Lichess browser stockfish solved this position instantly!

[Patishi]

HI everyone,
I thought that browser based analysis should be weaker than standard local GUI. But I was very surprised to see that lichess new stockfish 19 (94MB NNUE) managed to solve this position instantly running in the web browser. I have a 9800x3d cpu, and when I tried to solve this in the local GUI with stockfish 18 avx-512 version it took 1:30 - 2:00 minutes in the best case scenario (on some runs it took even longer).

On the lichess, on only 15/16 threads it simply solved it instantly. And I did multiple runs just to make sure it is consistent.

FEN
8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1
(hope you guys can see the image.. I don't know if I upload it correctly).

[Dicaste]

It's probably CLOUD analysis. It's been analyzed before so it's just memory.

[Patishi]

It's probably CLOUD analysis. It's been analyzed before so it's just memory.

I don't think so. Usually when it uses cloud analysis it says so, and I can definitely notice my CPU working when turning the engine on.
But I don't know, maybe it does retrieve the evaluation instantly from a cached database or something and than my local machine starts to think.

Disappointing

[AndrewGrant]

Tried to find mate for fun with Torch. No such luck.

info depth 66 seldepth 98 multipv 1 score cp 838 nodes 75934007313 nps 42886612 hashfull 998 tbhits 0 time 1770576 pv g4f6 h7g7 f6h5 g7g6 d1c2 g6h5 d7d8q h5g4 d6c6 g4h3 d8d1 b4a3 d1f1 h3g3 f1g1 g3f3 c2d1 f3f4 g1g4 f4e5 g4g3 e5d4 g3d6 d4e4 d6d5 e4f4 d5c4 f4g3 c4g4 g3h2 g4e2 h2h1 e2f1 h1h2 f1a6 a3c1 a6e2 h2g3 c6c5 c1a3 c5c6 a3c1 e2e1 g3g2 e1c3 e3e2 d1e2 c1f4 e2d1 g2h2 c3c2 h2g1 c2c5 g1g2 c6b7 g5h3 c5d5 g2g1 d5f3 f4d6 f3g4 g1h2 g4e2 h2g3 e2e3 g3h2 e3d2 h2g3 d2d6 h3f4 d6a3 g3f2 a3c5 f2e1 d1f3 e1d2 c5b4 d2c2

[peter]

Two times single thread (4.3GHz CPU) output, one of Reckless and one of Stockfish:

Code: Select all

8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1

Analysis by Reckless 0.10.0-dev-91b56c29:

2.d8D Sf7+ 3.Kd7 Sxd8 4.Kxd8 La5+ 5.Kd7 Lc7 6.Sxe3 Lxg3 7.Kc6 Kg6 8.Kb5 Sc7+ 9.Kxc5 
  Schwarz steht besser.  
  -/+ (-1.25)  Tiefe: 10/24   00:00:00
...
2.Sxe3 La5 3.Kc6 Sb8+ 4.Kb5 Ld8 5.Lc2+ Kh8 6.Lf5 Sf7 7.Kxc5 Lg5 8.Sc2 Kg7 9.Lh3 Sd8 10.Kd4 Ld2 11.Ke5 Sbc6+ 12.Kd6 Kf6 13.Lf1 Se5 14.Sb4 Sef7+ 15.Kd5 Le1 16.Sa2 c2 17.Ld3 Lxg3 
  Schwarz steht auf Gewinn.  
  -+ (-4.03)  Tiefe: 31/58   00:00:21  47150kN, tb=389758
2.Sf6+ 
  Schwarz steht auf Gewinn.  
  -+ (-3.67 ++)  Tiefe: 32/58   00:00:25  57586kN, tb=473738
...
2.Sf6+ 
  Weiß steht deutlich besser.  
  +- (1.71 ++)  Tiefe: 32/58   00:00:25  58201kN, tb=479335
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.Df6 Kxg3 7.Df1 c4+ 8.Kd5 Sc7+ 9.Kxc4 Ld6 10.Dg1+ Kf3 11.Ld1+ e2 12.Df1+ Ke4 13.Dxe2+ Kf5 14.Dd3+ Ke6 15.Dg6+ Kd7 16.Dg7+ Le7 17.Lg4+ Ke8 18.Lh5+ Kd7 19.Dxc3 Ke6 
  Weiß hat entscheidenden Vorteil.  
  +- (3.12)  Tiefe: 32/70   00:00:28  64055kN, tb=565275
...
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.Kc6 c4 7.Dd1+ Kxg3 8.Dg1+ Kf4 9.Dh2+ Kg4 10.De2+ Kf4 11.Dxc4+ Kg3 12.Dxa6 Sf3 13.Kd5 Le7 14.Ld1 Sh4 15.Dxh6 Sg2 16.Dg7+ Kf2 17.Dxc3 Lg5 18.De5 Lf4 19.De4 Lg5 20.Df3+ Kg1 21.Le2 Lh6 22.Df1+ Kh2 
  Weiß steht klar auf Gewinn.  
  +- (5.04 ++)  Tiefe: 35/31   00:00:53  119MN, tb=1430745


