Ok I see, what you want to do is what is usually done on the normal search for hash moves. Pretend that PxQ comes from the hash and do the same in your qsearch (search the 'hash' move, *then* generates the remaining moves). If you *know* that the move is doable (piece not pinned, etc) then you don't even have to check it's legality, so it will be faster than first generating all the captures.Michael Sherwin wrote:...
Concerning my second point (2.). If PxQ is possible and it is found early the capture can then just be made and -qsearch called. If when returning from -qsearch with a score >= beta then no more captures need to be generated. Simple really, but, will there be a speed gain?
Now you need a fast way to find the winning capture in your eval (should be easy if you allready have attacks in your eval).
HJ.