Article(id=1329419771717726439, tenantId=1146029695717560320, journalId=1324759914146529478, issueId=1329419577676641015, articleNumber=null, orderNo=null, doi=10.7544/issn1000-1239.202550472, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1750089600000, receivedDateStr=2025-06-17, revisedDate=1766073600000, revisedDateStr=2025-12-19, acceptedDate=1767715200000, acceptedDateStr=2026-01-07, onlineDate=1794887973002, onlineDateStr=2026-11-17, pubDate=1777564800000, pubDateStr=2026-05-01, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1794887973002, onlineIssueDateStr=2026-11-17, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1794887973002, creator=13701087609, updateTime=1794887973002, updator=13701087609, issue=Issue{id=1329419577676641015, tenantId=1146029695717560320, journalId=1324759914146529478, year='2026', volume='63', issue='5', pageStart='1095', pageEnd='1375', issueExtLink='null', onlineDate='null', pubDate='1777564800000', pubDateStr='2026-05-01', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1794887926739, creator='13701087609', updateTime=1794887926739, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext=null, issueFiles=null, downloadFileDto=null}, startPage=1337, endPage=1349, ext={EN=ArticleExt(id=1329419771902275816, articleId=1329419771717726439, tenantId=1146029695717560320, journalId=1324759914146529478, language=EN, title=Lightweight Embedded RDMA Software Stack for High-Speed Sensor Data Acquisition, columnId=1329419589001261979, journalTitle=Journal of Computer Research and Development, columnName=Network Technology, runingTitle=null, highlight=null, articleAbstract=

Remote direct memory access(RDMA)technology is widely used in high-performance computing and intelligent computing scenarios that require high bandwidth and low latency transmission, offering significant advantages over traditional TCP/IP protocol stacks in terms of bandwidth, efficiency, and latency. It has gradually become an industry consensus for high-speed sensors to access computing networks through RDMA technology. However, deploying RDMA in this scenario faces challenges due to the limited resources of embedded sensor nodes, making it difficult to deploy traditional software stacks. We conduct research on lightweight techniques for embedded RDMA software stacks, proposing a lightweight RDMA connection establishment method based on reliable UDP and a lightweight RDMA driver based on reliable Ethernet under conditions without high-performance CPU or operating system. An ultra-lightweight RDMA software stack suitable for embedded sensor nodes is designed. The experimental results show that the lightweight RDMA software stack for embedded systems can reduce the connection establishment overhead by approximately 94% and the software driver overhead by 84% compared with the standard RDMA software stack. The embedded sensor nodes and standard nodes use RDMA technology to perform sensor data acquisition and reliable transmission through one-level switching, with a system bandwidth of up to 98.04 Gbps, end-to-end latency as low as 4.58 μs, and CPU overhead of standard nodes within the system as low as 4.2%. In terms of protocol compatibility, as the number of queue pairs and data volume increase, the READ and WRITE bandwidths of embedded RDMA nodes can reach the peak performance of standard nodes. Meanwhile, the READ and WRITE latency remains at the microsecond level when the data volume is small.

, authors=Hui Yang1, Yutao Zhang1, Mandi Huang1, Yulei Yuan1, Qing Wang2, Tao Li1, *, authorsList=Hui Yang, Yutao Zhang, Mandi Huang, Yulei Yuan, Qing Wang, Tao Li, authorCompany=null, correspAuthors=Tao Li, authorNote=

born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

born in 2000. Master. His main research interests include RDMA and data center congestion control

born in 1999. Master. Her main research interest includes RDMA

born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

, correspAuthorsNote=null, copyrightStatement=Copyright © 2026 Journal of Computer Research and Development. All rights reserved., copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, fund=null), CN=ArticleExt(id=1329419775643595034, articleId=1329419771717726439, tenantId=1146029695717560320, journalId=1324759914146529478, language=CN, title=面向高速传感数据采集的轻量化嵌入式RDMA软件栈, columnId=1329419589143868317, journalTitle=计算机研究与发展, columnName=网络技术, runingTitle=null, highlight=null, articleAbstract=

远程直接内存访问(remote direct memory access,RDMA)技术广泛应用于超算以及智算所需的高带宽和低延迟传输,在带宽、效能和延迟方面与传统TCP/IP协议栈相比具备显著优势。高速传感器通过RDMA技术接入计算网络已逐渐成为工业界共识。然而,在该场景中应用RDMA存在嵌入式传感端资源受限、软件栈难以部署的问题。对嵌入式RDMA软件栈轻量化技术展开研究,提出了无高性能CPU和无操作系统条件下基于可靠UDP的轻量化RDMA建链方法和基于可靠以太网的轻量化RDMA驱动,设计了适用于嵌入式传感端的超轻量化RDMA软件栈。实验结果表明,轻量化嵌入式RDMA软件栈相较于标准RDMA软件栈可降低约94%的建链开销和84%的软件驱动开销,嵌入式传感端与标准端基于RDMA技术经由一级交换进行传感数据采集与可靠传输,系统带宽可达98.04 Gbps,端到端延迟低至4.58 μs,系统内标准端CPU开销低至4.2%。在协议兼容性方面,嵌入式RDMA端的READ和WRITE带宽随队列对(QP)数量和数据量的增加均能够达到标准端的峰值性能,READ和WRITE延迟在数据量较小时保持在微妙级水平。

, authors=杨惠1, 张毓涛1, 黄曼蒂1, 原玉磊1, 汪清2, 李韬1, *, authorsList=杨惠, 张毓涛, 黄曼蒂, 原玉磊, 汪清, 李韬, authorCompany=null, correspAuthors=李韬, authorNote=

作者贡献声明

杨惠和李韬提出并实现了基于可靠UDP的轻量化RDMA建链方法和基于可靠以太网的轻量化RDMA驱动,构建了论文整体框架,并撰写论文;张毓涛和黄曼蒂撰写论文,搭建原型系统与验证实验;原玉磊参与了轻量化RDMA建链和轻量化RDMA驱动的实现过程,提供了原型系统搭建条件;汪清和李韬负责论文审阅并提出指导意见。

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

1999年生。硕士。主要研究方向为RDMA

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

, correspAuthorsNote=
李韬()
, copyrightStatement=版权所有 © 2026 计算机研究与发展编辑部, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=oiM8cJJjYWHt5fqwrYjGrg==, magXml=z5FBxE+veq6IbCYR8OD22g==, pdfUrl=null, pdf=vu+PLVm5gYywbKdndmjCmw==, pdfFileSize=3352013, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=P+L2ba2PWGMmY0AHPXb0DQ==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=FKgsF63sYHlYzvb4H40RCw==, mapNumber=null, fund=null)}, authors=[Author(id=1329419776021082402, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=huihui19870124@126.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776088191268, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776021082402, language=EN, stringName=Hui Yang, firstName=Hui, middleName=null, lastName=Yang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

"}, bioImg=null, bioContent=

born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776146911525, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776021082402, language=CN, stringName=杨惠, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"ypiIpOemqXy1CYTh9ixIaw==","content":"

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

"}, bioImg=ypiIpOemqXy1CYTh9ixIaw==, bioContent=

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776205631783, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776281129257, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776205631783, language=EN, stringName=Yutao Zhang, firstName=Yutao, middleName=null, lastName=Zhang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 2000. Master. His main research interests include RDMA and data center congestion control

"}, bioImg=null, bioContent=

born in 2000. Master. His main research interests include RDMA and data center congestion control

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776344043818, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776205631783, language=CN, stringName=张毓涛, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"ttq5CX9XMCoxv9wI5iLS1Q==","content":"

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

"}, bioImg=ttq5CX9XMCoxv9wI5iLS1Q==, bioContent=

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776411152684, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776478261550, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776411152684, language=EN, stringName=Mandi Huang, firstName=Mandi, middleName=null, lastName=Huang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1999. Master. Her main research interest includes RDMA

"}, bioImg=null, bioContent=

born in 1999. Master. Her main research interest includes RDMA

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776536981807, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776411152684, language=CN, stringName=黄曼蒂, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"pOfGqR+ozokn2/PaeIAEHA==","content":"

1999年生。硕士。主要研究方向为RDMA

"}, bioImg=pOfGqR+ozokn2/PaeIAEHA==, bioContent=

1999年生。硕士。主要研究方向为RDMA

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776595702065, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776675393843, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776595702065, language=EN, stringName=Yulei Yuan, firstName=Yulei, middleName=null, lastName=Yuan, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

"}, bioImg=null, bioContent=

born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776746697012, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776595702065, language=CN, stringName=原玉磊, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"Njvv1BJ3o2+ydqnWQuvBJQ==","content":"

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

"}, bioImg=Njvv1BJ3o2+ydqnWQuvBJQ==, bioContent=

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776818000182, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776889303352, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776818000182, language=EN, stringName=Qing Wang, firstName=Qing, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, bio={"content":"

born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

"}, bioImg=null, bioContent=

born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776948023609, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776818000182, language=CN, stringName=汪清, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2天津大学电气自动化与信息工程学院 天津 300072, bio={"img":"DkxWaRK7oFQaWwKyL8V/8w==","content":"

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

"}, bioImg=DkxWaRK7oFQaWwKyL8V/8w==, bioContent=

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775937196318, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=2, ext=[AuthorCompanyExt(id=1329419775945584927, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072), AuthorCompanyExt(id=1329419775953973536, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2天津大学电气自动化与信息工程学院 天津 300072)])]), Author(id=1329419777006743867, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=taoli_network@163.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1329419777078047037, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419777006743867, language=EN, stringName=Tao Li, firstName=Tao, middleName=null, lastName=Li, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

"}, bioImg=null, bioContent=

born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419777149350206, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419777006743867, language=CN, stringName=李韬, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"0mbeWbIZdEs04XDZo5FjCw==","content":"

