High Performance Computing Cluster (HPC3) Launched

william Wan, head of teaching technologies

The New HPC3 Cluster 

HPC3 is the latest high performance computing (HPC) cluster implemented and maintained by ITSC with servers contributed by faculty members. It is run based on the "community cluster" model, with priority given to contributors while others can share when resources are idle, resulting in best utilization of resources and protection of individual contributors' rights to use. The HPC platform utilizes a large number of CPU and GPU processors to run computational jobs in parallel and solves multiple research problems at the same time. The objective of the centralized platform is to maximize the utilization and effectiveness of the cluster by providing centralized support and maintenance of hardware, software and user & quota management.

Applicable Research Areas

Combining the high speed interconnect with massive CPU / GPU cores, the HPC3 cluster is being used for multiple disciplines including but not limited to : Quantum Chemistry, Chemical Pathology, Metagenomics, Time Series Analysis, Computational Fluid Dynamics, Earth System Science, AI, Machine Learning & Data Science etc. It may help shorten the computational time from years to hours.

Hardware Configurations

The initial setup of the HPC3 cluster consists of 100 sets of CPU nodes, 3 sets of large memory nodes (1.5TB each) and 5 sets of GPU nodes (each with 8*RTX-2080Ti) connected by InfiniBand at 100Gbps. The mixture of different hardware suits different computational research needs.

High-Speed Research & Education Network

Research & education (R&E) network connectivity is crucial for researchers to transfer and collect data from different areas in the world to conduct their research. HKUST together with other universities in HK are connected with the global exchange network which speed up the data transfer capacity between HK and all over the world.