sting-sf-bh-11x ? Be afraid, be very much afraid !

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

Moderator: Ras

Dann Corbit
Posts: 12816
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by Dann Corbit »

Link to analysis results, modified source and binary (should run on most systems):

I have to make a few changes to Sting to get it to compile, link and run on my system.
And I also change the EPD format so that it is closer to the standard.
In the archive you will find at the root level vhard.epd which is a very hard to solve file of EPD problems.
To score 30/69 at three minutes per problem with 16 cores at 3.8GHz is an excellent score for a tactical engine.
sting-sf-bh-11x scores 40/69 at the stated conditions with UCI setting oldsting set to checked (scale your system's time and threads to mine for similar results).
But when I checked the blackhole checkbox (I left the oldsting setting checked also - not sure if you are supposed to do that) I got an astonishing 45/69 at three minutes with 16 cores at 3.8GHz.
Notes:
Sting does not scale well beyond 16 cores. 32 cores is right out.
Examine not only the solution count, but also how fast the problems were solved.
I recommend trying your own favorite tactical engine to see how well it performs.
Suppose you have an 8 core machine at 3.5GHz to get appropriate match on your system, multiply 3 minutes by (3.8/3.5) * (16/8)
so 2.1714285714285714285714285714286 * 180 seconds = 391 seconds per problem.
This buzzsaw of an engine is the neatest thing since sliced bread for an interesting category of hard problems.

Code: Select all

Analysis of G:\vhard.epd   
Analyzing engine: Sting
1/5/2026 6:21:13 PM Level: 180 Seconds
   Source                    : List of EPD/PGN files
      G:\vhard.epd
   Colours                   : White moves=True, Black moves=True
   Direction                 : forward
   Games (PGN)               : From 1 to 999999
   Moves                     : All moves
   Positions (EPD)           : From 1 to 999999 
   Use Engine(s)             : List
      Sting
      Sting-bh
   Use seconds per move      : 180 Seconds per move
   Analysis Lines            : Minimum search depth=2
   Activate abort analysis   : False
  1) Kc8-c7             g6-g7    
  2) Kc8-d7, Kc8-d8     Kc8-b7   
  3) Nc8-e7             Nc8-e7  * 122 Seconds
  4) Qa7xc7             Qa7xc7  * 0 Seconds
  5) Kd5-e6             Kd5-e6  * 11 Seconds
  6) Qh5xg6             Bh4-f6   
  7) Bb7-c6             Nd1-b2   
  8) Nh4xg6             Nh4xg6  * 0 Seconds
