Author: Ren Wu
Date: 11:42:04 01/29/01
Go up one level in this thread
On January 29, 2001 at 13:18:23, Robert Hyatt wrote: >On January 29, 2001 at 12:39:25, chip piller wrote: > >>I am reading a lot of posts with questions about SMP performance and specific >>CPU performance for chess programs. >>I am in the planning stage of a distributed chess program whose goal is to >>obtain improved performance by combining the processing power of several >>(2+)networked computers. I understand that there will be some interesting >>design issues for such a program, for example the hash and tree search handling. >>Does anyone here have any experience with or knowledge of a distributed chess >>program? Any suggestions/comments would be greatly appreciated. >>Chip > > >Look up "Sun Phoenix" and papers on distributed search written by Jonathan >Schaeffer at the university of Alberta. > >He worked on this for several years in fact, and produced reasonable >results. > As i remembered, the speed-up is not good when the number of computers increased, something like < 10x for unlimit number of computers. And i believe in 89 WCC, he used one or more computers to do tactic search only, becasue of this. But that was a long time ago, and with very slow network. with today's high speed nets, things may change a bit? Ren. >There are other parallel message-passing programs, but they are historically >based on non-network type connections (ie the hypercube machines, etc.)
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.