Author: Gian-Carlo Pascutto
Date: 09:05:23 01/21/02
Go up one level in this thread
On January 21, 2002 at 03:08:34, Ricardo Gibert wrote: >On January 20, 2002 at 09:47:29, Gian-Carlo Pascutto wrote: > >>On January 20, 2002 at 08:16:25, Ricardo Gibert wrote: >> >>>A typical EBF today is about 3 rather than 2, so a tripling of speed is needed >>>to add 1 ply of depth. >> >>A typical one yes, but the top commericals are much closer to 2 than to 3 >>it seems. >> >>ChessTiger's branching factor is simply amazing. >> >>-- >>GCP > >I have ChessTiger and have noted the same thing, but a EBF of 2 going to a depth >of 19 implies only about 2^20 nodes examined. That's about a million nodes, >which a really fast machine can do in about a second. Of course, CT can't get >that depth in anything near that time. That would be more like in Deep Blue >territory. That should tell you that something about the way nodes and/or depth >is being reckoned is not kosher. The effective branching factor usually improves as the depth gets bigger when using forward pruning with depth dependencies. This means that it is possible for CT to get an EBF of 4 or 5 in the first few plies but drop to 2 if the search goes deeper. I think this is illustrated with numbers in some of Heinz's papers. -- GCP
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.