Winning after Qf6+!

[peter]

Stockfish-dev has the following:

[pgn][Event ""]
[Site ""]
[Date ""]
[Round ""]
[White ""]
[Black ""]
[Result ""]
[FEN "1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - 0 1"]
[Setup "1"]

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4 Nd6 6. Bg3 Qd8 7. Rg7+
Kh8 8. Kg2 Ne8 9. Rxf7 Rxf7 10. Nxf7+ Kg8 11. Nxd8 Rxd8 12. h5 g5 13. Rf1 Nc7
14. f7+ Kg7 15. Be1 Rf8 16. Bd2 Rxf7 17. h6+ Kg8 18. Rxf7 Kxf7 19. Bh5+ Kg8 20.
Bg6 Be8 21. Bb1 Bh5 22. Bxg5 Ne8 23. Be7 Ng7 24. hxg7 Kxg7 25. Kf2 Kf7 26. Bc5
Bd1 27. Ke3 Bg4 28. Bb6 Bh5 29. Ba5 Bg4 30. Kd2 Bh5 31. Bd8 Bg4 32. Kc1 Be2 33.
Bc7 Bh5 34. Bb6 Bg4 35. Ba5 Bh5 36. Kd2 Bf3 37. Bc7 Bh5 38. Bb8 Bg4 39. Kc1 Kg7
40. Bd6 Kf7 41. Bb4 Bf3 42. Ba3 Kg7 43. Bd6 Kf7 44. Kd2 Bh5 45. Ba3 Kg7 46. Ke3
Kf7 47. Kf4 Be2 48. Bd6 Bd1 49. Ke3 Bg4 50. Bc5 Bh5 51. Ba3 Bd1 52. Kd2 Bh5 53.
Ke1 Bf3 54. Bc5 Bg4 55. Bd6 Bh5 56. Bb4 Bg4 57. Bc5 Bh5 58. Kf1 Bg4 59. Bd6 Bf3
60. Kf2 Bg4 61. Ba3 Bh5 62. Kg3 Kg7 63. Bd6 Bd1 64. Be7 Kf7 65. Ba3 Bh5 66. Kh2
Be2 67. Kg2 Bd3 68. Bxd3 cxd3 69. Kf2 { 1.23} [/pgn]

I see, so we'd have to have a closer look, if the material advantage, White has got after regaining the Queen (Bishop against 1 Pawn remaining) yet ends up in a fortress. That's what your output- line suggests, but then even the eval of your SF dev. would be too high, even more the one of amsFish.

Having had only Forward till now, I'd say maybe latest 33.g5 could be a white blunder leading to the fortress, not sure about that and the moves before.

Anyhow, we can agree, I guess, that the Fat- Fritz- Output isn't the best line after 25.Ng5 (?) Nxa3, which should be drawn by best play of both sides, can't we?
I'd still say 26...Nd6 makes things unnecessarily complicated, maybe the ? is too muc, ?! might be better.

[zullil]

Anyhow, we can agree, I guess, that the Fat- Fritz- Output isn't the best line after 25.Ng5 (?) Nxa3, which should be drawn by best play of both sides, can't we?
I'd still say 26...Nd6 makes things unnecessarily complicated, maybe the ? is too muc, ?! might be better.

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4 Nxd4

[d]1r3rk1/5p1R/1qb1pPp1/3pP1N1/p1pn3P/1pP5/1P2B3/3RB2K w - - 0 6
Stockfish-dev says:

+9.24 6. Rg7+ Kh8 7. cxd4 c3 8. Bxc3 Bb5 9. Rh7+ Kg8 10. Bg4 Qa6 11. Rg7+ Kh8 12. Rd2 a3 13. h5 a2 14. hxg6 a1=Q+ 15. Bd1 Qxd1+ 16. Rxd1 fxg6 17. Bb4 Bd3 18. Kg2 Bf1+ 19. Rxf1 Qe2+ 20. Rf2 Qh5 21. Rh7+ Qxh7 22. Nxh7 Kxh7 23. Bxf8 Rxf8 24. Kh3 Rc8 25. Rf3 Kg8 26. Kg4 Rc2 27. Rxb3 Rg2+ 28. Kf3 Rh2 29. Rb8+ Kf7 30. Rb7+ Kf8 31. Re7 Rh4 32. Ke3 g5 33. Rxe6 Rh3+ 34. Kd2 Rh4 35. Kd3 Rh3+ 36. Kc2 Re3 37. Kd2 Rf3 38. Re7 g4 39. Rg7 Ke8 40. Rxg4 Kd8 41. Ke2 Rf5 42. Rg8+ Kc7 43. Rg7+ Kc8 44. Ke3 (depth 53, 0:06:40)

[MikeB]

A much discussed position:

[d]1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - bm Qf6
Here we see Lc0 finding Qf6+ instantly. And the claim is made that its line is correct. But is it? The line given starts with Qf6+ Bxf6 gxf6+ Kg8 Ng5.

