Good Test.....g4......?

[rightrook]

[d]2r5/p2n2p1/1p1Bp1k1/4P1p1/PPp5/4P2P/4K1P1/2R5 w - - 0 1

Any programs actually play g4......the move that was played today by Anand...?

Just wondering if this is a good test move or not...

Robert

[Eelco de Groot]

I don't think that 31. g4 is a very difficult move to make. Toga needed 9 plies. Anand has a comfortable position here and he only needs a draw, to get 5 points. He only needs 6½ points total. As for Kramnik he needs to take chances with White in the next game, that is if he has not already resigned to losing the match without a further fight, so for Anand there enough possibilities for playing for a win with Black again. Kramnik is not to be envied! He seems in such a bad shape that maybe it did not come as much of a surprise to him. Really no idea what is going on and Kramnik will only talk about it after the match is over. Maybe Anand's preparation was just to good and he has been preparing this change in repertoire for a long time. Hat off to Vishy and his seconds, including Magnus! Curious what role he played!


[d]2r5/p2n2p1/1p1Bp1k1/4P1p1/PPp5/4P2P/4K1P1/2R5 w - -

Engine: Toga Mara Beta 6 (256 MB)
by Fabien Letouzey, Thomas Gaksch and Tam


9/20 0:00 -0.14 31.Kd2 a6 32.Kc3 b5 33.axb5 axb5
34.Ra1 Nb6 35.Ra7 Nd5+ 36.Kd4 (133.059)

9/20 0:00 -0.05 31.g4 a6 32.b5 axb5 33.axb5 Kh6
34.Kd2 Kh7 35.e4 Kg6 (163.723)

10/20 0:00 -0.14 31.g4 a6 32.b5 axb5 33.axb5 Kh7
34.Kd2 Kh6 35.Rf1 c3+ 36.Kc2 Kg6 (268.738)

11/26 0:01 0.00 31.g4 a6 32.b5 axb5 33.axb5 Kh7
34.Kd2 Kg6 35.Ke2 (665.718) 590

12/26 0:02 +0.33 31.g4 a6 32.b5 axb5 33.axb5 Kh7
34.e4 Kh6 35.Ke3 Kg6 36.Kd4 Rh8
37.Rxc4 Rxh3 (1.612.649) 660

13/29 0:03 +0.29 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.e4 g6 35.Ke3 Rh8 36.Kd4 Rxh3
37.Kxc4 Kg7 38.Kd4 (2.710.342) 688

14/31 0:08 +0.45 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.Rc2 Kg8 35.Kd2 Kf7 36.Kc3 Kg6
37.Kd4 Rh8 38.Rxc4 Rxh3 (6.241.845) 728

15/31 0:17 +0.48 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.Kd2 Kg6 35.Rc2 Kf7 36.Kc3 Rh8
37.Kxc4 Rxh3 38.Rf2+ Kg6 39.e4 Rh8 (12.896.813) 752

16/33 0:34 +0.48 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.Kd2 Kg6 35.Rc2 Kf7 36.Kc3 Rh8
37.Kxc4 Rxh3 38.Kd3 Rh1 39.Ke4 Kg6 (26.340.872) 761

17/39 1:16 +0.49 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.Kd2 Kg6 35.Rc2 Kf7 36.Kc3 Rh8
37.Kxc4 Rxh3 38.Kd3 Rh1 39.Rc7 Ke8
40.Ra7 (58.795.204) 766

best move: g2-g4 time: 2:34.125 min n/s: 761.955 CPU 100.0% n/s(1CPU): 761.955 nodes: 117.210.000

[Eelco de Groot]

I don't think that 31. g4 is a very difficult move to make. Toga needed 9 plies.

