Search

link to homepage

Institute for Advanced Simulation (IAS)

Navigation and service


Performance Analysis

The Jülich Supercomputing Centre has a long tradition in the development of performance tools for parallel programs. The current focus is on the automation of the performance analysis process. With the KOJAK toolset, we aimed at the development of a genericautomatic performance analysis environment for parallel programs. Performance problems are specified in terms of execution patterns that represent situations of inefficient behavior. These patterns are input for an analysis process that recognizes and quantifies the inefficient behavior in event traces. Mechanisms that hide the complex relationships within event pattern specifications allow a simple description of complex inefficient behavior on a high level of abstraction. With the Scalasca toolset, a successor to KOJAK, the main focus is on scalability in order to support analysis of parallel applicationsrunning on today's supercomputer consisting of many thousand processor cores. Future versions of Scalasca will be based on the new community instrumentation and run-time measurement infrastructure Score-P.

Collaborations:

  • The Scalasca toolset is developed in a tight collaboration with the Programming Laboratory of the German Research School for Simulation Sciences
  • VI-HPS: Virtual Institute - High Productivity Supercomputing (HGF)
  • EIC: Programming methods and performance tools for future IBM architectures
  • ECL: Performance modeling and tools for future cluster systems
  • IESP: International Exascale Software Project
  • EESI: European Exascale Software Initative (EU FP7)

Current research projects:

  • CATWALK: A Quick Development Path for Performance Models" (DFG SPPEXA)
  • H4H: Hybrid programming for heterogeneous architectures (EU ITEA2)
  • LMAC: Performance Dynamics of Massively Parallel Codes (BMBF)
  • DEEP: Scalasca support for OmpSs and the DEEP architecure/Intel MIC (EU FP7)
  • Mont-Blanc: Scalasca support for OmpSs and the Mont-Blanc architecure/ARM (EU FP7)

Concluded research projects:

  • PRIMA: Performance Refactoring on Instrumentation, Measurement and
    Analysis Technologies for Petascale Computing (US DOE)
  • TEXT: Tool support for MPI/SMPSs programming model (EU FP7)
  • HOPSA: Integration of system and application monitoring (EU RU FP7)
  • ParMA: Parallel Programming for Multi-core Architectures (EU ITEA2)
  • SILC: Scalable Performance-Analysis Infrastructure (BMBF)
  • eeClust: Energy-efficient cluster computing (BMBF)

Servicemeu

Homepage