Here Stockfish seems to show that Bd2 (rather than Ng5) is indeed winning. But how does White win after Ng5 Nxa3?

Qf6+ is best move , but it's just a draw after doing an overnight search with about 1.4 trillion nodes searched. Maybe if somebody searched 100 trillion nodes, they might find something different. That would require dually epyc processors for about a week.

Some quick 5 min analysis

dep score nodes time (not shown: tbhits knps seldep)
33 +50.04 6.25G 4:49.80 Qf6+ Bxf6 gxf6+ Kg8 Bf2 Ba8 Ng5 h5 Rg1 Nb4 cxb4 c3 Rxh5 cxb2 Rh7 b1=Q Rxb1 Qc6 Rg7+ Kh8 Rh7+
33 +36.40 6.44G 4:58.45 Rh6 Rh8 h4 Nxa3 Nh2 Nc2 Ng4 a3 h5 Be8 Bd2 a2 Nf6 Ra8 Kg2 a1=Q Rxa1 Nxa1 Bg4 Qc6 hxg6 fxg6 Rh1 Nc2 Qh2 Bxf6 Qh6+ Kg8 gxf6 Ra7 Bg5 Rf7 Bd2 Rb7 Bg5 Qc8 Rf1 Rf7 Bd2 Bd7 Rf3 Qf8 Qh3 h6 Bxe6 Qe8 Bxd7 Qxd7 e6 Qd8 Qh4
33 +28.77 6.45G 4:59.25 Bf2 Rh8 Nd2 Qd8 Rf3 Rf8 Nb1 Bxg5 Qg4 f5 exf6+ Bxf6 Rf1 Rb7 Bh4 Rbf7 Qxe6 Bd7 Rxf6 Bxe6 Rxf7+ Bxf7 Bxd8 Rxd8 Kg1 h5 Nd2 Re8 Kf2 Be6 Rg1 Rf8+ Bf3 Kh6 h4 Bg4 Ke2 Rf4 Rh1
33 +26.95 7.72G 5:58.99 Bd2 Nxa3 Rf1 Nc2 Bd1 a3 Bxc2 bxc2 bxa3 Qb1 Ng1 Qd1 Qh4 Rh8 Rhf3 Be8 Bc1 Rb1 Qf2 Rg8 Ne2 h5 gxh6+ Kh8 Rg3 g5 Rgg1 Qd3 Rg3 Qh7 Kg1 Rg6 Rf3 Rxh6 Rg3
33 +21.08 8.26G 6:24.08 Nd2 Qd8 Bg3 Nxa3 bxa3 Bxa3 Qf6+ Qxf6 exf6+ Kg8 Nb1 Bb2 Bd6 Rfd8 Bxb8 Rxb8 Kg2 a3 Nxa3 Bxa3 Rb1 Bd6 Bd1 Bd7 Rh4 e5 Bg4 Bc6 dxe5 Bxe5 Bf3 Bxc3 Rxc4 dxc4 Bxc6 b2 Bd5 Rc8 Kf3 Be5 Ke4 c3 Kxe5 c2

[jp]

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4 Nxd4

1r3rk1/5p1R/1qb1pPp1/3pP1N1/p1pn3P/1pP5/1P2B3/3RB2K w - - 0 6

Stockfish-dev says:

+9.24 6. Rg7+ Kh8 7. cxd4 c3 <snip> (depth 53, 0:06:40)

So this contradicts Peter's presumably asmFish result, right?

1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - bm Qf6

Qf6+ is best move , but it's just a draw after doing an overnight search with about 1.4 trillion nodes searched. Maybe if somebody searched 100 trillion nodes, they might find something different. That would require dually epyc processors for about a week.

But this contradicts Louis's SFdev search with 1.77 trillion nodes. Was your search with Honey?

[Ovyron]

Lol guys, lern2analyze, leaving any engine analyzing overnight has always been a waste of time, it gives the illusion of being good because of so many positions, but these discussions could have taken place in Rybka 3 times and people would have said things like "Rybka doesn't see it after X nodes, blah blah blah."

