Which engine finds that one?

[Andre]

[d]R7/4kp2/P3p1p1/3pP1P1/3P1P2/p6r/3K4/8 w - -[/d]

Best move is Kc2
Re8+ is a positional draw

[MikeGL]

Sting1 detects the fortress and shifts to Kc2 after
around 4 minutes in my 32bit dual core 2Ghz

[GARGAMEL]

hallo

Van Scheltinga,T - _, Kc2 1941
R7/4kp2/P3p1p1/3pP1P1/3P1P2/p6r/3K4/8 w - - 0 1

Analysis by Ivanhoe B46fE.02 x64:

1.Ta7+ Ke8 2.Kc2 Tf3 3.Tc7 a2 4.Kb2 a1T 5.Kxa1 Ta3+ 6.Kb2 Txa6
+/= (0.54) Tiefe: 6 00:00:00 7kN
+/= (0.39) Tiefe: 8 00:00:00 12kN
1.Te8+ Kxe8 2.a7 a2 3.a8D+ Ke7 4.Db7+ Kf8 5.Dc8+ Kg7 6.Dc1 Tb3 7.Da1 Tf3 8.Kc2 Txf4 9.Kc3 Tf2 10.Kb3 Th2 11.Df1 Tf2 12.Dg1 Th2 13.De1 Tf2 14.Dc1 Th2 15.Kb4 Th8 16.Da3 Th2 17.Da7 Tb2+ 18.Kc3 Tf2 19.Da4 Th2 20.Dd1 Tf2 21.Dc1 Th2 22.Kb3 Tf2 23.Dh1 Th2 24.Dg1 Tf2 25.Dh1
+/= (0.47 !) Tiefe: 9 00:00:00 23kN
+- (1.63) Tiefe: 34 00:01:23 1246mN
1.Kc2!! a2 2.Kb2 Ta3 3.Ka1 Kd7 4.a7 Kc6 5.Tc8+ Kb7 6.Tf8 Kxa7 7.Txf7+ Kb6 8.Tf6 Tf3 9.Txg6 Txf4 10.Txe6+ Kc7 11.g6 Tg4 12.Ta6 Kb7 13.Tf6 Kc8 14.Kxa2 Kd7 15.Ta6 Ke7 16.Ta7+ Kf8 17.g7+ Kg8 18.Kb3 Te4 19.Kb4 Txd4+ 20.Kc5 Tc4+ 21.Kxd5 Tc1 22.e6 Td1+ 23.Kc5 Tc1+ 24.Kd6 Td1+ 25.Ke5 Te1+ 26.Kf6 Tf1+ 27.Kg6 Tg1+ 28.Kf5 Tf1+ 29.Kg5 Tg1+ 30.Kf4 Tf1+ 31.Kg4 Tg1+ 32.Kh5 Te1
+- (1.71 !) Tiefe: 34 00:02:59 5047mN
+- (2.80) Tiefe: 34 00:12:33 16301mN

regards Anton

[Joerg Oster]

[d]R7/4kp2/P3p1p1/3pP1P1/3P1P2/p6r/3K4/8 w - -[/d]

Best move is Kc2
Re8+ is a positional draw

Stockfish 2.3.1 (1 GB Hash, 8 Threads) with the 50-moves-rule modification by Ryan Taker finds it after 36 sec at depth 33. Here the output after 2 min of analysis:

Code: Select all

info depth 38 seldepth 89 score cp 333 lowerbound nodes 1301905357 nps 11518432 time 113028 multipv 1 pv d2c2 a3a2 c2b2 h3a3 b2a1 e7d7 a6a7 d7e7 f4f5 e6f5 e5e6 f5f4 a8h8 e7e6 a7a8q a3a8 h8a8 e6f5 a8a7
info depth 38 currmove d2c2 currmovenumber 1
bestmove d2c2 ponder a3a2

[Jouni]

Well-known from a dozen testsuites. Rybka took about 6 meganodes

1.Kc2 a2 2.Kb2 Ra3 3.Ka1[] Kd7 4.a7
+- (1.53 ++) Depth: 24 00:00:40 6406kN, tb=52

[ernest]

Analysis by Ivanhoe B46fE.02 x64

Houdini seems to have a problem,
but indeed (1st try), depth 32!

Analysis by Ivanhoe B46fC w32:

Code: Select all

.............
1.Re8+ Kxe8 2.a7 a2 3.a8Q+ Ke7 4.Qa7+ Kf8 5.Qc5+ Kg7 6.Qc1 Rb3 7.Qa1 Rf3 8.Kc2 Rxf4 9.Kc3 Rf2 10.Kb3 Rh2 11.Qf1 Rf2 12.Qc1 Rh2 13.Qa1 Rf2 14.Kc3 Rh2 15.Qc1 Kf8 16.Qa3+ Kg7 17.Qa4 Rg2 18.Qd1 Rf2 19.Kb3 Rh2 20.Qc1 Rg2 21.Qe1 Rf2 22.Ka3 Rh2 23.Qf1 Re2 24.Qa1 Rf2 25.Qc1 Rg2 26.Qe1 Rh2 
  +-  (1.62)   Depth: 32   00:02:12  389mN
1.Kc2 
  +-  (1.70 !)   Depth: 32   00:05:54  1055mN
1.Kc2 
  +-  (1.78 !)   Depth: 32   00:06:59  1252mN
1.Kc2 
  +-  (1.90 !)   Depth: 32   00:08:02  1441mN
1.Kc2 
  +-  (2.08 !)   Depth: 32   00:09:26  1690mN

[ernest]

Rybka took about 6 meganodes

Which Rybka?
My Rybka 4.1 still likes a7 at depth 31 and 63 meganodes...

[ernest]

Sting1

The more recent Sting SF 2_VE also finds Kc2 after a few minutes.

[Jouni]

Just 4.1 MP finds it fast, but it varies a lot between tries.

[ernest]

Just 4.1 MP finds it fast, but it varies a lot between tries.

OK, verified (2nd try): indeed that SMP randomness...