Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

miniLB: A Performance Portability Study of Lattice-Boltzmann Simulations

Luigi Crisci, Biagio Cosenza, Giorgio Amati, Matteo Turisini

TL;DR

This paper presents miniLB, the first, to the best of the authors' knowledge, SYCL-based LBM mini-app, addressing the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems.

Abstract

The Lattice Boltzmann Method (LBM) is a computational technique of Computational Fluid Dynamics (CFD) that has gained popularity due to its high parallelism and ability to handle complex geometries with minimal effort. Although LBM frameworks are increasingly important in various industries and research fields, their complexity makes them difficult to modify and can lead to suboptimal performance. This paper presents miniLB, the first, to the best of our knowledge, SYCL-based LBM mini-app.miniLB addresses the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems. We analyze SYCL semantics for performance portability and evaluate miniLB on multiple GPU architectures using various SYCL implementations. Our results, compared against a manually-tuned FORTRAN version, demonstrate effectiveness of miniLB in assessing LBM performance across diverse hardware, offering valuable insights for optimizing large-scale LBM frameworks in modern computing environments.

miniLB: A Performance Portability Study of Lattice-Boltzmann Simulations

TL;DR

This paper presents miniLB, the first, to the best of the authors' knowledge, SYCL-based LBM mini-app, addressing the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems.

Abstract

The Lattice Boltzmann Method (LBM) is a computational technique of Computational Fluid Dynamics (CFD) that has gained popularity due to its high parallelism and ability to handle complex geometries with minimal effort. Although LBM frameworks are increasingly important in various industries and research fields, their complexity makes them difficult to modify and can lead to suboptimal performance. This paper presents miniLB, the first, to the best of our knowledge, SYCL-based LBM mini-app.miniLB addresses the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems. We analyze SYCL semantics for performance portability and evaluate miniLB on multiple GPU architectures using various SYCL implementations. Our results, compared against a manually-tuned FORTRAN version, demonstrate effectiveness of miniLB in assessing LBM performance across diverse hardware, offering valuable insights for optimizing large-scale LBM frameworks in modern computing environments.
Paper Structure (22 sections, 5 equations, 5 figures, 5 tables)

This paper contains 22 sections, 5 equations, 5 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Lattice Boltzmann Method
  4. SYCL
  5. Performance Portability
  6. miniLB Overview
  7. Computational Description
  8. Numerical precision
  9. FORTRAN-based Parallelization
  10. Use Cases
  11. Lid-Driven Cavity
  12. Von Karman Street
  13. SYCL Porting
  14. Kernel Parallelism
  15. Data Management and Access
  16. ...and 7 more sections

Figures (5)

  • Figure 1: Use case example outputs using the .vtk file with ParaView
  • Figure 2: Best SYCL configuration Million Lattice Update per seconds (MLUPs) for Lid-Driven Cavity and Von Karmann Street use cases
  • Figure 3: Best SYCL configuration speedup normalized over the Fortran best implementation
  • Figure 4: SYCL speedup over FORTRAN offload parallelism model by varying kernel and allocation type with column-major layout and in-order queues, on Lid-Driven Cavity
  • Figure 5: miniLB roofline models per target hardware