Author: Gerd Isenberg
Date: 12:53:02 11/30/02
Go up one level in this thread
On November 30, 2002 at 14:31:25, Russell Reagan wrote: >On November 30, 2002 at 12:39:40, Arshad F. Syed wrote: > >>Anyone done any research on this? Is it worth looking into? >> >>http://www.fys.ku.dk/~fischer/Temp/New%20Technology.htm >> >>Thanks, >>Arshad > >This sounds similar to the Kogge-Stone stuff. Different algorithms, but similar >in the fact that you don't need rotated bitboards, and you generate attacks by >computation rather than lookup. Reversed bitboards have nothing to to with Kogge-Stone or Flood Fill Algorithms. As far as i remember it is based on the x^(x-2) trick to generate attacks from single sliding pieces in all index increasing directions. It uses reversed bitboards to use the same trick in all other directions. On drawback is that you must handle also reversed attack bitboards. If you need to combine them, you have reverse one by mapping all bits i to bit 63-i. But the x^(x-2) trick is great and if i remember right, that side was the first time, i heard from this trick. I use the x^(x-2) idea bytewise (SIMD), in my mmx-routines to generate multiple rook attacks in the right board direction, even if multiple rooks are on the same rank: // input: mm6 occupied, mm1 rooks // output: mm0 right rook attacks movq mm0, mm6 ; occupied psubb mm6, mm1 ; occupied - rooks psubb mm6, mm1 ; occupied - 2*rooks pxor mm0, mm6 ; rightRookAttacks := occupied ^ (occupied - 2*rooks) Gerd
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.