Skip to content

Energy-Aware Profiling Tool Developer Guide¤

Welcome to the Developer Guide for our Energy-Aware Profiling Tool! This guide will walk you through the process of cloning the repository, customizing it for your needs, and adding new file modules to support additional device measurements.

Cloning the Repository¤

To get started, follow these steps to clone the repository:

  1. Open a terminal or command prompt.
  2. Navigate to the directory where you want to clone the repository.
  3. Run the following command: 
    git clone https://github.com/HPC-CRI/EA2P

Customizing the Tool¤

After cloning the repository, you may want to customize it according to your specific requirements. Here are some common customization tasks:

Configuration¤

  1. Modify the default configuration file config_energy.json located in the ea2p/src directory to adjust settings such as sampling interval, measurement thresholds, etc.

Visualization output¤

  1. Customize the output components named __record_data_to_file in the ea2p/src/power_meter.py file to tailor the output representation of energy profiling data as CSV, excel, json saving.

RAPL path for Intel Users¤

  1. Customize the intel RAPL directory to suit your energy profiling needs depending of your architecture using ea2p/src/utils.py files and set some others system configurations.

Adding a New File Module¤

If you need to support new device measurements, you can add a new file module to handle the data. Follow these steps:

  1. Create a new Python file in the ea2p/src/ directory for your new module. For example, FPGA_module.py.

  2. Implement the necessary functions to parse and process the data from the new device.

  3. Update the __init__.py file in the ea2p/src directory to include your new module. For example:

from .FPGA_module import XilinxPower

  1. Configure the ea2p/src/wrapper.py module to support your energy module. Especially the get_power_consumption function. Also, adapt the __set_power method in the same file to support auto-detection of your devices and make some initializations eventually.

  2. Finally, use the PowerMeter instance to profile your application by specifying your device in the list of devices in configuration the file.