Offline Energy-Optimal LLM Serving: Workload-Based Energy Models for LLM Inference on Heterogeneous Systems
ACM SIGEnergy Energy Informatics Review · 2024-12 · 4 (5) · 113-119
DOI: 10.1145/3727200.3727217 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM SIGEnergy Energy Informatics Review, online erschienen am 17.04.2025. Verfasst von Grant Wilkins und 2 weiteren Autor:innen. Die Arbeit wurde bisher 38-mal zitiert (Crossref) und nennt 47 Literaturangaben.
Abstract (Originalsprache)
The rapid adoption of large language models (LLMs) has led to significant advances in natural language processing and text generation. However, the energy consumed through LLM model inference remains a major challenge for sustainable AI deployment. To address this problem, we model the workload-dependent energy consumption and runtime of LLM inference tasks on heterogeneous GPU-CPU systems. By conducting an extensive characterization study of several state-of-the-art LLMs and analyzing their energy and runtime behavior across different magnitudes of input prompts and output text, we develop accurate ( R 2 > 0.96) energy and runtime models for each LLM. We employ these models to explore an offline, energy-optimal LLM workload scheduling framework. Through a case study, we demonstrate the advantages of energy and accuracy aware scheduling compared to existing best practices.
Zitieren
APA
Wilkins, G., Keshav, S., & Mortier, R. (2024). Offline Energy-Optimal LLM Serving: Workload-Based Energy Models for LLM Inference on Heterogeneous Systems. ACM SIGEnergy Energy Informatics Review, 4(5), 113-119. https://doi.org/10.1145/3727200.3727217
Vancouver
Wilkins G, Keshav S, Mortier R. Offline Energy-Optimal LLM Serving: Workload-Based Energy Models for LLM Inference on Heterogeneous Systems. ACM SIGEnergy Energy Informatics Review. 2024;4(5):113-119. doi:10.1145/3727200.3727217
BibTeX
@article{Wilkins2024,
author = {Wilkins, Grant and Keshav, Srinivasan and Mortier, Richard},
title = {Offline Energy-Optimal LLM Serving: Workload-Based Energy Models for LLM Inference on Heterogeneous Systems},
journal = {ACM SIGEnergy Energy Informatics Review},
year = {2024},
volume = {4},
number = {5},
pages = {113--119},
doi = {10.1145/3727200.3727217}
}
EUVIMED