A Review on Proprietary Accelerators for Large Language Models

2025/03/12 by Park, Sihyeong, Jemin Lee, Lee, Jemin +3
Computer Science · #Big Data and Digital Economy #FOS: Computer and information sciences #Hardware Architecture (cs.AR) #Natural Language Processing Techniques #Performance (cs.PF) #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2503.09650

openalex publication_date 2025/03/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

With the advancement of Large Language Models (LLMs), the importance of accelerators that efficiently process LLM computations has been increasing. This paper discusses the necessity of LLM accelerators and provides a comprehensive analysis of the hardware and software characteristics of the main commercial LLM accelerators. Based on this analysis, we propose considerations for the development of next-generation LLM accelerators and suggest future research directions.

Related

• Hardware Acceleration of LLMs: A comprehensive survey and comparison
• New Solutions on LLM Acceleration, Optimization, and Application
• A General-Purpose Device for Interaction with LLMs
• LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators
• HADES: Hardware Accelerated Decoding for Efficient Speculation in Large Language Models