Competitive edges

NCSA’s PSP has worked with more than one-third of the Fortune50 and nearly 60 percent of the nation’s leading manufacturers. Access’ Liz Murray chatted with PSP leaders Merle Giles and Evan Burness about the program’s history of success.

What makes the nation’s top companies want to partner with PSP?

Merle Giles: When NCSA was founded in 1986, one of its goals was to establish a relationship between science and industry. Illinois was a perfect place for this, as the university had been demonstrating preeminent scientific research for 100 years, and a number of Fortune 100® companies were headquartered in the state. Building a supercomputing center in Champaign-Urbana garnered a lot of attention from the beginning and our early work was indeed with the companies that were an easy drive away. We have extended these deep roots with Illinois industrial companies into engagements with global manufacturers, pharmaceutical companies, oil and gas corporations, and bio-tech companies that seek to accelerate their science and engineering.

Evan Burness: One of the main reasons why more and more companies are collaborating with NCSA is because HPC has graduated from being a research capability only to also a tool for production work, commercialization work, and really applied competencies. Companies are looking to us for guidance as they go through this transition, because this is all we do. We’re HPC experts, and experts in the areas around HPC, whereas the companies we work with are experts in making bulldozers, or pharmaceuticals, or in sequencing genomes. A partnership with PSP affords them the ability to remain focused on those things while still gaining a competitive edge through advanced computing and data technologies.

The University of Illinois has two new industry/manufacturing initiatives, UI Labs’ Digital Manufacturing Lab and the Illinois Manufacturing Lab. What role will NCSA and PSP play in those?

MG: The modern reality of advanced manufacturing is that it is increasingly digital. The early adopters have been the original equipment manufacturers (OEMs), with deep internal engineering and science expertise. As OEMs adopt more digital capabilities, demand increases for similar capabilities in the supply chain. NCSA is involved in serving small and medium manufacturers (SMMs) as well as OEMs.

With capital funding from the Illinois governor’s office and matching funds from the University of Illinois, the Illinois Manufacturing Lab is focused on how SMMs can lower uncertainty and optimize product design by using digital tools. NCSA has extensive experience with the fundamental digital building tools of manufacturers: fluids and structures, and we have the expertise to use these tools on supercomputers dedicated to industry.

The Digital Manufacturing Lab responds to the White House’s National Network for Manufacturing Innovation initiative, specifically the Digital Manufacturing and Design Innovation (DMDI) institute funding awarded in February and managed by the Department of Defense. NCSA is offering its expert services and resources to the Digital Manufacturing Lab for advanced modeling and simulation, data integration, and other digital design needs of the manufacturing community. This is an extension of the work we’ve been doing since our founding in 1986, and particularly in the past dozen years since we’ve had dedicated supercomputing for industry. Today, both iForge and Blue Waters are used to run the engineering community’s most demanding applications, with both high performance and scale.

EB: I think what is exciting about both UI Labs and Illinois Manufacturing Lab is that they allow NCSA to step outside its comfort zone in some really interesting ways. Supercomputing centers tend to sort of live in their own bubble, and in many ways they have to because HPC is such a specialized field. Nonetheless, the need for separate institutes like UI Labs and Illinois Manufacturing Lab—in terms of how we serve manufacturing, in both a narrow sense thru PSP, but also broader sense through the University of Illinois system and the state of Illinois—is based in the notion that manufacturing in the 21st century is going to look very different than manufacturing in the 20th century.

Both initiatives are very focused on this ability to better serve industry communities in ways that are very applied, and if you think about it, that begins to look like what we have been doing in PSP for quite some time. So for NCSA to contribute our supercomputing expertise to both projects is fantastic in and of itself, but it’s also a great way for us to get closer to this really cutting-edge manufacturing space.

What were some of the big PSP success stories of 2013?

