Event Coverage

Data, data, data: Inside Monash University’s NVIDIA-powered CAVE2 facility

By John Law - 30 Sep 2015

Data, data, data: Inside Monash University’s NVIDIA-powered CAVE2 facility

The facility which houses the CAVE2 supercomputer, located within the Clayton Campus of Monash University in Melbourne, Australia.

In less than a span of a month, we went from talking to the professor who is an authority on HPC visualization to visiting the actual site of where the facility was being house. We are, of course, talking about the CAVE2 facility that is housed within the Clayton Campus of Monash University at Melbourne, Australia.

This is what we came here to see: the CAVE2 facility, which is fully powered by NVIDIA's Quadro workstation GPUs.

To recap, we first heard of the facility – completely powered by NVIDIA’s Quadro GPUs – and how the good people at Monash University were using it to change the way people viewed data. Let’s be honest: any talk regarding data, codes, numbers, and calculations tend to be relatively long-winded and ultimately boring. But believe it or not, these codes and calculations actually serve to paint a bigger picture for the ones who understand it, both figuratively and literally. That’s pretty much where the CAVE2 facility comes in.

This is what the CAVE2 facility is comprised of.

The CAVE2, which is driven by the sheer prowess of 40 NVIDIA Quadro K5200 workstations graphics cards, is the every very essence, epitome, and definition of a machine that is made for the purpose of data visualization and HPC (that’s High Performance Computing, for those unfamiliar with the acronym). To summarize the entirety of the CAVE2, it is a facility comprising of 40 46-inch curved display which, when joined from bezel to bezel, comprises of a massive total of 84 million pixels, and close to 90 Teraflops of memory to allow the CAVE2 to process and visualize any data that is brought to their facility (within reason, of course). These screens are powered by a total of 20 Dell nodes, and each node powers up approximately four screens at any given time.

To be even more concise and clear about the CAVE2, the entire system can be used for numerous applications. Want to map out an entire room using nothing but Lidar (Light-based Radar)? Not a problem for the CAVE2. Thanks to the NVIDIA Quadro workstation graphics, the supercomputer can easily take the data of the required space and visualize it in real-time. The videos below show just how detailed these real-time rendered visuals are. We do apologize in advance for the glasses bit, as the entire demonstration of the following videos were all in 3D. By putting one part of the glasses over it, we were actually able to show you just how cool the experience really was.

It doesn’t stop there. The CAVE2 is also capable of visualizing wide environmental areas and subjects and render into real time, as performed by the video below.

This particular demonstration of the CAVE2 did actually intrigue us, as the applications of this could actually prove to be useful in law enforcement and post-mortem forensics. To elaborate on the latter: by actually rendering an entire area in real-time and at a particular time, forensics scientists could easily view different areas of the scene with relative ease and safety as well. As an example: a bomb has been detonated in a public area, and the forensics analyst needs to find the point of detonation. By mapping out the area using the CAVE2, the analyst can specifically identify where the explosion happened.

Again, that’s just one of the many applications. The CAVE2 can also be used to monitor and calculate the airflow of an area, allowing scientist or construction companies to easily determine an area in the city where the airflow would allow them to construct a new building without the fear of the building being worn down over time by air erosion or even being brought down by a strong wind.

One of the last demonstrations that was shown to us was the essence of data visualization: mapping out an entire biological entity in 4D, and even adding varying levels of details in order for the scientist to determine the different density levels of the subject. This ability of the mapping out an entire body of a rabbit and a rat – both with an immense and insane amount of detail – was the proof of the pudding for the CAVE2’s data visualization capabilities. As it that weren’t enough, the CAVE2 could even simulate a breathing pair of lungs on the 4D model itself!

The CAVE2 can easily visualize biological subjects, just like this 4D render of a rabbit's body and lungs.

All in all, it goes without saying that none of the CAVE2’s capabilities could’ve have been possible without the use of NVIDIA’s GPU technology. That being said, it’s also a nice and gentle reminder that NVIDIA isn’t just a name that’s exclusively associated with gaming any longer, but one that is also synonymous with the advancement of HPC, data visualizations, and the science as a whole.

For more news from NVIDIA, follow us here.