13 hard test positions July 2016

[peter]

That all sounds very interesting, Eelco!
Do you have a 64bit- compile of your latest Rainbow Serpent for free download

[Eelco de Groot]

Hi Peter,

I am not exactly sure if I can make that yet, because I was told my TDM-GCC is building statically linked executables, so that means it includes some libraries that other builds would get from their own operating system. It is not explicitly stated anywhere that a compile I make would not run on other systems. I only know from old experiences with MSVC 2005 that I had to make dynamically linked executables in my Windows XP environment if they were to run in 64 bit operating systems for instance. Here I am on Windows 10 64 bit so that is different from my 32 bit MSVC.

I will try to upload something maybe tonight, but for certain results, and something that would certainly run a bit faster too, it would be best to install your own TDM compiler (or any C++11 Posix threads compiler) and make a fresh compile with that on your own system. TDM-GCC is relatively easy to install because it is one of the few GCCs that have a Windows installer.


(trivia: I am very pleased with the nodes per second I get with TDM, especially if I skip nullmove a lot your NPS goes up Of course that is a completely meaningless measurement in this case, because in a nullmove node nothing gets stored (only in the call to verification search). So if I skip Nullmove, it goes to regular search and that will store a result and a move etc.)

[peter]

Many thanks, Eelco, but take your time, no hurry necessary.
BTW, I gave Sting a second try (I'm rather sure, I did let it compute once for about 5 minutes without anything but 0.00, but I'm not quite sure, that it was 7.2, and of course it simply could have been SMP- unfortune then and or good luck now:

