Fruit does not see draw - and resigns!

[mclane]

whats the name of the GUI Andreas used ?

[Tord Romstad]

i have to disagree - strongly !!

I don't understand what you disagree with - indeed I don't see anything in your post which contradicts anything I wrote. The example you give is not about an evaluation rule which is only relevant in obscure special cases which hardly ever occur in practice, but about very general rules which can be applied to a very wide class of endgames.

Tord

[Werner]

Hi- here Fruit sees at once the draw:

maybe you did not use egbbs or egtb or they are not correct intilized!

Werner

FEN: k7/P4p2/P7/1KB2p2/8/8/8/8 b - - 0 1

Fruit070327-SF1agg:
1 00:00 18 0 0.00 f5f4 Kb5a4
2 00:00 56 0 0.00 f5f4 Kb5a4 f4f3
3 00:00 120 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3
4 00:00 238 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2
5 00:00 441 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2
6 00:00 790 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6
7 00:00 1.767 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4
8 00:00 3.411 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5
9 00:00 6.904 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3
10 00:00 12.225 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4
11 00:01 23.764 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4 Ka3a4
12 00:01 39.062 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4 Ka3a4 f4f3
13 00:01 66.482 0 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4 Ka3a4 f4f3 Ka4a3
14 00:01 92.818 0 -M7 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4 Ka3a4 f4f3 Ka4a3
15 00:01 185.349 180.000 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4a3 f5f4 Ka3a4 f4f3 Ka4a3
16 00:01 265.291 185.350 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
17 00:02 380.064 220.525 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
18 00:02 514.943 255.000 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
19 00:03 718.535 255.429 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
20 00:03 919.199 285.900 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
21 00:05 1.189.117 278.456 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
22 00:06 1.497.465 240.377 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3
23 00:09 1.827.153 212.942 0.00 f5f4 Kb5a4 f4f3 Ka4a3 f3f2 Lc5xf2 f7f6 Ka3a4 f6f5 Ka4b3 f5f4 Kb3a4 f4f3 Ka4a3

[mclane]

i want to give another example.

in this game against virtual chess, Chess System Tal is in a weak situation.
Virtual chess will win, but then CSTAL has an idea...

[Event "15th WMCC"]
[Site "Paris"]
[Date "1997.11.01"]
[Round "10"]
[White "Virtual Chess 2 / P2-300"]
[Black "Chess System Tal / P2-300"]
[Result "1-0"]
[ECO "B92"]
[PlyCount "152"]
[EventDate "1997.10.26"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be2 e5 7. Nb3 Be7 8. O-O
O-O 9. Be3 Be6 10. Qd2 Nbd7 11. a4 Rc8 12. a5 Qc7 13. Rfd1 Rfd8 14. Qe1 Qc6 15.
Bf3 h6 16. h3 Bxb3 17. cxb3 Nc5 18. b4 Ne6 19. Be2 Nf4 20. Bxf4 exf4 21. Rd4
Bf8 22. Qd2 Qe8 23. Qxf4 Qe6 24. Qd2 Rc7 25. b5 axb5 26. Bxb5 Rdc8 27. Qd3 Qe5
28. Bc4 Kh8 29. Bd5 Nh5 30. g3 Nf6 31. Bc4 Qh5 32. Kg2 Qe5 33. Ra4 Qg5 34. Nb5
Rd7 35. f4 Qh5 36. b4 Qg6 37. Nc3 Nh5 38. Nd5 Be7 39. f5 Qg5 40. h4 Qc1 41. b5
Nxg3 42. Nxe7 Rxe7 43. Kxg3 Qg1+ 44. Kf3 Qh1+ 45. Ke3 d5 46. Rxd5 Qe1+ 47. Kf3
Qh1+ 48. Ke3 Rce8 49. Rd4 Qxh4 50. Bd5 Rc7 51. Rdc4 Rd7 52. a6 bxa6 53. Rxa6
Qe1+ 54. Kf3 Qh1+ 55. Ke2 Kh7 56. b6 Re5 57. Qf3 Qb1 58. Ra5 Qxb6 59. Qc3 Ree7
60. f6 gxf6 61. Rc6 Qb1 62. Kf2 Rb7 63. Ra2 Qh1 64. Qxf6 Qh2+ 65. Ke3 Qg3+ 66.
Kd4 Qg1+ 67. Rf2 Qa1+ 68. Kd3 Qxf6 69. Rcxf6 Rb1 70. Rh2 Rd1+ 71. Kc2 Kg7 72.
Rfxh6 Rg1 73. Rh7+ Kf8 74. Rf2 Rc7+ 75. Kd2 Rg7 76. Rh8+ Rg8 1-0