  9) Rg6-g4             Rg6-g4  * 5 Seconds
 10) Bb7-d5             Na7-b5   
 11) Nh4-g6             Nh4-g6  * 119 Seconds
 12) Ne3-f1             d4xe5    
 13) Be3-c1             Be3-c1  * 8 Seconds
 14) Re3-c3             Nh4-f5   
 15) Bc8xa6             Bc8xa6  * 0 Seconds
 16) Bc8xa6             Bc8xa6  * 0 Seconds
 17) Kc8-d7, Kc8-d8     Kc8-b7   
 18) Nc8-e7             Nc8-e7  * 0 Seconds
 19) Qa7xc7             Qa7xc7  * 0 Seconds
 20) Bb7-c6             Nd1-b2   
 21) Nh4xg6             Nh4xg6  * 0 Seconds
 22) h6-h7              Qh3-c3   
 23) Rg6-g4             Rg6-g4  * 0 Seconds
 24) Nh4-g6             Nh4-g6  * 104 Seconds
 25) e5-e6              e5xf6    
 26) Ba7-b8             g4-g5    
 27) Re3xd3             Re3xd3  * 93 Seconds
 28) Be3-c1             Be3-c1  * 17 Seconds
 29) Kc8-c7             g6-g7    
 30) Kc8-d7, Kc8-d8     Kc8-b8   
 31) Nc8-e7             Nc8-e7  * 104 Seconds
 32) Qa7xc7             Qa7xc7  * 0 Seconds
 33) Kd5-e6             Kd5-e6  * 0 Seconds
 34) Qh5xg6             Bh4-e1   
 35) Bb7-c6             Nd1-b2   
 36) Nh4xg6             Nh4xg6  * 0 Seconds
 37) Rg6-g4             Rg6-g4  * 0 Seconds
 38) Bb7-d5             Na7-b5   
 39) Nh4-g6             Nh4-g6  * 97 Seconds
 40) Ne3-f1             d4xe5    
 41) Be3-c1             Be3-c1  * 0 Seconds
 42) Qa3-f3             Qa3-f3  * 9 Seconds
 43) Kc8-c7             g6-g7    
 44) Kc8-d7, Kc8-d8     Kc8-b8   
 45) Qa7xc7             Qa7xc7  * 0 Seconds
 46) Kd5-e6             Kd5-e6  * 0 Seconds
 47) Qh5xg6             Bh4-e1   
 48) Bb7-c6             Nd1-b2   
 49) Nh4xg6             Nh4xg6  * 0 Seconds
 50) Rg6-g4             Rg6-g4  * 0 Seconds
 51) Bb7-d5             Na7-b5   
 52) Nh4-g6             Nh4-g6  * 98 Seconds
 53) Ba7-b8             g4-g5    
 54) Be2-c4             Be2-c4  * 104 Seconds
 55) Be3-c1             Be3-c1  * 0 Seconds
 56) Re3-c3             Nh4-f5   
 57) Kc8-c7             g6-g7    
 58) Nc8-e7             Nc8-e7  * 0 Seconds
 59) Qa7xc7             Qa7xc7  * 9 Seconds
 60) Kd5-e6             Kd5-e6  * 0 Seconds
 61) Bb7-c6             Nd1-b2   
 62) Nh4xg6             Nh4xg6  * 0 Seconds
 63) h6-h7              Qh3-c3   
 64) Rf6-h6             Rf6-h6  * 17 Seconds
 65) Rg6-g4             Rg6-g4  * 0 Seconds
 66) Rb3-c3             Rb3-c3  * 2 Seconds
 67) Nh4-g6             Nh4-g6  * 113 Seconds
 68) Be2-c4             Rc5xe5   
 69) Be3-c1             Be3-c1  * 0 Seconds
40 of 69 matching moves
1/5/2026 9:49:21 PM, Total time: 3:28:08 AM Rated time: 1:44:12 = 6252 Seconds

