Double Mach reflection, strong scaling and scaled-resolution series

Strong scalability, and throughput at a resolution growing with the number of processes, of an adaptive finite volume solver for the compressible Euler equations with strong shocks, and comparison of the dynamic load balancing strategies of samurai

Quick Facts

Type

Mini App

Status

Planned

Work Packages

WP1

Frameworks

Samurai

1. Overview

Strong scalability, and throughput at a resolution growing with the number of processes, of an adaptive finite volume solver for the compressible Euler equations with strong shocks, and comparison of the dynamic load balancing strategies of samurai

1.1. Repository

github.com/hpc-maths/samurai-euler

2. Technical Stack

2.1. Frameworks Used

  • Samurai

2.2. Parallel Frameworks

  • MPI

3. Methods & Algorithms

3.1. WP1: Discretization

  • mesh adaptation

  • multiresolution analysis

  • finite volume

  • dynamic load balancing

4. Benchmarking

4.1. Metrics

  • Regression Test Pass

  • Strong Scalability

  • Time To Solution

4.2. Benchmark Scope

  • Multi Node

5. Timeline

Milestone Date

Specification Due

N/A

Prototype Due

N/A

6. Team

Partners: IPP

Responsible: L. Gouarin