MG: In 2013 we had a really fun year for a lot of reasons. We hit a high in the number of PSP partners with 25. Those 25 partners were in sectors that include manufacturing and technology, oil and gas, and bioinformatics, with a very interesting mix of companies within each of those sectors. PSP added client consulting and expert assessment in new ways where we leveraged the broad strengths of NCSA, such as storage, networking, and application performance. Technical computing differs greatly from classical IT, and the PSP consulting team is very well suited to help companies bolster the performance of their most valuable scientists and engineers.

Another thing that was extremely rewarding for us was to bring the National Digital Engineering and Manufacturing Consortium (NDEMC) program to a successful completion. Through NDEMC we were able to serve SMMs directly, ultimately benefiting our large industry partners as well as the U.S. Department of Commerce, demonstrating how important advanced modeling and simulation is in small companies.

And we are most proud of how our work with OEMs in the past several years informed the national dialogue for DMDI, leading to a $320 million public-private partnership and a splendid new opportunity to pursue digital solutions for the nation’s entire manufacturing community.

EB: For me, I think the big successes for 2013 starts with the increased use of iForge. Our partners aren’t just using this supercomputer in large volumes they are using it for more varied production use cases and as a result are integrating it more and more into their broader production capabilities. The second notable success is we are starting to increase our use of the Blue Waters extreme scale system. This allows us to tackle some of our partners’ grand challenges. In some cases we’ve been able to accelerate solutions to problems that companies thought might be 10 years away from being solved.

As an example, PSP has performed some of the most advanced supercomputing ever done for commercial engineering workloads with some of the most widely used codes, such as the LS-DYNA code for structural mechanics, and Fluent and STAR-CCM+ codes for computational fluid dynamics. We also demonstrated some of the first use cases of Abaqus, the world’s most popular structural mechanics code, using both CPUs and GPUs to solve a real-world problem. These are big-time real breakthroughs; not in an abstract sense but in a very real sense.

How does supercomputing for the corporate sector differ from supercomputing for the academic sector?

EB: The most important point of differentiation is that with industry, there is an insatiable appetite for performance as well as an insatiable appetite for reliability. I think one of the key design features of iForge is its reliability, which is also a key design feature of any infrastructure or software that we recommend to industry partners. Our focus is packaging the very best of what NCSA has to offer and sharing the expertise we have gained through our cutting-edge work with HPC technologies, products, and software. That expertise translates into services and systems that we stand up at NCSA or help our industry partners stand up for themselves. Those things become a core strategic asset for them, the thing that is going to give them a leg up over their competition.

Could you give an example of a time when PSP was able to make the impossible possible for one of your industry partners?

EB: We had a company come to us last year requesting physical prototyping of a particular product they manufacture, with the goal of working towards the fully digital production of this item in the future. At the time this partner believed they were still many, many years away from accomplishing this switch due to the fact the application used to do this digital prototyping didn’t scale more than about a hundred cores simultaneously… at least at the time it didn’t. Thanks to Blue Waters, and with help from teams from LSTC and Cray, PSP pushed this particular real-world problem from this industry partner to over 15,000 cores. We have moved the football down to the one-yard line for them in terms of being able to do all of the physical prototyping in an entirely digital way. I can tell you that I don’t know if we’ve ever had a happier PSP partner than this company, and the goal this year is to get them over that one-yard line and into the end zone.

MB: One other example is our work with the National Cancer Institute on a shared memory machine a few years back. They played around for a little bit and I called them up and said, “You know what? I don’t think you even made the machine sweat. Step it up a bit.” So, while it had never been done before, they attempted stitching together 16 or 17 slices of a tumor in digital space. They put it all into memory, stitched it back together like a sliced carrot, oriented it in the right ways, and stuck a digital probe in it like a digital pathology task. And we accomplished it. There is something special about getting the right people in the right room with the right impossible problem; we can make amazing things happen in that space.

Disclaimer: Due to changes in website systems, we've adjusted archived content to fit the present-day site and the articles will not appear in their original published format. Formatting, header information, photographs and other illustrations are not available in archived articles.