High-Performance Computing Cluster

Introduction

• To address the growing complexity of research and rapid advances in artificial intelligence, the Department of Geography has established a dedicated high-performance computing cluster - GEOG HPC. This resource provides professors, researchers, and students with powerful parallel computing and massive storage capacity, fully supporting teaching and research activities. 

• The GEOG HPC System features cutting-edge infrastructure: advanced intelligent computing engine, next-generation scale-out storage, and the high-speed interconnect, to perform complex simulations, computations, and data analysis. 

• The GEOG HPC System will serve as the Department's research powerhouse, enabling complex geographic modeling, high-resolution remote sensing processing, data-intensive spatial analytics, smart cities and sustainable development, empowering geographic science research and promoting interdisciplinary innovation.

System Diagram

Hardware Environment

Type	Configuration
Front-end Node	1 Node, 40 Physical CPU Cores, 256GB System Memory, 25GbE Network
Computer Node	12 Nodes, 1920 Physical CPU Cores, 11.5TB System Memory, 28 GPU Cards
Storage Node	12 Nodes, 8.5PiB Usable Capacity

System Interconnect

• 200Gb/s HDR Infiniband Network

• 100GbE RoCE Network

• 25GbE Campus Network

Software Environment

• Operating System: Rocky Linux (x86-64)

• HPC Software Stack: OpenHPC

• Job Scheduler: SLURM