RAM Frame: Dynamic Mass Offset

I have a model in which the periods and effective mass modal participation is different for each of these dynamic load cases:

1. Modal Analysis ( Dynamic with No Ecc)

2. Dynamic with + Eccentricity (5%)

3. Dynamic with - Eccentricity (5%)

Does RAM shift the center of mass to alternate points offset ( eccentric locations) from the original calculated center of mass and modifying the mass moment of inertia? 

Also, if that is the case, then per Section 12.9.5 of ASCE 7-10 we would not need to apply "Ax" i.e. Amplification of Accidental torsion. My question is why do you mention here in this post ( link below) about Ax?


Please let me know. Thanks!