High Performance Computing Center North
Scalasca is a free and open-source software for measurement, analysis, and optimization of parallel program performance. It is licensed under the BSD-style license. Scalasca is mostly used for profiling scientific and engineering applications using OpenMP and/or MPI. It supports runtime analysis on supercomputers.
Extrae is the package devoted to generate Paraver trace-files for a post-mortem analysis.
Paraver is a flexible performance analysis tool developed to respond to the need to have a qualitative global perception of the application behavior by visual inspection and then to be able to focus on the detailed quantitative analysis of the problems.
(Click to enlarge the following)
Bring your own code if you have one you want to look at (MPI or OpenMP)!
Paraver can be run on Windows, Linux, and macOS. For the hands-on the participants will run it on their laptops. It can be downloaded from here, and should be easy to install.
Test examples and trace files used in the course can be downloaded from here: tools_170314.tar.gz.
Scalasca will be run on Abisko and/or Kebnekaise, so the participants who wishes to take part of this should make sure they have an account at HPC2N. We will create a course project for this course, so you need not worry about that part. Start by creating an account in SUPR if you do not have one there, and then email me at firstname.lastname@example.org so I can add you to the course project.
Participation is free. Lunch and coffee/tea will be provided.
Please bring a laptop, and as mentioned above, make sure you have an account at HPC2N if you want to participate in the hands-on part of the Scalasca training.
Prerequisites: Basic programming knowledge. Some knowledge of OpenMP and MPI.
Instructors: Jerry Eriksson, Pedro Ojeda May
Time and date: 14. March 2017, 09:00 - 16:00.
Location: MIT-Huset, Umeå University. Room MC323SEM (Directions to MIT huset here).
Deadline for registration: 7. March 2017. Please register using this form. All fields are mandatory.