Work

High-performance xPU Stencil Computations in Julia

Year: 2024

Type: article

Abstract: We present an efficient approach for writing architecture-agnostic parallel high-performance stencil computations in Julia, which is instantiated in the package ParallelStencil.jl.Powerful metaprogramming, costless abstractions and multiple dispatch enable writing a single code that is suitable for both productive prototyping on a single CPU thread and high performance production runs on GPU or CPU workstations or, if used in combination with distributed parallelization packages as ImplicitGlobalGrid.jl, on supercomputers.We demonstrate performance close to the theoretical upper bound on GPUs for a 3-D heat diffusion solver, which is a massive improvement over reachable performance with CUDA.jlArray programming.

Source: JuliaCon Proceedings

DOI: 10.21105/jcon.00138

Authors: Samuel Omlin, Ludovic Räss

Institutions: ETH Zurich, CSCS - Swiss National Supercomputing Centre, Swiss Federal Institute for Forest, Snow and Landscape Research, Laboratory of Hydraulics, Hydrology and Glaciology

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funders: Eidgenössische Technische Hochschule Zürich, Centro Svizzero di Calcolo Scientifico, National Supercomputing Centre Singapore

OpenAlex ID: W4403175100