Kindergarten bitboards and Xiangqi move genaration?

[Han Chengye]

Considering todays fast bitscans, and to find target squares in the set-wise world, e.g. undefended capture targets for a fast cut-off, the whole approach with precalculated movelists likely doesn´t pay off, also considering the huge amount of resources needed for all the tables.

got it, thx!

[Greg Strong]

Might be worth a look to see how these guys did it:
http://sourceforge.net/projects/xqwizard/
http://sourceforge.net/projects/ki11egg/
http://sourceforge.net/projects/xiangqi-engine/
http://sourceforge.net/projects/palmxq/

There used to be a program called ChessV that supported lots of chess variants with large boards but I can no longer find it.

ChessV is now located at http://www.samiam.org/chessv/

Xiangqi isn't one of it's variants, but its bitboard routines could easily accomodate Xiangqi (although they are far from optimal - they are just old-school rotated bitboards.)

[Han Chengye]

Han Cheng Ye, who are you?

a good question for no answer.
the most difficult thing in the world is "know thyself"

[liuzy]

韩成叶，你是哪位？你是象棋名手吗？

[hgm]

Nevertheless I still have the impression that mailbox in combination with differentially updated attack lists are very much faster, and are the way to go. Especially in Xiangqi, where bitboards exceed the word length even on 64-bit systems.

but this paper www.teu.ac.jp/gamelab/RESEARCH/gpw2006revised.pdf
suggest bitboards may be more faster.

This paper discusses bitboards in Shogi, but it is not clear at all what exactly is the alternative they compare it to (to conclude that bitbords are faster). What I had in mind (differentially updating attacks on occupied squares with the aid of doubly cross-linked lists) would in my estimate be a much faster technique than bitboards, especialy in Shogi, where sliders do not dominate the game so much, and you don't have the complication of the Cannon. So they are probably not comparing to that.