University of Illinois at Urbana-Champaign
The Grainger College of Engineering
Engineering IT Shared Services

Campus Research Computing Options

 

Research computing is complex and multi-faceted, incorporating a variety of resources and models. These can include high-performance computing, stand-alone computation systems, cloud computing, and more. Our goal for this article is to show you what is available and hopefully assist in assessing what solution(s) will enable your work. 

 

Top Computational FAQs

Are you working on Machine Learning or Artificial Intelligence simulations?
If you are, you likely already know that you'll likely require a system with powerful graphics cards, what you may not know is that there are several computational resources available that you can leverage instead of purchasing your own system.

Are the simulations or tasks you running easy for a regular computer to complete but needs to be ran hundreds to hundreds of thousands of times?
This is known as a "perfectly parallel" workload and is generally a good candidate for High-Throughput Computing solutions. 

What software does your project require to run your simulations or tasks?
Many computational resources have certain software packages natively available, some have the ability to add additional software by request or onto your own profile. 

 

Additional Questions

There are other questions to consider when determining the appropriate computational resource for your project, you can find some of them below.
If you would like to meet with someone to discuss your needs please email the Research Technology Facilitator, Ethan Conner (ewconner@illinois.edu). 

  • Are you working with sensitive information, whether it be the input or output data, or the code that you are running?
  • Will your simulations be leveraging a Message Passing Interface (MPI)?
  • Do you require a computational resource that does not have a wall time (termination of job if not completed in X time)?
  • How versed are your project members in Linux distributions or command line?

 

Computing Resources (At-a-Glance)

A full-sized version of this matrix can be found here.

  Illinois Campus Cluster Delta Nightingale Radiant HOLL-I HAL Innovative Systems Lab HT Condor Pilot ACCESS AWS AZURE
TYPE OF RESOURCE High-Performance Computing High-Performance Computing Secure High-Performance Computing High-Performance Computing AI High-Performance Computing High-Performance AI/GPU Computing EXPERIMENTAL High-Throughput Computing Varied Research Computation Services Varied Varied
WEBSITE Campus Cluster Delta | NCSA Nightingale | NCSA Radiant | NCSA HOLL-I | NCSA HAL Cluster | NCSA Wiki Innovative Systems Lab HTC Wiki ACCESS AWS at Illinois Azure at Illinois
CONTACT INFO help@campuscluster.illinois.edu help@ncsa.illinois.edu kindr@ncsa.illinois.edu lapine@illinois.edu htc-help@illinois.edu ACCESS Support Page aws-support@illinois.edu azure-support@illinois.edu
COST OF USE Varies by selected investment option.
Engineering users can use a college-level investment at no cost. 		 FREE (with approved allocation) Fees and Services Rates HOLL-I | NCSA FREE (with approved allocation) FREE (with approved allocation)  FREE  FREE (with approved allocation) Varies by selected options Varies by selected options
HARDWARE ICCP Node Information System Architecture System Architecture Technical Specifications HOLL-I | NCSA  HAL Cluster ISL Information Retired Campus Cluster Nodes (various configurations) Varies by resource N/A N/A
PRIMARY USE CASES The Illinois Campus Cluster provides access to computing and data storage resources and frees you from the hassle of administering your own compute cluster. Delta is a GPU based compute platform for Artificial Inteligence and Machine Learning, Simulation, and Data Science Nightingale is a high-performance compute cluster with storage for researchers to interact with protected data and perform analysis. (HIPAA, ePHI, CUI, FERPA approved) Radiant provides researchers a flexible, elastic, and scalable computing solution, using cloud-like virtualization, while remaining on-site at NCSA. HOLL-I (Highly Optimized Logical Learning Instrument) is a new service at NCSA that offers public access to an extreme scale machine-learning capability through a Cerebras CS-2 WSE.  The Hardware-Accelerated Learning (HAL) cluster was designed to speed up deep learning research. It integrates the most recent advances in computing, storage and interconnectivity technologies to create a purpose-built shared-use system.  Cutting-edge computing environments for experimentation and evaluation Jobs that are more ideally suited to a High Throughput (loosely coupled) environment The available resources in ACCESS have different uses, a list of resources and subsequent information can be found here. A variety of use cases can be found here A variety of use cases can be found here
HOW TO GET STARTED Engineering users can take advantage of a college-level investment at no cost. Email engrit-help@illinois.edu Delta Allocations Accessing the System Requesting Resources or Starting a project help@ncsa.illinois.edu - Email requesting for access to HOLL-I  New User Guide for HAL System Contact 
Vlad Kindratenko		 Request Access ACCESS Allocations: Get started Request an AWS Account Request an Azure Account
USER GUIDES / ONLINE SUPPORT Campus Cluster User Guide Delta User Guide Nightingale User Guide User Documentation Directory User Documentation 
(Requires HOLL-I account/access)		 HAL cluster Experimental systems, so very little support available Minimal on-campus support,  but user community has a   wealth of information. ACCESS Knowledge Base Getting started with AWS Azure Tutorials from Microsoft
SOFTWARE PACKAGES a variety of core software and libraries available  Complete List Here Software Software Openstack Components

