Author: GuyHaworth
Date: 00:45:25 07/19/02
Go up one level in this thread
Certainly, if you were to divide all depths in EGTs by 4, you could rely on being able to search 8-plies deep in order to progress the win from one 'identified, won' position to the next. This would, as someone said, trade 'time' for 'space'. As Tony Werten said, you might have less success finding that first 'won' position. This would save 2 bits per 8 or 10 bits per position - a 20-25% saving only. Just storing win/draw/loss data - 2 bits, saves 75-80%, but surprisingly, I don't think the compression is as helpful as you might expect it to be. Other finesses to the current 'Nalimov' EGT-format are possible but do not render major improvements: a) replacing the value 'broken' by the previous 'won/drawn/lost' value ... improves the compression b) changing the index function might clump 'broken' index-positions better and improve access a little. I am not optimistic about being able to store 7-man EGTs: it looks like a new storage technology will be needed, handing us that factor of 60 or more that we need. g
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.