Authors information

Name: Stefanos Georgiou

Affiliation: Athens University of Business and Economics

Contact details: sgeorgiou@aueb.gr, sgeorgiou@singularlogic.eu, and stefanos1316@gmai.com

Description

This reposotory was build for research purpose. Our aim is to compare the energy consumption of different tasks writtern in a variety of programming languages. In order to achieve our aim we used a dataset publicly available from Rosetta Code. We found all the collected tasks implemented in different langauges from a Git repo. Since Rosetta Code contains the amount of 655 programming languages we decicde to limit our scope and include only the most used ones.
To this line, we collected the 14 most used languages found from tiobe index June 2017. Thus, we developed a number of bash scripts to manage our dataset, compile, execute, and plot results out of the collected tasks.

Downloading

Since we are using a submodule in this repository we suggest the use of the following command:

  $ git clone --recursive https://github.com/stefanos1316/Rosetta-Code-Research.git

In case you used simple clone command, add the submdodules using the following commands inside the cloned repo.

  $ git submodule init

  $ git submodule update

How to Execute

First browse to Scripts directory and proceed as follows:

1. Execute the script.cleanAll in order to prepare the existing dataset and make it case insensitive by chaning all files from upper to lower case.
2. Execute the script.createNewDataSet to have a filtered dataset with all the selected tasks and programming languages that we will examine for this research.
3. Some of the Tasks cotains more that one implementation of the same languages, thus we had to manually drive through the directories and remove some of the executables. In the case of Java we also had to change the .java files name since it had more than one file of a selected task.
4. Afterwards, we had to add additional code inside each file in order to force a task to run around a million of times (some tasks are so small and are finishing faster than 1 second, thus our power analyzer cannot capture those results).
5. Execute the script.compileTasks in order to compile all tasks.
6. Execute the script.executeTasks[Remotly|Locally] in order to execute the tasks and collect results (in our case we used Watts Up Pro). On running the script.executeTasksRemotly, it is necessary to have a remote host with ssh where the Watts Up Pro is connected, otherwise the script.executeTasksLocally can be executed without the need of the remote host.
7. Execute the script.createPlottableData that creates the file with all the results from all the measurements (Energy/Performance). In the end, this script will call the script.plotGrpahs that is responsible to create plot the graphs.
8. The script.plotGraphs is executed by the script.createPlottableData and can provide graphs for both performance and energy consumption. Moreover, there are some default setting found in the end of script.createPlottableData, if anyone would like to change them and plot using different configuration parameters.

Note: For more information for executing scripts add the --help command line argument.