The revolution of eMobility has brought about new challenges in automotive NVH and acoustics with requirements to address high frequency (>2kHz) and broadband noise. Automotive NVH engineers are increasingly incorporating porous trim materials for noise suppression, given their absorptive, damping, and insulating properties. The need to more closely simulate the real world (fidelity) results in a challengingly significant vibroacoustic problem which demands the inclusion of noise level reduction earlier in the automotive design cycle and results in extensive engineering simulation, with systems of equations that have 10s and even 100s of millions of degrees-of-freedom (DOF). The design optimization of Body In White structures results in challengingly large NVH models. Typical vibroacoustic problems require the calculation of the full body vibration with the interior trim over a wide frequency range (up to several kHz). With multiple design iterations and durability studies, automotive OEMs must standardize their NVH models on the most performant finite element solvers to efficiently use time and resources. Traditionally, eigenvalue extraction approaches like Lanczos methods have been used to address low-frequency dynamics problems with moderately large finite element systems but have practical limitations (Lanczos is very slow compared to ACMS). The introduction of multi-level sub-structuring techniques like ACMS (fast eigenvalue solver) has enabled engineers to address the mid-frequency dynamic range, effectively extracting 10,000 plus modes on large system models and increasing performance using parallel processing. Additionally, structural-acoustic multi-physics solutions now employ a discrete Statistical Energy Analysis (SEA) method, which, combined with finite element modeling, addresses both the high-frequency content and the broadband noise. Addressing this time-consuming and resource-intensive structural engineering task enables NVH engineers to incorporate new physics while maintaining the productivity needed for their design cycles. OEMs can perform thousands of dynamic analyses runs when designing a vehicle. This work will showcase performance and efficiency enhancements with various application cases. The application of ACMS on an ~50M DOF body-in-white (BIW) FE model over a large frequency range will be discussed, including the relative time efficiencies versus Lanczos for various computer hardware configurations, including optimizations for high-performance computing (HPC) using both distributed memory parallel (DMP) and shared memory parallel (SMP) processing.