Author: Robert Hyatt
Date: 08:57:33 03/21/04
Go up one level in this thread
On March 21, 2004 at 08:57:41, Uri Blass wrote: >I thought that ETC is about finding if the next move cause cutoff so >I thought that the way it is done is by the following steps: > >1)generating list of moves > >2)calculating hash key after every legal move without making it(I an sure that >making every move is significantly more expensive espacially for me when I >update my attack tables during make move) Your idea is correct. You want to find a move that will cause a cutoff at the next ply, since you didn't get one at this ply. Do the minimal amount of work necessary to accomplish this test. Repetition test is cheap so doing that is also probably ok. But keep the cost down or it won't pay off. > >3)in case of hash hit generating a cut off if you can do it. > >I decided to look at fruit's code to see how it done there and I was surprised >to see that fruit simply makes all the moves and later checks also for mate or >for draws that are not based on hash tables. > >You of course get more information by this way and it may be productive to do >it(but I still think that making a move not in order to search later is not the >fastest way) but I wonder if it is the meaning of ETC. > >If yes then I think that the main benefit of tranposition cutoff in the middle >game is simply mate detection that leads to a smaller tree. > >Uri
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.