--------------------------------------------------------------------------------

Analysis of G:\vhard.epd   
Analyzing engine: Sting-bh
1/5/2026 6:21:13 PM Level: 1800 Seconds
   Source                    : List of EPD/PGN files
      G:\vhard.epd
   Colours                   : White moves=True, Black moves=True
   Direction                 : forward
   Games (PGN)               : From 1 to 999999
   Moves                     : All moves
   Positions (EPD)           : From 1 to 999999 
   Use Engine(s)             : List
      Sting
      Sting-bh
   Use seconds per move      : 180 Seconds per move
   Analysis Lines            : Minimum search depth=2
   Activate abort analysis   : False
  1) Kc8-c7             g6-g7    
  2) Kc8-d7, Kc8-d8     Kc8-d8  * 9 Seconds
  3) Nc8-e7             Nc8-e7  * 0 Seconds
  4) Qa7xc7             Qa7xc7  * 0 Seconds
  5) Kd5-e6             Kd5-e6  * 4 Seconds
  6) Qh5xg6             Qh5xg6  * 22 Seconds
  7) Bb7-c6             Nd1-b2   
  8) Nh4xg6             Nh4xg6  * 0 Seconds
  9) Rg6-g4             Rg6-g4  * 4 Seconds
 10) Bb7-d5             Na7-b5   
 11) Nh4-g6             Nh4-f5   
 12) Ne3-f1             d4xe5    
 13) Be3-c1             Be3-c1  * 3 Seconds
 14) Re3-c3             Re3-c3  * 36 Seconds
 15) Bc8xa6             Bc8xa6  * 0 Seconds
 16) Bc8xa6             Bc8xa6  * 24 Seconds
 17) Kc8-d7, Kc8-d8     Kc8-d7  * 141 Seconds
 18) Nc8-e7             Nc8-e7  * 25 Seconds
 19) Qa7xc7             Qa7xc7  * 0 Seconds
 20) Bb7-c6             Nd1-b2   
 21) Nh4xg6             Nh4xg6  * 0 Seconds
 22) h6-h7              h6-h7   * 145 Seconds
 23) Rg6-g4             Rg6-g4  * 0 Seconds
 24) Nh4-g6             Nh4-f5   
 25) e5-e6              e5xf6    
 26) Ba7-b8             Ba7-b8  * 39 Seconds
 27) Re3xd3             Re3xd3  * 134 Seconds
 28) Be3-c1             Be3-c1  * 2 Seconds
 29) Kc8-c7             Qd2-d8   
 30) Kc8-d7, Kc8-d8     Kc8-d7  * 74 Seconds
 31) Nc8-e7             Nc8-e7  * 0 Seconds
 32) Qa7xc7             Qa7xc7  * 0 Seconds
 33) Kd5-e6             Kd5-e6  * 0 Seconds
 34) Qh5xg6             Bh4-f6   
 35) Bb7-c6             Nd1-b2   
 36) Nh4xg6             Nh4xg6  * 0 Seconds
 37) Rg6-g4             Rg6-g4  * 0 Seconds
 38) Bb7-d5             Na7-b5   
 39) Nh4-g6             Nh4-f5   
 40) Ne3-f1             d4xe5    
 41) Be3-c1             Be3-c1  * 0 Seconds
 42) Qa3-f3             Qa3-f3  * 40 Seconds
 43) Kc8-c7             g6-g7    
 44) Kc8-d7, Kc8-d8     Kc8-d7  * 52 Seconds
 45) Qa7xc7             Qa7xc7  * 0 Seconds
 46) Kd5-e6             Kd5-e6  * 0 Seconds
 47) Qh5xg6             Qh5xg6  * 11 Seconds
 48) Bb7-c6             Nd1-b2   
 49) Nh4xg6             Nh4xg6  * 0 Seconds
 50) Rg6-g4             Rg6-g4  * 0 Seconds
 51) Bb7-d5             Na7-b5   
 52) Nh4-g6             Nh4-f5   
 53) Ba7-b8             Ba7-b8  * 0 Seconds
 54) Be2-c4             Kf2-e3   
 55) Be3-c1             Be3-c1  * 0 Seconds
 56) Re3-c3             Re3-c3  * 0 Seconds
 57) Kc8-c7             g6-g7    
 58) Nc8-e7             Nc8-e7  * 61 Seconds
 59) Qa7xc7             Qa7xc7  * 0 Seconds
 60) Kd5-e6             Kd5-e6  * 0 Seconds
 61) Bb7-c6             Nd1-b2   
 62) Nh4xg6             Nh4xg6  * 0 Seconds
 63) h6-h7              Qh3-c3   
 64) Rf6-h6             Rf6-f7   
 65) Rg6-g4             Rg6-g4  * 0 Seconds
 66) Rb3-c3             Rb3-c3  * 44 Seconds
 67) Nh4-g6             Nh4-f5   
 68) Be2-c4             Be2-c4  * 148 Seconds
 69) Be3-c1             Be3-c1  * 0 Seconds
45 of 69 matching moves
1/6/2026 1:17:31 AM, Total time: 6:56:18 AM Rated time: 1:28:58 = 5338 Seconds
The file stdat.7z holds the analysis results.
The src folder has the modified code and makefile along with the binary.
It may be interesting to see how it compares with just the BlackHole box checked and the OldSting box not checked.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
peter
Posts: 3526
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by peter »

Hi Dann!
There are 45 duplicates in vhard.epd, here's what epdutil.exe does with it, moving the duplicates to epdutil.log:

