Author: martin fierz
Date: 12:22:21 08/20/03
Go up one level in this thread
On August 19, 2003 at 13:39:34, Igor V. Korshunov wrote: >Hello! > >Is someone know links to papers about generating and compressing endgame >tablebases? > >I am going to implement tablebases for game "Russian checkers". >I do not want to reinvent well known things. This is very time consuming. > >The most important question for me is compressing and later fast access to >compressed tablebases. >I have not any ideas about it at the moment. > >Thanks for help, >Igor hi igor, definitely the right place for you to look is the chinook webpage at http://www.cs.ualberta.ca/~chinook/publications/ ; in particular the paper called "Solving Large Retrograde Analysis Problems Using a Network of Workstations". just forget about all the stuff they write about using a network and parallelization, and it's perfect. i reinvented the wheel in march 2002 and also computed the english checkers 8pc endgame database, because at that time, the chinook database was not yet available to the public (and had not been since they computed it around 1995, so i assumed it would "never" be). i relied heavily on this paper - and found it a very good and complete guide to endgame database computation. when you compress your database, don't use the chinook scheme of 5 win/loss/draw values per "uncompressed" byte (3*3*3*3*3 = 243 numbers) and 256-243=13 skip values; instead, use only 4 win/loss/draw values (=81 numbers) and use the much larger skip value range that you get this way. this improves the compression by around 20%. cheers martin
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.