If you know how to guide the engine then you can know the truth about the position without needing to analyze more than 1 trillion nodes. You can do it in 200 positions. But it requires more than the press of a button, and what 200 positions you need to check is the tricky part

[zullil]

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4 Nxd4

1r3rk1/5p1R/1qb1pPp1/3pP1N1/p1pn3P/1pP5/1P2B3/3RB2K w - - 0 6

Stockfish-dev says:

+9.24 6. Rg7+ Kh8 7. cxd4 c3 <snip> (depth 53, 0:06:40)

So this contradicts Peter's presumably asmFish result, right?

If I haven't muddled the lines, it seems to suggest that Ng5 is not as easily handled as was suggested.

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4

and now Nd6 is best, and might lead to a fortress-like draw.

[zullil]

Lol guys, lern2analyze, leaving any engine analyzing overnight has always been a waste of time, it gives the illusion of being good because of so many positions, but these discussions could have taken place in Rybka 3 times and people would have said things like "Rybka doesn't see it after X nodes, blah blah blah."

If you know how to guide the engine then you can know the truth about the position without needing to analyze more than 1 trillion nodes. You can do it in 200 positions. But it requires more than the press of a button, and what 200 positions you need to check is the tricky part

As is your pattern, another proclamation on the "right" way to do things, with (as always) nothing substantive to offer. No analysis. No evidence to support your claims. Please stick to working on your depth 22 opening book, so that you can win chess games with your special moves. That seems important to you.

I'm now taking the rare step of adding someone to my ignore list; you can keep Chessqueen company.

[peter]

Qf6+ is best move , but it's just a draw after doing an overnight search with about 1.4 trillion nodes searched. Maybe if somebody searched 100 trillion nodes, they might find something different. That would require dually epyc processors for about a week.

Some quick 5 min analysis

dep score nodes time (not shown: tbhits knps seldep)
33 +50.04 6.25G 4:49.80 Qf6+ Bxf6 gxf6+ Kg8 Bf2 Ba8 Ng5 h5 Rg1 Nb4 cxb4 c3 Rxh5 cxb2 Rh7 b1=Q Rxb1 Qc6 Rg7+ Kh8 Rh7+
33 +36.40 6.44G 4:58.45 Rh6 Rh8 h4 Nxa3 Nh2 Nc2 Ng4 a3 h5 Be8 Bd2 a2 Nf6 Ra8 Kg2 a1=Q Rxa1 Nxa1 Bg4 Qc6 hxg6 fxg6 Rh1 Nc2 Qh2 Bxf6 Qh6+ Kg8 gxf6 Ra7 Bg5 Rf7 Bd2 Rb7 Bg5 Qc8 Rf1 Rf7 Bd2 Bd7 Rf3 Qf8 Qh3 h6 Bxe6 Qe8 Bxd7 Qxd7 e6 Qd8 Qh4
33 +28.77 6.45G 4:59.25 Bf2 Rh8 Nd2 Qd8 Rf3 Rf8 Nb1 Bxg5 Qg4 f5 exf6+ Bxf6 Rf1 Rb7 Bh4 Rbf7 Qxe6 Bd7 Rxf6 Bxe6 Rxf7+ Bxf7 Bxd8 Rxd8 Kg1 h5 Nd2 Re8 Kf2 Be6 Rg1 Rf8+ Bf3 Kh6 h4 Bg4 Ke2 Rf4 Rh1
33 +26.95 7.72G 5:58.99 Bd2 Nxa3 Rf1 Nc2 Bd1 a3 Bxc2 bxc2 bxa3 Qb1 Ng1 Qd1 Qh4 Rh8 Rhf3 Be8 Bc1 Rb1 Qf2 Rg8 Ne2 h5 gxh6+ Kh8 Rg3 g5 Rgg1 Qd3 Rg3 Qh7 Kg1 Rg6 Rf3 Rxh6 Rg3
33 +21.08 8.26G 6:24.08 Nd2 Qd8 Bg3 Nxa3 bxa3 Bxa3 Qf6+ Qxf6 exf6+ Kg8 Nb1 Bb2 Bd6 Rfd8 Bxb8 Rxb8 Kg2 a3 Nxa3 Bxa3 Rb1 Bd6 Bd1 Bd7 Rh4 e5 Bg4 Bc6 dxe5 Bxe5 Bf3 Bxc3 Rxc4 dxc4 Bxc6 b2 Bd5 Rc8 Kf3 Be5 Ke4 c3 Kxe5 c2

Could you please have a look at the now about 100 times repeated lines after 24.Bd2 (!) instead of 24.Ng5? (or Bf2 (?) as in your output's line)

One thing at least hasn't changed since 2014, when Lyudmil came along with the position.
Nobody is willing to have a closer look at the lines but letting SF and SF branches ponder stand alone and believe in the wrong evals, no matter if now in win- percentage instead of cp.

