Frequently Asked Questions
How do I know I need to use high performance computing systems?
Typical "symptoms" include the following:
- it takes your computer hours to get your calculations done
- you cannot use your PC because it is running your jobs
- your PC does not have enough memory or storage to complete a job
- an application you need is not available for the type of operating system or hardware architecture you own
Who can use the high performance systems?
Accounts on systems are created for faculty members upon request. Students can get their accounts through their principal investigators (PIs). Apply for an account.
How small can my jobs be for me to still be eligible to use an HPC system?
There is no minimum size requirement. Clearly, it makes little sense to use a high performance system for a job you can easily run on your own PC. However, there are situations when a "small job" (a job that runs fast and consumes little resources) is sent to an HPC system. It might be a test job that is followed by a bigger job or it is a part of an array of 10s -100s of similar small jobs. Again, the answer is there is no minimum requirement.
Does my computing problem have to involve parallel computation in order for me to be able to use a HPC system?
HPC has typically been associated with symmetric parallel jobs but other computations might still find benefit and be appropriate for HPC. Some jobs might not necessarily be parallel but might require 1000s of runs. Other jobs might have high requirements in terms of memory or storage. All these characteristics make them strong HPC candidates. Some resources are optimize for highly parallel operations (URSA) while others like United Devices, ACSrocks, and Octans are better picks for less parallel jobs. Contact Research Cmputing via help@gsu.edu to get help when making this type of decision.
Can Research Computing help me work out how to parallelize my problem? How do I get help setting up a job (using commercial software package) / (using my own executable)?
Reearch Computing will discuss your situations with you. Contact Research Computing via help@gsu.edu.
What software packages does Research Computing already have available for me to use?
Gausian, Amber, AMMP, and SAS are available on URSA. Matlab is available on Octans (more info).
Can Research Computing install other commercial software packages so I can use HPC systems?
If appropriate licenses are in place, Research Computing accommodates requests for additional commercial SW packages.
Can I run my own executables on HPC resources?
Yes, you can. If you need help, contact us via help@gsu.edu.
How much CPU time can I expect to be able to secure for my job? Are there minimum guarantees?
The HPC subcommittee is defining a research computing allocation policy to define how much time each user can expect. Each computing resource has finite amount of computing time; allocations will be determined in a way that helps researchers get their work done.
Is there a waiting line for HPC resources?
HPC resources are shared. As with all shared resources, there are "lines." Users submit jobs to a scheduler and it gets queued. The scheduler also makes sure the user gets a fair share of the resource.
Is there a description of research computing resources that I can use for grants or papers?
Yes, there is an executive overview available. Download it from here: [PDF] [MS Word]
Is there a list of software licenses available to campus researchers?
There are a number of software licenses available to campus users. See Application Support for HPC.