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

"}, bioImg=0mbeWbIZdEs04XDZo5FjCw==, bioContent=

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])])], keywords=[Keyword(id=1329419777233236287, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, orderNo=1, keyword=sensor data), Keyword(id=1329419777291956544, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, orderNo=2, keyword=embedded), Keyword(id=1329419777346482497, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, orderNo=3, keyword=RDMA), Keyword(id=1329419777409397058, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, orderNo=4, keyword=lightweight), Keyword(id=1329419777476505923, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, orderNo=5, keyword=software stack), Keyword(id=1329419777539420484, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, orderNo=1, keyword=传感器数据), Keyword(id=1329419777598140741, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, orderNo=2, keyword=嵌入式), Keyword(id=1329419777661055302, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, orderNo=3, keyword=RDMA), Keyword(id=1329419777728164167, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, orderNo=4, keyword=轻量化), Keyword(id=1329419777791078728, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, orderNo=5, keyword=软件栈)], refs=[Reference(id=1329419780433490286, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=1, rfOrder=0, authorNames=null, journalName=null, refType=null, unstructuredReference=黄曼蒂, 李韬, 杨惠, 等. 以太网RDMA网卡综述[J]. 计算机研究与发展, 2025, 62(5): 1262−1289, articleTitle=null, refAbstract=null), Reference(id=1329419780496404847, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=1, rfOrder=1, authorNames=null, journalName=null, refType=null, unstructuredReference=Huang Mandi, Li Tao, Yang Hui, et al. Survey on Ethernet RDMA network interface card[J]. Journal of Computer Research and Development, 2025, 62(5): 1262−1289(in Chinese), articleTitle=null, refAbstract=null), Reference(id=1329419780584485232, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=2, rfOrder=2, authorNames=null, journalName=null, refType=null, unstructuredReference=Chowdhury A, Karmakar G, Kamruzzaman J, et al. Attacks on self-driving cars and their countermeasures: A survey[J]. IEEE Access, 2020, 8: 207308−207342, articleTitle=null, refAbstract=null), Reference(id=1329419780651594097, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=3, rfOrder=3, authorNames=null, journalName=null, refType=null, unstructuredReference=Xia Dan, Jiang Chun, Wan Jiafu, et al. Heterogeneous network access and fusion in smart factory: A survey[J]. ACM Computing Surveys, 2023, 55(6): 113.1−113.31, articleTitle=null, refAbstract=null), Reference(id=1329419780739674482, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=4, rfOrder=4, authorNames=null, journalName=null, refType=null, unstructuredReference=Luo Huimin, Zhang Jiao, Yu Mingxuan, et al. SeqBalance: Congestion-aware load balancing with no reordering in data center networks[J]. IEEE Internet of Things Journal, 2024, 12(13): 25707−25719, articleTitle=null, refAbstract=null), Reference(id=1329419780802589043, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=5, rfOrder=5, authorNames=null, journalName=null, refType=null, unstructuredReference=Xue Jiachen, Chaudhry M U, Vamanan B, et al. Dart: Divide and specialize for fast response to congestion in RDMA-based datacenter networks[J]. IEEE/ACM Transactions on Networking, 2018, 28(1): 322−335, articleTitle=null, refAbstract=null), Reference(id=1329419780869697908, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=6, rfOrder=6, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang Xiaoliang, Song Hexiang, Nguyen C, et al. Maximizing the benefit of RDMA at end hosts [C] // Proc of the 2021 IEEE Int Conf on Computer Communications. Piscataway, NJ: IEEE, 2021: 1−10, articleTitle=null, refAbstract=null), Reference(id=1329419780924223861, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=7, rfOrder=7, authorNames=null, journalName=null, refType=null, unstructuredReference=Mittal R, Shpiner A, Panda A, et al. Revisiting network support for RDMA [C] //Proc of the 2018 ACM SIGCOMM Conf. New York: ACM, 2018: 313−326, articleTitle=null, refAbstract=null), Reference(id=1329419780995527030, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=8, rfOrder=8, authorNames=null, journalName=null, refType=null, unstructuredReference=Li Yuliang, Miao Rui, Liu H H, et al. HPCC: High precision congestion control [C] // Proc of the 2019 ACM SIGCOMM Conf. New York: ACM, 2019: 44−58, articleTitle=null, refAbstract=null), Reference(id=1329419781054247287, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=9, rfOrder=9, authorNames=null, journalName=null, refType=null, unstructuredReference=Bai Wei, Abdeen S S, Agrawal A, et al. Empowering Azure storage with RDMA [C] // Proc of the 20th USENIX Symp on Networked Systems Design and Implementation (NSDI). Berkeley, CA: USENIX Association, 2023: 49−67, articleTitle=null, refAbstract=null), Reference(id=1329419781121356152, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=10, rfOrder=10, authorNames=null, journalName=null, refType=null, unstructuredReference=Kim D, Yu Tianlong, Liu H H, et al. FreeFlow: Software-based virtual RDMA networking for containerized clouds [C] // Proc of the 16th USENIX Symp on Networked Systems Design and Implementation. Berkeley, CA: USENIX Association, 2019: 113−126, articleTitle=null, refAbstract=null), Reference(id=1329419781196853625, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=11, rfOrder=11, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang Zilong, Luo Layong, Ning Qingsong, et al. SRNIC: A scalable architecture for RDMA NICs [C] // Proc of the 20th USENIX Symp on Networked Systems Design and Implementation. Berkeley, CA: USENIX Association, 2023: 1−14, articleTitle=null, refAbstract=null), Reference(id=1329419781268156794, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=12, rfOrder=12, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang Xizheng, Chen Guo, Yin Xijin, et al. StaR: Breaking the scalability limit for RDMA [C] // Proc of the 29th IEEE Int Conf on Network Protocols. Piscataway, NJ: IEEE, 2021: 1−11, articleTitle=null, refAbstract=null), Reference(id=1329419781347848571, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=13, rfOrder=13, authorNames=null, journalName=null, refType=null, unstructuredReference=Lee J B, Kim S J, Kim W H, et al. RDMA-based sampling port of ARINC-653[J]. IEEE Embedded Systems Letters, 2024, 16(4): 437−440, articleTitle=null, refAbstract=null), Reference(id=1329419781419151740, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=14, rfOrder=14, authorNames=null, journalName=null, refType=null, unstructuredReference=Feng Junjie, Chen Xianzhang, Liu Duo, et al. eRDAC: Efficient and reliable remote direct access and control for embedded systems[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2022, 41(11): 3685−3696, articleTitle=null, refAbstract=null), Reference(id=1329419781482066301, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=15, rfOrder=15, authorNames=null, journalName=null, refType=null, unstructuredReference=Schelten N, Steinert F, Knapheide J, et al. A high-throughput, eesource-efficient implementation of the RoCEv2 remote DMA protocol and its application[J]. ACM Transactions on Reconfigurable Technology, 2022, 16(1): 1−23, articleTitle=null, refAbstract=null), Reference(id=1329419781557563774, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=16, rfOrder=16, authorNames=null, journalName=null, refType=null, unstructuredReference=Cossettini A, Taranov K, Vogt C, et al. A RDMA interface for ultra-fast ultrasound data-streaming over an optical link [C] // Proc of the 2022 Design, Automation & Test in Europe Conf & Exhibition. Piscataway, NJ: IEEE, 2022: 80−83, articleTitle=null, refAbstract=null), Reference(id=1329419781624672639, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=17, rfOrder=17, authorNames=null, journalName=null, refType=null, unstructuredReference=AMD. AMD embedded RDMA enabled NIC LogiCORE IP product guide (PG332)[EB/OL]. (2024-11-22)[2025-06-10]. https://docs.xilinx.com/r/-en-US/pg332-ernic, articleTitle=null, refAbstract=null), Reference(id=1329419781704364416, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=18, rfOrder=18, authorNames=null, journalName=null, refType=null, unstructuredReference=InfiniBand. InfiniBand architecture specification[EB/OL]. [2025-06-10]. https://www.infinibandta.org/ibta-specification/, articleTitle=null, refAbstract=null), Reference(id=1329419781771473281, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=19, rfOrder=19, authorNames=null, journalName=null, refType=null, unstructuredReference=IETF Network Working Group. RFC5040: A remote direct memory access protocol specification[EB/OL]. (2007-10-01)[2025-06-10]. https://www.rfc-edi tor.org/rfc/rfc5040, articleTitle=null, refAbstract=null), Reference(id=1329419781842776450, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=20, rfOrder=20, authorNames=null, journalName=null, refType=null, unstructuredReference=NVIDIA. RDMA over converged ethernet(RoCE)[EB/OL]. [2025-06-10]. https://docs.nvidia.com/networking/display/onyxv3103004/rdma+over+converged+ethernet+(roce, articleTitle=null, refAbstract=null), Reference(id=1329419781918273923, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=21, rfOrder=21, authorNames=null, journalName=null, refType=null, unstructuredReference=He Zhiqiang, Wang Dongyang, Fu Binzhang, et al. MasQ: RDMA for virtual private cloud [C] // Proc of the 2020 ACM SIGCOMM Conf. New York: ACM, 2020: 1−14, articleTitle=null, refAbstract=null), Reference(id=1329419781981188484, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=22, rfOrder=22, authorNames=null, journalName=null, refType=null, unstructuredReference=OpenTSN 工作组. 银河衡芯HX-DP4C边缘智能网络处理器 [EB/OL]. 湖南: OpenTSN工作组, (2023-08-30)[2025-06-10]. https: //mp.weixin.qq.com/s/gh2no71EwjX0a-HGHLiPJg, articleTitle=null, refAbstract=null), Reference(id=1329419782056685957, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=22, rfOrder=23, authorNames=null, journalName=null, refType=null, unstructuredReference=OpenTSN Group. Yinhe Hengxin HX-DP4C Edge Intelligence Network Processor[EB/OL]. Hunan: OpenTSNGroup. (2023-08-30)[2025-06-10]. https://mp.weixin.qq.com/s/gh2no71EwjX0a-HGHLiPJg (in chinese, articleTitle=null, refAbstract=null), Reference(id=1329419782123794822, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=23, rfOrder=24, authorNames=null, journalName=null, refType=null, unstructuredReference=AMD. The AMD Virtex UltraScale+ FPGA VCU118[EB/OL]. [2025-06-10]. https://www.amd.com/zh-cn/products/adaptive-socs-and-fpgas/evaluation-boards/vcu118.html#tabs-e09e115b99-item-fb75e43802-tab, articleTitle=null, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)]), AuthorCompany(id=1329419775937196318, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=2, ext=[AuthorCompanyExt(id=1329419775945584927, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072), AuthorCompanyExt(id=1329419775953973536, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2天津大学电气自动化与信息工程学院 天津 300072)])], figs=[ArticleFig(id=1329419777921102153, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 1, caption=Standard RDMA software stack, figureFileSmall=JvpNkNElTsOghHw2IF+b3A==, figureFileBig=P+L2ba2PWGMmY0AHPXb0DQ==, tableContent=null), ArticleFig(id=1329419777984016714, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图1, caption=标准RDMA软件栈, figureFileSmall=JvpNkNElTsOghHw2IF+b3A==, figureFileBig=P+L2ba2PWGMmY0AHPXb0DQ==, tableContent=null), ArticleFig(id=1329419778168566091, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 2, caption=Overall architecture of embedded RDMA software stack, figureFileSmall=Y3PrTHF2MoKQ6FSuTQb3GQ==, figureFileBig=pEnTqUkbEMwEFN0yYRz5Cg==, tableContent=null), ArticleFig(id=1329419778239869260, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图2, caption=嵌入式RDMA软件栈总体架构, figureFileSmall=Y3PrTHF2MoKQ6FSuTQb3GQ==, figureFileBig=pEnTqUkbEMwEFN0yYRz5Cg==, tableContent=null), ArticleFig(id=1329419778306978125, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 3, caption=Connection establishment interaction content and message format, figureFileSmall=ADrhaOJZ8ynvD8yUYdcrjQ==, figureFileBig=tcSxL8DlMwgMxcs7ajJKJQ==, tableContent=null), ArticleFig(id=1329419778369892686, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图3, caption=建链交互内容及报文格式, figureFileSmall=ADrhaOJZ8ynvD8yUYdcrjQ==, figureFileBig=tcSxL8DlMwgMxcs7ajJKJQ==, tableContent=null), ArticleFig(id=1329419778445390159, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 4, caption=Connection establishment interaction workflow, figureFileSmall=/NdqzSV3NpQKYR5O8ikvqg==, figureFileBig=9YQoMqGPjRTKEXyKskX7pw==, tableContent=null), ArticleFig(id=1329419778508304720, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图4, caption=建链交互工作流程, figureFileSmall=/NdqzSV3NpQKYR5O8ikvqg==, figureFileBig=9YQoMqGPjRTKEXyKskX7pw==, tableContent=null), ArticleFig(id=1329419778579607889, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 5, caption=NACP message format, figureFileSmall=3feMnyCY1LAc60VVX5IEBQ==, figureFileBig=jGn1Am/JHzeasncrSlwEdg==, tableContent=null), ArticleFig(id=1329419778638328146, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图5, caption=NACP报文格式, figureFileSmall=3feMnyCY1LAc60VVX5IEBQ==, figureFileBig=jGn1Am/JHzeasncrSlwEdg==, tableContent=null), ArticleFig(id=1329419778705437011, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 6, caption=Mbox interaction mechanism, figureFileSmall=rY62HYDruhJq1n0luEjIPQ==, figureFileBig=nd9YpqhxrE+pAdz9z8w4Jw==, tableContent=null), ArticleFig(id=1329419778793517396, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图6, caption=Mbox交互机制, figureFileSmall=rY62HYDruhJq1n0luEjIPQ==, figureFileBig=nd9YpqhxrE+pAdz9z8w4Jw==, tableContent=null), ArticleFig(id=1329419778856431957, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 7, caption=Physical environment of experimental platform, figureFileSmall=FXl8ckCyCNUq1LH+2NMj2A==, figureFileBig=5iR9GbmXjNv7RZ64hzwpRQ==, tableContent=null), ArticleFig(id=1329419778927735126, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图7, caption=实验平台物理环境, figureFileSmall=FXl8ckCyCNUq1LH+2NMj2A==, figureFileBig=5iR9GbmXjNv7RZ64hzwpRQ==, tableContent=null), ArticleFig(id=1329419778990649687, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 8, caption=Comparison of protocol stack code size, figureFileSmall=UctXOZZKWhAVIXXWPaBkfQ==, figureFileBig=SrzPmYGXmv9V90xZgNdSHg==, tableContent=null), ArticleFig(id=1329419779066147160, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图8, caption=协议栈代码量对比, figureFileSmall=UctXOZZKWhAVIXXWPaBkfQ==, figureFileBig=SrzPmYGXmv9V90xZgNdSHg==, tableContent=null), ArticleFig(id=1329419779137450329, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 9, caption=Comparison of driver code size, figureFileSmall=tzRJhZybzllDSb6KBZTndQ==, figureFileBig=C1qx6KiHjjchbSzE0rthMA==, tableContent=null), ArticleFig(id=1329419779200364890, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图9, caption=驱动代码量对比, figureFileSmall=tzRJhZybzllDSb6KBZTndQ==, figureFileBig=C1qx6KiHjjchbSzE0rthMA==, tableContent=null), ArticleFig(id=1329419779259085147, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 10, caption=Standard host CPU utilization, figureFileSmall=8YTC9/Jiqpw2NmrkN47ucg==, figureFileBig=lsdm6wnKenrZMCNBc/pXIg==, tableContent=null), ArticleFig(id=1329419779317805404, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图10, caption=标准端CPU利用率, figureFileSmall=8YTC9/Jiqpw2NmrkN47ucg==, figureFileBig=lsdm6wnKenrZMCNBc/pXIg==, tableContent=null), ArticleFig(id=1329419779384914269, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 11, caption=Peak transmission bandwidth of the embedded RDMA under different QP counts and MTU sizes, figureFileSmall=qjRbhClabz/FjHMztvB8AQ==, figureFileBig=LzECF9cRu3XBR9MWf5kJtw==, tableContent=null), ArticleFig(id=1329419779452023134, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图11, caption=不同QP数量和MTU大小下的嵌入式RDMA传输带宽峰值, figureFileSmall=qjRbhClabz/FjHMztvB8AQ==, figureFileBig=LzECF9cRu3XBR9MWf5kJtw==, tableContent=null), ArticleFig(id=1329419779514937695, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 12, caption=Peak transmission bandwidth of the standard RDMA under different QP counts and MTU sizes, figureFileSmall=3K73c2YlPfnf3ToA5z/zbQ==, figureFileBig=erj8q8GyiVakyE+WMm5B8g==, tableContent=null), ArticleFig(id=1329419779598823777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图12, caption=不同QP数量和MTU大小下的标准RDMA传输带宽峰值, figureFileSmall=3K73c2YlPfnf3ToA5z/zbQ==, figureFileBig=erj8q8GyiVakyE+WMm5B8g==, tableContent=null), ArticleFig(id=1329419779649155426, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 13, caption=Transmission latency of the embedded RDMA under different MTU sizes, figureFileSmall=Ym+rQgg6fqPRnzDEk74Vow==, figureFileBig=XFACDIuAeh8Gs/pAFoJtPQ==, tableContent=null), ArticleFig(id=1329419779707875683, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图13, caption=不同MTU大小下的嵌入式RDMA传输延迟, figureFileSmall=Ym+rQgg6fqPRnzDEk74Vow==, figureFileBig=XFACDIuAeh8Gs/pAFoJtPQ==, tableContent=null), ArticleFig(id=1329419779766595940, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 14, caption=Transmission latency of the standard RDMA under different MTU sizes, figureFileSmall=UkW3Ez5ZgF2T4o9hp+tkiw==, figureFileBig=mhwCBZCBhv5u/Gwps8wEuA==, tableContent=null), ArticleFig(id=1329419779816927589, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图14, caption=不同MTU大小下的标准RDMA传输延迟, figureFileSmall=UkW3Ez5ZgF2T4o9hp+tkiw==, figureFileBig=mhwCBZCBhv5u/Gwps8wEuA==, tableContent=null), ArticleFig(id=1329419779913396582, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 15, caption=Variation trend of system transmission bandwidth with QP counts and message sizes, figureFileSmall=4HH2EkCEI/OK24aTz1yUrw==, figureFileBig=flF3mDekPkCjtMOpi+DTdw==, tableContent=null), ArticleFig(id=1329419779967922535, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图15, caption=系统传输带宽随QP数量和消息大小的变化趋势, figureFileSmall=4HH2EkCEI/OK24aTz1yUrw==, figureFileBig=flF3mDekPkCjtMOpi+DTdw==, tableContent=null), ArticleFig(id=1329419780026642792, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 16, caption=Variation trend of system transmission delay with message size, figureFileSmall=8f2tK5TuXIiN1PqfXADqGw==, figureFileBig=v+Um8Hi9X4nzptpLI3QT4Q==, tableContent=null), ArticleFig(id=1329419780085363049, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图16, caption=系统传输延迟随消息大小的变化趋势, figureFileSmall=8f2tK5TuXIiN1PqfXADqGw==, figureFileBig=v+Um8Hi9X4nzptpLI3QT4Q==, tableContent=null), ArticleFig(id=1329419780148277610, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Figure 17, caption=Stability evaluation of throughput and latency in stress test, figureFileSmall=PxeFP33XkRhj9EwtqxHqVA==, figureFileBig=Qt38Ho2EdqwjvRZwLSSvwA==, tableContent=null), ArticleFig(id=1329419780211192171, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=图17, caption=压力测试中吞吐量与延迟的稳定性评估, figureFileSmall=PxeFP33XkRhj9EwtqxHqVA==, figureFileBig=Qt38Ho2EdqwjvRZwLSSvwA==, tableContent=null), ArticleFig(id=1329419780265718124, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=EN, label=Table 1, caption=

