Scale-up 互连从铜到光的必然演进随着大模型架构从 Dense 向 MoE 演进，并融合训练与推理一体化范式，系统对互连网络在节点规模、带宽密度、通信时延和运行稳定性等方面提出更高要求。高性能互连已成为制约超大规模 AI 集群效率与扩展能力的关键基础设施。当前主流 Scale-up 方案依赖机柜内铜缆互连。尽管铜互连在成本与时延上具备优势，但其传输距离受限，迫使系统采用极致高密度、强耦合的“...
Source LinkScale-up 互连从铜到光的必然演进随着大模型架构从 Dense 向 MoE 演进，并融合训练与推理一体化范式，系统对互连网络在节点规模、带宽密度、通信时延和运行稳定性等方面提出更高要求。高性能互连已成为制约超大规模 AI 集群效率与扩展能力的关键基础设施。当前主流 Scale-up 方案依赖机柜内铜缆互连。尽管铜互连在成本与时延上具备优势，但其传输距离受限，迫使系统采用极致高密度、强耦合的“...
Source LinkDisclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.