Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Problem with short PV

Author: Dann Corbit

Date: 01:53:10 01/16/02

Go up one level in this thread


On January 16, 2002 at 03:39:40, Benny Antonsson wrote:

>On January 15, 2002 at 23:41:38, Robert Hyatt wrote:
>
>>On January 15, 2002 at 21:25:50, Benny Antonsson wrote:
>>
>>>How come, that in some searches I get a long and nice PV (quisce excluded) and
>>>in another I get only one move in the PV ? Am I doing something wrong ? I
>>>suppose the hashtable is inteferring in some way ?
>>
>>If you get an <EXACT> match from the hash table, how could you possibly
>>have a PB _beyond_ that node?
>>
>>This is just a part of hashing...

>So... saving the PV in the hashtable could solve the problem ?

How much pv will you store?  If you store several forward ply, it will
significantly cut down on your number of potential hash entries for a given hash
table size.

One idea might be to have a move number from 0 to 255 (for instance, created by
lexically sorting the moves and numbering the list) and store a single 8 bit
entry for the expected response.  As long as you also have a hash table entry
for the response, you could connect the dots until you run out of HT entries.

Don't know if this is a dumb idea or not.  I haven't tried it.



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.