# which is going to be faster on x86_64 with intel or amd intrinsics, int based 128 bit fixed point, or quad double floating point?