Efficient and Scalable Algorithms for Smoothed Particle Hydrodynamics on Hybrid Shared/Distributed-Memory Architectures

摘要

This paper describes a new fast and implicitly parallel approach toneighbour-finding in multi-resolution Smoothed Particle Hydrodynamics (SPH)simulations. This new approach is based on hierarchical cell decompositions andsorted interactions, within a task-based formulation. It is shown to be fasterthan traditional tree-based codes, and to scale better than domaindecomposition-based approaches on hybrid shared/distributed-memory parallelarchitectures, e.g. clusters of multi-cores, achieving a $40\times$ speedupover the Gadget-2 simulation code.