技术解析｜NVIDIA Cedar 模块设计：1.6Tbps 的背后

Ofweek光电信息网2025-07-24

芝能智芯出品在面向人工智能集群与高性能计算平台的硬件系统中，网络模块的性能正在成为瓶颈。NVIDIA 针对其 DGX H100/H200 系统开发了名为 Cedar 的定制化网络模块，具备 1.6Tbps 的总网络吞吐能力，远高于常规 PCIe 网卡解决方案。这个模块的硬件结构、互连方式、接口布局以及其在散热与系统集成方面的优势，值得我们参考。Part 1Cedar 模块的系统架构与设计逻辑...

网页链接

免责声明：本文观点仅代表作者个人观点，不构成本平台的投资建议，本平台不对文章信息准确性、完整性和及时性做出任何保证，亦不对因使用或信赖文章信息引发的任何损失承担责任。

精彩评论

我们需要你的真知灼见来填补这片空白

发表看法

{"i18n":{"language":"zh_CN"},"isChannel":false,"data":{"share":"https://www.laohu8.com/m/news/2553205585?lang=zh_CN&edition=full","thumbnail":"","is_english":false,"pubTime":"2025-07-24 16:53","share_image_url":"https://static.laohu8.com/9a95c1376e76363c1401fee7d3717173","id":"2553205585","market":"us","top_or_hot":-1,"title":"技术解析｜NVIDIA Cedar 模块设计：1.6Tbps 的背后","media":"Ofweek光电信息网","content":"<div>\n<p>芝能智芯出品在面向人工智能集群与高性能计算平台的硬件系统中，网络模块的性能正在成为瓶颈。NVIDIA 针对其 DGX H100/H200 系统开发了名为 Cedar 的定制化网络模块，具备 1.6Tbps 的总网络吞吐能力，远高于常规 PCIe 网卡解决方案。这个模块的硬件结构、互连方式、接口布局以及其在散热与系统集成方面的优势，值得我们参考。Part 1Cedar 模块的系统架构与设计逻辑...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025072417074694b7ee23&s=b\">网页链接</a>\n\n</div>\n","source":"tencent","html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>技术解析｜NVIDIA Cedar 模块设计：1.6Tbps 的背后</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n技术解析｜NVIDIA Cedar 模块设计：1.6Tbps 的背后\n</h2>\n\n<h4 class=\"meta\">\n\n\n2025-07-24 16:53 北京时间&nbsp;&nbsp;&nbsp;<a href=http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025072417074694b7ee23&s=b><strong>Ofweek光电信息网</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>芝能智芯出品在面向人工智能集群与高性能计算平台的硬件系统中，网络模块的性能正在成为瓶颈。NVIDIA 针对其 DGX H100/H200 系统开发了名为 Cedar 的定制化网络模块，具备 1.6Tbps 的总网络吞吐能力，远高于常规 PCIe 网卡解决方案。这个模块的硬件结构、互连方式、接口布局以及其在散热与系统集成方面的优势，值得我们参考。Part 1Cedar 模块的系统架构与设计逻辑...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025072417074694b7ee23&s=b\">网页链接</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","isBrief":false,"type":0,"news_type":1,"symbol":"IE00BD6J9T35.USD","symbol_name":"NEUBERGER BERMAN NEXT GENERATION MOBILITY \"A\" (USD) ACC","start_time":0,"source_url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025072417074694b7ee23&s=b","article_id":"2553205585","we_media_id":null,"thumbnails":[],"rights":{"source":"tencent","url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025072417074694b7ee23&s=b","rn_cache_url":null,"customStyle":"body{padding-top:10px;}#news_title{font-weight:bold;#titleStyle#;}#news_description