An established technique to solve time-dependent problems is the formulation of a complete frequency-domain analysis via Laplace or Fourier transforms, with subsequent ad hoc expression of results by numerical inversion. Although usually easy to implement, such a transform inversion is computationally intensive, if accurate results are desired, and liable to numerical instabilities. For diffusion-type problems, the Gaver-Stehfest algorithm seems well suited. General dynamics problems demand more robust algorithms usually based on Fourier series expansions, as firstly proposed by Dubner and Abate. Algorithms of either kind are already implemented in mathematical languages such as Matlab and Mathematica. The Fortran library has a Fourier-series algorithm proposed by Crump and improved by de Hoog et al. More recently, it has been proposed to solve transient problems of potential and elasticity by using an advanced mode superposition technique that applies to equilibrium-based finite element and boundary element models. One starts with a frequency-domain formulation that leads to a complex-symmetric (if viscous damping is included), effective stiffness matrix expressed as a frequency power series with generalized stiffness, dumping and mass matrices. After solution of the associated complex-symmetric, non-linear eigenvalue problem, one arrives at an advanced modal solution of the problem, which leads to the straightforward solution in the time domain and the immediate expression of any results of interest. Aim of the present research work is to compare the computational efficiency of the proposed advanced modal analysis with the techniques based on numerical inverse transforms, as applied to general, large scale problems. The relevant literature is reviewed and the main conceptual differences of the investigated methods are briefly outlined. All algorithms are implemented in Fortran so as to assure a common basis of comparison. Some initial results are displayed, as more definitive conclusions can only be expected after a large series of numerical simulations.