Code: Select all

2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - id "duplicate";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - id "duplicate";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - id "duplicate";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - id "duplicate";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - id "duplicate";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - id "duplicate";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - id "duplicate";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - id "duplicate";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - id "duplicate";
2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - id "duplicate";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - id "duplicate";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - id "duplicate";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - id "duplicate";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - id "duplicate";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - id "duplicate";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - id "duplicate";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - id "duplicate";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - id "duplicate";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - id "duplicate";
6rk/2p1np2/1r1pb3/1p2p3/p1PPP1p1/1P2N1Pq/P1BQRP2/3R2K1 w - - id "duplicate";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - id "duplicate";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - id "duplicate";
2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - id "duplicate";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - id "duplicate";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - id "duplicate";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - id "duplicate";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - id "duplicate";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - id "duplicate";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - id "duplicate";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - id "duplicate";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - id "duplicate";
7k/Bp1R1b1p/1p2pK1n/8/4P1P1/6pB/4p3/8 w - - id "duplicate";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - id "duplicate";
1rk1br1q/1pbp1n2/5p2/4P3/PpQR1PPN/4R1p1/1PK4n/6B1 w - - id "duplicate";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - id "duplicate";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - id "duplicate";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - id "duplicate";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - id "duplicate";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - id "duplicate";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - id "duplicate";
4n1k1/8/1b1P3P/1qB3Np/6PK/7Q/p2p1p2/8 w - - id "duplicate";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - id "duplicate";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - id "duplicate";
6r1/ppk2p1p/2b5/2R1pPP1/4P2P/8/4BK2/8 w - - id "duplicate";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - id "duplicate";
These are the 24 single ones in novhard.epd, that epdutil.exe keeps:

Code: Select all

2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id "ACT3.048";
2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - bm Kd7 Kd8; id "ACT4.094";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - bm Ne7; id "ACT5.100";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - bm Ke6; id "ACT2.074";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - bm Qxg6+; id "ACT4.066";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - bm Bd5; id "ACT5.095";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
6rk/2p1np2/1r1pb3/1p2p3/p1PPP1p1/1P2N1Pq/P1BQRP2/3R2K1 w - - bm Nf1; id "TTT2.085";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
1rk1br1q/1pbp1n2/5p2/4P3/PpQR1PPN/4R1p1/1PK4n/6B1 w - - bm Rc3; id "ACT2.045";
2B4n/p4p1P/n1p2Pp1/p1P3P1/P1q1NN2/4K3/P2P4/5k2 w - - bm Bxa6; id "ACT4.087";
2B4n/p4p1P/r1p2Pp1/p1P3P1/P1q1NN2/4K3/3P4/5k2 w - - bm Bxa6; id "ACT5.010";
4n1k1/8/1b1P3P/1qB3Np/6PK/7Q/p2p1p2/8 w - - bm h7+; id "ACT5.049";
6k1/8/ppp2pP1/4P1p1/7p/PPP1K3/8/8 w - - bm e6; id "ACT3.093";
7k/Bp1R1b1p/1p2pK1n/8/4P1P1/6pB/4p3/8 w - - bm Bb8; id "ACT3.008";
6nN/2p5/P1k2b2/2Bpp3/8/3bR3/8/K7 w - - bm Rxd3; id "ACT4.084";
1b2b1BN/2r1pp1n/2qpR2p/r1Np1Rnk/p5p1/Q5Kp/8/4B3 w - - bm Qf3; id "ACT3.089";
6r1/ppk2p1p/2b5/2R1pPP1/4P2P/8/4BK2/8 w - - bm Bc4; id "TTT1.092";
5Kbk/4B1pp/3N1R2/p3P2n/3P2p1/2P2p2/1P1pP1r1/1B4n1 w - - bm Rh6; id "ACT2.047";
5N2/rp6/pB1p1N2/pb2k1K1/3p2p1/1R2pp2/4B1n1/6bn w - - bm Rc3; id "ACT3.011";
Some are given even 3 or 4 times in vhard.epd, e.g. already the first one is in line 1, line 29, line 43 and line 57, have a look at pasted vhard.epd, as I downloaded it, after first one position as well as after its duplicates I created spaces here:

