Open JavaMast opened 3 years ago
Thanks you for testing these. Could you also provide some comparable data for the current stockfish master? These speeds look alright. The VNNI bringing no speedup is sad but expected, as the sparse implementation of the vector-matrix multiply is heavily bottnelecked by the amount of shuffling.
*Fat_Titz_x64_310821_ELTO_BMI2_RS was compiled with pos->nodes++;
** Stockfish = CMH Pruning Tweak
========== I unable to make SSE2 build and any 32-bit builds. You skipped support of these builds?
I unable to make SSE2 build and any 32-bit builds. You skipped support of these builds?
Yes, I don't care about that old stuff
i5-11400f