Comparison of CPU Load Between Reliable UDP Connection Establishment and Socket TCP Connection Establishment %

, figureFileSmall=null, figureFileBig=null, tableContent=
测试场景请求端 CPU 负载响应端 CPU 负载
轻量化RDMA+轻量级
可靠UDP的建链
22.811.8
标准RDMA+Socket TCP的建链54.026.9
), ArticleFig(id=1329419780328632685, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, language=CN, label=表1, caption=

可靠UDP建链与Socket TCP建链的CPU负载对比

, figureFileSmall=null, figureFileBig=null, tableContent=
测试场景请求端 CPU 负载响应端 CPU 负载
轻量化RDMA+轻量级
可靠UDP的建链
22.811.8
标准RDMA+Socket TCP的建链54.026.9
)], attaches=null, journal=Journal(id=1324759606947315908, delFlag=0, nameCn=计算机研究与发展, nameEn=Journal of Computer Research and Development, nameHistory1=null, nameHistory2=null, issn=1000-1239, eissn=null, cn=11-1777/TP, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=k2l8zgqwstuxxevFX2qxVA==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Computer Research and Development, journalRemark=null, publicationField=null, createdTime=1793776903141, updatedTime=1793850698612, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Electronic Information, subjectName=null, subjectCodeEn=Electronic Information, subjectNameEn=null, picCn=k2l8zgqwstuxxevFX2qxVA==, picEn=r6k0Ry+RD+c/VXj5QarmDw==, jcr=null, cjcr=null, exts=[JournalExt(id=1325069127710286688, language=CN, name=计算机研究与发展, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1793850698640, updatedTime=1793850698640, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://crad.ict.ac.cn/journalx/authorLogOn.action, submissionEditorUrl=https://crad.ict.ac.cn/journalx/editorLogOn.action, submissionReviewUrl=https://crad.ict.ac.cn/journalx/expertLogOn.action, submissionCeEditorUrl=https://crad.ict.ac.cn/journalx/editorInChiefLogOn.action, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1325069127777395553, language=EN, name=Journal of Computer Research and Development, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1793850698656, updatedTime=1793850698656, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://crad.ict.ac.cn/journalx/authorLogOn.action, submissionEditorUrl=https://crad.ict.ac.cn/journalx/editorLogOn.action, submissionReviewUrl=https://crad.ict.ac.cn/journalx/expertLogOn.action, submissionCeEditorUrl=https://crad.ict.ac.cn/journalx/editorInChiefLogOn.action, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1324759914146529478, websiteList=[Website(id=1325070300139892771, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1324759914146529478, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN, language=CN, createTime=1793850978169, createBy=18614031015, updateTime=1793851040259, updateBy=18614031015, name=计算机研究与发展-中文, tplId=1146099689490845704, title=计算机研究与发展, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1325070893751349360, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=articleTextType, value=kx, createTime=1793851119697, updateTime=1793851119697, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893730377837, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=banner, value=null, createTime=1793851119692, updateTime=1793851119692, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893772320883, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=grayFlag, value=0, createTime=1793851119702, updateTime=1793851119702, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893721989228, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=logo, value=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/file/pic?fileId=U27tDsy717YF7XSPA+Vwdw==, createTime=1793851119690, updateTime=1793851119690, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893784903797, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=minRunFlag, value=0, createTime=1793851119705, updateTime=1793851119705, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893742960751, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/file/pic, createTime=1793851119695, updateTime=1793851119695, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893780709492, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=silenceFlag, value=0, createTime=1793851119704, updateTime=1793851119704, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893738766446, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1793851119694, updateTime=1793851119694, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893755543665, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=themeColor, value=null, createTime=1793851119698, updateTime=1793851119698, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070893763932274, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300139892771, code=themeStyle, value=null, createTime=1793851119700, updateTime=1793851119700, creator=18614031015, updator=18614031015)]), Website(id=1325070300580294698, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1324759914146529478, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN, language=EN, createTime=1793850978274, createBy=18614031015, updateTime=1793851036638, updateBy=18614031015, name=计算机研究与发展-英文, tplId=1146101810881728533, title=Journal of Computer Research and Development, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1325070925288321148, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=articleTextType, value=kx, createTime=1793851127216, updateTime=1793851127216, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925263155321, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=banner, value=null, createTime=1793851127210, updateTime=1793851127210, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925309292671, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=grayFlag, value=0, createTime=1793851127221, updateTime=1793851127221, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925254766712, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=logo, value=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/file/pic?fileId=U27tDsy717YF7XSPA+Vwdw==, createTime=1793851127208, updateTime=1793851127208, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925326069889, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=minRunFlag, value=0, createTime=1793851127225, updateTime=1793851127225, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925279932539, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/file/pic, createTime=1793851127214, updateTime=1793851127214, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925317681280, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=silenceFlag, value=0, createTime=1793851127223, updateTime=1793851127223, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925271543930, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1793851127212, updateTime=1793851127212, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925292515453, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=themeColor, value=null, createTime=1793851127217, updateTime=1793851127217, creator=18614031015, updator=18614031015), WebsiteProps(id=1325070925300904062, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1325070300580294698, code=themeStyle, value=null, createTime=1793851127219, updateTime=1793851127219, creator=18614031015, updator=18614031015)])], journalTitle=计算机研究与发展, weixinUrl=null, journalUrl=https://crad.ict.ac.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Computer Research and Development, journalPhotoCn=k2l8zgqwstuxxevFX2qxVA==, journalPhotoEn=r6k0Ry+RD+c/VXj5QarmDw==, journalFirstLetter=J, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false, interPubPlatform=, interPubPlatformUrl=null), detailUrlCn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/10.7544/issn1000-1239.202550472, detailUrlEn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/10.7544/issn1000-1239.202550472, pdfUrlCn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/PDF/10.7544/issn1000-1239.202550472, pdfUrlEn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/PDF/10.7544/issn1000-1239.202550472, aliStartDate=0, aliEndDate=0, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1777564800000, fullTextJson=null, articleText=null, reference=null)
收藏切换
面向高速传感数据采集的轻量化嵌入式RDMA软件栈
杨惠 1 , 张毓涛 1 , 黄曼蒂 1 , 原玉磊 1 , 汪清 2 , 李韬 1, *
计算机研究与发展 | 网络技术 2026,63(5): 1337-1349
收起
收藏切换
计算机研究与发展 |网络技术 2026 , 63 (5) : 1337 -1349
面向高速传感数据采集的轻量化嵌入式RDMA软件栈
全屏
[Author(id=1329419776021082402, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=huihui19870124@126.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776088191268, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776021082402, language=EN, stringName=Hui Yang, firstName=Hui, middleName=null, lastName=Yang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

"}, bioImg=null, bioContent=

born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776146911525, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776021082402, language=CN, stringName=杨惠, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"ypiIpOemqXy1CYTh9ixIaw==","content":"

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

"}, bioImg=ypiIpOemqXy1CYTh9ixIaw==, bioContent=

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776205631783, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776281129257, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776205631783, language=EN, stringName=Yutao Zhang, firstName=Yutao, middleName=null, lastName=Zhang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 2000. Master. His main research interests include RDMA and data center congestion control

"}, bioImg=null, bioContent=

born in 2000. Master. His main research interests include RDMA and data center congestion control

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776344043818, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776205631783, language=CN, stringName=张毓涛, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"ttq5CX9XMCoxv9wI5iLS1Q==","content":"

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

"}, bioImg=ttq5CX9XMCoxv9wI5iLS1Q==, bioContent=

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776411152684, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776478261550, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776411152684, language=EN, stringName=Mandi Huang, firstName=Mandi, middleName=null, lastName=Huang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1999. Master. Her main research interest includes RDMA

"}, bioImg=null, bioContent=

born in 1999. Master. Her main research interest includes RDMA

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776536981807, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776411152684, language=CN, stringName=黄曼蒂, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"pOfGqR+ozokn2/PaeIAEHA==","content":"

1999年生。硕士。主要研究方向为RDMA

"}, bioImg=pOfGqR+ozokn2/PaeIAEHA==, bioContent=

1999年生。硕士。主要研究方向为RDMA

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776595702065, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776675393843, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776595702065, language=EN, stringName=Yulei Yuan, firstName=Yulei, middleName=null, lastName=Yuan, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

"}, bioImg=null, bioContent=

born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776746697012, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776595702065, language=CN, stringName=原玉磊, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"Njvv1BJ3o2+ydqnWQuvBJQ==","content":"

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

"}, bioImg=Njvv1BJ3o2+ydqnWQuvBJQ==, bioContent=

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])]), Author(id=1329419776818000182, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1329419776889303352, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776818000182, language=EN, stringName=Qing Wang, firstName=Qing, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, bio={"content":"

born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

"}, bioImg=null, bioContent=

born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419776948023609, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419776818000182, language=CN, stringName=汪清, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2天津大学电气自动化与信息工程学院 天津 300072, bio={"img":"DkxWaRK7oFQaWwKyL8V/8w==","content":"

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

"}, bioImg=DkxWaRK7oFQaWwKyL8V/8w==, bioContent=

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775937196318, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=2, ext=[AuthorCompanyExt(id=1329419775945584927, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072), AuthorCompanyExt(id=1329419775953973536, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775937196318, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2天津大学电气自动化与信息工程学院 天津 300072)])]), Author(id=1329419777006743867, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=taoli_network@163.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1329419777078047037, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419777006743867, language=EN, stringName=Tao Li, firstName=Tao, middleName=null, lastName=Li, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073, bio={"content":"

born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

"}, bioImg=null, bioContent=

born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

, aboutCorrespAuthor=null), CN=AuthorExt(id=1329419777149350206, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, authorId=1329419777006743867, language=CN, stringName=李韬, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, *, address=1国防科技大学计算机学院 长沙 410073, bio={"img":"0mbeWbIZdEs04XDZo5FjCw==","content":"

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

"}, bioImg=0mbeWbIZdEs04XDZo5FjCw==, bioContent=

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1329419775828144411, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, xref=1, ext=[AuthorCompanyExt(id=1329419775836533020, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073), AuthorCompanyExt(id=1329419775849115933, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1329419771717726439, companyId=1329419775828144411, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1国防科技大学计算机学院 长沙 410073)])])]
杨惠1 , 张毓涛1, 黄曼蒂1, 原玉磊1, 汪清2, 李韬1, *
作者信息
  • 1国防科技大学计算机学院 长沙 410073
  • 2天津大学电气自动化与信息工程学院 天津 300072
通讯作者:
作者简介:

作者贡献声明

杨惠和李韬提出并实现了基于可靠UDP的轻量化RDMA建链方法和基于可靠以太网的轻量化RDMA驱动,构建了论文整体框架,并撰写论文;张毓涛和黄曼蒂撰写论文,搭建原型系统与验证实验;原玉磊参与了轻量化RDMA建链和轻量化RDMA驱动的实现过程,提供了原型系统搭建条件;汪清和李韬负责论文审阅并提出指导意见。

1986年生。博士,副研究员。主要研究方向为RDMA、网络处理器

2000年生。硕士。主要研究方向为RDMA、数据中心拥塞控制

1999年生。硕士。主要研究方向为RDMA

1984年生。博士,副研究员。CCF会员。主要研究方向为高性能网络、网络安全

1982年生。博士,教授。主要研究方向为通信雷达信号处理、确定性网络、智能信息处理

1983年生。博士,副研究员。主要研究方向为高性能网络系统及芯片

Lightweight Embedded RDMA Software Stack for High-Speed Sensor Data Acquisition
Hui Yang1 , Yutao Zhang1, Mandi Huang1, Yulei Yuan1, Qing Wang2, Tao Li1, *
Affiliations
  • 1College of Computer Science and Technology, National University of Defense Technology, Changsha 410073
  • 2School of Electrical and Information Engineering, Tianjin University, Tianjin 300072
  • born in 1986. PhD, associate professor. Her main research interests include RDMA and network processor

    born in 2000. Master. His main research interests include RDMA and data center congestion control

    born in 1999. Master. Her main research interest includes RDMA

    born in 1984. PhD, associate professor. Member of CCF. His main research interests include high-performance networks and network security

    born in 1982. PhD, professor. Her main research interests include communication and radar signal processing, deterministic networking, and intelligent information processing

    born in 1983. PhD, associate professor. His main research interest includes high-performance network system and chip

出版时间: 2026-05-01 doi: 10.7544/issn1000-1239.202550472
文章导航
收藏切换

远程直接内存访问(remote direct memory access,RDMA)技术广泛应用于超算以及智算所需的高带宽和低延迟传输,在带宽、效能和延迟方面与传统TCP/IP协议栈相比具备显著优势。高速传感器通过RDMA技术接入计算网络已逐渐成为工业界共识。然而,在该场景中应用RDMA存在嵌入式传感端资源受限、软件栈难以部署的问题。对嵌入式RDMA软件栈轻量化技术展开研究,提出了无高性能CPU和无操作系统条件下基于可靠UDP的轻量化RDMA建链方法和基于可靠以太网的轻量化RDMA驱动,设计了适用于嵌入式传感端的超轻量化RDMA软件栈。实验结果表明,轻量化嵌入式RDMA软件栈相较于标准RDMA软件栈可降低约94%的建链开销和84%的软件驱动开销,嵌入式传感端与标准端基于RDMA技术经由一级交换进行传感数据采集与可靠传输,系统带宽可达98.04 Gbps,端到端延迟低至4.58 μs,系统内标准端CPU开销低至4.2%。在协议兼容性方面,嵌入式RDMA端的READ和WRITE带宽随队列对(QP)数量和数据量的增加均能够达到标准端的峰值性能,READ和WRITE延迟在数据量较小时保持在微妙级水平。

