[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"site-home-live-entry-setting":3,"site-header-contests":8,"site-header-camps":49,"study-material-detail-study-14":104},{"enabled":4,"title":5,"statusText":5,"buttonText":5,"campSlug":5,"stage":6,"classroomId":7,"hasClassroom":4},false,"",0,null,{"total":9,"list":10,"pageNum":44,"pageSize":45,"size":9,"startRow":44,"endRow":9,"pages":44,"prePage":6,"nextPage":6,"isFirstPage":46,"isLastPage":46,"hasPreviousPage":4,"hasNextPage":4,"navigatePages":47,"navigatepageNums":48,"navigateFirstPage":44,"navigateLastPage":44},3,[11,25,35],{"slug":12,"title":13,"logo":7,"liveUrl":14,"prize":15,"sponsor":16,"description":17,"organizer":18,"startTime":19,"endTime":20,"registrationDeadline":21,"status":22,"dashboardVisibility":23,"createdAt":24},"spring2026","2026春季启元人工智能大赛","https:\u002F\u002Fwww.bilibili.com\u002Fblackboard\u002Flive\u002Flive-activity-player.html?cid=1822703816&quality=0","200,000","麒麟软件有限公司","“2026春季启元人工智能大赛”是由启元实验室主办的面向全国人工智能领域开发者、科研人员及高校学生的权威赛事。大赛旨在促进与培养具备工程实践能力和创新能力的优秀人工智能领域开发者、推广国产软硬件技术与生态、助力国家人工智能产业高质量发展。\n各参赛选手需在规定的比赛周期内，依据赛事规则选定赛题并完成相应挑战，优胜者将共同角逐总额20万元的赛事奖金池，并且还将有机会获得实习和知名企业优质岗位内推资格。","启元实验室","2026-04-19 16:00:00","2026-07-13 16:00:00","2026-06-16 16:00:00","ongoing","private","2026-04-16 08:00:00",{"slug":26,"title":27,"logo":7,"liveUrl":14,"prize":28,"sponsor":7,"description":29,"organizer":18,"startTime":30,"endTime":31,"registrationDeadline":32,"status":33,"dashboardVisibility":23,"createdAt":34},"autumn2025","2025秋季启元人工智能大赛","150,000","“2025秋季启元人工智能大赛”是由启元实验室主办的面向全国人工智能领域开发者、科研人员及高校学生的权威赛事。大赛旨在促进与培养具备工程实践能力和创新能力的优秀人工智能领域开发者、推广国产软硬件技术与生态、助力国家人工智能产业高质量发展。","2025-11-08 08:00:00","2026-01-13 08:00:00","2025-12-30 08:00:00","ended","2025-11-03 08:00:00",{"slug":36,"title":37,"logo":7,"liveUrl":14,"prize":38,"sponsor":7,"description":39,"organizer":18,"startTime":40,"endTime":41,"registrationDeadline":42,"status":33,"dashboardVisibility":23,"createdAt":43},"summer2025","启元人工智能大赛(系统方向)","80,000","启元人工智能大赛由启元实验室主办，其中系统赛道基于启元实验室九源软件栈的相关开源项目，通过拉动开源社区力量，一方面促进国产智能计算生态的发展，另一方面也为人工智能系统方向的学生及从业者提供一个实践平台。","2025-07-08 08:00:00","2025-08-25 08:00:00","2025-07-22 08:00:00","2025-06-22 08:00:00",1,5,true,8,[44],{"total":45,"list":50,"pageNum":44,"pageSize":45,"size":45,"startRow":44,"endRow":45,"pages":44,"prePage":6,"nextPage":6,"isFirstPage":46,"isLastPage":46,"hasPreviousPage":4,"hasNextPage":4,"navigatePages":47,"navigatepageNums":103,"navigateFirstPage":44,"navigateLastPage":44},[51,62,73,84,94],{"id":52,"slug":53,"title":54,"logo":55,"publicLiveEmbedUrl":14,"description":56,"introduction":57,"organizer":18,"coOrganizer":5,"sponsor":16,"startTime":58,"endTime":59,"status":22,"dashboardVisibility":60,"createdAt":58,"updatedAt":61},"019eeeff-6b00-7d55-a629-efafded3e232","summer2026","2026年夏季InfiniTensor大模型与人工智能系统训练营","https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fuploads\u002F2026-06\u002F019eef7f-7816-7828-9309-79c2c61dd568.