Increasingly, sophisticated research use cases have resulted in increased researcher demand for diverse computation platforms. Single dimension number crunching is no longer a sufficient enabler, and contemporary research has developed a strong digital footprint that breaks the traditional HPC mould. More dimensions are needed to be added to the University research computing landscape.
The Virtual Research Desktop project (VRD project) seeks to deliver a unique capability in the Australian research computing context. While the project team have considered similar services provided by other national and international bodies and institutions, it is apparent that each of the solutions seeks to address overlapping sets of use cases and requirements. It is also evident that the way the service is provided – including the user experience and level of end-user support – differs across similar services. Nearly all related services are focused on Linux use cases, whereas the intent of the VRD initiative is to support both Linux and Windows use-cases.
Argus is the University of Sydney’s VRD project designed to bring remote, on-demand, interactive, graphically intensive compute environments. VRD supports common, convenient software and environments for both traditional and non-traditional users of research computing, including HPC services.
Currently Argus is only availabe to the following list of pilot researchers:
- Brain and Mind Centre
- School of Geosciences
- School of Chemistry
- School of Civil Engineering
- Preclinical Imaging Core Facility
- School of Physics
- School of Economics
- Forefront / Brain and Mind Centre
- Department of Environmental Sciences
- Mass Spectrometry Core Facility
- The Westmead Institute for Medical Research (Cytometry)
For more information about Argus please visit the Ask Sydney website.
Argus vs Artemis
Not all computing tasks are appropriate for HPC (Artemis); the University’s HPC system is designed for batch-oriented rather than interactive computing. It excels with workloads that are highly parallel (i.e. can processed by multiple CPUs simultaneously), or which require very large amounts of memory (2TB and beyond). There are many workloads that do not fit this model and which cannot make use of Artemis’ multi-node or high-memory capabilities. Many of these tasks still require computational capability that exceeds that available in common laptop or desktop computers.
...