Editorial

In recent years, the simulational community has seen the start of a shift of the computational paradigm from serial and moderately parallel to massively parallel calculations, even at the level of single CPU systems. A particularly clear-cut example of this development is the increasingly widespread use of systems based on graphics processing units for general purpose computing. These devices, whose theoretical peak performance exceeds that of current CPUs by orders of magnitude, are inherently massively parallel architectures. Harvesting the vast performance of these and similar systems employing clever parallelization techniques for the problems at hand constitutes one of the dominant challenges for the computational sciences in the years to come. This issue focuses on recent progresses in using GPU computing for computer simulations in physics, and covers statistical and condensed-matter physics, computational fluid dynamics, lattice field theory and astrophysics. Remarkable is the large number of contributions that feature codes that are actually used for production, which clearly shows that GPU computing has become part of mainstream high performance computing