Access

Introduction

The iVEC-run Pawsey Supercomputing Centre is supported by $80 million funding from the Australian Government. To help ensure that supercomputing resources are used effectively and efficiently, iVEC allocates compute time to researchers through project awards. Proposals for projects may be submitted through one of a number of schemes (detailed below) that iVEC operates or participates in.

Through these allocation schemes, iVEC strives to:

• maximise the research impact of Pawsey supercomputing resources;
• promote scientific advantage in priority domains, such as radio astronomy and geosciences;
• provide leading-edge, supercomputing resources for researchers in iVEC Partner institutions;
• enable wider adoption of and benefit from supercomputing across Australia.

When requesting access, prospective users should consider the range of resources available and the allocation schemes that we offer to gain access.

Resources Available

In 2015, the following Pawsey resources will be available:

• Magnus – iVEC’s petascale supercomputer, which supports capability-scale computational-research. It is a Cray XC30 system, with 1,488 compute nodes using the latest generation of Intel Xeon ‘Haswell’ CPUs, linked by the Cray Aries interconnect. Magnus, in its full petascale configuration, is available through merit-allocation schemes for the first time in the 2015 call. Those wishing to apply for time on Magnus are encouraged to apply for a Director’s Share allocation, now, to help them to complete their proposal during October.
• Fornax – iVEC’s data-intensive-research system, which supports accelerator-enabled and memory-intensive applications. It is an SGI cluster, with 96 large-memory nodes powered by NVIDIA M2090 ‘Fermi’ graphics processors (GPUs) on a Mellanox QDR Infiniband interconnect. Fornax will only be available until 30th June 2015, at which point it is due to be decommissioned. Please factor this into your plans.
• Galaxy – iVEC’s radio-astronomy-focused supercomputer. It is a Cray XC30 system, with 472 compute nodes based on Intel Xeon ‘Ivy Bridge’ processors and 64 compute nodes based on NVIDIA K20X ‘Kepler’ graphics processors, linked by the Cray Aries interconnect to deliver more than 250 Teraflops of compute capacity. Galaxy, through the ASTAC call, is available for January–June 2015.
• Zythos – an SGI UV2000 system. Zythos consists of 24 UV blades, and they appear as a single system with huge shared memory of 6TB. 20 of the blades each contain two 6-core Intel Xeon E5-4610 CPUs, and the other 4 blades each contain one 6-core Intel Xeon E5-4610 CPU and one Nvidia Tesla K20 GPU. Multiple users are permitted to run on Zythos at the same time.

You may apply to use any, or all of the machines. To do this, one should lodge multiple separate applications. To be competitive, we recommend that applications to use Fornax clearly express how the architecture will be utilised, and for Magnus how the interconnect will be utilised. Further details of the Pawsey supercomputers is available at http://portal.ivec.org/docs/Supercomputers/System_Descriptions. Additional national resources can be requested through NCMAS.

Note: Epic will be decommissioned at the end of 2014. Affected Project Leaders, for current Epic projects, have been or will be contacted in the near future to discuss options that exist after the Epic service is closed. See https://portal.ivec.org/docs/Time-line_for_withdrawal_of_Epic_supercomputer for the full announcement.

Allocation Schemes

Project proposals are invited via a number of schemes which are targeted at different research communities, fields, and levels of computational experience.

At the time of writing, compute-time allocations may be obtained through the following schemes:

• The National Computational Merit Allocation Scheme (NCMAS) – 15% of the compute time available on Pawsey supercomputers is allocated through this scheme, along with compute-time allocations from other Australian centres including NCI, VLSCI, and MASSIVE. The scheme operates six-monthly calls, which are open to the whole Australian research community and which provide substantial amounts of compute time for meritorious, computational-research projects.
• The iVEC Geosciences Merit Allocation Scheme – 25% of the compute time available on Pawsey supercomputers is allocated through this scheme. Open to the Australian geoscience research community, providing significant amounts of compute time for meritorious research projects in the domain. Beginning in 2015, this Scheme will call for applications every six months.
• The Astronomy Supercomputing Time Allocation Committee (ASTAC) – typically 5—25% of the compute time available on Pawsey supercomputers is allocated through this scheme, alongside compute-time allocations from other Australian centres including NCI and Swinburne University. The scheme operates six-monthly calls open to the Australian astronomy research community. Significant amounts of compute time are provided for meritorious research projects in the domain of astronomy. Pawsey resources are limited to radio astronomy research.
• The iVEC Partner Merit Allocation Scheme – 30% of the compute time available on Pawsey supercomputers is allocated through this scheme. Calls are open to researchers in iVEC Partner institutions and provide significant amounts of compute time for meritorious, computational research projects. The Partner institutions are CSIRO, Curtin University, Edith Cowan University, Murdoch University and The University of Western Australia. Beginning in 2015, this Scheme will call for applications every six months.
• The iVEC Director’s Allocation Scheme – 5% of the compute time available on Pawsey supercomputers may be allocated to short-term projects requiring modest amounts of compute time. The scheme is open to all Australian researchers and is intended to support researchers with less experience of supercomputer-scale computational research and to support speculative investigations designed to assess the potential of supercomputing for an application.

