Core Insights - The recent WAIC showcased the rising trend of supernodes, with multiple companies, including Huawei and Shanghai Yidian, presenting their supernode solutions, indicating a growing interest in high-performance computing [1][2][4] Group 1: Supernode Technology - Supernodes are designed to address the challenges of large-scale computing clusters by integrating computing resources to enhance efficiency and support models with trillions of parameters [1][2] - The technology allows for improved performance even when individual chip manufacturing processes are limited, marking a significant trend in the industry [1][5] - Supernodes can be developed through two main approaches: scale-out (horizontal expansion) and scale-up (vertical expansion), optimizing communication bandwidth and latency within the nodes [3][4] Group 2: Market Dynamics - The share of domestic AI chips in AI servers is increasing, with projections indicating a drop in reliance on foreign chips from 63% to 49% this year [6] - Companies like Nvidia are still focusing on the Chinese market, indicating the competitive landscape remains intense [6] - Domestic manufacturers are exploring alternative strategies to compete with established players like Nvidia, including optimizing for specific applications such as AI inference [6][8] Group 3: Innovation in Chip Design - Some domestic chip manufacturers are adopting sparse computing techniques, which require less stringent manufacturing processes, allowing for broader applicability in various scenarios [7] - Companies are focusing on edge computing and AI inference, aiming to reduce costs and improve efficiency in specific applications [8] - The introduction of new chips, such as the Homa M50, highlights the industry's shift towards innovative solutions that leverage emerging technologies like in-memory computing [8]

Core Insights - The recent WAIC highlighted the growing interest in supernodes, with companies like Huawei, ZTE, and H3C showcasing their advancements in this technology [3][4][5] - Supernodes are essential for managing large-scale AI models, enabling efficient resource utilization and high-performance computing [3][4][5] - The shift from traditional AI servers to supernode architectures is driven by the increasing complexity and size of AI models, which now reach trillions of parameters [4][5][9] Group 1: Supernode Technology - Supernodes integrate computing resources to create low-latency, high-bandwidth computing entities, enhancing the efficiency of AI model training and inference [3][4] - The technology allows for performance improvements even when individual chip manufacturing processes are limited, making it a crucial development in the industry [4][9] - Companies are exploring both horizontal (scale out) and vertical (scale up) expansion strategies to optimize supernode performance [5][9] Group 2: Market Dynamics - Domestic AI chip manufacturers are increasing their market share in AI servers, with the proportion of externally sourced chips expected to drop from 63% to 49% this year [10] - Companies like墨芯人工智能 are adopting strategies that focus on specific AI applications, such as inference optimization, to compete with established players like NVIDIA [10][11] - The competitive landscape is shifting, with firms like云天励飞 and后摩智能 targeting niche markets in edge computing and AI inference, avoiding direct competition with larger chip manufacturers [11][12][13] Group 3: Technological Innovations - The introduction of optical interconnects in supernode technology is a significant advancement, providing high bandwidth and low latency for AI workloads [6][9] - Companies are developing solutions that leverage optical communication to enhance the performance of AI chip clusters, addressing the limitations of traditional electrical interconnects [6][9] - The focus on sparse computing techniques allows for lower manufacturing process requirements, enabling more efficient AI model computations [11][12]

来源：内容 编译自 IEEE ，谢谢。 如果将过多的铜线捆扎在一起，最终会耗尽空间——前提是它们不会先熔合在一起。人工智能数据 中心在GPU和内存之间传输数据的电子互连方面也面临着类似的限制。为了满足人工智能的海量 数据需求，业界正在转向更大尺寸、更多处理器的芯片，这意味着在机架内实现更密集、更长距离 的连接。初创公司正在展示 GPU 如何摆脱铜互连，用光纤链路取而代之。 光纤链路对数据中心来说并不陌生。它们使用可插拔收发器在机架之间传输数据，将电信号转换为 光信号。为了提高能源效率，"将光学元件集成到芯片封装中一直是一个梦想，"加州大学圣巴巴拉 分校电气工程教授克林特·肖( Clint Schow)表示。这就是共封装光学器件（CPO），科技巨头们正 在全力支持它。英伟达 (Nvidia) 最近宣布量产一款网络交换机，该交换机使用嵌入在与交换机同 一基板上的光子调制器。"这震惊了整个行业，"加州桑尼维尔初创公司Avicena的首席执行官巴迪 亚·佩泽什基 (Bardia Pezeshki) 表示。 哥伦比亚大学电气工程教授、Xscape Photonics联合创始人Keren Bergman解释说， Nvid ...