zullil wrote:zullil wrote:Spacious_Mind wrote:
SF8 does not see Bg5 as best move. From someone else's test I know it only sees it as best move at 59 ply which you are confirming.
I have not tested Stockfish 8 on this position. As I
showed, the latest Cfish found Bg5 at depth 45, after about 20 minutes. Of course, since it used 20 threads, that search was highly non-deterministic. I have no idea how many nodes a single-threaded Cfish would need to find this move.
FWIW, here is a 16-thread search in which the latest SF-dev finds Bg5 at depth 52 (after almost 2.5 hours).
Code: Select all
info depth 52 seldepth 75 multipv 1 score cp 101 lowerbound nodes 193326637735 nps 23015160 hashfull 999 tbhits 0 time 8399969 pv c1g5
I went back and checked to see where I read 59 ply search with SF8 to find Bg5. My bad it was actually 51 ply on a Xeon E3-1225 v2 4 core 3.2 GHz and it took 45 hours to find it LOL. Amazing how quickly Hardware changes, I should have let my laptop run another 30 minutes and probably would have found in in less than 4.1/2 hours.
Anyway a question for all the learned programmers and gentlemen in this forum. What would it take for a top engine to consider the move Bg5?
Because when you look at a search 20 ply deep as shown below, you see that Bg5 is 7th choice with -0.51 whereas the best move is evaluated at -0.27 which is a difference of only 24 centipawns.
20/43 00:55 420,117k 7,558k -1.11 1.Rd1 f5 2.Qf3 d5 3.Nd2 Nc5 4.Nb3 Nxb3 5.axb3 Bb5 6.Re1 Qg8 7.Re7 Rf7 8.Rxf7 Qxf7 9.Bh3 Nd7 10.Qxf5 Nf6 11.Qg6 Qxg6
20/43 00:55 420,117k 7,558k -0.98 1.Qf3 d5 2.Nd2 Nd7 3.b4 Rad8 4.Bb2 Qc2 5.Bc3 Nd6 6.a4 f6 7.Rec1 Qh7 8.b5 cxb5 9.axb5 Bb7 10.Rxa7 Ne5 11.Bxe5
20/43 00:55 420,117k 7,558k -0.86 1.a3 d5 2.Nc3 Nc5 3.b4 Nd3 4.Re7 Bc4 5.Be3 Rd8 6.Rd1 Qg8 7.b5 cxb5 8.Qf5 Nc6 9.Rc7 Rac8 10.Rxc8 Rxc8 11.Nxd5 Nc5 12.Bf3 Qf8 13.Bg4 Ne6
20/43 00:55 420,117k 7,558k -0.81 1.Re3 d5 2.Nc3 Nd6 3.Re7 Bc8 4.Qf4 Nf5 5.Rc7 Na6 6.Rxc6 Bd7 7.Bxd5 Rad8 8.Be4 Bxc6 9.Bxf5 Qg8 10.Be3 Nc5 11.Qc7 Bf3 12.Bxc5 bxc5 13.Qxa7 Rb8 14.Qxc5 Rfd8
