Computer Chess Club Archives




Subject: Re: Matt Taylor's magic de Bruijn Constant

Author: Gerd Isenberg

Date: 23:11:47 07/13/03

Go up one level in this thread

>Interesting, but not easy I would think. Hashing ok (did you solve that problem
>actually?), but at some point you will have to know where the pieces are. Take
>Swap() for instance. Direct attacks in Swap() you could do without LastBit, but
>how about discovered ones? You would have to know where the pieces are.
>Best regards,

Finally you need a lookup to a 64*64 table with the unique move-key.
But only in doMove - not during generation - and there is only one lookup for
two squares.

One more reason is that 32-bit IMUL becomes faster with Opteron (single
directpath 3 cycles), but BSF is still vectorpath with 8/9 cycles latency.


This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.