Abstract

The rapid growth of large-scale distributed computing systems—such as cloud data centers, high-performance computing clusters, and edge-to-cloud infrastructures—has driven unprecedented computational capabilities. However, these systems also consume considerable energy, contributing to operational costs, environmental impact, and thermal management challenges. Energy-efficient algorithm design and optimization strategies aim to minimize energy consumption while maintaining performance, scalability, and reliability. This paper investigates the theoretical foundations and practical techniques for designing algorithms that are energy-aware, workload-adaptable, and cognizant of diverse resource constraints. By synthesizing research across scheduling, load balancing, data locality, and resource provisioning, we explore methods that dynamically adjust computational behavior to current energy states. The study also analyzes software–hardware co-design, predictive models, and adaptive systems that adjust parameters at runtime to optimize energy utilization. Through comprehensive literature review, modeling, and evaluation, we identify trade-offs between performance and energy savings, the impact of communication overhead on power usage, and the role of machine learning in workload prediction. Results highlight that energy-efficient strategies can significantly reduce power consumption without substantial performance loss. We conclude with recommendations for future research integrating intelligence, heterogeneity, and sustainability goals.