The merit-allocation schemes are summarized in the following table.

N.B. ASTAC is a call to the whole astronomy community, though Pawsey resources are only available for projects in the sub-domain of radio astronomy.

The Allocation Process

Each project proposal for use of Pawsey supercomputers, through whatever scheme, is assessed as to:

• its compliance with the eligibility of the scheme to which it is submitted;
• its technical feasibility – in terms of appropriateness for the resources requested;
• its scientific merit.

Only a project proposal that fulfils the constraints of the particular scheme and that is deemed to be technically feasible is eligible to receive a compute-time allocation on an Pawsey supercomputer. Within meritorious schemes, such as the iVEC Partner Merit Allocation Scheme or the National Computational Merit Allocation Scheme, the process is conducted by an independent panel of computational researchers who prioritise and award compute time to eligible projects according to their scientific merit.

Within the Director’s Allocation Scheme, iVEC aims to provide a modest compute-time allocation to all projects that fulfil the scheme’s criteria and are deemed technically feasible. However, iVEC expects that those researchers that hold a Director’s award will pursue more substantial allocations of compute time through a meritorious scheme, if such time is required. For the systems available at the time of writing, a typical Director’s award would be as follows:

• Epic – 10,000 SUs (that is, CPU hours) over three months;
• Fornax – 1,000 GPU hours (or 10,000 CPU hours) over three months;
• Magnus Phase 1 – 10,000 SUs (that is, CPU hours) over three months;
• Zythos – 10,000 SUs (that is, CPU hours) over three months.

Completing an application

To ensure iVEC has the information needed to properly assess a proposal, a researcher should ensure they:

• meet the criteria of the call to which they are applying;
• read the call text carefully and aim to address all points of consideration;
• provide clear and complete information.

More specifically, the application form requires the following information:

For a Director’s allocation, an applicant is not expected to provide the same level of information as for meritorious schemes. In particular, information pertaining to the project team needs only to provide sufficient detail to allow iVEC to confirm the team’s ability to achieve the project’s documented goals.

A project should have defined:

• time-frame (specified start and end point);
• scientific goals;
• expected research impact;
• technical approach (including, software, hardware, tools and techniques);
• resources (people effort and expertise);
• cost (compute time and short-term storage requirements).

iVEC periodically runs Merit Allocation Workshops (typically advertised here), at which we explain the compute-time application processes and provide assistance for those researchers that are preparing applications. Furthermore, researchers are encouraged to contact the iVEC Helpdesk should they have any questions or comments regarding compute-time award schemes or allocations.

Licensed Software

iVEC does not purchase licensed software apart from compilers, profilers, debuggers and similar.

Licensed software can be installed on behalf of users if it does not violate the terms of the license. A copy of the licence will need to be sent to iVEC with the request for installation or access to the software.

If a license server is required (such as FlexLM), the user needs to make appropriate arrangements with their own organisation. This will likely involve firewall changes in that organisation. iVEC can assist with providing the relevant network information as well as testing. iVEC will not host licenses on its own servers.

Notes

• When evaluating a proposal, the selection panel considers the proposer’s performance in any previous compute-time awards they may have received – in terms of fulfillment of goals and use of allocated resources.
• Allocation of compute time on Pawsey resources remains the discretion of the iVEC Executive Director: their decision regarding the assessment of a proposal or the allocation of compute time is final.

For assistance in completing your merit allocation application, please contact the iVEC Helpdesk on [email protected].

Useful Links

• iVEC Merit Allocation Portal.
• Understanding your Supercomputing Resource Allocation

Back to Main Page