Author: Jon Dart
Date: 13:16:05 03/21/01
I have been playing around with Phalanx a little bit and have been
very surprised at how fast it solves some tactical problems. Here's
some examples (these were run on a PIII/800 with Phalanx's
default hash size of 820k):
Arasan150.epd:
[D] 7k/3r2p1/7p/1pq1pp1P/p4n2/P1P2PQ1/1P1r4/1KN1R1R1 b - -
5 -> 0:01.06 132549 2 turns
6 369 130 159194 Qc5-d5 Kb1-a1 Rd7-a7 Re1-e3 Qd5-c5 Qg3-e1
6 -> 0:02.08 252709 0 turns
7 366 289 349190 Qc5-d5 Kb1-a1 Rd7-c7
7 371 430 541630 Pg7-g5 !
7 402 449 570746 Pg7-g5
7 -> 0:06.55 853331 1 turn
8 462 858 1082643 Pg7-g5 !!
8 755 1106 1354523 Pg7-g5
8 -> 0:14.97 1829711 0 turns
9 814 1914 2333154 Pg7-g5 Re1xe5 Qc5xe5
9 -> 0:34.67 4486506 0 turns
10 874 4491 5642519 Pg7-g5 !!
10 875 5570 6881403 Pg7-g5 Re1xe5 Qc5xe5
Depth=10, Value=875, Time=60.02, Last turn=4.30, Nodes=7454477, N/s=124199
PV = Pg7-g5 Re1xe5 Qc5xe5
[ Solution incorrect 0.00%, 0/1 ]
Processor time ......... 0:01:00.03
Average search depth ... 9.9
Average NPS report ..... 124199
Hash table age ......... 1
Good bye!
Arasan85.epd:
[D] r2qr1k1/p1p2pp1/5n2/2b2b1p/4PP2/2NP2QP/PP6/R1B1K1NR b KQ e3
5 -> 0:00.79 83493 0 turns
6 105 115 124323 Bf5xe4 !!
6 105 148 158930 Bf5xe4 Pd3xe4
6 -> 0:02.07 229015 0 turns
7 105 345 383179 Bf5xe4 Pd3xe4 Nf6xe4 Nc3xe4 Re8xe4 Ng1-e2
7 -> 0:07.01 806812 0 turns
8 165 1025 1133736 Bf5xe4 !!
8 165 1334 1480878 Bf5xe4 Pd3xe4
8 -> 0:26.57 2976057 0 turns
9 225 3371 3700630 Bf5xe4 !!
9 267 5209 5654886 Bf5xe4 Pd3xe4 Nf6xe4 Nc3xe4 Re8xe4 Ng1-e2
Qd8-d5 Rh1-h2
Depth=9, Value=267, Time=60.01, Last turn=0.11, Nodes=6517965, N/s=108614
PV = Bf5xe4 Pd3xe4 Nf6xe4 Nc3xe4 Re8xe4 Ng1-e2 Qd8-d5 Rh1-h2
[ Solution incorrect 0.00%, 0/1 ]
Processor time ......... 0:01:00.01
Average search depth ... 8.9
Average NPS report ..... 108614
Hash table age ......... 1
(Note Nxe4 is also an ok solution here).
Arasan87.epd:
[D] r3r1k1/pb1n1p2/1p1p1n1p/q1pPpN2/2P4p/2PBPP2/P1Q3PP/1R3RK1 w - -
-> easy move (1) Nf5xd6
-> easy move off
6 -> 0:00.53 66550 0 turns
7 94 196 235968 Nf5xd6 Qa5-a6 Nd6xe8 Ra8xe8 Rf1-d1 Ph4-h3
Pg2xh3
7 99 423 461155 Qc2-f2 !
7 106 540 578228 Qc2-f2 Nf6-h5 Qf2xh4 Nd7-f6 Pg2-g4 Qa5xa2
Nf5xh6 Kg8-f8 Pg4xh5
7 -> 0:05.53 594163 1 turn
8 121 989 1015117 Qc2-f2 Nf6-h5 Nf5xd6 Re8-b8 Qf2xh4 Nd7-f6
8 -> 0:11.73 1236434 0 turns
9 177 2737 2703873 Qc2-f2 Nf6-h5 Qf2xh4 Nd7-f6 Pg2-g4
9 -> 0:32.71 3369185 0 turns
Depth=10, Value=177, Time=60.01, Last turn=4.23, Nodes=5949974, N/s=99149
PV = Qc2-f2 Nf6-h5 Qf2xh4 Nd7-f6 Pg2-g4
[ Solution incorrect 0.00%, 0/1 ]
Processor time ......... 0:01:00.03
Average search depth ... 9.9
Average NPS report ..... 99149
This last move (Qf2) was actually played by Phalanx in a blitz game.
Crafty 18.1 takes almost two minutes to solve Arasan85, about
1:45 to solve Arasan87, and about 1:20 to solve Arasan150. Arasan
is also a lot slower than Phalanx on these tests, even though
its nps is about twice what Phalanx is getting.
--Jon
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.