Conventional maximum power point tracking (MPPT) algorithms are designed under assumption that all modules are uniformly insulated and therefore can be characterized by a single maximum power point (MPP). At increasing power level PV arrays become larger in area. Moreover, to achieve a reasonable voltage level modules are connected in series, creating strings. These two factors make any array susceptible for any mismatch in their current-voltage characteristics. Under partial shading condition (PSC), when PV modules receive different insulation, multiple maxima occur in power curve of the array. This leads to a tracking error of the MPP and results in significant decrease in efficiency of the system. This paper describes Global Maximum Power Point Searching (GMPPS) methods which are designed specifically for searching global MPP (GMPP). Three hardware methods, which are based on fast sweeping of the characteristics via additional circuitry are described and compared. Three-level boost (TLB) converter is utilized as a DC-DC converter performing GMPPS. Simulation results are provided with 15 kWp PV string model.