span{font-size:12px;#descriptionStyle#;}.footer-note{#statement#}","selectors":".mod-LoadTzbdNews, body","filters":".relate-stock, .hot-list, .recom-box, .wx-sou","directOrigin":true},"url":"https://stock-news.laohu8.com/highlight/detail?id=2553205585","pubTimestamp":1753347180,"columns":[],"sourceInfo":{"source_id":"tencent","name":"腾讯"},"weMediaInfo":null,"summary":"NVIDIA 针对其 DGX H100/H200 系统开发了名为 Cedar 的定制化网络模块，具备 1.6Tbps 的总网络吞吐能力，远高于常规 PCIe 网卡解决方案。连接器方面，Cedar 模块底部采用了定制化接口，类似 NVIDIA Grace Superchip 和 Hopper 系列处理器使用的高速连接基座，极大提升了信号完整性和模块稳定性。而 Cedar 模块可在一个独立模块中提供四路 400G 网络出口，大大降低了机箱网络接口复杂度，简化主板设计。","collect":0,"end_time":0,"defaultTopTitle":"qq.com","property":[],"viewcount":null,"language":"zh","relate_stocks":{"IE00BD6J9T35.USD":"NEUBERGER BERMAN NEXT GENERATION MOBILITY \"A\" (USD) ACC","2NVD.UK":"2X NVIDIA ETP","SG9999004303.SGD":"Nikko AM Shenton Global Opportunities SGD","LU2433249047.HKD":"THEMATICS META \"R/A\" (HKD) ACC","NVDD":"1倍做空NVDA ETF-Direxion","LU2092937148.SGD":"Blackrock ESG Multi-Asset A8 SGD-H","LU0820562030.AUD":"ALLIANZ INCOME AND GROWTH \"AMH2\" (AUDHDG) H2 INC","LU2250418816.HKD":"BGF WORLD TECHNOLOGY \"A\" (HKD) ACC","IE00BKPKM429.USD":"NEUBERGER BERMAN GLOBAL SUSTAINABLE EQUITY \"A\" (USD) ACC","IE00B19Z9505.USD":"美盛-美国大盘成长股A Acc","LU0965509010.AUD":"AB LOW VOLATILITY EQUITY PORTFOLIO \"AD\" (AUDHDG) INC","IE0005OL40V9.USD":"JANUS HENDERSON BALANCED \"A6M\" (USD) INC","NVD2.UK":"2X NVIDIA ETP","NVDY":"NVDA期权收益策略ETF-YieldMax","NVD":"2倍做空NVDA ETF-GraniteShares","NVD3.UK":"LS 3X NVIDIA","LU1366333091.USD":"FIDELITY GLOBAL FOCUS \"A\" (USD) ACC","LU0124676726.USD":"AB SICAV I - SUSTAINABLE US THEMATIC PORTFOLIO \"A\" (USD) ACC","BK4585":"ETF&股票定投概念","3NVD.UK":"LS 3X NVIDIA","LU0266013472.USD":"AXA WF - Framlington Longevity Economy A Cap USD","NVDX":"2倍做多NVDA ETF-T-Rex","LU1803068979.SGD":"FTIF - Franklin Technology A (acc) SGD-H1","LU0345768153.USD":"NINETY ONE GSF GLOBAL STRATEGIC MANAGED \"A\" (USD) ACC","NVDS.UK":"LS -1X NVIDIA","IE00BZ199S13.USD":"BNY MELLON