Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: odd multithreaded search behavior--explanation?

Author: Bruce Moreland

Date: 22:07:48 05/26/00

Go up one level in this thread


On May 26, 2000 at 05:19:53, Mike Curtis wrote:

I edited this a little bit because you had a typo in "process 1".

>One way to split the tree search is right at the root.
>
>process 1 => alpha_beta(MIN_EVAL, VALUE1)
>process 2 => alpha_beta(VALUE1, VALUE2)
>process 3 => alpha_beta(VALUE2, MAX_EVAL)

[snip]

>What do you think?

Yuck.  The reason is that process 2 is what you'd do already, if you use a
window.  Making the window narrower isn't that big a win, and that's all you'd
be able to do if you already have a program that has a window.  You can fiddle a
lot with the window and not really change anything.

So this would be inefficient.  You would also have an amusing problem if process
1 failed high, while 2 and 3 failed low, which can happen.

bruce



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.