[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f48vmA0ZJak35AWPFeqameyxq3bBX2JO2Rrt16hdD1vg":3},{"item":4,"related":44},{"id":5,"type":6,"title":7,"slug":8,"summary":9,"body":10,"coverUrl":11,"productScreenshots":12,"productLinks":13,"authorName":14,"authorUrl":15,"authorSubject":16,"category":17,"tags":22,"sourceLabel":35,"sourceName":36,"sourceUrl":37,"status":38,"seoTitle":7,"seoDescription":9,"canonicalUrl":35,"isFeatured":39,"sno":40,"sortOrder":41,"publishedAt":42,"updatedAt":43,"createdAt":43},"cd57a4b0-72ca-4f47-bb6c-c0bc50f0d215","article","CXL 内存池：一台服务器用不完的内存，能否分给另一台？","cxl-memory-pooling-and-sharing","CXL 内存池化让数据中心有机会按需分配扩展内存。本文区分 pooling 与 sharing，说明资源复用的价值，也提醒远端内存并不等同于本机 DRAM，收益取决于延迟、拓扑和管理软件。","一台服务器的内存暂时富余，另一台却正在为内存不足发愁。能否像调度网络带宽那样，把内存资源按需分配给不同主机？CXL（Compute Express Link）是一套面向处理器、加速器与内存设备的互连标准，其中的内存池化设想，就是让多台机器更灵活地使用一组可分配的内存资源。\n\n先区分两个经常混在一起的词：pooling（池化）与 sharing（共享）。池化通常意味着系统把某些内存容量分配给一台主机，之后也可能撤回并重新分配；在某个时刻，这块容量主要归一个主机使用。共享则强调多个主机能够同时访问同一资源，并需要处理一致性与协调问题。两者的目标和系统复杂度并不相同。\n\nCXL 提供主机与设备之间的高速互连及内存语义，使扩展内存不必只是传统意义上挂在远处的一块“磁盘”。在支持的系统架构里，内存设备可以通过 CXL 接入，资源管理软件再根据负载和策略进行分配。数据中心可据此把内存容量从闲置主机转给需求更高的工作负载，减少每台服务器为了应付峰值而永久多配内存的压力。\n\n但“连得上”不等于“和本机 DRAM 完全一样”。扩展内存的访问延迟、带宽、拓扑位置、设备容量以及错误恢复方式都可能不同。操作系统和应用是否把它当作普通内存使用，还是把它视作分层内存，也取决于平台软件栈。调度若没有考虑数据冷热，把频繁访问的数据放在更远的内存上，反而会拉低性能。\n\n内存池还需要一整套基础设施：支持 CXL 的处理器、交换设备与内存模块，固件和操作系统支持，以及负责分配、隔离与回收的管理软件。池化并非魔法式的“多机自动合并成一台超大电脑”。主机之间的安全边界、故障域、带宽争用和资源回收时机，都必须纳入设计。\n\n可以把它想成办公楼里的共享物料仓：各团队不必各自囤够全年峰值库存，而可以按需领用；但物料仓更远，领用需要规则，而且不能把别人正在使用的东西突然收走。CXL 内存池化的价值在于提高资源利用率，而不是消除本地与远端内存的差别。评估收益时，要看真实工作负载的延迟敏感程度、容量峰谷和整个系统的协调成本。\n\n来源：[Compute Express Link：CXL 3.0 White Paper](https:\u002F\u002Fcomputeexpresslink.org\u002Fwp-content\u002Fuploads\u002F2023\u002F12\u002FCXL_3.0_white-paper_FINAL-1.pdf)","\u002Fuploads\u002F2026-10-13\u002F4927fef3-e4ee-4731-94ba-9a41215d6da5.jpg",[],[],"Foundit","https:\u002F\u002Ffoundit.cn","foundit-ai-editorial",{"id":18,"name":19,"slug":20,"description":21},"6179d3b6-dc34-4483-9ded-3cd9f1b37a47","科普","abbreviation","介绍各领域新兴概念",[23,27,31],{"id":24,"name":25,"slug":26},"7c76bfc2-f80f-4ee0-a95d-27bd8708b434","技术","slug",{"id":28,"name":29,"slug":30},"8d35e42e-9f5b-4dd9-b30f-dc2c53d06096","硬件","hardware",{"id":32,"name":33,"slug":34},"4c2bbea6-eab7-40a8-8447-1de478ff7749","分析","analyse",null,"Compute Express Link Consortium","https:\u002F\u002Fcomputeexpresslink.org\u002Fwp-content\u002Fuploads\u002F2023\u002F12\u002FCXL_3.0_white-paper_FINAL-1.pdf","published",false,58,0,"2026-10-13T00:00:00.000Z","2026-10-13T02:13:41.400Z",[45,53,62],{"id":46,"type":6,"title":47,"slug":48,"summary":49,"coverUrl":50,"authorName":14,"sno":51,"publishedAt":42,"createdAt":52},"d4a1ef13-fa42-41f3-a9d2-635f47d801d0","RISC-V PMP：处理器怎样给不同特权级划出内存禁区？","riscv-pmp-physical-memory-protection","RISC-V PMP 由高特权级软件配置物理内存区域访问规则，并由处理器在较低权限代码访问时执行检查。本文区分 PMP 与虚拟内存，介绍它在固件和嵌入式安全中的作用及实现边界。","\u002Fuploads\u002F2026-10-13\u002Fecfdd306-c8d2-4bfa-984e-7f579fd86f61.jpg",41,"2026-10-13T02:13:39.212Z",{"id":54,"type":6,"title":55,"slug":56,"summary":57,"coverUrl":58,"authorName":14,"sno":59,"publishedAt":60,"createdAt":61},"56b94746-1551-4111-a1e9-6dc7e62e1604","RDMA：两台机器能否让网卡直接搬运注册内存？","rdma-registered-memory-and-remote-access","RDMA 让网卡在授权和注册内存范围内处理远端数据搬运。本文解释内存注册、访问权限与完成队列，避免把它误解为完全没有 CPU 参与。","\u002Fuploads\u002F2026-10-06\u002F86d526f9-9a46-46aa-8c81-1efd4aae1284.jpg",49,"2026-10-06T00:00:00.000Z","2026-10-06T07:51:32.204Z",{"id":63,"type":6,"title":64,"slug":65,"summary":66,"coverUrl":67,"authorName":14,"sno":68,"publishedAt":69,"createdAt":70},"67ba2085-7ee6-4379-9d0c-6372fed783ae","Intel TDX：云服务器管理员也看不到虚拟机里的数据吗？","intel-tdx-confidential-vm-explained","Intel TDX 通过处理器支持的隔离机制，将虚拟机运行环境划成受保护的 Trust Domain，降低宿主软件直接读取或篡改其内存的机会。本文说明机密计算解决什么信任问题，也谈远程证明、侧信道和来宾系统自身的风险。","\u002Fuploads\u002F2026-10-17\u002Ff8891699-ca7f-4edb-b64d-1b4db6421b0e.jpg",50,"2026-10-17T00:00:00.000Z","2026-10-17T03:08:39.718Z"]