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=NVzfZOYIHlQXv7MFHLCIDyq1g0c%3D","由启元实验室发起的InfiniTensor大模型与人工智能系统训练营，面向开源社区、高校及产业界伙伴开放。活动以「普及AI全栈知识技术、培养工程实践与创新型人才、推广国产软硬件技术生态」为核心目标，助力国家人工智能产业高质量发展。","...","2026-06-22 08:00:00","2026-10-19 08:00:00","public","2026-10-12 08:00:00",{"id":63,"slug":64,"title":65,"logo":66,"publicLiveEmbedUrl":7,"description":67,"introduction":57,"organizer":18,"coOrganizer":5,"sponsor":68,"startTime":69,"endTime":70,"status":33,"dashboardVisibility":23,"createdAt":71,"updatedAt":72},"019af3f3-e06a-7245-aa8d-37dc99a69c73","winter2025","2025年冬季InfiniTensor大模型与人工智能系统训练营","https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fuploads\u002F2025-12\u002F019afc69-b7ca-7805-91fb-605429671c01.png?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=aLVPGeaSuxrMWkA0%2FuFUP%2BSA%2BQU%3D","由启元实验室发起的InfiniTensor大模型与人工智能系统训练营，面向开源社区、高校及产业界伙伴开放。活动以「普及AI全栈知识技术、培养工程实践与创新型人才、推广国产软硬件技术生态」为核心目标，助力国家人工智能产业高质量发展。 \n","新华三技术有限公司","2025-12-25 08:00:00","2026-03-27 08:00:00","2025-12-06 08:00:00","2026-10-22 20:24:30",{"id":74,"slug":36,"title":75,"logo":76,"publicLiveEmbedUrl":7,"description":77,"introduction":57,"organizer":18,"coOrganizer":5,"sponsor":78,"startTime":79,"endTime":80,"status":81,"dashboardVisibility":23,"createdAt":82,"updatedAt":83},"01980342-208c-7803-b3f3-19e1d31960ba","2025年夏季InfiniTensor大模型与人工智能系统训练营","https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fuploads\u002F2025-07\u002F0198045e-28fe-78f0-95fe-23835ce58395.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=%2FWEj1cuYDG8cEmjwi3vCrq7EJsQ%3D","InfiniTensor大模型与人工智能系统训练营由启元实验室及清华大学 AI 系统方向团队发起，面向开源社区、高校和产业界，普及人工智能各方面知识和技术，培养相关人才。训练营联合开源项目开发者、学术界大牛、工业界资深工程师参与授课与项目指导。课程采用「线上直播授课 + 在线答疑 + 项目实践辅导 」的方式进行，确保每位同学都能获得充分成长。训练营全程免费，通过视频、直播课、习题和项目实践，在人工智能飞速发展的背景下，为学员提供一窥深层技术的机会。","北京积算科技有限公司","2025-07-14 08:00:00","2025-09-30 08:00:00","upcoming","2025-07-13 08:00:00","2026-10-22 20:24:58",{"id":85,"slug":86,"title":87,"logo":88,"publicLiveEmbedUrl":7,"description":89,"introduction":57,"organizer":18,"coOrganizer":5,"sponsor":5,"startTime":90,"endTime":91,"status":81,"dashboardVisibility":23,"createdAt":92,"updatedAt":93},"01980cf9-a7d8-7796-ae86-0bbe05334add","winter2024","2024年冬季InfiniTensor大模型与人工智能系统训练营","https:\u002F\u002Ftesting-cdn.infinitensor.com\u002Fuploads\u002F2025-07\u002F01981269-ce23-7624-a287-9f07dd0d85f6.jpg","InfiniTensor大模型与人工智能系统训练营由清华大学发起，面向开源社区、高校和产业界，普及人工智能各方面知识和技术，培养相关人才。训练营全程免费，通过视频、直播课、习题和项目实践，在人工智能飞速发展的背景下，为学员提供一窥深层技术的机会。报名学员请阅读“导学”课程页的训练营介绍。","2025-01-05 08:00:00","2025-03-22 08:00:00","2025-07-15 08:00:00","2026-10-22 20:24:39",{"id":95,"slug":96,"title":97,"logo":98,"publicLiveEmbedUrl":7,"description":99,"introduction":57,"organizer":18,"coOrganizer":5,"sponsor":5,"startTime":100,"endTime":101,"status":81,"dashboardVisibility":23,"createdAt":92,"updatedAt":102},"01980c20-363a-7dc3-8f5a-58a68e08dcc4","summer2024","2024年夏季InfiniTensor大模型与人工智能系统训练营","https:\u002F\u002Ftesting-cdn.infinitensor.com\u002Fuploads\u002F2025-07\u002F01981269-fbfa-788a-afcd-17fa9be74ff9.jpg","InfiniTensor大模型与人工智能系统训练营由清华大学发起，面向开源社区和高校，普及人工智能各方面知识和技术，培养相关人才。