传感器数据  /  嵌入式  /  RDMA  /  轻量化  /  软件栈

Remote direct memory access(RDMA)technology is widely used in high-performance computing and intelligent computing scenarios that require high bandwidth and low latency transmission, offering significant advantages over traditional TCP/IP protocol stacks in terms of bandwidth, efficiency, and latency. It has gradually become an industry consensus for high-speed sensors to access computing networks through RDMA technology. However, deploying RDMA in this scenario faces challenges due to the limited resources of embedded sensor nodes, making it difficult to deploy traditional software stacks. We conduct research on lightweight techniques for embedded RDMA software stacks, proposing a lightweight RDMA connection establishment method based on reliable UDP and a lightweight RDMA driver based on reliable Ethernet under conditions without high-performance CPU or operating system. An ultra-lightweight RDMA software stack suitable for embedded sensor nodes is designed. The experimental results show that the lightweight RDMA software stack for embedded systems can reduce the connection establishment overhead by approximately 94% and the software driver overhead by 84% compared with the standard RDMA software stack. The embedded sensor nodes and standard nodes use RDMA technology to perform sensor data acquisition and reliable transmission through one-level switching, with a system bandwidth of up to 98.04 Gbps, end-to-end latency as low as 4.58 μs, and CPU overhead of standard nodes within the system as low as 4.2%. In terms of protocol compatibility, as the number of queue pairs and data volume increase, the READ and WRITE bandwidths of embedded RDMA nodes can reach the peak performance of standard nodes. Meanwhile, the READ and WRITE latency remains at the microsecond level when the data volume is small.

