Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hashing is a complicated affair ?

Author: rasjid chan

Date: 16:10:59 04/05/04

Go up one level in this thread


On April 05, 2004 at 18:58:57, Andrew Wagner wrote:

This is what I learn, maybe from a cursory read of crafty's commentary
(not source) and off I go... for Everest. It seems others are saying
MUCH MORE.

Rasjid




>On April 05, 2004 at 18:42:57, rasjid chan wrote:
>
>>On April 05, 2004 at 15:59:40, Dann Corbit wrote:
>>
>>What fruits! I can't yet digest the apple.
>>
>>On a more serious note, it seems there MAY BE much more in hashing
>>than what I know - UB, LB, EX. I need time to see what all these mean.
>
>UB = Upper bound, LB = Lower bound, EX = exact.
>
>When you store a value in the hash table, sometimes it will not be exact, so you
>store some flag along with it that says what kind of position it is. If you just
>failed high, all you know is that the score is at least X. If if failed low, all
>you know is the score is at most X. And if the score is between alpha and beta,
>it's exact.
>
>>
>>Rasjid
>>
>>
>>
>>>On April 05, 2004 at 15:24:11, rasjid chan wrote:
>>>
>>>>On April 05, 2004 at 14:44:04, Dann Corbit wrote:
>>>>
>>>>Maybe you miss my point. Actually I have no references about the
>>>>technical "intricacies" about hash table implementation and
>>>>I rediscover "new" things about hashing all the time which is BAD.
>>>>So I post this to hope someone just confirm with me once if my
>>>>analysis could be way wrong!
>>>
>>>Probably, I cannot answer well about that.
>>>
>>>I have many kinds of hash table.  One kind scores only exact scores and it is
>>>permanent (I save it to disk and restore it from disk).  These are also divided
>>>by piece count.
>>>
>>>The other kind of table stores edge values.  There is only one of these and it
>>>is regenerated for every game.
>>>
>>>I keep as much flag information as possible (this is a null move, a lower bound,
>>>an upper bound, unknown, qsearch, EGTB hit, invalid, killer,...)
>>>
>>>What you describe sounds OK to me.  Since you ask the question, I think you must
>>>see some problem.
>>>
>>>
>>>>I did have all the assert()s, you mentioned and I may be the top in using
>>>>assert().
>>>
>>>Have you seen the source code for fruit?



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.