Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bit board representation

Author: Robert Hyatt

Date: 14:01:44 05/31/01

Go up one level in this thread


On May 31, 2001 at 10:47:27, Bas Hamstra wrote:

>>I'm not a factor of 3 slower. That is your imagination working.  Somethings
>>are faster in bitmaps, some slower.  You want to harp on the move generation.
>>That is less than 10% of my execution time, so it doesn't count.  Evaluation
>>is very good in bitmaps.
>
>For most things, yes, but not everything. If you want to use AttacksTo info
>heavily, like some tactical programs do, you have a problem doing that with
>bitboards. You could do it either "on the fly"  using a couple of dozens
>AttacksTo()'s and end up with a *very* low nps. The alternative would be to
>calculate AttackTo tables (or even keep them incrementally). Now that is a
>problem with bitboards too. It *can* be done, of course, but as a result for
>instance Crafty could then divide it's nps by 4 (if not more).
>
>(and remember I use bitboards myself)


I used to incrementally update the attacks_to and attacks_from.  It made my
program not quite twice as slow, NPS-wise, once I used rotated bitmaps to do
the updates.  In fact, it was something on the order of 50-60% faster when I
stopped the incremental update.  But _if_ I wanted that information, I would
certainly go back to doing it that way.  I just happen to not like using that
info in the evaluation.  I did for a while and didn't like the result.  After
taking it out, getting rid of the incremental update was the next natural step.



>
>
>Best regards,
>Bas.



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.