Code: Select all

2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id "ACT3.048";

2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - bm Kd7 Kd8; id "ACT4.094";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - bm Ne7; id "ACT5.100";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - bm Ke6; id "ACT2.074";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - bm Qxg6+; id "ACT4.066";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - bm Bd5; id "ACT5.095";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
6rk/2p1np2/1r1pb3/1p2p3/p1PPP1p1/1P2N1Pq/P1BQRP2/3R2K1 w - - bm Nf1; id "TTT2.085";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
1rk1br1q/1pbp1n2/5p2/4P3/PpQR1PPN/4R1p1/1PK4n/6B1 w - - bm Rc3; id "ACT2.045";
2B4n/p4p1P/n1p2Pp1/p1P3P1/P1q1NN2/4K3/P2P4/5k2 w - - bm Bxa6; id "ACT4.087";
2B4n/p4p1P/r1p2Pp1/p1P3P1/P1q1NN2/4K3/3P4/5k2 w - - bm Bxa6; id "ACT5.010";
2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - bm Kd7 Kd8; id "ACT4.094";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - bm Ne7; id "ACT5.100";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
4n1k1/8/1b1P3P/1qB3Np/6PK/7Q/p2p1p2/8 w - - bm h7+; id "ACT5.049";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
6k1/8/ppp2pP1/4P1p1/7p/PPP1K3/8/8 w - - bm e6; id "ACT3.093";
7k/Bp1R1b1p/1p2pK1n/8/4P1P1/6pB/4p3/8 w - - bm Bb8; id "ACT3.008";
6nN/2p5/P1k2b2/2Bpp3/8/3bR3/8/K7 w - - bm Rxd3; id "ACT4.084";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id "ACT3.048";

2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - bm Kd7 Kd8; id "ACT4.094";
2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - bm Ne7; id "ACT5.100";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - bm Ke6; id "ACT2.074";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - bm Qxg6+; id "ACT4.066";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - bm Bd5; id "ACT5.095";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
6rk/2p1np2/1r1pb3/1p2p3/p1PPP1p1/1P2N1Pq/P1BQRP2/3R2K1 w - - bm Nf1; id "TTT2.085";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
1b2b1BN/2r1pp1n/2qpR2p/r1Np1Rnk/p5p1/Q5Kp/8/4B3 w - - bm Qf3; id "ACT3.089";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id "ACT3.048";

2K5/p1p4k/p1p5/2p3p1/6Pb/6p1/p5P1/B7 w - - bm Kd7 Kd8; id "ACT4.094";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - bm Ke6; id "ACT2.074";
2q1k1b1/7p/4p1r1/Pp1pP2Q/1P1P2PB/2P5/1KP5/8 w - - bm Qxg6+; id "ACT4.066";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
5r2/NBk4P/8/8/1P2K3/pp6/1p6/1N6 w - - bm Bd5; id "ACT5.095";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
7k/Bp1R1b1p/1p2pK1n/8/4P1P1/6pB/4p3/8 w - - bm Bb8; id "ACT3.008";
6r1/ppk2p1p/2b5/2R1pPP1/4P2P/8/4BK2/8 w - - bm Bc4; id "TTT1.092";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
1rk1br1q/1pbp1n2/5p2/4P3/PpQR1PPN/4R1p1/1PK4n/6B1 w - - bm Rc3; id "ACT2.045";
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id "ACT3.048";