8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1

Analysis by Stockfish 19:

2.Sf6+ Kg7 3.Sh5+ Kg6 4.Sf4+ Kg7 
  Die Stellung ist ausgeglichen.  
  = (0.00)  Tiefe: 10/12   00:00:00
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Sf4+ Kg7 
  Die Stellung ist ausgeglichen.  
  = (0.00)  Tiefe: 11/8   00:00:00
2.d8D Sf7+ 3.Kd7 Sxd8 4.Kxd8 La5+ 5.Ke7 Kg6 6.Sxe3 Sb4 7.Sc4 
  Schwarz steht deutlich besser.  
  -/+ (-1.56)  Tiefe: 12/14   00:00:00
...
2.Sxe3 La5 3.Le2 Sb8 4.Lb5 Sf7+ 5.Kxc5 Se5 6.La4 Sbxd7+ 7.Kb5 Ld8 8.Sd5 Sf6 9.Sxc3 Le7 10.Lc2+ Kg7 11.Lb3 Ld6 12.Le6 Kg6 13.Se2 Kg5 14.Sd4 Se4 15.Se2 Sf3 16.Kc6 Lxg3 17.Kd5 Sf6+ 18.Kc5 
  Schwarz steht klar auf Gewinn.  
  -+ (-4.97)  Tiefe: 37/69   00:01:32  200MN, tb=2225211
2.Sf6+ 
  Schwarz steht auf Gewinn.  
  -+ (-3.97 ++)  Tiefe: 38/59   00:01:43  223MN, tb=2448774
...
2.Sf6+ 
  Weiß steht besser.  
  +/- (1.08 ++)  Tiefe: 38/59   00:01:44  224MN, tb=2454757
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.Df6 Kxg3 7.Df1 Sc7 8.Kxc7 Se6+ 9.Kb6 Sd4 10.Ld1 c2 11.Dg1+ Kf4 12.Dh2+ Ke4 13.Lxc2+ Kf3 14.Dh1+ Kf2 15.Dh4+ Ke2 16.Lh7 Kd2 17.Dxh6 La3 18.Lb1 Lb4 19.Dg5 La3 20.La2 Kd3 21.Dg6+ Kd2 22.Lc4 e2 
  Weiß steht deutlich besser.  
  +- (2.03)  Tiefe: 38/59   00:01:44  225MN, tb=2485650
...
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.Df6 Kxg3 7.Df1 c4+ 8.Kd5 Lc5 9.Dg1+ Kf3 10.Ld1+ e2 11.Df1+ Lf2 12.Dxe2+ Kg2 13.Dg4+ Lg3 14.Kxc4 Sb8 15.Lc2 Kf2 16.Df5+ Ke2 17.Dd3+ Kf2 18.Dd4+ Kf3 19.Dg1 Kf4 20.Db6 Kf3 21.Dd4 Kg2 22.De3 Sc6 
  Weiß steht klar auf Gewinn.  
  +- (5.19 ++)  Tiefe: 40/56   00:01:56  250MN, tb=3035034

And one of Lc0 with RTX5070 Nvidia GPU, because not as deterministic as A-B single thread, 3 times repeated, each time at once at best move stable and about as fast on winning line and eval as this one was:

Code: Select all

8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1

Analysis by Lc0 v0.32.1trt:

