IDC Enviorment and Computing Devices
Air Cooling IDC Enviorment
Part of A100 GPUs
Part of Hxx GPU Module
Part of GPU/Storage Nodes
AI Supercomputing Platform
The research team of Professor Xiaolin Andy Li at the Hangzhou Institute of Medicine, Chinese Academy of Sciences, possesses extensive resources in AI computing and high-performance computing (HPC):
The research team currently operates over 30 units of CPU/GPU servers, all-flash/hybrid-flash storage systems, and core support equipment. The infrastructure primarily includes Hxxx compute nodes, Hxx compute nodes, A100 compute nodes, A40 compute nodes, A30 compute nodes, and RTX 3090 compute nodes, with the total number of GPUs exceeding 150. The high-performance SSD and hybrid flash storage systems provide a raw hardware capacity of 3PB. Furthermore, the cluster undergoes annual planned expansions to adequately meet the ongoing scientific research demands of the group.
The intelligent scheduling platform TensorStack significantly empowers our foundational infrastructure by refining resource allocation and scheduling to drastically improve hardware utilization. It seamlessly integrates critical functionalities including debug environments, job submission & workflow customization, and model/service deployment. To date, the platform has contributed to the production of over 30 high-quality research papers, all published in major journals and conferences. Furthermore, it facilitates the translation of research into applications, supporting the filing of more than 10 patent applications by the team (all approved by the National Patent Office). Through its robust computational support, TensorStack ensures the orderly development and version iteration of the team’s research product, ScholarClub
Supercomputing+TensorStack:
AI for Science, Supercharged by an Intelligent Scheduling Platform
