The demand for energy-efficient computing continues to rise as embedded and energy-constrained systems dominate modern applications. Traditional performance metrics such as time and space complexity fail to capture computation cost when energy is the primary constraint. This study introduces an energy-aware benchmarking framework focused on three primary data structures, which are HashMap, Balanced Tree (B-Tree), and Skip List. Latency, throughput, and energy consumption were used to define the performance metrics of interest. The benchmarking was conducted on a Linux operating system through various workloads and dataset sizes with energy consumption data collected through the Running Average Power Limit (RAPL) counters. Related performance metrics such as CPU frequency, cache misses, memory footprint, and I/O usage were collected through system profiling utilities. Experimental results included 29,997 benchmark samples; the XGBoost classifier achieved a very high structure identification accuracy of 99.53%, while RandomForest achieved 98.77%. Regression models attained R² = 0.966 for energy prediction and R² = 0.691 for latency prediction. The proposed energy-aware benchmarking framework provides a systematic method to quantify trade-offs between speed and energy cost, identify thresholds for data structure suboptimality, and enable dynamic, energy-efficient data structure selection in constrained environments.