Article(id=1325490478069711388, tenantId=1146029695717560320, journalId=1324759914146529478, issueId=1325489890678420381, articleNumber=null, orderNo=null, doi=10.7544/issn1000-1239.202440833, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=1730304000000, receivedDateStr=2024-10-31, revisedDate=1750003200000, revisedDateStr=2025-06-16, acceptedDate=1752681600000, acceptedDateStr=2025-07-17, onlineDate=1793951156393, onlineDateStr=2026-11-06, pubDate=1767196800000, pubDateStr=2026-01-01, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1793951156393, onlineIssueDateStr=2026-11-06, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1793951156393, creator=13701087609, updateTime=1793951156393, updator=13701087609, issue=Issue{id=1325489890678420381, tenantId=1146029695717560320, journalId=1324759914146529478, year='2026', volume='63', issue='1', pageStart='1', pageEnd='272', issueExtLink='null', onlineDate='null', pubDate='1767196800000', pubDateStr='2026-01-01', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1793951016348, creator='13701087609', updateTime=1793952522017, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext={EN=IssueExt(id=1325496205974839692, tenantId=1146029695717560320, journalId=1324759914146529478, issueId=1325489890678420381, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1325496205974839693, tenantId=1146029695717560320, journalId=1324759914146529478, issueId=1325489890678420381, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null, downloadFileDto=null}, startPage=28, endPage=40, ext={EN=ArticleExt(id=1325490479697101341, articleId=1325490478069711388, tenantId=1146029695717560320, journalId=1324759914146529478, language=EN, title=Design and Implementation of Chained DMA Controller Based on PCIe, columnId=1325489901852041729, journalTitle=Journal of Computer Research and Development, columnName=Computer Architecture, runingTitle=null, highlight=null, articleAbstract=

In recent years, although Chinese domestic field programmable gate array (FPGA) manufacturers have developed rapidly, they still face challenges when deploying FPGA heterogeneous accelerators in data centers. Compared with international manufacturers such as Xilinx (now AMD) and Intel, domestic manufacturers generally lack solutions for high-speed transmission between PCIe devices and hosts, especially in the field of high-performance direct memory access (DMA) controller design, where there are obvious shortcomings. To solve this problem, we design and implement a PCIe-based multi-channel chained DMA controller. By using an independent descriptor controller to manage each channel, sharing data movers, and reducing the consumption of FPGA logic resources, this design improves resource efficiency. The adoption of a chain structure for descriptor management reduces CPU interrupt pressure while meeting the requirements for continuous high-speed transmission between hosts and devices. An innovative architecture for asynchronous internal information pre-processing is developed, enabling data stream processing that significantly improves bandwidth utilization and transmission performance. Testing results show that under PCIe Gen3x8, the DMA bandwidth between the hosts and the domestic FPGA accelerator reaches 6.91 GBps (86% utilization rate), supporting up to 16 channels with channel balancing implementation. This design effectively enables large-scale deployment of domestic FPGA heterogeneous accelerators in data center scenarios.

, authors=Hongliang Wang1, Zhenhua Guo1, Qi Mu1, 3, Shengcai Lu1, Yaming Xu2, Quanquan Yu1, authorsList=Hongliang Wang, Zhenhua Guo, Qi Mu, Shengcai Lu, Yaming Xu, Quanquan Yu, authorCompany=null, correspAuthors=null, authorNote=

born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

, 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=1325490484688323142, articleId=1325490478069711388, tenantId=1146029695717560320, journalId=1324759914146529478, language=CN, title=基于PCIe的链式DMA控制器设计与实现, columnId=1325489904427344387, journalTitle=计算机研究与发展, columnName=体系结构, runingTitle=null, highlight=null, articleAbstract=

近年来,国产可编程门阵列(field programmable gate array,FPGA)厂商虽发展迅速,但在数据中心部署FPGA异构加速器时仍面临挑战。相较于赛灵思、英特尔等国际厂商,国产厂商普遍缺乏PCIe设备与主机间高速传输的解决方案,尤其在高性能直接内存访问(direct memory access,DMA)控制器设计领域存在明显短板。为解决该问题,设计并实现了基于PCIe的多通道链式DMA控制器。采用独立描述符控制器管理各通道,共享数据搬移器,降低对FPGA逻辑资源的消耗。采用链式结构实现描述符管理,减少中断对CPU的压力,满足主机与设备连续高速传输的需求。创新性地构建内部信息异步与预处理的架构,实现数据流水化处理,显著提升带宽利用率以及传输性能。经测试,在PCIe Gen3x8下,主机与国产FPGA加速器之间的DMA带宽高达6.91 GBps(利用率86%),支持多达16通道且实现通道负载均衡,该设计有效支撑了国产FPGA异构加速器在数据中心场景下的规模化部署。

, authors=王洪良1, 郭振华1, 牟奇1, 3, 卢圣才1, 徐亚明2, 于泉泉1, authorsList=王洪良, 郭振华, 牟奇, 卢圣才, 徐亚明, 于泉泉, authorCompany=null, correspAuthors=null, authorNote=

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

, correspAuthorsNote=null, copyrightStatement=版权所有 © 2026 计算机研究与发展编辑部, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=dEI8fafN431+Oax8DpJUyQ==, magXml=5ghr6f7BU0dymmE7z04EbQ==, pdfUrl=null, pdf=x1hEF+NHLOFT5ciX4NJSwg==, pdfFileSize=1283687, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=9ZQio5FhW5oJSoeBpxi0+g==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=DuGrQ/oCoBUpEzY8KQSM3g==, mapNumber=null, fund=null)}, authors=[Author(id=1325490485409743441, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=wanghl01@ieisystem.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1325490485493629523, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485409743441, language=EN, stringName=Hongliang Wang, firstName=Hongliang, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

"}, bioImg=null, bioContent=

born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490485556544084, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485409743441, language=CN, stringName=王洪良, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"UaLBYk6iYaONP4yreJ6gtg==","content":"

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

"}, bioImg=UaLBYk6iYaONP4yreJ6gtg==, bioContent=

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490485627847254, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490485812396632, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485627847254, language=EN, stringName=Zhenhua Guo, firstName=Zhenhua, middleName=null, lastName=Guo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

"}, bioImg=null, bioContent=

born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490485875311193, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485627847254, language=CN, stringName=郭振华, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"4kkMzth5/em2+OLgQnaUbA==","content":"

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

"}, bioImg=4kkMzth5/em2+OLgQnaUbA==, bioContent=

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490485934031451, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486009528926, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485934031451, language=EN, stringName=Qi Mu, firstName=Qi, middleName=null, lastName=Mu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101
3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641, bio={"content":"

born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

"}, bioImg=null, bioContent=

born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486076637791, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485934031451, language=CN, stringName=牟奇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1浪潮电子信息产业股份有限公司 济南 250101
3华南理工大学计算机科学与工程学院 广州 510641, bio={"img":"tw7zkmtxWyqLl/bDgKr9tg==","content":"

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

"}, bioImg=tw7zkmtxWyqLl/bDgKr9tg==, bioContent=

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)]), AuthorCompany(id=1325490485090976333, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=3, ext=[AuthorCompanyExt(id=1325490485099364942, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641), AuthorCompanyExt(id=1325490485111947855, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3华南理工大学计算机科学与工程学院 广州 510641)])]), Author(id=1325490486189884001, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486256992867, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486189884001, language=EN, stringName=Shengcai Lu, firstName=Shengcai, middleName=null, lastName=Lu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

"}, bioImg=null, bioContent=

born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486336684644, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486189884001, language=CN, stringName=卢圣才, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"8xKkuKzUhI19IYTIhWyTKQ==","content":"

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

"}, bioImg=8xKkuKzUhI19IYTIhWyTKQ==, bioContent=

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490486395404902, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486466708072, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486395404902, language=EN, stringName=Yaming Xu, firstName=Yaming, middleName=null, lastName=Xu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085, bio={"content":"

born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

"}, bioImg=null, bioContent=

born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486533816937, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486395404902, language=CN, stringName=徐亚明, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2浪潮 (北京)电子信息产业有限公司 北京 100085, bio={"img":"QvSSvtsECXCZsd7ROw3t+A==","content":"

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

"}, bioImg=QvSSvtsECXCZsd7ROw3t+A==, bioContent=

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490485002895946, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=2, ext=[AuthorCompanyExt(id=1325490485011284555, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085), AuthorCompanyExt(id=1325490485019673164, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2浪潮 (北京)电子信息产业有限公司 北京 100085)])]), Author(id=1325490486596731499, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, orderNo=5, 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=1325490486684811885, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486596731499, language=EN, stringName=Quanquan Yu, firstName=Quanquan, middleName=null, lastName=Yu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

"}, bioImg=null, bioContent=

born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486760309358, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486596731499, language=CN, stringName=于泉泉, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"FEbTgGsuFVzc2iz8IffB7w==","content":"

