Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bitboard VS array board ,speed difference in movegen()

Author: TEERAPONG TOVIRAT

Date: 09:05:12 02/26/01

Go up one level in this thread


On February 25, 2001 at 05:44:43, Ralf Elvsén wrote:

>On February 25, 2001 at 05:19:54, TEERAPONG TOVIRAT wrote:
>
>>
>>>and 16 make/unmake alternately. (sqrt(1*16) = 4). So maybe
>>>17/2 = 8.5 is better than 4? Why not measure these numbers instead
>>
>>The problem  is we should use geometric mean ( y = sqrt (x))
>>or arithmetic mean  (y= x/2) .
>>I'm not sure. Perhaps, some math experts here can help.
>>However, I prefer geometric mean...
>>Thanks for you opinion.
>>Teerapong
>
>It is definitely the geometric mean for the branching factor.
>So in average you have (1 + 16) moves per 2 nodes. Then of
>course you must take the arithmetic mean to get 8.5 :)

I understand your point. But the basic idea of my test is try to imitate
the real life, in other word " in vivo" test.
If you loop 8.5 times,finally,you would get  (8.5^ply) nodes instead of
approximately (4^ply) nodes. However, I'll accept any figure that
produces the nearest  total searching nodes as my program
normally does.
Perhaps,this is the best solution between us ie. count  nodes then
compare them with  real situation.
Nice to discuss with you and everyone here :)
Teerapong




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.