Facilities Statement
The Advanced Research Computing Center (ARCC) located at the University of Wyoming (UW) was established in 2010 to serve as a unified research support group for students, staff and faculty. In 2018 ARCC was moved from Infromation Technology to the Office of Research and Economic Development to better support research activities. All ARCC operations promote the growth of UW’s research and education activities. Resources provided by the department include high performance computing (HPC), large research data storage, and user training and consultation.
Data Center and Infrastructure
The division of Information Technology opened it’s 6,000 sqft data center in the winter of 2008. Computational resources are housed in liquid-cooled enclosures and includes highly redundant infrastructure for power, cooling, and security. Hardware is interconnected using InfiniBand and high-speed ethernet.
Research Computing Systems
Specialized ARCC staff members perform on-site administration, maintenance and support on all ARCC hosted systems and research support resources. This includes UW’s primary HPC cluster, Medicinebow and several other clusters and specialized research support systems.
Medicinebow HPC Cluster Configuration
Medicinebow is a x86_64 based HPC running RHEL 9. Job scheduling configuration is based on the condo model which encourages researchers to purchase nodes for use in our HPC environment granting priority and predictable access. Medicinebow is open to all facets of research. ARCC also supports several specialty clusters and research services used by UW research organizations. Medicinebow leverages fair-share mechanisms to distribute computational workloads between varied research projects using weighted scheduling parameters (job size, age, etc.) to maximize efficient cluster utilization. Medicinebow is backed by VAST performance data storage platform hosting over 3 PB of high-performance storage.
The Medicinebow cluster hardware is heterogeneous, comprising several specialty partitions to address a wide variety of scientific pipelines and workloads. The entire cluster hosts over 300 nodes including specialized nodes for mass memory computation (4TB RAM), and a wide variety of GPU nodes. The system utilizes SLURM workload manager and LMOD environment modules to provide a robust and flexible user experience. Medicinebow supports a wide range of compilers (GNU, Intel oneAPI, and NVIDIA HPC SDK) and containerization frameworks.
MedicineBow Hardware Table
| Partitions | # Nodes |
Cores / Node |
RAM / Node |
GPU Type |
GPU(s) / Node |
Processor |
TensorCores / GPU | CUDACores / GPU |
|---|---|---|---|---|---|---|---|---|
| mb | 25 |
96 | 1024G | N/A | N/A | 2x 4-Core / 96-Thread 4th Gen AMD EPYC 9454 | N/A | N/A |
| mb-a30 | 8 |
96 | 768G | A30 | 8 | 2x 4-Core / 96-Thread 4th Gen AMD EPYC 9454 | 224TC/GPU | 3804 FP32 CUDA/GPU |
| mb-l40s | 4 |
96 | 764G | L40S | 8 | 2x 4-Core / 96-Thread 4th Gen AMD EPYC 9454 | ||
| mb-h100 | 6 |
96 | 764G | H100 | 8 | 2x 4-Core / 96 Thread 4th Gen AMD EPYC 9454 | ||
| mb-A6000 | 1 |
64 | 1024G | A6000 | 4 | AMD | ||
| beartooth | 2 |
56 | 256G | N/A | N/A | Intel Icelake | N/A | N/A |
| beartooth | 6 |
56 | 515G | N/A | N/A | Intel Icelake | N/A | N/A |
| beartooth | 8 |
56 | 1024G | N/A | N/A | Intel Icelake | N/A | N/A |
| beartooth-gpu | 4 |
56 | 1024G | A30 | 4 | Intel Icelake | ||
| teton | 175 |
32 | 128G | N/A | N/A | Intel Broadwell | N/A | N/A |
| teton | 56 |
40 | 192 or 768G | N/A | N/A | Intel Cascade Lake | N/A | N/A |
| teton | 8 |
43 | 1024G | N/A | N/A | Intel Broadwell | N/A | N/A |
| teton | 2 |
48 | 4096G | N/A | N/A | AMD EPYC | N/A | N/A |
| teton-gpu | 2 |
32 | 512G | T4 | 4 | Intel Icelake | ||
| wildiris | 12 |
GannettPeak HPC Cluster Configuration
The GannettPeak Compute Environment (Advanced Infrastructure to Accelerate Impact of AI through Applications and Innovation for Wyoming) is a high-performance computing (HPC) cluster consisting of NVIDIA’s Hopper GPUs with ARM CPUs onto a single chip with fast memory interconnect. This unique architecture addresses critical computing challenges related to data movement and energy consumption through the Grace Hopper GH200 architecture, which integrates high-bandwidth coherent data transfers between CPU and GPU. The cluster was funded through an award from the NSF. This cluster is made of 2 login nodes, 12 single GPU Gracehopper Nodes, and 24 dual GPU Gracehopper Nodes. Nodes are named based on usage and compute components.