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

"}, bioImg=FEbTgGsuFVzc2iz8IffB7w==, bioContent=

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])])], keywords=[Keyword(id=1325490486856778351, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, orderNo=1, keyword=PCIe), Keyword(id=1325490488479974000, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, orderNo=2, keyword=FPGA), Keyword(id=1325490488580637297, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, orderNo=3, keyword=heterogeneous acceleration), Keyword(id=1325490488664523378, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, orderNo=4, keyword=multi-channel), Keyword(id=1325490488765186675, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, orderNo=5, keyword=chained DMA), Keyword(id=1325490488844878452, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, orderNo=1, keyword=PCIe), Keyword(id=1325490490791035509, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, orderNo=2, keyword=FPGA), Keyword(id=1325490490967196278, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, orderNo=3, keyword=异构加速), Keyword(id=1325490491273380471, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, orderNo=4, keyword=多通道), Keyword(id=1325490492951102072, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, orderNo=5, keyword=链式DMA)], refs=[Reference(id=1325490511812887203, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Kastner. RIFFA: Reusable integration framework for FPGA accelerators [CP/OL]. San Francisco, CA: GitHub, (2015-05-04) [2025-05-08]. https://github.com/KastnerRG/riffa, articleTitle=null, refAbstract=null), Reference(id=1325490511951299236, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=2, rfOrder=1, authorNames=null, journalName=null, refType=null, unstructuredReference=scryver. XAPP1052 [CP/OL]. San Francisco, CA: GitHub, (2019-01-08) [2025-05-08]. https://github.com/scryver/XAPP1052, articleTitle=null, refAbstract=null), Reference(id=1325490512358146726, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=3, rfOrder=2, authorNames=null, journalName=null, refType=null, unstructuredReference=王齐. PCI Express体系结构导读[M]. 北京: 机械工业出版社, 2016, articleTitle=null, refAbstract=null), Reference(id=1325490512655942311, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Wang Qi. Introduction to PCI Express Architecture [M]. Beijing: Machinery Industry Press, 2016 (in Chinese), articleTitle=null, refAbstract=null), Reference(id=1325490512727245480, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=PCI-SIG. PCI Express Base Specification Revision 3.0[S/OL]. Beaverton, OR: PCI-SIG, 2010[2025-05-08]. https://www.pcisig.com, articleTitle=null, refAbstract=null), Reference(id=1325490513129898665, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Budruk R, Anderson D, Shanley T. PCI Express System Architecture[M]. Boston, MA: Addison-Wesley Professional, 2003, articleTitle=null, refAbstract=null), Reference(id=1325490513939399338, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Kavianipour H, Muschter S, Bohm C. High performance FPGA-based DMA interface for PCIe[J]. IEEE Transactions on Nuclear Science, 2014, 61(2): 745−749, articleTitle=null, refAbstract=null), Reference(id=1325490515642286763, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Rota L, Caselle M, Chilingaryan S, et al. A PCIe DMA architecture for multi-Gigabyte per second data transmission[J]. IEEE Transactions on Nuclear Science, 2015, 62(3): 972−976, articleTitle=null, refAbstract=null), Reference(id=1325490516099465900, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Zazo J F, Lopez-Buedo S, Audzevich Y, et al. A PCIe DMA engine to support the virtualization of 40 Gbps FPGA-accelerated network appliances[C]//Proc of the 2015 Int Conf on ReConFigurable Computing and FPGAs (ReConFig). Piscataway, NJ: IEEE, 2015: 1−6, articleTitle=null, refAbstract=null), Reference(id=1325490516560839341, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Wu Deze, Duan Guodong, Ren Minhua, et al. Research and design of descriptor-based DMA for 10 Gigabit Ethernet controller[C]//Proc of the 6th Int Conf on Electronic Engineering and Informatics (EEI). Piscataway, NJ: IEEE, 2024: 855−859, articleTitle=null, refAbstract=null), Reference(id=1325490516967686830, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Forencich A, Snoeren A C, Porter G, et al. Corundum: An open-source 100-Gbps Nic[C]//Proc of the 28th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2020: 38−46, articleTitle=null, refAbstract=null), Reference(id=1325490517269676719, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Kubálek J, Cabal J, Špinler M, et al. DMA Medusa: A vendor-independent FPGA-based architecture for 400 Gbps DMA transfers[C]//Proc of the 29th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2021: 258−258, articleTitle=null, refAbstract=null), Reference(id=1325490517357757104, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Kasai S, Osana Y. A driver-based approach for DMA transfer between FPGA-GPU[C]//Proc of the 10th Int Symp on Computing and Networking Workshops (CANDARW). Piscataway, NJ: IEEE, 2022: 103−108, articleTitle=null, refAbstract=null), Reference(id=1325490517793964721, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Zabołotny W M. Versatile DMA engine for high-energy physics data acquisition implemented with high-level synthesis[J]. Electronics, 2023, 12(4): 883, articleTitle=null, refAbstract=null), Reference(id=1325490519811424946, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Yokono T, Yamabe Y, Tanaka K, et al. FPGA-based accelerators system with low latency autonomous DMA engine[C]//Proc of the 30th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2022: 1−1, articleTitle=null, refAbstract=null), Reference(id=1325490520293769909, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Zhang Honglin, Yang Haibo, Liu Chengcheng, et al. Design of a common high-performance data acquisition system for physics experiments at HIAF[J]. IEEE Transactions on Nuclear Science, 2024, 72: 713−719, articleTitle=null, refAbstract=null), Reference(id=1325490520759337654, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Kobayashi R, Fujita N, Yamaguchi Y, et al. Multi-hybrid accelerated simulation by GPU and FPGA on radiative transfer simulation in astrophysics[J]. Journal of Information Processing, 2020, 28: 1073−1089, articleTitle=null, refAbstract=null), Reference(id=1325490521187156663, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Zhao Yingxiao, Liu Xin, Yang Jiong. A resource and timing optimized PCIe DMA architecture using FPGA internal data buffer[J]. IEICE Electronics Express, 2019, 16(1): 20180858, articleTitle=null, refAbstract=null), Reference(id=1325490521921159864, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Cheng Kun, Liao Shengkai, Shen Qi, et al. Design and implementation of high-throughput PCIe with DMA architecture between FPGA and powerPC[J]. arXiv preprint, arXiv: 1809.07702, 2018, articleTitle=null, refAbstract=null), Reference(id=1325490522726466234, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Zabołotny W M, Gumiński M, Kruszewski M, et al. Control and diagnostics system generator for complex FPGA-based measurement systems[J]. Sensors, 2021, 21(21): 7378, articleTitle=null, refAbstract=null), Reference(id=1325490524420965051, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Shanehsazzadeh F, Sadri M S. Area and performance evaluation of central DMA controller in Xilinx embedded FPGA designs[C]//Proc of the 2017 Iranian Conf on Electrical Engineering (ICEE). Piscataway, NJ: IEEE, 2017: 546−550, articleTitle=null, refAbstract=null), Reference(id=1325490524873949885, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=王彦伟, 李仁刚, 徐冉, 等. 基于可重构架构的数据中心异构加速软硬件系统级平台[J]. 计算机研究与发展, 2025, 62(4): 963−977, articleTitle=null, refAbstract=null), Reference(id=1325490525717004990, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=21, rfOrder=22, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang Yanwei, Li Rengang, Xu Ran, et al. Data center heterogeneous acceleration software-hardware system-level platform based on reconfigurable architecture[J]. Journal of Computer Research and Development, 2025, 62(4): 963−977 (in Chinese), articleTitle=null, refAbstract=null), Reference(id=1325490526518117055, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=Chandra K, Jagtap A P, Ranjan N, et al. Design of PCIe-DMA bridge interface for high speed ethernet applications[C]//Proc of the 2019 2nd Int Conf on Advanced Computational and Communication Paradigms (ICACCP). Piscataway, NJ: IEEE, 2019: 1−5, articleTitle=null, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)]), AuthorCompany(id=1325490485002895946, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=2, ext=[AuthorCompanyExt(id=1325490485011284555, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085), AuthorCompanyExt(id=1325490485019673164, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2浪潮 (北京)电子信息产业有限公司 北京 100085)]), AuthorCompany(id=1325490485090976333, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=3, ext=[AuthorCompanyExt(id=1325490485099364942, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641), AuthorCompanyExt(id=1325490485111947855, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3华南理工大学计算机科学与工程学院 广州 510641)])], figs=[ArticleFig(id=1325490493441835641, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 1, caption=PCIe hierarchical structure, figureFileSmall=Ef/9iEx1rKUAeOaxOmFY/A==, figureFileBig=9ZQio5FhW5oJSoeBpxi0+g==, tableContent=null), ArticleFig(id=1325490493794157178, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图1, caption=PCIe层次结构, figureFileSmall=Ef/9iEx1rKUAeOaxOmFY/A==, figureFileBig=9ZQio5FhW5oJSoeBpxi0+g==, tableContent=null), ArticleFig(id=1325490498223342204, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 2, caption=PCIe packet header format, figureFileSmall=8tnYN7Ch+MJVhFXf80nYIA==, figureFileBig=SItJcZhuJjrL4os6B9qUSQ==, tableContent=null), ArticleFig(id=1325490498412085885, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图2, caption=PCIe 包头格式, figureFileSmall=8tnYN7Ch+MJVhFXf80nYIA==, figureFileBig=SItJcZhuJjrL4os6B9qUSQ==, tableContent=null), ArticleFig(id=1325490498512749182, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 3, caption=Schematic diagram of system to FPGA internal bus address space, figureFileSmall=JWzTEl2x9pQoFSJ/WCdN4A==, figureFileBig=850ebskbcS3L9rPhxjed8w==, tableContent=null), ArticleFig(id=1325490498630189695, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图3, caption=系统到FPGA内部总线地址空间示意图, figureFileSmall=JWzTEl2x9pQoFSJ/WCdN4A==, figureFileBig=850ebskbcS3L9rPhxjed8w==, tableContent=null), ArticleFig(id=1325490498814739072, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 4, caption=Schematic diagram of descriptor configuration table format, figureFileSmall=h5TtbLP/SP/3257z92fdew==, figureFileBig=N3MMKbEvbMgf30weILRGhQ==, tableContent=null), ArticleFig(id=1325490498907013761, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图4, caption=描述符配置表格式示意图, figureFileSmall=h5TtbLP/SP/3257z92fdew==, figureFileBig=N3MMKbEvbMgf30weILRGhQ==, tableContent=null), ArticleFig(id=1325490498999288450, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 5, caption=Schematic diagram of the configured descriptor, figureFileSmall=Bz95lMOvFbt+ijM7+P4D5w==, figureFileBig=rUefjm4IqFQwySJiNuRlew==, tableContent=null), ArticleFig(id=1325490499238363779, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图5, caption=配置后的描述符示意图, figureFileSmall=Bz95lMOvFbt+ijM7+P4D5w==, figureFileBig=rUefjm4IqFQwySJiNuRlew==, tableContent=null), ArticleFig(id=1325490499372581509, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 6, caption=DMA controller architecture diagram, figureFileSmall=VCWZz2ZpcWYMQPiD7NK3Bg==, figureFileBig=7HooLG6Afb2WbmtDUGszHA==, tableContent=null), ArticleFig(id=1325490499712320134, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图6, caption=DMA控制器架构图, figureFileSmall=VCWZz2ZpcWYMQPiD7NK3Bg==, figureFileBig=7HooLG6Afb2WbmtDUGszHA==, tableContent=null), ArticleFig(id=1325490499804594823, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 7, caption=Multi-channel and scheduling balance, figureFileSmall=duL7bNH6I1eClHFVaNfqCg==, figureFileBig=ugNGTFr8vHUITTfZJGMy/g==, tableContent=null), ArticleFig(id=1325490500119167624, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图7, caption=多通道与调度均衡, figureFileSmall=duL7bNH6I1eClHFVaNfqCg==, figureFileBig=ugNGTFr8vHUITTfZJGMy/g==, tableContent=null), ArticleFig(id=1325490500249191049, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 8, caption=H2C data mover, figureFileSmall=w9/TEyB+KEePoOeqV+IYVw==, figureFileBig=f00pJwqHFl0ryo+Jz4GVmA==, tableContent=null), ArticleFig(id=1325490501939495562, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图8, caption=H2C数据搬移器, figureFileSmall=w9/TEyB+KEePoOeqV+IYVw==, figureFileBig=f00pJwqHFl0ryo+Jz4GVmA==, tableContent=null), ArticleFig(id=1325490502585418379, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 9, caption=C2H data mover, figureFileSmall=HADS7/fiFwG/jeI+0fW5lw==, figureFileBig=3nhJqVYylUMbGFFqEhNICg==, tableContent=null), ArticleFig(id=1325490503050986124, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图9, caption=C2H数据搬移器, figureFileSmall=HADS7/fiFwG/jeI+0fW5lw==, figureFileBig=3nhJqVYylUMbGFFqEhNICg==, tableContent=null), ArticleFig(id=1325490503516553869, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 10, caption=Tag management and disorderly reorganization, figureFileSmall=q5rDZv/zC2GeSecnT2O2wQ==, figureFileBig=ErdxsPhxJ5pTsL7LGJ7Ang==, tableContent=null), ArticleFig(id=1325490503852098190, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图10, caption=Tag管理与乱序重组, figureFileSmall=q5rDZv/zC2GeSecnT2O2wQ==, figureFileBig=ErdxsPhxJ5pTsL7LGJ7Ang==, tableContent=null), ArticleFig(id=1325490504317665935, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 11, caption=Hardware board diagram, figureFileSmall=0cpE5P9xCUrQKu5l3k6Zkg==, figureFileBig=R4H9s6v4L5/D8MNrrfUVgg==, tableContent=null), ArticleFig(id=1325490506410623634, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图11, caption=硬件板卡图, figureFileSmall=0cpE5P9xCUrQKu5l3k6Zkg==, figureFileBig=R4H9s6v4L5/D8MNrrfUVgg==, tableContent=null), ArticleFig(id=1325490506800693907, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 12, caption=DMA performance test chart under four channels, figureFileSmall=Ti9UpgCMmNbo2KnOmV2IcA==, figureFileBig=xLiPAq1ddk3AhbieV1iTYw==, tableContent=null), ArticleFig(id=1325490507228512917, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图12, caption=4通道下DMA性能测试图, figureFileSmall=Ti9UpgCMmNbo2KnOmV2IcA==, figureFileBig=xLiPAq1ddk3AhbieV1iTYw==, tableContent=null), ArticleFig(id=1325490507622777494, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 13, caption=Comparative diagram of DMA performance under different modes, figureFileSmall=7xWzQ2WyUqnHXXtP5XdINg==, figureFileBig=N/5dbzw7HqkgoAbNu/d2/A==, tableContent=null), ArticleFig(id=1325490508063179415, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图13, caption=不同模式下DMA性能对比图, figureFileSmall=7xWzQ2WyUqnHXXtP5XdINg==, figureFileBig=N/5dbzw7HqkgoAbNu/d2/A==, tableContent=null), ArticleFig(id=1325490508142871193, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Figure 14, caption=Comparative diagram of performance for XDMA and IDMA, figureFileSmall=6c/8Meqr5QE2DBlBpO67/g==, figureFileBig=d3rDjFJHPDzSdWSqPOok9Q==, tableContent=null), ArticleFig(id=1325490508478415514, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=图14, caption=XDMA与IDMA性能对比图, figureFileSmall=6c/8Meqr5QE2DBlBpO67/g==, figureFileBig=d3rDjFJHPDzSdWSqPOok9Q==, tableContent=null), ArticleFig(id=1325490508562301595, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Table 1, caption=Control Registers, figureFileSmall=null, figureFileBig=null, tableContent=
偏移寄存器描述
0x0000Descriptor L描述符表存储地址低32 b。
0x0004Descriptor H描述符表的存储地址高32 b。
0x0010LAST_PTR表示最后1个读描述符ID(0~127)。
0x0014FIRST_PTR表示第1个读描述符ID(0~127)。
0x0018CONTROL[31:2] 保留。
[1]Ctrl。当设置为1,本次DMA不上报中断;当设置为0,在最后1个描述符完成后发送中断。
[0]Done。当设置为1,每个完成描述符在状态表中Done 位置写1;当设置为0,仅在DMA_LAST_PTR对应的Done位置写1。
0x001CDMA_START写该寄存器任意值来触发DMA开始。
), ArticleFig(id=1325490508658770588, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=表1, caption=