CSoft 1.5 for CS-2 WSE (more information in User Documentation)

 Software for HAL

Several technologies including test openstack, gpu, Kubernetes, etc

 HT Condor pilot shares many software packages with the Engineering Linux Environment Varies by resource, a list and subsequent info on resource providers can be found here. AWS Products (external site) Azure Products (external site)
DATA STORAGE OPTIONS 7GB Quota (additional storage available for purchase) File Systems Fees and Services: Storage Rates 1TB per project on TAIGA as part of access, additional storage available at standard TAIGA rates. Access to any existing TAIGA storage  N/A Varies on project scope usually 1-2TB per project  10GB initial quota Varies by resource Scalable to the project via EC2 and other AWS offerings Scalable to the project

If none of the computing resources listed above meet your needs, Engineering IT is available to help create and support a solution tailored to your project’s computing requirements. Email engrit-help@illinois.edu to get started.

 

Grainger College of Engineering's Computational Resources

Illinois Campus Cluster Program

The Grainger College of Engineering (GCoE) has an investment into the Illinois Campus Cluster Program (ICCP); this resource is available to faculty, staff, and students in the GCoE at no cost for both research and instructional purposes. (There are currently a total of 92 CPU and 15 GPU nodes, containing 2956 CPU cores, 26688GB RAM, and 60 GPUs.)

  • 76 CPU nodes - 2x Intel Xeon E5-2690 v3 (12-core), 1824 cores, 19456 GB RAM (256 GB ea.)
  • 8 CPU nodes - 2x Intel Xeon E5-2680 (14-core), 224 cores, 2048 GB RAM (256 GB ea.)
  • 8 CPU nodes, 2x Intel E5-2680 v4 (14-core), 224 cores, 512 GB RAM (64 GB ea.)
    • Dedicated to instructional use.
  • 7 GPU nodes, 2x Intel Xeon Gold 6148 (20-core), 280 cores, 1344 GB RAM (192 GB ea.), 14x V100 GPUs (2 ea.)
  • 5 GPU nodes, 1x AMD EPYC 7763 (64-core), 320 cores total, 2560 GB RAM (512GB ea.), 40x A10 GPUs (8 ea.)
  • 3 GPU nodes, 2x Intel Xeon E5-2680 (14-core), 84 cores, 768 GB RAM (192 GB ea.), 6x P100 GPUs
    • Dedicated to instructional use.

High-Throughput Condor Pilot

Additionally, GCoE has a 4,300-core High-Throughput Computing resource available based on HT Condor. Both of these resources can be used at zero cost to anyone affiliated with the college.

 

Grant and Proposal Support

In order to cite the Illinois Campus Cluster Program in your work, or to prepare a budget for your grant proposal, please take advantage of the prepared documentation and sample budget template information here: https://campuscluster.illinois.edu/resources/grant-proposal-support/

 

Training Resources