Code: Select all

1B1k4/3p4/1Kp5/PpP2p2/1PqP1P2/8/1P6/8 w - - 0 1

Analysis by Sting SF 7.2 64bit:
...
1.a6 Kc8 2.a7 Da2 3.Lc7 Da4 4.Le5 Da1 5.Lc7 Da2 6.Ld8 Da4 7.Lc7 Da1 8.Ld6 Da2 9.Lc7 
  =  (0.00)   Tiefe: 36   00:00:24  241MN
1.a6 
  =  (0.08 ++)   Tiefe: 37   00:00:28  286MN
1.a6 
  =  (0.16 ++)   Tiefe: 37   00:00:29  292MN
1.a6 Kc8 2.a7 Da2 3.Lc7 Da1 4.b3 Da3 5.Le5 Da2 6.Lb8 Da3 7.Le5 
  =  (0.00)   Tiefe: 37   00:00:33  334MN
...
1.a6 Kc8 2.a7 Da2 3.Lc7 Da1 4.b3 Da2 5.Le5 Da3 6.Lf6 Da2 7.Le5 
  =  (0.28)   Tiefe: 40   00:01:18  827MN
...
1.a6 Kc8 2.a7 Da2 3.Lc7 Da1 4.b3 Da3 5.Le5 Da2 6.Lf6 Da3 7.Lh4 Da2 8.Lf2 Da3 9.Le3 Da2 10.Ld2 Da1 11.Lc3 Da3 12.Le1 Da1 13.Lg3 Da2 14.Lf2 
  +-  (1.65)   Tiefe: 41   00:01:32  981MN
...
1.a6 Kc8 2.a7 Da2 3.Lc7 Da1 4.b3 Da3 5.Le5 Da2 6.Lf6 Da3 7.Lh4 Da1 8.Le1 Da2 9.Ld2 Da3 10.Lc3 Da2 11.Lb2 Kd8 12.Kb7 Dxb2 13.a8D+ Ke7 14.d5 Dxb3 15.d6+ Kf7 16.Dd8 De6 17.Kc7 
  +-  (1.65)   Tiefe: 42   00:01:42  1098MN
...
1.a6 Kc8 2.a7 Da2 3.Ld6 Da1 4.Lf8 Da4 5.Le7 Da1 6.Lh4 Da2 7.Lg5 Da1 8.b3 Da2 9.Lf6 Da3 10.Lh4 Da2 11.Lf2 Da3 12.Le1 Da1 13.Lh4 Da3 14.Le1 
  +-  (5.41)   Tiefe: 43   00:02:35  1829MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da1 4.b3 Da3 5.Lg7 Da2 6.Lf6 Da3 7.Lh4 Da1 8.Lf6 Da2 9.Lh4 Da3 10.Lg3 Da1 11.Le1 Da2 12.Lh4 
  +-  (5.21)   Tiefe: 44   00:03:32  2730MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da1 4.Lf6 Da4 5.b3 Da3 6.Lh4 Da2 7.Lf2 Da3 8.Le3 Da2 9.Ld2 Da1 10.Le1 Da2 11.Lf2 
  +-  (5.21)   Tiefe: 45   00:03:52  2983MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da4 4.Lf6 Da1 5.b3 Da2 6.Le7 Da1 7.Ld6 Da2 8.Lf8 Da3 9.Lh6 Da1 10.Lg5 Da2 11.Le7 
  +-  (5.69)   Tiefe: 46   00:04:24  3439MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da4 4.Lf6 Da1 5.b3 Da2 6.Lh4 Da3 7.Lg3 Da1 8.Lf2 Da2 9.Lh4 
  +-  (5.85)   Tiefe: 47   00:05:14  4187MN
...
1.a6 Kc8 2.a7 Da2 3.Lc7 Da1 4.b3 Da3 5.Le5 Da2 6.Lf6 Da3 7.Lh4 Da2 8.Lf2 Da1 9.Le1 Da2 10.Ld2 Da1 11.Le3 Da3 12.Lc1 Da1 13.Ld2 Da2 14.Le1 Da1 15.Lg3 Da2 16.Lf2 
  +-  (5.93)   Tiefe: 48   00:08:41  7848MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da1 4.Lf6 Da4 5.Lh4 Da1 6.Lf2 Da4 7.b3 Da1 8.Le1 Da2 9.Ld2 Da1 10.Lc1 d5 11.Ld2 Da2 12.Le1 Da1 13.Lh4 Da2 14.Lf2 Da3 15.Le3 Da1 16.Lc1 Kd7 17.Lb2 Da2 18.Kb7 Dxb2 19.a8D Dxd4 20.Dc8+ Ke7 21.Dc7+ Kf8 22.Dxc6 Dg7+ 23.Kb6 Dg1 24.Dxd5 Kg7 25.Dxf5 Dh2 26.Kb7 Kg8 27.c6 Dh1 28.Dxb5 Kf8 29.Df5+ 
  +-  (7.59)   Tiefe: 49   00:14:13  13903MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da1 4.Lf6 Da4 5.Le7 Da1 6.Lh4 Da4 7.Lf2 Da2 8.Le1 Da4 9.Lf2 
  +-  (11.11)   Tiefe: 50   00:27:50  29543MN
...
1.a6 Kc8 2.a7 Da2 3.Le5 Da4 4.Lf6 Da1 5.Lh4 Da4 6.b3 Da3 7.Lg3 Da2 8.Lf2 Da1 9.Le1 Da2 10.Lh4 Da1 11.Lf6 Da2 12.Lh4 
  +-  (11.91)   Tiefe: 51   00:45:59  50493MN
...
1.a6 
  +-  (15.05 ++)   Tiefe: 52   01:12:34  79858MN

I gave it a second try after starting the Fritz14-GUI for new already and again Sting 7.2 began to raise the eval in depth 40 after about 50" this time, so I'd say, it wasn't 7.2 I tried last time.
Congrats to Marek Kwiatkowski too for this version of his engine

[peter]

but I'm not quite sure, that it was 7.2

Had a look at Marek's posting about his new version again:

http://www.talkchess.com/forum/viewtopi ... 45&t=61012

He gives the position of this study too as an example of 7.2's progress there

[Vinvin]

I ran an overnight analyze on this 12 positions ( #3 was avoided).
Conditions : 10 min/positions, CPU : 6 cores 4 GHz, 8 GB HT, 6 pcs Syzygy, 4 runs.
Analyzing engine: Stockfish_160922_x64_modern_fast

Code: Select all

   Use seconds per move      : 600 Seconds per move
   Analysis Lines            : Minimum search depth=12
   Activate abort analysis   : True. Minimal analysis depth 24. Abort, if the correct move is held 3 additional ply 