2N3q1/pp1N4/5K2/7k/7P/3Q4/1P6/r7 w - - bm Ne7; id "ACT5.100";
2k5/Q1np1N2/2p2Bpp/1p1P4/pPP1p2P/P5q1/8/1K6 w - - bm Qxc7+; id "ACT4.035";
2nB1k2/6pp/3p2p1/n2K4/3Qp1N1/1Np4R/rb1p4/1Rq3r1 w - - bm Ke6; id "ACT2.074";
4K3/1B1p4/pn1Npb2/k3p3/P1pn4/P7/7B/3N4 w - - bm Bc6; id "ACT3.029";
4k2q/p4p1B/1pp3pP/2N2pP1/2P2P1N/8/PP2P3/2K5 w - - bm Nxg6; id "ACT5.011";
4n1k1/8/1b1P3P/1qB3Np/6PK/7Q/p2p1p2/8 w - - bm h7+; id "ACT5.049";
5Kbk/4B1pp/3N1R2/p3P2n/3P2p1/2P2p2/1P1pP1r1/1B4n1 w - - bm Rh6; id "ACT2.047";
5N1r/5n1n/ppp3R1/5K2/7k/6p1/6PN/8 w - - bm Rg4+; id "ACT3.049";
5N2/rp6/pB1p1N2/pb2k1K1/3p2p1/1R2pp2/4B1n1/6bn w - - bm Rc3; id "ACT3.011";
6bk/5pp1/3N1p1b/8/6pN/5pPp/4pP1P/4K3 w - - bm Ng6+; id "ACT3.051";
6r1/ppk2p1p/2b5/2R1pPP1/4P2P/8/4BK2/8 w - - bm Bc4; id "TTT1.092";
1k5b/p2K3P/P6p/p4pp1/1p6/4B3/2P5/8 w - - bm Bc1; id "ACT3.084";
Peter.
Dann Corbit
Posts: 12816
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by Dann Corbit »

Yikes!
I noticed that watching it analyze, but I thought I was just seeing the same moves because I played that set over and over.
:shock:
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
peter
Posts: 3526
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by peter »

Anyhow, with the remaining 24 I get these results with 30 threads of a 16x4.3GHz CPU and 1 minute per position, first in Shredder- GUI because of the beautiful tables of solutions and time per solution position- wise:

Code: Select all

Lc0v0.32.1 6147500PT 
Bisher gelöst: 16 von 24  ;  5:32m

         1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
 -------------------------------------------------------------------------------------
   0 |   0   0   7   -   -   -   -   0   -  15   0   2   4   1   0  41  11   4   0   0
  20 |   -   0   -   -

  TotTime: 6:40m    SolTime: 5:32m

Code: Select all

Sting 11 Black Hole Dann Corbit's compile
Bisher gelöst: 14 von 24  ;  13:02m

         1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
 -------------------------------------------------------------------------------------
   0 |   -   2   0   3  22   -   -   0   1   -   -   -   4  25   0   0   -   -  17   -
  20 |  46  30   -  27
 
  TotTime: 15:58m    SolTime: 13:02m 

Code: Select all

Sting 11 Old Sting Dann Corbit's compile
Bisher gelöst: 12 von 24  ;  14:24m

         1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20
 -------------------------------------------------------------------------------------
   0 |   -   -   0   1  24   -   -   0   3   -   -   -   5  34   0   0   -   -   -   -
  20 |  30   -  43   1

  TotTime: 17:49m    SolTime: 14:24m
And with MEA

Code: Select all

                                                  Max   Total   Time   Hash          
    Engine             Score   Found  Pos   ELO  Score   Rate    ms     Mb  Cpu      
 1  Sting11DC            145     13    24  2416    270  53.7%  60000  8192   30
 2  Lc0v0.32.1-6147500PT 145     16    24  2416    270  53.7%  60000  8192    2
 3  TheHuntsman1         130     12    24  2164    270  48.1%  60000  8192   30
As for MEA I gave 5 to 20 points per solution, according to 4 levels of difficulty from not so hard to medium to hard to very hard.
This way you see the distinction between e.g. Lc0 and Sting getting zero as for the score. On the other hand with all positions of same amount of rewarding points, distinction would seem much bigger actually and Lc0 would change places with Sting. In Shredder GUI you could make distincion better by counting exact time to solution, EloStatTS does that even per position, pity Sting doesn*t take all possible threads, 8 ist the maximum to be set.
So I just had two more runs to compare Huntsman to Rems, but there you see the other one even bigger problem, even with very near to each other number of solutions you have a very big error bar:

