Author: Dieter Buerssner
Date: 13:47:28 06/05/04
Go up one level in this thread
On June 05, 2004 at 16:22:18, Michel Langeveld wrote: >On June 05, 2004 at 15:26:53, Anthony Cozzie wrote: > >>We need to get Nalimov to write some W/L/D tables. Preferably in C rather than >>C++, considering that 2/3 of all instructions in Zappa are from tbindex.cpp :) >> >>*pokes Eugene* >If you have the API working with the original Namilov tablebases, you can >generate these easily yourself. I agree. It took me a few days atound christmas 2002. This included lots of debugging and verification code. The basic idea of W/D/L tables (that can be probed from RAM) is perhaps doable in one day, for an efficient programmer. Say, at least the basics. Understanding the Nalimov indexing scheme, and beeing able to code something as space efficient, will take much longer (when you want to use the idea of the unblockable checks). But an indexing scheme, that is a bit less efficient, can be done rather fast. It is not needed, to understand the Nalimov scheme, to do such a task. Eugene's code will access the TBs very transparently for any user (=programmer). So it is no big problem, to translate to any scheme, you might want to use. Knowledge of the articles by Heinz and Balimov will be helpful. But even without this, it should be possible soon. When I tried, I did not know those articles, but later they helped me, to squeeze out some space. My first application of Eugene's code was, to write a code generator for a perfect (game theoretically, not in the sense of always find the shortest mate) KPK eval function. It took only an hour (or maybe few hours). "code generator" means writing a program, that puts out an initialized array in C here - with W/D (L is not possible) info for KPK here. Regards, Dieter
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.