Author: chip piller
Date: 15:06:59 02/09/01
Go up one level in this thread
On February 09, 2001 at 16:44:32, Robert Hyatt wrote: >On February 09, 2001 at 15:40:28, chip piller wrote: > >>I posted a similar question recently, noting that there is a lot of interest in >>fast cpu's and SMP machines to get improved performance for a chess program. >> >>I am working on (planning stage now) a distributed chess project. There is a >>project page at http://sourceforge.net/projects/dchess but we have no code yet. >> >>Our approach will be a distributed chess analysis program with low bandwidth >>requirements. >>Chip > ^^^^^^^^^^^^^^ > >alpha/beta is going to make this requirement _very_ difficult to address. >The search, by definition, is very high-bandwidth. > >That is the stumbling block any distributed algorithm has to overcome. The >more you reduce the bandwidth between nodes, the _larger_ the tree becomes. I suspect that what you say is correct. I am admittedly rather naiive and underinformed at this point on the theory and limitations of chess game tree searches. But I consider this to be in my favor, if I knew more about this I probably would not even attempt a distributed chess program. I'll post results (good or bad) when I have something up and running. Chip
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.