Author: Robert Hyatt
Date: 11:35:41 06/02/05
Go up one level in this thread
On June 02, 2005 at 11:41:58, Gian-Carlo Pascutto wrote: >On June 02, 2005 at 07:46:16, emerson tan wrote: > >>would hydra be more powerful if the team use OPTERON rather than XEON? > >The limit is probably the cards themselves or the interfacing, rather than the >software search. > >So they can use slower, cheaper, second grade CPU's for that :) > >-- >GCP This is also known as "the deep blue paradox". Once the chess processors are built, they can search at a certain speed. That is, it takes N units of time to complete a search to depth P. The software part of the search needs to provide a new position to search every N units of time to keep the chess processor 100% busy. If you speed the software search up with a faster cpu, you will gain a ply of search in the software. But now you are providing positions too quickly for the chess hardware to search them, so you have to reduce the depth P by 1 ply so that the searches are completed more quickly. Net gain... zero.
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.