Author: Darrel Briley
Date: 13:14:44 08/10/05
Go up one level in this thread
On August 10, 2005 at 06:10:32, Eelco de Groot wrote:
>On August 10, 2005 at 05:20:20, Darrel Briley wrote:
>
>>[D]r1b1kb1r/3n1p1p/1q3p2/4p3/1p1p4/3B1N2/PP2QPPP/R1BR2K1 w kq - 0 16
>>
>>From a game last night on ICC. Shredder 9 played 16. Bc4 as first move out of
>>book. Looking at it after the game it seems 16. Nxd4 is a shot. Shredder took
>>about seven and a half minutes to see it(AthlonXP 2600+ 512Mb hash). How long
>>do other engines take?
>>
>>complete score...
>>
>>[TimeControl "2700+15"]
>>
>>1. d4 Nf6 2. Nf3 d5 3. c4 e6 4. Nc3 c6 5. e3 Nbd7 6. Bd3 dxc4 7. Bxc4 b5 8.
>>Bd3 a6 9. e4 c5 10. e5 cxd4 11. Nxb5 axb5 12. exf6 Qb6 13. O-O gxf6 14. Qe2
>>b4 15. Rd1 e5 16. Bc4 Be7 17. Nh4 Ba6 18. Qf3 Ra7 19. Bxa6 Rxa6 20. Bh6 Bf8
>>21. Bxf8 Nxf8 22. Rac1 Qb8 23. Rc6 Qa8 24. Qxf6 Qxc6 25. Qxe5+ Qe6 26. Qxh8
>>f6 27. Nf3 Kf7 28. h4 Rd6 29. Re1 Qxa2 30. h5 Re6 31. Rd1 Re4 32. Rc1 Qxb2
>>33. Rc5 Qb1+ 34. Kh2 Re8 35. Rc7+ Re7 36. Rc8 Re8 37. Rxe8 Kxe8 38. Qxf6 Qe4
>>39. Nxd4 Nd7 40. Qd6 Qe5+ 41. Qxe5+ Nxe5 42. Nb3 Nd3 43. Kg3 Ke7 44. f4 Kd6
>>45. Kf3 Kd5 46. g4 Ne1+ 47. Kg3 Nc2 48. f5 Nd4 49. Nxd4 Kxd4 50. f6 b3 51.
>>f7 b2 52. f8=Q b1=Q 53. Kh4 Kd5 54. Qf7+ Ke5 55. Qc7+ Kf6 56. Qf4+ Kg7 57.
>>h6+ Kg6 58. Qg5+ Kf7 59. Qf5+ Qxf5 60. gxf5 Kf6 61. Kg4 Ke7 62. Kg3 Kf6
>>{Game drawn by mutual agreement}
>>1/2-1/2
>>
>>Odd play at end due to missing semicolon separating TB paths
>>
>> DB
>
>15..e5 is clearly a big blunder. Pro Deo 1.1 Default needs little time to find
>16.Nxd4, 36 seconds, but so does Shredder 9 UCI, 36 seconds on my Athlon 3200+,
>100 Mb HT. Are you sure your set-up is right? 7 minutes to find it seems not
>exactly right.
>
>Regards, Eelco
>
>
>Pro Deo 1.1 default 200 Mb HT
>
>00:00:00.3 -0,82 4 9172 Bf4 Ba6 a4 bxa3ep bxa3
>00:00:00.3 -0,78 4 10014 a4 bxa3ep bxa3 Bb4
>00:00:00.3 -0,78 4 12037 Nd2 Bb7 a4 bxa3ep bxa3
>00:00:00.3 -0,52 4 15457 Bf5 Be7 Qc4 Nc5
>00:00:00.3 -0,58 5 40674 Bf5 Be7 Qc2 Nc5 Bxh7
>00:00:00.3 -0,48 5 54165 Bb5 Qe6 Bc4 Qe7 Bd2
>00:00:00.3 -0,39 5 68846 Bxh7 b3 Bb1 Ba6 Qe4
>00:00:00.3 -0,66 6 95670 Bxh7 Nc5 Bd3 Nxd3 Rxd3 Ba6 Qe4
>00:00:00.4 -0,57 6 134176 Bb5 Qe6 Bc4 Qe7 Bd2 Bb7
>00:00:00.6 -0,64 7 348475 Bb5 Be7 Bh6 Ba6 Qe4 Rc8 Bg7
>00:00:00.8 -0,51 7 514536 Bc4 Bc5 Qe4 Bb7 Qf5 Qd6 Bh6
>00:00:00.9 -0,48 7 657466 Bd2 Qe6 Bc4 Qd6 Rac1 Bb7 Ne1
>00:00:01.3 -0,57 8 972214 Bd2 Bb7 Bf5 Be7 Rdc1 Nc5 Qc4 Bxf3
>00:00:02.2 -0,43 8 1883205 Bb5 Qe6 Bc4 Qe7 Qe4 Ra5 Bd2 Qd8
>00:00:03.2 -0,54 9 3012178 Bb5 Qe6 Bc4 Qe7 Qe4 Ra7 Nd2 Bb7 Qg4
>00:00:08.1 -0,49 10 8321477 Bb5 Be7 Bd2 Qd6 Qc4 Qe6 Qc7
>00:00:12.8 -0,42 10 13427526 Bf5 Be7 Re1 Ba6 Qe4
>00:00:29.0 -0,41 11 31854499 Bf5 Be7 Qe4 Ra5 Qg4 Bf8 Re1
>00:00:35.8 -0,41 11 39706878 Nxd4
>00:00:40.5 0,03 11 44668494 Nxd4 Qxd4 Be3 Qd5
>00:01:35.2 0,71 12 103770043 Nxd4 Bb7 Nf5 Rg8
Eelco,
The results on my system are very strange, and I've as yet been unable to
determine the problem. I deleted Shredder's PL2 file and reran the position.
