推论优化系统可拓展美超微的领先GPU服务器组合，为客户提供前所未有的人工智能推论、培训与深度学习解决方案选择，包括单根、双根、纵向扩展和横向扩展设计

Designed for the Next-Generation of AI, New HGX-2 System with 16 Tesla V100 GPUs and NVSwitch leverages over 80,000 Cuda Cores to deliver unmatched performance for deep learning and compute workloads

拥有超过 0.5PB闪存的1U服务器提供超过1000万IOPS性能，使其成为IOPS密集型和低延迟应用、数据处理和超融合基础架构(HCI)的理想解决方案

通过认证的超融合BigTwin服务器支持最多中央处理器和扩展卡及最大内存和NVMe存储，将每台虚拟机的性能提升10倍，成本节省9倍

该片描述了美超微、英特尔和NASA的使命，即借助节能和减少电子垃圾，最大限度地减少当今数据中心对于环境的影响

On Display at the Flash Memory Summit, new family of Latency, Bandwidth, Density and Thermal Optimized solutions supporting new EDSFF form factor allow rack level pooling of NVMe resources to reduce costs and improve efficiency and utilization

Resource-Saving Architecture optimizes power, cooling, shared resources and refresh cycle costs to Save Millions in TCO and reduce Total Cost for the Environment (TCE)

在2018年RSA新加坡大会上，美超微将展示专为软件定义网络(SDN)、网络功能虚拟化(NFV)、软件定义广域网(SD-WAN)和虚拟客户端设备(vCPE)应用而设计的紧凑、密集、低功耗和低延迟系统

全闪存 NVMe 和持久化内存支持服务器和存储的市场先行者为 SNIA 带来了丰富的系统级设计和开发经验

First-to-Market Solutions for Content Creation & Embedded/IoT Applications that require a Cost-Optimized, Multi-Core, Power Efficient, & High-Speed Processing Platform