This paper explores and evaluates various approaches to accelerate Electromagnetic Transient (EMT) simulations of power systems using Graphical Processing Units (GPUs). Existing EMT simulation methods face computational challenges in systems with extensive renewable energy sources due to the complexity and switching dynamics of the system. The paper focuses on simulation methods based upon specialized GPU solvers to handle simulations of large and complicated power systems (e.g., with extensive switching components) with computational efficiency. Results from benchmark systems show significant speedups, particularly for large networks with high-frequency switching events.