]3Bk3/1r2p3/r2p1P2/b1pP2p1/1pP2PPp/pP5P/P2K4/8 w - - 0 1 > >Crafty 18.13: > >12 3.21 -- 1. Bxe7 >12 6.48 -11.36 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. g6 Rxe7 5. Kd1 f3 6. g7 Rxg7 > 7. Ke1 Rg2 >12-> 15.39 -11.36 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. g6 Rxe7 5. Kd1 f3 6. g7 Rxg7 > 7. Ke1 Rg2 >13 16.30 -- 1. Bxe7 >13 24.18 -11.98 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. Ke2 Rxe7+ 5. Kf2 Re3 6. g6 Rxh3 > 7. g7 Rg3 8. g8=Q+ Rxg8 9. Kf3 >13-> 37.73 -11.98 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. Ke2 Rxe7+ 5. Kf2 Re3 6. g6 Rxh3 > 7. g7 Rg3 8. g8=Q+ Rxg8 9. Kf3 >14 38.74 -- 1. Bxe7 >14 2:08 -13.17 1. Bxe7 gxf4 2. Bxd6 Rxd6 3. g5 Rxd5+ > 4. cxd5 c4 5. Kc2 c3 6. Kd3 Bd8 > 7. d6 Bxf6 8. gxf6 >14 6:11 -12.97 1. f7+ Kxf7 2. f5 Kf6 3. Ke2 Ke5 > 4. Kd2 Ra8 5. Ke3 Rxd8 6. Kf3 Kd4 > 7. Ke2 Rbd7 <HT> >14 8:13 -12.94 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ba5 <HT> >14-> 8:15 -12.94 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ba5 6. Kd3 <HT> >15 9:35 -12.94 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ba5 6. Kf3 Kd4 > 7. Ke2 Rb8 8. Kd2 Rf8 >15-> 10:03 -12.94 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ba5 6. Kf3 Kd4 > 7. Ke2 Rb8 8. Kd2 Rf8 >16 16:04 -13.09 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ra8 6. Ke2 Kf4 > 7. Kf2 Bb6 8. Kg2 Ke3 9. Kh2 >16-> 17:47 -13.09 1. f5 Bxd8 2. f7+ Kxf7 3. Ke2 Kf6 > 4. Kf2 Ke5 5. Ke3 Ra8 6. Ke2 Kf4 > 7. Kf2 Bb6 8. Kg2 Ke3 9. Kh2 > >After 2 minutes (14 plies) the evaluation drops under the >limit of -12.40 and stays below it. 14 plies sounds logical >because one move (2 plies) before the solution was found at >ply 16. > >But now look at the output of Crafty 18.14. It is the reason >why I wrote this posting: > >12 3.76 -- 1. Bxe7 >12 8.10 -11.36 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. g6 Rxe7 5. Kd1 f3 6. g7 Rxg7 > 7. Ke1 Rg2 >12-> 19.26 -11.36 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. g6 Rxe7 5. Kd1 f3 6. g7 Rxg7 > 7. Ke1 Rg2 >13 20.32 -- 1. Bxe7 >13 30.06 -11.98 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. Ke2 Rxe7+ 5. Kf2 Re3 6. g6 Rxh3 > 7. g7 Rg3 8. g8=Q+ Rxg8 9. Kf3 >13-> 40.30 -11.98 1. Bxe7 gxf4 2. g5 Rxe7 3. fxe7 Ra7 > 4. Ke2 Rxe7+ 5. Kf2 Re3 6. g6 Rxh3 > 7. g7 Rg3 8. g8=Q+ Rxg8 9. Kf3 >14 41.27 -- 1. Bxe7 >14 2:26 -13.17 1. Bxe7 gxf4 2. Bxd6 Rxd6 3. g5 Rxd5+ > 4. cxd5 c4 5. Kc2 c3 6. Kd3 Bd8 > 7. d6 Bxf6 8. gxf6 >14 5:36 -12.99 1. f7+ Kxf7 2. f5 Kf6 3. Kd3 Ke5 > 4. Ke2 Bxd8 5. Ke3 Ra8 6. Kf3 Kd4 > 7. Ke2 Ke4 8. Ke1 <HT> >14 8:18 -12.93 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Ra8 > 4. Kd2 Kf6 <HT> >14-> 8:18 -12.93 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Ra8 > 4. Kd2 Kf6 <HT> >15 9:58 -12.99 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Kf6 > 4. Ke4 Raa7 5. Kd3 Ke5 6. Ke2 Ra6 > 7. Ke3 Ra8 <HT> >15-> 10:33 -12.99 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Kf6 > 4. Ke4 Raa7 5. Kd3 Ke5 6. Ke2 Ra6 > 7. Ke3 Ra8 <HT> >16 10:38 -- 1. f5 >16 54:16 -13.09 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Kf6 > 4. Kd2 Ke5 5. Ke3 Ra8 6. Ke2 Kf4 > 7. Kf2 Bb6 8. Kg2 Ke3 9. Kh2 >16-> 56:35 -13.09 1. f5 Bxd8 2. f7+ Kxf7 3. Kd3 Kf6 > 4. Kd2 Ke5 5. Ke3 Ra8 6. Ke2 Kf4 > 7. Kf2 Bb6 8. Kg2 Ke3 9. Kh2 >17 56:42 -- 1. f5 > >After 2.5 minutes (14 plies) the evaluation drops under the >limit of -12.34 and stays below it. > >XXXXXXXXXXXXXXXXXXX BIG QUESTION XXXXXXXXXXXXXXXXXXXXXXXX >Why do we have such a big difference in the first position? >The difference in the second position is vey small. >XXXXXXXXXXXXXXXXXXX BIG QUESTION XXXXXXXXXXXXXXXXXXXXXXXX > > >Crafty was running on a 333 MHz PII. > >Olaf Jenkner
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.