Code: Select all

    Program                                    Elo   +/-  Matches  Score   Av.Op.   S.Pos.   MST1    MST2   RIndex

  1 TheHuntsman1-30t                         : 3512  107     20    53.4 %   3488    13/ 24    6.5s   31.0s   0.73
  2 RemsM-091224-30t-32-6-2000               : 3488  103     20    46.6 %   3512    13/ 24   13.4s   34.8s   0.58

MST1  : Mean solution time (solved positions only)
MST2  : Mean solution time (solved and unsolved positions)
RIndex: Score according to solution time ranking for each position
So the relation between distinction and error bar is even bigger counting the exact time indices not only per engine but per engine and per position. That*s the main statistical problem with positional testing anyhow, there have to be compared positions solved by one engine to positions solved by the other one, not just sums of solvings and sums of time.
And that*s the reason why you need a minimum of positions solved by both (or some of all, if more of them) engines compared to each other, if these sample size gets too small, statistical relevance of the results is insignificant. With this kind of positions (look at the Shredder- tables, the positions solved by Lc0 are quite different ones than the ones solved by Sting and again other ones than those solved by Huntsman, so even if it was the 69 without duplicates but all of this kind of anti engine puzzles only yet still, meaning so much hardware- time necessary to get numbers of solved positions at all even from the best in this one suite performing engines, statistical significance and comparability to other one suites, not to talk about transferability to game playing results, the pure statistical relevance of its own also would be very low still, regards
Peter.
peter
Posts: 3526
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by peter »

Found out latest Sting only seems to use 8 threads maximum in Fritz (necessary for EloStatTS), in Task Manager yet 30 are running, so I added it together with Lc0, still 30 threads for all A-B-engines, Lc0 with RTX5070 GPU, 60"/pos:

Code: Select all

    Program                                    Elo   +/-  Matches  Score   Av.Op.   S.Pos.   MST1    MST2   RIndex

  1 Lc0v0.32.1-6147500PT                     : 3540   58     61    57.7 %   3486    16/ 24    5.4s   23.6s   0.57
  2 TheHuntsman1-30t                         : 3505   60     58    51.0 %   3498    13/ 24    6.5s   31.0s   0.59
  3 Sting11x-BlackHole-30t                   : 3479   61     56    45.7 %   3509    12/ 24   10.4s   35.2s   0.44
  4 RemsM-091224-30t-32-6-2000               : 3472   57     55    44.8 %   3508    13/ 24   13.4s   34.8s   0.44

MST1  : Mean solution time (solved positions only)
MST2  : Mean solution time (solved and unsolved positions)
RIndex: Score according to solution time ranking for each position
And the Fritz- results position- wise, ranking only out of numbers of solutions and total time (in EloStatTS out of position- wise calculated time indices as well):

Code: Select all

Position        Lc0v0.32.1	TheHuntsman1	Rems M-091224   Sting 11x
ACT3.048	00:00:00	00:01:00	00:00:18	00:01:00
ACT4.094	00:00:00	00:00:00	00:01:00	00:01:00
ACT5.100	00:00:05	00:01:00	00:00:51	00:00:00
ACT4.035	00:01:00	00:00:35	00:01:00	00:00:01
ACT2.074	00:01:00	00:00:00	00:01:00	00:00:04
ACT4.066	00:01:00	00:01:00	00:01:00	00:01:00
ACT3.029	00:01:00	00:01:00	00:01:00	00:01:00
ACT5.011	00:00:00	00:01:00a	00:00:56	00:00:00
ACT3.049	00:01:00	00:01:00	00:01:00	00:00:05
ACT5.095	00:00:14	00:00:01	00:01:00	00:01:00
ACT3.051	00:00:00	00:00:23	00:01:00	00:01:00
TTT2.085	00:00:01	00:01:00	00:00:01	00:01:00
ACT3.084	00:00:03	00:00:00	00:00:00	00:00:01
ACT2.045	00:00:02	00:00:00	00:01:00	00:01:00
ACT4.087	00:00:00	00:00:00	00:00:09	00:00:00
ACT5.010	00:00:34	00:00:06	00:00:13	00:00:00
ACT5.049	00:00:13	00:01:00	00:01:00	00:01:00
ACT3.093	00:00:03	00:01:00	00:00:00	00:01:00
ACT3.008	00:00:00	00:01:00	00:00:03	00:00:18
ACT4.084	00:00:00	00:00:00	00:00:00	00:01:00
ACT3.089	00:01:00	00:00:00	00:00:06	00:00:52
TTT1.092	00:00:00	00:01:00	00:00:00	00:01:00
ACT2.047	00:01:00	00:00:09	00:00:10	00:00:20
ACT3.011	00:01:00	00:00:00	00:01:00	00:00:15

