Rybka 3 with new GUI: are they serious?!

[smirobth]

The second one is awesome, thanks a lot George!

Yes, awesome indeed. Here is another hard one that computers don't solve based in part on the same theme.
[d] 8/8/K7/1p6/3B1p1p/5P2/1P5p/5B1k w - - 0 1
1.Bh3! wins

[Albert Silver]

How else they can solve NOLOT-3 in practical play time? Or find the following mate in practical time again? We need their ELO to increase!

I tested them with Rybka Winfinder (64-bit) on a slow Athlon64 2.2GHz with 512MB hash.

[D]r2qk2r/ppp1b1pp/2n1p3/3pP1n1/3P2b1/2PB1NN1/PP4PP/R1BQK2R w KQkq - 0 1

1.Nxg5!!

[+0.84] d=21 1.Nxg5 Bxd1 2.Nxe6 Qb8 3.Nxg7 Kf7 4.Bh6 Bg4 5.O–O Kg8 6.Rf4 Bg5 7.Rxg4 Bxh6 (1:40.09)
[+0.78] d=21 1.Nxg5 (1:18.25)
[+0.39] d=21 1.Nxg5 (1:14.19)
[+0.42] d=21 1.Nxg5 (1:08.17)
[+0.39] d=21 1.h3 Bxf3 2.gxf3 Nf7 3.Qc2 Bh4 4.Qg2 Qe7 5.Bd2 Na5 6.O–O–O O–O–O 7.Nh5 Nc4 (0:47.49)
[+0.39] d=20 1.h3 Bxf3 2.gxf3 Nf7 3.Qc2 Bh4 4.Qg2 Qe7 5.Bd2 Na5 6.O–O–O O–O–O 7.Nh5 Nc4 (0:22.26)
[+0.39] d=19 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 Bxg3 6.Rxg3 Qh4 7.Qf2 Qe7 (0:08.56)
[+0.39] d=18 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 Bxg3 6.Rxg3 Qh4 7.Qf2 Qe7 (0:04.13)
[+0.39] d=17 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 Bxg3 6.Rxg3 Qh4 7.Qf2 Qe7 (0:02.00)
[+0.39] d=16 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 Bxg3 6.Rxg3 Qh4 7.Qf2 Qe7 (0:01.04)
[+0.33] d=15 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 g6 6.Be3 Ng5 (0:00.32)
[+0.33] d=14 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 O–O 5.Qc2 g6 6.Be3 (0:00.19)
[+0.33] d=13 1.h3 Bxf3 2.gxf3 Nf7 3.f4 Bh4 4.Rg1 Qe7 5.f5 exf5 (0:00.13)
[+0.33] d=12 1.h3 Bxf3 2.gxf3 Qd7 3.f4 Ne4 4.Nxe4 dxe4 5.Bxe4 O–O–O 6.Qg4 (0:00.11)
[+0.22] d=12 1.h4 Nxf3 2.gxf3 Bxh4 3.fxg4 Bxg3 4.Kd2 Bh4 5.Kc2 O–O 6.Be3 (0:00.09)
[+0.11] d=12 1.Bxg5 Bxg5 2.h3 Bxf3 3.Qxf3 Rf8 4.Qh5 Kd7 5.Qxh7 Qe7 6.Qg6 (0:00.04)
[+0.16] d=11 1.Bxg5 Bxg5 2.h3 Bxf3 3.Qxf3 Rf8 4.Qh5 Kd7 5.Qxh7 Qe7 (0:00.02)
[+0.16] d=10 1.Bxg5 Bxg5 2.h3 Bxf3 3.Qxf3 Rf8 4.Qh5 Kd7 5.Rf1 (0:00.00)
[+0.16] d=9 1.Bxg5 Bxg5 2.h3 Bxf3 3.Qxf3 Rf8 4.Qh5 Kd7 (0:00.00)
[+0.16] d=8 1.Bxg5 Bxg5 2.h3 Bxf3 3.Qxf3 Rf8 (0:00.00)