训练营全程免费，通过课程、习题和项目实践，在人工智能飞速发展的背景下，为学员提供一窥深层技术的机会。","2024-07-20 08:00:00","2024-09-29 08:00:00","2026-10-22 20:24:49",[44],{"currentItem":105,"processedData":125},{"img":106,"keywords":107,"treeType":44,"description":108,"showStatus":44,"updateTime":109,"type":110,"title":111,"content":112,"nextArticle":113,"createTime":109,"sortOrder":117,"canonicalPath":118,"id":119,"contentType":120,"prevArticle":121},"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_fm.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=TB3bD2NYSbUunFncGumtNMW0X84%3D","InfiniTensor,冬季训练营,集群通信,分布式并行,网络互联","围绕互联、集群与通信主题，介绍分布式系统中的网络通信基础与并行协同机制。","2026-10-22 20:38:11",2,"InfiniTensor 冬季训练营通信与并行第一课《互联、集群与通信》","![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw01.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=k6wg4riuHcYMvlJDXtG%2Fchhhaz0%3D)\r\n\r\n\r\n\r\n2026 年 1 月 14 日，2025 冬季 InfiniTensor 大模型与人工智能系统训练营通信与并行系列第一课《互联、集群与通信》正式开讲。\r\n\r\n本节课讲述了通信在通用场景及深度学习领域的必要性，讲解计算机单机和跨机通信的方式、技术，介绍数据中心的分类、服务器类型、网络拓扑结构以及拓展模式等内容。\r\n\r\n### 通信的本质与需求\r\n- 通信的本质就是信息交换与传输\r\n- 通用场景下的通信需求（互联网访问、数据交换）\r\n- AI 领域的特殊通信需求\r\n\r\n### 通信的必要性\r\n- 模型规模增长趋势（从 GPT-3 到现代大模型）\r\n- 训练数据规模扩大带来的挑战\r\n- 算力与内存需求增长\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw02.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=r%2FonGG6ESpJz%2BLP4CvMG3TZqlRA%3D)\r\n\r\n### 计算机通信体系结构\r\n- 单机多 GPU 通信\r\n- 多机跨网络通信\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw03.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=5t62RW840bJQGelABaJ4a7XnHy8%3D)\r\n\r\n\r\n### 单机通信技术详解\r\n- PCIe vs 老式总线（PCI\u002FAGP）\r\n- PCIe 特点：点对点、Full-duplex、串行总线\r\n- PCIe 多通道扩展（x1, x2，...，x16）\r\n- 传输速率单位换算（GT\u002Fs → GB\u002Fs）\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw04.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=e%2FOOn7IZula34tZtMUBniaOcbmA%3D)\r\n\r\n- 编码开销与有效带宽计算\r\n\r\n### GPU 间数据传输优化\r\n- 传统拷贝路径：GPU→CPU→GPU\r\n- DMA（Direct Memory Access）技术\r\n- Peer-to-Peer（P2P）直连技术\r\n- PCIe 形成树状结构\r\n- PCIe Switch 拓展连接更多设备\r\n- 跨 CPU 通信路径分析\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw05.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=F8Jnf4oMBSqx9HgrKHVgC0FSjNc%3D)\r\n\r\n\r\n### 多 CPU 与 NUMA 架构\r\n- 多 CPU 互联技术\r\n- Intel CPU QPI\u002FUPI\r\n- SMP\u002FUMA：共享内存与总线\r\n- AMP\u002FNUMA：非均匀内存访问\r\n- NUMA 优化策略：局部性优化\r\n\r\n### GPU 高速互联技术\r\n- GPU 间直连通信\r\n- 全连接拓扑（Fully Connected Mesh）\r\n- 带宽对比：NVLink vs PCIe\r\n- NVSwitch 扩展技术\r\n\r\n### 跨机通信与网络技术\r\n- 网络接口卡（NIC）\r\n- 有线与无线网卡\r\n- 协议栈处理与 CPU 参与\r\n- RDMA（Remote Direct Memory Access）\r\n\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw06.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=044GgMZZDBLnCI744kAJktvuWmE%3D)\r\n- 零拷贝与内核旁路\r\n- RDMA实现方式：\r\n ① InfiniBand（原生RDMA）\r\n ② RoCE（RDMA over Converge Etnernet）\r\n ③ iWARP（基于TCP\u002FIP）\r\n\r\n### 数据中心与集群架构\r\n#### 1. 