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 generic automatic 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 focus is on scalability in order to support analysis of parallel applications running on today's supercomputer consisting of many thousand processor cores. The latest versions of Scalasca are based on the community-maintained instrumentation and run-time measurement infrastructure Score-P.

Collaborations:

  • The Scalasca toolset is developed in collaboration with Laboratory for Parallel Programming of Technische Universität Darmstadt
  • VI-HPS: Virtual Institute - High Productivity Supercomputing (HGF)
  • NHR4CES: National High Performance Computing Center for Computational Engineering Sciences
  • ETP4HPC: European Technology Platform (ETP) for High-Performance Computing (HPC)
  • JLESC: Joint Laboratory for Extreme Scale Computing

Current research projects:

  • EUPEX: European Pilot for Exascale (EuroHPC JU)
  • ExtraNoise: Performance analysis of HPC applications in noisy environments (DFG/RFBR)
  • DEEP-SEA: Programming Environment for European Exascale Systems (EuroHPC JU)
  • POP: Performance Optimisation and Productivity (EU H2020)

Concluded research projects:

  • SCIPHI: Score-P and Cube extensions for Intel PHI (Intel Corporation)
  • RAPID: Runtime Analysis of Parallel applications for Industrial software Development (Siemens AG)
  • Score-E: Scalable Tools for the Analysis and Optimization ofEnergy Consumption in HPC (BMBF)
  • CATWALK: A Quick Development Path for Performance Models" (DFG SPPEXA)
  • DEEP: Scalasca support for OmpSs and the DEEParchitecture/Intel MIC (EU FP7)
  • Mont-Blanc: Scalasca support for OmpSs and the Mont-Blanc architecture/ARM (EU FP7)
  • LMAC: Performance Dynamics of Massively Parallel Codes (BMBF)
  • H4H: Hybrid programming for heterogeneous architectures (EU ITEA2)
  • PRIMA: Performance Refactoring on Instrumentation, Measurement and Analysis Technologies forPetascale Computing (US DOE)
  • HOPSA: Integration of system and application monitoring (EU RU FP7)
  • TEXT: Tool support for MPI/SMPSs programming model (EU FP7)
  • eeClust: Energy-efficient cluster computing (BMBF)
  • SILC: Scalable Performance-Analysis Infrastructure (BMBF)
  • ParMA: Parallel Programming for Multi-coreArchitectures (EU ITEA2)
Last Modified: 28.03.2022