Computer Chess Club Archives




Subject: Re: CilkChess

Author: Don Dailey

Date: 16:42:23 02/16/99

Go up one level in this thread

On February 14, 1999 at 17:46:38, Dave Gomboc wrote:

>Are Don Dailey and Co. still using mtd(f)?  Are they also using ETC?
>Dave Gomboc

Yes, I still use mtd(f).   I can also switch to a PVS search but it
is a very clear slowdown, no matter what I do, so mtd wins.

The basic implementation of mtd is simpler than any other aspiration
based search, but the problems it introduces ends up making it harder
to implement in the long run.  As we've talked about before on this
group, you have a problem with stable PV's and other things.  We have
gradually worked around these problems.  PVS seems to be an even bigger
win on a parallel program so I am forced to keep it in the program
even though I'm not particularly fond of it.  The speedup is not
trivial or I would throw it out in a heartbeat!

- Don

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.