控制寄存器

, figureFileSmall=null, figureFileBig=null, tableContent=
偏移寄存器描述
0x0000Descriptor L描述符表存储地址低32 b。
0x0004Descriptor H描述符表的存储地址高32 b。
0x0010LAST_PTR表示最后1个读描述符ID(0~127)。
0x0014FIRST_PTR表示第1个读描述符ID(0~127)。
0x0018CONTROL[31:2] 保留。
[1]Ctrl。当设置为1,本次DMA不上报中断;当设置为0,在最后1个描述符完成后发送中断。
[0]Done。当设置为1,每个完成描述符在状态表中Done 位置写1;当设置为0,仅在DMA_LAST_PTR对应的Done位置写1。
0x001CDMA_START写该寄存器任意值来触发DMA开始。
), ArticleFig(id=1325490509296304797, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Table 2, caption=Descriptor Format, figureFileSmall=null, figureFileBig=null, tableContent=
偏移寄存器描述
0x00LOW_SRC_ADDR源地址低32 b,PCIe系统内存地址。
0x04HIGH_SRC_ADDR源地址高32 b。
0x08LOW_DEST_ADDR目的地址低32 b,FPGA内部总线地址。
0x0CHIGH_DEST_ADDR目的地址高32 b。
0x10LENGTH[31] 1:指示该描述符含有下一个可执行链表信息;0:表示无。
[30:29]为下一个链表信息的控制信号。
[30] Ctrl。当设置为1,执行完链表不上报中断;当设置为0,在最后1个描述符完成后上报中断。
[29]Done。当设置为1,为每个完成的描述符在状态表中Done位置写1;当设置为0,仅在DMA_ LAST_PTR对应的Done位置写1。
[28] 保留。
[27:0] 传输大小,单位为B,不能为0。
0x14DESC_ID[31:16] magic值,该数值用于验证驱动描述符是否配置正确。
[15:8] 下一个链表的last_ptr。
[7] 保留。
[6:0] 指定描述符的ID,最大为127。
0x18Next Desc Addr L下一链表描述符存储低位地址
0x1CNext Desc Addr H下一链表描述符存储高位地址
), ArticleFig(id=1325490510990803614, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=表2, caption=

