Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Analyzing in Rebel 9

Author: Ed Schröder

Date: 02:54:28 01/02/98

Go up one level in this thread


>Posted by Ernst Walet on January 01, 1998 at 16:58:25:

>I have analyzed the end position of the second game of the rematch
>between Kasparov and Deep Blue with Rebel9 and had some strange results.

>When i use power analyze on Qb6xc6 and Qb6-e3 separately it gives the
>following score on the PV:

>PLY            Qb6xc6       Qb6-e3

>10             -0.82        -2.34
>11             -1.00        -1.76
>12             -1.23        -1.52
>13             -1.37        -1.41
>14             -1.38        -1.18
>15             -1.56        -1.76

>When i use power analyze on Qb6xc6 and Qb6-e3 together it gives the
>following score on the PV:

>PLY            move

>10             Qb6xc6       -0.81
>11             Qb6xc6       -1.00
>12             Qb6xc6       -1.23
>13             Qb6xc6       -1.37
>14             Qb6xc6       -1.43
>15             Didn't wait because i expected it to switch PV on ply 14

>With normal analyze it swiches from Qb6xc6, score -1.56 to Qb6-e3,score
>-1.53 on the 15th ply.

>Why not the same score and ply depth switching between these two moves?

>has it to do with hash-tables?

Your guess is right.

Besides transpositions the hash table is also used to guide the search.

Whenever you change the "move order" of a search this may cause
different moves and/or scores. Not much but still...

This is even true if you only make some minor changes to the move
generator to improve sorting or so. Result: other moves, other scores.

However no big differences.

- Ed -




>beste regards, Ernst.



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.