Naum needs a bit longer though, 19 plies even in two best moves mode:


2r5/p2n2p1/1p1Bp1k1/4P1p1/PPp5/4P2P/4K1P1/2R5 w - -

Engine: Naum 3.1 (256 MB)
by Aleksandar Naumov

19 0:58 +0.36 31.Kd2 a6 32.Ra1 Kf5 33.Rf1+ Kg6 34.Kc3 b5 35.a5 Re8 36.e4 Rd8 37.Rf2 Rc8 38.Rd2 Kh5 39.Rd1 g4 40.h4 (55.350.662) 941

19 0:46 +0.28 31.b5 g4 32.Kd2 gxh3 33.gxh3 Kh5 34.Rg1 g5 35.Rg4 Rd8 36.Kc3 Rc8 37.Kc2 c3 38.e4 Kh6 39.h4 gxh4 40.Rxh4+ (43.475.224) 932
___________________________________________________

19 1:07 +0.42 31.g4 a6 32.b5 axb5 33.axb5 Kf7 34.Kd2 Kg6 35.Kc3 Nc5 36.Rc2 Nb3 37.Ra2 Na5 38.Ra1 Kh6 39.Bb4 Nb3 40.Ra6 (64.294.754) 948

19 0:58 +0.36 31.Kd2 a6 32.Ra1 Kf5 33.Rf1+ Kg6 34.Kc3 b5 35.a5 Re8 36.e4 Rd8 37.Rf2 Rc8 38.Rd2 Kh5 39.Rd1 g4 40.h4 (55.350.662) 941
___________________________________________________

20 1:11 +0.38 31.g4 a6 32.b5 axb5 33.axb5 Kf7 34.Kd2 Kg6 35.Kc3 Nc5 36.Rc2 Nb3 37.Ra2 Na5 38.Ra1 Kh6 39.Bb4 Nb3 40.Ra6 Nc1 (68.105.430) 952

20 1:23 +0.30 31.Kd2 a6 32.Ra1 Kf5 33.Rf1+ Kg6 34.Kc3 b5 35.a5 Re8 36.Rd1 Kf5 37.Bc7 Nf8 38.Rf1+ Ke4 39.Rf7 Ng6 40.Rxg7 Ne7 (79.508.586) 957 TB:3
___________________________________________________

21 1:42 +0.58 31.g4 a6 32.b5 axb5 33.axb5 Kf7 34.Kd2 Kg6 35.Rc2 Rd8 36.Kc3 Rh8 37.Kxc4 Rxh3 38.Kd3 Rh8 39.Rc7 Rd8 40.Kc3 Kf7 41.Kd4 (99.260.508) 971 TB:3

21 2:22 +0.37 31.Kd2 a6 32.e4 b5 33.axb5 axb5 34.Ra1 Rc6 35.g4 Nf6 36.Ke3 Ne8 37.Bc5 Nc7 38.Kd4 Na6 39.Ra3 Nxc5 40.bxc5 Rc7 41.Ra6 (138.420.743) 974 TB:3
___________________________________________________

22 2:44 +0.70 31.g4 a6 32.b5 axb5 33.axb5 Kf7
34.Kd2 Kg6 35.Rc3 Nf6 36.exf6 Rd8
37.fxg7 Rxd6+ 38.Ke2 Kxg7 39.Rxc4 Rd5
40.Rc6 Rxb5 41.Rxe6 Rb1 (161.703.098) 982 TB:3

22 4:40 +0.34 31.Kd2 a6 32.Ra1 Kf5 33.Rf1+ Kg6
34.Kc3 b5 35.a5 Re8 36.Rd1 Kf5
37.Bc7 Nf8 38.Rf1+ Kg6 39.e4 Rc8
40.Bb6 Nd7 41.Bd4 Rd8 (272.700.251) 971 TB:107

best move: g2-g4 time: 5:02.156 min n/s: 971.165 nodes: 287.830.078 TB: 107

[ArmyBridge]

Toga Mara?? I tought that Toga Mara was a Rybka 3 clone??

[Golem]

Toga Mara?? I tought that Toga Mara was a Rybka 3 clone??

Of course, it is !

[Ovyron]

Toga Mara?? I tought that Toga Mara was a Rybka 3 clone??

There are two, the legit Toga Mara and a renamed Rybka 3.

