SF and Sting-sf knowledge missing

[Vinvin]

Here, the eval should be draw, because after 1..e2?! -> Bg4! 2.e1Q h3! and black cannot mate.
The score (+6.50) doesn't change from depth 12 to depth 75 ??!!
[d]8/8/8/5Bp1/7k/4p3/5PKP/8 b - - 0 1

This lack knowledge prevents the engines to find the move Rb1!! in this position :
[d]1R6/8/8/5bp1/4p2k/8/B1p2PKP/8 w - - 0 1

In this position, engines should play the only winning move ...g4!!, too :
[d]8/8/8/5Bp1/7k/7P/4pPK1/8 b - - 0 4

[Ralf Müller]

I wonder if there exists an engine who an solve this fortresses.

[hgm]

It is, in any case, a fortress that would be easily recognized by the algorithm I proposed:

The trigger is again "single piece and no passers". Further testing then reveals the black King is cut off from the white King, (in fact it cannot move at all here), all white pieces are protected, and the Pawn protected by only a King is not reachable by the black King.

Issuing the "fortress challenge" then allows black to place the Queen (his only mobile piece) anywhere he wants, but white will then always have a move to another position that also satisfies the fortress condition. So black cannot smoke out the white King from the fortress before the search reaches a leaf, and there the fortress will be evaluated as a draw (e.g. as +0.9).

[Eelco de Groot]

I do not see any increase in score anymore in the second position after check with 1.Rh8 so that seems to me also a fortress, then there is no need to sac the rook. Or is there another win?

If 1. Rh8+

[d]7R/8/8/5bp1/4p2k/8/B1p2PKP/8 b - -

Engine: Sf20150607_015 MOD MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

47/100 6:02 +4.34 1...Kg4 2.Rf8 c1Q 3.h3+ Kf4 4.Be6 Ke5
5.Bxf5 Qa3 6.Rf7 Qf3+ 7.Kg1 e3
8.Re7+ Kf6 9.Rxe3 Qd1+ 10.Kg2 Kxf5
11.Rg3 Qd5+ 12.Kg1 Qc6 13.Kf1 Qh1+
14.Ke2 Qe4+ (1.578.815.884) 4351

[Dirt]

I wonder if there exists an engine who an solve this fortresses.

If you are taking about the first position, Stockfish 6 does evaluate it as under 1.00, at least with Syzygy bases. It took 83 plies on my machine.

[Vinvin]

I do not see any increase in score anymore in the second position after check with 1.Rh8 so that seems to me also a fortress, then there is no need to sac the rook. Or is there another win?

If 1. Rh8+

7R/8/8/5bp1/4p2k/8/B1p2PKP/8 b - -

Engine: Sf20150607_015 MOD MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

47/100 6:02 +4.34 1...Kg4 2.Rf8 c1Q 3.h3+ Kf4 4.Be6 Ke5
5.Bxf5 Qa3 6.Rf7 Qf3+ 7.Kg1 e3
8.Re7+ Kf6 9.Rxe3 Qd1+ 10.Kg2 Kxf5
11.Rg3 Qd5+ 12.Kg1 Qc6 13.Kf1 Qh1+
14.Ke2 Qe4+ (1.578.815.884) 4351

Interesting, Rh8 seems draw !
After 1.Rh8+ Kg4 2. Rf8 c1=Q 3. h3+ Kf4 4. Be6 Ke5 5. Bxf5 Qa3 6. Rf7 Qf3+ 7.
Kg1 e3 8. Re7+ Kf6 9. Rxe3 Qxf5 10. h4 *
That's a fortress !

But the point still valide for the 2 other positions :
For the 1st eval is wrong, 6.50 instead of 0.00.
3rd position, without EGTB, a lot of engines give the same eval for h3 (losing after g4) and Bg4 (draw).