数据中心类型：\r\n- 通算中心（CPU 为主）\r\n- 超算中心（CPU+GPU）\r\n- 智算中心（GPU\u002FASIC 为主）\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw07.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=o1y67E2ADWLu08XH%2BOM3gXMf9GU%3D)\r\n\r\n#### 2. 服务器类型与机架设计：\r\n- 塔式服务器、机架式服务器、刀片服务器\r\n#### 3. 散热系统：\r\n- 风冷系统\r\n- 液冷系统（冷板式、沉浸式\u002F浸没式）\r\n\r\n### 集群拓展与网络拓扑\r\n#### 1. 算力扩展\r\n- Scale Up：纵向扩展\r\n- Scale Out：横向扩展\r\n- Scale Across：跨域扩展\r\n#### 2. 超节点（Superpod）\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw08.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=vjtwwL8BmJpJmx4hfkv26shijzM%3D)\r\n\r\n#### 3. 网络拓扑结构：\r\n- 三层树状结构（接入\u002F汇聚\u002F核心层）\r\n- 南北流量 vs 东西流量\r\n- Spine-Leaf 结构\r\n- 其他高性能计算网络拓扑\r\n\r\n### 算网协同与在网计算\r\n- 算网协同：网络感知计算需求、动态资源调整\r\n- 在网计算：网络设备数据处理能力、计算任务卸载至网络层，边传输，边计算\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw09.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=8nbIoee6GWyFMAGAP3rVVTmK96Y%3D)\r\n\r\n\r\n\r\n### 总结与展望\r\n本节课梳理了深度学习通信的核心技术与硬件架构。后续课程将深入训练与推理中的通信算法与优化实践。\r\n\r\n\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_01_zw6.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=ZwRth4vKI1lmA8Y4THFk9als6XI%3D)\r\n\r\n\r\n### 💡 学习贴士\r\n更多细节及完整课程内容，请观看直播或查看课程回放。\r\n\r\n链接：https:\u002F\u002Fwww.infinitensor.com\u002Fcamp\u002Fwinter2025\r\n\r\n### 📚 训练营报名持续进行中\r\n扫码报名与加入群聊，或点击 **「阅读原文」** 直达官网报名\r\n\r\n\r\n![](https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_01_zw7.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=uxAgm6FPaH%2Bpzv9S8ucllx2qHnM%3D)",{"publishTime":7,"createTime":109,"canonicalPath":114,"id":115,"title":116},"\u002Fstudy-material\u002Fstudy-13","study-13","InfiniTensor 冬季训练营前沿模型第一课《大模型学习方法》",95,"\u002Fstudy-material\u002Fstudy-14","study-14","markdown",{"publishTime":7,"createTime":109,"canonicalPath":122,"id":123,"title":124},"\u002Fstudy-material\u002Fstudy-15","study-15","InfiniTensor 冬季训练营推理引擎第二课《大模型原理与结构》",{"htmlContent":126,"headings":127},"\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw01.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=k6wg4riuHcYMvlJDXtG%2Fchhhaz0%3D\" alt=\"\">\u003C\u002Fp>\n\u003Cp>2026 年 1 月 14 日，2025 冬季 InfiniTensor 大模型与人工智能系统训练营通信与并行系列第一课《互联、集群与通信》正式开讲。\u003C\u002Fp>\n\u003Cp>本节课讲述了通信在通用场景及深度学习领域的必要性，讲解计算机单机和跨机通信的方式、技术，介绍数据中心的分类、服务器类型、网络拓扑结构以及拓展模式等内容。\u003C\u002Fp>\n\u003Ch3 id=\"通信的本质与需求\">\u003Ca class=\"anchor\" href=\"#通信的本质与需求\" aria-label=\"跳转到标题：通信的本质与需求\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>通信的本质与需求\u003C\u002Fh3>\u003Cul>\n\u003Cli>通信的本质就是信息交换与传输\u003C\u002Fli>\n\u003Cli>通用场景下的通信需求（互联网访问、数据交换）\u003C\u002Fli>\n\u003Cli>AI 领域的特殊通信需求\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"通信的必要性\">\u003Ca class=\"anchor\" href=\"#通信的必要性\" aria-label=\"跳转到标题：通信的必要性\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>通信的必要性\u003C\u002Fh3>\u003Cul>\n\u003Cli>模型规模增长趋势（从 GPT-3 到现代大模型）\u003C\u002Fli>\n\u003Cli>训练数据规模扩大带来的挑战\u003C\u002Fli>\n\u003Cli>算力与内存需求增长\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw02.