This position offers a good chance to see how much engines prune, if we compare with raw perft values:
Code: Select all
k1b5/1p1p1p1p/pPpPpPpP/P1P1P1P1/8/8/8/K1B5 w - - 0 1
JetChess 1.0.0.0
perft(1) = 8
perft(2) = 8
perft(3) = 67
perft(4) = 67
perft(5) = 636
perft(6) = 636
perft(7) = 6,351
perft(8) = 6,351
perft(9) = 65,912
perft(10) = 65,912
perft(11) = 697,423
perft(12) = 697,423
perft(13) = 7,475,537
perft(14) = 7,475,537
perft(15) = 80,762,412
perft(16) = 80,762,412
perft(17) = 877,372,612
perft(18) = 877,372,612
perft(19) = 9,569,117,249
perft(20) = 9,569,117,249
perft(21) = 104,678,687,584
perft(22) = 104,678,687,584
perft(23) = 1,147,749,457,358
perft(24) = 1,147,749,457,358
perft(25) = 12,607,474,602,517
perft(26) = 12,607,474,602,517
perft(27) = 138,688,607,524,558
perft(28) = 138,688,607,524,558
perft(29) = 1,527,429,467,586,307
perft(30) = 1,527,429,467,586,307
perft(31) = 16,838,019,116,987,973
perft(32) = 16,838,019,116,987,973
Furthermore, in the classical chess engines output of what I call (nominal depth)/(selective depth) = nd/sd (please correct me if I am wrong), I have always seen sd(nd) >= sd(nd - 1)... until now. Here are some examples of your SF output:Uri Blass wrote:[...]
33/56 00:00 68k 4,551k +1.44 1.Kb2 Kb8 [...]
34/60 00:00 659k 3,312k +1.44 1.Kb2 Kb8 [...]
[...]
49/90 00:00 1,003k 3,203k +1.44 1.Kb2 Kb8 [...]
50/92 00:00 3,465k 4,170k +1.44 1.Kb2 Kb8 [...]
[...]
57/101 00:01 5,445k 4,338k +1.44 1.Kb2 Kb8 [...]
58/101 00:15 77,466k 4,955k +1.44 1.Kb2 Kb8 [...]
[...]
Code: Select all
13/27 and 14/22
16/30 and 17/26
19/37 and 20/21
22/40 and 23/36
24/36 and 25/34
30/54 and 31/52
41/78 and 42/76
44/86 and 45/80
46/86 and 47/84
51/96 and 52/94
53/98 and 54/92
58/101 and 59/96
Regards from Spain.
Ajedrecista.