Warehouse-Scale Video Acceleration

IEEE Micro(2022)

引用 1|浏览16
暂无评分
摘要
Video processing is foundational to a spectrum of important workloads: video sharing, video conferencing, photos/video archival, virtual/augmented reality, cloud gaming, and live streaming. The exponential growth in these workloads, coincident with a slowing Moore’s law, poses significant challenges to deliver more computing at higher efficiencies. In this article, we present the design of a new accelerator—the video coding unit—targeted at warehouse-scale (cloud) video transcoding. Our deployment at scale in Google, the first of its kind, has demonstrated significant benefits on several products (YouTube, Meet, Stadia, Photos, etc.) serving billions of users.
更多
查看译文
关键词
slowing Moore law,live streaming,virtual-augmented reality,photos-video archival,cloud gaming,video conferencing,video sharing,video processing,warehouse-scale video acceleration,warehouse-scale video transcoding,video coding unit
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要