Poster: performance modeling and computational quality of service (CQoS) in synergia2 accelerator simulations

  • Authors:

  • Steve Goldhaber;Stefan Muszala;Nanbor Wang;James F. Amundson;Eric G. Stern;Boyana Norris;Daihee Kim

  • Affiliations:

  • Tech-X Corporation, Boulder, CO, USA;Tech-X Corporation, Zurich, Switzerland;Tech-X Corporation, Boulder, CO, USA;Fermi National Accelerator Laboratory, Batavia, IL, USA;Fermi National Accelerator Laboratory, Batavia, IL, USA;Argonne National Laboratory, Argonne, IL, USA;Binghamton University, State University of New York, Binghamton, NY, USA

  • Venue:
  • Proceedings of the 2011 companion on High Performance Computing Networking, Storage and Analysis Companion
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

High-precision accelerator modeling is essential for particle accelerator design and optimization. However, this modeling presents a significant computational challenge. We discuss performance modeling of and computational quality of service (CQoS) results from Synergia2, an advanced particle accelerator simulation code developed under the ComPASS SciDAC-2 accelerator modeling project. Understanding both strong and weak scaling behavior is crucial both to designing code optimizations and to adding new functionality. We ported Synergia2 both to linux clusters and to Surveyor, an IBM® Blue Gene®/P (BG/P) system at Argonne[5]. We then performed detailed profiling which led to a heuristic performance model for a benchmark simulation. The 3-D Hockney space-charge solver was modeled in detail using a component-based approach as it represents the main scaling challenge. These processes helped identify performance bottlenecks and code optimizations. We applied CQoS methods to maintain optimal memory layout and used a component approach to maintain the required CQoS in the space-charge solver.