The ENGINES CRACKERS 2 (Update)

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

Moderators: hgm, Rebel, chrisw

Jouni
Posts: 3281
Joined: Wed Mar 08, 2006 8:15 pm

Re: The ENGINES CRACKERS 2 (Update)

Post by Jouni »

Crystal solves in 30s 1 3 10 11 12 13 15. Note, that position 3 = position 18!
Jouni
Glarean
Posts: 262
Joined: Sun Oct 05, 2008 1:04 pm
Location: Switzerland
Full name: Walter Eigenmann

Re: The ENGINES CRACKERS 2 (Update)

Post by Glarean »

Spliffjiffer wrote: Mon Sep 20, 2021 8:41 pmthis is probably a draw and i cant see what Ba3 makes better than Nxa4 honestly... i think your specific problem might indicate that a sample of a problem might tell sth to us ...be not sure, investigate, be unclear, look deeper... and find nothing ;-)...but dig deeper...i follow your samples since many, many...a lot of years... years and im still impressed...(30 yeras lets say).maybe..there is a tiny thing left...your positions.. they are good, but not completely well analysed...ill wanna say this to u ...good work, fantastic findouts, poor investigations....your not the guy who find outs sth...you are a man who can make sth stand but not being able to investigate a lot !!?...the positions that you were ale to allocate, i will not put in question, these are very good positiosns...but the researches...i put heavily in question....theses are poore but u say they are heavaly well done and time requestingly deducateted...i see this as an abuse of guys that take hours to make sth real in terms of computer-analysis...i honor u very much Walter, but your abilities to analyse a position tends to nothing, in my considerations...this position tells everything...can u make a workaround?...i heavily doubht...i give all honor since say 30 yerars but your abilties to determine a pos were not getting better
Sorry, but I can't do anything with your stammering....
WHICH analyses exactly are wrong?
Position 2: First try an overnight analysis, then we'll talk further...
Glarean
Posts: 262
Joined: Sun Oct 05, 2008 1:04 pm
Location: Switzerland
Full name: Walter Eigenmann

Re: The ENGINES CRACKERS 2 (Update)

Post by Glarean »

Jouni wrote: Mon Sep 20, 2021 9:07 pmNote, that position 3 = position 18!
Oh no! You're right!
I'll have to replace number 18 soon....

Many thanks: Walter

.
Spliffjiffer
Posts: 416
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: The ENGINES CRACKERS 2 (Update)

Post by Spliffjiffer »

stammering,...yes, i know...u wanna say that im wimmering around...but my story is this: in pos 2 Ba3 is not better than Nxa4...no 24h...your turn..give sth...
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Glarean
Posts: 262
Joined: Sun Oct 05, 2008 1:04 pm
Location: Switzerland
Full name: Walter Eigenmann

Re: The ENGINES CRACKERS 2 (Update)

Post by Glarean »

.
Spliffjiffer wrote: Tue Sep 21, 2021 2:48 amstammering,...yes, i know...u wanna say that im wimmering around...
No, I want to say that you stammered. And I recommend: think first, then formulate clearly.
Spliffjiffer wrote: Tue Sep 21, 2021 2:48 ambut my story is this: in pos 2 Ba3 is not better than Nxa4...no 24h...your turn..give sth...
My story is that you should first do a 10-hour Stockfish analysis here before you join the discussion on this position.
Until then, I assume that Carlsen's move is marginally superior.
Spliffjiffer
Posts: 416
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: The ENGINES CRACKERS 2 (Update)

Post by Spliffjiffer »

thx for your story, help and that stammered stuff...youre a great man...and i idiot thought it were ardy difficult after say 30 sec :-)
btw: pkenominal winningline that you are willing to provide...sry stumbering articulated
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Vinvin
Posts: 5228
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: The ENGINES CRACKERS 2 (Update)

Post by Vinvin »

3 runs with 4 engines for the 19 positions :
- Crystal 3.2 x64 avx2
- ShashChess17.1-x86-avx2
- SugaR AI ICCF 2.40 avx2
- Stockfish_21091708_x64_avx2

Conditions :
Interface : Arena 3.5.1
CPU = 5950X@4GHz
16 threads per engine
15 minutes (900 seconds) per position
All 6 man EGTB + one 7-man KRPPKRP
65535 MB HT

See the tables of all results at the end.

Average per engine : the leader is Crystal 3.2 with 16.7 on 19.

