Author: Dagh Nielsen
Date: 12:06:52 12/06/05
Go up one level in this thread
On December 06, 2005 at 05:42:11, Vincent Lejeune wrote: >On December 06, 2005 at 05:24:26, Steve Lim wrote: > >>On December 06, 2005 at 05:21:34, Daniel Shawul wrote: >> >>>On December 06, 2005 at 05:15:54, Steve Lim wrote: >>> >>>>Title says it all. Am just wondering how they differ from Eugene's EGTBs? >>> They store only Win Draw Loss information, hence they are smaller and some >>>of them can be loaded in to RAM, which makes them faster. >>> >>>> >>>>Thanks. >> >> >>Ahhh.. similar to the old Distance to Conversion vs Distance to Mate. >> >>This is DTC I guess. > >I'm trying to be more clear ... > >In the Bitbase endgame file only results is store, so there's only 3 values >possible for each position 1) win 2) lost 3) draw >In comparison, in the 5 pieces Nalimov files there's 256 (255?) possible value >from -127 to +127 (-127 = mated in 127 moves ; 0 = draw ; +127 = mate in 127 >moves, not very sure about this values, but it's the logic) > >In bitbase one can store 4 positions per bytes but one don't know the best move. >In Nalimov (5 pieces) one can store 1 positions per byte but one knows the best >move(s). > >remark : > I pointed Nalimov "5 pieces" because some 6 pieces endgame require more than 1 >bytes (mate > 127 moves). On one side that means that more pieces on the board >give bigger size difference between Nalimov and bitbase but the drawback will be >that with the bitbase, computer will not be able to find the mate because it's >so long that the file with draw/win/lose will not be enough to find the path to >the mate in 1. Hi, I'm not sure I understand all the advantages of bitbases in full, but I'm just wondering if one could not use an approach like this: (1) Use bitbases during search (2) Once won/lost position is reached, switch to Nalimov table bases. The point would be to speed up things, but I don't know if I got something wrong. Regards, Dagh Nielsen
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.