Author: leonid
Date: 08:21:05 05/19/01
Go up one level in this thread
On May 19, 2001 at 10:26:58, Heiner Marxen wrote: >On May 19, 2001 at 08:36:17, leonid wrote: > >>On May 18, 2001 at 22:03:13, Heiner Marxen wrote: >> >>>On May 17, 2001 at 13:50:30, leonid wrote: >>> >>>>Hi! >>>> >>>>If somebody would like very easy mate, then this veriation from this first is >>>>exactly this. >>>> >>>>[D]qr3rqB/3k3B/N1qqq1NQ/Q1n1n2Q/1qQ1Qq1Q/2RNR2Q/Q2K3b/1q3q1b w - - >>>> >>>>Please indicate your result. >>>> >>>>Thanks, >>>>Leonid. >>> >>>Mate in 8. Two key moves. On K7/600 with 350 MB hash Chest after 441.12 sec >>>(7.4 min) mutters: >> >>Excellent! Mine by brute force reached first solution (not all) in 4 min 10 sec. >> >>> >>>Nxc5+ Qbxc5 Nxe5+ Qfxe5 Q6xe6+ Qdxe6 Nxc5+ Kc8 Qhxe6+ Qcxe6 Qxe6+ Qexe6 >>>Qexe6+ Qxe6 Qxe6# >>> >>>Nxe5+ Qfxe5 Nxc5+ Qbxc5 Q6xe6+ Qdxe6 Nxc5+ Kc8 Qhxe6+ Qcxe6 Qxe6+ Qexe6 >>>Qexe6+ Qxe6 Qxe6# >>> >>>As often, symetric solutions to a (nearly) symetric problem. >>>Timing info reveals that Chest does feel quite comfortly: >>> >>>depth seconds >>># 3 0.02 0.96 72- 0 >>># 4 0.24 1.12 667- 0 >>># 5 2.05 1.28 6153- 0 >>># 6 13.72 1.53 44092- 0 >>># 7 79.46 2.02 265586- 0 >>># 8 441.02 2.70 1524472- 9 >> >>Here I am confused. What is your node/per/second for 8 moves? > >The above numbers were intended to show the effective branching factor. >The "nodes" above are "generated hash table entries", not searched positions. >Those can be seen (indirectly) from the numbers below. > >>Mine is 93k per second when it looked 8 moves deep and by brute force. I am so >>exact in indication since search 7 moves deep, for instance, give different >>number. It is 152k. NPS for 5 and 6 moves deep is also different. Strange but >>true. >> >>Cheers, >>Leonid. >> >>> >>> white black quotients >>>mvx 8: 32 38 [ 32.000 1.188] 39 >>>mvx 7: 708 722 [ 18.632 1.020] 91 >>>mvx 6: 7324 8085 [ 10.144 1.104] 185 >>>mvx 5: 57939 69121 [ 7.166 1.193] 1052 147 >>>mvx 4: 423641 537738 [ 6.129 1.269] 5565 198 >>>mvx 3: 2906635 3507479 [ 5.405 1.207] >>>mvx 2: 14814509 2992977 [ 4.224 0.202] >>>mvx 1: 275837 0 [ 0.092 ] > >Summing the above move numbers (every move makes a new position searched) >I get 25602785 / 441.02 = 58054 N/s. Aha! Very close to my numbers once again. Your went down a little bit because of heavy use of hash. Leonid. > >Heiner > > >>>Also, black likes to deliver mate-in-1 to white: >>>anti: 2529687 try, 1637033 lists [64.71%], 2376273 moves [93.94%], 1461034 succ >>>[57.76%] >>> >>>Cheers, >>>Heiner
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.