You are here

High Performance Computing (HPC)

What is the HPC service?

HPC is the use of large-scale, high-throughput  computers  and parallel processing techniques for solving complex computational problems. The terms “HPC” and “supercomputing” are sometimes used interchangeably. HPC machines are based on a Unix/Linux operating system and involve a batch system or scheduler to maximise the compute power. 

HPC systems are most often useful for researchers:  

  • With parallelisable tasks (parallelism can be achieved on a fine or coarse grain level — on a fine level you have a program which is intrinsically parallel, while on a coarse grain level you might run many instances of the same program for parameter studies)
  • With high throughput compute requirements
  • With computing jobs needing more memory (RAM) than available on their local systems.

 

Intersect manages a HPC facility named Orange that was commissioned in 2013. Further information on Orange technical specifications can be found here.  Intersect is also a partner of the National Computational Infrastructure (NCI) peak facility based at the Australian National University. The NCI peak facility is named Raijin and further information can be found here.

To make the right choice of HPC facility:

  • Investigate the software available on Raijin and Orange.  Complete lists of all software installed are available at the NCI software list - there is a drop-down menu from which you can select almost all HPC machines in Australia, including Intersect's Orange.  Please note that software packages marked with a yellow dot carry license restrictions and advice may be required on whether you can use this software.  Also, it may be possible to request new software packages.  
  • Compare the hardware specifications of Raijin and Orange.  
  • Consider the memory requirements of your research problem. Orange and Raijin are both distributed memory systems.  However, Orange has some nodes with a larger RAM allocation which are suited to certain classes of research applications where access to large amounts of memory is a key performance requirement.
  • Also consider NCI's Specialised Facilities.  NCI offers specialised Bioinformatics and Imaging/Visualisation facilities. If your research is in either of these areas, then NCI could well be the right choice for you.

 

Who can access the HPC service?

  • For researchers from Intersect member organisations, HPC resources are merit-allocated and fully subsidised.  
  • For non-Intersect members, Intersect Affiliates or members wanting additional resources, HPC time can be purchased, please email us at time@intersect.org.au

 

How do I access the service?  

There are several methods to access Intersect’s HPC resources:   


1. Merit Subsidies for HPC Resources

What is merit subsidy?

Intersect supports high-impact research through subsidised allocations of its HPC service available to researchers within Intersect member organisations. Larger proposals for significant levels of resources on the Orange and Raijin machines are requested through the annual Resource Allocation Round. Small, ad hoc usage may be approved outside the annual resource allocation round - refer Small Ad Hoc Usage below for information on smaller-scale “experimental” proposals.   

As a merit-based system, applications are reviewed by our HPC staff (for technical complexity and track record) and by the Intersect Resource Allocation Committee (for research merit).  Important factors the Committee considers to determine the allocation of resources include: 

  • quality of the research argument
  • track record of previous granted resources
  • suitability of the chosen hardware and software.  
     

The Intersect Resource Allocation Round (RAR) is aligned with the National Computational Merit Allocation Scheme (NCMAS). We encourage users with significant allocation requests (> 50k Core Hours per quarter) and who consider their projects to meet the criteria of NCMAS, to apply for both the Intersect RAR and NCMAS in order to maximise the chance of receiving compute time. The same project information can be used to apply for RAR and NCMAS.  Applications that are not successful for NCMAS will be reconsidered in the Intersect RAR if a separate application is lodged.  

To maximise HPC usage on the NCI Raijin machine, we evaluate the usage of each project after 6 weeks (middle of each quarter) and may re-distribute resources if necessary.

How do I apply?

Applications for merit-based resource subsidies (Orange and Raijin) are made through NCI's forms:

New project applications:    https://my.nci.org.au/mancini/login?next=/mancini/project/propose/

         ***Please select "INTERSECT (NSW)" as Scheme/Partner***

Create a new ID at: https://my.nci.org.au/mancini/signup/0

To be added to an existing project, use this form: https://my.nci.org.au/mancini/project/

Note that an application must be made by academic staff at Intersect member institutions. We also welcome PhD students making use of the facilities but require that the lead CI on an application be an academic staff member.

Indicative dates for 2016:

1 October 2015                     -                Applications Open
31 October 2015                  -                 Applications Close
Last week of 2015               -                 Award Notifications

Acknowledging Intersect

If you have used merit resources on Orange or Raijin via the Intersect partner share, we request that you acknowledge us. The proposed text is:

Computational (and/or storage) resources used in this work were provided by Intersect Australia Ltd.

The full policy is available here.


2. Small, Ad Hoc Usage

For new users needing to test the suitability of HPC, Intersect is able to allocate a small HPC resource at numerous times throughout the year.  Ad-hoc projects have a fairly limited quota, but they are enough for you to install your application, submit a job, and see if it runs smoothly.  An ad-hoc project will provide 1,000 Service Units (SU) of computing resources. One SU corresponds to one core-hour, so if you run a parallelised program that uses four cores and runs for one hour, you will have used 4 SUs of your quota.

For an ad-hoc project:

  • Contact your institution’s eResearch Analyst or Intersect HPC specialist to request an ad-hoc project. This enables us to know that an application is coming.
  • Go to the following link and register using the Application Form for New User Registration at NCI NF system.  
  • Once you have submitted this form, you will be issued with an "identity number".  Send this identity number to your Intersect contact and they will process your application.
  • When the application has been processed, you will receive an email and SMS with your account username and password. This indicates that you are ready to use the HPC system. At this point it is best to let your Intersect contact know so they can assist you.
  • Once you have successfully completed your evaluation using an ad-hoc project, you can apply for a much larger quota through the merit-based resource allocation scheme

 

An ad-hoc allocation can be made outside the regular annual Resource Allocation Round (October application for January start). Note that ad hoc projects are specifically for test driving the systems. 

NCI also runs some start-up projects on Raijin. The default limit is 1k SU/quarter. These should be discussed directly with NCI. 

 

What training on HPC is available?

Intersect offers training courses on HPC topics as below:

  • Intensive Beginner to Intermediate HPC - Introduction to Unix for HPC: a one-and-a-half day course introduction to the command line and our HPC computing facilities. A basic familiarity with Windows or Mac OS X is assumed.
  • Intermediate HPC - From Unix to HPC: a 3-hour course introduction to HPC computing facilities for those with Command Line Unix Experience.
  • Advanced HPC - Parallel Programming: An introduction to Parallel Programming in C or Fortran using MPI and OpenMP.  


Where can I find more detailed information on using HPC?

Check out Time.intersect.org.au/supercomputing for practical information including:

  • Accessing HPC machines
  • Setting up a software environment
  • Compiling code
  • Running jobs
  • Using disk space.  
     

What costs are involved?

  • Intersect Member: Fully subsidised through membership
  • For non-Intersect members: By commercial negotiation.