Welcome to OSC! If you are new to supercomputing, new to OSC, or simply interested in getting an account (if you don't already have one), we have some resources to help you.
Welcome to OSC! This document provides some useful links to get you started.
If you are new to supercomputing, or to OSC, please visit http://www.osc.edu/supercomputing/getting-started for information about how our supercomputers are configured, and some guidance about appropriate usage of the various resources.
Logging in to OSC resources is primarily accomplished via SSH (or “secure shell”). This is a method of getting a command line interface on the HPC, which is running a version of the Linux operating system. For more details on how to connect to our systems, please visit
If that sounds too intimidating, and you would like to try a zero-client web-browser approach to using OSC, you can log in to OnDemand:
SFTP (Secure File Transfer Protocol) is the primary method of moving files to and from OSC storage systems. SCP (Secure CoPy) is another common method. Small files can be moved to and from OSC systems inside of OnDemand. Moving large files to or from OSC via the login nodes can experience difficulty, since there are compute restrictions on those nodes, so the recommendation is to connect to sftp.osc.edu, or scp.osc.edu. If you have a large amount of data to move, you can schedule transfers via Globus Online.
Batch scripts – instructions to be executed by the node(s) assigned to you when your request for resources can be met by the scheduler – can get very complicated, so it’s best to refer to the documentation for the software you want to use to get started.
We also have a more detailed guide on the batch system that we recommend new users read. It provides information about scheduling, and how to write scripts to accomplish various tasks.
You can find events at the center (including training) by visiting osc.edu/e , system notices are at osc.edu/n , we provide important information in the “message of the day” (visible when you log in), and you can get notices by following @HPCNotices on Twitter. Help can be found by contacting our help desk weekdays, 9AM to 5PM.
Support: 1-800-686-6472, 614-292-1800 (local), email@example.com
Main Supercomputing page: http://www.osc.edu/supercomputing/
ARMSTRONG Client Portal: https://armstrong.osc.edu/
Using the HPC or “supercomputer” is a little different from running programs on your desktop. When you login you’ll be connected to one of the system’s “login nodes”. These nodes serve as a staging area for you to marshal your data and submit jobs to the batch scheduler. Your job will then wait in a queue along with other researchers' jobs. Once the resources it requires become available, the batch scheduler will then run your job on a subset of our hundreds of “compute nodes”. You can see the overall structure in the diagram below.
An important point about the diagram above is that OSC clusters are a collection of shared, finite resources. When you connect to the login nodes, you are sharing their resources (CPU cycles, memory, disk space, network bandwidth, etc.) with a few dozen other researchers. The same is true of the file servers when you access your home or project directories, and can even be true of the compute nodes.
For most day-to-day activities you should not have to worry about this, and we take precautions to limit the impact that others might have on your experience. That said, there are a few use cases that are worth watching out for:
The login nodes should only be used for light computation; any CPU- or memory-intensive operations should be done using the batch system. A good rule of thumb is that if you wouldn't want to run a task on your personal desktop because it would slow down other applications, you shouldn't run it on the login nodes. (See also: Interactive Jobs.)
I/O-intensive jobs should copy their files to fast, temporary storage, such as the local storage allocated to jobs or the Lustre parallel filesystem.
When running memory-intensive or potentially unstable jobs, we highly recommend requesting whole nodes.
If you request partial nodes, be sure to consider the amount of memory available per core. (See: HPC Hardware.) If you need more memory, request more cores. It is perfectly acceptable to leave cores idle in this situation; memory is just as valuable a resource as processors.
In general, we just encourage our users to remember that what you do may affect other researchers on the system. If you think something you want to do or try might interfere with the work of others, we highly recommend that you contact us at firstname.lastname@example.org.
There are two ways to connect to our systems. The traditional way will require you to install some software locally on your machine, including an SSH client, SFTP client, and optionally an X Windows server. The alternative is to use our zero-client web portal, OnDemand.