MOBILITY INNOVATION \"B\" (USD) ACC","LU0061474960.USD":"天利环球焦点基金AU Acc","NVDA":"英伟达","IE0004086264.USD":"BNY MELLON GLOBAL OPPORTUNITIES \"A\" (USD) ACC","LU1720051108.HKD":"ALLIANZ GLOBAL ARTIFICIAL INTELLIGENCE \"AT\" (HKD) ACC","LU1914381329.SGD":"Allianz Best Styles Global Equity Cl ET Acc H2-SGD","NVDU":"2倍做多NVDA ETF-Direxion","LU0157215616.USD":"FIDELITY GLOBAL FOCUS \"A\" INC","LU2357125470.USD":"BNP PARIBAS ENERGY TRANSITION \"CLASSIC RH\" (USDHDG) ACC","SGXZ23171101.USD":"NIKKO AM SHENTON GLOBAL OPPORTUNITIES (USD) ACC","NVDS":"1.5倍做空NVDA ETF-Tradr","LU2083900584.USD":"HSBC GIF US INCOME FOCUSED \"AM2\" (USD) INC","LU1718418525.SGD":"JPMorgan Investment Funds - Global Select Equity A (acc) SGD","IE00BK4W5L77.USD":"HSBC GLOBAL FUNDS ICAV US EQUITY INDEX \"HC\" (USD) ACC","LU2360032135.SGD":"ALLSPRING GLOBAL EQUITY ENHANCED INCOME \"A\" (SGDHDG) INC","LU1066051811.HKD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM2\" (HKD) INC","LU2286300806.USD":"Allianz Cyber Security AT Acc USD","LU0057025933.USD":"SUSTAINABLE GLOBAL THEMATIC PORTFOLIO \"AX\" (USD) ACC","NVIW.SI":"NVDA 3xLongSG261006","LU2462157665.USD":"ALLIANZ GLOBAL INCOME \"A\" (USD) INC","SNVD.UK":"LS -1X NVIDIA","LU0107464264.USD":"abrdn SICAV I - GLOBAL INNOVATION EQUITY \"A\" (USD) ACC","LU2168564065.EUR":"AZ ALLOCATION - TREND \"AAZ\" (EUR) ACC","LU1935042215.USD":"MANULIFE GF GLOBAL MULTI-ASSET DIVERSIFIED INCOME  \"AA\" (USD) INC A"},"translate_title":"Technical Analysis | NVIDIA Cedar Module Design: Behind 1.6 Tbps","themeId":null,"isJumpTheme":false,"ttsUrl":null,"symbols_score_info":{"NVD3.UK":0.6,"NVDS.UK":0.6,"NVDS":0.6,"NVDA":1,"3NVD.UK":0.6,"SNVD.UK":0.6,"NVDU":0.6,"NVIW.SI":0.6,"NVD":0.6,"NVD2.UK":0.6,"NVDX":0.6,"NVDD":0.6,"NVDY":0.6,"2NVD.UK":0.6},"content_text":"芝能智芯出品在面向人工智能集群与高性能计算平台的硬件系统中，网络模块的性能正在成为瓶颈。NVIDIA 针对其 DGX H100/H200 系统开发了名为 Cedar 的定制化网络模块，具备 1.6Tbps 的总网络吞吐能力，远高于常规 PCIe 网卡解决方案。这个模块的硬件结构、互连方式、接口布局以及其在散热与系统集成方面的优势，值得我们参考。Part 1Cedar 模块的系统架构与设计逻辑NVIDIA Cedar 模块最早在 DGX H100 系统中亮相，其最大的特点在于抛弃了常规的 PCIe 网卡形式，而是采用类似于 GPU SXM 接口的模块化结构。这一选择的核心目的是提升网络接口的密度与带宽集成能力，同时降低散热复杂性。模块在物理结构上呈扁平横置状态，从上部观察，其主核心是四颗 NVIDIA ConnectX-7 网卡芯片。每颗网卡具备 400Gbps 的网络带宽，四颗并联则构成了 1.6Tbps 的总带宽能力。与传统分布在不同主板插槽中的网络卡相比，这种“板上集成”设计节省了大量的主板空间，并大幅提升了端口密度。