Knight v Bishop (part 17)

Black has just captured on d8 in Spanton (1890) - Karl-Heinz Stolzenwald (1623), Bad Wörishofen (Bavaria) 2019. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

White has much the better pawn-structure, but no pawn-majority to work with. Stockfish12 and Komodo11.01 reckon White is slightly better.

K-HS offered a draw as I thought about my next move, but I had no intention of accepting.

27.Nc1 Ke6 28.Nd3 Be7 29.Kf1 f5 30.f3 Bd6 31.Ke2 fxe4!?

Undoubling pawns is natural, and the text is Komodo11.01's choice, but it does free the f3 square for use by White's pieces. On the other hand, if Black does not capture on e4, White can capture on f5 at a time of his choosing and use the e4 square.

32.fxe4 Be7!?

Stockfish12 much prefers 32...Kf6, but Komodo11.01 is fine with the text.

33.Kf3 h5 34.Nc1 Kf6 35.Ne2 Kg5 36.g3

*****

*****

*****

*****

*****

36...Bd6?

Correct is 36...fxg3, which I intended meeting with 37.Nxg3 and Nf5. White has the easier game, but Black is not significantly worse, according to the engines.

37.h4+ Kf6?

Better, but almost certainly still losing, is 37...Kg6 or 37...Kh6, both of which allow Black to defend the h5 pawn without exchanging minor pieces.

38.gxf4 exf4 39.Nxf4 Ke5

39...Bxf4 40.Kxf4 is hopeless for Black.

40.Ke3 Kf6 41.Nxh5+ Kg6 42.Nf4+ Kf6 43.Nd3 Bf8 44.Kf4 Kg6 1-0

Knight v Bishop (part 16)

Black has just captured on e8 in Spanton (1859) - Thorsten Meyer (2028), Malta (Buġibba) 2017. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

White has more space in the centre, but the bishop is at least semi-bad thanks to the white queenside pawns that are fixed, or at least semi-fixed, on light squares. Stockfish12 reckons the position is equal; Komodo11.01 gives White a slight edge.

22.f3 g6 23.Kf2 Nf6 24.Ba4 Kf8 25.Ke3 Ke7 26.a3 b6 27.b4 axb4 28.axb4 Ng8 29.g4 Nh6 30.Bc2 Kf6 31.h4 Ng8 32.g5+ Ke7 33.h5!?

The engines reckon White is slightly better after 33.f4 or 33.Kd4.

33...Kf8 34.h6 Ne7 35.Kd4 f6 36.gxf6!?

A simpler way to maintain the balance is 36.f4, but the text should result in equality too.

36...Nf5+ 37.Ke4?

Correct is 37.Bxf5 gxf5, when 38.f4 is a simple draw, but also equal is 38.Kc3!? Kf7 39.Kb3 Kxf6 (not 39...Kg6 40.Ka4 Kxh6? - taking the f pawn still draws - 41.Kb5 Kg6 42.Kc6 h5 43.Kxc7 h4 44.Kxf6 Kc7 45.h3 d6 46.h2 d7 47.h1=Q d8=Q+, when White is winning, according to the engines) 40.Ka4 f4 41.Kb5 Ke7 42.Kc6 Kd8.

37...Nxh6 38.Kf4 Kf7 39.Kg5 Ng8 40.b5 h6+ 41.Kf4 Kxf6

Black is winning easily. The game finished:

42.Bd3 Kf7 43.Kg3 Nf6 44.Kh4 Nd7 45.f4 Kf6 46.Kg4 g5 47.Be2 Nc5 48.fxg5+ hxg5 49.Bd1 Nd3 50.Be2 Ne5+ 51.Kh5 g4!? 52.Kh4 Kf5 53.Kg3 Ke4 54.Bf1 Kd4 55.Kf4 Nxc4 56.Kxg4 Ne3+ 0-1

Titles

THE international chess federation, known by its French acronym Fide, awards four titles for over-the-board play: grandmaster (GM), international master (IM), Fide master (FM) and, since 2002, candidate master (CM).

To get the GM title you generally need three norms* and have a rating of at least 2500, for the IM title you generally need three norms and a rating of at least 2400, and to get the FM title you generally need to reach a rating of 2300 (no norms).

The Fide-recognised organisation for correspondence chess is the International Correspondence Chess Federation.

