More demand for HPC means big changes in IT

For the last few decades, supercomputing has been fusing computation into science and engineering. It has become a powerful tool that has allowed scientists to advance more rapidly in research and answer questions that were impossible to answer through theory and experimentation.

With the advancement of technology, high performance computing (HPC) is now experiencing massive growth from all academic fields. “If you look at all HPC centers from a decade ago, we were very focused on the physical sciences — mathematics, engineering, things that were equation-based. Now all areas are looking or HPC,” said Dan Stanzione, Executive Director of the Texas Advanced Computing Center (TACC) at The University of Texas at Austin.

When asked about the future of high performance computing, he says classic HPC problems were based on science that had a physical design like designing airplanes, designing better materials and modeling reservoirs for oil. But now the real growth drivers in the area are with life sciences.

“If you look at how life sciences was done 20 years ago, it was a lot of wet lab and bench science. Now with imaging, genomics and proteomics, there is so much digital data that comes out of the life sciences,” Stanzione said. “Everything from basic biology to engineering new crops to the way we deliver health care is being radically changed through computing technology.”

The increasing need for HPC doesn’t stop there. Life sciences may have been the growth driver in the area in the last decade, but humanities and social sciences are now moving into the supercomputing and big data space. “We are seeing a sub-fusing of almost all areas of academic inquiry. The social sciences are becoming much more data driven, and are grabbing big pieces of data from the internet on how people behave and respond. Even the humanities and arts have become more and more digital,” Stanzione said.

One of the reasons is that high performance computing has advanced the digital visualization capabilities of the past. Stanzione says when researchers look at scientific data and render it in 3D, it goes beyond science. One way that academics in the humanities are working with HPC scientists is by building 3D reconstructions and restorations of great artworks and architecture before the pieces are lost or degraded overtime.

“All of this is preserved digitally, and the best way to show them off is through the same facilities we use for scientific computing,” Stanzione explained.

More Diversity Means More Challenges

As systems are advancing, it’s not just about high processing computing any more. Cloud and big data are also playing a role and this has raised more than a few challenges. The first challenge is diversity. HPC, cloud and big data require specialized platforms that require specialized skills. Beyond where you put the hardware, Stanzione says is the questions of how much in-house expertise you need to take advantage of the platforms capabilities — and the big question of how do you pay for it.

Then there is the challenge of compatibility. While a lot of scientific computing is suitable for the cloud, there are a number of processes conducted on high performance systems that aren’t. And then there is the question of accessibility. Organizations and research institutions are looking at what is private cloud, what is hybrid cloud and what should they take to the public cloud.

Finally, there is the economic challenge. “The cloud is a consumption model and most universities run on a capital model. You might have a budget of a million dollars and so you plan to do as much computation that fits into that million-dollar budget. But when you open it up to faculty out to the cloud and they start bringing in bills every month, you may not necessarily have the budget to cover it — and it’s hard to predict how much they are going to use,” Stanzione explained.

Be Flexible and Prepared to Adapt

With technology constantly changing and demands evolving, Stanzione says the key to the next 5 years is flexibility. With new technology comes opportunities to do things in new ways — such as breaking down the traditional discipline walls within universities.

“My whole career has been inter- and multi-disciplinary, but it’s still a huge challenge to get researchers to embrace that,” he explained. “The big challenges that society faces involve blending computation and a number of academic disciplines.”