CPPJoules is a energy measurement tool designed to measure the energy consumption during the execution of a piece of C++ code. CPPJoules leverages the Intel Running Average Power Limit (RAPL) interface, Intel Powergadget, Nvidia Management Library (NVML) for reporting energy consumption of different system-on-chip (SoC) power domains.

Requirements

Github Repo

Direct Installer Link

For Linux

1. Click here to download CPP Joules for Linux

The file installer.sh will be downloaded.

2. Now naviagte to the folder where the file installer.sh has been downloaded .Open that folder in you terminal run the command :
   chmod +x installer.sh
   ./installer.sh

   Copy code

   You will be prompted to enter your user password

3. The library will be installed in /usr/bin
4. Check g++ version by running this code:
   g++ --version

   Copy code

If the version is less than v11.0 then run the below code :

sudo apt update
sudo apt install g++-11

Copy code
5. Now create a test.cpp file and copy the below test code :
//test.cpp
#include<cppJoules.h>

int main(){
    EnergyTracker tracker;
    tracker.start();
    //API Call
    tracker.stop();
    tracker.calculate_energy();
    tracker.print_energy();
    return 0;
}
Copy code

Or you can try with one of our test scripts from here

6. Now compile the code with this code :
   g++ test.cpp -lCPP_Joules

   Copy code
7. And run it
   sudo ./a.out

   Copy code
8. You will get the output as :
   Time 9.00326
   core-0 8386636
   nvidia_gpu_0 164299
   package-0 66858350
   uncore-0 33936
   

For Windows

1. Click here to download CPP Joules for Windows

The file installer.bat will be downloaded.

2. Right click on the installer.bat file and click run as administrator.
3. The folder cppJoules will be created.
4. Now place this cppJoules folder in the same location where your cpp file is present.
5. Example :
• Create a folder testCPPJoules and place the above cppJoules folder within it.
• Now create a test.cpp file within the folder testCPPJoules .
• Now you can directly use the header in your cpp file
//test.cpp
#include"cppJoules/include/cppJoules.h"
#include<iostream>

int main(){
    EnergyTracker tracker;
    tracker.start();
    //API Call
    tracker.stop();
    tracker.calculate();
    tracker.print_energy();
    return 0;
}
Copy code

Or you can try with one of our test scripts from here

6. Now compile the file with this command :
   g++ test.cpp -lCPP_Joules -L./cppJoules

   Copy code
7. Then run the executable :
   ./a.exe

   Copy code
8. The output
   Time 9.02878
   DRAM-0 0
   GT-0 0
   IA-0 16
   Processor-0 50
   nvidia_gpu_0 30633
   
   The values of DRAM-0 , GT-0 and IA-0 are in joules. The value of nvidia_gpu_0 is in microjoules.

   Note : The g++ compiler version has to be above v11.0.0. Install from here
   Note : If PowerGadget v3.6 is not present, then install from here

• Measurement of Energy

  CPPJoules leverages Intel's RAPL and NVIDIA's NVML libraries to measure energy consumption for both the CPU and GPU components. For CPU energy measurement, the tool utilizes Intel's RAPL interface, which provides detailed energy traces data across various domains, including the CPU package, core, uncore, and DRAM. This is achieved through the Powercap interface on Linux and Intel Power Gadget on Windows.

  
  

Test code for array concatenation ( in linux )



Test code for bubble sort ( in linux )