20/43 00:55 420,117k 7,558k -0.80 1.a4 d5 2.Nc3 Nc5 3.b4 Nd3 4.Rd1 Bc4 5.b5 Rd8 6.Bf4 cxb5 7.Bc7 Re8 8.Bxd5 Bxd5 9.Nxd5 Na6 10.Qf3 Nb2 11.Rd4 Nxa4 12.Qxf7 Rf8 13.Qd7 N4c5 14.Qd6 Qc2 15.Rf4 Rxf4
20/43 00:55 420,117k 7,558k -0.76 1.Qh4 d5 2.Qe7 Kg8 3.Bf4 Qf5 4.Nd6 Nxd6 5.Qxd6 Bb5 6.a4 Bc4 7.a5 bxa5 8.Rxa5 Qg4 9.b3 Bxb3 10.f3 Qc8 11.Re7 Bc4 12.g4 Re8 13.Rc7 Qd8 14.Qxd8 Rxd8
20/43 00:55 420,117k 7,558k -0.67 1.Bd2 d5 2.Bb4 Rd8 3.Nc3 Nc5 4.Bxc5 bxc5 5.Re7 Qg8 6.Qf4 Bc8 7.Qxf7 Qxf7 8.Rxf7 Na6 9.Re1 Bg4 10.Ree7 Rg8 11.Rxa7 Rxa7 12.Rxa7 Nb4 13.Na4 c4 14.Rc7 d4 15.Bxc6 Bxh5
20/43 00:55 420,117k 7,558k -0.51 1.Bg5 f5 2.Qf3 d5 3.Nf6 gxf6 4.Bxf6+ Rxf6 5.Re8+ Qg8 6.Rxg8+ Kxg8 7.Re1 Nd7 8.Re7 Nbc5 9.b4 Re6 10.Rxe6 Nxe6 11.Qxf5 Ndf8 12.Qf6 Bc4 13.Qxh6 Kf7 14.Qe3 Re8 15.Bh3 Bxa2 16.Qf3+ Ke7
20/43 00:55 420,117k 7,558k -0.40 1.Nd2 Nc5 2.Nb3 Nxb3 3.axb3 d5 4.Re7 Bc8 5.Qf4 Qf5 6.Qxf5 Bxf5 7.Raxa7 Rxa7 8.Rxa7 d4 9.Bf4 c5 10.Bd6 Rd8 11.Bc7 Rc8 12.Bxb6 Nd7 13.Ba5
20/43 00:55 420,117k 7,558k -0.34 1.Qf4 d5 2.Nd6 Nxd6 3.Qxd6 Rg8 4.Bf4 Qf5 5.Re5 Qg4 6.Bxh6 Qd7 7.Qxd7 Nxd7 8.Re7 gxh6 9.Rxd7 Kg7 10.Rc1 Bb5 11.b3 Rgc8 12.Bh3 Rd8 13.a4 Rxd7 14.Bxd7
20/43 00:55 420,117k 7,558k -0.33 1.b3 d5 2.Ba3 Rd8 3.Qf4 Nd7 4.Nd6 Nxd6 5.Bxd6 Nc5 6.Bxc5 bxc5 7.Rac1 c4 8.Bf1 f6 9.bxc4 Qg8 10.Re7 Re8 11.Rc7 Rac8 12.Rd7 Re4 13.Qf5 Bxc4 14.Bxc4 Rxc4 15.Rxc4 dxc4 16.Rxa7
20/43 00:55 420,117k 7,558k -0.30 1.b4 d5 2.Nc3 Bc4 3.Bf4 Qg8 4.b5 Na5 5.Rad1 Re8 6.a4 Re6 7.Rxe6 fxe6 8.Re1 Nd7 9.bxc6 Nxc6 10.Nxd5 Bxd5 11.Bxd5 exd5 12.Qxd7 Rc8 13.Bd6 d4
20/43 00:55 420,117k 7,558k -0.29 1.Be3 d5 2.Nf6 gxf6 3.Bd4 Qg7 4.Qf5 Nd6 5.Qf4 Nd7 6.Qxd6 Ne5 7.Bxe5 fxe5 8.Qxc6 Rad8 9.Rad1 f5 10.Rxd5 Bb7 11.Qxb7 Qxb7 12.Rxd8 Qg7 13.Rxf8+ Qxf8 14.Rxe5
20/43 00:55 420,117k 7,558k -0.27 1.Nc3 d5 2.b4 Bc4 3.Bf4 Qg8 4.Rad1 Re8 5.b5 Rxe1+ 6.Rxe1 Nc5 7.Be5 Ne6 8.bxc6 Nxc6 9.Nxd5 Nxe5 10.Rxe5 Bxd5 11.Bxd5 Rc8 12.Bxe6 fxe6 13.Kg2
So why does it take say 51 ply to revisit a 7th choice move and then decide it actually was the best move, when in reality it should have at least been seen as best at 26 ply?
Amusing as well as confounding is this position.
