Usage Guide¶

This guide walks you through the steps to use the Energy Pipeline for detecting energy regressions in your codebase.

🧩 Overview¶

The pipeline automates:

You need to create a config file in YAML format that follows your config.schema.yml structure. This includes:

Example:

TODO add example

Ensure:

Run the system setup:

sudo ./system_setup.sh first-setup
reboot
sudo ./system_setup.sh setup

You can later revert changes:

sudo ./system_setup.sh revert-setup

Before real measurements:

python main.py stability-test

This checks system temperature and performance counter stability.

python main.py measure --config path/to/config.yml

To ensure results follow Git commit order:

python main.py sort path/to/results.csv /path/to/repo sorted_results.csv

To visualize energy trends and regressions:

python main.py plot path/to/sorted_results.csv

This creates one PNG per energy metric (energy-pkg, energy-core, energy-gpu), with:

You can use Docker to isolate the environment:

docker buildx build -t pipeline .
docker run -it --privileged pipeline:latest

Make sure to mount required volumes and copy your config + repo path into the container.

tmux new -s energy
# [run your pipeline]
# Detach with Ctrl+B D
# Reattach with: tmux attach -t energy

Step	Command
Stability test	`python main.py stability-test`
Run pipeline	`python main.py measure --config config.yml`
Sort results	`python main.py sort results.csv /path/to/repo sorted.csv`
Plot results	`python main.py plot sorted.csv`