计算机科学
互连
钥匙(锁)
嵌入式系统
带宽(计算)
炸薯条
延迟(音频)
连接(主束)
计算机体系结构
工程类
计算机网络
操作系统
电信
结构工程
作者
Ying Wei,Yi Chieh Huang,Haiming Tang,Nithya Sankaran,Ish Chadha,Dai Dai,Olakanmi Oluwole,Vishnu Balan,Edward Lee
标识
DOI:10.1109/isscc42615.2023.10067395
摘要
NVLink-C2C is the enabler for Nvidia's Grace-Hopper and Grace Superchip systems, with 900GB/s link between Grace and Hopper, or between two Grace chips. The connection provides a unified, cache-coherent memory address space that combines system and HBM GPU memories for simplified programmability. This coherent, high-bandwidth, low-power, low latency connection between CPU and GPUs is key to accelerating the most complex AI and HPC workloads.
科研通智能强力驱动
Strongly Powered by AbleSci AI