  Solution                 1st run               2nd run               3rd run               4th run
 1) b2-b4                b2-b4   * 593 Sec     b2-b4   * 164 Sec     b2-b4   * 48 Sec      b2-b4   * 60 Sec  
 2) a7-a8Q, a7-a8R       Bc7-b6                Bc7-b6                Bc7-b6                Bc7-b6            
 3) Be3-d4               Rh7-a7                Rh7-a7                Rh7-a7                Rh7-a7            
 4) a5-a6                Kb6-b7                Kb6-b7                a5-a6   * 0 Sec       Kb6-b7            
 5) f5xg6                f5xg6   * 6 Sec       Bc7-e5                f5xg6   * 7 Sec       f5xg6   * 9 Sec   
 6) Ne5xd3               Ne5xd3  * 32 Sec      Ne5xd3  * 0 Sec       Ne5xd3  * 0 Sec       Ne5xd3  * 166 Sec 
 7) g3-g4                Be2-d1                Be2-d1                Be2-d1                Rh1-h3            
 8) .. Be1xh4            Be1xh4  * 2 Sec       Be1xh4  * 150 Sec     Be1xh4  * 244 Sec     Be1xh4  * 16 Sec  
 9) Nc3-e4               Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec   
10) Rf4xf5               Rf4xf5  * 10 Sec      Rf4xf5  * 5 Sec       Rf4xf5  * 4 Sec       Rf4xf5  * 8 Sec   
11) Bd3xh7               Bd3xh7  * 64 Sec      Bd3xh7  * 54 Sec      h4-h5                 Bd3xh7  * 59 Sec  
12) Qe2xd2, Ne5-g4       Rd1xd2                Rd1xd2                Ne5-g4  * 174 Sec     Rd1xd2   

[mcostalba]

I ran an overnight analyze on this 12 positions ( #3 was avoided).
Conditions : 10 min/positions, CPU : 6 cores 4 GHz, 8 GB HT, 6 pcs Syzygy, 4 runs.
Analyzing engine: Stockfish_160922_x64_modern_fast

Code: Select all

   Use seconds per move      : 600 Seconds per move
   Analysis Lines            : Minimum search depth=12
   Activate abort analysis   : True. Minimal analysis depth 24. Abort, if the correct move is held 3 additional ply 

  Solution                 1st run               2nd run               3rd run               4th run
 1) b2-b4                b2-b4   * 593 Sec     b2-b4   * 164 Sec     b2-b4   * 48 Sec      b2-b4   * 60 Sec  
 2) a7-a8Q, a7-a8R       Bc7-b6                Bc7-b6                Bc7-b6                Bc7-b6            
 3) Be3-d4               Rh7-a7                Rh7-a7                Rh7-a7                Rh7-a7            
 4) a5-a6                Kb6-b7                Kb6-b7                a5-a6   * 0 Sec       Kb6-b7            
 5) f5xg6                f5xg6   * 6 Sec       Bc7-e5                f5xg6   * 7 Sec       f5xg6   * 9 Sec   
 6) Ne5xd3               Ne5xd3  * 32 Sec      Ne5xd3  * 0 Sec       Ne5xd3  * 0 Sec       Ne5xd3  * 166 Sec 
 7) g3-g4                Be2-d1                Be2-d1                Be2-d1                Rh1-h3            
 8) .. Be1xh4            Be1xh4  * 2 Sec       Be1xh4  * 150 Sec     Be1xh4  * 244 Sec     Be1xh4  * 16 Sec  
 9) Nc3-e4               Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec   
10) Rf4xf5               Rf4xf5  * 10 Sec      Rf4xf5  * 5 Sec       Rf4xf5  * 4 Sec       Rf4xf5  * 8 Sec   
11) Bd3xh7               Bd3xh7  * 64 Sec      Bd3xh7  * 54 Sec      h4-h5                 Bd3xh7  * 59 Sec  
12) Qe2xd2, Ne5-g4       Rd1xd2                Rd1xd2                Ne5-g4  * 174 Sec     Rd1xd2   

So the real monster here is the 7:

[d] r1b4r/p4pk1/P3p1p1/1pBpPnPp/2pP1K2/2P3P1/4BP2/R6R w - - 0 1

Also Dann Corbit searched for half an hour until depth 52 with no result.

Do you know from which game is this one?

