EN Bereichsnavigation EN

Performance > TAU

 

Quick start guide

The following instructions are for those who want to get Scalasca up and running quickly.

  • module load PrgEnv-xxx where xxx=cray or pgi or intel or gnu.
  • module load tau
  • module help tau ; echo $TAU_MAKEFILE
  • tau_f90.sh -c fcode.f90
  • tau_f90.sh -o myprogram fcode.o
  • /usr/bin/time -p aprun -n128 ./myprogram
  • pprof -s profile.0.0.0 or paraprof

Read on for complete details.

Description

Tau is a performance analysis tool developed by  the University of Oregon, Los Alamos National Laboratory, and Research Centre Juelich for CSCS production systems.


Tau is a portable profiling and tracing toolkit for performance analysis of parallel programs written in Fortran, C, C++.

What can TAU do?

It can generate detailed profiles of the following :

  • MPI communication (calls, buffer sizes, communication patterns)
  • PAPI hardware performance counters
  • Memory used per compute node

In addition, TAU could be configured for MPI and OpenMP combined profiling and I/O profiling. TAU provides interoperability with a range of instrumentation libraries, post-processing and tracing tools.  For instance, TAU MPI traces could be converted to the Open Trace Format (OTF) traces that could be visualized using the Vampir tracing tool.

TAU provides post-processing and analysis tools in two formats: text-based and visualization. The text-based tool called pprof can produce output for individual tasks as well as aggregated information. The visualization tool called paraprof is written using Java therefore offline analysis could also be performed on users personal platforms where these tools are available by copying over the profile files generated by the TAU tool.

Usage

    • First, load the modulefile to get the correct environment variables set :

                       module load tau

    • Compile and link your application for measurement :

                       tau_f90.sh fcode.f90 -o fexe

    or

                       tau_cc.sh -o cexe ccode.c -o cexe

    • Run the instrumented executable :

                       /usr/bin/time -p aprun -n 128 ./fexe

    • Upon successful execution, the performance output files will be generated and will start with "profile". Explore the tau report with

                       pprof -s profile.*

    or

                       paraprof

    For more information, see Troubleshooting below.

    Troubleshooting

    This page describes the basics of Tau. For further information please check :

    To top