[d]2r4k/1p1rbpp1/3p3p/PP1N1P1n/R1BRP2P/3Q2P1/6K1/2q5 b - - 0 41

sorry but i had no faster PC available at the moment

FEN: 2r4k/1p1rbpp1/3p3p/PP1N1P1n/R1BRP2P/3Q2P1/6K1/2q5 b - - 0 41

Glaurung_Lodz_2007:
2 00:00 398 0 +0,90 41. ... De1 42.Dc3
3 00:00 665 0 +3,19 41. ... De1 42.Sb6 Txc4 43.Taxc4
3 00:00 1.036 0 +2,37 41. ... Ld8 42.Dd1 Db2+ 43.Td2
3 00:00 1.592 31.840 +2,33 41. ... Kh7 42.Dd1 Dxd1 43.Txd1
4 00:00 3.088 19.300 +3,37 41. ... Kh7 42.Sb6 Txc4 43.Taxc4
4 00:00 5.590 34.937 +1,84 41. ... Ld8 42.De3 Dxe3 43.Sxe3 Kg8
5 00:00 8.155 38.833 +1,78 41. ... Ld8 42.Dc3 Dxc3 43.Sxc3 Lf6 44.Td3
6 00:00 10.769 39.885 +1,62 41. ... Ld8 42.Dc3 Dxc3 43.Sxc3 Lf6 44.Td3 Kg8
7 00:00 31.900 59.074 +2,15 41. ... Ld8 42.a6 bxa6 43.bxa6 Sf6 44.a7 Ta8 45.Sxf6 Lxf6
8 00:01 69.792 67.107 +2,03 41. ... Ld8 42.a6 bxa6 43.bxa6 Sf6 44.Dc3 Dxc3 45.Sxc3 Lb6 46.Td2
9 00:02 168.971 77.155 +2,45 41. ... Ld8 42.Dd2 Dxd2+ 43.Txd2 Sf6 44.Sxf6 Lxf6 45.a6 bxa6 46.bxa6 Kg8 47.a7 Ta8
10 00:10 848.561 84.015 +2,17 41. ... Ld8 42.Dd2 Dxd2+ 43.Txd2 Sf6 44.Sxf6 Lxf6 45.a6 bxa6 46.bxa6 Kg8 47.a7 Ta8 48.Tda2
11 00:17 1.537.154 87.737 +2,23 41. ... Ld8 42.Dd2 Dxd2+ 43.Txd2 Sf6 44.Lb3 Sxd5 45.Lxd5 Kg8 46.a6 bxa6 47.bxa6 Ta7 48.Tc4 Txc4 49.Lxc4
12 00:33 2.977.740 89.475 +2,19 41. ... Ld8 42.Dd2 Dxd2+ 43.Txd2 Sf6 44.Lb3 Tb8 45.Td3 Sxd5 46.Lxd5 Kg8 47.Kf3 Lf6 48.a6 bxa6 49.bxa6
13 01:47 9.513.273 88.693 +2,31 41. ... Ld8 42.Tb4 Da1 43.a6 Sf6 44.Tb1 Da5 45.Sxf6 Lxf6 46.axb7 Txb7 47.Txd6 Tbc7 48.Tc1 Kg8

after exchanging the sacced pieces the black queen can look for a perpetual check.
IMO the right strategy if the opponent leads +2.x pawns and you
have nothing in return, despite to sac a knight for chances for a perpetual-check.

