NNs seem to solve the Gusev Qxe5!! easily.

[MikeGL]

Just rechecked this Gusev immortal using latest SF13+.
Was surprised to see latest SF dev, as of today May 01 2021, being able to find Qxe5 to be the best (together with Qa3) in
less than 3 minutes on my 32-bit 11-year-old PC. Yes, Engine hashes was cleared before the run.

[d]4q1kr/p6p/1prQPppB/4n3/4P3/2P5/PP2B2P/R5K1 w - - 0 1

Code: Select all

  26/14	134.325.920	0.00	Qa3 Rxe6 Qxa7 Qe7 Qa8+ Qe8 Qb7 Qf7 Qa6 Qe7
 26/18	134.325.920	0.00	Qxe5 fxe5 Rf1 Rc7 Bd1 Rg7 Bb3 Re7 Bd5 Qd8 Rf7 Qc8 Rf1
 26/41	134.325.920    -0.07	Qd4 Rxe6 Rf1 Kf7 Bd1 Qb5 Bb3 Rhe8 h3 Qd3 Bf4 Qxd4+ cxd4 Nd3 Bh6 g5 Bxg5 Kg6 Bxe6 Rxe6 d5 Rxe4 Bxf6 Nf4 Bh8 Nxd5 Rd1 Nf4 h4 Re8 Bc3 Ne2+ Kf2 Nxc3 bxc3 Re4 Rd6+ Kf5 Rh6 Rc4 Rxh7
 26/41	134.325.920    -0.59	Qd5 Rxe6 Bd1 Kf7 Bb3 Qc6 Qxc6 Nxc6 Be3 Ke7 Bxe6 Kxe6 Kf2 Ne5 Rd1 h5 Bd4 g5 a4 Rd8 Ke2 g4 h3 Nf3 hxg4 hxg4 Ke3 f5 Kf4 fxe4
 26/39	134.325.920    -2.10	Qb4 Rxe6 Bb5 a5 Bxe8 axb4 Ba4 Kf7 Bb3 bxc3 bxc3 g5 Rf1 Rg8 h4 gxh4+ Kh1 Rg3 Bf4 Rxc3 Bxe6+ Kxe6 Bxe5 fxe5 Rb1 Kf6 Rxb6+ Kg5 Rb4 Kf4 Kg2 Rc2+ Kh3 Rxa2 Kxh4
 26/40	134.325.920    -2.56	Qd2 Rxe6 Rf1 Kf7 Bd1 b5 b4 Qc6 Qd4 Qb6 Bb3 Nc4 Be3 Rd8 Qxb6 axb6 Bd4 Rxe4 Rxf6+ Ke8 Bxc4 bxc4 Rxb6 Ra8 Kf2 Rxa2+ Kg3 g5 h3 Rc2 b5
 26/44	134.325.920    -2.88	Qd1 Rxe6 Qd5 Kf7 Bd1 Qc6 Bb3 Qxd5 Bxd5 g5 Rf1 Rg8 h4 Ke7 hxg5 fxg5 Bxe6 Kxe6 Bf8 h5 a4 g4 Bh6 h4 a5 Nf3+ Kf2 bxa5 Ra1 Rb8 Rxa5 Rxb2+ Ke3 h3 Bf4 h2 Rh5 Ra2 Rh6+ Kf7

[Spliffjiffer]

Just rechecked this Gusev immortal using latest SF13+.
Was surprised to see latest SF dev, as of today May 01 2021, being able to find Qxe5 to be the best (together with Qa3) in
less than 3 minutes on my 32-bit 11-year-old PC. Yes, Engine hashes was cleared before the run.

thats not the real surprise imo, its more surprising that with older sergio nets a winning-score of about 5.00++ is shown within seconds and Qa3 is still given 0.00 (...in MPv=3 eg...Gusevs Qxe5 is found very quickly with a 0.00 score as far as i remember (default net))