Code: Select all

                       Cryst3.2   Shah17.1   Sug-2.40    SF-210917
#found                   16,7      14,0        13,3         16,0                        
Avg time                 142       119         119          142  
Avg Time w/penalty       343       561         620          403 
Sort by number of times the solutions is found (on the 12 runs) :

Code: Select all

Solutions       #found
 15) Nf4-g6      12
 10) Nc5-e4      12
  3) .. c7-c6    12
  8) Ne3-f5      12
 12) Rg5-g4      12
  7) a5-a6       12
  9) .. d6-d5    12
 13) Rd8-d5      12
 14) Kc1-b1      11
  6) .. h7-h5    10
  5) Nf3-g5      10
 16) Kd6-e6       9
  1) .. Qf7xf3    9
 11) g4-g5        8
 17) b3-b4        8
 18) Bd4-e5       7
 19) Re1-c1       7
  2) Bb2-a3       5
  4) Ra3xc3       0
All results :

Code: Select all

Solution           CrystCrystCryst     ShashShashShash     SugaRSugaRSugaR     StockStockStock    #found   Avg time
-------------------------------------------------------------------------------------------------------------------
  1) .. Qf7xf3       101   16  540       372  266  217                            50   34   51       9       183,0
  2) Bb2-a3            2                      788                                138   60  160       5       229,6
  3) .. c7-c6          0   17    0         3    8    8        56   22  131        62   99   37      12        36,9
  4) Ra3xc3                                                                                          0       -----
  5) Nf3-g5          685  560  336       719                 122   44   45         7   10   20      10       254,8
  6) .. h7-h5         14   22    7         3  208   50       118                  21   50    3      10        49,6
  7) a5-a6            53   44   45        22    5    2        78   10  154       164  173  298      12        87,3
  8) Ne3-f5            7    3   43        57    4   21        43    3   24        11  270    4      12        40,8
  9) .. d6-d5         32   99   80        35   59   13        67  145  251        91  159   91      12        93,5
 10) Nc5-e4            0    0    0         0    0    0         0    0    0        35   12  386      12        36,1
 11) g4-g5             0                   0   12    3        46                 397   11  732       8       150,1
 12) Rg5-g4           16  141   89         1    0    0       161  290   42         1    2    1      12        62,0
 13) Rd8-d5            1    1    0         0    0    7         3    5    0       835  486  538      12       156,3
 14) Kc1-b1          109   43  324            244   47       242  102   77         9    0    0      11       108,8
 15) Nf4-g6            0    2    0         0    0    0         0    6    0       110   31  152      12        25,1
 16) Kd6-e6            6    0  120         0        50              0            188  117  100       9        64,6
 17) b3-b4             8  754  778                 780         0    0    0              0            8       290,0
 18) Bd4-e5          367   74   57            201                      829       278  273            7       297,0
 19) Re1-c1            4  641  717       776                 557  504  566                           7       537,9
-------------------------------------------------------------------------------------------------------------------
Avg time              78  151  196       142  128   86       107   87  163       150  105  172      12       130,3
Avg Time w/penalty   169  411  449       578  568  537       552  628  680       410  284  514               481,8
#found                18   16   16        14   14   14        14   13   13        16   17   15     180        15,0
#not found             1    3    3         5    5    5         5    6    6         3    2    4                 4,0
The positions in EPD format :

Code: Select all

1rb1r3/5qk1/3b2p1/1PnPp1Pp/4B3/pP3Q2/P2N1B1P/1KR3R1 b - - bm Qxf3 ;
r2q1rk1/1p2p1bp/2np2p1/2n2b2/p1P5/1PN1QN1P/PB2BPP1/3R1RK1 w - - bm Ba3 ;
8/1pp1p3/6p1/3P4/pP3P2/8/k1K3P1/8 b - - bm c6 ;
7k/pp1pBp1N/1p3Pp1/b3K1PR/P3P2p/R1n4P/2P1N1P1/3q4 w - -  bm  Rxc3;
r1bqr1k1/pp1nb1p1/2n3P1/3pPp1p/1p1P3P/P4N2/3N1P2/RBBQK2R w KQ - bm Ng5 ;
4r3/4b2p/1pk2pb1/p1p5/P2pNB2/1P3P2/2P2KPP/3R4 b - - bm h5;
7k/1p5p/2qpn1PP/Pp1p4/8/3B2p1/3B1pP1/R4K2 w - - bm a6 ;
3kB3/5K2/7p/3p4/3pn3/4NN2/8/1b4B1 w - - bm  Nf5 ;
r2qr1k1/1ppb1pp1/p2p1bnp/P7/2NPP3/1B3P2/1P4PP/R1BQ1RK1 b - - bm d5;
1q6/4p3/6Q1/2N1n3/p2P2P1/P2PP1k1/4Kp1p/2B5 w - - bm Ne4+ ;
3r1r2/pb2nkp1/1pnb1p1p/2pNpP2/q1P1P1PP/3P1N2/P3BR2/3RQ1BK w - - bm g5 ;
8/qp1N1bn1/3p4/Bn1p2R1/3pp2k/6R1/1pp1KP2/8 w - - bm R5g4+ ;
3R4/4K1b1/8/N1n1k3/n1R1pp2/2P3P1/2N1p1P1/q7 w - - bm Rd5+ ;
2k1rb2/2q2pBr/p2p1PpP/1pnP2P1/5Q2/P7/1PP5/2KR1B2 w - - bm Kb1 ;
4r1nk/2p1q2p/8/3Q2PP/2P1pN2/8/5R2/5K2 w - - bm Ng6+ ;
8/bp4Pk/3K1B2/R6p/2pNp1P1/p7/3p1Pn1/2r4q w - -  bm  Ke6 ;
4r3/2kbr1p1/2p2p2/p2p1P2/P2B2Pp/KP2P2P/8/1RR5 w - - bm b4 ;
6k1/2p2p1p/4pP2/8/1pPB1P1P/7p/1KP1R3/7r w - - bm Be5;
r1n5/k3pp2/2Rr4/1p6/1p6/p4pp1/QP4Pn/4RBK1 w - - bm Rec1 ;
Glarean
Posts: 262
Joined: Sun Oct 05, 2008 1:04 pm
Location: Switzerland
Full name: Walter Eigenmann