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=r%2FonGG6ESpJz%2BLP4CvMG3TZqlRA%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch3 id=\"计算机通信体系结构\">\u003Ca class=\"anchor\" href=\"#计算机通信体系结构\" aria-label=\"跳转到标题：计算机通信体系结构\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>计算机通信体系结构\u003C\u002Fh3>\u003Cul>\n\u003Cli>单机多 GPU 通信\u003C\u002Fli>\n\u003Cli>多机跨网络通信\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw03.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=5t62RW840bJQGelABaJ4a7XnHy8%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch3 id=\"单机通信技术详解\">\u003Ca class=\"anchor\" href=\"#单机通信技术详解\" aria-label=\"跳转到标题：单机通信技术详解\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>单机通信技术详解\u003C\u002Fh3>\u003Cul>\n\u003Cli>PCIe vs 老式总线（PCI\u002FAGP）\u003C\u002Fli>\n\u003Cli>PCIe 特点：点对点、Full-duplex、串行总线\u003C\u002Fli>\n\u003Cli>PCIe 多通道扩展（x1, x2，...，x16）\u003C\u002Fli>\n\u003Cli>传输速率单位换算（GT\u002Fs → GB\u002Fs）\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw04.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=e%2FOOn7IZula34tZtMUBniaOcbmA%3D\" alt=\"\">\u003C\u002Fp>\n\u003Cul>\n\u003Cli>编码开销与有效带宽计算\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"gpu-间数据传输优化\">\u003Ca class=\"anchor\" href=\"#gpu-间数据传输优化\" aria-label=\"跳转到标题：GPU 间数据传输优化\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>GPU 间数据传输优化\u003C\u002Fh3>\u003Cul>\n\u003Cli>传统拷贝路径：GPU→CPU→GPU\u003C\u002Fli>\n\u003Cli>DMA（Direct Memory Access）技术\u003C\u002Fli>\n\u003Cli>Peer-to-Peer（P2P）直连技术\u003C\u002Fli>\n\u003Cli>PCIe 形成树状结构\u003C\u002Fli>\n\u003Cli>PCIe Switch 拓展连接更多设备\u003C\u002Fli>\n\u003Cli>跨 CPU 通信路径分析\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw05.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=F8Jnf4oMBSqx9HgrKHVgC0FSjNc%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch3 id=\"多-cpu-与-numa-架构\">\u003Ca class=\"anchor\" href=\"#多-cpu-与-numa-架构\" aria-label=\"跳转到标题：多 CPU 与 NUMA 架构\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>多 CPU 与 NUMA 架构\u003C\u002Fh3>\u003Cul>\n\u003Cli>多 CPU 互联技术\u003C\u002Fli>\n\u003Cli>Intel CPU QPI\u002FUPI\u003C\u002Fli>\n\u003Cli>SMP\u002FUMA：共享内存与总线\u003C\u002Fli>\n\u003Cli>AMP\u002FNUMA：非均匀内存访问\u003C\u002Fli>\n\u003Cli>NUMA 优化策略：局部性优化\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"gpu-高速互联技术\">\u003Ca class=\"anchor\" href=\"#gpu-高速互联技术\" aria-label=\"跳转到标题：GPU 高速互联技术\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>GPU 高速互联技术\u003C\u002Fh3>\u003Cul>\n\u003Cli>GPU 间直连通信\u003C\u002Fli>\n\u003Cli>全连接拓扑（Fully Connected Mesh）\u003C\u002Fli>\n\u003Cli>带宽对比：NVLink vs PCIe\u003C\u002Fli>\n\u003Cli>NVSwitch 扩展技术\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"跨机通信与网络技术\">\u003Ca class=\"anchor\" href=\"#跨机通信与网络技术\" aria-label=\"跳转到标题：跨机通信与网络技术\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>跨机通信与网络技术\u003C\u002Fh3>\u003Cul>\n\u003Cli>网络接口卡（NIC）\u003C\u002Fli>\n\u003Cli>有线与无线网卡\u003C\u002Fli>\n\u003Cli>协议栈处理与 CPU 参与\u003C\u002Fli>\n\u003Cli>RDMA（Remote Direct Memory Access）\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw06.