Energy Consumption Analysis of Large Language Models Across CPU and GPU Using Diverse Metric Types
Bibliographic record
Abstract
With the growing use of artificial intelligence models, such as Large Language Models (LLMs) and transformers, in both academic and industrial fields, energy consumption has risen, making energy efficiency increasingly important. Most existing studies use resource utilization as the main indicator for evaluating energy consumption of machine learning models. However, this approach may overlook other key factors influencing energy efficiency. Moreover, the prediction models proposed by many related studies lack coverage of a wider range of tasks or models. This paper proposes a multi-level energy analysis framework. First, we integrate data collection and correlation analysis to study the relationships between energy consumption and specific dynamic and static metrics. Second, we expand the scope of energy and performance analysis to include a broader range of LLMs and tasks. Finally, we combine the implementations of the previous objectives to construct and validate our energy models. Our experimental results demonstrate that our models robustly predict energy consumption across diverse NLP tasks. Developing such a framework helps more effectively analyze the energy efficiency of LLMs, understand energy consumption patterns, and provide developers with actionable strategies for optimizing energy use in AI systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".