Skip to content

Getting Started at the Center for High Performance Computing

Welcome to the Center for High Performance Computing (CHPC) at the University of Utah. The CHPC is the university’s home for research computing and data. Our vision is to support the ever-increasing—and increasingly diverse—computing needs of researchers. To this end, we provide computational resources, data storage and transfer services, virtual machines, resources for the analysis of protected health information, and advanced training and user support to researchers in our community. Whether you’re studying the structure and function of proteins, analyzing patterns in the formation of sea ice, or examining speech in audio recordings, the CHPC is here to help.

Our dedicated team of scientists and engineers, system administrators, network engineers, software developers, security professionals, and students is always willing to help if you have any questions or concerns. The best way to contact us is to email us at helpdesk@chpc.utah.edu. Please include as much information as possible when contacting us so we can provide timely support. Additionally, you are welcome to us send feedback on this guide, especially if you have suggestions for improvements.

  This guide is for users new to the CHPC and is intended to provide an overview of our systems and resources. It is not meant to eliminate the need to read over our other documentation or policies, or to replace the documentation about specific resources.

On this page

The table of contents requires JavaScript to load.

Accounts

All users need a CHPC account. We describe the process for getting a CHPC account on our Accounts page. Accounts are available to researchers working at the University of Utah, faculty members from Utah institutions of higher education, and collaborators from other institutions. If you do not have an active University of Utah account (a uNID, like u0123456, and the ability to log in to services like Campus Information Services), you must first be added as an affiliate; this process is also described on the Accounts page. We also support accounts for classes at the request of instructors.

Once you have a CHPC account, you can access our resources in the using your campus credentials. If you need to change your password, you can do so through Campus Information Services.

If you require access to the Protected Environment (PE), there is an additional application process, described on the PE page. This is required because the CHPC must confirm that you have access to individual projects in the PE.

Training and support

We understand that our users come from different academic backgrounds and have different levels of experience with research computing. Regardless of whether you’re an undergraduate student participating in your first research experience or a faculty member already familiar with computational research—or if you find yourself anywhere in between—we’re here to support you.

Each semester, the CHPC’s User Services team gives a series of presentations. We have designed these presentations to help you learn about CHPC resources and other topics that may help with your research, including navigating the command line, writing scripts, and using the scheduler to submit your work to computational resources. If you are new to computational research, we recommend attending our introductory presentations to learn the fundamentals.

  We hold all presentations remotely through videoconferencing software; there is no cost to attend and registration is generally not required.

The scientists and engineers on the User Services team have experience performing computational research and are happy to help if you have questions about using CHPC resources for your research. Please contact us at helpdesk@chpc.utah.edu if you have any questions or concerns.

Resources

The CHPC has several computing environments. Reseachers may use the General Environment (GE) for research that does not involve sensitive data. We also mantain a Protected Environment (PE) for researchers with sponsored research projects and work that is sensitive in nature, including projects involving Protected Health Information (PHI). Both the GE and PE include high-performance computing clusters, Windows resources, virtual machines, and several storage options.

  If you are connecting to resources in the PE, you must be on the campus network. If you’re connecting from off campus, you will need to use the university’s VPN service. Additionally, you must use multi-factor authentication to connect to resources in the PE.

Clusters

The largest computational resources at the CHPC are the high-performance computing clusters. These are groups of computers (the computers are often called servers or nodes) that are used to perform calculations. A supercomputer is not one individual computer, but a group of connected computers, which is where the name “cluster” comes from. Researchers run their computations on the clusters by requesting resources—CPU cores, memory, and GPUs—through the scheduler.

At the CHPC, the clusters are operated on the condominium model, where research groups can purchase nodes that are added to a cluster. When a research group is not using its computational resources, they are made available to the community, though the owner group always has priority over guest users. This model helps ensure that nodes are not left idle and increases overall throughput and accessibility. Additionally, the CHPC owns nodes and provides priority access to department-owned nodes through allocations. Allocations are required to use CHPC-owned nodes with priority on recent clusters; they are not required on older systems (see the cluster user guides for details). Research groups may request an allocation by submitting a proposal, which is reviewed by a committee of researchers. Small allocations are available to new principal investigators; these are reviewed internally by CHPC staff. When you interact with the scheduler, you must specify an account, partition, and quality of service (QoS), which together describe the resources you are requesting (such as CHPC-owned nodes with an allocation or owner nodes).

CHPC-owned (“general”) nodes
Allocated clusters*

Groups with allocations have priority access to the general nodes of allocated clusters.

All users without allocation have “freecycle” (guest) access to general nodes. Freecycle jobs can be preempted (canceled) by jobs from a group with allocation.

Unallocated clusters*

All users can submit jobs to the general nodes of unallocated clusters. These are not subject to preemption.

*See the cluster user guides for details

Research group-owned (“owner”) nodes
All clusters

Research groups that have purchased nodes have priority access to the nodes owned by the group.

All users have guest access to owner nodes. Guest jobs on owner nodes can be preempted (canceled) by jobs from the group that owns the nodes.

Accessing the clusters

The nodes on the clusters run a Linux-based operating system. For many users who have worked with computers running Windows or macOS, it can seem like an unfamiliar environment. The CHPC presents an introductory Linux course, which can help you learn to navigate the system, edit your files, and run programs. We recommend attending this presentation series if you have not used Linux before.

