Towards Extreme-Scale Computing with High-Order Discontinuous Methods

Abstract

Exa-scale computers are expected to appear in the next decade. State-of-the-art CFO algorithms used today will need a drastic overhaul to take advantage of the massively parallel exa-scale computers with mixed CPU and GPU architectures perhaps with hundreds of millions of compute cores. In the present project, we examine several critical issues important for scalability with high-order discontinuous methods, and develop algorithms balancing scalability, with accuracy, efficiency and robustness. The last decade has seen significant progresses on adaptive high-order CFO methods in many aspects, including new formulations, efficient time integration algorithms, hp-adaptations, shock-capturing, high-order mesh generation and visualization. Their potential has been demonstrated in computational aeroacoustics, large eddy simulation (LES) and direct numerical simulation (DNS) of vortex dominated turbulent flows at moderate Reynolds numbers.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 30, 2020
Accession Number
AD1107168

Entities

People

  • Zhi J. Wang

Organizations

  • University of Kansas

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Accuracy
  • Air Force Research Laboratories
  • Algorithms
  • Computational Fluid Dynamics
  • Computers
  • Efficiency
  • Engineering
  • Experimental Data
  • Flow
  • Fluid Dynamics
  • Fluid Flow
  • Large Eddy Simulation
  • Reynolds Number
  • Scalability
  • Scientific Research
  • Simulations
  • Turbulent Flow

Fields of Study

  • Physics

Readers

  • Computational Fluid Dynamics (CFD)
  • Distributed Systems and Data Platform Development