sensor data  /  embedded  /  RDMA  /  lightweight  /  software stack
杨惠, 张毓涛, 黄曼蒂, 原玉磊, 汪清, 李韬. 面向高速传感数据采集的轻量化嵌入式RDMA软件栈. 计算机研究与发展, 2026 , 63 (5) : 1337 -1349 . DOI: 10.7544/issn1000-1239.202550472
Hui Yang, Yutao Zhang, Mandi Huang, Yulei Yuan, Qing Wang, Tao Li. Lightweight Embedded RDMA Software Stack for High-Speed Sensor Data Acquisition[J]. Journal of Computer Research and Development, 2026 , 63 (5) : 1337 -1349 . DOI: 10.7544/issn1000-1239.202550472
远程直接内存访问(remote direct memory access,RDMA)凭借其零拷贝、内核旁路和CPU卸载的设计思想,可实现高带宽、低延迟、低CPU开销的网络传输服务,并已在高性能计算、数据中心[1]等领域得到广泛关注和研究。随着无人驾驶[2]、工业物联网[3]等技术的快速发展,各类传感设备的种类、数量和传输带宽持续增加,自动驾驶紧急避障和工业产线急停等安全关键场景下,对智能无人信息处理平台的传感数据传输和融合处理能力提出了严苛的要求。
传统总线方式的传感数据采集与处理存在结构烟囱化、宽带受限、可扩展性较差等问题,难以实现高通量传感数据的融合传输。传统以太网通信方式网络协议栈的软件处理开销过大,计算资源占用率高且延迟高,难以满足高清摄像头视频流和相控阵雷达等传感数据的高带宽和低延迟融合传输需求[4]。将在高性能计算等领域得到广泛关注RDMA技术扩展应用到高速传感数据采集与融合传输场景,可满足智能无人等系统中高带宽、低延迟、低算力开销的网络传输需求,实现多源传感数据的高性能可靠传输。
然而,当前RDMA相关研究主要聚焦面向数据中心应用领域的网卡计算资源[5]和存储资源管理[6]、流量控制[7]和拥塞控制[8]、分布式存储系统[9]和多租户性能隔离[10]等相关优化工作,在面向高速多源传感数据采集等嵌入式场景下的RDMA相关研究较少。现有数据中心领域相关研究主要关注RDMA硬件层面优化,通过将RDMA网卡相关处理逻辑和存储资源上载软件[11]、转移至对端[12]等方式尽可能减少硬件处理开销,这在数据中心场景中具备高性能CPU和完备操作系统的情况下是可行的,但在接入高速传感数据采集的嵌入式场景中,受限于嵌入式系统有限的计算和存储资源,嵌入式软件常使用轻量化操作系统或在无操作系统的裸机环境下执行,并不具备高性能CPU以支持通用操作系统和标准RDMA软件栈。
现有嵌入式领域相关研究主要面向航空电子系统[13]、分布式系统集群[14]、网络连接加速器[15]、超声数据采集[16]等领域展开。在软件层面,文献[13]直接针对航空电子系统通信语义将RDMA libverbs软件库进行迁移应用;ERNIC[17]使用Microblaze软核处理器和裁剪版Ubuntu操作系统实现一定程序轻量化;eRDAC[14]则通过轻量化中断处理程序获取远程访问完成通知以消除完成队列(completion queue,CQ)相关数据结构和处理线程的存储和计算开销。文献[13]直接复用标准RDMA软件栈,并未做轻量化处理,难以满足实际嵌入式场景需求;ERNIC实际上仅对RDMA软件栈依赖的底层支撑操作系统进行轻量化,未对RDMA软件栈本身做轻量化处理;eRDAC从CQ的角度对RDMA完成处理进行了轻量化实现,但只聚焦该单一模块优化,未从整体软件架构层面进行轻量化设计,对于高速传感数据采集这一传输导向明确的嵌入式场景而言仍存在功能冗余。
在此背景下,为应对在高速传感数据采集领域中RDMA软件栈面临的问题挑战,满足无人驾驶、工业物联网等嵌入式场景下的传感数据高性能传输需求,本文以高速传感数据采集场景为导向,基于RISC-V微核处理器设计基于可靠UDP的轻量化建链和基于可靠以太网的轻量化RDMA软件驱动,并实现了不依赖于高性能CPU和操作系统的超轻量化RDMA软件栈解决方案,为构建软硬件协同的嵌入式RDMA端系统提供软件层面技术支撑,并为无人驾驶等智能边缘设备的高性能传感数据传输提供新的思路和解决方法。
RDMA通过网络协议栈的硬件卸载实现,允许网络设备在不经过双端CPU参与的情况下直接访问另一个网络设备的内存空间,从而在显著降低延迟和CPU开销的同时提高传输带宽。在实现形式上,RDMA协议实现方式主要包括InifiniBand[18],iWARP[19],RoCE[20]三类。其中,InifiniBand采用全自定义网络协议栈,需专用硬件网卡和交换机作为硬件支撑,具备成本极高、性能极佳、不兼容以太网的通信特点,常用于超算领域;iWARP实现了以太网兼容,并通过TCP/IP协议栈保证传输可靠性,但TCP/IP协议栈的硬件卸载实现在成本和复杂度上仍然较高,在数据中心等领域尚未大规模应用;RoCEv1同样进行了以太网兼容性设计,RoCEv2则进一步使用UDP/IP作为网络层和传输层协议从而支持跨子网通信,并采用流量控制和拥塞控制等方式保证传输可靠性,目前凭借较低的实现复杂度和较好的以太网兼容性已经在数据中心领域得到了广泛关注。
RDMA将网络协议栈相关报文处理以及可靠传输等功能卸载到网卡硬件实现,用户态RDMA驱动程序直接将RDMA操作下发至硬件网卡以启动RDMA传输,实现内核旁路,无需操作系统内核态参与。启动传输后由RDMA硬件网卡直接进行远端内存读写,无需进行用户态到内核态的多次拷贝和系统调用。从通信原语上,RDMA为应用程序提供RDMA WRITE/READ等单边原语以及SEND/RECV双边原语。其中,单边原语在传输过程中不需要通信双方CPU参与,常用于数据搬移;双边原语则涉及对端CPU参与下发RECV操作并处理对应操作完成信息,常用于通信双方状态信息同步。从传输方式上,RDMA提供可靠连接(reliable connected,RC)、不可靠连接(unreliable connected,UC)、不可靠数据报(unreliable datagram,UD)等多种传输方式。其中,RC传输方式需在通信双方之间建立一对一连接并提供可靠传输服务,最大消息长度可达2 GB;UD传输方式则支持一对多通信,不需要保证传输可靠性且仅支持双边原语,最大消息长度仅4 KB。
标准RDMA软件栈包括RDMA双端建链协议栈、RDMA用户态和内核态驱动,整体工作流程如图1所示,主要包括前期配置、数据传输、资源回收3部分[21]。
前期配置阶段主要完成RDMA资源准备、双端建链协商以及队列对(queue pair,QP)状态转换,由RDMA内核态驱动参与完成。其中,资源准备主要涉及RDMA应用程序打开RDMA硬件网卡,并创建用于网卡进行内存数据读写访问的内存区域(memory region,MR)、CQ、QP等资源。通信双端建链协商主要包括基于通信管理器(communication manager,CM)和基于Socket TCP/IP协议栈2种方式,其中CM建链需要一个UD类型QP专用于处理CM相关报文,且具备专为CM建链定制的编程接口,整体实现较为复杂。相比之下,Socket建链方式只需按照标准Socket API进行通信即可,更适合面向嵌入式场景进一步优化,基于Socket协议栈提供的网络通信接口进行双端建链交互,调用RDMA驱动程序提供的verbs接口完成RDMA控制路径相关操作。具体Socket建链交互信息可见2.3节。建链完成后,通信双方已经得知与对端进行RDMA操作对应使用的QP相关信息,并进行QP状态转换,正常情况下QP在创建后依次从复位(reset,RST)状态转换至初始化(initialized,INIT)状态、准备接收(ready to receive,RTR)状态和准备发送(ready to send,RTS)状态。
数据传输阶段主要包括软件下发工作请求和获取完成信息,以及硬件网卡执行数据传输等环节。在软件层面,RDMA应用程序首先向QP中的接收队列(receive queue,RQ)下发操作请求,预先为接收通信对端发来的SEND操作做好准备,之后可根据需求向QP中的发送队列(send queue,SQ)下发WRITE和READ等操作,并由硬件网卡根据请求进行报文封装和解析等操作。具体而言,应用程序调用相关函数后,相关工作请求(work request,WR)会被RDMA用户态驱动程序解析为工作队列元素(work queue element,WQE)放入相应SQ或CQ中,并通过Doorbell机制通知硬件网卡有待处理的WQE。网卡根据Doorbell机制和对应WQE解析需完成的RDMA操作请求构造RDMA请求报文并接收对端网卡响应;在报文封装和解析过程中进行远端内存读写,并在本次操作执行完成后构造完成队列元素(completion queue element,CQE)放入对应CQ。应用程序可通过轮询或中断等方式获取CQ中的操作。
在资源回收阶段,RDMA操作已全部执行完成,RDMA应用程序依次销毁QP,CQ,MR等内存资源,并关闭RDMA硬件网卡设备,避免产生内存泄漏等问题。此过程同样需要RDMA内核态驱动程序参与。
在高速传感数据采集领域中应用RDMA技术时,需重点考虑其嵌入式场景特点:首先,嵌入式系统受限于算力和存储资源,如采用低功耗处理器ARM Cortex-M系列等,难以直接运行标准RDMA软件栈;其次,面向高速传感数据采集的嵌入式系统,功能相对单一,无需支持完备的RDMA通信原语和传输方式;最后,嵌入式系统多运行轻量级实时操作系统,缺乏通用操作系统的多任务调度和虚拟内存管理等高级功能。鉴于上述嵌入式系统场景特点,进行完整的RDMA驱动程序和建链协议栈的直接移植难以满足场景需求且并不必要[14]。
因此,需面向高速传感数据采集的嵌入式场景实现轻量化RDMA软件栈,并满足以下设计目标:1)保证RDMA标准兼容性。保证RDMA标准兼容性一方面可以保证嵌入式端与商用RDMA网卡的互操作性,有利于高速数据采集嵌入式端直接接入现有RDMA通信网络;另一方面可以实现嵌入式RDMA软件栈对硬件标准RDMA网卡中的典型处理逻辑,如RDMA报文封装解析以及可靠传输保证等的兼容配置。2)支持硬件灵活配置。在高速传感数据采集场景中,不同类型传感设备往往具备不同的流量特征,如温湿度传感器等可能进行周期性数据采集并将其打包为独立的数据单位进行块式传输,摄像头则进行音视频流数据的持续采集并将其以连续字节流的形式进行流式传输。即便同类型传感设备,以摄像头为例,不同分辨率对应的帧图像大小均有不同,这需要嵌入式RDMA软件栈实现灵活的硬件RDMA网卡处理逻辑传输配置。3)避免嵌入式端弱核影响传输性能。在1.2节标准RDMA工作流程中,每次RDMA传输前的操作启动和传输后的完成处理均需要软件栈参与,在RDMA操作频繁的情况下,尤其是对流式传感数据传输,弱核参与会显著影响全系统传输性能,这需要嵌入式RDMA软件栈进行合理的软件功能卸载以提升系统整体传输效率。
标准RDMA软件栈实现依赖于操作系统,并需要RDMA应用程序、RDMA用户态驱动、内核态驱动和用于建链的Socket TCP/IP协议栈协同工作。轻量化嵌入式RDMA软件栈不依赖操作系统实现,无需区分用户态和内核态,从而实现嵌入式RDMA软件栈的轻量化设计。