It awards five titles for correspondence play: grandmaster, senior international master, international master, master and expert.

The titles are, somewhat inconsistently, abbreviated to GM, SIM, IM, CCM and CCE, the CCs standing for correspondence chess.

As a rough guide, the ICCF title of CCM is equal to the Fide title of FM, but is achieved not by rating but by obtaining two or more norms in tournaments with a total of at least 24 games.

I today completed my first CCM norm, albeit in a tournament which is ongoing and in which I only played eight games.

(My two wins are annotated at https://beauchess.blogspot.com/2020/10/morphing-french-xvi.html and https://beauchess.blogspot.com/2020/11/lockdown-correspondence-chess.html).

MT-Gutiérre/SF4 Board 3, Carlos Flores Gutiérrez Team Tournament SF4 Board 3

							TD Lorenz, Bernhard
Minor category K							1	2	3	4	5	6	7	8	9	CCM	CCE	Score	Wins	SB	RG	Place
1		212116		Spanton, Tim	2260	England 1		½	½	½	½	1	½	½	1	5	4	5	2	14	0	1
2		910125	CCE	Dzenis, Janis (Tukmus)	2252	Latvia	½		½	½	½	½	½	.	½	5		3.5	0	11.25	1	2
3		86216	CCE	Meißner, Rüdiger	2271	Germany 1	½	½		½	.	½	½	½	½	5		3.5	0	10.75	1	3
4		810272		Flew, Nick	2000	Wales 2	½	½	½		0	½	½	½	½	4½	3½	3.5	0	10.75	0	4
5		241683	CCE	Rodolfo Masera, Luca	2290	Italia Azzurra	½	½	.	1		.	.	.	1	5		3	2	9.75	4	5
6		90161	IM	Maurer, Serge	2221	Canada	0	½	½	½	.		½	½	½			3	0	8.5	1	6
7		460338		Kontulainen, Hanne	2192	Finland 2	½	½	½	½	.	½		.	.	5	4	2.5	0	9.25	3	7
8		421093	IM	Nowak, Ireneusz	2275	Poland	½	.	½	½	.	½	.		.			2	0	7.5	4	8
9		884012	CCE	Quanzaah, Emmanuel	2092	Africa	0	½	½	½	0	½	.	.		5		2	0	6.75	2	9

*A norm is a high-rating performance, the exact requirements of which have varied over time and depend on the title being sought.

Knight v Bishop (part 15)

Black has just captured on d8 in Ayodeji Jeje (187) - Spanton (174), Sheffield 2017. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

Rival pawn-majorities usually favour the bishop, but Black's majority is damaged by including doubled e pawns. In addition, Black has two isolated pawns, and the bishop's scope is hampered by the e4 pawn. However, Black has more space in the centre, and the bishop currently dominates the knight. Stockfish12 gives White a slight edge, but Komodo11.01 reckons the position is equal.

25.Ke2 Kd7 26.Kd2 Kd6 27.Ne2 f5!?

Another pawn on a light square, but it frees the bishop from having to be able to defend e4.

28.c4 e5 29.g3 g5!?

The engines are not keen on this, preferring 29...exf4.

30.Kc3

The engines give 30.fxg5 hxg5 31.h4 with advantage for White.

30...exf4 31.exf4!?

This is the engines' choice.

31...a5 32.a3 Be8 33.Kd2 a4!? 34.bxa4

White gets a pair of protected passed pawns after 34.b4? cxb4 35.axb4, but Black's split passed pawns, supported by a bishop, are more powerful.

34...Bxa4 35.Nc3!?

Apparently blundering a pawn, but after ...

35...Bb3

(35...Bc6 is more solid) ... White has a close-to-winning resource.

*****

*****

*****

*****

*****

36.Nd1??

White's position is hopeless after this.

Also bad is 36.g4? gxf4 37.gxf5 Ke5 38.f6 Bxc4.

But 36.fxg5 hxg5 37.g4! fxg5? 38.Nxe4+ wins for White, according to the engines. Black therefore has to play 37...Ke6 (or 37...Ke5) 38.gxf5+ Kxf5 39.Ke3 Bxc4 40.Nxe4 Bd5 41.Nxc5, when 41...Bc6 seems to hold for Black despite being a pawn down. In this line the engines reckon the more prosaic 37.h4 gxh4 38.gxh4 Bxc4 39.h5 draws. 