GannettPeak Hardware Table and Configuration
| Hostnames | Usage | Chip GPU | RAM |
|---|---|---|---|
| ai4wy-log1, ai4wy-log2 | Login Nodes | N/A | N/A |
| ai4wy-101 through ai4wy-112 | Compute Nodes | Single GH200 Superchip including 1 NVIDIA H100 GPU & 72 ARM Neoverse V2 Cores | 480 GB ECC On-Board LPDDR5X Memory and 144GB ECC HBM3 |
| ai4wy-201 through ai4wy-224 | Compute Nodes | Dual GH200 Superchip including 2 NVIDIA H100 GPU & 144 ARM Neoverse V2 Cores | 480 GB ECC On-Board LPDDR5X Memory and 144GB ECC HBM3 (144GB per GPU) |
High Performance Data storage
Our facility houses two main research storage resources in addition to housing data for direct use within Medicinebow. Both are available to all researchers at UW for both short- and long-term archive data storage. The first is data-Alcova, a peta-scale capable system. Both Medicinebow and the data-Alcova file storage use the VAST data platform, a highly scalable, in-line, block-based data storage service, built for performance. Connected to the UWyo network, this data storage offers blazing-fast transfer speeds starting at 100 GB/s. Data-Alcova also facilitates the publication of datasets curated by UW Libraries.
S3 is ARCC’s other primary research storage resource. Inferred by it's name, ARCC's S3 is ceph object storage (basically a cloud-like storage platform for cloud computing and large data) that provides it's users with a low-cost storage option. The S3 service also provides additional functionality to allow access to wide scale of large datasets for web-based applications.
All ARCC hosted data storage services support GridFTP via Globus data transfer servers connected to the UW Science DMZ (100 Gbps Internet2 link). Data-Alcova storage allows for SMB/CIFS and NFS, and MedicineBow storage faciliates access over NFS to make it for researchers to transfer data from their daily work environment.
Storage Services and Accessibility
| Service | Accessible using: | SMB | CIFS | NFS | Globus |
|---|---|---|---|---|---|
| MedicineBow | x | ||||
| Data-Alcova | x | x | x | x | |
| S3 (Pathfinder) | x |
Personnel and Support
The University of Wyoming is served by UW ARCC (Advanced Research Computing Center) and Central Information Technology (UWIT). ARCC’s team consists of 8 FTEs and reports to the Faculty Director for Computing Resources. UWIT’s staff consists of 95 FTEs across multiple service areas including desktop support, networking, security, enterprise systems, and application development and support. In addition to computational resources, UW ARCC personnel focus on IT services directly supporting research and UW researchers. Members of our team assists researchers by offering guidance, consultation and training for any users - both new and experienced – who may seek help incorporating HPC and research specific technologies in their research pipeline. UW ARCC works in collaboration with the UW Libraries Digital Scholarship Center to facilitate training towards the growth in understanding and use of HPC across UW and throughout our state.
Downloadable Facilities Document
Researchers who need a downloadable facilities document to add to grant proposals or other research documents may download a PDF copy of our facilities statement