Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: quick bitboard question

Author: Robert Hyatt

Date: 20:56:23 08/26/05

Go up one level in this thread


On August 26, 2005 at 20:29:30, Brian Richardson wrote:

>On August 26, 2005 at 16:14:33, Robert Hyatt wrote:
>
>>On August 26, 2005 at 15:57:09, Charles Roberson wrote:
>>
>>>
>>>  Bob,
>>>
>>>  In Telepath (bitboard), I matched the ordering from NoonianChess
>>>(nonbitboard). While this allowed some code reuse, it was somewhat confusing.
>>>
>>>   The pattern was A1=0, H1=7, A2=8, H2=15 .......
>>>
>>> In hind sight, It may have been better to use A1=7, H1=0, A2=15, H2=8.
>>>
>>>   Think of each system and how mentally easy or hard it is to calculate
>>>   diagonals. The reason I say "mentally easy" is you have to use your own
>>>   brain to debug as you well know.
>>>
>>>   Charles
>>
>>That is what I used in Crafty originally, which might be why many used the same
>>numbering.  The problem is that this requires you to "mentally" invert things,
>>which is not necessarily a bad thing.  I have considered keeping this the same,
>>and just having to "mentally mirror" things since in a 64 bit value, bit 0 is
>>the right-most bit...
>>
>>might be the best idea to leave the bit numbers the same, in fact...  and just
>>deal with the visualization issues mentally and not make that many drastic
>>changes to the engine...
>
>Tinker also uses Crafty's bitboard numbering (albeit non-rotated), and yes,
>sometimes it is a bit of a pain to visualize...
>But, I would just like to suggest some caution about the bitboard renumbering
>idea.  They are used in soooo many places and there are mask constants and such
>all over;  missing just one will mess things up in hard to find ways.  I
>understand that you will be essentially rewriting the eval, but even so it is
>just hard to see the benefit relative to the risk, IMO.


I had actually already reached that decision, looking at the code.  Much was
pretty easy to change, but changing all the rotated bitmap stuff is a pain, and
even pawn move generation would be altered.  So I've scrapped this, and will
just continue to put up with 63-square for the BSF/BSR stuff.

:)




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.