Sorry I don't know if Talkchess has a FEN diagram functionality but here is the FEN:
FEN: 3r2kq/p2prp1p/1p4pP/2nR4/1Q6/1B3RP1/P4PK1/8 w - - 0 1
Winning move was Rxc5.
Analysis with 20 ply again:
20/45 00:24 208,442k 8,595k -0.29 1.Rg5 Qa1 2.Rd5 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rcxd7 Rh5 12.Rf6 Rf8 13.Rxa7 Rb5 14.Rf3 Kg7 15.Rd7 h5 16.Rfd3
20/45 00:24 208,442k 8,595k -0.27 1.Rd4 Rde8 2.Rd5 Qa1 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rxa7 Rh5 12.Rxb6 Rf5 13.f4 d5
20/45 00:24 208,442k 8,595k -0.27 1.Kg1 Rde8 2.Kg2 Qa1 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rxa7 Rh5 12.Rxb6 Rf5
20/45 00:24 208,442k 8,595k -0.19 1.Rf4 Qa1 2.Rf3 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rxa7 Rh5 12.Rxb6 Rf5 13.Rbb7
20/45 00:24 208,442k 8,595k -0.18 1.Kh1 Rde8 2.Kg2 Qa1 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rdxd7 Rf8
20/45 00:24 208,442k 8,595k -0.18 1.Bd1 Qa1 2.Bb3 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rdxd7 Rf8 12.Rd5 f5 13.Rxa7 f4 14.gxf4 Rh4 15.Rdd7 Rfxf4
20/45 00:24 208,442k 8,595k -0.18 1.Qb5 Qa1 2.Qb4 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rdxd7 Rf8 12.Rd5 f5 13.Rxa7 f4 14.gxf4
20/45 00:24 208,442k 8,595k -0.17 1.Kh3 Qa1 2.Kg2 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rdxd7 Rh5
20/45 00:24 208,442k 8,595k -0.14 1.Rfd3 Kf8 2.Rf3 f5 3.Qc4 Ke8 4.Re3 Nxb3 5.Rxb3 Qf6 6.Rbd3 Re6 7.Qc7 Qe7 8.Qxa7 Re2 9.Qa4 Qe4+ 10.Qxe4+ fxe4 11.R3d4 Ke7
20/45 00:24 208,442k 8,595k -0.10 1.Kf1 Rde8 2.Kg2 Qa1 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe2 8.Qf6 Qe5 9.Qxe5 Rxe5 10.Rc3 Rh5 11.Rc7 Rxh6 12.Rdxd7 Rf8 13.Rd5 a5 14.Rb7
20/45 00:24 208,442k 8,595k -0.10 1.Bc2 Rde8 2.Bb3 Qa1 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe2 8.Qf6 Qe5 9.Qxe5 Rxe5 10.Rc3 Rh5 11.Rc7 Rxh6 12.Rdxd7 Rf8 13.Rd5 a5 14.Rb7 f5 15.Rxb6
20/45 00:24 208,442k 8,595k -0.10 1.Bc4 Qa1 2.Bb3 Rde8 3.Qc4 Rf8 4.Rd1 Qb2 5.Qf4 Nxb3 6.axb3 Rfe8 7.Rd6 Qe5 8.Qxe5 Rxe5 9.Rc3 Rh5 10.Rc7 Rxh6 11.Rdxd7 Rf8 12.Rd5 f5 13.Rxa7 f4