You can access OnDemand by pointing a web browser to ondemand.osc.edu. Documentation is available here. Any newer version of a common web brower and Java 1.6 or better should be sufficient to connect.
In order to use our systems, you'll need two main pieces of software: an SFTP client and an SSH client.
SFTP ("SSH File Transfer Protocol") clients allow you transfer files between your workstation and our shared filesystem in a secure manner. We recommend the following applications:
SSH ("Secure Shell") clients allow you to open a command-line-based "terminal session" with our clusters. We recommend the following options:
A third, optional piece of software you might want to install is an X Windows server, which will be necessary if you want to run graphical, windowed applications like MATLAB. We recommend the following X Windows servers:
The primary way you'll interact with the OSC clusters is through the SSH terminal. See our supercomputing environments for the hostnames of our current clusters. You should not need to do anything special beyond entering the hostname.
Once you've established an SSH connection, you will be presented with some informational text about the cluster you've connected to followed by a UNIX command prompt. For a brief discussion of UNIX command prompts and what you can do with them, see the next section of this guide.
To transfer files, use your preferred SFTP client to connect to:
Since process times are limited on the login nodes, trying to transfer large files directly to glenn.osc.edu or oakley.osc.edu may terminate partway through. The sftp.osc.edu is specially configured to avoid this issue, and so we recommend it for all your file transfers.
With an X Windows server you will be able to run graphical applications on our clusters that display on your workstation. To do this, you will need to launch your X Windows server before connecting to our systems. Then, when setting up your SSH connection, you will need to be sure to enable "X11 Forwarding".
For users of the command-line
ssh client, you can do this by adding the "
-X" option. For example, the below will connect to the Oakley cluster with X11 forwarding:
$ ssh -X email@example.com
If you are connecting with PuTTY, the checkbox to enable X11 forwarding can be found in the connections pane under "Connections → SSH → X11".
For other SSH clients, consult their documentation to determine how to enable X11 forwarding.
The Ohio Supercomputer Center provides cycles to both academic and industrial clients. The methods for gaining access to the systems are different for each user community. Primarily, our users are Ohio-based and academic, and the vast majority of our cycles will continue to be consumed by Ohio-based academic users. Those cycles are allocated to academic PIs via an allocations process overseen by the Statewide Users Group, which evaluates applications via a peer review process. There are two basic classes of academic allocations: an allocation of a number of resource units (RUs) that never expire, and an annual allocation process that awards an RU allocation for the upcoming calendar year. The annual allocation process is designed for "discovery level" users, who consistently consume large amounts of resources, and allow the Allocations Committee to better allocate the limited Center resources. (You may find more information about "RU" calculation and charging here.)
Other users (non-Ohio academic or industrial/commercial) interested in using Center resources may purchase cycles at a rate set by the expected total consumption. Expert consulting support is also available.
|Allocation Approval||SUG||SUG||OSC Director|
|Call for Proposals||At any time||November||At any time|
For users interested in gaining access to larger resources, please contact OSC Help. We can assist you in applying for resources at an NSF or XSEDE site.
Once a project has been created, the PI can create accounts for users via the ARMSTRONG portal, or add existing users to that project
I need additional resources for my existing project
If you already have a project, but are running low on allocated resources, you can click on the links above for your project type to request additional resources.
I wish to use OSC to support teaching a class
We provide special classroom allocations for this purpose. You may apply here.
I don't think I fit in the above categories
Please contact us.
Once your account is set-up you will receive a welcome letter from us containing your username and default password, which will allow you to register your ARMSTRONG account.
The ARMSTRONG portal provides many services to OSC users, including:
To register for your ARMSTRONG account as an OSC user, click the "Register Now" link under "OSC HPC Clients" in the lower-left portion of the webiste or click here. You will be asked for the username and password provided in your welcome letter, after which you will be guided through the ARMSTRONG account creation process.
Note: Your ARMSTRONG account is not the same as your HPC account. Changing the password for one will not change the password for the other, and you cannot log into ARMSTRONG using your HPC credentials.
