Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: math mistake

Author: Gian-Carlo Pascutto

Date: 01:49:23 01/11/02

Go up one level in this thread


On January 10, 2002 at 22:00:08, Vincent Diepeveen wrote:

>Your drums forgot one thing. It is a cluster, not a shared memory
>machine. So expect a speedup of like 2.0 or something out of n processors.

No. I took this into account when adding in the branching factor.

As I understand it, Deep Blue was essentially a cluster too (no shared
hashtables for the hardware nodes), so the branching factor should be
comparable.

On a shared memory machine, I would have assumed a branching factor
of 2.5 or 3.0. It would have gone several plies deeper then.

The problem was that on a shared memory machine of so many processors
I have no idea about the actual parallel efficiency. But we know what
branching factor it translated to for DB, so it makes sense to use that
number instead.

--
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.