Argonne National Laboratory

An MPI/OpenACC Implementation of a High Order Electromagnetics Solver with GPUDirect Communication

TitleAn MPI/OpenACC Implementation of a High Order Electromagnetics Solver with GPUDirect Communication
Publication TypeJournal Article
Year of Publication2015
AuthorsOtten, M, Gong, J, Mametjanov, A, Vose, A, Levesque, J, Fischer, P, Min, M
JournalInternational Journal of High Performance Computing Applications
Pagination1-15
Date Published03/2015
Other NumbersANL/MCS-P5431-0415
AbstractWe present performance results and an analysis of an MPI/OpenACC implementation of an electromagnetic solver based on a spectral-element discontinuous Galerkin discretization of the time-dependent Maxwell equations. The OpenACC implementation covers all solution routines, including a highly tuned element-by-element operator evaluation and a GPUDirect gather-scatter kernel to effect nearest-neighbor flux exchanges. Modifications are designed to make effective use of vectorization, streaming, and data management. Performance results using up to 16,384 GPUs of the Cray XK7 supercomputer Titan show more than 2.5× speedup over CPU-only performance on the same number of nodes (262,144 MPI ranks) for problem sizes of up to 6.9 billion grid points. We discuss performance enhancement strategies and the overall potential of GPU-based computing for this class of problems.
URLhttp://hpc.sagepub.com/content/early/2016/02/01/1094342015626584.full.pdf+html
DOI
PDFhttp://www.mcs.anl.gov/papers/P5341-0415.pdf