<![CDATA[This study aims to evaluate the performance and energy efficiency of distributed unsupervised learning algorithms on two types of clusters, namely Single Board Computers (SBC) and Personal Computers (PC), using Apache Spark. Two algorithms were tested—k-means and Gaussian Mixture Model (GMM)—executed across varying dataset sizes and numbers of processor cores to observe scalability. The results show that PCs consistently achieved faster execution times, particularly with k-means on large datasets. On the other hand, SBCs demonstrated higher energy efficiency in all scenarios, with energy savings of up to 93% for k-means and 86% for GMM compared to the highest-consumption configuration on PC. These findings affirm the potential of SBCs as a low-power and cost-efficient solution for green or sustainable computing, particularly for learning, academic experimentation, and small-scale edge computing development, and are relevant to sustainability efforts through their contribution to the Sustainable Development Goals (SDGs).]]>
Copyrights © 2026
