Supertactical position shootout - 3 engines (2-10 ply)

[Martin Thoresen]

[d]1r6/1Ppr4/1P1p3R/K1PNqk2/2Bpn1pB/4pn1p/1Q4b1/4R3 w - - 0 1

White to move

----- Ply = 2 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit 1-0
Glaurung 2.0.1 64-bit 4CPU 0-1
----- Ply = 3 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit 0-1
Glaurung 2.0.1 64-bit 4CPU 0-1
----- Ply = 4 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit 1-0
Glaurung 2.0.1 64-bit 4CPU 0-1
----- Ply = 5 -----
Rybka 2.3.2a 64-bit 4CPU 1-0
Hiarcs X54 Hypermodern 32-bit 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
----- Ply = 6 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit 1-0
Glaurung 2.0.1 64-bit 4CPU 1-0
----- Ply = 7 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit 1-0
Glaurung 2.0.1 64-bit 4CPU ½-½
----- Ply = 8 -----
Rybka 2.3.2a 64-bit 4CPU 1-0
Hiarcs X54 Hypermodern 32-bit ½-½
Glaurung 2.0.1 64-bit 4CPU ½-½
----- Ply = 9 -----
Rybka 2.3.2a 64-bit 4CPU ½-½
Hiarcs X54 Hypermodern 32-bit ½-½
Glaurung 2.0.1 64-bit 4CPU 1-0
----- Ply = 10 -----
Rybka 2.3.2a 64-bit 4CPU 0-1
Hiarcs X54 Hypermodern 32-bit ½-½
Glaurung 2.0.1 64-bit 4CPU 1-0

Score White: 48%, 27 Games


Rybka points with white 2.5
Rybka points with black 6.5

Hiarcs points with white 5.5
Hiarcs points with black 3.5

Glaurung points with white 5.0
Glaurung points with black 4.0

[Martin Thoresen]

Updated results 4-13 ply:

Shoot out:
----- Ply = 4 -----
Deep Shredder 11 64-bit 4CPU 1-0
Loop M1-T 64-bit 4CPU 0-1
Naum 2.2 64-bit 4CPU 0-1
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU 0-1
Rybka 2.3.2a 64-bit 4CPU 0-1
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 0-1
Scorpio 2.0 4CPU 1-0
----- Ply = 5 -----
Deep Shredder 11 64-bit 4CPU 0-1
Loop M1-T 64-bit 4CPU 0-1
Naum 2.2 64-bit 4CPU 0-1
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU 0-1
Rybka 2.3.2a 64-bit 4CPU 0-1
Zappa Mexico II 64-bit 4CPU ½-½
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU 1-0
----- Ply = 6 -----
Deep Shredder 11 64-bit 4CPU ½-½
Loop M1-T 64-bit 4CPU 0-1
Naum 2.2 64-bit 4CPU 0-1
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU 0-1
Rybka 2.3.2a 64-bit 4CPU ½-½
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 0-1
Scorpio 2.0 4CPU 0-1
----- Ply = 7 -----
Deep Shredder 11 64-bit 4CPU 0-1
Loop M1-T 64-bit 4CPU 1-0
Naum 2.2 64-bit 4CPU ½-½
Toga II 1.4.2SE 4CPU 0-1
Bright 0.3b 4CPU 0-1
Rybka 2.3.2a 64-bit 4CPU 0-1
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU 1-0
----- Ply = 8 -----
Deep Shredder 11 64-bit 4CPU ½-½
Loop M1-T 64-bit 4CPU 1-0
Naum 2.2 64-bit 4CPU 0-1
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU 1-0
Rybka 2.3.2a 64-bit 4CPU ½-½
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 0-1
Scorpio 2.0 4CPU ½-½
----- Ply = 9 -----
Deep Shredder 11 64-bit 4CPU ½-½
Loop M1-T 64-bit 4CPU 1-0
Naum 2.2 64-bit 4CPU 0-1
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU ½-½
Rybka 2.3.2a 64-bit 4CPU 0-1
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU ½-½
Scorpio 2.0 4CPU ½-½
----- Ply = 10 -----
Deep Shredder 11 64-bit 4CPU ½-½
Loop M1-T 64-bit 4CPU 1-0
Naum 2.2 64-bit 4CPU ½-½
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU 1-0
Rybka 2.3.2a 64-bit 4CPU ½-½
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU 1-0
----- Ply = 11 -----
Deep Shredder 11 64-bit 4CPU 0-1
Loop M1-T 64-bit 4CPU 1-0
Naum 2.2 64-bit 4CPU 1-0
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU ½-½
Rybka 2.3.2a 64-bit 4CPU ½-½
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU 1-0
----- Ply = 12 -----
Deep Shredder 11 64-bit 4CPU 1-0
Loop M1-T 64-bit 4CPU ½-½
Naum 2.2 64-bit 4CPU 1-0
Toga II 1.4.2SE 4CPU 1-0
Bright 0.3b 4CPU ½-½
Rybka 2.3.2a 64-bit 4CPU 0-1
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU 1-0
----- Ply = 13 -----
Deep Shredder 11 64-bit 4CPU 1-0
Loop M1-T 64-bit 4CPU 0-1
Naum 2.2 64-bit 4CPU 1-0
Toga II 1.4.2SE 4CPU 0-1
Bright 0.3b 4CPU ½-½
Rybka 2.3.2a 64-bit 4CPU 1-0
Zappa Mexico II 64-bit 4CPU 0-1
Glaurung 2.0.1 64-bit 4CPU 1-0
Scorpio 2.0 4CPU ½-½