It still takes a very long time to find 16. Nxd4. It takes 3:46 to see the
move, and much longer to see it as clearly winning.
Shredder 9.1 UCI
9/30 0:01 +0.30 16.Bc4 Be7 17.Bh6 Bb7 18.Qd3 Ra5 19.Nd2 f5 (485.081) 397
10/24 0:01 +0.34 16.Bc4 Be7 17.Bh6 Bb7 18.Rac1 Rg8 19.Nh4 Nc5 20.f3 Ne6
(714.366) 408
11/33 0:03 +0.56 16.Bc4 Bc5 17.Bh6 Bb7 18.Nh4 Ba6 19.Qe4 Rc8 20.Bg7 Bb7
21.Qf5 (1.339.126) 417
12/33 0:04 +0.65 16.Bc4 Bc5 17.Bh6 Bb7 18.Nh4 Ba6 19.Nf5 Rg8 20.Ng7+ Ke7
21.Rac1 Bxc4 22.Qxc4 Raf8 (2.033.143) 425
13/33 0:08 +0.71 16.Bc4 Bc5 17.Bh6 Ba6 18.Qe4 Ke7 19.Bd5 (3.766.578) 430
14/38 0:15 +0.69 16.Bc4 Bc5 17.Bh6 Ba6 18.Qe4 Bb7 19.Qf5 Rg8 20.Qxh7
(6.592.341) 436
15/42 0:34 +0.63 16.Bc4 Bc5 17.Bh6 Ba6 18.Qe4 Ke7 19.Bd5 Rag8 20.Nh4 Ke8
21.Nf5 Rg6 22.Rac1 Rhg8 23.g3 (15.240.429) 437
16/45 1:16 +0.71 16.Bc4 Bc5 17.Bh6 Rg8 18.Nh4 Nf8 19.Nf5 Rg6 20.Bb5+ Nd7
21.Bd3 (33.610.501) 439
17/46 2:41 +0.46-- 16.Bc4 Bc5 (71.693.923) 443
17/46 2:55 +0.46 16.Bc4 Be7 17.Bh6 Bb7 18.Nh4 Rc8 19.Rdc1 Rg8 20.Qh5 Rxc4
21.Rxc4 Qe6 22.Rac1 Qg4 (77.836.587) 444
17/48 3:46 +0.47++ 16.Nxd4 (100.340.432) 442
17/48 3:50 +0.82++ 16.Nxd4 (102.101.332) 442
17/48 9:47 +2.47 16.Nxd4 Ba6 17.Nf5 Nc5 18.Be3 Bxd3 19.Rxd3 Qe6 20.Bxc5 Qxf5
21.Rf3 Qe6 22.Bxb4 Bxb4 23.Qb5+ (254.916.898) 433
best move: Nf3xd4 time: 9:55.750 min n/s: 433.558 CPU 99.6% n/s(1CPU):
435.299 nodes: 258.279.158
(The original figure of seven and a half minutes was taken from Chesspartner
after some backsolving)
So I tried Pro Deo 1.1 in the Shredder Classic GUI as well, 100Mb hash:
Pro Deo 1.1
8.00 0:01 -0.57 16.Bd2 Bb7 17.Bf5 Be7 18.Rdc1 Nc5 19.Qc4 Bxf3 (972.214) 972
8.03 0:02 -0.43 16.Bb5 Qe6 17.Bc4 Qe7 18.Qe4 Ra5 19.Bd2 Qd8 (1.883.187) 941
9.01 0:03 -0.54 16.Bb5 Qe6 17.Bc4 Qe7 18.Qe4 Ra7 19.Nd2 Bb7 20.Qg4
(3.012.014) 1004
10.01 0:08 -0.49 16.Bb5 Be7 17.Bd2 Qd6 18.Qc4 Qe6 19.Qc7 (8.322.077) 1040
10.04 0:13 -0.42 16.Bf5 Be7 17.Re1 Ba6 18.Qe4 (13.423.966) 1032
11.01 0:30 -0.41 16.Bf5 Be7 17.Qe4 Ra5 18.Qg4 Bf8 19.Re1 (31.893.162) 1063
11.17 0:36 -0.41 16.Nxd4 (38.970.943) 1082
11.17 0:41 +0.03 16.Nxd4 Qxd4 17.Be3 Qd5 (44.108.249) 1075
12.01 1:41 +0.71 16.Nxd4 Bb7 17.Nf5 Rg8 18.Ng3 Rg7 (104.264.682) 1032
13.01 4:27 +0.95 16.Nxd4 Ba6 17.Nf5 Rg8 18.Bxa6 Qxa6 19.Qh5 Qc4 20.Ne3 Qe4
21.Rxd7 Kxd7 22.Qxf7+ (276.038.016) 1033
14.01 7:57 +1.00 16.Nxd4 Ba6 17.Nf5 Qc6 18.Bxa6 Rxa6 19.Qh5 h6 20.Nxh6
(498.860.017) 1045
best move: Nf3xd4 time: 8:01.516 min n/s: 1.045.828 nodes: 498.860.017
Here my numbers are very much the same as yours. Very odd.
I've checked everything I can think of, but have found no cause of the very long
solve time for Shredder 9 UCI. My system is very slightly overclocked (1917Mhz
Barton core 11.5 x 166) running at 2070Mhz (11.5 x 180). but has been running
completely stable at these settings for 2 years, nevertheless I even reduced bus
speed, memory timings and core voltage to default, the only result being it
takes even longer to find the move. Very odd. If anyone has any ideas please
let me know.
DB
P.S. I've also tried it with various hash sizes, and also in the Chessbase GUI
with the same result.
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.