Parallel Performance Evaluation Using TAU
August 21-22, 2019
Individual Meetings (by appointment):
Parallel Performance Evaluation Using TAU
To meet the needs of computational scientists to evaluate the performance of their parallel, scientific applications, we present the TAU Performance System and its interfaces to other tools such as PAPI, Score-P, Scalasca, OTF and Vampir. This one day workshop will cover performance evaluation of applications on Tri-lab OCF platforms. This workshop will focus on performance data collection, analysis, and performance optimization. After describing and demonstrating how performance data (both profile and trace data) can be collected in a straightforward manner using TAU's (Tuning and Analysis Utilities) automated instrumentation, the workshop will cover how to analyze the performance data collected and drill down to find performance bottlenecks and determine their causes. The workshop will include some sample codes that illustrate the different instrumentation and measurement choices available to the users. Topics will cover generating performance profiles and traces with OpenMP instrumentation using the OpenMP Tools API with Intel compilers, memory utilization, I/O, and hardware performance counters data using PAPI. Hardware counter data can show not only which routines are taking the most time, but why? For example, because of cache misses, TLB misses, excess address arithmetic, or poor branch prediction behavior. We will demonstrate scalable tracing using Score-P and OTF and visualization using the Vampir trace visualizers. Performance data analysis using ParaProf and PerfExplorer will be demonstrated using TAU's database technology (TAUdb). The workshop will also feature cross experiment analysis including comparing the effects of multi-core architectures on code performance. The following new and advanced features will be covered: Using TAU in the Extreme-Scale Scientific Software Stack (E4S https://e4s.io) container environment.
We will attempt to collect and analyze performance data for additional user codes during the hands-on portion of the workshop. Users and developers are welcome to contact the instructor ahead of time to begin collecting data so as to have it on hand for the workshop.
Additional information about TAU can be found at tau.uoregon.edu
August 21 Workshop: For hands-on exercises, attendees will need to bring their own laptop. Laptop requirements:
Temporary workshop accounts will be provided for accessing Livermore Computing (LC) clusters during the workshop. Attendees who have their own LC computer accounts may use those if desired.
August 22 Individual Sessions: Interested developers can schedule individual meeting times with the instructor, 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 firstname.lastname@example.org
|Agenda:||August 21 Workshop: (T1879 classroom D)
August 22 Individual Sessions: (B453 R1020)
|Bio:||Sameer Shende (email@example.com) 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.|
August 21 Workshop: You must register in advance. Registration is limited to LLNL employees, students and collaborators. Please contact Blaise Barney (925-422-2578 / firstname.lastname@example.org) to reserve a seat.
August 22 Individual Sessions: Interested developers can schedule one-on-one meeting times with the instructor. These individual sessions must be scheduled in advance by contacting Blaise Barney (925-422-2578 / email@example.com).
|Level/Prerequisites||Familiarity with running parallel programs in an HPC environment is highly recommended.|
|Questions?||Please call or send e-mail to Blaise Barney (925-422-2578 / firstname.lastname@example.org)|