Another G/15 on FICS, I play Black, this time I found it. Interesting that first 3 moves I wanted to play were right, then 4th move was actually leading only to perpetual, but there is a winning move here as well. I didn’t get there because he played 3rd move differently, it was move losing quickly. I think OTB I would see the right continuation, maybe straight from the beginning.
Can you see the both lines, winning and perpetual? Just in case, the answer is in the comment. Black to move:

April 3, 2009 at 4:23 pm
32. … Ne3! 33. fxe3 Bxe3+ 34. Kh2 Bf2 35. Qf1 (White played 35. Be5)
perpetual – 35. … Qg3+ 36. Kh1 Qh4+ 37. Kg2 Qg3+
win – 35. … Bg3+ 36. Kg1 Bh2+ 37. Kxh2 Qxf1
April 3, 2009 at 9:42 pm
I looked at 32. Nxf variation, which I spotted right away (also took a much shorter look at 32…Ne3 afterwards). I can see Black winning the f-pawn in the Nxf variation but that is it (because 32…Ng5 opens up a lot of threats)
In your variation Ng5 leads to Qf4+, followed by BxQ. I didn’t look at Bf2 because I want to go out and water before it gets dark, but that’s nice you saw that you had time to play it.
April 3, 2009 at 10:15 pm
After Bf2, it is a stock combo. G15 probably prompted you to make/find the biggest sac because it could either win or perpetual (which saves on time).
April 4, 2009 at 12:21 am
linuxguy – I spent 5 min 25 sec checking this combo. I made sure that there are no counter-threats like Ng5. I actually thought that I win a queen after 32. … Ne3 33. fxe3 Bxe3+ 34. Kh2 Bf2 35. Qf1 Qg3+ 36. Kh1 Qh4+ 37. Qh3, not seeing that king can go to g2. Funny that in the previous game with this guy I saced Bxh7, and it wasn’t correct, I lost. So, this time I wanted to make it again, but do it right. At least, there was perpetual and a win nearby too
.
April 4, 2009 at 2:24 am
I was visualizing the Kg2 when you did that. hehe.
Pretty awesome combo you found, actually, and 5.5 minutes is excellent. Like you said last time, you have to know it’s there, that that is the position some sac must come from.
The important thing was to worry more about White getting activated, and getting the bishop down in there before White gets loose. IOW, it was really a preventative sac. I give you even more credit for realizing the endgame is not desireable in any case for Black. The h-pawn is not queenable (wrong bishop) and the other pawns are sorta “bad”, but even if that is not the case, this position simply wasn’t meant to end up an endgame for either side, too many threats.
April 4, 2009 at 11:29 am
linuxguy – thanks. I was attacking before – 30. Ne6 f3 31. gxf3 Qxf3 and just knew that Q+N is deadly. Nevertheless Crafty recommends 32. Qb2 instead of Qe1 with 0.08 estimate, no combos here. It was actually G/20, that’s why I had some time to think. I feel I should play more G/15, G/20 games.
April 6, 2009 at 12:06 am
I forgot that after Bf2 the queen has to go to f1 to defend against the Qf2+ followed by Qh2 mate.
April 6, 2009 at 12:07 pm
chesstiger – anyway this move doesn’t save White, just delays the final.