描述符格式

, figureFileSmall=null, figureFileBig=null, tableContent=
偏移寄存器描述
0x00LOW_SRC_ADDR源地址低32 b,PCIe系统内存地址。
0x04HIGH_SRC_ADDR源地址高32 b。
0x08LOW_DEST_ADDR目的地址低32 b,FPGA内部总线地址。
0x0CHIGH_DEST_ADDR目的地址高32 b。
0x10LENGTH[31] 1:指示该描述符含有下一个可执行链表信息;0:表示无。
[30:29]为下一个链表信息的控制信号。
[30] Ctrl。当设置为1,执行完链表不上报中断;当设置为0,在最后1个描述符完成后上报中断。
[29]Done。当设置为1,为每个完成的描述符在状态表中Done位置写1;当设置为0,仅在DMA_ LAST_PTR对应的Done位置写1。
[28] 保留。
[27:0] 传输大小,单位为B,不能为0。
0x14DESC_ID[31:16] magic值,该数值用于验证驱动描述符是否配置正确。
[15:8] 下一个链表的last_ptr。
[7] 保留。
[6:0] 指定描述符的ID,最大为127。
0x18Next Desc Addr L下一链表描述符存储低位地址
0x1CNext Desc Addr H下一链表描述符存储高位地址
), ArticleFig(id=1325490511095661216, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=EN, label=Table 3, caption=Performance Comparison for DMA Specification, figureFileSmall=null, figureFileBig=null, tableContent=
规格SGDMA(英特尔)XDMA(赛灵思)IDMA(本文)
测试平台英特尔Arria赛灵思
Virtex7
同创
Titan3
SR-IOV虚拟化不支持不支持支持
虚拟设备数量128
最大描述符数12864128
链式不支持支持支持
最大通道数1416
带宽利用率/%81.36086
驱动零拷贝不支持支持支持
非零拷贝支持不支持支持
描述符轮询不支持支持支持
LUT使用数量1.2万单通道2万
4通道3.6万
单通道1.6万
16通道1.7万
), ArticleFig(id=1325490511431205538, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, language=CN, label=表3, caption=

DMA规格性能对比

, figureFileSmall=null, figureFileBig=null, tableContent=
规格SGDMA(英特尔)XDMA(赛灵思)IDMA(本文)
测试平台英特尔Arria赛灵思
Virtex7
同创
Titan3
SR-IOV虚拟化不支持不支持支持
虚拟设备数量128
最大描述符数12864128
链式不支持支持支持
最大通道数1416
带宽利用率/%81.36086
驱动零拷贝不支持支持支持
非零拷贝支持不支持支持
描述符轮询不支持支持支持
LUT使用数量1.2万单通道2万
4通道3.6万
单通道1.6万
16通道1.7万
)], 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/issn1000-1239.202440833, detailUrlEn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/10.7544/issn1000-1239.202440833, pdfUrlCn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/PDF/10.7544/issn1000-1239.202440833, pdfUrlEn=https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/PDF/10.7544/issn1000-1239.202440833, aliStartDate=0, aliEndDate=0, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1767196800000, fullTextJson=null, articleText=null, reference=null)
收藏切换
基于PCIe的链式DMA控制器设计与实现
王洪良 1 , 郭振华 1 , 牟奇 1, 3 , 卢圣才 1 , 徐亚明 2 , 于泉泉 1
计算机研究与发展 | 体系结构 2026,63(1): 28-40
收起
收藏切换
计算机研究与发展 |体系结构 2026 , 63 (1) : 28 -40
基于PCIe的链式DMA控制器设计与实现
全屏
[Author(id=1325490485409743441, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=wanghl01@ieisystem.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1325490485493629523, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485409743441, language=EN, stringName=Hongliang Wang, firstName=Hongliang, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

"}, bioImg=null, bioContent=

born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490485556544084, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485409743441, language=CN, stringName=王洪良, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"UaLBYk6iYaONP4yreJ6gtg==","content":"

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

"}, bioImg=UaLBYk6iYaONP4yreJ6gtg==, bioContent=

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490485627847254, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490485812396632, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485627847254, language=EN, stringName=Zhenhua Guo, firstName=Zhenhua, middleName=null, lastName=Guo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

"}, bioImg=null, bioContent=

born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490485875311193, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485627847254, language=CN, stringName=郭振华, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"4kkMzth5/em2+OLgQnaUbA==","content":"

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

"}, bioImg=4kkMzth5/em2+OLgQnaUbA==, bioContent=

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490485934031451, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486009528926, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485934031451, language=EN, stringName=Qi Mu, firstName=Qi, middleName=null, lastName=Mu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101
3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641, bio={"content":"

born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

"}, bioImg=null, bioContent=

born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486076637791, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490485934031451, language=CN, stringName=牟奇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1浪潮电子信息产业股份有限公司 济南 250101
3华南理工大学计算机科学与工程学院 广州 510641, bio={"img":"tw7zkmtxWyqLl/bDgKr9tg==","content":"

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

"}, bioImg=tw7zkmtxWyqLl/bDgKr9tg==, bioContent=

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)]), AuthorCompany(id=1325490485090976333, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=3, ext=[AuthorCompanyExt(id=1325490485099364942, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641), AuthorCompanyExt(id=1325490485111947855, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485090976333, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3华南理工大学计算机科学与工程学院 广州 510641)])]), Author(id=1325490486189884001, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486256992867, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486189884001, language=EN, stringName=Shengcai Lu, firstName=Shengcai, middleName=null, lastName=Lu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

"}, bioImg=null, bioContent=

born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486336684644, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486189884001, language=CN, stringName=卢圣才, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"8xKkuKzUhI19IYTIhWyTKQ==","content":"

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

"}, bioImg=8xKkuKzUhI19IYTIhWyTKQ==, bioContent=

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])]), Author(id=1325490486395404902, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, 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=1325490486466708072, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486395404902, language=EN, stringName=Yaming Xu, firstName=Yaming, middleName=null, lastName=Xu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085, bio={"content":"

born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

"}, bioImg=null, bioContent=

born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486533816937, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486395404902, language=CN, stringName=徐亚明, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2浪潮 (北京)电子信息产业有限公司 北京 100085, bio={"img":"QvSSvtsECXCZsd7ROw3t+A==","content":"

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

"}, bioImg=QvSSvtsECXCZsd7ROw3t+A==, bioContent=

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490485002895946, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=2, ext=[AuthorCompanyExt(id=1325490485011284555, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085), AuthorCompanyExt(id=1325490485019673164, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490485002895946, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2浪潮 (北京)电子信息产业有限公司 北京 100085)])]), Author(id=1325490486596731499, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, orderNo=5, 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=1325490486684811885, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486596731499, language=EN, stringName=Quanquan Yu, firstName=Quanquan, middleName=null, lastName=Yu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101, bio={"content":"

born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

"}, bioImg=null, bioContent=

born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

, aboutCorrespAuthor=null), CN=AuthorExt(id=1325490486760309358, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, authorId=1325490486596731499, language=CN, stringName=于泉泉, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1浪潮电子信息产业股份有限公司 济南 250101, bio={"img":"FEbTgGsuFVzc2iz8IffB7w==","content":"

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

"}, bioImg=FEbTgGsuFVzc2iz8IffB7w==, bioContent=

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1325490484914815559, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, xref=1, ext=[AuthorCompanyExt(id=1325490484923204168, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1Inspur Electronic Information Industry Co., Ltd., Jinan 250101), AuthorCompanyExt(id=1325490484931592777, tenantId=1146029695717560320, journalId=1324759914146529478, articleId=1325490478069711388, companyId=1325490484914815559, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1浪潮电子信息产业股份有限公司 济南 250101)])])]
王洪良1 , 郭振华1, 牟奇1, 3, 卢圣才1, 徐亚明2, 于泉泉1
作者信息
  • 1浪潮电子信息产业股份有限公司 济南 250101
  • 2浪潮 (北京)电子信息产业有限公司 北京 100085
  • 3华南理工大学计算机科学与工程学院 广州 510641
作者简介:

1991年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

1988年生。博士,高级工程师。主要研究方向为计算机体系结构、分布式系统、多模态人工智能

1989年生。硕士,工程师。CCF会员。主要研究方向为计算机体系结构、高性能计算

1986年生。硕士,工程师。主要研究方向为计算机体系结构、数字芯片设计、高性能计算

1986年生。学士,工程师。主要研究方向为计算机体系结构、计算机网络、高性能计算

1987年生。硕士,工程师。主要研究方向为云计算服务器、存储服务器领域的系统设计和硬件开发

Design and Implementation of Chained DMA Controller Based on PCIe
Hongliang Wang1 , Zhenhua Guo1, Qi Mu1, 3, Shengcai Lu1, Yaming Xu2, Quanquan Yu1
Affiliations
  • 1Inspur Electronic Information Industry Co., Ltd., Jinan 250101
  • 2Inspur (Beijing) Electronic Information Industry Co., Ltd., Beijing 100085
  • 3School of Computer Science and Engineering, South China University of Technology, Guangzhou 510641
  • born in 1991. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

    born in 1988. PhD, senior engineer. His main research interests include computer architecture, distributed systems, and multimodal artificial intelligence

    born in 1989. Master, engineer. Member of CCF. His main research interests include computer architecture and high-performance computing

    born in 1986. Master, engineer. His main research interests include computer architecture, digital chip design, and high-performance computing

    born in 1986. Bachelor, engineer. His main research interests include computer architecture, computer network, and high-performance computing

    born in 1987. Master, engineer. His main research interest includes system design and hardware development in the fields of cloud computing servers and storage servers

出版时间: 2026-01-01 doi: 10.7544/issn1000-1239.202440833
文章导航
收藏切换

近年来,国产可编程门阵列(field programmable gate array,FPGA)厂商虽发展迅速,但在数据中心部署FPGA异构加速器时仍面临挑战。相较于赛灵思、英特尔等国际厂商,国产厂商普遍缺乏PCIe设备与主机间高速传输的解决方案,尤其在高性能直接内存访问(direct memory access,DMA)控制器设计领域存在明显短板。为解决该问题,设计并实现了基于PCIe的多通道链式DMA控制器。采用独立描述符控制器管理各通道,共享数据搬移器,降低对FPGA逻辑资源的消耗。采用链式结构实现描述符管理,减少中断对CPU的压力,满足主机与设备连续高速传输的需求。创新性地构建内部信息异步与预处理的架构,实现数据流水化处理,显著提升带宽利用率以及传输性能。经测试,在PCIe Gen3x8下,主机与国产FPGA加速器之间的DMA带宽高达6.91 GBps(利用率86%),支持多达16通道且实现通道负载均衡,该设计有效支撑了国产FPGA异构加速器在数据中心场景下的规模化部署。

