Author: Ian Osgood
Date: 20:15:47 03/24/04
Go up one level in this thread
On March 24, 2004 at 17:25:49, Mikael Bäckman wrote:
>On March 24, 2004 at 16:58:02, Andreas Guettinger wrote:
>
>>On March 24, 2004 at 15:23:26, Dann Corbit wrote:
>>
>>>A lot of people seem to use the 16x8 board with 0x88.
>>>
>>>Very vew people seem to use the other half for anything useful.
>>>
>>>I am wondering, does anyone use the other half of the board to store data about
>>>the square it duplicates?
>>>
>>>Seems like a very obvious enhancement to me.
>>
>>Interestingly I had exactly the same thoughts some days ago.
>>
>>I looked at some of my arrays[FIELDS] for piece values, mobility values etc
>>where the right part of the tables are always empty. I wondered if you could not
>>just put the values for white in the left half, the values for black in the
>>right hlaf of the tables and just use (sq+8) to access the values for black?
>>
>>regards
>>Andy
>
>
>I do this for my evaluation piecesquare values and attacktables. It works fine,
>but the code can get rather difficult to read. For example here's a bit of
>Cheplas kingsafety code which tests who controls the '2 forward' squares:
>
>sq = kingSq[1];
>if (OnBoard(sq + N + N)) {
> t = attCount[sq+N+N+8] - attCount[sq+N+N];
> if (t > 0) ac += t;
> if (OnBoard(sq + N + NW)) {
> t = attCount[sq+N+NW+8] - attCount[sq+N+NW];
> if (t > 0) ac += t;
> }
> if (OnBoard(sq + N + NE)) {
> t = attCount[sq+N+NE+8] - attCount[sq+N+NE];
> if (t > 0) ac += t;
> }
>}
>
>
>That '+8' is so easy to forget...
>
>/Mikael
#define WhiteAttacks(sq) attCount[sq]
#define BlackAttacks(sq) attCount[(sq)+8]
I use board interleaving for all the eval's piece-square tables in my Forth 0x88
program.
As of yet, I haven't found anything useful to do with the other half of the main
board. (the squares are encoded [000c cppp] where c is color and p is piece.
Ian
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.