November 15-20, 2015, Austin, Texas

Vampir meets SC15

Vampir comes to SC15 in Austin! Our performance visualizer for parallel programs provides graphical visualization of massively parallel applications. Learn how we walk the road to exascale at our booth #1351 from November 16 to November 19. Additionally, you can engage with our experts in various activities of the SC15 technical program. We showcase the latest version of our tool suite and introduce a sampling technology that drastically simplifies performance optimization.

At SC15 we unveil Vampir 9 that introduces new performance charts, customizable performance metrics, extended grouping and filtering, and new support for sampling traces.
Selected highlights include:
  • Combined visualization of instrumented and sampled events ensures a coherent performance analysis experience
  • New sampling related statistics and counter metrics allow an even deeper understanding of measurement data and application behavior alike
  • Enhanced visualization of customizable process groups allows easy analysis of idle times and further behavior that impacts performance.
  • New process filter management allows comparative analysis of statistics from different process groups at the same time.
  • Support for traces generated by Score-P 2.0

Vampir Website


Score-P is the primary code instrumentation and run-time measurement framework for Vampir 9, and also works natively with Scalasca, TAU, and Periscope. The latest version 2.0 provides numerous improvements and new features.
Selected highlights include:
  • A new measurement feature records the full calling-context information for each event. This in combination with an all-new sampling technology allows Score-P to measure application behavior with very low overheads. Careful handling of parallel paradigm invocations ensures accurate timings for these events.
  • Memory allocations and de-allocations are now monitored.
  • A new timer infrastructure allows users to select the timer at measurement-time.
  • New low-overhead timers improve the measurement of very fine-grain application behavior.
  • New compiler wrappers greatly simplify the integration of Score-P into common build systems.
  • Added support for CUDA 7.0 and onwards.

Score-P Website

Our events on SuperComputing 2015

TimeRoomTypeTitle
Sun, 15th November
08:30 AM - 05:00 PM
14TutorialEfficient Parallel Debugging for MPI, Threads, and Beyond
Sun, 15th November
01:30 PM - 05:00 PM
13ABTutorialTowards Comprehensive System Comparison: Using the SPEC HPG Benchmarks for Better Analysis, Evaluation, and Procurement of Next-Generation HPC Systems
Mon, 16th November
08:30 AM - 12:00 PM
14TutorialPractical Hybrid Parallel Application Performance Engineering
Mon, 16th November
09:00 AM - 05:30 PM
Hilton Salon AWorkshopSixth Annual Workshop for the Energy Efficient HPC Working Group (EE HPC WG)
Mon, 16th November
09:00 AM - 05:30 PM
Hilton Salon BWorkshopESPT2015: Extreme-Scale Programming Tools
Wed, 18th November
05:30 PM - 07:00 PM
Hilton Salon CBirds of a FeatherOpenACC API 2.5: User Experience, Vendor Reaction, Relevance, and Roadmap
Thu, 19th November
02:00 PM - 02:30 PM
18ABPaperState of the Practice: Measuring Systems
Thu, 19th November
05:15 PM - 07:00 PM
Level 4 - LobbyPosterA Deadlock Detection Concept for OpenMP Tasks and Fully Hybrid MPI-OpenMP Applications
Thu, 19th November
05:15 PM - 07:00 PM
Level 4 - LobbyPosterIncreasing Fabric Utilization with Job-Aware Routing
Thu, 19th November
05:15 PM - 07:00 PM
Level 4 - LobbyPosterAnalysis of Node Failures in High Performance Computers Based on System Logs
Thu, 19th November
05:15 PM - 07:00 PM
Level 4 - LobbyPosterVerification of Resilient Communication Models for the Simulation of a Highly Adaptive Energy-Efficient Computer

Screenshots

  • Counter Data
  • Message Matrix
  • Performance Data Overlay
  • Performance Data Overlay
  • Multiple Timelines
  • Understanding your program?
  • Message Stats
  • Counter-Call-Tree
  • Process Stats
  • Summary
  • Function summary