Author: leonid
Date: 10:36:44 01/03/01
Go up one level in this thread
On January 03, 2001 at 12:21:10, Heiner Marxen wrote: >On January 03, 2001 at 09:25:29, leonid wrote: > >>On January 02, 2001 at 21:22:01, Dann Corbit wrote: >>>Under standard chess, it is a mate in 11 (as shown above correctly by Chest). >> >>Is is still pretty good. Search by brute force 11 moves deep should take pretty >>long time to solve. It probably used some kind of selective search and >>successfully. > >K6-2/400 with 10MB hash: 374 sec >PIII/550 with 30+MB hash: 177 sec >Speed factor for hash (transposition table): 35.6 (conservative estimate). > >You should really use a transposition table for positions with so few pieces, >it makes a great difference (hint, hint :-). More memory then also helps. > >Here are the move execution counts for the different levels (depth in moves): > >mvx 11: 18 22 [18.000 1.222] mvskip lvskip >mvx 10: 468 216 [21.273 0.462] >mvx 9: 5121 1400 [23.708 0.273] 79 >mvx 8: 29838 5123 [21.313 0.172] 1088 1 >mvx 7: 85729 13523 [16.734 0.158] 4948 45 >mvx 6: 211779 35455 [15.661 0.167] 15705 103 >mvx 5: 512410 96046 [14.452 0.187] 62779 368 >mvx 4: 1325249 303453 [13.798 0.229] 171602 1027 >mvx 3: 3751315 1617681 [12.362 0.431] 459 1 >mvx 2: 2386674 1787280 [ 1.475 0.749] >mvx 1: 56337 0 [ 0.032 ] > >In [..] you find the quotients between successive plies. > >(EGTBs are even better, of course) > >Heiner Thanks Heiner for numbers! It is interesting for me to glance on them since I have no counters of this kind in my mate solver. Have them only for the second part of my program. Those few last years I was very little with my program. Expect to come back and put all hash, and anything else, when I will rewrite it for expected 64 bits chip and the most probable for Linux. For now money and few other hobbies keep me in safe distance from my chess program. Leonid.
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.