The game finished:

36...Bxd1 37.Kxd1 g4 38.Kd2 Kc6 39.Kc3 Kb6 40.Kc2 Ka5 41.Kb3 h5 42.a4 e3 43.Kc3 Kxa4 44.Kd3 Kb3 45.Kxe3 Kxc4 46.Kd2 Kb3 47.Kc1 Kc3 48.Kd1 Kb2 49.Kd2 c4 50.Ke3 c3 51.Kd4 c2 52.Ke5 c1=Q 53.Kxf5 Qc2+ 54.Kg5 Qxh2 55.f5 Qxg3 0-1

Knight v Bishop (part 14)

Black has just captured on d6 in Spanton (168) - Larry Marden (173), Southern Counties Chess Union U180 2016. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

This is another ending arising from the Exchange Variation of the Spanish. Black's pawn-structure is fairly typical, but White's queenside has been severely compromised. The analysis engines Stockfish12 and Komodo11.01 reckon Black has the upper hand.

34.Kc1?!

My original notes show I had intended playing 34.a5, after which the engines continue 34...b5 35.Kc1 Be6 36.Nf4 c4 37.Kd2, when Stockfish12 has Black winning but Komodo11.01 gives Black just a slight edge.

34...a5 35.f4 c4 36.Nb2?!

The engines much prefer 36.Nf2, albeit strongly favouring Black.

36...b5 37.Kd2 Kc5

Even better, according to the engines, is 37...g5!?, eg 38.hxg5 fxg5 39.fxg5 Ke5 40.Ke3 Bg6, or 38.g3 gxh4 39.gxh4 bg6 40.Ke3 f5!? 41.e5+ Kd5 42.Kf3 Bf7 43.Ke3 Bh5 and 44...Ke4 etc.

38.g4?!

The engines prefer remaining flexible on the kingside, eg 38.Nd1 g5!? 39.g3 Bg6 40.Nf2 gxh4 41.gxh4 Kd6, when they reckon Black has the upper hand but is not (yet) winning.

38...h6?!

The engines still very much like pushing the g pawn, eg 38...g5!? 39.hxg5 fxg5 40.fxg5 (40.f5 Kd6 41.Ke3 Ke5 42.Kf3 h5 looks very good for Black) Kd6, after which the white kingside pawns are extremely weak.

39.g5?

This makes it easy, but Black is also winning after 39.Nd1!? Kd6, according to the engines. However it is still a game.

39...fxg5 40.fxg5 hxg5 41.hxg5 Kd6 42.Ke3 Ke5 43.Kf3 Be8 44.Ke3 Bd7 45.Kf3 Be6 46.Ke3 Bg4 0-1

The final position is a particularly horrible zugzwang.

Knight v Bishop (part 13)

White has just captured on f3 in Pedro Ramon Martinez Reyes (2246) - Spanton (1995), Gibraltar 2014. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

White is winning, according to Stockfish12; has a slight edge, according to Komodo11.01. The latter verdict is probably correct.

28...Kf7 29.h4 g6 30.Kg4 h5+ 31.Kf3 Be7 32.g3 e5!?

This is the engines' choice. It puts a central pawn on the same colour-complex as the bishop but opens a way for the black king to advance.

33.Nc3 Ke6 34.Ke4 Bc5 35.Nd1 Kf6 36.Ne3 Ke6 37.c3 b6 38.d4!?

Stockfish12 strongly dislikes this, but it is hard to see how else White makes progress.

38...exd4 39.cxd4

Now White is the player with the isolani, but White has created a 3-2 kingside majority.

39...Be7 40.Nc4 Bf6?!

The engines prefer 40...Bd6, and if 41.f4, which seems a good answer to the text, 41...c5 42.d5+ Kf6 43.Ne5 Bxe5 44.fxe5+ Ke7 appears to hold, although Komodo11.01 for a long time reckons White is winning.

41.Ne5?!

Almost certainly stronger is 41.f4, which threatens the winning 42.Ne5 Bxe5 43.fxe5! (43.dxe5 only draws), which is a winning pawn-ending thanks to the white king having an invasion route via f4. The engines reckon Black's best chance is 41...c5, but the simple reply 42.dxc5 wins as 42...bxc5 is met by 43.Ne5, when either the g6 pawn falls or White gets a passed pawn after 43...Bxe5 44.fxe5.

