Sphere-Decoding (SD) methods are branch-and-bound-like techniques used for optimal detection of digital communications signals over in wireless MIMO (Multiple input Multiple Output) channels. These methods look for the optimal solution in a tree of partial solutions; the size of the tree depends on the parameters of the problem (dimension of the channel matrix, cardinality of the alphabet), and such search can be much more expensive depending on these parameters. This search often has to be carried out in real time. This paper presents parallel versions of the Sphere-Decoding method for different parallel architectures with the goal of reducing the computation time.