One million jobs run on the Research Campus Grid
Over the last few years the University of Newcastle’s High Performance Computing environment which is known as the Research Computing Grid (RCG), has supported numerous research groups in running simulations.
The groups benefit by being able to use either the multiple CPU cores of the RCG in parallel for a single large simulation or, running dozens of small independent simulations at the same time. Using the RCG for either scenario provides a valuable reduction in time taken to produce the end results. Without the RCG, the only other option would be to run them one at a time on a desktop, or not at all.
During November, job number 1,000,000 ran on the grid! Some of the interesting jobs run during the month include:
- Modelling the air flow through small diffuser augmented wind turbines.
- Simulations of water inundation in soil embankments.
- Developing a model for detection of Schizophrenia in human genomes.
- Deep learning for pattern recognition in images using the Kepler GPU cards.
Recently, researchers from the School of Environmental and Life Sciences have been running tidal models of Lake Macquarie. In order to be able to increase the number of weeks modelled and to use a high resolution of 1 metre, they estimated that their largest simulation would take 5 months to run on a desktop. In some scenarios, running a simulation for this length of time is just not feasible.
It was at this point that the ARCS team were able to get involved and offer a much more viable solution. By using the RCG during an off-peak period, ARCS were able to dedicate 512 CPU cores to the task it took just 18 hours to complete the entire simulation! The below image was the produced result (Image is highly reduced for this newsletter).