Computer Chess Club Archives


Search

Terms

Messages

Subject: Please help for my hash table implementation

Author: Pham Hong Nguyen

Date: 23:37:38 04/26/99



I am trying to build a simple chess program. I use hash table following the
lecture “Strategy and board game programming” of UCI. Element of hash table is
defined:

typdef struct {
	long checksum;
	short	depth;
	enum {EXACT, LOWER_BOUND, UPPER_BOUND} entry_type;
	short eval;
} hashtable;

I hope that it would save time as much as some one reported about 90%. But it
has saved only 30% at 7 ply (with quiescence search). Some data as following
(without Null move):

No hash:
number of evaluation:	18.160.786

Small hash table:
	hash size (element):	64471
	number of evaluation:	13.389.557
	collision cases:	1.030.140
	successful cases:	268.593

Big hash table:
	hash size (element):	1.496.480
	number of evaluation:	10.921.967
	collision cases:	679.605
	successful cases:	273.021

When I use hash table with Null move, it saves more 5% than Null move only.

My questions are:
- Is it normal or may I do some thing wrong?
- How may I do to improve?

Any help would be much appreciated.

Pham Hong Nguyen (sorry for my English)
(From Vietnam)




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.