2.Sf6+ Kg7 
  Schwarz hat minimalen Vorteil.  
  = (-0.30)  Tiefe: 1/2   00:00:05  2Knoten
2.Sf6+ Kg7 3.Sh5+ 
  Die Stellung ist ausgeglichen.  
  = (-0.20)  Tiefe: 2/3   00:00:05  3Knoten
...
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.De8 
  Schwarz steht besser.  
  -/+ (-0.88)  Tiefe: 5/9   00:00:05  17Knoten
...
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Sf7+ 6.Ke6 Sxd8+ 7.Kf5 e2 8.Le4 e1S 9.Ld5 c2 10.Lc4 c1S 11.Lb5 Sc7 12.La4 Se2 13.Ld1 Sf3 14.Lxe2 Sde6 15.Lxf3# 
  Schwarz steht etwas besser.  
  =/+ (-0.32)  Tiefe: 14/28   00:00:05  210Knoten
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Sf7+ 6.Ke6 Sxd8+ 7.Kf5 e2 8.Le4 e1S 9.Ld5 c2 10.Lc4 c1S 11.Lb5 Sc7 12.La4 Se2 13.Ld1 Sf3 14.Lxe2 Sc6 15.Lxf3# 
  Weiß steht besser.  
  +/- (0.78)  Tiefe: 15/29   00:00:05  244Knoten
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Sf7+ 6.Ke6 Sxd8+ 7.Kf5 e2 8.Le4 e1S 9.Ld5 c2 10.Lc4 c1S 11.Lb5 Sc7 12.La4 Se2 13.Ld1 Sf3 14.Lxe2 Sf7 15.Lxf3# 
  Weiß hat entscheidenden Vorteil.  
  +- (3.42)  Tiefe: 16/29   00:00:05  332Knoten
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Sf7+ 6.Ke6 Sxd8+ 7.Kf5 e2 8.Le4 e1S 9.Ld5 c2 10.Lc4 c1S 11.Lb5 Sc7 12.La4 Se2 13.Ld1 Sf3 14.Lxe2 La3 15.Lxf3# 
  Weiß steht klar auf Gewinn.  
  +- (4.77)  Tiefe: 17/30   00:00:05  399Knoten
...
2.Sf6+ Kg7 3.Sh5+ Kg6 4.Lc2+ Kxh5 5.d8D Kg4 6.Kc6 La3 7.Dd1+ Kxg3 8.Dg1+ Kf4 9.Dh2+ Kf3 10.Ld1+ Ke4 11.Dh1+ Ke5 12.Dd5+ Kf6 13.Dd6+ Se6 14.Kd7 Sac7 15.Lc2 e2 16.Dg3 c4 17.Dxc3+ Kf7 18.Df3+ Kg7 19.Dxe2 Lc1 20.Lf5 Lf4 21.Lxe6 
  Weiß steht klar auf Gewinn.  
  +- (4.96)  Tiefe: 23/47   00:00:26  218kN, tb=1916

[Eelco de Groot]

Is Nf6 actually move two of the study? Here below I use it as first move as in the article (if you search for "Plaskett" there are tons of results, also in the forum) .

I get a curious result with Stockfish (Dev version from shortly before Stock 19) if White plays in the cooked version after 4...Kg4 the move that in the beginning was thought the strongest, namely 5.Qf6 This still wins, according to some where others thought there was only a draw. But this was at times where all GMs mainly used Chessbase I think, before Rybka, probably a version of Fritz 10 or something. But tactically that was good.

https://en.wikipedia.org/wiki/Plaskett%27s_Puzzle

1.Nf6+ Kg7 2.Nh5+ Kg6 3.Bc2+ Kxh5 4.d8Q Kg4 (cook) 5.Qf6 (Ehn and Kastner considered this still winning but not intuititively obvious anymore) Kxg3 (why not) 6.Qf1

This is not forced but in some depths, also SF thought it good enough. But after ten hours the evaluation went down. Still at +4 something but it was +7 (*)

So at least I think the Wikipedia article should get an update about the 5.Qf6 line. Because it seems cooked also