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=044GgMZZDBLnCI744kAJktvuWmE%3D\" alt=\"\">\u003C\u002Fp>\n\u003Cul>\n\u003Cli>零拷贝与内核旁路\u003C\u002Fli>\n\u003Cli>RDMA实现方式：\u003Cbr> ① InfiniBand（原生RDMA）\u003Cbr> ② RoCE（RDMA over Converge Etnernet）\u003Cbr> ③ iWARP（基于TCP\u002FIP）\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"数据中心与集群架构\">\u003Ca class=\"anchor\" href=\"#数据中心与集群架构\" aria-label=\"跳转到标题：数据中心与集群架构\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>数据中心与集群架构\u003C\u002Fh3>\u003Ch4 id=\"1-数据中心类型\">\u003Ca class=\"anchor\" href=\"#1-数据中心类型\" aria-label=\"跳转到标题：1. 数据中心类型：\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>1. 数据中心类型：\u003C\u002Fh4>\u003Cul>\n\u003Cli>通算中心（CPU 为主）\u003C\u002Fli>\n\u003Cli>超算中心（CPU+GPU）\u003C\u002Fli>\n\u003Cli>智算中心（GPU\u002FASIC 为主）\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw07.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=o1y67E2ADWLu08XH%2BOM3gXMf9GU%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch4 id=\"2-服务器类型与机架设计\">\u003Ca class=\"anchor\" href=\"#2-服务器类型与机架设计\" aria-label=\"跳转到标题：2. 服务器类型与机架设计：\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>2. 服务器类型与机架设计：\u003C\u002Fh4>\u003Cul>\n\u003Cli>塔式服务器、机架式服务器、刀片服务器\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch4 id=\"3-散热系统\">\u003Ca class=\"anchor\" href=\"#3-散热系统\" aria-label=\"跳转到标题：3. 散热系统：\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>3. 散热系统：\u003C\u002Fh4>\u003Cul>\n\u003Cli>风冷系统\u003C\u002Fli>\n\u003Cli>液冷系统（冷板式、沉浸式\u002F浸没式）\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"集群拓展与网络拓扑\">\u003Ca class=\"anchor\" href=\"#集群拓展与网络拓扑\" aria-label=\"跳转到标题：集群拓展与网络拓扑\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>集群拓展与网络拓扑\u003C\u002Fh3>\u003Ch4 id=\"1-算力扩展\">\u003Ca class=\"anchor\" href=\"#1-算力扩展\" aria-label=\"跳转到标题：1. 算力扩展\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>1. 算力扩展\u003C\u002Fh4>\u003Cul>\n\u003Cli>Scale Up：纵向扩展\u003C\u002Fli>\n\u003Cli>Scale Out：横向扩展\u003C\u002Fli>\n\u003Cli>Scale Across：跨域扩展\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch4 id=\"2-超节点-superpod\">\u003Ca class=\"anchor\" href=\"#2-超节点-superpod\" aria-label=\"跳转到标题：2. 超节点（Superpod）\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>2. 超节点（Superpod）\u003C\u002Fh4>\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw08.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=vjtwwL8BmJpJmx4hfkv26shijzM%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch4 id=\"3-网络拓扑结构\">\u003Ca class=\"anchor\" href=\"#3-网络拓扑结构\" aria-label=\"跳转到标题：3. 网络拓扑结构：\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>3. 