Author: GuyHaworth
Date: 03:54:36 09/19/03
Go up one level in this thread
I'm afraid I know little of the details of EN's organisation of compression and
disc-files.
Some facts and questions:
a) Rob Hyatt did some experiments for EN and recommended 8KB block sizes
- presumably minimises 'time' at runtime
- larger block sizes mean smaller files, so filesize is not the goal
Q1? Is this "result = 8KB blocks" or "unit of compression = 8KB block"?
b) Google leads to some useful sources on LZ77, which Kadatch added to.
c) I guess EN has an index of the first 'index position' in each block
- I believe he does not have to "aim and adjust" to find the right data
It may not be that LZ77/Kadatch is optimal for EGTs but why wait? It seems ok.
An LZ77/K-compressed EGT is better than no EGT at all.
One way you can improve compression is to consider 'broken values' as
'wildcards' as they aren't going to be consulted. This is bound to improve the
string-matching on which LZ77 depends.
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.