[d]8/8/n2K3p/2p3n1/1b6/2p1p1k1/2B5/5Q2 b - -

33 555:17 +4.48 6...c4+ 7.Kd5 Pc7+ 8.Kxc4 Ld6 9.Ld1 Ph3 10.Df3+ Kh2 11.Dxe3 Lg3 12.Dxh6 c2 13.Lxc2 Pe8 14.Kd3 Lf4 15.Dg6 Pd6 16.Lb3 Lg3 17.Ke3 Pf4 18.Dh7+ Ph3 19.Dc2+ Pf2 (28.960.094.726) 869
33 555:17 +10.44 6...Pf3 7.Ld1 c4+ 8.Kd5 Pc7+ 9.Kc6 Pe5+ 10.Kxc7 h5 11.De1+ Kf4 12.Dh4+ Kf5 13.Lc2+ Ke6 14.Dxh5 Ld6+ 15.Kb6 Pd7+ 16.Kc6 Pe5+ 17.Kb5 e2 18.Dxe2 La3 19.De4 Lc1 (28.960.094.726) 869
32 555:17 +12.55 6...Pc7 7.Dg1+ Kf4 8.Dh2+ Kf3 9.Ld1+ Ke4 10.Dg2+ Kf5 11.Kxc7 Pe6+ 12.Kc6 h5 13.Lc2+ Kf6 14.Df3+ Kg7 15.Dxe3 Pd4+ 16.Kd5 Pxc2 17.De7+ Kg6 18.De4+ Kg7 19.Dxc2 h4 (28.960.094.726) 869
32 555:17 +12.98 6...La3 7.Dg1+ Kf3 8.Ld1+ e2 9.Df1+ Kg3 10.De1+ Kh3 11.Lxe2 c2 12.Dc3+ Kg2 13.Dxc2 Pb4 14.Df5 c4 15.Df1+ Kg3 16.Ke5 Pc6+ 17.Kd5 Pe7+ 18.Kxc4 Pc8 19.Kb3 Le7 (28.960.094.726) 869
32 555:17 +13.04 6...e2 7.Dxe2 c4+ 8.Kd5 Lc5 9.Dxc4 Pc7+ 10.Kxc5 Pce6+ 11.Kd6 Kg2 12.Dxc3 Pf8 13.Kd5 Pge6 14.Dd2+ Kg3 15.Dxh6 Kf2 16.Df6+ Ke2 17.Kc4 Ke1 18.Df3 Kd2 (28.960.094.726) 869
32 555:17 +13.49 6...Ph3 7.De1+ Kf3 8.Ld1+ Kg2 9.Dxe3 Pg1 10.Lc2 h5 11.Le4+ Kf1 12.Dc1+ Kf2 13.Df4+ Ke2 14.Kd5 Pc7+ 15.Dxc7 Kd1 16.Dh2 Pe2 17.Lf3 Kc2 18.Dxe2+ Kb3 19.Le4 Ka4 (28.960.094.726) 869


(*) before it went down:

[d]8/8/n2K3p/2p3n1/1b6/2p1p1k1/2B5/5Q2 b - -

Engine: Stockfish dev-20260726-23cf5d82 (1024 MB)
gemaakt door the Stockfish developers (see AUTHORS f

28 144:40 +6.91 6...c4+ 7.Kd5 Pc7+ 8.Kxc4 Pf3 9.Kxb4 Pd5+
10.Kc5 Pf6 11.De2 Kf4 12.Ld1 Pd7+
13.Kc4 Pfe5+ 14.Kxc3 Ke4 15.De1 Pf8
16.Df1 Peg6 17.Lc2+ Ke5 18.Lxg6 Pxg6
19.Db5+ Kf4 (6.133.288.803) 706

28 144:40 +7.23 6...Pf3 7.Ld1 c4+ 8.Kd5 Pc7+ 9.Kc6 Pe5+
10.Kxc7 h5 11.De1+ Kf4 12.Dh4+ Kf5
13.Lc2+ Ke6 14.Dxh5 Lc5 15.Dh3+ Kd5
16.Dg2+ Ke6 17.De4 Ld6+ 18.Kb6 e2
19.Dxe2 Lb8 (6.133.288.803) 706

