1402 Software Design and Implementation for Next-Generation Parallel Architectures

  • David Littlewood, Sandia National Laboratories
  • Henry Tufo, University of Colorado, Boulder
  • Hiroshi Okuda, University of Tokyo
  • Reese Jones, Sandia National Laboratories

Software development for high-performance scientific computing continues to evolve in response to increased parallelism and the advent of on-node accelerators, in particular GPUs. To run effectively at scale, codes must negotiate massive parallelism, data locality, and hardware resilience, among other concerns. In addition, the drive for improved performance can result in increased code complexity, which may affect developer productivity and long-term software maintainability. This minisymposium focuses on the design and implementation of engineering mechanics applications and related computational / communication kernels on next-generation supercomputing platforms, in particular approaches that replace or complement the traditional MPI programming model. Examples include strategies for performance portability across disparate hardware architectures, algorithmic advances for solvers, load balancing, and asynchronous tasking.

© WCCM-APCOM 2022. All Rights Reserved.