当前您的位置： 首页 > 英文站 > Research Support > Core Facilities > Cores > High Performance Computing
Research Support
High Performance Computing
High Performance Computing (HPC) has been designed and constructed to meet the computing needs across SJTU since 2018. It has become one of the largest and most complete computing cores among universities in China. We have built the top 3 power- ful π2.0 supercomputer system, the most powerful AI computing system and the largest cloud computing system in China’s universities. In addition to provide computing resources, we endeavor to offer custom development and performance optimization support based on user needs.
Director: Junwei Zhao
jCloud2.0
jCloud 2.0 is equipped with 12320 CPU cores, 124TB memory, 12PB storage capacity and 100G network. We provide stable and efficient cloud host, cloud storage, dedicated VPC network, cloud database, scientific research application and other services, and cloud solutions for data and business.
High Performance Computing π 2.0
High performance computing π 2.0, double precision theoretical peak performance of 2.1 PFlops, is the first supercomputing built with Intel Cascade Lake CPU, with 656 computing nodes and a total of 26000 CPU cores. The lustre storage system based on Intel full flash NVMe SSD is used for the first time, with a storage capacity of 10PB.
Artificial Intelligence Computing
The artificial intelligence computing core is composed of eight DGX-2, and the in-depth learning tensor compu- ting capacity can reach 16 PFLOPS; by carrying the NVIDIA NVSwitch innovative technology, the bandwidth between GPUs is as high as 2.4 TB/s; with the scalable architecture, the complexity and scale of the model are not limited by the limitations of the traditional architecture, and meets many complex AI challenges.
Contact
Junwei Zhao
Email: paul_zjw@sjtu.edu.cn
闵行基地：上海市闵行区东川路800号
瑞金基地：上海市黄浦区瑞金二路197号
COPYRIGHT © 上海交通大学转化医学研究院
