Author: Dann Corbit
Date: 11:23:19 01/30/02
Go up one level in this thread
On January 30, 2002 at 08:51:17, Vincent Diepeveen wrote: [snip] >all assumptions are of course wrong: > > - EGTB ==> O(1) lookup, whereas in databases no such thing happens soon, > i miss a managed binary tree concept in your writings so you need O(n) > for every lookup AFAIK. Unless you use hashing, in which case it is O(1) > - you waste bits. If you store a position you can better store the > hash like i do in my hashtable and then add a score to it. So you > would need 64 bits + 8 bits score at most. Side to move of course > doesn't need to get stored, nor bounds or whatever, nor best move, > waste of bits it is. The format for the key is almost irrelevant. It can be changed at any time. >Most important is however that you do it all for nothing because: > > - it will never save time putting your stuff in hashtable, but it will > instead waste memory as the the search space is just too big to store > in a few terabyte of memory, especially if you use 250 bits a position. Amazing that you can know this before the experiment even takes place. Who said anyone was going to use 250 bits for a position? It was just a demonstration of the concept. Apparently, you haven't quite got it yet.
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.