20/45 00:24 208,442k 8,595k -0.08 1.Rd2 Kf8 2.Qc4 f5 3.Rd5 Ke8 4.Re3 Nxb3 5.Rxb3 Qf6 6.Rbd3 Re6 7.Qc7 Qe7 8.Qxa7 Re2 9.Qa4 Qe4+ 10.Qxe4+ Rxe4 11.Kf3 Ra4 12.Rd6 Rxa2 13.Rf6 Ke7 14.Rxb6 Ra5 15.Re3+ Kf7
20/45 00:24 208,442k 8,595k -0.07 1.Qg4 Nxb3 2.axb3 Qb2 3.Qc4 Rf8 4.Qc7 Re1 5.Qc3 Qa1 6.Qxa1 Rxa1 7.Rxd7 b5 8.Rb7 a6 9.Re3 Ra3 10.Rd3 Ra5 11.Rbd7 Ra1 12.Rb7 Ra2 13.Re3 Ra1
20/45 00:24 208,442k 8,595k -0.04 1.a3 Rc8 2.Rdd3 Rf8 3.Qh4 Ree8 4.Qf4 Qe5 5.Bxf7+ Kh8 6.Qxe5+ Rxe5 7.Bc4 Rff5 8.Rxf5 Rxf5 9.Rf3 Rxf3 10.Kxf3 Na4 11.Bb5 Nb2 12.Bxd7 Nc4 13.Ke4 a5 14.Be6 Nxa3 15.Ke5
20/45 00:24 208,442k 8,595k 0.00 1.Qd2 Qa1 2.Qg5 Rde8 3.Rd1 Qb2 4.Rd2 Qb1 5.Rd1 Qe4 6.Qf6 Ne6 7.Rd5 Rc8 8.Rd1 Rce8
20/45 00:24 208,442k 8,595k 0.00 1.Rxc5 bxc5 2.Qxc5 Qe5 3.Bxf7+ Kh8 4.Qxa7 Qd6 5.Qa4 Re5 6.Qc4 Rh5 7.Qe2 Rf8 8.Bxg6 Rxh6 9.Rxf8+ Qxf8 10.Qe5+ Qg7 11.Qe8+ Qg8 12.Qe5+ Qg7
20/45 00:24 208,442k 8,595k 0.00 1.Kh2 Qa1 2.Rd1 Qe5 3.Rd5 Qa1
20/45 00:24 208,442k 8,595k 0.00 1.a4 Rf8 2.a5 Re6 3.Qb5 Rf6 4.Re3 Re6 5.Rf3
20/45 00:24 208,442k 8,595k 0.00 1.Qa3 d6 2.Qb4 Kf8 3.Qf4 Nxb3 4.axb3 Qa1 5.Rxd6 Rxd6 6.Qxd6 Qe5 7.Qd8+ Re8 8.Qd7 Re7 9.Qd8+
20/45 00:24 208,442k 8,595k 0.00 1.Qc4 Rf8 2.Qb4 Rd8
20/45 00:24 208,442k 8,595k 0.00 1.Qh4 Rde8 2.Qf4 Nxb3 3.axb3 Qb2 4.Rfd3 d6 5.Rd2 Qa1 6.Rd1 Qc3 7.R1d3 Qb2
20/45 00:24 208,442k 8,595k 0.00 1.Rdd3 Rf8 2.Qh4 Ree8 3.Qf4 Qe5 4.Bxf7+ Kh8 5.Qxe5+ Rxe5 6.Bc4 Rd8 7.Rd6 Rh5 8.Rdf6 Rxh6 9.Rf8+ Rxf8 10.Rxf8+ Kg7 11.Rf7+ Kh8 12.Rf8+
20/45 00:24 208,442k 8,595k 0.00 1.Qf4 Rf8 2.Qd6 Rfe8 3.Rxc5 bxc5 4.Rxf7 Rxf7 5.Qxd7 Ref8 6.Bxf7+ Rxf7 7.Qd8+ Rf8 8.Qd5+ Rf7 9.Qa8+ Rf8 10.Qd5+
20/45 00:24 208,442k 8,595k 0.00 1.Rd6 Qe5 2.Rd5 Qh8
20/45 00:24 208,442k 8,595k 0.00 1.Rd1 Rf8 2.Qh4 Ree8 3.Qf4 Nxb3 4.Rxd7 Nc5 5.Rxf7 Ne6 6.Rxf8+ Rxf8 7.Qd6 Rxf3 8.Qxe6+ Rf7 9.Qc8+ Rf8 10.Qe6+ Rf7
At 20 ply it ranks it 10th with exact same eval as first place Rd1. In fact if you take the same 24 centipawn difference from the previous example then in this example you should have at least 24 candidate moves that should have been equally searched deeper with nothing cut off.