PCIe  /  FPGA  /  异构加速  /  多通道  /  链式DMA

In recent years, although Chinese domestic field programmable gate array (FPGA) manufacturers have developed rapidly, they still face challenges when deploying FPGA heterogeneous accelerators in data centers. Compared with international manufacturers such as Xilinx (now AMD) and Intel, domestic manufacturers generally lack solutions for high-speed transmission between PCIe devices and hosts, especially in the field of high-performance direct memory access (DMA) controller design, where there are obvious shortcomings. To solve this problem, we design and implement a PCIe-based multi-channel chained DMA controller. By using an independent descriptor controller to manage each channel, sharing data movers, and reducing the consumption of FPGA logic resources, this design improves resource efficiency. The adoption of a chain structure for descriptor management reduces CPU interrupt pressure while meeting the requirements for continuous high-speed transmission between hosts and devices. An innovative architecture for asynchronous internal information pre-processing is developed, enabling data stream processing that significantly improves bandwidth utilization and transmission performance. Testing results show that under PCIe Gen3x8, the DMA bandwidth between the hosts and the domestic FPGA accelerator reaches 6.91 GBps (86% utilization rate), supporting up to 16 channels with channel balancing implementation. This design effectively enables large-scale deployment of domestic FPGA heterogeneous accelerators in data center scenarios.