41...Bxe5 42.dxe5

White is winning, according to Komodo11.01; has a slight edge, according to Stockfish12. This time Stockfish12 is probably correct.

42...c5 43.f3 b5 44.g4

Komodo11.01 reckons White wins with 44.a3, meeting Stockfish12's 44...a6! with 45.g4, but then Stockfish12's 45...c4! 46.bxc4 hxg4 draws, as Komodo11.01 comes to see after 47.fxg4 Kd4 48.c3 Kxc3 Kxe5. In this line, the try 47...h5 48.Kd3 gxh5 49.gxh5 Kxe5 50.Kxc3 is also a draw.

44...hxg4?

44...c4 had to be played, after which 45.bxc4 hxg4! 46.fxg4 bxc4 is drawn, as in the second part of the previous note.

45.fxg4 a5?!

Pushing the c pawn no longer draws as White can ignore it and instead create a second passed pawn of his own, eg 45...c4 46.h5 c3 47.Kd3 gxh5 (47...b4 48.h6 Kf7 49.e6+!) 48.gxh5 Kxe5 49.b4 Kf5 50.Kxc3 Kg5 51.Kd4 Kxh5 52.Kc5 Kg4 53.Kxb5 Kf4 54.Kc6 Ke5 55.Kxc7 Kd5 56.Kb7 Kc4 57.a3 etc. However, it may be it is easier for White to go wrong than is the case after the text.

46.h5 gxh5 47.gxh5 c4 48.bxc4 b4 49.h6 1-0

Knight v Bishop (part 12)

Black has just captured on d4 in Spanton (2026) - Anton Wallis (2200), Liechtenstein 2005. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

Rival pawn-majorities usually favour the bishop, but the analysis engines Stockfish12 and Komodo11.01 reckon the position is equal. I guess what they do not like about the black set-up is that the bishop is obstructed by the semi-fixed d pawn.

27.Kd3 Bc1

The engines prefer centralisation with 27...Kd6!?, one point being 28.Kxd4?! is met by 28...Bg7+ and 29...Bxb2.

*****

*****

*****

*****

*****

28.Kxd4?!

This exchange of pawns almost certainly favours Black. The engines reckon White has a small edge - very small, according to Komodo11.01 - after 28.b3. Also possible is 28.b4!?

28...Bxb2+ 29.Kc5

AW said afterwards he thought this was wrong, but it is the engines' choice.

29...Kd7 30.Kb4?!

I cannot recall what I was hoping to achieve with this retreat. It was surely time to activate the knight by 30.Nd2 or 30.Ng3.

30...Kd6 31.Ng3 Bd4

*****

*****

*****

*****

*****

32.Ne2?!

This is the engines' second choice, but they much-prefer 32.Ka5!?, and if 32...Bb6+ then 33.Ka6!?, which they reckon is dead-equal. I doubt if this sequence would ever have occurred to me.

32...Bb6

The problem with White's last move is that Black is now well-placed to advance his queenside pawns.

33.Nf4 Ke5 34.Ne2 Kd6 35.Nf4 Be3?!

The logical 35...a5+ is good for Black.

36.Ne2?

Correct is 36.Nd3, the point being that if, as in the game, 36...Bd2+ 37.Kb3, Black cannot advance with 37...Kc5 or 37...Ke5. And 36...a6?, intending ...c5, is bad because of 37.Ka5.

36...Bd2+ 37.Kb3?!

Probably better is 37.c3, although the engines much prefer Black.

37...Kc5 38.Ka3?

AW could not understand why I played this ground-giving move; neither can I. However the position is difficult for White in any event, eg the engines give best play as 38.c3 Be3 39.Ng3 h5, when Black is winning (Stockfish12) or at least has the upper hand (Komodo11.01).

38...Kc4 39.Kb2 a5 40.Kb1 a4 41.Kb2 b4 42.Kb1 c5 43.Kb2 Be3 44.Kb1 a3 45.Ka1 Bd4+ 46.Kb1 Bb2 47.f4 Kb5 48.g4 c4

Position after 48...c4. How should White proceed?

