How does Parallelism impact the Energy Efficiency and Performance of High-Performance Scientific Software? The Case of Haddock

Replication Package of the study:

Title: "How does Parallelism impact the Energy Efficiency and Performance of High-Performance Scientific Software? The Case of Haddock"
Authors: Vincenzo Stoico • Dmitrijs Voronovs • Ivano Malavolta • Patricia Lago
Paper: preprint_to_publish

Structure of the Repository

• cli: Folder containing the command-line tool. The tool can orchestrate the experiment and query SLURM to retrieve job measurements and data.
• data: Folder including the measurements of our experiment. The files follow different naming conventions. The files called [machine_names]_[file.id].txt contain the measurements sorted per machine, while those named data_[daa|dpp].csv include data per workflow. The Focus Group - HADDOCK.csv file contain the results of the focus group.
• notebooks: Folder containing the Jupyter notebooks to replicate the data analysis.
• runs: Folder including the logs of each run of our experiment.
• requirements.txt: File including the list of packages required for data analysis.

Install Data Analysis packages

Install Jupyter Notebook and required libraries

pip install jupyter
pip install -r requirements.txt

Launch Jupyter Notebook from the upmost folder of this repository (./haddock-tactics-rep-pkg)

jupyter notebook