Power flow computations are a cornerstone of many simulations regarding the electric grid. This thesis evaluates the landscape of power flow computation methods with a focus on practical computational performance in large-scale simulations, as they occur in modern distribution grid planning. The investigation involves various model assumptions, different algorithms, implementation details, and unconventional computational optimization methods. As a result, the implementations devised in this thesis are up to a thousand times faster for large scale grid simulations than established solutions.