On a decent quad, it should easily find this in less than 15 mins

[D]8/8/7K/8/p1p5/2Bp4/pn1N4/k3N3 w - - 0 1

1.Nxd3!! mate in 13

RybkaWinFinderv2.2.x64:
4 00:00 1.368 43.776 +0.60 Nd2xc4 Ka1b1
5 00:00 2.662 56.789 +0.60 Nd2xc4 Ka1b1 Bc3xb2
6 00:00 5.525 88.400 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3
7 00:00 11.036 102.735 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q
8 00:00 20.697 122.507 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2
9 00:00 37.276 142.961 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2 Bb2xa1
10 00:01 69.661 157.120 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2 Bb2xa1 Ka2xa1 Nd2c4
11 00:01 127.790 167.336 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2 Bb2xa1 Ka2xa1 Nd2c4 Ka1a2 Nd3f2
11 00:01 189.815 167.995 +4.00 Bc3e5 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Be5f4+ d3d2 Nc3xa2+ Kc1b1 Na2c3+ Kb1c1 Kh6g5
12 00:02 216.144 166.539 +4.00 Bc3e5 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Be5f4+ d3d2 Nc3xa2+ Kc1b1 Na2c3+ Kb1c1 Ne1f3
13 00:03 443.147 171.757 +0.40 Bc3e5 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Be5f4+ d3d2 Nc3xa2+ Kc1d1 Ne1f3 Nb2d3 Bf4g5
13 00:03 443.147 171.757 +0.50 Nd2xc4
13 00:03 443.147 171.757 +0.70 Nd2xc4
13 00:04 717.523 187.291 +0.70 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2 Bb2xa1 Ka2xa1 Nd2c4 Ka1b1 Nc4e3 a4a3
13 00:05 950.227 186.977 +4.00 Bc3f6 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Bf6g5+ d3d2 Nc3xa2+ Kc1d1 Ne1f3 Nb2d3 Nf3xd2
14 00:06 1.018.418 185.364 +4.20 Bc3f6 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Bf6g5+ d3d2 Nc3xa2+ Kc1d1 Ne1f3 Nb2d3 Nf3xd2 Kd1c2
15 00:09 1.547.880 184.091 +4.00 Bc3f6 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Bf6g5+ d3d2 Nc3xa2+ Kc1d1 Ne1f3 Nb2d3 Nf3xd2 Kd1c2
16 00:13 2.378.958 184.270 +4.00 Bc3f6 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Bf6g5+ d3d2 Nc3xa2+ Kc1b1 Na2c3+ Kb1c1 Ne1f3 Nb2d3
17 00:23 4.087.199 186.394 +0.40 Bc3f6 a4a3 Nd2e4 Ka1b1 Ne4c3+ Kb1c1 Bf6g5+ d3d2 Nc3xa2+ Kc1b1 Na2c3+ Kb1c1 Ne1f3 Nb2d3
17 00:24 4.087.199 186.394 +0.50 Nd2xc4
17 00:31 4.087.199 186.394 +0.40 Nd2xc4
17 00:38 7.276.464 196.158 +0.50 Nd2xc4 Ka1b1 Bc3xb2 d3d2 Ne1d3 a2a1Q Nc4xd2+ Kb1a2 Bb2xa1 Ka2xa1 Nd2c4 Ka1b1 Nc4a3+ Kb1a2
17 00:52 7.276.464 196.158 +0.60 Ne1xd3
17 00:52 7.276.464 196.158 +M381 Ne1xd3
17 00:52 10.107.723 198.088 -M2 Ne1xd3 c4xd3 Bc3h8 a4a3 Kh6g7 Nb2d1 Kg7g6+ Nd1b2 Kg6f6 Nb2d1 Kf6f5+ Nd1b2 Kf5e5 Nb2d1 Ke5e4+ Nd1b2 Ke4d4 Nb2d1 Kd4xd3+ Nd1b2+ Kd3c3

Albert

[smirobth]

The second one is awesome, thanks a lot George!

Yes, awesome indeed. Here is another hard one that computers don't solve based in part on the same theme.
[d] 8/8/K7/1p6/3B1p1p/5P2/1P5p/5B1k w - - 0 1
1.Bh3! wins

I should have said "some programs don't solve". Fritz has a terrible time for some reason, but I just checked with other prorams and they don't!?

[Dann Corbit]

If Rybka 3 is delayed until new GUI is ready we must wait years! I think
chessok has no skills to make good interface. And do we need new one -
there is already 2 interfaces without any major flaws or missing functions...

Jouni

The big money will come from a GUI upgrade. Look at ChessMaster -- not the strongest engine by far, but larger sales than everything else put together.

Personally, I want both (improved GUI and improved engine).

I guess that GUI development will have no impact whatsoever on engine development, because different teams will be working on the GUI and the Engine. However, I suspect that new engine features (tied to the GUI) are going to be present in the engine.

[Tanya Deborah]

When Elo increase it generally means better positional knowledge understanding. The problem is that some times this diminish the tactical abilities of the program.

Note that, Rybka Winfinder is clearly inferior than Rybka 2.3.2a, but it can solve the second position where 2.3.2a fails completely.


8/8/7K/8/p1p5/2Bp4/pn1N4/k3N3 w - - 0 1

Analysis by Rybka WinFinder 2.2 32-bit:

