energyCnV 1.0.3

Track and visualize GPU energy consumption using PyNVML, with MongoDB integration.

EnergyMonitor

Overview

EnergyMonitor is a Python-based tool that tracks GPU power consumption using NVIDIA Management Library (NVML) and stores energy usage data in a MongoDB database. It allows users to log in, track energy consumption per project, and store run-specific energy data.

Features

• User Authentication: Secure login using bcrypt.
• Energy Monitoring: Tracks GPU power usage using NVML.
• Project-Based Tracking: Associates energy consumption data with different projects.
• Database Storage: Saves energy consumption history in MongoDB.

Installation Requirements

Before using the tool, install the required dependencies:

pip install pymongo bcrypt pynvml

Ensure you have:

• A MongoDB database (e.g., MongoDB Atlas or a local instance).
• NVIDIA GPU and drivers installed.
• NVML library (included in the NVIDIA drivers).

Usage

1. Import and Initialize

from energy_monitor import EnergyMonitor
monitor = EnergyMonitor()

2. Login

Before tracking energy consumption, log in with:

monitor.login("username", "password")

3. Start Monitoring

Start tracking energy consumption for a specific project:

monitor.start("ProjectName")

If the project does not exist, it prompts the user to create one.

4. Stop Monitoring

To stop tracking and save energy data:

monitor.stop()

This logs the total energy used and duration to the database.

5. Update Energy Consumption

This function continuously updates power consumption:

monitor.update_energy()

This should be run in a loop while monitoring is active.

Functions

login(username, password)

Inputs:

• username (str): The username of the user.
• password (str): The corresponding password.

Functionality:

• Authenticates the user with a hashed password.
• Stores the user ID after successful login.

start(project_name)

Inputs:

• project_name (str): Name of the project to track energy usage.

Functionality:

• Initializes GPU energy monitoring.
• Checks if the project exists; if not, prompts user to create one.

stop()

Functionality:

• Stops monitoring and logs energy consumption data.
• Stores run details like duration, timestamp, and energy used.

update_energy()

Functionality:

• Continuously updates GPU energy consumption while running.
• Runs in a loop to track power usage per second.

Database Structure

Each user document in MongoDB follows this structure:

{
    "_id": ObjectId("..."),
    "username": "user123",
    "password": "hashed_password",
    "projects": {
        "ProjectName": {
            "run1": {
                "timestamp": "2025-03-08T12:00:00Z",
                "duration": 3600,
                "energy": 500000,
                "energy_kwh": 0.138
            }
        }
    }
}

Notes

• Make sure you log in before starting monitoring.
• Ensure your GPU supports NVML.
• The program exits on incorrect login.

License

This project is open-source. Feel free to modify and extend it!