White to Play and Win

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Terry McCracken
Posts: 16465
Joined: Wed Aug 01, 2007 4:16 am
Location: Canada

Re: White to Play and Win

Post by Terry McCracken »

PauloSoare wrote:
Terry McCracken wrote:
tjfroh wrote:Terry,

It takes Deep Rybka 3 25 seconds to see the winning move Bxh6. It then finds a better route by checking the black king with Bh7+ after 86 seconds.

Tj
Checking h7 isn't a better move. I doubt any computer/software combo can find 19. Bxh6!! with a positive winning score.
Terry, Q6600, HT=2048, Chessbase gui


New game
r1b2rk1/1p1nb1p1/1q2p2p/p2pP3/P7/3B1NP1/1PQ3KP/R1B2R2 w - - 0 1

Analysis by Deep Rybka 3:

1.Bc1-d2 Qb6-c5 2.Bd3-h7+
=/+ (-0.39) Depth: 6 00:00:00 9kN
1.Bd3-b5 Rf8-f7 2.Bc1-f4
=/+ (-0.37) Depth: 6 00:00:00 17kN
1.Bd3-b5 Rf8-f7 2.Bc1-f4 Qb6-c5 3.Qc2-d2
=/+ (-0.38) Depth: 7 00:00:00 23kN
1.Bd3-b5 Rf8-f5 2.Bb5-d3 Rf5-f7
=/+ (-0.48) Depth: 8 00:00:00 51kN
1.Bd3-b5 Rf8-f5 2.Bc1-d2 Qb6-c5 3.Qc2-d3 Qc5-a7
=/+ (-0.35) Depth: 9 00:00:00 91kN
1.Bd3-b5 Rf8-f5 2.Bc1-d2 Be7-c5 3.Bb5-d3 Rf5-f8 4.Ra1-c1 Bc5-e3
=/+ (-0.32) Depth: 10 00:00:01 161kN
1.Bc1-d2
= (-0.15 !) Depth: 10 00:00:02 260kN
1.Bc1-d2 Be7-b4 2.Bd2xb4 Qb6xb4 3.Ra1-c1 Ra8-b8 4.Bd3-h7+ Kg8-h8 5.Bh7-d3 b7-b6
= (-0.21) Depth: 10 00:00:02 416kN
1.Bc1-d2 Be7-b4 2.Bd2-c3 Bb4xc3 3.b2xc3 Qb6-d8 4.Bd3-h7+ Kg8-h8
= (-0.24) Depth: 11 00:00:03 584kN
1.Bc1-d2 Be7-b4 2.Bd2-c3 Bb4xc3 3.b2xc3 Qb6-c5 4.Ra1-e1 b7-b6 5.Nf3-d4 Rf8xf1 6.Kg2xf1 Qc5-e7
=/+ (-0.27) Depth: 12 00:00:04 793kN
1.Bc1-d2 Rf8xf3 2.Rf1xf3[] Nd7xe5 3.Bd3-h7+ Kg8-h8 4.Rf3-b3 Be7-b4 5.Bd2xb4 a5xb4 6.Ra1-e1 Ne5-g4
=/+ (-0.39) Depth: 13 00:00:16 3096kN
1.Bd3-b5
= (-0.07 !) Depth: 13 00:00:20 3706kN
1.Bd3-b5 Rf8-f5 2.Bc1-d2 Be7-b4 3.Bb5-d3 Rf5-f8 4.Bd2xb4
=/+ (-0.27) Depth: 13 00:00:22 3970kN
1.Bc1xh6 g7xh6[] 2.Bd3-h7+
= (0.00) Depth: 13 00:00:27 4802kN
1.Bc1xh6
= (0.20 !) Depth: 14 00:00:44 7022kN
1.Bc1xh6
+/= (0.40 !) Depth: 14 00:00:54 8188kN
1.Bc1xh6
+/- (0.80 !) Depth: 14 00:01:12 10723kN
1.Bc1xh6
+- (1.60 !) Depth: 14 00:01:55 17297kN
1.Bc1xh6 Rf8-f5 2.Bd3xf5[] e6xf5 3.Bh6-g5 Be7-c5 4.Qc2xf5 Nd7-f8 5.Qf5-d3 d5-d4 6.Bg5-d2 Bc8-d7 7.Ra1-c1 Nf8-e6
+- (2.16) Depth: 14 00:03:10 29976kN
1.Bd3-h7+ Kg8-h8[] 2.Bc1xh6[] Rf8-f5 3.Bh7xf5[] e6xf5 4.Bh6-g5 Be7-c5 5.Ra1-c1
+- (2.55) Depth: 14 00:03:11 30089kN