PCIe  /  FPGA  /  heterogeneous acceleration  /  multi-channel  /  chained DMA
王洪良, 郭振华, 牟奇, 卢圣才, 徐亚明, 于泉泉. 基于PCIe的链式DMA控制器设计与实现. 计算机研究与发展, 2026 , 63 (1) : 28 -40 . DOI: 10.7544/issn1000-1239.202440833
Hongliang Wang, Zhenhua Guo, Qi Mu, Shengcai Lu, Yaming Xu, Quanquan Yu. Design and Implementation of Chained DMA Controller Based on PCIe[J]. Journal of Computer Research and Development, 2026 , 63 (1) : 28 -40 . DOI: 10.7544/issn1000-1239.202440833
得益于强大的并行计算能力和灵活性,可编程门阵列(field programmable gate array,FPGA)异构加速器受到微软、腾讯、字节跳动、阿里等互联网巨头的青睐,并被广泛应用于数据中心的大规模部署。例如,微软利用FPGA加速搜索算法,计算速度较CPU提升了40倍,系统整体性能提升了100%,同时服务器数量减少了50%。字节跳动将FPGA加速器用于视频压缩和图像处理任务,显著降低了数据中心的存储和计算压力。腾讯利用FPGA进行网络数据卸载,致力于探索智能网卡的整体解决方案与硬件新形态。阿里则推出了GPU与FPGA的异构计算解决方案,为零售系统提供加速支持。清华大学基于FPGA开发大模型推理框架,显著提升了性价比与能效比。此外,科大讯飞、联捷科技、华大基金、百度等公司使用FPGA加速语音识别、图像处理、科学计算、自动驾驶、人工智能等业务。因此,FPGA异构加速在学术界和工业界均展现出广阔的应用前景。
在科技贸易战的背景下,部署自主可控FPGA异构加速器的需求愈发迫切。数据中心部署的FPGA加速器普遍采用高性能、大容量芯片,而当前英特尔和赛灵思长期垄断市场,国产FPGA芯片在性能、容量及生态建设等方面存在显著差距,使得国产FPGA异构加速器的实际部署面临严峻挑战。本文调研了紫光同创、安路科技、高云、易灵思、智多晶、京微齐力6家国内民用厂商及其产品,发现目前国内FPGA产品的最高工艺水平为12 nm,查找表(look up table,LUT)最大规模为60万,最高Serdes速率为32 Gbps,最高MAC规格为100 Gbps,PCIe规格最高支持Gen4x8。相较之下,国外产品已采用7 nm先进工艺,LUT最大规模达1800万,Serdes速率达112 Gbps,MAC规格提升至400 Gbps,PCIe规格支持Gen5x16。参数对比凸显国内产品全维度落后态势。自2016年起,国内企业开始基于英特尔Arria10/Stratix10、赛灵思VU37P等国外平台开发加速器产品,而英特尔与赛灵思也相继推出PAC卡、U280等自有解决方案。科技贸易战爆发后,国内企业开始尝试国产化替代,例如采用同创PG2L100H芯片研发FPGA加速器原型。FPGA异构加速应用的核心难点在于通用高规格高性能直接内存访问(direct memory access,DMA)控制器,该技术是主机与FPGA间数据传输的关键,如赛灵思XDMA、英特尔SGDMA,而国产FPGA厂商尚未攻克相关技术,仅能提供PCIe硬核以及简易的演示程序,无法满足数据中心的应用要求,导致国产加速器难以部署。
当前PCIe DMA控制器设计领域存在2个参考设计方案:开源架构RIFFA[1]和赛灵思早期参考设计XDMA [2]。然而,两者在规格性能指标上存在显著缺陷,主要表现为:带宽利用率低、缺乏链式传输支持、缺乏多通道支持或者调度不均衡、目的地址寻址机制受限等问题,无法满足复杂的应用场景。由于PCIe协议复杂且可借鉴的成熟设计稀缺,开发高性能高规格DMA控制器需要重点解决以下挑战:
1)链式传输机制。链式DMA机制相较于传统的块DMA以及分散聚合DMA更加复杂,在实现时整体架构设计难度也更高。在实际运行过程中,链式DMA可以连续不断搬运数据,当描述符执行完成时,如何优化中断上报机制成为关键问题。需要设计一套新的消息机制,既能防止因大量中断上报而导致的系统性能降低,又能及时有效地通知主机数据传输的完成情况,确保主机与 DMA 控制器之间的高效协作。
2)多通道资源优化与调度均衡。多通道设计面临双重技术问题。其一,硬件资源消耗与通道数量呈线性增长关系,以赛灵思XDMA为例,单通道消耗约2万 LUT,4通道时消耗约3.6万LUT,显著制约了通道扩展性。其二,多通道间需要设计调度机制,解决各个通道描述符执行不均衡现象,该问题影响系统多任务的公平性。
3)流水化架构设计。在DMA设计中,涉及大量的数据处理操作,如去帧头、组帧、数据预取、数据移位等,处理操作涉及控制信息计算、状态转移等,若采用传统的控制信息跟随数据的方式处理,将导致延迟增加、性能降低,因此,设计一套高效的流水化数据处理架构是提升性能的关键。
4)乱序接收重组机制。根据PCIe协议规范,非转发型事务需要返回完成报文,如果多个非转发型事务并发,返回的完成报文会产生乱序。因此,需要设计一套机制完成返回报文的缓存管理以及重新定序。虽然主机支持请求事务乱序可以提高处理器的性能,但这无疑为用户设计 DMA 控制器增加了难度。
为了解决上述问题,设计并实现了基于PCIe的高规格、高性能DMA控制器,并在国产FPGA紫光同创平台进行测试验证,同时兼容英特尔、赛灵思、同创、安路等国内外一系列平台。实验结果表明,基于该DMA控制器的FPGA异构加速器设备与主机通信性能优越、功能完善,能够满足多种复杂应用场景的需求,填补了国内在该领域的空白。本文主要有4个贡献。
1)详细设计并定义链式DMA控制寄存器及描述符的格式,其中涵盖了控制寄存器与描述符的偏移量、寄存器名称及其具体功能说明。为了便于理解,本文以主机内存数据向FPGA内部地址空间的传输为例,详细阐述了DMA配置的具体流程,并提供了完整的操作步骤说明。
2)针对多通道系统的特点,对其实现架构进行了深入设计优化,在保证通道独立性的基本前提下,显著降低逻辑资源使用率。具体优化策略主要包括2个方面:首先,采用FPGA片上块RAM替代传统的寄存器组方案,有效解决了多通道场景下大量寄存器组导致的资源浪费问题。其次,基于状态机设计了各通道描述符轮询机制,结合共享式数据搬移器架构实现了通道间传输资源的时分复用。
3)针对PCIe协议中请求与响应独立处理的局限性,创新性地引入了一套控制信息管理机制。通过在系统架构中引入专门的控制信息处理模块,实现了对各类处理的异步调度。该机制能够提前解析并分发控制指令至各功能模块,在数据到达之前完成必要的预处理准备工作。这种改进显著优化了数据处理流程,使得各模块能够无缝衔接地开展工作,最终实现了高效的流水线式数据处理。
4)设计了一套基于PCIe的Tag管理与返回包乱序重组机制。该机制主要包括Tag的分发与回收、返回包数据缓存以及排序等功能模块。具体实现上,采用基于报文并发的Tag管理机制,为每个请求分配一个唯一的Tag标识,并预先为其分配对应的缓存空间。在缓存后端,实现返回包数据的乱序重组功能,当相关数据包接收完成后,完成Tag的释放和回收操作。
通过上述机制的设计与实现,成功提高了系统的带宽利用率,降低了系统延迟以及资源使用率。
PCI Express总线是一种基于点对点串行连接机制的高速互连技术,采用差分信号传输方式,显著提升了抗干扰性能和数据传输的完整性和可靠性。根据实际设计需求,PCIe物理链路可灵活配置为x1~x32等多种链路宽度模式,以适应不同的带宽需求。此外,其传输速率可根据应用场景选择不同的代数标准(如Gen1~Gen5),从而实现更高的数据传输效率。
PCIe层次结构如图1所示,主要包括事务传输层、数据链路层和物理层。数据报文在设备核心中生成后,依次通过事务传输层、数据链路层和物理层进行封装和处理,最终通过串行差分信号线发送至目标设备。接收端的数据则依次经过物理层、数据链路层和事务传输层的解封装和处理,最终将解析后的数据送入设备核心中,从而完成整个数据的高效传递[3]。
事务层包(transaction layer packet,TLP)是用户层与事务层之间交互的主要机制,其数据头由3或4个双字(double word,DW)组成,如图2所示。其中,保留字段用R标识,其余字段的定义由具体的事务类型决定。根据PCIe协议的规定,Fmt字段用于指示包的格式,Type字段用于指示包的类型,通过Fmt与Type的不同组合可以定义多种类型的数据包[4]。本文设计的DMA控制器中,涉及的主要TLP类型包括存储器读(memory read,Mrd)、存储器写(memory write,Mwr)以及完成数据(completion with data, CplD)。具体而言,当PCIe主设备需要访问目标设备的存储器空间时,会使用Non-Posted总线事务向目标设备发送Mrd请求,目标设备接收到该请求后会通过CplD包将存储器中的数据返回给主设备。在PCIe总线中,存储器写操作采用Posted总线事务,主设备仅需发送Mwr包即可完成存储器写入操作,无需等待目标设备的响应报文。值得注意的是,Non-Posted事务要求目标设备必须返回响应报文以确认数据已成功接收,而Posted事务则不需要目标设备的显式响应 [5]。
AXI协议是ARM公司提出的AMBA(advanced microcontroller bus architecture)协议的核心组成部分,代表了一种高性能片内总线架构。该协议以其高带宽、低延迟和高效能著称,已成为现代嵌入式系统设计中的重要技术。AXI协议定义了3种接口类型:AXI-MM接口专为高性能存储器映射操作设计,适用于需要地址指定的高速数据传输场景,如内存访问等。AXI-Lite作为AXI-MM的精简版本,主要用于访问低速外设中的寄存器资源。AXI-Stream接口则专注于高效的流数据传输,特别适用于主从设备之间的连续数据读写操作,其特点在于无需地址标识,从而实现了更高的传输效率。
DMA是一种高效的数据传输机制,它能够在外部设备与内存以及内存与内存之间实现快速的数据传输操作。与传统的CPU控制数据传输方式不同,DMA传输过程完全独立于CPU运行,仅需CPU在传输开始和结束时进行少量的初始化和收尾处理。这种设计使得CPU在整个数据传输过程中能够专注于执行其他任务,从而实现了任务处理与数据交换的并行执行。当系统配置了DMA控制器后,CPU只需向DMA控制器发送一条指令,并设置源地址、目的地址及数据长度等参数,DMA控制器便会自主接管后续的数据传输任务。在数据传输过程中,DMA控制器会高效地完成数据从源地址到目的地址的迁移工作。一旦数据传输完成,DMA控制器会通过中断机制通知CPU,以便系统能够及时进行后续的数据处理工作。这种机制显著降低了CPU的负载率,使CPU能够将更多资源投入到核心计算任务中。因此,DMA技术在现代计算机系统中得到了广泛应用,并成为提升系统性能的重要手段之一。
DMA根据工作模式可以分为块DMA、分散聚合DMA(scatter gather DMA,SGDMA)以及链式DMA。块DMA在进行数据传输时,要求源和目标物理地址空间必须是连续的。然而,在实际应用中,系统分配的存储空间往往并非物理连续,因此块DMA方式需要将传输任务分割为多个独立操作。当数据分布于非连续物理地址空间时,块DMA模式需反复进行配置并频繁触发中断机制,这不仅降低了CPU的工作效率,也影响了数据传输的整体效率。SGDMA通过构建一个链表来描述非连续物理存储空间。CPU首先配置链表,然后将链表头地址写入DMA控制器寄存器。DMA控制器会逐一解析并执行链表中的每个描述符,在完成每一块物理连续数据的传输后无需立即触发中断,而是根据链表继续传输下一块物理连续的数据。仅在所有数据传输完成后才统一触发中断。显然,SGDMA模式相较于块DMA模式具有更高的效率。链式DMA作为SGDMA的扩展实现,在链表的基础上增加了指针,该指针位于链表末尾。当当前链表执行完毕后,DMA控制器可以根据末尾指针指向下一个链表继续执行。这种机制虽然在控制器实现上更为复杂,但提升了数据传输的灵活性和效率。
在DMA控制器开发领域,由于设计复杂度高、研发周期长,国际厂商赛灵思在2017年前仅提供简易的XAPP1052参考设计方案。该方案采用传统中断机制,可对事务包进行基础处理,配套提供FPGA源码及Linux驱动程序。其开源性使得科研人员能够基于该参考设计开展功能扩展与优化研究,成为学术界主要研究对象。2017年后,赛灵思推出升级版XDMA控制器,通过引入MSI/MSI-X中断机制、多通道架构及链式DMA传输技术显著完善功能规格。新方案提供图形化配置界面,支持参数快速设置与系统集成,大幅降低开发门槛。然而,XDMA以加密源代码形式交付,用户无法像早期XAPP1052方案那样进行定制开发。同时,由于XDMA功能完备性已满足多数应用场景需求,用户端的定制化开发需求锐减,直接导致学术界相关研究方向热度下降,后续相关研究主要基于赛灵思XDMA及英特尔SGDMA等厂商提供的闭源解决方案开发应用。
在架构优化层面,研究者多基于赛灵思参考设计进行改良,如Kavianipour等人[6]通过简化用户接口实现750~790 MBps带宽,Rota等人[7]通过FPGA内部存储描述符方式将块DMA升级为SGDMA,在Gen2x8下将带宽提升至3461 MBps,Zazo等人[8]则在官方参考设计的基础上扩展了对SR-IOV虚拟化的支持。在专用领域优化方面,文献[9]针对10 Gbps以太网控制器开发了支持平衡任务优先级的SGDMA,Forencich等人[10]在刚玉平台实现了100 Gbps网络定制DMA,支持传输、接收、完成和事件队列,对于网络专用DMA设计具有参考意义,而文献[11]中则采用英特尔官方多DMA控制器并行的方式,突破了400 Gbps网络传输瓶颈。在异构计算领域,Kasai等人[12]利用XDMA验证了FPGA-GPU直传可行性,但传输效率仍未达实用标准。此外,文献[13]中尝试用高层次综合语言开发DMA控制器,虽然解决了传统的开发效率问题,但是实际性能不足。
当前主流方案多依赖赛灵思XDMA或英特尔SGDMA直接开发应用[14-17],自主架构设计研究明显不足。尤其在国内,同创、安路等厂商尚未形成成熟商用的DMA控制器产品,这一现状导致2个关键问题:其一,现有改良方案存在平台依赖性强、功能扩展性差等固有缺陷[18-22];其二,国际技术垄断加剧了我国在该领域的技术风险。因此,研发具备全功能规格、性能高、跨平台兼容的自主化DMA控制器,已成为突破异构计算通信瓶颈和保障产业链安全的核心攻关方向之一。
