So Good

Hmm after 16 hours, 12 minutes and 22 seconds it is back to 0,00 again... keep on searching

White to move

Originally posted by heinzkat
[fen]8/qb6/8/2p5/2Q5/8/8/3BKn1k w - -[/fen]

White to move

1. Kf2 Bg2 2. Bf3 Qg7 3. Qh4+ Nh2 4. Qh8 (can't allow Qd4+) - Ugly tho 😞


Mechanical from here, ...Qg5 5. Qh6 Qg8 6.Qc1+
Winning the queen after chasing black up the h file onto g4/g5

-GIN

Why not 4... Qg6?

Originally posted by heinzkat
Why not 4... Qg6?

4...Qg6 is fine, followed by 5. Qh7 Qg5

Good catch.
Same principles still apply.

Try giving this one to the engines (not sure how to post the diagram, so I'll tell you where the pieces are):


White: Kf6, Rf2, Pd5, Qb5

Black: Ke4



White to move and mate in three.
Most engines only find mate-in-four

Shredder only finds the mate-in-three after 1 minute 33 seconds!

Thanks for posting the diagram 🙂

Originally posted by heinzkat
[fen]8/8/5K2/1Q1P4/4k3/8/5R2/8 w - -[/fen]

My custom Stockfish finds it in less than a second. Very good.
Rybka 32 didn't find it at all. (5 minutes, I'm not waiting longer)
Rybka 64 on 4 cores found it in less than a second as well, with only 2 tree tries.
(I'm sure Vasik would be excited about that!)


Stockfish-16x-NowakowskiEdition:
1 00:00 63 0 +96.81 Kf6g6
2 00:00 84 0 +96.81 d5d6 Ke4e3 Rf2e2+ Ke3f3
2 00:00 178 0 +96.86 Rf2d2 Ke4e3 Qb5e2+ Ke3f4
3 00:00 932 0 +96.86 Rf2d2 Ke4f4 Qb5d3
3 00:00 1.321 0 +101.82 d5d6 Ke4e3 d6d7 Ke3xf2 d7d8Q Kf2e3 Qb5d3+ Ke3f2
4 00:00 2.591 0 +101.87 d5d6 Ke4e3 d6d7 Ke3xf2 d7d8Q Kf2e3 Kf6g5
5 00:00 7.157 0 +102.08 d5d6 Ke4e3 d6d7 Ke3xf2 d7d8Q Kf2g3 Qb5e2 Kg3h4
6 00:00 20.690 1.379.333 +M6 d5d6 Ke4e3 d6d7 Ke3xf2 d7d8Q Kf2g3 Qb5e2 Kg3h4 Qe2h2+ Kh4g4 Qd8d1+
6 00:00 30.233 975.258 +M5 Rf2e2+ Ke4d4 d5d6 Kd4c3 d6d7 Kc3d4 d7d8Q+ Kd4c3 Qd8d2+
7 00:00 54.123 1.151.553 +M5 Rf2e2+ Ke4d4 d5d6 Kd4c3 d6d7 Kc3d4 d7d8Q+ Kd4c3 Qd8d2+
8 00:00 82.755 1.334.758 +M5 Rf2e2+ Ke4d4 d5d6 Kd4c3 d6d7 Kc3d4 d7d8Q+ Kd4c3 Qd8d2+
8 00:00 91.406 1.171.871 +M4 Qb5c4+ Ke4e3 Rf2g2 Ke3f3 Qc4e2+ Kf3f4 Rg2g4+
9 00:00 97.895 1.255.064 +M4 Qb5c4+ Ke4e3 Rf2g2 Ke3f3 Qc4e2+ Kf3f4 Rg2g4+
10 00:00 107.072 1.372.717 +M4 Qb5c4+ Ke4e3 Rf2g2 Ke3f3 Qc4e2+ Kf3f4 Rg2g4+
10 00:00 122.234 873.100 +M3 Rf2g2 Ke4d4 Rg2g3 Kd4e4 Qb5c4+
11 00:00 123.117 879.407 +M3 Rf2g2 Ke4d4 Rg2g3 Kd4e4 Qb5c4+
....Continued
100 00:00 342.412 535.856 +M3 Rf2g2 Ke4d4 Rg2g3 Kd4e4 Qb5c4+

FEN: 8/8/5K2/1Q1P4/4k3/8/5R2/8 w - - 0 1

Rybka 3 Human 1-cpu w32:

1 05:01 11.467.038 90.459 +M4 Qb5e2+
2 05:01 11.582.799 90.306 +M4 Rf2c2
3 05:01 11.525.397 90.342 +M4 Qb5c5
4 05:01 11.433.206 90.342 +M4 Qb5b4+
5 05:01 11.390.392 87.950 +M4 Qb5c4+

Originally posted by Nowakowski
My custom Stockfish

Nowakowski's Improvement Program?

Originally posted by heinzkat
Nowakowski's Improvement Program?

nope, my ICCF assistant.

Originally posted by Nowakowski
My custom Stockfish finds it in less than a second. Very good.
Rybka 32 didn't find it at all. (5 minutes, I'm not waiting longer)
Rybka 64 on 4 cores found it in less than a second as well, with only 2 tree tries.
(I'm sure Vasalich would be excited about that!)


Stockfish-16x-NowakowskiEdition:
1 00:00 63 0 +96.81 Kf6g6
2 00:0 ...[text shortened]... Qb5c5
4 05:01 11.433.206 90.342 +M4 Qb5b4+
5 05:01 11.390.392 87.950 +M4 Qb5c4+

That's impressive. I had a friend analyze the position with Fritz (I think Fritz 11) and it couldn't find it either, nor could Chessmaster.

Originally posted by Der Bayer
That's impressive. I had a friend analyze the position with Fritz (I think Fritz 11) and it couldn't find it either, nor could Chessmaster.

if interested i can give you a copy of my SF alongside arena so you can toy with it
yourself.

Originally posted by Nowakowski
if interested i can give you a copy of my SF alongside arena so you can toy with it
yourself.

I appreciate the offer, but for simple analysis purposes I think my Shredder is adequate. It's only in tricky puzzles like this one where it doesn't do well.

Just out of curiosity though, have you ever played your rybka against your customized engine to see which one comes out on top?

Originally posted by Der Bayer
I appreciate the offer, but for simple analysis purposes I think my Shredder is adequate. It's only in tricky puzzles like this one where it doesn't do well.

Just out of curiosity though, have you ever played your rybka against your customized engine to see which one comes out on top?

I haven't, but I'm sure Rybka would tear it apart. Rybka sits around 2900 elo in well
balanced systems, whereas I don't think my stockfish would break 2780 in well
balanced systems.

My Stockfish is useful for ending analysis, its development was specific for enormous
tree's and huge hash files for increased horizon scalability. Rybka is built with positional
scoring on tier based logic. My SF would get slaughtered.

However, if compared to one another in creating ending tablebases, my SF can run a
6 piece positional solvent in 35% less time.

ADD: This is a good site showing you how the engines stack up, note that the SF
which is in 2nd is the standard compiled version, and is probably stronger than mine
in tournament conditions.
http://www.compchess.de/html/elostat.html


-GIN