28 144:40 +7.48 6...Ph3 7.De1+ Pf2 8.Dxe3+ Kg2 9.De2 c4+
10.Kd5 Lc5 11.Kxc4 Kg3 12.De5+ Kf3
13.Dh5+ Kg3 14.Dxh6 Pg4 15.Dxa6 Lf2
16.Kxc3 Le1+ 17.Kb3 Lf2 18.Dg6 Le3
19.Lf5 Kf2 (6.133.288.803) 706

28 144:40 +8.64 6...Pc7 7.Dg1+ Kf4 8.Dh2+ Kf3 9.Ld1+ Ke4
10.Dg2+ Kf5 11.Kxc7 c2 12.Dg4+ Kf6
13.Lxc2 Pf7 14.Df4+ Ke7 15.Dxe3+ Kf6
16.Kc6 La5 17.Dxc5 Le1 18.Kd5 h5
19.De3 La5 (6.133.288.803) 706

28 144:40 +8.76 6...Pb8 7.Ld1 Kh2 8.De2+ Kh1 9.Dxe3 La5
10.Kxc5 Lc7 11.Kb5 Lh2 12.Dxc3 Kg2
13.Dd2+ Kh3 14.De2 Lc7 15.Dd3+ Lg3
16.Df1+ Kh2 17.Lc2 Lh4 18.De2+ Kg3
19.De5+ Kh3 (6.133.288.803) 706

28 144:40 +10.89 6...La3 7.Dg1+ Kf4 8.Dh2+ Kg4 9.De2+ Kf4
10.Dxa6 Pf3 11.Kd5 Kg3 12.De2 Lc1
13.Le4 Pd4 14.De1+ Kg4 15.Dxc1 e2
16.Dxc3 Kg5 17.Ld3 h5 18.Dd2+ Kf6
19.Lxe2 Pb3 (6.133.288.803) 706


(Stockfish decidedly went down in the evaluation, still at + 4 but that is not Mate!

It could well be with just 1024 MB I get search instabilities but the rest of the Multi PV lines seem winning for White. Just not the main line)

[peter]

Is Nf6 actually move two of the study?

No. Maybe you mean a version, Uri posted once in a while according to my database with one more added before original starting position, and then there was another one similiar position in fora afair, avoiding the longer way to mate deviating from author's main line at 4th move of Black's (see below at 4...Kg4 instead of the thematic way 4...Nf7+). If you search for the position here, you'll find quite some more, I guess.

"Plaskett's Puzzle", as it's called sometimes too,

https://en.wikipedia.org/wiki/Plaskett%27s_Puzzle

composed by G.van Breukelen, is in HHdbVI like this:
[pgn][Event "no ty"]
[Site "; Panorama 11 /1976"]
[Date "1976.??.??"]
[Round "?"]
[White "Van_Breukelen=G"]
[Black "(+0047.24d6h7) M"]
[Result "1-0"]
[WhiteFideId "-1"]
[BlackFideId "-1"]
[SetUp "1"]
[FEN "8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1"]
[PlyCount "27"]
[GameId "2352874051060524"]
[EventDate "1976.??.??"]

1. Nf6+ Kg7 2. Nh5+ Kg6 (2... Kh8 3. d8=Q+) 3. Bc2+ Kxh5 4. d8=Q $1 Nf7+ (4... Kg4 5. Qf6 Kxg3 6. Qe5+ Kf2 7. Qh2+ Ke1 8. Qg1+ Ke2 9. Qg4+ Ke1 10. Ke5) 5. Ke6 Nxd8+ 6. Kf5 e2 7. Be4 e1=N $1 8. Bd5 $1 c2 9. Bc4 c1=N $1 10. Bb5 Nc7 11. Ba4 $1 Ne2 12. Bd1 Nf3 13. Bxe2 Nce6 14. Bxf3# 1-0
[/pgn]
And in YACPDB:
https://www.yacpdb.org/#search/OFwvM1Az ... LzEvMS8w/1

This one authors thematic #14 isn't the longest defend of Black's, with 4...Kg4 instead of ...Nf7+ it's a mate in 32 overall. See Slowchess after some Forward- Backward here:

SlowChess Blitz 2.9 avx2 (32768 MB hash, 31 threads)
FEN: 8/3P3k/n2K3p/2p3n1/1b4N1/2p1p1P1/8/3B4 w - - 0 1

