Direct simulation with the lattice Boltzmann code BEST of developed turbulence in channel flows In spite of the dramatic increase of the performance of recent supercomputers the direct numerical simulation (DNS) of turbulent flows is still an expensive venture in view of the high memory and CPU time requirements. Today, the DNS is limited to very low Reynolds number and simple flow geometries which are far away from technical relevance. Therefore, there is an increasing demand in the development of numerical schemes to simulate fluid flows resolving the turbulent scales in order to exploit existing and future supercomputers more efficiently. In that context, the lattice Boltzmann method have challenged the traditionally used DNS method based on FV or pseudospektral . The potential of these LBM have been clearly shown in various publications. The goal of the present paper is to specify the advantages of the LBM to DNS more quantitatively and to make use of the LBM to investigate new phenomena related to wall bounded turbulence.