Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hashtable size: diminishing returns?

Author: Gordon Rattray

Date: 06:18:06 03/28/00

Go up one level in this thread


On March 28, 2000 at 07:59:43, Tom Kerrigan wrote:

>On March 27, 2000 at 12:26:02, Gordon Rattray wrote:
>
>>How much benefit can one expect by supplying more RAM for hash tables?  I'd
>>guess that moving from, say, 32Mb to 128Mb would be a significant improvement
>
>I'm not sure about this. At least with my program, having _A_ hash table is a
>significant improvement, but the size doesn't matter that much. Anything over,
>say, 1MB just changes things by a few percent here and there.
>
>-Tom

What I don't understand is this...  I'm sure I've read somewhere that the search
efficiency of Fritz degenerates significantly once the hash table has been
filled.  This suggests to me that a bigger hash table would allow it to search
more before the efficiency drops.  Also the drop may be slightly less?!

Also, as a simplified example, wouldn't a program perform better if it can keep
a hash table entry for all positions within a 6 ply search as opposed to 5 ply?
Surely, a further search to 7 ply would then be quicker for the former case (all
6 ply entries)?

Please note that my understanding of chess algorithms is limited and mainly
involves alpha-beta.

Gordon





This page took 0.01 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.