Allinea Performance Reports installations at NSC


Directory list:

VersionDescription
18.0.2 Allinea reports version 18.0.2
5.1 Allinea reports version 5.1
6.0.4-47391 Allinea reports version 6.0.4-47391
6.1 Allinea reports version 6.1
7.0 Allinea reports version 7.0

Level of support

Tier 2 NSC has only limited experience with this software, but we will try to help as much as possible. We have ran some tests, for example, if the program comes with a test suite, but they may be far from exhaustive. We will try to install and test new versions, as soon as we can.

Please see the page describing our software support categories for more information.
You can also contact support@nsc.liu.se for further information.

Official homepage: www.allinea.com/products/performance-reports

Allinea Performance Reports is an advanced profiling tool for scalar, multi-threaded and large-scale parallel applications. Features are described on the official homepage.

Please contact NSC Support if you have any questions or problems.

How to run

Load the module corresponding to the version you want to use, e.g.,

   module load allinea-reports/7.0

It adds the performance reports binary path.

Simply run your application through perf-report to collect the overall statistics of your code in the way of:

perf-report ${YOUR_SIMULATION_COMMAND}

Basically, perf-report is designed to provide the simple run-time performance collection of large-scale MPI applications. Therefore, report generation for the MPI simulation is straightforward:

perf-report mpiexec.hydra --bootstrap slurm -np ### ./your_executable

If you intend to collect the performance measurement of your serial run, add ‘–no-mpi’ option. That is to say,

perf-report --no-mpi ./your_executable

When using our mpprun wrapper, you have to launch your application as following:

mpprun -pr ./your_executable

To learn more about mpprun, please consult our mpprun documentation or the more detailed mpprun tutorial section, where the performance report is covered in Example 4: Profiling a parallel program with perf-report.

At the end of the simulation, perf-report produces report files, one HTML and one text-based file which are identical. The report presents the overall performance of your code, including cost for scalar/vector operations, MPI cost, I/O overhead, memory usage, and thread parallelism.

This HTML report is visible by any web browser. Otherwise, you can read the text-based report.

For ThinLink user: If you intend to load the HTML report in the ThinLink session, you shall choose “Konqueror” as the browser for opening this file. Simple double-clicking from the file manager will show you the source of this HTML file by “Geany” software (which is set to operate HTML files by default).