Efficiency Analizer

Python tooling to run benchmark suites on multiple remote devices, collect energy/thermal data through an APCape power board, and generate per-test CSVs and plots.

How it works

• main.py iterates over the configs defined in configs.py, runs each test suite on the target devices via SSH, and streams power readings from the APCape (default APCape8L).
• Power/temperature/frequency samples are written to CSV, then matplotlib produces summary plots for each test and a combined view.
• Each config in all_configs/ defines the commands to run, upload/delete rules, and the devices involved.

Requirements

• APCape power board reachable at the configured server/port (defaults are in each config, e.g. beaglebone.local:6526 and APCape8L).
• SSH access to the target devices defined under all_configs/ (key-based auth recommended).
• Python 3.10+ with the packages in requirements.txt (pip install -r requirements.txt).
• pmlib server running on the APCape host.

Setup

1. (Optional) Create a virtualenv:

   python -m venv .venv
   .\\.venv\\Scripts\\activate

2. Install deps:

   pip install -r requirements.txt

3. Review device settings in all_configs/devices.py and the individual config files. Update IPs, usernames, and temp thresholds to match your lab.
4. Enable/disable suites by editing the configs mapping in configs.py (remove entries you don’t want to run).

Running

python main.py

Results land under all_configs/results/<run-group>/<config-name>-<config-test-name>/ with:

• *.csv power/temp/frequency snapshots
• plots/ PNG summaries (per test and aggregate)
• stdout/ folders if enabled by the config

Creating new configs

Copy an existing file in all_configs/, adjust the tests dict and devices, then add it to configs in configs.py.

License

MIT License. See LICENSE.