Architecture design and optimization strategy of heterogeneous AI computing power operation platform

doi:10.12267/j.issn.2096-5931.2022.03.002

Information and Communications Technology and Policy ›› 2022, Vol. 48 ›› Issue (3): 7-12.doi: 10.12267/j.issn.2096-5931.2022.03.002

Previous Articles Next Articles

Architecture design and optimization strategy of heterogeneous AI computing power operation platform

GUO Liang¹, ZHAO Jinghua¹, ZHAO Jizhuang²

1. Cloud Computing & Big Data Research Institute, China Academy of Information and Communications Technology, Beijing 100191, China
2. Research Academy of China Telecom, Beijing 102200, China

Received:2022-01-30 Online:2022-03-15 Published:2022-03-25

Abstract

Abstract:

Heterogeneous AI computing power operation platform is an important carrier for efficient and reliable processing and application of diversified data, which can meet the diversified demands of various upper applications for computing resources and computing resources. This paper introduces the formation mechanism, technical architecture and functional composition of heterogeneous AI computing power operation platform, and suggests that heterogeneous AI computing power operation platform could be optimized and developed from the three aspects of technical ability improvement, application ecology improvement and standardized evaluation.

Key words: heterogeneous AI computing power operation platform, heterogeneous adaptation, technology fusion, application of wisdom

CLC Number:

TP393.0

GUO Liang, ZHAO Jinghua, ZHAO Jizhuang. Architecture design and optimization strategy of heterogeneous AI computing power operation platform[J]. Information and Communications Technology and Policy, 2022, 48(3): 7-12.

Add to citation manager EndNote|Ris|BibTeX

URL:

http://ictp.caict.ac.cn/EN/10.12267/j.issn.2096-5931.2022.03.002

http://ictp.caict.ac.cn/EN/Y2022/V48/I3/7

Figures/Tables 2

References 12

[1]	开放数据中心委员会. 数据中心算力白皮书[R], 2020.
[2]	IDC,浪潮. 2021—2022年中国人工智能计算力发展评估报告[R], 2021.
[3]	中国联通算力网络产业技术联盟. 异构算力统一标识和服务白皮书[R], 2021.
[4]	工业和信息化部. 工业和信息化部关于印发“十四五”大数据产业发展规划的通知[EB/OL]. (2022-01-18)[2022-01-30]. http://www.gov.cn/zhengce/zhengceku/2021-11/30/content_5655089.htm.
[5]	工业和信息化部. 工业和信息化部关于印发“十四五”信息通信行业发展规划的通知[EB/OL]. (2022-01-18)[2022-01-30]. http://www.gov.cn/zhengce/zhengceku/2021-11/16/content_5651262.htm.
[6]	江西省人民政府. 江西省人民政府办公厅关于印发江西省“十四五”新型基础设施建设规划的通知[EB/OL]. (2022-01-18)[2022-01-30]. http://www.jiangxi.gov.cn/art/2021/11/8/art_4968_3711667.html.
[7]	徐敬蘅. 面向异构系统的大气模式并行优化方法研究[D]. 清华大学, 2019.
[8]	郑宗生, 胡晨雨, 姜晓轶. 基于改进的最大均值差异算法的深度迁移适配网络[J]. 计算机应用, 2020, 40(11):3107-3112.
[9]	阳王东, 王昊天, 张宇峰, 等. 异构混合并行计算综述[J]. 计算机科学, 2020, 47(8): 5-16+3.
[10]	孟伟, 袁丽雅, 韩炳涛, 等. 深度学习推理侧模型优化架构探索[J]. 信息通信技术与政策, 2020, 46(9):42-47.
[11]	艾瑞咨询. 2021年中国数据库行业研究报告[R], 2021.
[12]	王月, 柯芊. 智能计算中心: 人工智能时代的算力基石[J]. 中国电信业, 2021(S1):11-15.

Architecture design and optimization strategy of heterogeneous AI computing power operation platform

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 2

References 12

Related Articles 2

Recommended Articles

Metrics

Comments

[1]	ZHANG Ming, WANG Xudong, ZHAI Shuang. Green energy-saving technology of data storage in new data center in the background of carbon peaking and carbon neutrality [J]. Information and Communications Technology and Policy, 2022, 48(3): 69-73.
[2]	Bo YANG. Promoting development of digital economy by improving efficiency and performance of internet network [J]. Information and Communications Technology and Policy, 2022, 48(1): 69-74.