2018 Impact factor 4.843
Particles and Fields

EPJ B Colloquium: Large scale simulations on GPU clusters

Optimal partitioning for 1024 processors of an irregular domain representing a full coronary tree

Graphics Processing Units (GPU) are currently used as a cost-effective platform for computer simulations and big-data processing. Large scale applications require that multiple GPUs work together, but the efficiency obtained with cluster of GPUs is, at times, suboptimal because the GPU features are not exploited at their best.

In this EPJ B Colloquium, Massimo Bernaschi and colleagues describe how it is possible to achieve an excellent efficiency for applications in statistical mechanics, particle dynamics and networks analysis by using suitable memory access patterns and mechanisms like CUDA streams, profiling tools, etc. Similar concepts and techniques may be applied also to other problems like the solution of Partial Differential Equations.

Editors-in-Chief
L. Baudis, G. Dissertori, K. Skenderis and D. Zeppenfeld
The author would like to thank two anonymous referees for pointing out several shortcomings in a previous version of this paper and for suggestions to improve its clarity.

J. H. Field

ISSN: 1434-6044 (Print Edition)
ISSN: 1434-6052 (Electronic Edition)

© Società Italiana di Fisica and
Springer-Verlag