Carlos Esparza
Carlos Esparza
> I'm curious, since you delved deep into this rabbit hole - did you find that there was any significant benefit of changing \( to \%( anywhere? Barely. I just...
> @ces42 It looks like we are still calculating threatenedPieces but no longer using it. Did you intend to use it for something or can it be removed? No. That...
No, but maybe I'll just wait for https://github.com/official-stockfish/Stockfish/pull/5988 to get merged (some day :P) and rebase on master then. That will show the right diff.
I like this idea -- I swapped threads and hash in speedtest several times today
This could be merged now (in theory). Is there a place where I could document the new options?
@mstembera I counted the calls where the accumulator is updated by 3 moves at once (happens relatively often) but only very few of them (~0.2% I think) involved one side...
I think that should work. However backwards updates take up approx. 5x less time than forward updates so I don't think this optimization for backwards updates would gain enough to...
I tried replacing the DirtyPiece copies by references and, surpisingly, that resulted in a (small but measurable) slowdown. I like the other changes though.
Nice find! I don't understand it either (and it might be specific to my system), but if I benchmark your changes I get ``` Result of 40 runs ================== base...
@vondele done