Dresden 2011 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

DY: Fachverband Dynamik und Statistische Physik

DY 40: Posters II

DY 40.39: Poster

Thursday, March 17, 2011, 17:00–19:00, P3

Multicanonical and Wang-Landau simulations on GPU — •Taras Yavors’kii and Martin Weigel — Institut für Physik, KOMET 331, Johannes Gutenberg-Universität Mainz, Staudinger Weg 7, 55127 Mainz

Over the last couple of years it has been realized that the vast computational power of graphics processing units (GPUs) could be harvested for purposes other than the video game industry. This power, which is advertised to exceed that of current CPUs by a large factor, results from the relative simplicity of the GPU architectures as compared to CPUs, combined with a large number of parallel processing units on a single chip. To benefit from this setup for general computing purposes, the problems at hand need to be prepared in a way to profit from the inherent parallelism and hierarchical structure of memory accesses. In this contribution, we discuss how suited this massively parallel setup is for simulations of the multicanonical or Wang-Landau type, which have proven useful for the study of systems with complex energy landscapes and first-order phase transitions. In contrast to canonical simulations, such approaches appear to require knowledge of the current value of a global reaction coordinate such as energy or magnetization prior to each spin flip. This seems to preclude their efficient implementation on GPUs. We assess the performance potential of different parallel implementations of such algorithms on GPU and compare to the speedups previously gained for simulations in the canonical ensemble.