Characterizing Communication Patterns in Distributed Large Language Model Inference
IEEE Micro · 2026-07 · 46 (4) · 85-94
DOI: 10.1109/mm.2026.3672150 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Micro, 2026-07. Verfasst von Lang Xu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 12 Literaturangaben. Gefördert von: National Science Foundation, XRAC.
Abstract (Originalsprache)
Large language models (LLMs) built on transformer architectures have transformed natural language processing, achieving remarkable performance across diverse applications. While distributed inference frameworks enable practical deployment of these models, inter-GPU communication creates significant performance constraints that limit service quality in real-world systems. This article investigates communication dynamics in distributed LLM serving, analyzing how various parallelization approaches coordinate data exchange between GPU workers during inference. We study transformer-based models as representative examples of contemporary architectures widely used in operational deployments. Our work combines detailed profiling measurements with predictive analytical models to characterize communication behavior across different parallelization configurations. These insights offer practical recommendations for selecting appropriate parallelization schemes in production LLM services and identify key opportunities for optimizing inference frameworks and communication infrastructure.
Förderung
- National Science Foundation
- XRAC
Zitieren
APA
Xu, L., Suresh, K. K., Anthony, Q., Alnaasan, N., & Panda, D. K. (2026). Characterizing Communication Patterns in Distributed Large Language Model Inference. IEEE Micro, 46(4), 85-94. https://doi.org/10.1109/mm.2026.3672150
Vancouver
Xu L, Suresh KK, Anthony Q, Alnaasan N, Panda DK. Characterizing Communication Patterns in Distributed Large Language Model Inference. IEEE Micro. 2026;46(4):85-94. doi:10.1109/mm.2026.3672150
BibTeX
@article{Xu2026,
author = {Xu, Lang and Suresh, Kaushik Kandadi and Anthony, Quentin and Alnaasan, Nawras and Panda, Dhabaleswar K.},
title = {Characterizing Communication Patterns in Distributed Large Language Model Inference},
journal = {IEEE Micro},
year = {2026},
volume = {46},
number = {4},
pages = {85--94},
doi = {10.1109/mm.2026.3672150}
}
EUVIMED