Tempest Research Cluster
Overview
Tempest is a high performance computing (HPC) research cluster at MSU and the largest supercomputer in Montana. Tempest features powerful high-memory compute systems, GPUs, a fast storage system, support for long-running workloads, and an interactive web interface with graphical desktops, RStudio, JupyterLab, file transfer, and easy workload submission functionality . RCI provides training and full support for the Tempest HPC system. Tempest is available via a buy-in model with a free access tier for all MSU researchers and instructors.
Tempest accounts can be requested by completing the New User Request Survey
Quickstart and Documentation
Getting started in the Terminal
Installing software on Tempest
Domain-specific workflows and documentation
Technical Specifications
Tempest is an HPC cluster consisting of EPYC X86 CPU and Nvidia GPU nodes. In total, Tempest currently has 12,264 logical CPU cores, 55.9TB of ECC memory, 26 Nvidia A40 GPUs, 14 Nvidia A100 GPUs, and 12 Nvidia H100 GPUs. For storage, Tempest currently has 600TB of all-flash high-speed storage. Tempest components are interconnected by 100/50/25Gb Ethernet with RDMA.
If you would like to see a more detailed description of cluster resources, please see our technical description of computational resources.
If you would like to add a section about RCI into your next grant, feel free to use our template cluster description for research grants.
Getting Access
Tempest has two tiers of access: a free-access tier, and a buy-in tier
Free Access Tier
Each PI or class at MSU is eligible for priority access to 256 CPUs, 1024GB of memory, and 2 GPUs (of which up to 1 can be an A100 GPU) for MSU research or instruction. In addition to this limited priority acces, all MSU researchers also recieve fairshare access to a portion of Tempest and premtible access to any idle resoruces. Availability of free tier resources is subject to change.
Buy-In Tier
Those who buy-in to Tempest receive resource allocations in addition to the free tier allocation. Buy-in supports the maintenance of the cluster. Each buy-in group receives absolute priority access to a resource allocation commensurate to their level of contribution in addition to the free access tier. Contributors also have the ability to "burst" priority workloads to 200% of their allocation (up to a $30,000 value in additional resources). Additionally, all Tempest users are able to run unlimited "unsafe" and "legacy" jobs on any resources not being used for priority jobs.
|
Resource |
$5000 |
$10000 |
$15000 |
$20000 |
$25000 |
$30000 |
|
CPUS |
85 |
170 |
255 |
340 |
425 |
510 |
|
A40 GPU |
1 |
2 |
3 |
4 |
5 |
6 |
|
Memory |
340 Gb |
680 Gb |
1020 Gb |
1360 Gb |
1700 Gb |
2040 Gb |
Contributors also have the option to optimize their buy-in for either A40 or A100 GPUs. Those including tempest funding in grant proposals should list it as a line item under contracted services in the grant budget.
Over $30,000 resource allocation rates differ.
To learn more about buying-in to Tempest, email [email protected].
Citation
All Tempest users are requested to cite Tempest and RCI in publications resulting from research involving computational work performed on the system. Please use the following for citation purposes when referring to the use of Tempest computing resources in your research:
Computational efforts were performed on the Tempest High Performance Computing System, operated and supported by University Information Technology Research Cyberinfrastructure at Montana State University.