嵌入式RDMA软件栈整体功能架构包括4个主要模块:1)地址解析协议(address resolution protocol,ARP)请求响应模块。嵌入式RDMA端作为一个网络通信设备,应支持ARP协议以便处理ARP请求/响应,方便网络设备间地址解析与通信。2)存储空间管理模块。嵌入式RDMA软件栈需要在前期配置阶段创建MR,CQ,QP资源,并进行相关内存空间地址信息维护。由于嵌入式RDMA软件栈相关存储资源均使用物理地址,因此无需维护内存转换表(memory translation table,MTT)表项信息。3)通信双端建链交互模块。该模块主要负责定义建链结构体(包括通信对端内存地址等信息),完成UDP报文的封装与解析,并实现丢包重传处理,详见2.3节。4)RDMA嵌入式端硬件配置模块。该模块主要负责嵌入式端RDMA硬件网卡处理逻辑所需元数据信息配置以及RDMA传输参数配置,详见2.4节。
嵌入式RDMA软件栈的工作流程为:在系统上电后首先对所采用的RISC-V微核进行初始化配置,由ARP请求响应模块构造ARP请求并处理响应,更新ARP地址解析信息。由存储空间管理模块进行MR等地址信息初始化,并由硬件RDMA配置模块通过自定义以太网报文对硬件RDMA处理逻辑正常通信所需QPC(QP context)、CQC(CQ context)、SGID(RDMA网络节点标识符)、SMAC(源MAC地址)等元数据信息进行配置。随后,由通信双端建链交互模块根据存储空间管理模块相关内存信息进行本端建链结构体初始化,并据此构造UDP报文进行双端通信,建链交互完成后由RDMA嵌入式端硬件配置模块进行QPC配置并实现QP状态转换。此外,RDMA嵌入式端硬件配置模块还会进行WQE携带数据量等传输参数配置。
相较于标准RDMA软件栈,嵌入式RDMA软件栈直接从嵌入式场景下的硬件配置需求出发进行设计,无需操作系统参与,省去了系统调用、虚实地址转换、资源申请注册及回收操作,并通过轻量化UDP建链、硬件卸载RDMA传输请求下发和传输完成处理进一步实现了嵌入式RDMA软件栈轻量化。在此对嵌入式RDMA软件栈未设计专门的资源申请注册即回收机制的合理性进行说明:1)高速传感数据采集应用场景固定,如自动驾驶传感数据采集设备往往只更新不移除;2)资源管理本身不依赖于操作系统,无需操作系统相关的资源申请和注册、管理等操作;3)传感数据采集设备移除时直接更新QP配置即可实现资源重新分配,无需复杂注册回收流程。RDMA轻量化软件栈总体架构如图2所示。
RDMA协议标准并未规定双端进行信息交互的建链方式或建链协议栈,本文针对高速传感数据采集嵌入式场景,提出一种基于可靠UDP的轻量化RDMA建链方案,以尽可能实现建链轻量化。从可选建链方式来看,TCP 卸载引擎虽能卸载 TCP/IP 协议栈,但因协议复杂性导致硬件成本过高;RDMA CM 建链依赖完整 RDMA 协议栈,与轻量化设计背道而驰;UDP 卸载引擎虽硬件复杂度低、算力占用少,适配嵌入式场景,但原生 UDP 协议不可靠,无法保障关键控制信息传输。对此,基于可靠UDP的轻量化RDMA建链可在不依赖于协议栈的情况下直接进行UDP报文构造和解析,并通过扩展序列号字段校验和超时重传保证传输可靠性。该方式针对RDMA建链这一特定需求进一步轻量化,避免了建链协议栈带来的系统资源开销。
如图3所示为RDMA轻量化软件栈中建链交互内容和交互报文格式。为实现RDMA协议标准兼容,建链交互内容包括:1)网络通信相关信息,即SGID和本地标识符(local identifier,LID),分别用于支持RDMA通信设备跨网和子网内通信。2)注册MR相关信息,即虚拟地址(virtual address,VA)和远程访问密钥(remote key,RKey),方便交互节点凭借Rkey校验权限和VA地址信息直接进行远端内存访问。3)QP相关信息,即QP编号(QP number,QPN),在每个RDMA设备创建多个QP的情况下,QPN交互对于通信双方建立连接和正确进行数据包处理是必要的。此外,在RC传输方式下,通信双方交互分组序列号(packet sequence number,PSN),用于检测报文是否按序接收进而进行可靠传输保证。在双端交互报文格式上,基于可靠UDP的RDMA建链遵循UDP/IP标准协议格式,帧类型、首部长度和上层协议等字段可直接指定,IP首部校验和、UDP首部校验和这2个字段根据校验算法进行填充。此外,为保证传输可靠性,在UDP报文首部基础上增加2B扩展序列号字段,该字段根据当前系统时间随机生成,并根据响应报文回送序列号进行可靠传输校验。
在具体实现上,建链交互模块主要通过函数connect_send,connect_recv,connect_interact实现双端通信功能并为本端其他模块提供connect_remote_query查询接口。函数connect_send负责UDP建链请求报文各字段配置、校验和计算、序列号生成,并通过RISC-V处理器提供的底层报文发送函数rv_send进行报文发送。函数connect_recv则负责通过函数rv_recv进行报文接收和识别,对收到的UDP建链响应报文进行校验检查,并根据校验结果解析报文建链结构体信息或返回错误信息。函数connect_interact则作为建链交互模块对外提供的调用接口,根据主函数提供的本端建链结构体信息和QPN调用函数connect_send发送报文,并循环调用函数connect_recv进行响应报文接收和校验,并在校验错误或超时未收到响应时重传报文。如图4所示为基于可靠UDP的轻量化RDMA建链工作流程示意图。
基于可靠UDP的轻量化RDMA建链只需在商用RDMA网卡的上层软件中将RDMA应用程序中TCP Socket函数替换为UDP Socket函数,并在UDP载荷中增加序列号字段即可,无需修改RDMA硬件网卡逻辑或RDMA标准报文交互流程,因此完全兼容标准RDMA协议,实现嵌入式端与接入商用RDMA网卡标准端交互。
RDMA内核态驱动完成RDMA工作流程前期配置阶段的资源准备和QP状态转换部分以及资源回收阶段,RDMA用户态驱动程序则主要完成数据传输阶段的操作请求下发和完成通知处理部分。本文提出一种针对高速传感数据采集嵌入式场景,基于可靠以太网的轻量化RDMA软件驱动,包括轻量化RDMA元数据信息配置方法以实现前期配置阶段的资源准备和QP状态转换部分、轻量化RDMA传输参数配置以及硬件卸载机制以实现数据传输阶段的操作请求下发和完成通知处理部分,从而避免了将标准RDMA驱动程序直接进行嵌入式移植存在的实现复杂度高、功能冗余、资源占用高等方面问题。
轻量化RDMA元数据信息配置主要用于将QP,CQ,MR等资源相关参数配置给嵌入式端RDMA硬件网卡处理逻辑,以保证网卡进行资源访问和RDMA报文处理。由于仅涉及嵌入式端软硬件间通信,因此在配置方式上,轻量化RDMA元数据信息配置无需考虑双端RDMA通信兼容性问题,采用自定义网络附加控制协议(network attached control protocol,NACP)和Mlbox交互机制联合实现。NACP协议遵循以太网报文格式,并定义序列号字段和元数据信息配置命令字段作为以太网报文有效载荷。如图5所示,其中,序列号字段与2.3节中UDP载荷扩展字段作用一致,用于保证NACP传输可靠性。元数据信息配置命令字段根据配置地址、配置数据量等信息进行了字段定义,使用字段Type区分配置硬件元数据信息的请求/响应报文和查询硬件元数据信息的请求/响应报文,使用字段Burst/Tag说明配置或查询数据字节数,使用字段Addr区分配置或查询地址。轻量化RDMA软件驱动提供的配置请求报文和网卡硬件的响应报文携带配置和查询的元数据信息。在交互机制上,元数据信息配置涉及QPC、CQC、SGID、SMAC、内存保护表(memory protection table,MPT)用于校验内存访问权限等多项配置,Mlbox交互机制用于传输元数据配置相关说明信息。
如图6所示,以QPC配置为例,软硬件通过NACP配置请求/响应报文首先进行QPC元数据信息的交互,之后进行QPC配置掩码信息交互,最后进行QPC配置来描述信息cmd_data的信息交互,获取当前硬件是否处于可配置状态,并根据查询状态决定继续查询或通过cmd_data完成本次元数据信息配置。上述cmd_data主要包括用于指定本次配置或查询对应的QPN/CQN等的目标编号Num,用于区分具体元数据信息类型的操作码Opcode以及用于指示当前硬件配置状态的状态标识位Comp。
在具体实现上,不同元数据信息配置流程基本一致。以需要实现QP状态转换且配置流程较为复杂的QPC配置为例,元数据信息配置主要包括QPC数据结构配置、NACP命令构造、NACP报文收发处理3部分。QPC数据结构配置包括用于QP状态转换的qpc_config_INIT_RTR,qpc_config_RTR_RTS等函数,这些函数会进一步调用函数qpc_config进行QPC、QPC掩码以及cmd_data结构体赋值。NACP命令构造包括作为外部函数调用接口以及用于NACP命令构造的函数qpc_encap,用于掩码和cmd_data信息NACP命令构造的send_datamask,read_cmd_data,send_cmd_data等函数,这些函数除了进行对应NACP命令构造外,还会调用底层NACP报文收发接口进行报文收发。NACP报文收发处理使用函数nacp_write_send作为外部调用接口,该函数调用请求报文封装函数进行NACP写请求报文封装,调用响应报文解析函数进行NACP写响应报文处理,并在响应报文超时或校验失败时进行请求报文重传以保证配置可靠性。
标准RDMA用户态驱动需要参与传输启动和传输完成处理,为避免传感端的弱核处理器参与数据传输过程而影响RDMA传输性能,本文将上述RDMA用户态驱动参与的相关操作卸载到嵌入式端RDMA硬件网卡处理逻辑中实现,同时为嵌入式RDMA软件栈提供传输参数配置接口,兼顾传输性能和灵活性。
轻量化RDMA传输参数配置,主要用于传感端数据传输相关系数设置,包括QPN和WQE传输数据大小、传感端数据存储的DDR(double data rate SDRAM)首地址与长度、传输启停状态等。传输参数配置同样使用NACP报文格式,通过以太网帧类型区分传输参数配置报文和上述元数据配置报文,配置内容则主要取决于高速传感数据采集端的流量特性。值得注意的是,嵌入式RDMA软件栈在收到通信对端的UDP采集控制报文后才可进行硬件传输启停参数配置。
在具体实现上,传输参数配置与上述元数据配置流程类似,调用NACP命令构造以及NACP报文交互相关函数即可。传输启停状态配置涉及用于通信对端进行UDP报文交互的相关函数,其中函数collect_recv用于对收到的UDP传输启停控制报文进行识别和校验,函数req_pkt_parse用于解析控制报文携带的传输配置信息,函数resp_pkt_send用于构造对应UDP传输控制响应报文,函数collect_config用于进行硬件传输启停状态配置。传感端卸载标准RDMA用户态驱动,由网卡硬件完成 RDMA传输启动和完成处理操作,避免传感端弱核处理器参与数据传输过程而影响RDMA传输性能,嵌入式端网卡根据软件配置的传输参数进行WQE构造、SQ Doorbell生成、CQE解析以及CQ Doorbell下发。
轻量化RDMA元数据配置和传输参数配置,基于嵌入式传感端软硬件交互需求,通过对高速传感数据采集场景的轻量化软件栈设计,实现了与标准RDMA驱动程序一致的底层功能,在最小化软件开销的前提下,实现RDMA传输协议,并在支持硬件灵活配置的同时规避了传感端弱算力对传输性能的影响。
