View source for Directory of Computing Resources for Statistical Research


'''Computing Resources''' are essential for modern statistical research and data science. The department, university, and national organizations provide several High-Performance Computing (HPC) clusters to scale up your workflows, run massive simulations, or train deep learning models. 

This page provides an overview of the computing clusters available to members of the department, detailing who manages them, hardware capabilities, and where to find setup instructions.

<div style="background-color: #e2eef9; border-left: 5px solid #005A9C; padding: 15px; margin-bottom: 20px; margin-top: 15px;">
'''Tip: Choosing the Right Cluster'''<br>
Start local before scaling up. Use the department-managed clusters for prototyping, data preparation, and medium-scale CPU jobs. For massive parallelism or extensive GPU needs that exceed department limits, you should apply for time on external clusters like PSC Bridges-2.
</div>

== Available Computing Clusters ==
'''Note: We are going to be decommissioning the old hydra servers. Please move your workflows to the leviathan servers!'''
{| class="wikitable" style="width: 100%; border-collapse: collapse; text-align: left; box-shadow: 0 2px 5px rgba(0,0,0,0.1);"
|- style="background-color: #f8f9fa; border-bottom: 2px solid #dee2e6;"
! style="padding: 12px; width: 10%;" | HPC Cluster Name
! style="padding: 12px; width: 15%;" | Management
! style="padding: 12px; width: 10%;" | Has GPUs?
! style="padding: 12px; width: 8%;" | Sharing Policy
! style="padding: 12px; width: 20%;" | How to access
! style="padding: 12px; width: 37%;" | Setup & Documentation Articles
|- style="border-bottom: 1px solid #dee2e6;"
| style="padding: 12px; font-weight: bold;" | Leviathans
| style="padding: 12px;" | Statistics Department
| style="padding: 12px;" | No
| style="padding: 12px;" | FCFS
| style="padding: 12px;" | Anytime direct ssh to server, e.g. <code>ssh [userid]@leviathan1.stat.cmu.edu</code>
| style="padding: 12px;" | 
<categorytree mode="pages" depth="1" hideroot="on">Department Servers</categorytree>
|- style="border-bottom: 1px solid #dee2e6;"
| style="padding: 12px; font-weight: bold;" | Ghidorahs
| style="padding: 12px;" | Statistics Department
| style="padding: 12px; color: #28a745; font-weight: bold;" | Yes
(3 nodes)
| style="padding: 12px;" | FCFS
| style="padding: 12px;" | Anytime direct ssh to server, e.g. <code>ssh [userid]@ghidorah1.stat.cmu.edu</code> 
| style="padding: 12px;" | 
<categorytree mode="pages" depth="1" hideroot="on">Department Servers</categorytree>
|- style="border-bottom: 1px solid #dee2e6; background-color: #fdfdfd;"
| style="padding: 12px; font-weight: bold;" | Wright Cluster
| style="padding: 12px;" | Collaboration between several CMU departments
| style="padding: 12px; color: #28a745; font-weight: bold;" | Yes
(3 nodes)
| style="padding: 12px;" | Slurm
| style="padding: 12px;" | See instructions [[Wright GPU Cluster|here]]
| style="padding: 12px;" | 
<categorytree mode="pages" depth="1" hideroot="on">Wright Cluster</categorytree>
|- style="border-bottom: 1px solid #dee2e6; background-color: #fdfdfd;"
| style="padding: 12px; font-weight: bold;" | PSC Clusters
| style="padding: 12px;" | External (Pittsburgh Supercomputing Center)
| style="padding: 12px; color: #28a745; font-weight: bold;" | Yes
(Many nodes)
| style="padding: 12px;" | Slurm
| style="padding: 12px;" | See instructions [[PSC|here]]
| style="padding: 12px;" | 
<categorytree mode="pages" depth="1" hideroot="on">PSC</categorytree>
|}

<div style="background-color: #fff3cd; border-left: 5px solid #ffc107; padding: 15px; margin-top: 20px;">
'''Note on Accounts:''' Department-managed clusters usually require you to request access via the department IT staff or faculty sponsor. External resources like PSC require writing a brief research proposal and creating an ACCESS account.
</div>

== Technical Specifications ==

The following table summarizes the hardware specifications per node for each resource type:

{| class="wikitable" style="width: 100%; text-align: left;"
|-
! Cluster
! Number of Nodes
! CPUs per Node
! RAM per Node
! GPUs per Node
! Notes
|-
| '''Leviathans'''
| 11
| 16 cores (AMD Ryzen 9 9950x @ 5.7GHz)
| 128GB (Nodes 1-6)<br>196GB (Nodes 7-11)
| 0
| 
|-
| '''Ghidorahs'''
| 3
| 24 cores (Intel Core i9-13900KF: 8 P-cores, 16 E-cores)
| 64GB
| 1 (NVIDIA RTX 4090 24GB)
| 
|-
| '''Wright Cluster'''
| 3
| 64 cores
| 1TB
| 8 (NVIDIA L40 48GB)
| See [https://www.cmu.edu/dietrich/statistics-datascience/resources/docs/wright-gpu-cluster-policy.pdf Wright GPU Cluster Policy - CMU]
|-
| '''PSC Clusters'''
| >500
| Varies
| Varies
| Varies
| See [https://www.psc.edu/resources/bridges-2/ Bridges-2 Nodes]
|}

== Software ==
The servers within Statistics & Data Science offer a variety of software. Sudo access, also referred to as the ability to install software, on department servers is not allowed. If a software package you would like to use is not listed below, please contact remarks@stat.cmu.edu.
{| class="wikitable"
|+
!
!Ghidorahs
!Leviathans
|-
|Python
|X
|X
|-
|R
|X
|X
|-
|Apptainer
|X
|X
|-
|Quarto
|
|X
|-
|Gurobi 
|
|X
|-
|Rstudio-Server
|
|X
|-
|Matlab
|
|X
|-
|Julia
|
|X
|}

== Other Resources ==
* [[Open OnDemand]]
* [[Open OnDemand Wright]]
* [[Monitoring with Grafana]]
* [https://util.stat.cmu.edu/grafana/dashboards/f/efud3z15idnuoa/ Grafana Monitor]