In this paper, we present and justify an effective strategy for performing three-dimensional (3D) inertial-confinement-fusion (ICF) capsule simulations. We have evaluated a frequently used strategy in which two-dimensional (2D) simulations are rotated to 3D once sufficient relevant 2D flow physics has been captured and fine resolution requirements can be restricted to relatively small regions. This addresses situations typical of ICF capsules which are otherwise prohibitively intensive computationally. We tested this approach for our previously reported fully 3D simulations of laser-driven reshock experiments where we can use the available 3D data as reference. Our studies indicate that simulations that begin as purely 2D lead to significant underprediction of mixing and turbulent kinetic energy production at later time when compared to the fully 3D simulations. If, however, additional suitable nonuniform perturbations are applied at the time of rotation to 3D, we show that one can obtain good agreement with the purely 3D simulation data, as measured by vorticity distributions as well as integrated mixing and turbulent kinetic energy measurements. Next, we present results of simulations of a simple OMEGA-type ICF capsule using the developed strategy. These simulations are in good agreement with available experimental data and suggest that the dominant mechanism for yield degradation in ICF implosions is hydrodynamic instability growth seeded by long-wavelength surface defects. This effect is compounded by drive asymmetries and amplified by repeated shock interactions with an increasingly distorted shell, which results in further yield reduction. Our simulations are performed with and without drive asymmetries in order to compare the importance of these effects to those of surface defects; our simulations indicate that long-wavelength surface defects degrade yield by approximately 60% and short-wavelength drive asymmetry degrades yield by a further 30%.