实验物理环境如图7所示,实验平台主要包括用于运行嵌入式RDMA轻量化软件栈的DP4C开发板[22]、用于模拟采集传感数据生成和接入的PC主机和ALINX FH1219适配模块、用于运行硬件传输相关卸载模块和RDMA处理逻辑的VCU118开发板[23],以及由飞腾D2000 CPU、华为Atlas 300I Pro推理卡和NVIDIA Mellanox ConnectX-5网卡构成的RDMA通信对端。其中VCU118开发板作为嵌入式端RDMA网卡硬件逻辑载体,将DP4C下发的建链交互报文和UDP传输启停控制报文转发至RDMA通信对端,将DP4C相关元数据配置报文和传输参数配置报文解析后,进行嵌入式端RDMA网卡硬件模块配置,实现RDMA请求响应报文的收发处理。RDMA通信对端由飞腾D2000支持标准RDMA驱动程序和操作系统运行以及与嵌入式端的建链和传输控制交互,以及由商用Mellanox网卡负责与嵌入式端的RDMA通信,验证嵌入式RDMA软件栈的功能正确性和标准RDMA协议兼容性,由华为Atlas 300I Pro推理卡完成从嵌入式端传输来的视频流的实时目标检测和分析处理。通信双端通过100 Gbps光纤互联,理论带宽最大可支持100 Gbps。
在此对承载嵌入式RDMA软件栈的DP4C开发板做进一步说明。DP4C开发板使用自研RISC-V芯片,该芯片集成3个通用RISC-V核处理引擎以及1个AI处理引擎,并提供配套软件开发环境,如上文所述函数rv_send,rv_recv以及系统打印和字符串处理等功能函数。嵌入式RDMA轻量化软件栈在Linux平台经C语言编程开发后,通过配套RISC-V交叉编译链和DP4C配置程序部署至DP4C开发板运行。
由于嵌入式相关软件代码部署在RISC-V芯片微核上执行,无法直接与标准RDMA软件栈对比CPU利用率或内存占用率,因此本文从芯片占用面积、功耗和存储资源3个方面对其轻量化实现后的资源开销进行评估。
在芯片占用面积方面,DP4C开发板中RISC-V芯片基于130 nm工艺流片,芯片总面积100 mm2,4个处理引擎总面积约37 mm2,嵌入式RDMA软件栈只使用其中一个RISC-V核处理引擎,实际执行占用面积仅约9 mm2。在功耗方面,DP4C芯片整体功耗约1.4 W,嵌入式RDMA软件栈单个RISC-V核运行所占功耗远低于此,可以保证低功耗运行。
如图8所示,在存储资源方面,对于通信双端建链,使用cloc代码统计工具对Linux6.15版本下的标准Socket TCP/IP建链协议栈代码量进行了统计,标准TCP建链协议栈内核态代码约24148行,主要包括TCP连接管理\状态机\拥塞控制等核心程序,用户态Socket封装代码约2552行。相比于标准TCP建链协议栈,基于无操作系统可靠UDP的轻量化RDMA建链协议(可看作用户态代码)代码总量仅约1371行,可降低约94%的建链开销。如图9所示,标准RDMA用户态驱动程序代码约12909行,内核态驱动程序代码约3320行,而基于可靠以太网的轻量化RDMA软件驱动代码总量仅约2503行,可降低约84%的软件驱动开销。上述结果表明,本文设计的嵌入式RDMA软件栈可大幅降低RDMA软件栈所需存储空间和计算资源,并完整支持通信双端RDMA传输,达到了较好的轻量化效果,满足嵌入式应用场景特性。
以上主要对嵌入式端RDMA轻量化软件栈资源开销进行评估,为对比嵌入式RDMA轻量级可靠UDP建链和标准RDMA基于socket TCP建链的CPU开销,分别搭建2个环境。环境1:请求端使用轻量化RDMA网卡,响应端为Mellanox商用百G网卡,双端通过轻量级udp建链。环境2:请求端使用Mellanox商用百G网卡,响应端同样为Mellanox商用百G网卡,双端通过socket TCP完成建链。为了更客观地统计双端CPU负载情况,我们将运行在请求端轻量化RDMA的RISC-V微核上的建链程序,代理运行在响应端商用CPU上,通过远程配置请求端,在同一CPU上统计负载占用率。环境1中,请求端轻量化RDMA建链CPU负载占用率为22.8%,响应端的CPU负载为11.8%。环境2中,建链请求端的CPU负载占用率为54.0%,响应端的CPU负载为26.9%。
测试结果如表1所示,轻量化RDMA +轻量级可靠UDP的建链相较于标准RDMA+ Socket TCP建链,能够显著降低CPU负载。
为了进一步验证轻量化RDMA应用于高速传感数据采集场景带来的低网络处理开销优势,本节在不同带宽模拟视频流输入情况下,对通信对端的CPU利用率进行评估,实验结果如图10所示,通信对端的CPU利用率相关开销稳定在4.2%~5.6%之间,主要用于CQE轮询以确保视频流实时分析处理。对于块式传输数据而言,通信对端的CPU利用率约占0%。相比于基于Socket TCP的数据传输(使用iPerf在相同消息大小和MTU为4096 B的配置下进行测试),基于嵌入式RDMA软件栈的高速数据采集系统在保证低CPU网络处理开销方面优势显著,能够满足嵌入式应用场景需求。
本节对嵌入式RDMA轻量化软件栈配置下的系统传输性能进行测试。3.2节中,采用PC模拟视频采集数据,通过SDI接口接入传感端的功能测试环境,系统传输性能受限于输入的PC模拟采集数据生成的视频流带宽,不能测试出系统传输性能上限。因此,本节性能测试环境将模拟传感数据首先存入嵌入式端DDR,通过嵌入式RDMA将存入传感端DDR的视频流数据与对端进行RDMA通信来完成传输性能测试,重点评估嵌入式RDMA软件栈工作条件下的双端RDMA实际传输性能,主要分为传输带宽性能评估和传输延迟测量。为验证系统传输能力,在传感终端和标准算力终端之间接入交换机,支持单接口100 Gbps交换能力,以下带宽及延迟均基于下一跳交换进行。
传输带宽性能评估旨在研究面向高速传感数据采集的嵌入式RDMA软件栈是否可满足多源传感数据的高带宽、高实时传输需求。实验使用perftest设置不同消息大小,并分别选取MTU为512 B/1024 B/2048 B/4096 B及QP数量为1/10/30的情况下对其传输延迟进行测试,结果如图11所示。分析表明,系统吞吐量随QP数量、MTU尺寸及消息大小的增加而显著提升。在系统时钟频率为250 MHz的条件下,当QP数量为30、MTU设置为4096 B、消息大小约为1 MB时,最大单向吞吐量达到98.04 Gbps。这一结果验证了嵌入式RDMA轻量化软件栈在并发连接场景下具有良好的传输性能和扩展性,能够有效满足高带宽传感数据传输需求。我们同时注意到单QP条件下的峰值性能已逼近多QP场景下的性能,这在高速数据采集应用中具有十分重要的意义,因为嵌入式RDMA数据采集端通常使用较少的QP对进行海量传感数据融合传输。
如图12所示,标准RDMA协议栈在不同QP数量和MTU配置下的峰值传输带宽表现出与嵌入式RDMA协议栈相似的变化趋势。然而在QP数量为1时,标准RDMA协议栈WRITE与READ操作的峰值带宽仅有54.91 Gbps和34.70 Gbps,比嵌入式RDMA协议栈分别低52%和148%,充分说明嵌入式RDMA协议栈在高速数据采集应用中的优越性。
传输延迟测试用于验证面向高速传感数据采集的嵌入式RDMA软件栈是否能够满足高速传感数据的低延迟传输需求。测试条件同上,分别选择较小消息(256 B)以测试端到端WRITE,READ,SEND,RECV传输的最低延迟。在此说明,perftest 测试所得传输延迟为端到端总延迟,包含终端应用下发 WQE 与 Doorbell、RDMA 报文处理逻辑解析 WQE 并构造报文、链路传输、接收全部预期 ACK、RDMA 报文处理逻辑生成 CQE,以及终端应用接收并完成 CQE 处理的全流程耗时。如图13所示,传输数据量(消息大小)是影响端到端延迟的关键因素,且在系统时钟频率250 MHz的配置下,当采用QP数量为1、MTU设置为2048 B、消息大小约为256 B时,过一级交换,WRITE操作的端到端传输延迟最低仅为4.58 μs,其中单跳交换延迟约2 μs。这一结果验证了算控网络在满足高带宽传感数据接入的同时,能够有效满足高速传感数据的低延迟实时传输需求。
如图14所示,标准RDMA协议栈在不同MTU大小下的延迟表现与嵌入式实现基本相当,然而标准RDMA协议栈在不同操作类型下随MTU尺寸变化的延迟表现存在显著差异,在较小MTU配置下,READ操作的延迟明显高于WRITE与SEND/RECV操作,这是因为READ操作需要发起端首先发送READ请求报文至对端,等待对端返回数据报文,此双向交互过程在MTU较小时会产生更多的报文数量与更频繁的上下文切换,显著增加了端到端延迟。相比之下,嵌入式RDMA中MTU对于延迟的影响相对较小。
为验证嵌入式RDMA端与商用通信对端的协议兼容性,我们分别测试了二者的WRITE操作吞吐量和READ操作吞吐量随消息大小和QP数量的变化趋势,我们将MTU固定为4096 KB以获取峰值带宽。如图15所示,随着QP数量和数据量增加,吞吐量稳定提高,在足够数量的QP以及大数据量的情况下,嵌入式RDMA端的WRITE和READ性能均能够达到商用通信对端峰值性能(98 Gbps及以上)。单QP条件下,嵌入式RDMA端发起对商用通信对端的READ请求,带宽提升速率明显低于商用通信对端对嵌入式RDMA端的READ带宽提升速率,这是因为嵌入式RDMA中,传输状态及上下文信息均保存在网卡本地,减少了因访问CPU DDR内存带来的延迟,因此当嵌入式RDMA被读时,性能更优。
嵌入式RDMA端与商用通信对端的延迟趋势测试是在QP为1的条件下进行的。如图16所示,当传输数据量较小时(≤104 KB),二者的WRITE和READ延迟始终维持在微秒级水平,随着数据量增大,二者延迟呈现同步增大趋势。得益于嵌入式 RDMA 端将部分软件栈功能卸载至硬件,嵌入式 RDMA 端主动向商用对端发起的WRITE操作延迟,以及商用对端向嵌入式 RDMA 端发起的WRITE操作延迟均相对较低。上述关于带宽和延迟的观察结果充分证明嵌入式RDMA能够实现与标准RDMA的高带宽轻量化传输兼容性,在主动WRITE和被动READ方面性能更优。
我们分别进行了2组长达72 h的压力测试,用于模拟并验证嵌入式RDMA在高速传感场景下的运行带宽和延迟稳定性。图17展示了72 h内带宽与延迟的动态变化情况,测试结果表明,在整个测试周期内系统带宽始终稳定在约91 Gbps的高位水平,波动幅度小于0.1%,表现出卓越的带宽稳定性。在另一组实验中,传输延迟在整个测试过程中始终维持在约5μs的稳定水平,未出现明显的性能抖动或异常波动。需要说明的是,由于系统启动会有短暂的突发状况,因此前期极少报文延迟达到了10.3 μs,随着带宽的稳定,延迟快速收敛至5 μs。这一长时间连续测试结果充分验证了嵌入式RDMA协议栈在持续高负载条件下的可靠性与鲁棒性,能够满足高速传感场景下对传输系统长期稳定运行的严苛要求,为实际工程部署提供了有力的性能保障。
本文对RDMA标准软件栈和处理流程进行了详细分析,并据此提出面向高速传感数据采集的嵌入式RDMA轻量化软件栈技术及实现方法,从RDMA建链和RDMA软件驱动2个方面进行RDMA软件栈轻量化。其中,基于可靠UDP的轻量化RDMA建链保证了通信双端前期配置阶段信息交互;基于可靠以太网的RDMA元数据配置保证了嵌入式端的RDMA标准兼容性;基于可靠以太网的RDMA传输参数配置和传输控制软件功能选择性卸载,在保障扩展灵活性的同时避免了传感端弱核影响嵌入式端传输性能,满足了嵌入式系统应用场景需求,提升全系统能力。实验结果表明,相较于标准 RDMA 软件栈,本文所设计的轻量化嵌入式 RDMA 软件栈可在无高性能 CPU、无操作系统的约束下正常运行,为高速传感数据采集场景的高带宽、低延迟网络传输提供了可靠的软件支撑,并构建了从传感端到标准端的完整系统解决方案。