Score White: 48%, 90 Games


Deep Shredder points with white 5.0
Deep Shredder points with black 5.0

Loop points with white 5.5
Loop points with black 4.5

Naum points with white 4.0
Naum points with black 6.0

Toga points with white 8.0
Toga points with black 2.0

Bright points with white 4.0
Bright points with black 6.0

Rybka points with white 3.0
Rybka points with black 7.0

Zappa points with white 0.5
Zappa points with black 9.5

Glaurung points with white 6.5
Glaurung points with black 3.5

Scorpio points with white 7.5
Scorpio points with black 2.5




Toga scored most with white (8.0) while Zappa had a special 9.5 score with black!

[jdart]

Depth to solve doesn't really mean much. When an engine says "depth 4" how deep it is actually searching and how "widely" it searches (i.e. how many nodes it is pruning) will differ from one engine to another. What really counts is how fast a solution is found : in a real game, you have a limited time to move and it does not matter if you find the right move by reaching a shallow or deep nominal depth.

[Martin Thoresen]

Depth to solve doesn't really mean much. When an engine says "depth 4" how deep it is actually searching and how "widely" it searches (i.e. how many nodes it is pruning) will differ from one engine to another. What really counts is how fast a solution is found : in a real game, you have a limited time to move and it does not matter if you find the right move by reaching a shallow or deep nominal depth.

Jon,

True. But I don't think this position has any practical value, I just find it interesting that it is "perceived" so differently by different engines. For instance, Zappa didn't even manage to win this position by playing white. Maybe Zappa needs more plies (than 13) in order to do that. Toga, on the other hand, only won twice with black, the rest was either white wins or draws.

Since this is a most unclear position, it is difficult to draw any conclusions from my scientific tests.

What seems to be the case is that 1.c6 is a bad move. 1.Rh5+ seems to be a better starting point.

Regards,
Martin

[Ignacio]

What seems to be the case is that 1.c6 is a bad move. 1.Rh5+ seems to be a better starting point.

1.c6! is a good move. Example: 1... Nxh4 2.cd7 c6 3.Rh5+ Ng5 4.Ka6

[Martin Thoresen]

What seems to be the case is that 1.c6 is a bad move. 1.Rh5+ seems to be a better starting point.

1.c6! is a good move. Example: 1... Nxh4 2.cd7 c6 3.Rh5+ Ng5 4.Ka6

Yes, you're right, I mixed the moves up. I meant that c6 is the good move, while Rh5+ seems to lose.

Here is a depth 16 analysis by Rybka 2.3.2a 64-bit 4CPU @ 3.4 GHz:

1r6/1Ppr4/1P1p3R/K1PNqk2/2Bpn1pB/4pn1p/1Q4b1/4R3 w - - 0 1

1. +- (1.67): 1.c6 cxb6+ 2.Qxb6 Rdxb7 3.cxb7 Nxh4 4.Qc7 Rxb7 5.Qxb7 Kg5 6.Qh7 Ng3 7.Rf6 Ngf5
2. = (0.00): 1.Rxe3 cxb6+ 2.Ka6 dxe3 3.Nxe3+ Kf4 4.Nd5+ Qxd5 5.Bxd5 Nxc5+ 6.Kb5 Bf1+ 7.Bc4 Bxc4+
3. ³ (-0.57): 1.Rh5+ Kg6 2.Rxe5 Nxe5 3.Bb5 cxb6+ 4.cxb6 Nc5 5.Nc7 Rf7 6.Bg3 Nc6+ 7.Bxc6 Bxc6
4. -+ (-1.98): 1.Qb5 cxb6+ 2.cxb6 Rdxb7 3.Bd3 g3 4.Rh5+ Kg4 5.Rxe5 Nxe5 6.Be7 Nc5 7.Be2+ Kf5
5. -+ (-2.84): 1.Qb1 cxb6+ 2.cxb6 Nxh4 3.Rh5+ Kg6 4.Rxe5 dxe5 5.Nc7 Kh5 6.Bb5 Rd6 7.Qa2 Rxb7