An unsound sac for draw chances.

[mclane]

i have to disagree - strongly !!

I don't understand what you disagree with - indeed I don't see anything in your post which contradicts anything I wrote. The example you give is not about an evaluation rule which is only relevant in obscure special cases which hardly ever occur in practice, but about very general rules which can be applied to a very wide class of endgames.

Tord

you said:

Most programmers tend to concentrate on more general chess knowledge, which can be applied in a wider variety of positions.

and

all sorts of obscure endgames

the positions IS a general chess situation.
no matter how many pawns are placed on the a-file.

[Werner]

same here
Werner

Fruit by Fabien Letouzey, see http://www.fruitchess.com/
CPU0: AuthenticAMD AMD64 Family 15 Model 43 Stepping 1 2700 MHz
CPU1: AuthenticAMD AMD64 Family 15 Model 43 Stepping 1 2700 MHz
GUI: Tablebases mit max. 0 Steinen gefunden! [Cache: 16 MB + intern 0.00 MB]
GUI: Shredderbases mit 0 Steinen gefunden! (0 MB)
Engine: Fruit 070327 (256 MB)
von Fabien Letouzey, France
Stellung: 0 1 bm Ta7;
18.01 0:01 0.00 1...Ta7 2.Ld6 f4 3.Lxf4 f5 4.Ld6 f4 5.Lxf4 Txa6 (887.571) 582 TB:662
19.01 0:03 0.00 1...Ta7 2.Ld6 f4 3.Lxf4 f5 4.Ld6 f4 5.Lxf4 Txa6 (1.521.284) 503 TB:2.127
20.01 0:03 0.00 1...Ta7 2.Ld6 f4 3.Lxf4 f5 4.Ld6 f4 5.Lxf4 Txa6 (1.732.846) 506 TB:2.140
21.01 0:05 0.00 1...Ta7 2.Ld6 f4 3.Lxf4 f5 4.Ld6 f4 5.Lxf4 Txa6 (2.468.710) 474 TB:5.046
22.01 0:06 0.00 1...Ta7 2.Ld6 f4 3.Lxf4 f5 4.Ld6 f4 5.Lxf4 Txa6 (2.904.278) 442 TB:5.473
Bester Zug: Td7-a7 Zeit: 0:08.047 min K/s: 422.053 CPU 100.0% K/s(1CPU): 422.053 Knoten: 3.330.000 TB: 6.697

[Tord Romstad]

i have to disagree - strongly !!

I don't understand what you disagree with - indeed I don't see anything in your post which contradicts anything I wrote. The example you give is not about an evaluation rule which is only relevant in obscure special cases which hardly ever occur in practice, but about very general rules which can be applied to a very wide class of endgames.

Tord

you said:

Most programmers tend to concentrate on more general chess knowledge, which can be applied in a wider variety of positions.

and

all sorts of obscure endgames

the positions IS a general chess situation.
no matter how many pawns are placed on the a-file.

No, positions with multiple doubled rook pawns, no pawns on other files, a bishop of the wrong color, and the defending king placed in front of the pawns are extremely rare, and having knowledge about this exceptional case isn't worth even a single rating point. This is not a matter of opinion, but a simple statistical fact. Have a look at a big PGN database if you want, or try to deactivate this piece of knowledge in Glaurung (or any other program which has it), play a few million games, and see that you can't measure any difference in strength.

Tord

[Tord Romstad]

i want to give another example.

Another example of what? There is nothing in this game which has any relevance to what I (or anybody else) have written in this thread.

Tord

[Dr.Ex]

[d]kn6/2B5/8/1P6/1K6/P7/8/8 b - - 0 1

Tord is speaking about such cases.
He would find the draw immediately, an engine without that knowledge like Spike might play Nd7 and lose.

