Author: Peter Skinner
Date: 09:49:14 01/06/06
Go up one level in this thread
On January 06, 2006 at 11:25:58, Robert Hyatt wrote: >On January 06, 2006 at 10:42:43, Peter Skinner wrote: > >>On January 06, 2006 at 09:10:06, Torstein Hall wrote: >> >>>This must be a book learning, not positional learning, at least for all CB >>>engines not having position learning built in! So Fritz can in my view not learn >>>anything. >>> >>>Torstein >> >>Technically anything that runs in the Chessbase interface has positional >>learning. >> >>It is just some engines have it internally, and some the interface does it for >>it. Obviously one is better than the other, but both "have it". >> >>Peter > > >I'm not sure how that would work. Positional learning is normally a product of >fiddling with hash table entries. I don't believe the GUI can do this. In >fact, it probably doesn't even know where the hash table sits in memory, much >less how to make sensible adjustments to certain entries... Well Chessbase's interface controls the hash table for the engine. You can globally set one hashtable size for all engines to use, so it must know where it sits and how to utilize it. This is where I have a problem with it. Hash tables, book usage, book learning, tb access, and time management should be the engines' job, not the interface. This is also why I have a problem with the UCI protocal, due to the amount of options it gives up to an interface. The great thing about the Winboard protocal is that the engine controls just about everything and Winboard itself is just an interface. Nothing more... Peter
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.