Using double precision:
The results should be consistent using either method as shown above. However, using quad precision, these computations do not give consistent results:
I'm not sure what's under the hood, but it seems to me that either computation should give the same answer.
Customer support service by UserEcho