Take this a step further when you search 33 ply deep you have 28 equally evaluated move choices all rated with 0.00. With a 29th choice at -0.08 within 8 centipawns of being equally as good.
33/66 57:32 33,722,153k 9,766k -1.41 1.Rh5 gxh5 2.Rxf7 Rxf7 3.Qf4 Ne6 4.Bxe6 dxe6 5.Qg5+ Qg7 6.hxg7 Rd5 7.Qe3 Rdf5 8.Qxe6 Rxf2+ 9.Kh3 Kxg7 10.Qe5+ Kg6 11.Qd6+ R2f6 12.Qd3+ Rf5 13.Qd6+ Kg7 14.Qd4+ Kg8 15.Qd8+ Rf8 16.Qd7 Ra5 17.Kh4 Rf6 18.Qd8+ Kf7 19.Qd7+ Kg6 20.Qe7 Rff5 21.Qd6+ Kg7 22.Qd7+ Rf7 23.Qd4+ Rf6 24.Qd7+ Kh6 25.Qd2+ Kg6 26.Qd7 Rf7 27.Qd6+ Kf5 28.Kxh5 Ke4+ 29.Kh4
33/66 57:32 33,722,153k 9,766k -0.08 1.g4 Rf8 2.Rd1 Qe5 3.Re3 Qf6 4.Rxe7 Qxe7 5.Qd4 Ne6 6.Bxe6 fxe6 7.Qxd7 Qf6 8.f4 g5 9.Rd6 gxf4 10.Qxe6+ Qxe6 11.Rxe6 f3+ 12.Kf2 Rf4 13.g5 Rg4 14.Re7 Rxg5 15.Rxa7 Rh5 16.Ra8+ Kf7 17.Ra7+ Kg6 18.Ra6 Rb5 19.a4 Rb2+ 20.Kxf3 Kg5 21.a5 Rb3+ 22.Kg2 Ra3 23.Rxb6 Rxa5 24.Rb7 Kxh6 25.Rf7 Rg5+ 26.Kh2 Rh5+ 27.Kg3 Rd5 28.Kg2 Kh5 29.Rg7 Rd2+ 30.Kh1
33/66 57:32 33,722,153k 9,766k 0.00 1.Rc3 Re4 2.Bc4 Qf6 3.Rf3 Qe7 4.Qc3 Ne6 5.Bb3 d6 6.Rfd3 Qf8 7.Rd1 Nc5 8.Rxd6 Rxd6 9.Rxd6 Re8 10.Bd5 Qxh6 11.Qa3 Qh5 12.Qxa7 Qf5 13.Qxb6 Ne4 14.Rd8 Rxd8 15.Qxd8+ Kg7 16.Bxe4 Qxe4+ 17.Kh2 h5 18.Qd2 h4 19.Qf4 hxg3+ 20.Kxg3 Qd5 21.a4 Qb3+ 22.Kg2 Qd5+ 23.Kg3
33/66 57:32 33,722,153k 9,766k 0.00 1.Ba4 Re4 2.Qb5 Rxa4 3.Rxc5 bxc5 4.Qxa4 Qd4 5.Qxa7 Re8 6.a4 f5 7.Qb6 Kf7 8.a5 Re1 9.Qb3+ c4 10.Qb4 Re2 11.a6 Ra2 12.a7 Rxa7 13.Re3 d6 14.Qb8 Re7 15.Rxe7+ Kxe7 16.Qb7+ Ke8 17.Qa8+ Kf7 18.Qb7+ Ke8
33/66 57:32 33,722,153k 9,766k 0.00 1.Bd1 Kf8 2.Bb3 Ke8 3.Qa3 f5 4.Qxa7 Qf6 5.Qa3 Rc8 6.Bc4 Re4 7.Bb5 Rd8 8.Qc1 Ke7 9.Qd2 Re5 10.Re3 Re4 11.Rf3
33/66 57:32 33,722,153k 9,766k 0.00 1.Rfd3 Kf8 2.Rf3 Ke8 3.Qa3 f5 4.Qxa7 Qf6 5.Qa3 Rc8 6.Bc4 Re4 7.Bb5 Rd8 8.Qc1 Ke7 9.Qd2 Re5 10.Re3 Re4 11.Rf3
