Download and Install

Score-P

First, you need to download and install Score-P. On a typically Debian based System you simply need to type the following in your shell:

wget https://www.vi-hps.org/cms/upload/packages/scorep/scorep-6.0.tar.gz
tar -xf scorep-6.0.tar.gz
cd scorep-6.0/
mkdir build
cd build
../configure --enable-shared
make 
sudo make install

This will install Score-P into the typical locations on your System.

You can change the install location by specifying

../configure --prefix=/some/location/

Please be sure to set the $LD_LIBRARY_PATH and $PATH adequately. Please also make sure, that the gcc-compiler plugin is used. E.g.

scorep-info config-summary | grep -A3 "Score-P (GCC plug-in):"

should deliver something like:

    Score-P (GCC plug-in):
      GCC plug-in support:      yes, using the C++ compiler and -I/usr/lib/gcc/x86_64-linux-gnu/9/plugin/include
      Compiler used:            g++-9 -std=c++11

If you get something else, please install the gcc plugin headers, e.g. by doing

sudo apt install gcc-9-plugin-dev

for gcc 9.

Ubuntu

For Ubuntu LTS based systems, there is a PPA available: https://launchpad.net/~andreasgocht/+archive/ubuntu/scorep. However, please be aware, that this PPA is not officially supported.

Cube

To visualise the profile you might need CUBE. To install please do:

wget http://apps.fz-juelich.de/scalasca/releases/cube/4.4/dist/cubegui-4.4.4.tar.gz
tar -xf cubegui-4.4.4.tar.gz
cd cubegui-4.4.4
mkdir build
cd build
../configure
make 
sudo make install

This will install Cube into the typical locations on your System.

You can change the install location by specifying

../configure --prefix=/some/location/

However, please be sure to set the $LD_LIBRARY_PATH and $PATH adequately.

Installing the Score-P Python bindings

To install the python bindings you simply need to clone the repository, and install it using pip:

git clone https://github.com/score-p/scorep_binding_python
cd scorep_python_bindings/
pip3 install .

The bindings are supposed to be python 2.7 compatible. However, I really recommend using Python 3.5 upwards.