参考文献 引证文献
排序方式:
1
黄曼蒂, 李韬, 杨惠, 等. 以太网RDMA网卡综述[J]. 计算机研究与发展, 2025, 62(5): 1262−1289
Huang Mandi, Li Tao, Yang Hui, et al. Survey on Ethernet RDMA network interface card[J]. Journal of Computer Research and Development, 2025, 62(5): 1262−1289(in Chinese)
2
Chowdhury A, Karmakar G, Kamruzzaman J, et al. Attacks on self-driving cars and their countermeasures: A survey[J]. IEEE Access, 2020, 8: 207308−207342
3
Xia Dan, Jiang Chun, Wan Jiafu, et al. Heterogeneous network access and fusion in smart factory: A survey[J]. ACM Computing Surveys, 2023, 55(6): 113.1−113.31
4
Luo Huimin, Zhang Jiao, Yu Mingxuan, et al. SeqBalance: Congestion-aware load balancing with no reordering in data center networks[J]. IEEE Internet of Things Journal, 2024, 12(13): 25707−25719
5
Xue Jiachen, Chaudhry M U, Vamanan B, et al. Dart: Divide and specialize for fast response to congestion in RDMA-based datacenter networks[J]. IEEE/ACM Transactions on Networking, 2018, 28(1): 322−335
6
Wang Xiaoliang, Song Hexiang, Nguyen C, et al. Maximizing the benefit of RDMA at end hosts [C] // Proc of the 2021 IEEE Int Conf on Computer Communications. Piscataway, NJ: IEEE, 2021: 1−10
7
Mittal R, Shpiner A, Panda A, et al. Revisiting network support for RDMA [C] //Proc of the 2018 ACM SIGCOMM Conf. New York: ACM, 2018: 313−326
8
Li Yuliang, Miao Rui, Liu H H, et al. HPCC: High precision congestion control [C] // Proc of the 2019 ACM SIGCOMM Conf. New York: ACM, 2019: 44−58
9
Bai Wei, Abdeen S S, Agrawal A, et al. Empowering Azure storage with RDMA [C] // Proc of the 20th USENIX Symp on Networked Systems Design and Implementation (NSDI). Berkeley, CA: USENIX Association, 2023: 49−67
10
Kim D, Yu Tianlong, Liu H H, et al. FreeFlow: Software-based virtual RDMA networking for containerized clouds [C] // Proc of the 16th USENIX Symp on Networked Systems Design and Implementation. Berkeley, CA: USENIX Association, 2019: 113−126
11
Wang Zilong, Luo Layong, Ning Qingsong, et al. SRNIC: A scalable architecture for RDMA NICs [C] // Proc of the 20th USENIX Symp on Networked Systems Design and Implementation. Berkeley, CA: USENIX Association, 2023: 1−14
12
Wang Xizheng, Chen Guo, Yin Xijin, et al. StaR: Breaking the scalability limit for RDMA [C] // Proc of the 29th IEEE Int Conf on Network Protocols. Piscataway, NJ: IEEE, 2021: 1−11
13
Lee J B, Kim S J, Kim W H, et al. RDMA-based sampling port of ARINC-653[J]. IEEE Embedded Systems Letters, 2024, 16(4): 437−440
14
Feng Junjie, Chen Xianzhang, Liu Duo, et al. eRDAC: Efficient and reliable remote direct access and control for embedded systems[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2022, 41(11): 3685−3696
15
Schelten N, Steinert F, Knapheide J, et al. A high-throughput, eesource-efficient implementation of the RoCEv2 remote DMA protocol and its application[J]. ACM Transactions on Reconfigurable Technology, 2022, 16(1): 1−23
16
Cossettini A, Taranov K, Vogt C, et al. A RDMA interface for ultra-fast ultrasound data-streaming over an optical link [C] // Proc of the 2022 Design, Automation & Test in Europe Conf & Exhibition. Piscataway, NJ: IEEE, 2022: 80−83
17
AMD. AMD embedded RDMA enabled NIC LogiCORE IP product guide (PG332)[EB/OL]. (2024-11-22)[2025-06-10]. https://docs.xilinx.com/r/-en-US/pg332-ernic
18
InfiniBand. InfiniBand architecture specification[EB/OL]. [2025-06-10]. https://www.infinibandta.org/ibta-specification/
19
IETF Network Working Group. RFC5040: A remote direct memory access protocol specification[EB/OL]. (2007-10-01)[2025-06-10]. https://www.rfc-edi tor.org/rfc/rfc5040
20
NVIDIA. RDMA over converged ethernet(RoCE)[EB/OL]. [2025-06-10]. https://docs.nvidia.com/networking/display/onyxv3103004/rdma+over+converged+ethernet+(roce
21
He Zhiqiang, Wang Dongyang, Fu Binzhang, et al. MasQ: RDMA for virtual private cloud [C] // Proc of the 2020 ACM SIGCOMM Conf. New York: ACM, 2020: 1−14
22
OpenTSN 工作组. 银河衡芯HX-DP4C边缘智能网络处理器 [EB/OL]. 湖南: OpenTSN工作组, (2023-08-30)[2025-06-10]. https: //mp.weixin.qq.com/s/gh2no71EwjX0a-HGHLiPJg
OpenTSN Group. Yinhe Hengxin HX-DP4C Edge Intelligence Network Processor[EB/OL]. Hunan: OpenTSNGroup. (2023-08-30)[2025-06-10]. https://mp.weixin.qq.com/s/gh2no71EwjX0a-HGHLiPJg (in chinese
23
AMD. The AMD Virtex UltraScale+ FPGA VCU118[EB/OL]. [2025-06-10]. https://www.amd.com/zh-cn/products/adaptive-socs-and-fpgas/evaluation-boards/vcu118.html#tabs-e09e115b99-item-fb75e43802-tab
2026年第63卷第5期
96
45
0
引用本文
BibTeX
文章信息
doi: 10.7544/issn1000-1239.202550472
  • 接收时间:2025-06-17
  • 首发时间:2026-11-17
  • 出版时间:2026-05-01
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-06-17
  • 修回日期:2025-12-19
  • 录用日期:2026-01-07
基金
作者信息
    1国防科技大学计算机学院 长沙 410073
    2天津大学电气自动化与信息工程学院 天津 300072

通讯作者:

参考文献
分享链接
https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/10.7544/issn1000-1239.202550472
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