JSweep: A Patch-centric Data-driven Approach for Parallel Sweeps on Large-scale Meshes

TL;DR

JSweep introduces a patch-centric data-driven framework for parallel sweeps on large-scale meshes, enabling scalable high-performance computations in complex multi-physics simulations.

Contribution

It presents a novel patch-centric data-driven abstraction and runtime system that effectively parallelizes sweep computations on large, complex meshes.

Findings

Scales to tens of thousands of cores with good efficiency

Supports unstructured and deforming meshes

Achieves high performance in real-world applications

Abstract

In mesh-based numerical simulations, sweep is an important computation pattern. During sweeping a mesh, computations on cells are strictly ordered by data dependencies in given directions. Due to such a serial order, parallelizing sweep is challenging, especially for unstructured and deforming structured meshes. Meanwhile, recent high-fidelity multi-physics simulations of particle transport, including nuclear reactor and inertial confinement fusion, require {\em sweeps} on large scale meshes with billions of cells and hundreds of directions. In this paper, we present JSweep, a parallel data-driven computational framework integrated in the JAxMIN infrastructure. The essential of JSweep is a general patch-centric data-driven abstraction, coupled with a high performance runtime system leveraging hybrid parallelism of MPI+threads and achieving dynamic communication on contemporary multi-core clusters. Built on JSweep, we implement a representative data-driven algorithm, Sn transport, featuring optimizations of vertex clustering, multi-level priority strategy and patch-angle parallelism. Experimental evaluation with two real-world applications on structured and unstructured meshes respectively, demonstrates that JSweep can scale to tens of thousands of processor cores with reasonable parallel efficiency.