There are a few options for connecting to the clusters. If you are familiar with the command line, you can use SSH to connect to a login node. An SSH client is installed by default on macOS and most Linux distributions. A native SSH client is also available on Windows (provided you are using Windows 10 or newer), though it may first need to be enabled in your system settings.

You may also use FastX or the Open OnDemand web portal to connect. You can use FastX, which includes support for graphical software, in a web browser or install it on your computer. Open OnDemand, which works in your web browser, includes many commonly used graphical applications; this is a great way to interact with the CHPC’s resources, especially if you are not yet comfortable with the command line. We describe access methods in further detail on the Accessing CHPC Resources page.

Using the scheduler

Computational work on the clusters is managed by the scheduler, Slurm. Researchers request resources (part of a node or multiple nodes) through the scheduler, which allocates them as they become available. If you will be performing computations on the CHPC’s clusters, you will need to learn how to use the scheduler. CHPC staff offer presentations on the scheduler and have produced training videos to make it more accessible.

  When you first connect to a cluster, you are connecting to a shared resource called a login node or an interactive node. Many users connect to the same login nodes, and you should not run large computations—which might interfere with others’ work—on login nodes. You must use the scheduler to request dedicated computational resources (also called compute nodes) to run large computations.

You have some flexibility with how you interact with the scheduler. Many researchers write scripts to submit their computations. These include instructions for the scheduler regarding the resources required for the job. Scripts allow researchers to submit work without needing to run individual commands. Other researchers request resources from the command line and run commands directly from the compute nodes. Others use Open OnDemand to request resources from the scheduler and run graphical software like Jupyter from a web browser.

 

Why do I need to learn to use all this software?

We know that learning to use new software can be challenging, especially if you—like many new CHPC users—don’t have a background in research computing. It can take time and quite a bit of effort to learn how to work effectively in a high-performance computing environment. We are here to help! Please let us know if you have any questions about working with our systems.

We believe that many of the foundational skills we teach in our introductory presentations, such as navigating the command line and writing shell scripts, are invaluable skills for researchers. Additionally, the CHPC’s systems use industry-standard software and the concepts you will learn while working with our systems are transferrable. According to the TOP500 project, every one of the top 500 supercomputers in the world is running Linux. Additionally, according to SchedMD, the company that develops and supports Slurm, “Slurm is used by approximately 65% of supercomputers” among the top 500. The software we use at the CHPC is also common to many other universities and research institutions. The skills you will learn while working with the CHPC’s resources will be useful throughout your research career.

Windows servers

While the clusters are the largest computational resources available at the CHPC, we also offer access to Windows servers in the General Environment and Protected Environment. These are shared, large-memory nodes with popular software like SAS, Stata, MATLAB, and Mathematica. The Windows servers are most commonly used for large-memory statistical analyses. Please contact us if you have other Windows needs.

Accessing the Windows servers

You can connect to Windows resources using the remote desktop protocol (RDP). Clients for RDP are available on all major operating systems. Instructions for connecting are given on the Remote Desktop page.

  You must be on the campus network to connect to Windows servers. If you’re connecting from off campus, you will need to use the university’s VPN service.

Storage services

The CHPC is the ideal home for your research data. We store petabytes of research data and offer several storage solutions. For more information, please see the Storage page. Storage options at the CHPC include

  • Home directories. By default, every CHPC user has 50 GB of home directory space. Research groups may purchase home directory space, which includes backups.
      Home directories are not backed up by default. Purchased home directory space includes backups.
  • Group space. Research groups may purchase storage. Group storage may or may not include backups; this affects the purchase price.
  • Archive storage. We offer an archive storage solution based on object storage. Users can move data to and from archive storage as needed. This space is not mounted on CHPC systems; it is a standalone system accessible through the S3 API.
  • Scratch. Scratch space is available to all users; there is no cost to use it. Scratch space is used for the storage of intermediate files while a job is running. Files are removed if they have not been accessed within 60 days. The scratch space is not backed up.
      Scratch space is for temporary file storage only. Files in scratch spaces are automatically deleted after a period of inactivity. Scratch space is not backed up.
  • Temporary files. There are several options for temporary file storage. Slurm jobs have access to the local disk on the nodes on which they are running. This space is not intended for long-term storage and is not backed up.

Data transfer

There are several options for transferring your data to and from CHPC resources. We describe these on the Data Transfer page. Options include mounting filesystems on your local computer, using graphical or command-line tools to copy files, using dedicated hardware and parallel transfer tools for large transfers, and providing a guest transfer account to a colleague.

Responsibilities

The CHPC is an important but limited resource for researchers at institutions of higher education in Utah. It is our responsibility to safeguard our facilities and ensure they are being used properly. We cannot do this job alone; we depend on your assistance. Responsible use and cooperation help us maximize availability for you and other researchers. When signing up for an account, all users agree that

In obtaining this account I agree to use the Center for High Performance Computing resources solely for the purposes connected with my University of Utah or Utah State University affiliation(s) and agree not to allow my access to be used in a manner which could permit unauthorized use of the computer. During the use of this account, certain proprietary software may be made available. Availability of said software is for my use only and may not be copied to any other machine(s) or made available to any other person(s).

If the CHPC determines that you have not used its resources in an appropriate fashion, you could lose your account, your current allocation, and your ability to qualify for future allocations.

We all have a part to play when it comes to ensuring CHPC resources remain available to all researchers in our community. Please help us by protecting your account, abiding by software license restrictions, and using our resources only for research and/or classroom applications.

Last Updated: 11/12/24