great! fastmath=True
is also important to have a shot at activating simd instructions.
1 Like
@evalparse This thread is rather informative about SVML optimization https://github.com/numba/numba/issues/5562 if you still trying to make it work
Perhaps I’m missing something here, but aren’t we missing the final calculation of the Euclidean distance transform by not taking the square root of the summed squares?