TalkChess.com

A variation of the classical approach using x^(x-1) separation and pure calculation for orthogonal lines and tiny lookup for diagonals.
Very nice!

In the null-move context, to determine attack/defense targets and trajectories see also Botvinnik-Markoff Extension. Botvinnik already proposed Vector Attacks aka 15x15 boards with superimposed lookup as used in Pioneer.

Rebel wrote: ↑

Wed Mar 10, 2021 5:43 pm

Best computer chess website on the Internet Gerd.

Thanks for the compliments, Ed. But don't exaggerate

Can you please provide a link to the wrong pseudo code in CPW? And a link to the discussion here as well. I really hope, after more than 10 years of editing CPW, it is a little bit more than nostalgic stuff ;-) Don't get me wrong, the CPW is a great resource to get started with things, and to get t...

The chess programming Wiki should be enough, but apart from looking through it for nostalgic purposes, I think it is often too general with regard of topic discussion, and it seems to have at least some mistakes (for example, a few days ago I found a discussion on transposition tables, where the ps...

BTW ~bits + 1 is also known under the name -bits (2-complements system). Two's complement as increment of the ones' complement is quite obvious. Adding something to its ones' complement leaves all bits in a word set. Adding one overflows to zero. (x + ~x) + 1 == 0 I wonder why there were some ones'...

Sorry. I do not catch the idea. Not how to do it (that is what I feel you tried to explain me), I do not know what I need to do it! :P Soberango go piece by piece and move by move and test if it is allowed or not. Now I understand how to know all the moves that each piece could do. What is the idea...

Converting target squares of a bitboard into a list is referred as Bitboard Serialization in CPW. The necessary bit scan might also be done without special bsf, aka trailing zero count instruction (using compiler intrinsics for X64 or ARM ), but portable and quite efficiently by Multiplication of th...

Possibly noob question: It seems everyone is using bitboards combined with look up tables to generate moves. While I understand that for things like "is this square occupied" bitboards are great, I don't quite see the advantage of using them for move generation. For example, if I have a rook I can ...

From ML perspective it gets even more clear - HCE can only handle linear dependencies between static eval and results while NNUE thankfully to it's hidden layers is capable of revealing non linear dependencies, e.g. non-deterministic behavior in similar positions. Nonlinear dependencies have nothin...