Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Branching factor again ;-)

Author: Dann Corbit

Date: 12:11:03 08/26/02

Go up one level in this thread


On August 26, 2002 at 14:02:54, Steffen Basting wrote:

>Hello!
>It is running on a PIII/850. I know the NPS are really slow, this is because of
>my VERY slow evaluation. I've got another version with a less complex evaluation
>which reaches about 200.000 nodes per second (but doesn't play as human-like as
>this version, which is my target ;-)).
>I think hashtables are the only way to make big progress now. I just implemented
>sort of history heuristic which makes my program reach the next ply about 5 -
>10% faster.

I recommend the Zobrist hash.  Other hashes are a bit faster to generate from
scratch, but Zobrist is really nice when you make/unmake a move since you can do
an update from the partial change.

Many open source programs use Zobrist hashing.  It should be easy to find
examples.  Bruce Moreland has a nice writeup on the complexities of hashing. (If
I recall correctly).



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.