Installation and Running of SKA SDP Benchmark tests
ReFrame is developed in Python and hence, Python 3.6+ and several python modules are needed to run the tests. We strongly recommend to use conda environment to create virtual environments. Installation instructions for Spack is provided separately here in Installation section.
Clone this repository:
git clone https://gitlab.com/ska-telescope/sdp/ska-sdp-benchmark-tests.git
Install conda (if not already installed):
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh bash Miniconda3-latest-Linux-x86_64.sh
More instructions on installing conda can be found here
Create and activate
ska-sdp-benchmark-testsenvironment. The conda environment files are placed in folder
share/conda/in the root of the repository. Currently only x86 and ppc64le architectures are supported. For instance, if the system under test has x86 architecture, we can create conda environment using:
cd ska-sdp-benchmark-tests conda env create -f share/conda/environment-x86.yml conda activate ska-sdp-benchmark-tests
Install all dependencies (like ReFrame) using the share/setup-env.sh script:
This script installs the correct ReFrame version, wraps the reframe binary around a shell function and modifies the
PYTHONPATH appropriately to get all modules on the path. Once this environment is set, we can invoke
reframe from anywhere and run tests by providing an absolute path to test files.
It also fetches and installs the SDP Perfmon toolkit, compiles it and installs it into the current conda environment.
If the perfmon is installed, this script requires a working compiler toolchain and CMake installation.
If you don’t want to install the perfmon, set
"INSTALL_PERFMON=no" before sourcing the script.
The installation of SDP Perfmon needs CMake >= 3.17.5.
Running SKA SDP Benchmark tests
As stated in the Installation instructions, once we set up environment for SKA SDP Benchmark tests using the script
setup-env.sh, we can invoke reframe from anywhere in the system. All the documentation provided in the following subsections assume that this environment is not set up and hence all tests are invoked from repository root. For example, if a test
mytest.py is located at an absolute path
/path/to/mytest.py, we can execute that test using either:
cd ska-sdp-benchmark-tests conda activate ska-sdp-benchmark-tests source share/setup-env.sh reframe -c /path/to/mytest.py -r