[44] M32 01:00.7 2072.52M Nf6+ Kg7 Nh5+ Kg6 Bc2+ Kxh5 d8=Q Kg4 Qf6 Kxg3 Qe5+ Kf3 Bd1+ Kg2 Qxe3 c4+ Kd5 Bc5 Qxc3 Nc7+ Kxc4 Nce6 Qf6 Bd4 Qxh6 Kg3 Qh5 Kf4 Kd5 Kg3 Qg4+ Kf2 Kc4 Ba7 Qh4+ Kg2 Bg4 Bf2 Qh6 Bg3 Bxe6 Bf4 Kd3 Kg3 Qf6 Bc1 Qe5+ Bf4 Qe1+ Kg2 Qe2+ Kg3 Qg4+ Kh2 Qxf4+ Kg2 Qxg5+ Kh2 Kc4 Kh1 Bd5+ Kh2 Qg2#

[Eelco de Groot]

[d]8/8/n2K1Q1p/2p3n1/1b6/2p1p1k1/2B5/8 w - -

Engine: Stockfish dev-20260726-23cf5d82 (1024 MB)
gemaakt door the Stockfish developers (see AUTHORS f

38 281:22 +12.27 6.De5+ Kg2 7.Dxe3 Pf7+ 8.Ke7 Pg5
9.De2+ Kg3 10.Kd6 Pf3 11.Dxa6 Pd4
12.Kd5 Kf2 13.Dd3 La3 14.Dxc3 Lb4
15.Dh3 Pe2 16.Dh2+ Ke3 17.Dxh6+ Kf2
18.Ld3 Pc3+ 19.Ke5 (13.357.225.494) 791

38 281:22 +7.56 6.Df1 c4+ 7.Kd5 Pc7+ 8.Kxc4 Pce6
9.Dg1+ Kf3 10.Ld1+ e2 11.Df1+ Kg3
12.De1+ Kh3 13.Dxe2 Ld6 14.La4 Kh4
15.Ld7 c2 16.Dxc2 Pc5 17.Df2+ Kh5
18.Lf5 Pf7 19.Kd5 (13.357.225.494) 791

38 281:22 0.00 6.Ke5 e2 7.Df4+ Kg2 8.Dg4+ Kf2
9.Dh4+ Kf1 10.Dh1+ Kf2 11.Dh2+ Kf1 (13.357.225.494) 791

38 281:22 0.00 6.Lg6 e2 7.De5+ Kf2 8.Dh2+ Ke3
9.De5+ (13.357.225.494) 791

38 281:22 0.00 6.Dg7 e2 7.De5+ Kf2 8.Df5+ Ke1 9.Df4 La3
10.Dg3+ Kf1 11.Df4+ (13.357.225.494) 791

38 281:22 0.00 6.Df5 e2 7.De5+ Kf2 8.Df5+ Ke1 9.Df4 La3
10.Dg3+ Kf1 11.Df4+ Ke1 (13.357.225.494) 791

Thanks Peter for clearing it up. I blame it on Stockfish 19 Slowchess Mate could be added to the article. 6.Qe5+ is the best move Stock 19 now agrees when going one ply back (see above) then but not certain about 6.Qf1 Trying to reach depth 39 but now ten hours underway trying to resolve new depth 39 for 6.Qe5+ Fail High. Ten hours. As I was seeing search explosions pretty much all the time there must be bug. Nooob's bot found bug. New issue but it was there before Stockfish 18 already Humans no longer check or too many Stockfish versions? Or now somehow the thing surfaces because of other changes?

New issue https://github.com/official-stockfish/S ... ssues/7156

[peter]

6.Qe5+ is the best move Stock 19 now agrees when going one ply back (see above) then but not certain about 6.Qf1 Trying to reach depth 39 but now ten hours underway trying to resolve new depth 39 for 6.Qe5+ Fail High.