33/66 57:32 33,722,153k 9,766k 0.00 1.Rd4 Kf8 2.Rd6 f5 3.Qc4 Nxb3 4.Qxb3 Qe5 5.Qa3 Qc5 6.Qb2 Kg8 7.Qb3+ Kh8 8.Qb2+ Qe5 9.Qxe5+ Rxe5 10.Ra3 Ra5 11.Rxa5 bxa5 12.Kf3 Kg8 13.Ra6 Ra8 14.Rd6 Rd8
33/66 57:32 33,722,153k 9,766k 0.00 1.Rd2 Kf8 2.Rd6 f5 3.Qc4 Nxb3 4.Qxb3 Qe5 5.Qa3 Qc5 6.Qb2 Kg8 7.Qb3+ Kh8 8.Qb2+ Qe5 9.Qxe5+ Rxe5 10.Ra3 Ra5 11.Rxa5 bxa5 12.Kf3 Kg8 13.Ra6 Ra8 14.Rd6 Rd8
33/66 57:32 33,722,153k 9,766k 0.00 1.Kf1 Rde8 2.Kg2 Qa1 3.Qf4 Nxb3 4.axb3 f5 5.Qc4 Qf6 6.Rxd7+ Kf8 7.Rxe7 Qxe7 8.Qd4 Kg8 9.Qc4+ Qf7 10.Qc6 Qe6 11.Qc3 Qe7 12.Qc4+
33/66 57:32 33,722,153k 9,766k 0.00 1.Kh1 Qa1+ 2.Kg2 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Kg1 Re4 2.Qa3 Qa1+ 3.Kg2 Nxb3 4.Qxb3 Re7 5.Qc4 Rf8 6.Rc3 Qe1 7.Qd4 Qe4+ 8.Qxe4 Rxe4 9.Rxd7 Ra4 10.a3 b5 11.Rcd3 a6 12.Ra7 Rb8 13.Rf3 Rf8 14.Rd3
33/66 57:32 33,722,153k 9,766k 0.00 1.a3 Rc8 2.Rdd3 Rf8 3.Rd1 Re2 4.Qf4 Nxb3 5.Rxd7 Qe5 6.Qxe5 Rxe5 7.Rxb3 Ra5 8.Rb7 Rc8 9.Rf3 Rf8 10.Rb3
33/66 57:32 33,722,153k 9,766k 0.00 1.Kh3 Re4 2.Qb5 Re6 3.Rxc5 bxc5 4.Bxe6 dxe6 5.Qb7 Rf8 6.Rxf7 Rxf7 7.Qc8+ Rf8 8.Qxe6+ Rf7 9.Qe8+ Rf8 10.Qe6+
33/66 57:32 33,722,153k 9,766k 0.00 1.Kh2 Qa1 2.Kg2 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Rg5 Qa1 2.Rd5 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Rf4 Qa1 2.Rf3 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Bc4 Qa1 2.Bb3 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Bc2 Rde8 2.Bb3 Qa1 3.Qf4 Nxb3 4.axb3 f5 5.Qc4 Qf6 6.Rxd7+ Kf8 7.Rxe7 Qxe7 8.Qd4 Kg8 9.Qc4+ Qf7 10.Qc6 Qe6 11.Qc3 Qe7 12.Qc4+
33/66 57:32 33,722,153k 9,766k 0.00 1.Qb5 Qa1 2.Qb4 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Rxc5 bxc5 2.Qxc5 Qe5 3.Bxf7+ Kh8 4.Qxa7 d5 5.Qb6 Qd6 6.Qd4+ Qe5 7.Qb6
33/66 57:32 33,722,153k 9,766k 0.00 1.Qg4 Nxb3 2.axb3 Qb2 3.Qf4 Rf8 4.Rfd3 Re1 5.Rd1 Re6 6.Rxd7 Rf6 7.Qe3 Re6 8.Qf4
