High Performance Computing (HPC); HyPER C3 – Community Cloud HPC Cluster  

HyPER C3 provides a community high-performance computing (HPC) cluster to support compute-intensive scholarship at Emory. HyPER C3 is a shared resource available to faculty groups within Emory's AI.Humanity Initiative. Dr. van Assen as AI.Humanity Faculty have full access to the cluster, extending to the entire lab.   

  The cluster offers:  

• A team of HPC engineers, coordinated with central and school IT teams, to manage daily cluster administrations and help investigators with onboarding, training on using the cloud cluster, advice on leveraging the resources for projects, and designing solutions sets.   

• Scientific compute nodes: A state-of-the-art selection of high-performance compute nodes available in AWS, including general-purpose CPUs and GPU-based processors, high-end GPU instances and highmemory nodes, served through a head/logon node and several partitions (or queues) suitable for a variety of scientific computing use cases.   

• Compute-optimized storage: A tiered storage configuration, including Lustre or comparable fast-access drives, active and temporary file storage for immediate computation purposes at an individual or group permission level, and pathways to long-term / archival storage.   

• User-friendly interface: Fronted by Slurm (AWS ParallelCluster) and other common HPC tools, and authenticated with Emory credentials from both on and off-campus (through VPN), offering a familiar user experience with proven functionality and secure operational performance.  

• Scientific software: Access to popular software packages, specialized AI/ML resources, cluster/parallel computing platforms, and container deployment technology.   

• Regulatory compliance: Implemented with technical and process controls and policies compliant with HIPAA and NIST 800-171 standards, thereby suitable for the analysis of sensitive data such as ePHI (with teams following the appropriate rules of behavior) and offering a robust level of protection and disaster recovery.  

• As part of the account creation process, the service provisions a set of virtual private cloud environments that provide geographic redundancy for the data and services. Through this automated process, the environments are set to an approved configuration with minimal risk of variations. With all steps automated, there is a greater level of assurance around firewall rules, network configurations, and environment setups.  

In addition to these Emory-specific customizations, the Emory research team benefits from: (a) Amazon’s elasticity, providing investigators with the opportunity to scale up or scale down their infrastructure based on needs. As such, the team is not paying for unused or idle infrastructure; (b) ability to tap into Amazon’s technology optimized for specific research workloads, such as high memory computing cores and high-speed solid-state drives (SSD); and (c) quickly spin up computing resources within minutes to increase the time for investigators and their team to conduct their science.  

In support of this service, Emory has dedicated technical resources to help researchers and their teams. Emory has purchased AWS Enterprise support to provide 24x7 support for the service and is also sponsoring training and leading a cloud community of practice, which includes participation from investigators, IT organizations, and scientific cores.  

 Page In Progress