面向工业大模型的算力网络架构与关键技术
Computing Power Network Architecture and Key Technologies for Large-Scale Industrial Models
沈昕炎 1林亚捷 1许方敏 1赵成林1
作者信息
摘要
随着第四次工业革命的推动,工业生产逐渐迈入了数字化和智能化的时代.在这一时代背景下,工业大模型作为推动工业创新的核心引擎,扮演着越来越重要的角色.同时,工业大模型的广泛应用也给算力网络提出了更为复杂和严峻的实时性需求.本文深入研究了工业大模型在数字化转型中的关键问题,着眼于其对算力网络的需求,分析了工业大模型在实时性方面的复杂需求,提出了面向工业大模型的算力网络架构,并对算力网络的关键技术进行了介绍,为工业大模型的高效运行提供了技术支持,为工业数字化、智能化的快速发展提供了助力.
Abstract
As the Fourth Industrial Revolution progresses,industrial production has gradually entered the era of digitization and intelligence.In this era,the industrial large-scale models,as the core engine driving industrial innovation,plays an increasingly important role.At the same time,the widespread application of industrial large-scale models also poses more complex and severe real-time requirements for computing power networks.This article delves into the key issues of industrial large-scale models in digital transformation,focusing on their demand for computing power networks.It analyzes the complex real-time requirements of industrial large-scale models,proposes a computing power network architecture for industrial large-scale models,and introduces the key technologies of computing power networks,providing technical support for the efficient operation of industrial large-scale models and promoting the rapid development of industrial digitization and intelligence.
关键词
工业大模型/算力网络/算力感知/分布式计算Key words
Industrial large-scale models/Computing power networks/Computing power perception/Distributed computing引用本文复制引用
出版年
2024
NETL
NSTL
万方数据