[meschle]

Glaurung 2.1 clone finds in 10 secs - quicker than rybka 3 on single cpu hardware



Glaurung:
2 00:00 115 3.709 +0.19 Ke2d2 Kg6f5
2 00:00 198 6.387 +0.21 e3e4 c4c3
3 00:00 423 13.645 +0.43 e3e4 a7a6 Ke2d2
4 00:00 837 27.000 +0.47 e3e4 a7a6 Ke2d2 b6b5
5 00:00 1.780 37.872 +0.52 e3e4 a7a6 Ke2e3 c4c3 Ke3d3
6 00:00 3.914 63.129 +0.13 e3e4 g5g4 h3xg4 Kg6g5 Ke2f3 c4c3
6 00:00 11.373 145.807 +0.41 Ke2f3 c4c3 Kf3e4 a7a6 Ke4d3 Kg6f5
7 00:00 20.148 161.184 +0.52 Ke2f3 Kg6h5 Kf3e4 g5g4 Ke4d4 g4xh3 g2xh3
8 00:00 29.225 169.912 +0.31 Ke2f3 a7a6 Kf3e4 b6b5 a4a5 Rc8h8 Ke4d4 Kg6f5
9 00:00 43.008 196.383 +0.23 Ke2f3 a7a6 Kf3e4 b6b5 Ke4d4 b5xa4 Rc1xc4 Rc8xc4+ Kd4xc4 Kg6f5 b4b5
9 00:00 62.380 235.396 +0.50 e3e4 a7a6 Ke2e3 b6b5 Ke3d4 b5xa4 Rc1xc4 Rc8xc4+ Kd4xc4 Nd7b6+ Kc4c5 Nb6d7+ Kc5d4
10 00:00 70.993 227.541 +0.45 e3e4 a7a6 Ke2e3 b6b5 Ke3d4 b5xa4 Rc1xc4 Rc8xc4+ Kd4xc4 g5g4 h3xg4 a4a3
11 00:00 126.014 298.611 +0.17 e3e4 g5g4 h3xg4 Kg6g5 Ke2f3 c4c3 Bd6e7+ Kg5g6 Kf3f4 a7a6 Be7d6 c3c2 Kf4e3
11 00:00 198.581 373.975 +0.25 Ke2d2 a7a6 Kd2c3 b6b5 a4xb5 a6xb5 Kc3d4 Rc8h8 Bd6c5 Rh8d8 Bc5d6
12 00:00 419.049 439.715 +0.43 Ke2d2 a7a6 Kd2c3 Kg6f5 Kc3d4 b6b5 e3e4+ Kf5g6 a4a5 Nd7b8 Kd4c3 Nb8c6 Rc1f1
13 00:01 598.879 479.103 +0.37 Ke2d2 a7a6 Kd2c3 Kg6f5 Kc3d4 b6b5 e3e4+ Kf5g6 Rc1c3 b5xa4 Rc3xc4 Rc8xc4+ Kd4xc4 Nd7b6+ Kc4c3 g5g4 h3xg4
14 00:02 1.223.258 543.670 +0.60 Ke2d2 a7a6 Kd2c3 Kg6f5 Rc1f1+ Kf5g6 b4b5 a6a5 Rf1c1 g5g4 h3xg4 Nd7c5 Bd6xc5 Rc8xc5 Rc1e1 Rc5c8
15 00:04 2.450.367 591.876 +0.56 Ke2d2 a7a6 e3e4 b6b5 a4xb5 a6xb5 Rc1a1 g5g4 h3xg4 Rc8d8 Kd2c3 Rd8c8 Ra1a5 Rc8h8 Bd6c5
16 00:05 3.592.319 613.128 +0.47 Ke2d2 a7a6 e3e4 b6b5 a4xb5 a6xb5 Rc1a1 g5g4 h3xg4 Rc8c6 Kd2c2 Nd7f6 e5xf6 Rc6xd6 e4e5 Rd6d5 Ra1f1 Rd5xe5
16 00:07 4.774.120 623.578 +0.56 e3e4 g5g4 h3xg4 a7a6 b4b5 a6xb5 a4xb5 Kg6g5 Ke2e3 Kg5xg4 Ke3d4 g7g5 Rc1xc4 Rc8xc4+ Kd4xc4 Kg4g3 Kc4d4
16 00:10 6.917.212 637.883 +0.66 g2g4 Kg6f7 b4b5 Kf7g6 Ke2d2 Kg6f7 e3e4 Kf7g6 Kd2e3 Rc8h8 Ke3d4 Rh8xh3 Rc1xc4 Rh3h2 Kd4d3 Kg6f7
17 00:11 7.294.271 641.254 +0.66 g2g4 Kg6f7 b4b5 Kf7g6 Ke2d2 Kg6f7 e3e4 Kf7g6 Kd2e3 Rc8h8 Ke3d4 Rh8xh3 Rc1xc4 Rh3h1 Rc4c3 Rh1d1+ Kd4c4 Rd1h1
18 00:15 10.014.158 660.042 +0.78 g2g4 a7a6 b4b5 a6a5 e3e4 Kg6f7 Ke2e3 Rc8h8 Rc1h1 Rh8c8 Ke3d4 c4c3 Rh1f1+ Kf7g6 Rf1c1 c3c2 Kd4d3 Rc8h8 Rc1xc2 Rh8xh3+ Kd3d4
19 00:23 15.875.600 672.865 +0.86 g2g4 a7a6 b4b5 a6xb5 a4xb5 Kg6f7 Rc1c2 Kf7e8 Ke2d2 Ke8f7 Kd2c3 Rc8h8 Kc3xc4 Rh8xh3 Kc4d3 Rh3h8 Rc2c7 Rh8d8 Kd3d4
20 00:31 21.318.687 685.973 +0.96 g2g4 a7a6 b4b5 a6xb5 a4xb5 Kg6f7 Rc1c2 Kf7e8 Rc2a2 Nd7c5 Ra2a7 g7g6 Ke2d2 Nc5e4+ Kd2c2 Ke8d8 Bd6e7+ Kd8e8 Be7a3 Ke8d8 Ba3d6 Ne4f2 Ra7b7