After 1.Nf6+ Kg7 2.Nh5+ Kg6 3.Bc2+ Kxh5 4.d8Q Kg4 5.Qf6 Kxg3
6.Qf1+!? (instead of 6.Qe5+):

8/8/n2K3p/2p3n1/1b6/2p1p1k1/2B5/5Q2 b - -

Engine: SlowChess2.9 (32768 MB)
von Jonathan Kreuzer

48/76 3:45 +M38 6...c4+ 7.Kd5 La5 8.Dg1+ Kf3 9.Ld1+ e2
10.Df1+ Kg3 11.De1+ Kh3 12.Lxe2 Sb4+
13.Kxc4 c2 14.Dc3+ Kh2 15.Kb5 Se4
16.Db2 h5 17.Kxa5 Sc6+ 18.Kb6 Sd4
19.Ld3 Sd2 (7.772.695.720) 34406
TB:217.939.053

MultiPV=2 one ply earlier:

49 7:55 +M27 6.De5+ Kf3 7.Ld1+ Kg2 8.Dxe3 c4+ 9.Kd5 Lc5 10.Dxc3 Sc7+ 11.Kxc4 Sce6 12.Df6 Ld4 13.Dxh6 Kg3 14.Dh5 Kf4 15.Kd5 Kg3 16.Dg4+ Kf2 17.Kc4 Lb6 18.Dh4+ Kf1 19.Lg4 (16.252.889.456) 34172 TB:430.939.747
48 7:55 +M39 6.Df1 c4+ 7.Kd5 La5 8.Dg1+ Kf3 9.Ld1+ e2 10.Df1+ Kg3 11.De1+ Kh3 12.Lxe2 Sb4+ 13.Kxc4 c2 14.Dc3+ Kh2 15.Kb5 Se4 16.Db2 h5 17.Kxa5 Sc6+ 18.Kb6 Sd4 19.Ld3 (16.252.889.456) 34172 TB:430.939.747

With this one second primary line in joint output, SlowChess doesn't resolve the iteration within 20 minutes neither, 6.Qf1- line still was grey at copying and no new update so far, stopped then, regards

[Eelco de Groot]

Thanks Peter. What I can see, on this end, Stockfish did not get into endless loop in quiescence or something, there was a 39 ply result, no Fail High found apparently, because eval lower after somewhere 30 hours total.


[d]8/8/n2K1Q1p/2p3n1/1b6/2p1p1k1/2B5/8 w - -

Engine: Stockfish dev-20260726-23cf5d82 (1024 MB)
gemaakt door the Stockfish developers (see AUTHORS f

39 1826:03 +8.97 6.De5+ Kg2 7.Dxe3 Pf7+ 8.Ke7 Pg5
9.De2+ Kg3 10.Kd6 c4+ 11.Kd5 Pc5
12.Dxc4 Pce6 13.Dxb4 Pf4+ 14.Kc4 Pg2
15.Dxc3+ Kg4 16.Ld1+ Kf4 17.Df6+ Kg3
18.Dxh6 Ph3 19.Kc3 (146.642.931.858) 1338

39 1826:03 +8.47 6.Df1 c4+ 7.Kd5 Pc7+ 8.Kxc4 Ld6
9.Ld1 Ph3 10.Df3+ Kh2 11.Dxe3 Lg3
12.Dxh6 c2 13.Lxc2 Pe8 14.Kd3 Pd6
15.Ke2 Kg2 16.Dd2 Pg1+ 17.Kd1+ Kf3
18.Dd5+ Kf2 19.Lb3 (146.642.931.858) 1338

39 1826:03 0.00 6.Ke5 e2 7.Df4+ Kg2 8.Dg4+ Kf2
9.Dh4+ Kf1 10.Dh1+ Kf2 (146.642.931.858) 1338

39 1826:03 0.00 6.Lg6 e2 7.De5+ Kf2 8.Dh2+ Ke3
9.De5+ (146.642.931.858) 1338

39 1826:03 0.00 6.Dg7 e2 7.De5+ Kf2 8.Df5+ Ke1 9.Df4 La3
10.Dg3+ Kf1 11.Df4+ Ke1 (146.642.931.858) 1338

39 1826:03 0.00 6.Df5 e2 7.De5+ Kf2 8.Df5+ Ke1 9.Df4 Pc7
10.Dg3+ Kf1 11.Kxc7 e1D 12.Ld3+ De2
13.Kb6 Dxd3 14.Dxd3+ Kf2 15.Dc2+ Ke3
16.Dc1+ Kd3 17.Dd1+ Ke3 (146.642.931.858) 1338