While it has been shown that sphere detectors for multiuser detection (MUD) and multiple-input multiple-output (MIMO) systems can be efficiently designed, a significant hindrance to their feasibility arises from the preprocessing requirements of these algorithms. This paper investigates a number of approaches that may be taken to address this issue, and evaluates their performance in fixed point and reduced precision floating point simulations to determine the most suitable approach. A very large scale integrated circuit (VLSI) implementation strategy is also proposed to show how the chosen method may be efficiently built in hardware.