三星和AMD正在共同开发先进的AI内存,其中HBM4在量产中,而 HBM4E针对NVIDIA的AI平台. Samsung and AMD are co-developing advanced AI memory, with HBM4 in mass production and HBM4E targeting NVIDIA’s AI platform.

三星和AMD扩大了他们的合作伙伴关系,共同开发下一代AI内存,包括HBM4用于AMD的MI455X GPU以及EPYC CPU的高级DDR5,现在HBM 4正在批量生产中,提供高达13Gbps速度和3.3TB/s带宽. Samsung and AMD have expanded their partnership to co-develop next-generation AI memory, including HBM4 for AMD’s MI455X GPU and advanced DDR5 for EPYC CPUs, with HBM4 now in mass production offering up to 13 Gbps speeds and 3.3 TB/s bandwidth. 三星还推出了HBM4E,其目标是16Gbps和4.0TB/s的带宽,该设备为NVIDIA的Vera Rubin AI平台设计. Samsung also unveiled HBM4E targeting 16 Gbps and 4.0 TB/s bandwidth, designed for NVIDIA’s Vera Rubin AI platform. 两家公司计划探索造协作,并优化人工智能工作负载的完整计算堆. The companies plan to explore foundry collaborations and optimize the full computing stack for AI workloads. 三星展示了新的存储和存储技术,包括LPDDR6 DRAM和PM9E3/PM9 E1 NAND,旨在实现设备上的人工智能和个人人工智能超级计算,同时通过AI驱动的数字双胞胎推进制造. Samsung showcased new memory and storage technologies, including LPDDR6 DRAM and PM9E3/PM9E1 NAND, aimed at on-device AI and personal AI supercomputing, while advancing manufacturing through AI-driven digital twins.