Author: Ricardo Gibert
Date: 08:14:48 06/05/99
Go up one level in this thread
On June 04, 1999 at 16:37:15, KarinsDad wrote: >On June 04, 1999 at 16:14:37, Dann Corbit wrote: > >>BTW, Peter Klausler's new encoding for CDB 2.0 may take about 8 bytes per >>position. >>;-) > >Magic. > >I wonder how he is doing it. Taking a sequence of moves off of a given known >position? Or is this the size of the database / number of positions (i.e. the >average)? > >KarinsDad :) If you download his database program, the text file called "internal" contains the following text that is exerpted: "POSITION/MOVE GRAPH CDB represents chess positions and the moves that link them. It has four different representations, which complicates the code enormously, but which allow large databases to be kept in physical memory without thrashing. The vast majority of positions are represented with the "Tiny" form. A Tiny position is one that has exactly one known incoming move and one known outgoing move. A Tiny position occupies only 12 bytes. It contains only a link to the preceding position, a 16-bit chess move code, some flags, and the hash table overhead. Reconstructing an actual chess position from a Tiny position is a recursive process: first reconstruct the single preceding position, then apply the incoming move."
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.