[Mike S.]

[d]kn6/2B5/8/1P6/1K6/P7/8/8 b - - 0 1

I found many positive examples of engines which find 1...Na6+ without tablebases (and 64 MB hash tables):

Analysis by Crafty 20.14:

1...Sa6+ 2.Ka5 Sxc7 3.a4 Ka7 4.Kb4 Se6 5.Ka5 Kb7 6.Kb4 Ka7
= (0.09) Tiefe: 4/6 00:00:00
= (0.00) Tiefe: 27/22 00:00:16 56537kN

Analysis by Hiarcs 9:

1...Sa6+ 2.bxa6
² (0.69) Tiefe: 5/11 00:00:00
= (0.05) Tiefe: 20/36 00:00:12 8879kN

Analysis by Fritz for Fun 4:

1...Sa6+ 2.Ka5 Sxc7 3.a4 Kb7 4.Kb4 Sd5+ 5.Kc5 Sb6 6.a5 Sd7+ 7.Kd6 Sf6 8.b6 Se4+ 9.Kd5 Sd2 10.a6+ Kxa6 11.b7
= (0.00) Tiefe: 7/11 00:00:00 2kN
= (0.00) Tiefe: 25/37 00:00:12 15819kN

Analysis by Shredder 9 SE:

1...Sa6+ 2.bxa6 Ka7 3.Ka5 Ka8 4.Kb5 Ka7 5.a4 Ka8 6.a5 Ka7 7.Kc5
² (0.42) Tiefe: 4/6 00:00:00
= (0.05) Tiefe: 35/35 00:00:17 21453kN

Analysis by The King 3.23:

1...Sa6+ 2.bxa6 Ka7 3.Kb5 Ka8 4.a7 Kxa7 5.Kc5 Ka6 6.Ld6 Kb7 7.Kc4 Ka8 8.Le5
= (0.06) Tiefe: 3 00:00:00
= (0.12) Tiefe: 22 00:00:13 6574kN

Analysis by Fruit 05/11/03:

1...Sa6+ 2.Kc4 Sxc7 3.b6 Kb8 4.bxc7+ Kxc7 5.a4 Kb7
² (0.38) Tiefe: 2/5 00:00:00
= (0.00) Tiefe: 31/43 00:00:18 33175kN

Analysis by Fruit 2.3.1 nb:

1...Sa6+ 2.bxa6 Ka7 3.Kb3 Ka8 4.a4 Ka7 5.Ka2 Ka8 6.Kb3
= (0.00) Tiefe: 1 00:00:00
= (0.00) Tiefe: 38 00:00:20 41424kN

Analysis by Toga II 1.3x4 (without bitbases):

1...Kb7
+- (3.43) Tiefe: 1/5 00:00:00
+- (3.43) Tiefe: 1/5 00:00:00
1...Sa6+ 2.bxa6 Ka7 3.Ka5 Ka8 4.Le5 Ka7 5.Lg3 Ka8 6.Lf2 Kb8 7.Kb5 Ka8 8.Lh4 Ka7 9.Le1 Ka8 10.Kb4 Ka7 11.Lf2+ Ka8 12.Kc3
² (0.46) Tiefe: 1/5 00:00:00
= (0.00) Tiefe: 19/33 00:00:05 8054kN

Analysis by Alaric 707 (has a kpk bitbase):

1...Sa6+ 2.bxa6 Ka7 3.Kb5 Ka8 4.a4 Ka7 5.Lb6+ Ka8 6.Kb4 Kb8 7.Ka3 Ka8 8.Ka2 Kb8 9.Ka1 Ka8 10.Kb1 Kb8 11.Kc1 Ka8 12.Kd1 Kb8 13.Ke1 Ka8 14.Kf1 Kb8 15.Kg1 Ka8 16.Kh2 Kb8 17.Lg1
= (0.01) Tiefe: 9 00:00:00 11kN
= (0.01) Tiefe: 31 00:00:18 29354kN