Total time : 	00:09:21	00:12:18	00:13:52	00:14:01
Total solved: 	16	        13	        13	        12
Peter.
Dann Corbit
Posts: 12816
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by Dann Corbit »

Peter,
what kind of GPU do you have?
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
peter
Posts: 3526
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by peter »

Dann Corbit wrote: Wed Jan 07, 2026 12:10 pm Peter,
what kind of GPU do you have?
peter wrote: Wed Jan 07, 2026 1:08 am ... Lc0 with RTX5070 GPU...
Peter.
Paloma
Posts: 1221
Joined: Thu Dec 25, 2008 9:07 pm
Full name: Herbert L

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by Paloma »

What caught my attention:
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id “ACT3.048”;

I searched in yacpdb and found the same position but with wBishop on g1 ?!?
Mueller, Dieter 1989/1990 mate in 3 or 4
Too easy for a test position.

But without the bishop (ACT3.048), Kc7 is only worth +3.?? to 5.??, with no progress in the further course of the game.
Chacal X
Posts: 131
Joined: Mon Mar 10, 2025 9:06 pm
Full name: flavio thomazini

Re: sting-sf-bh-11x ? Be afraid, be very much afraid !

Post by Chacal X »

It's a very powerful program, I just regret that I can't play against other programs on PlayChess; it doesn't load in the PlayChess machine room...
User avatar
Ajedrecista
Posts: 2181
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: sting-sf-bh-11x? Be afraid, be very much afraid!

Post by Ajedrecista »

Hello:
Paloma wrote: Sat Jan 10, 2026 1:15 pm What caught my attention:
2K3b1/8/pPP1p1P1/2p1k2N/2P1pN1P/6PP/n1pQ2p1/1q3r2 w - - bm Kc7; id “ACT3.048”;

I searched in yacpdb and found the same position but with wBishop on g1 ?!?
Mueller, Dieter 1989/1990 mate in 3 or 4
Too easy for a test position.

But without the bishop (ACT3.048), Kc7 is only worth +3.?? to 5.??, with no progress in the further course of the game.
I found the version without the white bishop on g1 at PDB P1050474, claimed to be a checkmate in three moves, but was busted by Chest 5.2 (no checkmate in nine moves or less, indeed). It can not be found anymore at the Meson Chess Problem Database pointed by PDB, where it was supposedly once, so it looks like it was a typo or a misprint of the true position that was corrected at Meson, but not at PDB. This bishop attacking d4 looks crucial in some lines.

OTOH, the version with the white bishop on g1 and solution Kd7 can be found at YACPDB #227032, PDB P1367401 (claimed solution Kc7, which results in a longer checkmate according to Chest 5.2: 1.- Kc7, Qxb6+; 2.- Kxb6, Rb1+; 3.- Kc7, Rb7+; 4.- cxb7, e3; 5.- Qxe3+, Kf5; 6.- g4#) and Meson #32780 (with Kd7 correct solution).

FWIW, ACT means 'Alternative Chess Test' and ACT3 is the third version of ACT. All ACT versions can be downloaded from here.

Other insights are welcome, of course.

Regards from Spain.

Ajedrecista.