[ernest]

Analysis by Rybka 3 32-bit, Core 2 Duo @3GHz (512MB hash):
1st try
1.g4
+/= (0.45 !) Depth: 20 00:01:21 6752kN
2nd try
1.g4
+/= (0.51 !) Depth: 20 00:01:17 6309kN, tb=1
3rd try
1.g4
+/= (0.60 !) Depth: 21 00:03:10 15781kN, tb=2
4th try
1.g4
+/= (0.45 !) Depth: 18 00:00:30 2579kN

[Eelco de Groot]

Toga Mara?? I tought that Toga Mara was a Rybka 3 clone??

There are two, the legit Toga Mara and a renamed Rybka 3.

There are at least three versions then, if I include my Toga Mara Beta 6 that I used for the analysis in the post above, and which is just a test of Taban's material.cpp in Toga Checkov. But any version that you will find at the moment will not be a real Toga. It is very unfortunate that these Rybka clones exist but I don't think that the author if Toga Mara is involved. Let me quote the last discussion there was about Toga Mara's code on the Toga Forum, it is private, that also means anything Tam posted there does not have much of an audience as hardly anybody has access, so I hope he does not mind that I post this:

Posted by Eelco de Groot on Wed Oct 01, 2008 6:04 am

Eelco,

Help me out here.Am I supposed to "patent" a code even when I release open source just to claim as the orginator? I thought this is against the spirit of open source. I know that once I release the entire code, the whole lot of engine programmers who had "retired" on seeing Rybka 3 strength will get back to business.

Cheers!

For the hackers/crackers: Common guys, be original, don't use Tam or Toga Mara.

Hi Tam,

Sorry if my suspicions were all unfounded. I really would like nothing more than the Toga Mara to succeed! But I think I have just read one too many Ben Lau posts in the Rybka forum, about his next great engine that just turned out to be a stupid clone of the Strelka clone.

The Rybka 3 clones that were distributed as Togas were not reassuring me much..

I don't think and did not say that your material imbalance code did not have great potential, or I would never have said it could be something that is part of Rybka 3's secret. I hardly dare to believe it is truely all your code, really I am It would be easy to verify this if the Toga Mara code would become public, I would believe what Vas would have to say about it.

It is also that your descriptions of how your new code works have been far less convincing than the published code itself but maybe that is just a language problem. I really would enjoy seeing some more

My regards,
Eelco

Just to bring you people up to speed a bit. My own version of Toga Mara Beta, if I may call it that, is just Toga II 1.4beta5c, with my Checkov modifications of Beta 4 and Taban's material.cpp which is the only part of Toga Mara that we know something about. I modified Tam's code with a lot of basically randomly chosen constants and this still seems not worse than the old material.cpp. So I am convinced there is some powerful material there, literally, for which Taban should get the credit. Taban is his surname in the code, I suppose Tam is a shortened version of it.

Mare van Toga Mara, Tidings of Toga Mara Regards,
Eelco