Introduction

About the Loki and Vali Clusters

The Loki and Vali clusters are high-performance computing (HPC) systems at Augusta University, designed to facilitate cutting-edge research by providing computational resources to faculty, students, and research teams. These clusters are optimized for large-scale parallel computing and data-intensive applications across multiple disciplines, including bioinformatics, physics, engineering, and artificial intelligence.

Both clusters are part of Augusta University’s HPC infrastructure, managed by Information Technology Services (ITS), and are equipped with centrally managed resources, storage, software environments, and technical support. The clusters support a range of computational workloads, including simulation, modeling, machine learning, and large-scale data analysis.

Loki and Vali Namesake

Loki

In Norse mythology, Loki is a legendary mischievous shape-shifter who walks the fine line between chaos and ingenuity. He is unpredictable, clever, and always pushing boundaries sometimes for good, often for trouble. The Loki cluster embodies this same spirit: a system built for flexibility, capable of adapting to a variety of research needs, from simulation to artificial intelligence. Like its namesake, Loki thrives in complex, dynamic environments where creativity and innovation drive discovery. The Loki cluster is designed to empower exploratory research, experimental workflows, and cutting-edge computational science.

Vali

On the other hand, Vali is a child of Loki, capable of utilizing mischief if needed, but favoring renewal hoping to restore balance. Unlike the cunning Loki, Vali is straightforward, relentless, and unwavering. His strength lies in precision and efficiency. The Vali cluster reflects these qualities: optimized for stability, high-throughput computing, and data-intensive tasks that demand reliability. The Vali cluster is designed to support and sustain classical and emerging computational research.






Loki and Vali Cluster Comparison

Feature

Loki Cluster

Vali Cluster

Compute Focus

High-performance computing for PhD researchers and PIs

Academic-focused cluster for students in specific courses

Scheduler

SLURM

SLURM

Storage

Lustre & NFS (/scratch, /work, /project)

NFS-based (/home, /software)

Compute Resources

Mixed CPU/GPU architecture with high-memory and AI/ML-optimized nodes

Primarily CPU-based but includes dedicated GPU nodes

Node Specifications

Nodes from 20 to 256-core, memory from 91GB to 1.5TB, multiple GPU nodes

Nodes with 32 to 64 cores, memory ranging from 128GB to 1TB, includes dedicated GPU nodes

Partitions (Queues)

  • cpu_high_mem_q, cpu_middle_mem_q, cpu_normal_q,

  • gpu_high_ai_q, gpu_middle_ai_q, gpu_normal_q, interactive_q

cpu_normal_q, cpu_middle_mem_q, cpu_high_mem_q, gpu_normal_q, interactive_q

Ideal Use Cases

  • AI/ML (17 apps): Deep learning and GPU-accelerated workflows

  • Bioinformatics (31 apps): Genomics and life science research

  • Compilers (3 apps): Code compilation and optimization

  • Data Science (5 apps): Data processing and analysis

  • Languages (7 apps): Programming language support

  • Tools (13 apps): General-purpose utilities

  • Libraries (4 apps): Core scientific libraries

  • Math (4 apps): Mathematical and statistical computing

  • Numerical Libraries (1 app): High-performance routines

  • Development (1 app): Software development support

  • AI/ML (1 app): Basic machine learning support

  • Computational Chemistry (1 app): Chemistry and materials modeling

  • Compilers (4 apps): Programming and instructional tools

  • Data Science (3 apps): Analytics and simulations

  • Languages (9 apps): Language and runtime environments

  • Tools (8 apps): Utilities and development helpers

  • Libraries (2 apps): Supporting scientific workflows

  • Math (6 apps): Core mathematical functions

  • Numerical Libraries (2 apps): Scientific computing support

  • Toolchain (1 app): Build and compile infrastructure

  • System (1 app): System-level utilities

  • MPI (4 apps): Parallel computing tools

  • Development (1 app): Application development resources

Acceptable Use

All users of the Loki and Vali clusters must comply with Augusta University’s Acceptable Use of Information Technology policy. This policy outlines the responsible use of the university’s IT resources, including:

  1. Compliance with Laws and Policies: Users must adhere to all applicable laws, regulations, and university policies when utilizing IT resources.

  2. Responsible Use: IT resources should be used in a manner that supports the university’s mission of education, research, and service.

  3. Security Measures: Users are responsible for safeguarding the integrity and security of IT resources, including protecting account credentials and ensuring devices are secure and free from viruses.

  4. Prohibited Activities: Unauthorized access, distribution of malicious software, and activities that interfere with the use of IT resources by others are strictly prohibited.

By adhering to these guidelines, users contribute to a secure and efficient computing environment that supports Augusta University’s academic and research endeavors.

Citations and Acknowledgements

To support the growth and continued funding of HPC resources, researchers are encouraged to acknowledge the use of the Loki and Vali clusters in their publications and presentations.

Acceptable citation examples include:

  • “The authors acknowledge the support of the Augusta University High Performance Computing Services (AUHPCS) for providing computational resources contributing to the results presented in this publication/report.”

If your work acknowledges the use of these clusters, please send a brief email to auhpcs_support@augusta.edu

Cluster Description for Grant Submissions

Augusta University’s High Performance Computing (HPC) cluster supports a broad range of research across disciplines such as bioinformatics, genomics, population sciences, mathematics, chemistry, physics, data science, AI, and digital forensics. It is also well-suited for complex simulations, modeling, visualization, and rendering workloads.

Key System Specifications

Compute Power:

  • 1032 CPU cores

  • 93,184 CUDA cores

  • 8,192 Tensor cores

  • 100.8 TFLOPS theoretical peak performance

Node Breakdown:

  • General, middle-memory, and high-memory CPU nodes

  • GPU nodes with Quadro RTX6000, Tesla T4, and A100 GPUs

Storage & Interconnects:

  • 240TB Lustre scratch

  • 1PB Lustre project storage

  • 15TB CPU RAM, 528GB GPU RAM

  • HDR InfiniBand and 10 Gb/s Ethernet

This infrastructure provides the performance and flexibility needed for cutting-edge, multidisciplinary research.

For further assistance, visit our Support Page or contact our team at auhpcs_support@augusta.edu.