Re: The ENGINES CRACKERS 2 (Update)

Post by Glarean »

Vinvin wrote: Wed Sep 22, 2021 2:49 am 3 runs with 4 engines for the 19 positions :
- Crystal 3.2 x64 avx2
- ShashChess17.1-x86-avx2
- SugaR AI ICCF 2.40 avx2
- Stockfish_21091708_x64_avx2

Conditions :
Interface : Arena 3.5.1
CPU = 5950X@4GHz
16 threads per engine
15 minutes (900 seconds) per position
All 6 man EGTB + one 7-man KRPPKRP
65535 MB HT
Thanks for the tests and the result reports!
Addition: No. 18 has since been replaced by a study by Illescas.
Vinvin
Posts: 5228
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: The ENGINES CRACKERS 2 (Update)

Post by Vinvin »

There's a problem with position#3 : 8/1pp1p3/6p1/3P4/pP3P2/8/k1K3P1/8 b - - bm c6 ;
Engines can choose randomly between 1…c6! and 1…Ka3? because it's difficult to see Ka3 is losing.
A lot of engines gives 0.00 to both moves.
peter
Posts: 3185
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: The ENGINES CRACKERS 2 (Update)

Post by peter »

Vinvin wrote: Wed Sep 22, 2021 12:02 pm There's a problem with position#3 : 8/1pp1p3/6p1/3P4/pP3P2/8/k1K3P1/8 b - - bm c6 ;
Engines can choose randomly between 1…c6! and 1…Ka3? because it's difficult to see Ka3 is losing.
A lot of engines gives 0.00 to both moves.
You're right as for using the position in automatically evaluated single best move suites.
To be sure, if it's not solved "for wrong reasons" (in this case accidentally), evaluating 1.Ka3? wrongly 0.00 too, you'd have to let run engine in MultiPV=2 mode or give in 1...Ka3 and look at output. As long as 2.g4! isn't found, 0.00 get's stuck rather long as for my trials with most fishes. Best for long while is often

1) 1...c6... 0.00
2) 1...Ka3 2.g4...0.03, which I wouldn't call "solved" neither

Crystal 280821 at 32 threads:


[fen]8/1pp1p3/6p1/3P4/pP3P2/8/k1K3P1/8 b - -[/fen]

Code: Select all

NNUE evaluation using nn-33495fe25081.nnue enabled
-----
 26	 0:01 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (68.955.342) 44660
 26	 0:01 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (68.955.342) 44660
-----
 27	 0:02 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (97.804.362) 43643
 27	 0:02 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (97.804.362) 43643
-----
 28	 0:03 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (134.291.597) 42137
 28	 0:03 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (134.291.597) 42137
-----
 29	 0:04 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (185.063.012) 40283
 28	 0:04 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (185.063.012) 40283
-----
 29	 0:05 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (207.409.634) 40604
 29	 0:05 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (207.409.634) 40604
-----
 30	 0:07 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (303.699.992) 40342
 29	 0:07 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (303.699.992) 40342
-----
 30	 0:07 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (304.926.142) 40344
 30	 0:07 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (304.926.142) 40344
-----
 31	 0:09 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (400.808.853) 41058
 30	 0:09 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (400.808.853) 41058
-----
 31	 0:10 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (450.397.210) 41038
 31	 0:10 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (450.397.210) 41038
-----
 32	 0:14 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (569.667.224) 40496
 31	 0:14 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (569.667.224) 40496
-----
 32	 0:15 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (628.935.480) 40537
 32	 0:15 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (628.935.480) 40537
-----
 33	 0:28 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (1.092.943.496) 37992
 32	 0:28 	 0.00 	1...Ka3 2.Kc3 c6 3.dxc6 bxc6 4.g3 Ka2 5.Kc2 Ka3 (1.092.943.496) 37992
-----
 33	 4:21 	 0.00 	1...c6 2.dxc6 bxc6 3.g3 Ka3 4.Kc3 Ka2 (10.430.105.514) 39856
 33	 4:21 	+13.67	1...Ka3 2.g4 (10.430.105.514) 39856
Bester Zug: c7-c6 Zeit: 4:21.688 min  K/s: 39.856.569  Knoten: 10.430.105.514 
The fail low (from White's side) at 1...Ka3? was solved by stopping analysis- mode, probably it would have worked that way much sooner too regards
Peter.