DMA控制器的实现基础是控制寄存器与描述符设计,本文结合PCIe与通用DMA技术,自主设计了一套多通道链式DMA控制寄存器与描述符方案。控制寄存器分为读描述符控制寄存器和写描述符控制寄存器,分别对应H2C(host to card)和C2H(card to host)方向。读写描述符存储在PCIe系统内存中,每个描述符表最多可存储128个描述符,每个描述符由8个双字组成。描述符存储表采用状态表加描述符表的结构形式,其中状态表位于描述符表之前。状态表的起始地址为H2C 描述符表基地址或C2H 描述符表地址,描述符表的起始地址为基地址加固定偏移地址0x200。DMA开始寄存器用于触发DMA控制器来获取描述符表。因此,在配置DMA传输时,设置开始寄存器必须作为最后一步操作。
表1描述了DMA控制寄存器的配置。DMA描述符控制寄存器位于PCIe的基地址寄存器(base address register,BAR)空间上。H2C控制寄存器的偏移地址范围为0x0000~0x01FF,共支持16个通道,每个通道占用0x20的地址空间。C2H控制寄存器的偏移地址范围为0x0200~0x03FF,同样支持16个通道,每个通道占用0x20的地址空间。
表2为DMA描述符格式,共8个描述符,每个描述符占用32 b。H2C和C2H各有一套描述符格式。每个描述符不仅记录了当前DMA传输所需的地址、数据量等信息,还在链表末尾包含下一个链表的起始位置及相关信息,从而实现链式DMA传输,减少了CPU对中断的频繁处理。
本节示例展示了3个数据块从主机系统内存移动到FPGA总线地址空间的过程。图3详细说明了PCIe和FPGA总线地址空间中数据块的位置和大小分布,而图4为具体的描述符格式示意图。
主机配置执行的步骤如下:
步骤1.计算分配所需的内存。
① 状态表中的每个条目为4 B,128个条目需要512 B的内存。
②每个描述符是32 B。3个描述符需要96 B的内存,状态和描述符表的总内存分配为608 B。
步骤2.在PCIe地址空间中分配608 B的内存。此示例中分配内存的起始地址是0xf000_0000,将此地址写入H2C描述符控制寄存器中。
步骤3.基地址为0xF000_0000,那么第1个描述符的地址为0xF000_0000+0x200。
①在0xF000_0204中写入高32 b源地址0x0。
②在0xF000_0200中写入低32 b源地址0x1000_0000。
③在0xF000_0200c中写入高32 b目的地址0x0。
④在0xF000_0208中写入低32 b目的地址0x5000_0000。
⑤在0xF000_0210中写入0x0001_0000,长度64 KB。
⑥在0xF000_0214中写入描述符ID 0。
⑦在0xF000_0218中写入magic值ad4b。
步骤4.重复步骤1~3,写入另外2个描述符,写完后如图5所示。
步骤5.通过BAR0来配置H2C控制寄存器基地址,本例使用64 b地址。
①在0x0000_0004中写入0x0000_0000,高32 b读描述符基地址。
②在0x0000_0000中写入0xF000_0000,低32 b读描述符基地址。
③描述符地址在此基础上加0x200(无需驱动操作)。
步骤6.写寄存器H2C_LAST_PTR, 在0x0000_0010写入0x2。
步骤7.写寄存器H2C_FIRST_PTR, 在0x0000_0014写入0x0。
步骤8.写寄存器H2C_CONTROL, 在0x0000_0018写入0x0。
步骤9.写寄存器H2C_DMA_START,在0x0000_001c写任意值,启动DMA读。
PCIe作为高速串行总线协议,其分层协议为外设互联提供了高带宽、低延迟的通信基础,而直接内存访问技术作为提升系统数据吞吐效率的核心机制,其实现方式并未在PCIe协议中做强制性规定,需结合具体应用场景进行设计与实现。本文根据自定义控制寄存器、描述符以及运行流程,实现了多通道链式DMA控制器,其中DMA控制器逻辑实现框架如图6所示。
PCIe硬核模块实现物理层、数据链路层及事务传输层协议处理,DMA通过主机请求、设备请求、设备返回、主机返回及中断请求等接口与硬核进行交互,接口遵循AXI-Stream协议规范。DMA控制器核心组成包含描述符管理、数据搬移器、整帧缓存、状态上报、接口转换、中断管理、接口标准化等模块,各模块间通过流模式进行数据传输以保障传输效率。基地址寄存器管理模块负责将主机对BAR空间的访问转换为APB总线操作,其中BAR0用于DMA功能配置,BAR1~BAR5负责用户业务参数配置。描述符管理模块通过APB总线与主机交互,实现描述符存储与调度管理功能。数据搬移器接收并解析描述符指令,除常规数据搬运外,主机端的描述符数据同样需通过数据搬移器传输至设备端。整帧缓存模块对Mwr TLP包实施至少单包缓存后再进行完整帧输出,该机制旨在避免流模式传输时帧内出现多拍无效间隔导致的PCIe链路错误。状态上报模块将每个描述符完成状态信息封装为Mwr TLP包回传主机,主机为每个描述符预先分配状态标志存储空间,当描述符处理完成后即向对应主机地址写入完成标志。由于数据搬移器的输入/输出数据流未处理突发传输及地址边界对齐等问题,接口标准化模块负责将流数据转换为符合AXI-MM标准的协议格式。
针对多通道的特点,本文在实现上做了如下优化,如图7所示。为节省逻辑资源,本文将一块片上RAM分为16份,分别存储16个通道的控制寄存器,同时,将DMA启动标识缓存到FIFO中,通过状态机将配置的控制寄存器逐个读出,生成H2C描述符,该描述符为特殊描述符,利用数据搬移器将主机中的描述符表中的数据搬移到设备中,这里与普通数据共用搬移器以节省资源。描述符数据在解析后,分别存储到深度为16×128(通道×单次最大描述符个数)RAM中,按照轮询调度算法调度每个通道。当每个通道最后一个描述符被执行后,需要检查是否有链式标志位,如果有,需要生成新的描述符,将下一个链表内的描述符搬运到设备中继续执行。
本文创新性地提出了一种基于流水化架构的数据搬移器设计方案,通过将预处理后的描述符信息进行异步分发,使各个数据关键处理单元能够并行执行。该架构采用多级流水线缓存结构,构建了描述符解析与数据处理相分离的协同工作机制,相较于传统串行处理模式中控制信息与数据流同步传递的方式,实现了处理单元间的零等待数据传递,这一架构突破性地解决了数据搬移过程中控制流与数据流耦合度过高的瓶颈问题,为高速数据传输控制系统提供了新的设计范式。
如图8所示,在H2C数据搬移器架构中,描述符预处理模块首先解析原始描述符,动态计算目标地址、数据长度等关键参数,并通过异步分发机制将控制信息广播至流水线各处理单元,同时基于PCIe协议(包括最大读请求大小限制与4 KB地址对齐规则),执行跨界拆分逻辑,将连续内存访问请求拆分为多个符合规范的Mrd TLP。当CplD报文到达接收端后,处理流程依次包含:1)帧头剥离,提取有效负载数据。2)依据Tag标识符将分片数据路由至独立缓存分区,实现乱序到达分片的动态重组。3)根据描述符总长度将多个Tag携带数据组成一包数据以完成数据拼接,同时产生字节掩码。4)针对目的地址非0的情况,通过移位对齐,确保数据与地址对应。5)采用多级流水式位宽转换器,将重组数据按目标存储器的位宽要求(如64 b/128 b/256 b)进行位宽转化。整个数据通路通过握手协议实现背压控制与吞吐量优化,消除流水线间气泡,实现全流水处理。
如图9所示,在C2H数据搬移器架构中,描述符首先通过预处理模块进行参数解析,系统会同步计算出存储器读取操作的源地址/长度以及流水线处理模块所需的源地址/长度等配置,这些控制信息通过异步分发机制分别传输至存储器读控制管理单元和流水线处理模块。跨界拆分模块依据PCIe协议的最大载荷及4 KB地址边界约束条件执行跨界地址拆分算法。从存储器获取的原始数据流需经过多层处理:1)通过读地址与长度将描述符规定的所有数据读出。2)对读出的原始数据增加掩码标志,然后移位,将掩码标志为0的空数据移除。3)根据系统总线位宽(如512 b)与存储总线位宽(如256 b)的差异进行自适应位宽转换。4)经处理后的数据流暂存于高速FIFO缓冲队列,同时设计预缓存机制,缓存2拍数据,优化组帧模块取数据效率。5)根据拆分后的目的地址/长度等将缓存的数据按Mwr TLP协议规范进行组装。
在PCIe DMA读事务中,设备端依据最大读请求大小与4 KB地址对齐规则对DMA描述符进行动态分片,生成符合规范的Mrd TLP。主机端响应时,根据读完成边界(read completion boundary,RCB)参数对返回数据进行多次分片,形成多个CplD报文。由于PCIe协议允许不同的CplD乱序传输,Tag作为CplD报文中的唯一标识符,成为解决乱序问题的关键要素。
如图10所示,本文采用基于累加计数器的Tag顺序分配策略,通过参数化配置Tag位宽动态调整Tag池容量以适配不同场景的并发请求需求。每个Tag关联独立分配的块RAM存储区,其深度由下式计算:深度=Tag数×lb(最大读请求×8/数据位宽), 确保不同Tag的数据分片物理隔离。每个存储区配置专属状态标志,实时跟踪分片到达状态。当某Tag所有分片到达后标志被置位,触发该Tag数据读取,按分片序号升序重构原始数据流,读取完毕后,Tag管理模块计数器加1,完成Tag回收。
本文设计并实现了一种基于PCIe的DMA控制器,并在自主研发的硬件板卡FP600加速器上完成了功能验证。如图11所示为硬件板卡的实物图。实验采用同创Titan3系列FPGA芯片PG3T600P,该芯片理论上支持PCIe Gen4x8接口。由于当前FPGA厂商尚未提供适用于PCIe Gen4x8的硬核,本研究选用Gen3x8硬核进行测试及对比,理论最大传输速率为8 GBps。除国产平台,本文DMA控制器也在赛灵思平台进行了测试,其测试性能与国产平台几乎一致。此外,本文还分别在英特尔Arria 10平台和赛灵思Virtex7平台上测试了英特尔SGDMA和赛灵思XDMA的性能表现,最终给出了3种DMA控制器的规格参数及性能指标对比表。
本文设计的DMA控制器最大支持16个通道传输。为了便于说明和验证性能,选择其中4个通道进行测试,多通道DMA控制器测试结果如图12所示。图中展示了CH0~CH3四个通道的DMA读写带宽实验结果,所有数据均在4个通道同时工作的情况下获取。
从图12可以看出,随着数据包长度的增加,带宽呈现显著增长趋势。这一现象表明,在数据传输过程中采用一次性执行128个描述符与链式机制的组合策略能够有效提升系统性能。当4个通道同时工作时,每个通道的带宽性能基本一致,能够有效共享PCIe总线带宽。这一现象与设计中各通道基于描述符轮询使用数据搬移器的机制完全吻合。具体而言,搬移器通过一种高效的轮询机制来管理多通道的数据传输任务。其核心工作原理是:搬移器会按照预设的顺序循环遍历每一个通道的描述符队列,逐一检查并处理其中待完成的数据搬移任务。这种机制确保了每个通道都能获得均等的处理机会,从而实现了多通道之间的公平带宽分配。
主机端的数据拷贝模式会影响DMA传输的执行效率,本文分别测试了零拷贝与非零拷贝模式下的DMA读写带宽性能,实验结果如图13所示。实验结果表明,得益于控制信息预处理异构分发机制和数据处理流水化架构设计,本文设计的DMA在性能上体现出显著优势。在非零拷贝下,H2C方向的最大带宽达到6.91 GBps,约为理论带宽的86%;C2H方向的最大带宽为6.31 GBps,约为理论带宽的79%。在零拷贝模式下,H2C方向的最大带宽为5.67 GBps,C2H方向的最大带宽为5.37 GBps。测试过程中,自研加速器板卡与主机协商确定的最大有效负载为256 B,最大读请求为512 B。因此,H2C方向的最大带宽高于C2H方向的结果符合预期。从实验曲线可以看出,在初始阶段,H2C方向的带宽低于C2H方向的。这是因为C2H方向仅涉及设备向主机发送数据的过程,而H2C方向需要设备发送读请求并处理主机返回的数据包,其流程更为复杂。因此,在初始阶段,C2H方向的效率更高。
此外,图13还显示,在1 MB数据量以下的场景中,零拷贝模式展现出更高的性能优势;而当数据量超过1 MB时,非零拷贝模式则更具效率优势。从技术原理角度来看,零拷贝机制以部分灵活性和兼容性为代价,换取了显著的性能提升,特别适用于对吞吐量和延迟敏感的应用场景,以及需要实时处理的小文件传输需求。相比之下,非零拷贝机制则在通用性和开发便捷性方面表现更为突出,能够更好地满足高性能、大规模数据传输的应用需求。基于上述分析,建议根据具体的业务场景和性能要求,选择最合适的拷贝模式以实现最优效果。
英特尔的SGDMA控制器相对简单,缺乏链式传输和多通道支持等高级功能,本文未将其作为重点对比对象。在零拷贝模式下,本文设计的IDMA与赛灵思XDMA的性能对比数据如图14所示。可以看出,在数据量为1 KB至64 MB的范围内,IDMA效率明显优于XDMA,最大带宽高出约16%。
此外,本文对DMA控制器的规格和性能与赛灵思、英特尔的闭源产品进行了全面对比,具体结果见表3。通过对比可以看出,IDMA在规格和性能方面均达到了或超越了国外厂商的水平。特别是在多通道设计上,本文的设计明显优于赛灵思的方案,且资源占用更为节省。赛灵思的XDMA在单通道下占用2万LUT资源,4通道下占用3.6万LUT资源。IDMA在单通道下仅占用1.6万LUT资源,16通道下也仅占用1.7万LUT资源,资源增长幅度仅为0.1万。相比之下,XDMA最多仅支持4通道设计,并且通过大量占用逻辑资源来实现通道扩展。IDMA通过优化多通道实现架构,成功避免了这一问题。
本文在对PCIe协议和AXI协议总线进行深入理论研究的基础上,设计并实现了一种基于FPGA的多通道链式DMA控制器,并在自主研发的加速器FP600上成功验证。该设计有效解决了数据中心服务器与国产FPGA设备之间的高效数据传输及业务加速需求。在本设计中,对寄存器配置、描述符管理、系统架构等关键部分进行了详细介绍,并提供了具体的设计思路和使用流程步骤。此外,详细阐述了多通道的具体实现架构、通道均衡机制以及资源优化方案。通过设计Tag管理与返回包缓存机制,实现了数据包的乱序重组。同时,采用信息预处理和异步分发机制,实现了数据的流水化处理,显著提高了带宽利用率,使实际带宽基本接近理论极限。本文设计的DMA控制器支持最大16个通道链式传输,实验与分析表明,在H2C方向的最大带宽为6.91 GBps,约为理论带宽的86%;在C2H方向的最大带宽为6.31 GBps,约为理论带宽的79%。在系统性能与规格方面,本文设计的DMA控制器与英特尔的SGDMA和赛灵思的XDMA相比均具有竞争力,自主研发的DMA控制器不仅适用于数据中心FPGA加速器,还可广泛应用于云计算平台、智能网卡、人工智能等多种场景。
作者贡献声明:王洪良、郭振华提出研究思路,设计系统整体架构并撰写论文;王洪良、牟奇、卢圣才负责工程设计开发、实验以及架构优化;徐亚明、于泉泉负责部分论文内容撰写并提出修改建议。

