Privacy & Legal Notice

Livermore Computing Training Announcement

Title: New Features in the TAU Performance System
Dates / Locations: General Presentation: Jan 22, 2014, 10am - noon; Building 453, Room 1012 (Black Diamond Room)

Individual Sessions: Time slots for these must be scheduled in advance
January 22, 2014, 1pm-5pm: Building 453, Room 1010
January 23, 2014, 8am-noon, 1pm-5pm: Building 453, Room 1016

Presenter: Sameer Shende, University of Oregon / Paratools, Inc.
Description: The presentation portion of this seminar will highlight the new features in TAU to better support the needs of computational scientists to evaluate parallel application performance.
  • TAU's source analysis and instrumentation capabilities have been enhanced significantly by integrating the ROSE compiler from LLNL in the PDT package.
  • TAU features a wrapper generator for instrumenting libraries without accessing the source code.
  • TAU now features support for event-based sampling to pinpoint performance bottlenecks at the loop and statement level.
  • TAU's memory debugging capabilities now include support for evaluating heap memory utilization, memory leak detection, and runtime bounds checking.
  • TAU supports power profiling using the RAPL interface on modern Intel processors and can identify the location and nature of power spikes in applications using marker events.
  • It supports accelerators such as the Intel Xeon Phi coprocessor, NVIDIA, and AMD GPUs with support for instrumentation of OpenMP, CUDA, OpenCL, and OpenACC programs.
  • TAU's OpenMP instrumentation capabilities have been enhanced using the OpenMP Tools API (OMPT) interface for tracking the performance of each OpenMP construct at the library layer.
  • It features updated support for rewriting binaries using DyninstAPI (U. Wisconsin, UMD), PEBIL (SDSC), and MAQAO (Intel Exascale Labs) packages.
  • Using the Score-P measurement library, TAU can generate callpath profiles and event-traces natively in the OTF2 format that does not require merging or conversion prior to loading in Vampir. The performance data generated can be stored in TAUdb, a new performance data management framework for cross-experiment data analysis and data mining.

Following the general presentation, attendees can schedule individual meeting times with the speaker, which can include hands-on work with their codes/projects (bring your own laptop if you wish to do this). These individual sessions must be scheduled in advance by contacting Blaise Barney at 422-2578 or

Additional information about TAU can be found at

About the Speaker: Sameer Shende serves as the Director of the Performance Research Laboratory at the University of Oregon and the President of ParaTools, Inc. He received his Ph.D. from the University of Oregon and B.Tech from the Indian Institute of Technology, Bombay. He has helped develop the TAU Performance System, Program Database Toolkit (PDT), Parallel Tools Runtime Environment (PToolsRTE), and the HPCLinux distribution. His areas of interest include tools and techniques for performance instrumentation, measurement, performance analysis, runtime systems, and compiler optimizations."
Fee: No cost
Level/Prerequisites: Having a basic familiarity of performance tools for High Performance Computing would be beneficial.
Registration: No registration is required for the general presentation, however those wishing to schedule individual sessions with the speaker must do this in advance by contacting Blaise Barney at 422-2578 or
Questions: Please contact Blaise Barney at 422-2578 or