网络拓扑结构：\u003C\u002Fh4>\u003Cul>\n\u003Cli>三层树状结构（接入\u002F汇聚\u002F核心层）\u003C\u002Fli>\n\u003Cli>南北流量 vs 东西流量\u003C\u002Fli>\n\u003Cli>Spine-Leaf 结构\u003C\u002Fli>\n\u003Cli>其他高性能计算网络拓扑\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"算网协同与在网计算\">\u003Ca class=\"anchor\" href=\"#算网协同与在网计算\" aria-label=\"跳转到标题：算网协同与在网计算\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>算网协同与在网计算\u003C\u002Fh3>\u003Cul>\n\u003Cli>算网协同：网络感知计算需求、动态资源调整\u003C\u002Fli>\n\u003Cli>在网计算：网络设备数据处理能力、计算任务卸载至网络层，边传输，边计算\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_14_zw09.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=8nbIoee6GWyFMAGAP3rVVTmK96Y%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch3 id=\"总结与展望\">\u003Ca class=\"anchor\" href=\"#总结与展望\" aria-label=\"跳转到标题：总结与展望\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>总结与展望\u003C\u002Fh3>\u003Cp>本节课梳理了深度学习通信的核心技术与硬件架构。后续课程将深入训练与推理中的通信算法与优化实践。\u003C\u002Fp>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_01_zw6.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=ZwRth4vKI1lmA8Y4THFk9als6XI%3D\" alt=\"\">\u003C\u002Fp>\n\u003Ch3 id=\"学习贴士\">\u003Ca class=\"anchor\" href=\"#学习贴士\" aria-label=\"跳转到标题：💡 学习贴士\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>💡 学习贴士\u003C\u002Fh3>\u003Cp>更多细节及完整课程内容，请观看直播或查看课程回放。\u003C\u002Fp>\n\u003Cp>链接：\u003Ca href=\"https:\u002F\u002Fwww.infinitensor.com\u002Fcamp\u002Fwinter2025\">https:\u002F\u002Fwww.infinitensor.com\u002Fcamp\u002Fwinter2025\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch3 id=\"训练营报名持续进行中\">\u003Ca class=\"anchor\" href=\"#训练营报名持续进行中\" aria-label=\"跳转到标题：📚 训练营报名持续进行中\">\u003Cspan class=\"octicon octicon-link\">\u003C\u002Fspan>\u003C\u002Fa>📚 训练营报名持续进行中\u003C\u002Fh3>\u003Cp>扫码报名与加入群聊，或点击 \u003Cstrong>「阅读原文」\u003C\u002Fstrong> 直达官网报名\u003C\u002Fp>\n\u003Cp>\u003Cimg src=\"https:\u002F\u002Foss.infinitensor.com\u002Fprod\u002Fimages\u002FKnowledge_Base\u002FKB_01_zw7.jpg?Expires=1792752838&OSSAccessKeyId=LTAI5t9R6uzvTa69CT92cFvV&Signature=uxAgm6FPaH%2Bpzv9S8ucllx2qHnM%3D\" alt=\"\">\u003C\u002Fp>\n",[128,130,132,134,136,139,142,145,147,149,153,156,159,161,164,167,170,172,174,177],{"id":129,"text":129,"level":9},"通信的本质与需求",{"id":131,"text":131,"level":9},"通信的必要性",{"id":133,"text":133,"level":9},"计算机通信体系结构",{"id":135,"text":135,"level":9},"单机通信技术详解",{"id":137,"text":138,"level":9},"gpu-间数据传输优化","GPU 间数据传输优化",{"id":140,"text":141,"level":9},"多-cpu-与-numa-架构","多 CPU 与 NUMA 架构",{"id":143,"text":144,"level":9},"gpu-高速互联技术","GPU 高速互联技术",{"id":146,"text":146,"level":9},"跨机通信与网络技术",{"id":148,"text":148,"level":9},"数据中心与集群架构",{"id":150,"text":151,"level":152},"1-数据中心类型","1. 数据中心类型：",4,{"id":154,"text":155,"level":152},"2-服务器类型与机架设计","2. 服务器类型与机架设计：",{"id":157,"text":158,"level":152},"3-散热系统","3. 散热系统：",{"id":160,"text":160,"level":9},"集群拓展与网络拓扑",{"id":162,"text":163,"level":152},"1-算力扩展","1. 算力扩展",{"id":165,"text":166,"level":152},"2-超节点-superpod","2. 超节点（Superpod）",{"id":168,"text":169,"level":152},"3-网络拓扑结构","3. 网络拓扑结构：",{"id":171,"text":171,"level":9},"算网协同与在网计算",{"id":173,"text":173,"level":9},"总结与展望",{"id":175,"text":176,"level":9},"学习贴士","💡 学习贴士",{"id":178,"text":179,"level":9},"训练营报名持续进行中","📚 训练营报名持续进行中"]