参考文献 引证文献
排序方式:
1
Kastner. RIFFA: Reusable integration framework for FPGA accelerators [CP/OL]. San Francisco, CA: GitHub, (2015-05-04) [2025-05-08]. https://github.com/KastnerRG/riffa
2
scryver. XAPP1052 [CP/OL]. San Francisco, CA: GitHub, (2019-01-08) [2025-05-08]. https://github.com/scryver/XAPP1052
3
王齐. PCI Express体系结构导读[M]. 北京: 机械工业出版社, 2016
Wang Qi. Introduction to PCI Express Architecture [M]. Beijing: Machinery Industry Press, 2016 (in Chinese)
4
PCI-SIG. PCI Express Base Specification Revision 3.0[S/OL]. Beaverton, OR: PCI-SIG, 2010[2025-05-08]. https://www.pcisig.com
5
Budruk R, Anderson D, Shanley T. PCI Express System Architecture[M]. Boston, MA: Addison-Wesley Professional, 2003
6
Kavianipour H, Muschter S, Bohm C. High performance FPGA-based DMA interface for PCIe[J]. IEEE Transactions on Nuclear Science, 2014, 61(2): 745−749
7
Rota L, Caselle M, Chilingaryan S, et al. A PCIe DMA architecture for multi-Gigabyte per second data transmission[J]. IEEE Transactions on Nuclear Science, 2015, 62(3): 972−976
8
Zazo J F, Lopez-Buedo S, Audzevich Y, et al. A PCIe DMA engine to support the virtualization of 40 Gbps FPGA-accelerated network appliances[C]//Proc of the 2015 Int Conf on ReConFigurable Computing and FPGAs (ReConFig). Piscataway, NJ: IEEE, 2015: 1−6
9
Wu Deze, Duan Guodong, Ren Minhua, et al. Research and design of descriptor-based DMA for 10 Gigabit Ethernet controller[C]//Proc of the 6th Int Conf on Electronic Engineering and Informatics (EEI). Piscataway, NJ: IEEE, 2024: 855−859
10
Forencich A, Snoeren A C, Porter G, et al. Corundum: An open-source 100-Gbps Nic[C]//Proc of the 28th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2020: 38−46
11
Kubálek J, Cabal J, Špinler M, et al. DMA Medusa: A vendor-independent FPGA-based architecture for 400 Gbps DMA transfers[C]//Proc of the 29th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2021: 258−258
12
Kasai S, Osana Y. A driver-based approach for DMA transfer between FPGA-GPU[C]//Proc of the 10th Int Symp on Computing and Networking Workshops (CANDARW). Piscataway, NJ: IEEE, 2022: 103−108
13
Zabołotny W M. Versatile DMA engine for high-energy physics data acquisition implemented with high-level synthesis[J]. Electronics, 2023, 12(4): 883
14
Yokono T, Yamabe Y, Tanaka K, et al. FPGA-based accelerators system with low latency autonomous DMA engine[C]//Proc of the 30th Annual Int Symp on Field-Programmable Custom Computing Machines (FCCM). Piscataway, NJ: IEEE, 2022: 1−1
15
Zhang Honglin, Yang Haibo, Liu Chengcheng, et al. Design of a common high-performance data acquisition system for physics experiments at HIAF[J]. IEEE Transactions on Nuclear Science, 2024, 72: 713−719
16
Kobayashi R, Fujita N, Yamaguchi Y, et al. Multi-hybrid accelerated simulation by GPU and FPGA on radiative transfer simulation in astrophysics[J]. Journal of Information Processing, 2020, 28: 1073−1089
17
Zhao Yingxiao, Liu Xin, Yang Jiong. A resource and timing optimized PCIe DMA architecture using FPGA internal data buffer[J]. IEICE Electronics Express, 2019, 16(1): 20180858
18
Cheng Kun, Liao Shengkai, Shen Qi, et al. Design and implementation of high-throughput PCIe with DMA architecture between FPGA and powerPC[J]. arXiv preprint, arXiv: 1809.07702, 2018
19
Zabołotny W M, Gumiński M, Kruszewski M, et al. Control and diagnostics system generator for complex FPGA-based measurement systems[J]. Sensors, 2021, 21(21): 7378
20
Shanehsazzadeh F, Sadri M S. Area and performance evaluation of central DMA controller in Xilinx embedded FPGA designs[C]//Proc of the 2017 Iranian Conf on Electrical Engineering (ICEE). Piscataway, NJ: IEEE, 2017: 546−550
21
王彦伟, 李仁刚, 徐冉, 等. 基于可重构架构的数据中心异构加速软硬件系统级平台[J]. 计算机研究与发展, 2025, 62(4): 963−977
Wang Yanwei, Li Rengang, Xu Ran, et al. Data center heterogeneous acceleration software-hardware system-level platform based on reconfigurable architecture[J]. Journal of Computer Research and Development, 2025, 62(4): 963−977 (in Chinese)
22
Chandra K, Jagtap A P, Ranjan N, et al. Design of PCIe-DMA bridge interface for high speed ethernet applications[C]//Proc of the 2019 2nd Int Conf on Advanced Computational and Communication Paradigms (ICACCP). Piscataway, NJ: IEEE, 2019: 1−5
2026年第63卷第1期
188
78
0
引用本文
BibTeX
文章信息
doi: 10.7544/issn1000-1239.202440833
  • 接收时间:2024-10-31
  • 首发时间:2026-11-06
  • 出版时间:2026-01-01
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-10-31
  • 修回日期:2025-06-16
  • 录用日期:2025-07-17
基金
作者信息
    1浪潮电子信息产业股份有限公司 济南 250101
    2浪潮 (北京)电子信息产业有限公司 北京 100085
    3华南理工大学计算机科学与工程学院 广州 510641
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/jsjyjyfz/CN/10.7544/issn1000-1239.202440833
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