1.Nxc4 Kb1
+/= (0.60) Depth: 4 00:00:00
1.Nxc4 Kb1 2.Bxb2
+/= (0.60) Depth: 5 00:00:00
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3
+/= (0.70) Depth: 6 00:00:00 6kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q
+/= (0.70) Depth: 7 00:00:00 11kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2
+/= (0.70) Depth: 8 00:00:00 21kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2 5.Bxa1
+/= (0.70) Depth: 9 00:00:00 37kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2 5.Bxa1 Kxa1 6.Nc4
+/= (0.70) Depth: 10 00:00:00 70kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2 5.Bxa1 Kxa1 6.Nc4 Ka2 7.Nf2
+/= (0.70) Depth: 11 00:00:01 128kN
1.Be5 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bf4+ d2 5.Nxa2+ Kb1 6.Nc3+ Kc1 7.Kg5
+- (4.00) Depth: 11 00:00:02 190kN
1.Be5 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bf4+ d2 5.Nxa2+ Kb1 6.Nc3+ Kc1 7.Nf3
+- (4.00) Depth: 12 00:00:03 216kN
1.Be5 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bf4+ d2 5.Nxa2+ Kd1 6.Nf3 Nd3 7.Bg5
+/= (0.40) Depth: 13 00:00:05 443kN
1.Nxc4
+/= (0.50) Depth: 13 00:00:05 443kN
1.Nxc4
+/= (0.70) Depth: 13 00:00:06 443kN
1.Nxc4
+/= (0.70) Depth: 13 00:00:06 443kN
1.Nxc4
+/= (0.70) Depth: 13 00:00:06 443kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2 5.Bxa1 Kxa1 6.Nc4 Kb1 7.Ne3 a3
+/= (0.70) Depth: 13 00:00:08 718kN
1.Bf6 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bg5+ d2 5.Nxa2+ Kd1 6.Nf3 Nd3 7.Nxd2
+- (4.00) Depth: 13 00:00:11 950kN
1.Bf6 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bg5+ d2 5.Nxa2+ Kd1 6.Nf3 Nd3 7.Nxd2 Kc2
+- (4.20) Depth: 14 00:00:12 1018kN
1.Bf6 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bg5+ d2 5.Nxa2+ Kd1 6.Nf3 Nd3 7.Nxd2 Kc2
+- (4.00) Depth: 15 00:00:19 1548kN
1.Bf6 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bg5+ d2 5.Nxa2+ Kb1 6.Nc3+ Kc1 7.Nf3 Nd3
+- (4.00) Depth: 16 00:00:29 2379kN
1.Bf6 a3 2.Ne4 Kb1 3.Nc3+ Kc1 4.Bg5+ d2 5.Nxa2+ Kb1 6.Nc3+ Kc1 7.Nf3 Nd3
+/= (0.40) Depth: 17 00:00:49 4087kN
1.Nxc4
+/= (0.50) Depth: 17 00:00:52 4087kN
1.Nxc4
+/= (0.40) Depth: 17 00:00:55 4087kN
1.Nxc4
+/= (0.40) Depth: 17 00:00:59 4087kN
1.Nxc4
+/= (0.40) Depth: 17 00:01:08 4087kN
1.Nxc4 Kb1 2.Bxb2 d2 3.Nd3 a1Q 4.Nxd2+ Ka2 5.Bxa1 Kxa1 6.Nc4 Kb1 7.Na3+ Ka2
+/= (0.50) Depth: 17 00:01:23 7276kN
1.Nxd3
+/= (0.60) Depth: 17 00:01:53 7276kN
1.Nxd3
+- (#381) Depth: 17 00:01:53 7276kN
1.Nxd3
+- (#381) Depth: 17 00:01:53 7276kN
1.Nxd3
+- (#381) Depth: 17 00:01:53 7276kN
1.Nxd3 cxd3 2.Bh8 a3 3.Kg7 Nd1 4.Kg6+ Nb2 5.Kf6 Nd1 6.Kf5+ Nb2 7.Ke5 Nd1 8.Ke4+ Nb2 9.Kd4 Nd1 10.Kxd3+ Nb2+ 11.Kc3
-+ (-#2) Depth: 17 00:01:53 10108kN

(,JRC 09.04.2008)

Regards,
Tanya Deborah

[Sylwy]

When Elo increase it generally means better positional knowledge understanding......."

Regards,
Tanya Deborah

Nope ! An ELO increase generally means a deeply search and -in very rare situations- somewhat a best evaluation function . Positional knowledges ( & tactical-of courses )means a very elaborated evaluation function. Von Neumann's architecture of our computers limits this desideratum.


Regards,
Silvian

[ml]

Fritz 5.32 finds Bh3 in less than a second and evaluates it in favour of white. Fritz 10 and 11 also find Bh3 immediately but after a while end up scoring it as a draw. This might be one of those cases where a weaker engine can produce a better evaluation than an overall stronger one.

[Dr.Wael Deeb]

Fritz 5.32 finds Bh3 in less than a second and evaluates it in favour of white. Fritz 10 and 11 also find Bh3 immediately but after a while end up scoring it as a draw. This might be one of those cases where a weaker engine can produce a better evaluation than an overall stronger one.

It depends on the position the wise man said

[ml]

Fritz 5.32 finds Bh3 in less than a second and evaluates it in favour of white. Fritz 10 and 11 also find Bh3 immediately but after a while end up scoring it as a draw. This might be one of those cases where a weaker engine can produce a better evaluation than an overall stronger one.

It depends on the position the wise man said

True.

[oreopoulos]

By they way. Convecta created ChessAssistant which has a zillion more powerfull features than chessbase. Chessbase is plain marketing and kinderganrden features.

If one can bring chess software to a next level, its them.
Its also obvious that if one program has 10 functions and the other 10000 then its obvious you will have a harder time finding the 10000 functions.