Author: Andreas Mader
Date: 23:21:28 11/26/97
Go up one level in this thread
On November 26, 1997 at 16:09:35, Willie Wood wrote:
>
>On November 26, 1997 at 10:25:05, Andreas Mader wrote:
>
>>
>>Nimzo98 uses mobility terms for sure and it is a very fast program, too.
>>I was involved in the design of the mobility evaluation, funny things
>>can happen if you evaluate the mobility of a piece too high :). The user
>>can change the weight of four general mobility factors ('normal'
>>mobility and 'x-ray' mobility in midgame and endgame) if he/she wants to
>>do experiments.
>>
>>Andreas
>
>
>Does Nimzo use bitmaps? Apparently, as I understand it, the use of
>bitmaps can eliminate the time penalty for mobility eval. I'm
>interested to see whether non-bitmap programs use mobility, and if
>there's another approach to calculation other than counting the attacked
>squares.
>
>WW
Yes, Nimzo uses bitmaps whereever they are useful! The mobilty function
doesn't do a simple 'field counting' of every piece type, it's a little
bit more complex. One more reason why Nimzo is so fast is the code
optimizing (on the base of processor cycles). As far as I know this can
only be done when you write in assembler. When I first saw the
evaluation function (the one that is done in every node and not only at
the root) I couldn't believe that 200K NPS are possible (on a P233), now
I know better...
Andreas
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.