The lattice Boltzmann method (LBM) is becoming increasingly popular for the computational simulation of fluid flow. This approach is based on kinetic theory, and considers the evolution of dis-tributions of particles on a lattice whose collective behaviour represents that of the equations governing the motion of fluids. The use of the LBM is attractive as it has a relatively fast execution speed and is highly suited to parallel implementation on multiprocessor computers. In this paper, the most advanced formulation currently available, the generalized lattice Boltzmann equation (GLBE), or multiple relaxation time model, is discussed. The results of large eddy simulations of fully developed turbulent channel flow obtained using this model are presented and shown to be in good agreement with benchmark data. Stud-ies to compare the speed and numerical stability of the GLBE with other methods have been undertaken and clearly demonstrate its superiority over earlier versions of the LBM. Finally, results from performance tests using up to 1024 processors of a massively parallel supercomputer are provided, showing the expected near-linear scaling for large problems.