[Dann Corbit]

It is from this game:

[Event "Stockfish TCEC S92 vs Komodo 10, Slow 4"]
[Site "Ultimatum-PC"]
[Date "2016.07.13"]
[Round "1"]
[White "Komodo 10 64-bit"]
[Black "Stockfish 120716 64 BMI2"]
[Result "1-0"]
[ECO "C12h"]
[Annotator "0.00;0.00"]
[PlyCount "281"]
[TimeControl "25200+42"]

1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.Bg5 Bb4 5.e5 h6 6.Be3 Ne4 7.Qg4 Kf8 8.a3 Bxc3+
9.bxc3 Nxc3 10.Ne2 h5 11.Qf3 Nxe2 12.Bxe2 Nc6 13.h4 Ne7 14.Bg5 c5 15.O-O
Qc7 16.c3 Bd7 17.a4 Nc6 18.Qe3 Rb8 19.Bd3 Rc8 20.Rfb1 Na5 21.Qg3 b6 22.Ba6
Re8 23.Bd3 Rc8 24.Qf3 Be8 25.Ra2 Nc4 26.Qg3 Na5 27.Rc1 Bc6 28.Rf1 Bd7 29.
Rd1 Nc6 30.Rb1 Na5 31.Rd1 Nc6 32.Ba6 Rb8 33.Be2 Re8 34.Qe3 Ne7 35.Rc1 Nf5
36.Qf3 Kg8 37.Bd3 Qc6 38.Raa1 c4 39.Bc2 Bc8 40.Rcb1 Bd7 41.g3 Rb8 42.Bc1
Ra8 43.Ba3 Qc8 44.Bb4 Bc6 45.a5 b5 46.a6 Bd7 47.Bc5 Qe8 48.Bd1 Qc8 49.Bc2
Qe8 50.Qd1 Qb8 51.Qd2 Qd8 52.Bd1 Bc6 53.Bf3 Rh6 54.Re1 Kh8 55.Qg5 Qxg5 56.
hxg5 Rh7 57.Kg2 Bd7 58.Rh1 g6 59.Be2 Ng7 60.Rh2 Kg8 61.Rh3 Bc8 62.Kf3 Nf5
63.Kf4 Rh8 64.Rhh1 Kg7 65.Bf3 Kg8 66.Bd1 Kg7 67.Bf3 Kg8 68.Be2 Kg7 69.g4
hxg4 70.Bxg4 Rxh1 71.Rxh1 Bd7 72.Bxf5 gxf5 73.Ra1 Kg6 74.Rh1 Kg7 75.Be7
Rg8 76.Ra1 Bc6 77.Re1 Bd7 78.Ra1 Bc6 79.Re1 Bd7 80.f3 Bc6 81.Ra1 Bd7 82.
Bc5 Ra8 83.Rh1 Bc6 84.Be7 Rg8 85.Rf1 Bd7 86.Bc5 Ra8 87.Rc1 Bc6 88.Rh1 Bd7
89.Ra1 Bc6 90.Rb1 Kg6 91.Ra1 Kg7 92.Be7 Rg8 93.Rf1 Bd7 94.Bc5 Ra8 95.Rc1
Bc6 96.Rh1 Bd7 97.Rg1 Kg6 98.Rb1 Bc6 99.Rh1 Kg7 100.Rd1 Kg6 101.Rb1 Kg7
102.Be7 Rg8 103.Rh1 Bd7 104.Kg3 Bc6 105.Kf4 Bd7 106.Rh2 Bc8 107.Rb2 Bd7
108.Kg3 Bc6 109.Bc5 Kg6 110.Bxa7 Kxg5 111.Bc5 f4+ 112.Kf2 Rh8 113.Kg2 Kf5
114.a7 f6 115.exf6 Kxf6 116.Ra2 Ba8 117.Ra6 Rg8+ 118.Kf2 Rh8 119.Kg2 Rg8+
120.Kf2 Rh8 121.Rb6 Rh2+ 122.Ke1 Rc2 123.Rb8 Bc6 124.Rf8+ Kg5 125.Rc8 Bb7
126.Be7+ Kg6 127.Rb8 Bc6 128.Rb6 Ba8 129.Bb4 Ra2 130.Rxe6+ Kf5 131.Re5+
Kf6 132.Re7 Kg5 133.Rh7 Bc6 134.Kd1 Kg6 135.Rc7 Ba8 136.Bc5 Kh5 137.Rh7+
Kg6 138.Rh8 Bc6 139.Rc8 Ra1+ 140.Kd2 Ra2+ 141.Kc1 1-0

