We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.DC

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Distributed, Parallel, and Cluster Computing

Title: Distributed Parallelization of xPU Stencil Computations in Julia

Authors: Samuel Omlin (1), Ludovic Räss (2 and 3), Ivan Utkin (2 and 3) ((1) Swiss National Supercomputing Centre (CSCS), ETH Zurich, Lugano, Switzerland, (2) Laboratory of Hydraulics, Hydrology and Glaciology (VAW), ETH Zurich, Zurich, Switzerland, (3) Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Birmensdorf, Switzerland)
Abstract: We present a straightforward approach for distributed parallelization of stencil-based xPU applications on a regular staggered grid, which is instantiated in the package ImplicitGlobalGrid.jl. The approach allows to leverage remote direct memory access and enables close to ideal weak scaling of real-world applications on thousands of GPUs. The communication costs can be easily hidden behind computation.
Comments: submitted to JuliaCon Proceedings 2022
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Computational Physics (physics.comp-ph); Geophysics (physics.geo-ph)
Cite as: arXiv:2211.15716 [cs.DC]
  (or arXiv:2211.15716v1 [cs.DC] for this version)

Submission history

From: Samuel Omlin [view email]
[v1] Mon, 28 Nov 2022 19:02:59 GMT (230kb,D)

Link back to: arXiv, form interface, contact.