1.Bd3-h7+ +- (2.75 !) Depth: 15 00:03:51 37184kN1.Bd3-h7+ Kg8-h8[] 2.Bc1xh6[] Rf8-f5 3.Bh7xf5[] e6xf5 4.Bh6-g5 Be7-c5 5.Ra1-c1 Qb6-e6 6.Bg5-d2 Bc5-e7 7.Qc2-c7 Kh8-g8 8.Bd2-g5 Be7-b4[] 9.Qc7-d8+ Nd7-f8[] 10.Rf1-d1 Qe6-a6[] 11.Qd8xd5+ Bc8-e6[] 12.Qd5-d3 Be6-d7[] 13.Qd3xa6 Ra8xa6[] 14.b2-b3 Bd7-c6[] 15.Bg5-d2
+- (2.68) Depth: 15 00:04:05 39828kN
1.Bd3-h7+ Kg8-h8[] 2.Bc1xh6[] Rf8-f5 3.Bh7xf5[] e6xf5 4.Bh6-g5 Be7-c5 5.Ra1-c1 Qb6-e6 6.Bg5-d2 Bc5-e7 7.Qc2-c7 Kh8-g8 8.Bd2-g5 Be7-b4[] 9.Qc7-d8+ Nd7-f8[] 10.Rf1-d1 Qe6-a6[] 11.Qd8xd5+ Bc8-e6[] 12.Qd5-d3 Be6-d7[] 13.Qd3xa6 Ra8xa6[] 14.b2-b3 Bd7-c6[] 15.Bg5-d2
+- (2.68) Depth: 16 00:04:36 45053kN
1.Bd3-h7+ Kg8-h8[] 2.Bc1xh6[] Rf8-f5 3.Bh7xf5[] e6xf5 4.Bh6-g5 Be7-c5 5.Ra1-c1 Qb6-e6 6.Bg5-d2 Bc5-e7 7.Qc2-c7 Kh8-g8 8.Bd2-g5 Be7-b4[] 9.Qc7-d8+ Nd7-f8[] 10.Rf1-d1 Qe6-a6[] 11.Qd8xd5+ Bc8-e6[] 12.Qd5-d3 Be6-d7[] 13.Qd3xa6 Ra8xa6[] 14.b2-b3 Bd7-c6[] 15.Bg5-d2
+- (2.68) Depth: 17 00:05:43 56229kN

(Paulo, Internet 12.10.2008)
I'm impressed that Deep Rybka 3? 64bit? found a positive score in a reasonable amount of time!

I should have said that either Bxh6 or Bh7+ as they are basically equel.

If I weren't so pressed right now I'd want Deep Rybka 3 64bit!
Damn it's a good engine!

Terry
royb
Posts: 566
Joined: Thu Mar 09, 2006 12:53 am

Re: White to Play and Win

Post by royb »

Here are the results from my 1.6 GHz system (single core used):

Rybka3-64-bit 1 core (32 MB hash used):