Once you've created your ARMSTRONG account, you should change your HPC password to something besides the default. To do this, mouse over the "HPC Accounts" menu at the top of the page and choose "Change HPC Password/Shell".
Note: HPC password changes and resets may take up to one hour to process. During this time your old password will continue to work. If your password still has not changed after an hour, contact us at firstname.lastname@example.org.
Principal Investigators of OSC projects are responsible for updating their authorized user list, their outside funding sources, and their publications and presentations that acknowledge OSC. All of these tasks can be accomplished through the ARMSTRONG portal (https://armstrong.osc.edu). PIs are also responsible for monitoring their project's Resource Unit balance and for submitting a proposal to the Allocations Committee of the Statewide Users Group before the balance goes negative. Projects with negative balances are subject to restrictions, causing submitted jobs to have lower priority in the queue.
Eligible Principal Investigators must be tenure-track faculty members or permanent research scientists at Ohio colleges or universities. All other researchers, such as postdoctoral researchers, graduate students, and undergraduate students, may be authorized users under a project led by an eligible PI.
Principal Investigators who plan to submit a proposal for additional resource units should have the following information ready when they fill out the application form at https://www.osc.edu/supercomputing/additional_resources :
Once a proposal has been vetted for sending out for review, any restrictions that have been incurred owing to a negative balance will be lifted. The Allocations Committee meets early in each odd-numbered month, and awards are made after the meeting.
Please disregard any RU messages or instructions to apply for additional resources. If you have any billing questions, please consult your contact with OSC.
OSC HPC resources use an operating system called "Linux", which is a UNIX-based operating system, first released on 5 October 1991. Linux is by a wide margin the most popular operating system choice for supercomputing, with over 90% of the Top 500 list running some variant of it. In fact, many common devices run Linux variant operating systems, including game consoles, tablets, routers, and even Android-based smartphones.
While Linux supports desktop graphical user interface configurations (as does OSC) in most cases, file manipulation will be done via the command line. Since all jobs run in batch will be non-interactive, they by definition will not allow the use of GUIs. Thus, we strongly suggest new users become comfortable with basic command-line operations, so that they can learn to write scripts to submit to the scheduler that will behave as intended. We have provided some tutorials explaining basics from moving about the file system, to extracting archives, to modifying your environment, that are available for self-paced learning. We also occasionally offer training classes on this topic.
Welcome to OSC! This document includes some useful information to help your classroom use of our resources be effective.
Access: We suggest that students consider connecting to https://ondemand.osc.edu/ to access the HPC resources. All production supercomputing resources can be accessed via that website, without having to worry about client configuration.
Resources: We currently have two production clusters, Oakley and Glenn, with Nvidia GPUs available on both (but in larger quantities on Oakley). Recent usage patterns have shown that jobs on Glenn typically execute faster. However, both systems have "debug" queues during typical business hours that allow small jobs of less than 1 hour to start much quicker than they might otherwise.
If you need to reserve access to particular resources, please contact OSC Help (preferably with at least two weeks lead time) so that we can put in the required reservations to ensure resources are available during lab or class times.
Software: We have a list of supported software, including sample batch scripts, in our documentation. If you have specific needs that we can help with, let OSC Help know.
Students: We have a single-page guide for new students to help them figure out the basics of using OSC at http://www.osc.edu/new_user. Included are basics on getting connected, HPC system structure, file transfers, and batch systems.
Support: You can find events at the center (including training) by visiting osc.edu/e , system notices are at osc.edu/n , we provide important information in the “message of the day” (visible when you log in), and you can get notices by following @HPCNotices on Twitter. Help can be found by contacting OSC Help weekdays, 9AM to 5PM.
OSC Help: 1-800-686-6472, 614-292-1800 (local), email@example.com
Main Supercomputing page: http://www.osc.edu/supercomputing/
ARMSTRONG Client Portal: https://armstrong.osc.edu/