*****

*****

*****

*****

*****

49.c3

This is the engines' choice but it seems White could continue to mark time, for example with 49.Ng1!? The engines' main line then runs 49...Kc5 50.Nf3 Bc3 51.Kc1, which they reckon is winning for Black. However after 51...Bd4 52.Kb1 Be3 53.f5 gxf5 54.gxf5! Kd6 55.Ka1 there seems no way for Black to make progress, eg 55...Ke7 is met by 56.c3! bxc3 ( 56...b3 57.Ne5) 57.Kb1 Kd6 58.Kc2, when White seems to be holding.

*****

*****

*****

*****

*****

49...b3

Not 49...bxc3? 50.Kc2 Kb4? (50...Kc5 is dead-equal, according to the engines) 51.Nd4, when White is winning, eg 51...Kc5 52.Nf3 h6 53.h4 Kd6 54.h5 gxh5 55.gxh5 Kc5 56.e5 Kd5 57.f5 Kc5 58.f6 exf6 59.f6! Kd6 60.Ne5!

Even worse is 49...Bxc3?? 50.Nxc3 bxc3 51.Kc2 with a winning pawn-ending for White.

50.f5 Kc5 51.fxg6 fxg6 52.axb3?!

This hastens the end, but White is lost anyway.

52...cxb3 53.e5 Kd5 54.Nf4+ Kc6 55.Ne2 g5 56.Nd4+ Kd5 57.e6 Kd6 58.Nxb3 Bxc3 59.Nc5

Or 59.Ka2 Bb4 60.Nd4 Kd5 etc.

59...Bf6

Not 59...Bb4? 60.Ne4+.

60.Nd7 Be7 61.Kc2 Kxe6 62.Nb6 Ke5 63.Nc4 Kf4 0-1

Knight v Bishop (part 11)

Black has just captured on e5 in Spanton (2112) - M Soobik (2257), IECG (correspondence) 2005. Who stands better, and by how much?

*****

*****

*****

*****

*****

*****

*****

*****

*****

*****

This ending is another arising from the Exchange Variation of the Spanish. White, as usual, has the better pawn-structure, although having all the white kingside pawns on light squares means the bishop has targets. In addition, most black pawns are on dark squares, which means they do not impede 

the bishop, and the black king is centralised, while both white pieces are on the back rank.

The analysis engines Stockfish12 and Komodo11.01 reckon the position is equal.

Nevertheless this is one of those positions where, unless Black has an immediate win, it is only White who, thanks to the knight's flexibility compared with the bishop's monochrome character, can realistically hope to generate winning chances with normal play.. 

32.Ng2 Be8 33.Kd2 g6 34.h6!?

Safer is 34.hxg6, but I felt every exchange of pawns would bring a draw nearer, and after the text it is possible Black will have problems defending h7.

34...g5 35.c3 Bd7

Black is not able to directly go after the h6 pawn, eg 35...Ke6 36.Ne3 Kf7? fails to 37.Nd5 as 37...Kg6 38.Nxc7 Bc6 (or 38...Bf7) can be met by 39.Nd5.

36.Ke3 c5 37.Kf2 b5 38.Kg3 Be6

MS offered a draw.

If 38...Ke6, trying to go after the h6 pawn, the engines point out 39.a4!, when White is very much in business, eg 39...bxa4 40.Ne3 Bb5 41.f4, when Black is a pawn up but has four isolated pawns and massive problems holding his position together.

39.Ne3 Bf7 40.Nf5 Be8 41.Ng7 Bd7 42.Nf5 Be8

Not 42...Bxf5 43.gxf5, eg 43...a4 44.Kg4 b4 45.Kg3 Kd6 46.f4 gxf4+ 47.Kxf4 Ke7 48.e5 etc.

43.Ne7

The knight has got into Black's position, but the game is dead-equal, according to the engines.

Black now sets a little trap.

43...Bf7!? 44.Nc6+ Kd6

*****

*****

*****

*****

*****

45.Na7

Not 45.Nxa5? Kc7, after which the knight cannot escape.

45...Be8 46.Nc8+ Ke5 47.Nb6 Bf7 48.Nc8 Be8 49.a4!? b4!?

Also equal is 49...bxa4 as Black still cannot make a passer from his queenside pawn-majority. After the text, however, the a4 pawn remains a target for the bishop.

50.Nb6 Bf7!?

Setting another trap.