连接器方面，Cedar 模块底部采用了定制化接口，类似 NVIDIA Grace Superchip 和 Hopper 系列处理器使用的高速连接基座，极大提升了信号完整性和模块稳定性。这种设计绕开了标准 PCIe 总线，通过主板直连方式构建更短的传输路径，从而减少延迟并提升吞吐效率。在模块侧面还设有光纤收发接头接口，采用 Flyover cable（飞越线缆）技术将数据从模块导出至系统背部的光纤接口阵列。这种布局既优化了空间利用，也有助于降低内部线缆的布线复杂度。在管理接口方面，模块提供了两个独立的带外管理（OOB）端口，用于系统级的远程控制和健康监测，满足数据中心自动化运维的需求。Part 2散热与系统集成的工程优化与许多使用垂直 PCIe 插槽的传统网卡相比，Cedar 模块在冷却方面展示了显著优势。由于其模块化布局为水平排布，整块网络模块可以在上方加装单一大面积的散热器或液冷冷板。相比于为每一张 PCIe 卡单独部署散热系统的方式，这种整合式冷却架构能够极大地减少冷却元件的数量与液冷管路布线。实际硬件上观察，模块顶部的铝制散热器覆盖全部四颗 ConnectX-7 芯片与电源管理电路，形成整体热路径，有助于均衡热负载。此外，由于模块以扁平形式直接嵌入主板或机箱托架，其热量可通过冷板直接导向液冷回路，避免了多次换热所造成的能量损耗。这一架构也为未来更高热密度的网络系统预留了空间。例如，在液冷方案成熟后，Cedar 模块可以直接上方加装板式冷却装置，实现一体化散热，兼顾空间与可靠性。从系统集成角度来看，Cedar 模块为高密度 DGX 服务器架构带来更大弹性。在传统架构中，八路 GPU 系统往往因网卡数量限制而受限于网络接口数量和拓扑设计。而 Cedar 模块可在一个独立模块中提供四路 400G 网络出口，大大降低了机箱网络接口复杂度，简化主板设计。其布线采用背部输出形式，在整机布局中避免了前端信号干扰，并为多节点机柜部署提供便利。此外，由于模块采用统一封装规范，也便于未来在不同 DGX 型号或 MGX 平台间快速移植。小结NVIDIA Cedar 模块代表了超高带宽网络接口设计的一种方向，即通过高度集成、定制封装与优化布线实现带宽密度最大化与散热效率提升的统一，打破了传统 PCIe 网络架构在规模化部署上的限制，使得 DGX 系列服务器具备更强的集群通信能力与散热弹性。随着 HGX B300 和 ConnectX-8 的逐步普及，这种以模块化方式实现超带宽互联的设计理念将可能延伸至更多服务器平台。Cedar 模块虽属过渡产品，但其所体现的工程思维和热-电-互连一体化设计，值得广泛借鉴。","kind":"news","is_publish_news":true,"is_publish_highlight":false,"is_publish_live":false,"is_publish_wemedia":null,"editions":null,"column":"","sentiment":"0","news_tag":"","news_rank":0,"symbols":[],"gpt_button":0,"need_auth":false,"code":"91000000","status":"200"},"commentList":[],"isCommentEnd":true,"newsSizeData":{"likeSize":0,"commentSize":0,"repostSize":0,"favoriteSize":0,"likeStatus":false,"favoriteStatus":false},"APP":{"userAgent":"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)","isDev":false,"isTTM":false,"tenantId":"TBCN","deviceId":"web-server-community-laohu8-v3","version":"4.36.2","shortVersion":"4.36.2","platform":"web","vendor":"web","appName":"laohu8","isIOS":false,"isAndroid":false,"isTiger":false,"isTHS":false,"isWeiXin":false,"isWeiXinMini":false,"isWeiBo":false,"isQQ":false,"isBaiduSwan":false,"isBaiduBox":false,"isDingTalk":false,"isToutiao":false,"isOnePlus":false,"isHuaWei":false,"isXiaomi":false,"isXiaomiWebView":false,"isOppo":false,"isVivo":false,"isSamsung":false,"isMobile":false},"href":"/m/news/2553205585","isCrawlerRequest":true}