Author: Tony Werten
Date: 11:33:14 01/27/01
Go up one level in this thread
On January 27, 2001 at 14:19:25, Alvaro Jose Povoa Cardoso wrote: >On January 27, 2001 at 09:48:16, Robert Hyatt wrote: > >>On January 27, 2001 at 07:49:23, Alvaro Jose Povoa Cardoso wrote: >> >>>Can some one please explain in detail the inner workings of ETC? >>>Do we use it before we start looping through all moves at a particular node and >>>after the hash table look up? >>>Do we need to store additional stuff in the hash table for ETC? >>>Are there any freeware engines with free source code that use ETC? >>> >>>Thank you >>>Alvaro Cardoso >> >> >>When you get to a node X, and you don't find a hash entry that helps, then >>you generate the list of moves from that position, update the hash signature >>for each move in turn, and see if the resulting entry in the hash table would >>be useful. If you find a "yes" then you make that move first and continue >>searching knowing you will get a quick answer back due to that hash hit... >> >>It seems to save about 10% of the tree size. But the loop (for me) costs >>about the same. Making it a "wash". And in the case of "wash" I always go >>for the simple "wash" which means no ETC. > >Is it convenient to limit ETC to nodes say 2, 3 or 4 plies before quiescence >searh? >Will this keep ETC advantages and eliminate some of the loop overhead? The papers describing ETC say you must limit it some way. If I remember correctly they said don't do it the last 2 plies. Tony > >Alvaro Cardoso
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.