[Vinvin]

I ran an overnight analyze on this 12 positions ( #3 was avoided).
Conditions : 10 min/positions, CPU : 6 cores 4 GHz, 8 GB HT, 6 pcs Syzygy, 4 runs.
Analyzing engine: Stockfish_160922_x64_modern_fast

Code: Select all

   Use seconds per move      : 600 Seconds per move
   Analysis Lines            : Minimum search depth=12
   Activate abort analysis   : True. Minimal analysis depth 24. Abort, if the correct move is held 3 additional ply 

  Solution                 1st run               2nd run               3rd run               4th run
 1) b2-b4                b2-b4   * 593 Sec     b2-b4   * 164 Sec     b2-b4   * 48 Sec      b2-b4   * 60 Sec  
 2) a7-a8Q, a7-a8R       Bc7-b6                Bc7-b6                Bc7-b6                Bc7-b6            
 3) Be3-d4               Rh7-a7                Rh7-a7                Rh7-a7                Rh7-a7            
 4) a5-a6                Kb6-b7                Kb6-b7                a5-a6   * 0 Sec       Kb6-b7            
 5) f5xg6                f5xg6   * 6 Sec       Bc7-e5                f5xg6   * 7 Sec       f5xg6   * 9 Sec   
 6) Ne5xd3               Ne5xd3  * 32 Sec      Ne5xd3  * 0 Sec       Ne5xd3  * 0 Sec       Ne5xd3  * 166 Sec 
 7) g3-g4                Be2-d1                Be2-d1                Be2-d1                Rh1-h3            
 8) .. Be1xh4            Be1xh4  * 2 Sec       Be1xh4  * 150 Sec     Be1xh4  * 244 Sec     Be1xh4  * 16 Sec  
 9) Nc3-e4               Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec       Nc3-e4  * 0 Sec   
10) Rf4xf5               Rf4xf5  * 10 Sec      Rf4xf5  * 5 Sec       Rf4xf5  * 4 Sec       Rf4xf5  * 8 Sec   
11) Bd3xh7               Bd3xh7  * 64 Sec      Bd3xh7  * 54 Sec      h4-h5                 Bd3xh7  * 59 Sec  
12) Qe2xd2, Ne5-g4       Rd1xd2                Rd1xd2                Ne5-g4  * 174 Sec     Rd1xd2   

So the real monster here is the 7:

r1b4r/p4pk1/P3p1p1/1pBpPnPp/2pP1K2/2P3P1/4BP2/R6R w - - 0 1

Also Dann Corbit searched for half an hour until depth 52 with no result.

Do you know from which game is this one?

the #12 is a real headache for Stockfish too. Here it was very lucky because of SMP but SF is very bad for this position.
[d]r4rk1/pb3pp1/1p1qpR2/2ppN3/3P4/2P4P/PP1pQ1P1/3R2K1 w - - bm Qxd2 Ng4;

[peter]

Morgähn!
(That's "Guten Morgen + gähnen", in English about good morn-yawning )

the #12 is a real headache for Stockfish too. Here it was very lucky because of SMP but SF is very bad for this position.
[d]r4rk1/pb3pp1/1p1qpR2/2ppN3/3P4/2P4P/PP1pQ1P1/3R2K1 w - - bm Qxd2 Ng4;

That's from Carlsen-Tomashevsky Tata Steel 2016, isn't it?
Am not sure, but think I got this one from you and here already then. Thank you once again for collecting and sharing

[Vinvin]

Morgähn!
(That's "Guten Morgen + gähnen", in English about good morn-yawning )

the #12 is a real headache for Stockfish too. Here it was very lucky because of SMP but SF is very bad for this position.
[d]r4rk1/pb3pp1/1p1qpR2/2ppN3/3P4/2P4P/PP1pQ1P1/3R2K1 w - - bm Qxd2 Ng4;

That's from Carlsen-Tomashevsky Tata Steel 2016, isn't it?
Am not sure, but think I got this one from you and here already then. Thank you once again for collecting and sharing

Yes, Good memory !
I found the first post by Marc Moisan here : http://www.talkchess.com/forum/viewtopi ... 58&t=59033