Nowadays we have LC0 and it even finds 22.Qf6! and 24.Bd2!, yet nobody is willing to see at least the difference to 24.Ng5 (?).
To you and Louis, Mike, I'm outa here again.
Sorry, but it's getting too time- consuming, and I don't talk about hardware- time.

[zullil]

Qf6+ is best move , but it's just a draw after doing an overnight search with about 1.4 trillion nodes searched.

Stockfish-dev seems pretty happy with

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Bd2.

See here.

[zullil]

Qf6+ is best move , but it's just a draw after doing an overnight search with about 1.4 trillion nodes searched. Maybe if somebody searched 100 trillion nodes, they might find something different. That would require dually epyc processors for about a week.

Some quick 5 min analysis

dep score nodes time (not shown: tbhits knps seldep)
33 +50.04 6.25G 4:49.80 Qf6+ Bxf6 gxf6+ Kg8 Bf2 Ba8 Ng5 h5 Rg1 Nb4 cxb4 c3 Rxh5 cxb2 Rh7 b1=Q Rxb1 Qc6 Rg7+ Kh8 Rh7+
33 +36.40 6.44G 4:58.45 Rh6 Rh8 h4 Nxa3 Nh2 Nc2 Ng4 a3 h5 Be8 Bd2 a2 Nf6 Ra8 Kg2 a1=Q Rxa1 Nxa1 Bg4 Qc6 hxg6 fxg6 Rh1 Nc2 Qh2 Bxf6 Qh6+ Kg8 gxf6 Ra7 Bg5 Rf7 Bd2 Rb7 Bg5 Qc8 Rf1 Rf7 Bd2 Bd7 Rf3 Qf8 Qh3 h6 Bxe6 Qe8 Bxd7 Qxd7 e6 Qd8 Qh4
33 +28.77 6.45G 4:59.25 Bf2 Rh8 Nd2 Qd8 Rf3 Rf8 Nb1 Bxg5 Qg4 f5 exf6+ Bxf6 Rf1 Rb7 Bh4 Rbf7 Qxe6 Bd7 Rxf6 Bxe6 Rxf7+ Bxf7 Bxd8 Rxd8 Kg1 h5 Nd2 Re8 Kf2 Be6 Rg1 Rf8+ Bf3 Kh6 h4 Bg4 Ke2 Rf4 Rh1
33 +26.95 7.72G 5:58.99 Bd2 Nxa3 Rf1 Nc2 Bd1 a3 Bxc2 bxc2 bxa3 Qb1 Ng1 Qd1 Qh4 Rh8 Rhf3 Be8 Bc1 Rb1 Qf2 Rg8 Ne2 h5 gxh6+ Kh8 Rg3 g5 Rgg1 Qd3 Rg3 Qh7 Kg1 Rg6 Rf3 Rxh6 Rg3
33 +21.08 8.26G 6:24.08 Nd2 Qd8 Bg3 Nxa3 bxa3 Bxa3 Qf6+ Qxf6 exf6+ Kg8 Nb1 Bb2 Bd6 Rfd8 Bxb8 Rxb8 Kg2 a3 Nxa3 Bxa3 Rb1 Bd6 Bd1 Bd7 Rh4 e5 Bg4 Bc6 dxe5 Bxe5 Bf3 Bxc3 Rxc4 dxc4 Bxc6 b2 Bd5 Rc8 Kf3 Be5 Ke4 c3 Kxe5 c2

Could you please have a look at the now about 100 times repeated lines after 24.Bd2 (!) instead of 24.Ng5? (or Bf2 (?) as in your output's line)

One thing at least hasn't changed since 2014, when Lyudmil came along with the position.
Nobody is willing to have a closer look at the lines but letting SF and SF branches ponder stand alone and believe in the wrong evals, no matter if now in win- percentage instead of cp.

Nowadays we have LC0 and it even finds 22.Qf6! and 24.Bd2!, yet nobody is willing to see at least the difference to 24.Ng5 (?).
To you and Louis, Mike, I'm outa here again.
Sorry, but it's getting too time- consuming, and I don't talk about hardware- time.

Thanks, Peter. I agree that 1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Bd2 wins.

I think 3. Ng5 only draws, but don't see any alternative to 5...Nd6 in this line.

1. Qf6+ Bxf6 2. gxf6+ Kg8 3. Ng5 Nxa3 4. Rxh7 Nb5 5. h4

And then the draw is not completely clear yet.