Computer Chess Club Archives




Subject: Re: MTD(f) and hash table size

Author: Tord Romstad

Date: 10:51:58 08/17/03

Go up one level in this thread

On August 16, 2003 at 18:34:26, Dan Andersson wrote:

> Depth new >= depth old is a pretty crappy replacement scheme IMO. You risk
>filling the hash with worthless nodes. And that will kill MTD(f) as it will mean
>you have to do the same work repeatedly in the new null window search, and it
>may harm other search implementations. I actually prefer a repalce always scheme
>to that one.
> Better to use a two slot scheme. Probe both slots for a hit. One slot is always
>replace the other is depth or deeper. If you don't get a hit you replace
>according to the two schemes.

Thanks.  I have now tried this, and indeed it works much better.  Unfortunately
the search
still suddenly slows down after a few minutes, but I can now search one or two
deeper before the wall appears.


This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.