4 -0.42 Bd2
5 -0.37 Bd2 Qc5 (0.07)
6 -0.44 Bd2 Qc5 Bh7+ (0.12)
6 -0.37 Bb5 Rf7 Bf4 (0.17)
7 -0.39 Bb5 Rf7 Bf4 Qc5 Qg6 (0.23)
8 -0.34 Bb5 Rf7 Bd2 Qc5 Qb3 (0.40)
9 -0.31 Bb5 Rf5 Bd3 Rf7 (0.96)
10 -0.34 Bb5 Rf5 Bd3 Rxf3 Rxf3 Nxe5 Bh7+ Kh8 Rb3 Qc6 Qxc6 bxc6 Bc2 Bf6 (1.89)
10 -0.32 Bd2 Rxf3 Rxf3 Nxe5 (4.58)
10 -0.11 b3 (7.64)
10 -0.17 b3 Kh8 Bb5 Rf5 Bb2 Qc5 Qe2 Qc7 (9.42)
11 -0.17 b3 Kh8 Bb5 Rf5 Bb2 Qc5 Qe2 Qc7 (11.24)
12 -0.09 b3 Kh8 Bb5 Rf5 Bb2 Qc5 Bd3 Qxc2+ Bxc2 Rf8 (15.79)
13 -0.12 b3 Kh8 Bb5 Rf5 Bb2 Qc5 Bd3 Qxc2+ Bxc2 Rf8 Nh4 Bxh4 (23.70)
13 +0.00 Bxh6 gxh6 Bh7+ (32.56)
14 +0.20 Bxh6 (74.00)
14 +0.40 Bxh6 (111.86)
14 +0.80 Bxh6 (162.97)
14 +1.25 Bxh6 gxh6 Bh7+ Kh8 Qg6 Nf6 exf6 Qxb2+ Nd2 Qxd2+ Kh1 Bxf6 Rf4 Bd7 Bg8 Rf7 Bxf7 Bg7 Bxe6 Bxe6 Qxe6 Qc2 Raf1 Qc6 Qxc6 bxc6 Rc1 Ra6 Rf7 Rb6 (251.35)
15 +1.45 Bxh6 (343.94)

And here are the results for a dual-core search (each Rybka3 gets 32 MB hash):

Rybka3-64-bit 2-cores:

4 -0.39 Bd2
5 -0.39 Bd2 Bb4 (0.10)
5 -0.34 Bb5 Rf7 (0.10)
6 -0.39 Bb5 Rf7 Bf4 (0.10)
7 -0.45 Bb5 Rf7 Bf4 Qc5 Qg6 Nf8 (0.16)
8 -0.26 Bb5 Rf7 Bd2 Qc5 Qg6 (0.46)
9 -0.45 Bd2 Qc5 Bh7+ Kh8 Qg6 Qb6 Qc2 Qc6 (1.21)
10 -0.25 Bd2 (1.75)
10 -0.22 Bd2 Bb4 Bxb4 Qxb4 Rac1 Rb8 Rfe1 b6 (2.44)
11 -0.25 Bd2 Bb4 Bxb4 Qxb4 Rac1 Rb8 Rfe1 b6 Nh4 (3.00)
12 -0.31 Bd2 Rxf3 Rxf3 Nxe5 Bh7+ Kh8 Rf4 Bd7 Bc3 Rc8 Raf1 Bf6 Qb1 Qa6 Bd4 (13.98)
12 -0.10 b3 Bb4 Bh7+ Kh8 (25.70)
13 +0.00 b3 Kh8 Bb5 Rd8 Bb2 Rf8 Bd4 (39.23)
14 -0.10 b3 Kh8 Bb5 Rf5 Bb2 Qd8 Bd3 Rf8 (48.84)
14 +0.20 Bxh6 (70.26)
14 +0.40 Bxh6 (87.53)
14 +0.80 Bxh6 (110.48)
14 +0.99 Bxh6 gxh6 Bh7+ (142.22)
15 +1.19 Bxh6 (184.86)
15 +1.39 Bxh6 (233.97)
15 +1.57 Bxh6 gxh6 Bh7+ Kh8 Qg6 Nf6 exf6 Qxb2+ Nd2 Qxd2+ Kh1 Bxf6 Rf4 Bd7 Bg8 Rf7 Bxf7 Bg7 Bxe6 Bxe6 Qxe6 Rb8 Raf1 Qd3 Qb6 Qa6 Qxa6 bxa6 Rd1 d4 (311.48)
16 +1.57 Bxh6 gxh6 Bh7+ Kh8 Qg6 Nf6 exf6 Qxb2+ Nd2 Qxd2+ Kh1 Bxf6 Rf4 Bd7 Bg8 Rf7 Bxf7 Bg7 Bxe6 Bxe6 Qxe6 Rb8 Raf1 Qd3 Qb6 Qa6 Qxa6 bxa6 Rd1 d4 (394.33)

In both cases, without a lot of time spent, Rybka-3 sees that Bxh6 is very strong.