*****

*****

*****

*****

*****

51.Kf2

51.Nd7+ Ke6 (51...Kd6? 52.Nxf6) 52.Nxc5+? loses to 52...Kd6 as the knight falls, eg 53.Nb7+ Kc7 (53...Kc6?? 54.Nd8+) 54.Nxa5 bxc3 55.bxc3 Kb6 56.Nxc4+ Bxc4, when White has three pawns for the bishop but is completely lost.

51...Be6 52.Ke3 Kd6 53.f4 Kc6!?

53...Bxg4 54.Nxc4+ Kc7 (only move) 55.fxg5 fxg5 56.Nxa5 is better for White, according to Komodo11.01 at first, but the engine soon agrees with Stockfish12 that Black has more-or-less full compensation for being a pawn down.

53...gxf4+ 54.Kxf4 Ke7 is also equal, according to the engines.

54.f5

The drawish nature of the position can be gauged from the fact that 54.Nd5!? Bxd5 55.cxd5+ Kxd5 is still equal, according to the engines, despite Black having won a pawn.

54...Bf7

Not 54...Kxb6? 55.fxe6 Kc7 56.e5! fxe5 57.Ke4 Kd6 58.Kf5 etc.

55.Nc8

Position after 55.Nc8

*****

*****

*****

*****

*****

55...Kd7

55...Kc7?! seems to lose to an amazing line from the engines, viz 56.Ne7 Kd6 57.Ng6! Bg8 58.Kf3! Bf7 59.e5+! fxe5 60.Nf8 Bg8 61.f6! Bd5+ 62.Ke3 Bg8 63.f7! Bxf7 64.Nxh7, when Black is temporarily a pawn up but will lose thanks to White's passed h pawn. This line is not all forced, but appears convincing.

56.Na7 Kd6 57.Kf3 Bg8 58.Kf2 Bf7 59.Ke3

*****

*****

*****

*****

*****

59...bxc3?

59...Bg8 60.Nb5+ Ke5 (only move) 61.cxb4 cxb4 62.Nd4 Kd6 63.Nf3 Ke7 64.Kd4 Kd6 65.Nd2 c3! 66.bxc3 bxc3 67.Kxc3 Ke5! 68.Nc4+ Bxc4 69.Kxc4 Kxe4 70.Kc5! Kf4 71.Kd6 Kxg4 72.Ke6 Kh5 73.Kxf6 g4 74.Ke5 g3 75.f6 g2 76.f7 g1=Q 77.f8=Q is a long drawing line given by the engines. Even if there is a flaw in their analysis, Black has to go down this line as the text loses.

60.bxc3 Bg8 61.Nb5+ Ke5

So far, as in the drawing line given in the previous note, but the exchange on c3 means White has an extra, winning, option.

62.Na3 Bf7 63.Nb1 Kd6 64.Nd2 Bg8 65.Nf3 Ke7

Black has had to mark time while the knight redeploys ready for pushing the e pawn.

66.Ke2!

The immediate 66.e5 only draws after 66...Bd5 67.exf6+ (67.e6 Bc6 is also a draw) Kxf6 68.Nd2 Ke5.

66...Bf7

66...Kd6 loses to 67.e5+, eg 67...fxe5 68.Nxg5 Ke7 69.Ke3 Kd6 70.Ne4+ Kc6 71.Nf6 etc.

67.e5 Bxd5 68.exf6+ Kxf6 69.Nd2 Kf7

If 69...Ke5, we reach the same position as in the line with 66.e5, except that here White has the move 70.Ke3, after which Black is in zugzwang, eg 70...Bf7 71.Ne4 or 70...Kf6 71.Ne4+.

70.Ke3 Bc6 71.Ne4 Ba4 72.Nxg5+ Kg8 73.Kd2 Bc6 74.Kc1 Bd5 75.Kb2 Ba8 76.Ka3 Bc6 77.f6 Be8 78.Ne4 Bd7

78...Kf7 79.Nd6+ Kf8 80.Nxe8 Kxe8 81.g5, or 79...Kxf6 80.Nxe8+ Kg5 81.Ka4 Kh6 82.Nf6 Kg5 83.Nxh7+.

79.g5 1-0

Material is level in the final position, but Black is lost. The engines' main line runs 79...Bf5 80.Nd6+ Bc2 81.Nxc4, when one plausible continuation runs 81...Kf7 82.Kb2 Bd1 83.Nd6+  Kg8 84.f7+ Kf8 85.g6 etc.