33/66 57:32 33,722,153k 9,766k 0.00 1.Qd2 Qa1 2.Qg5 Rde8 3.Rd1 Qb2 4.Rd2 Qa1 5.Rd1
33/66 57:32 33,722,153k 9,766k 0.00 1.a4 Rf8 2.a5 Re6 3.Qb5 Rf6 4.Re3 Re6 5.Rf3
33/66 57:32 33,722,153k 9,766k 0.00 1.Qa3 d6 2.Qb4 Nxb3 3.axb3 Kf8 4.Rxd6 Rxd6 5.Qxd6 Qe5 6.Qd8+ Re8 7.Qd7 Re7 8.Qd8+
33/66 57:32 33,722,153k 9,766k 0.00 1.Qc4 Rf8 2.Qb4 Rd8
33/66 57:32 33,722,153k 9,766k 0.00 1.Qh4 Rde8 2.Qb4 Qa1 3.Qf4 Nxb3 4.axb3 f5 5.Qc4 Qf6 6.Rxd7+ Kf8 7.Rxe7 Qxe7 8.Qd4 Kg8 9.Qc4+ Qf7 10.Qc6 Qe6 11.Qc3 Qe7 12.Qc4+
33/66 57:32 33,722,153k 9,766k 0.00 1.Rdd3 Kf8 2.Rde3 f5 3.Rxe7 Kxe7 4.Re3+ Kf8 5.Qh4 Re8 6.Qc4 Re6 7.Qf4 Rxe3 8.Qd6+ Re7 9.Qb8+ Re8 10.Qd6+ Re7
33/66 57:32 33,722,153k 9,766k 0.00 1.Qf4 Rf8 2.Qd6 Rfe8 3.Qf4 Nxb3 4.axb3 Qa1 5.Rfd3 d6 6.Rd1 Qc3 7.R1d3 Qa1
33/66 57:32 33,722,153k 9,766k 0.00 1.Rd6 Qe5 2.Rd5 Qh8
33/66 57:32 33,722,153k 9,766k 0.00 1.Rd1 Kf8 2.Re1 f5 3.Rxe7 Kxe7 4.Re3+ Kf8 5.Qh4 Re8 6.Qc4 Re6 7.Qf4 Rxe3 8.Qd6+ Re7 9.Qb8+ Re8 10.Qd6+ Re7
So how does SF determine what moves to ignore?
Another strange thing is when you manually enter 1. Rxc5 bxc5 2. Qh4 Rde8 3. Rf6 from the start position it takes 26 ply and 26 seconds for SF8 to see that 3. ... Kf8 is the best move. And it sees what a blunder the move 3. ... Rf8 was in 5 seconds and 22 ply. So it amazes me that 3. ... Rf8 was played??
26/61 00:26 297,464k 11,057k -0.41 3. ... Kf8 4.Qf4 Qg8 5.Qc7 c4 6.Qxc4 Rd8 7.Qd4 Ke8 8.Qxa7 Qf8 9.Qc7 Qxh6 10.Bxf7+ Rxf7 11.Rxf7 Kxf7 12.Qxd8 Qd2 13.Qa8 Qc1 14.Qd5+ Ke7 15.Qe5+ Kf7 16.Qd6 Qc6+ 17.Qxc6 dxc6 18.Kf3 Ke7 19.Ke3 Kd6 20.Kd4
22/48 00:05 53,624k 10,593k -3.05 3. ... Rf8 4.Qg5 Re6 5.Bxe6 dxe6 6.Qf4 Rc8 7.g4 g5 8.Qxg5+ Kf8 9.Qf4 Kg8 10.g5 Rf8 11.Qf3 a6 12.Qc6 c4 13.Qxc4 a5 14.Qc7 a4 15.Qa7 a3 16.Qxa3
If nothing else these Alpha games are interesting for exploring weaknesses.
Regards