From Science Node: “Can academia work with industry?”

Science Node bloc
From Science Node

07 Oct, 2019
Alisa Alering

AI and the data explosion means industry needs academic HPC more than ever.

As data accumulates at breakneck speed and demand for AI innovation grows, some of the world’s most prominent companies are looking to high-performance computing (HPC) to solve their toughest problems. Academic research centers have the compute power and the expert know-how to make those solutions a reality. But can these disparate sectors successfully work together?

To find out, we talk with Brendan McGinty, director of the pioneering industry program at the National Center for Supercomputing Applications (NCSA). Heading up a program with 30+ years of experience partnering with Fortune 50 companies, McGinty is well-placed to tell academic HPC centers why they might want to consider industry partnerships—and what to watch out for.

Brendan McGinty, director of NCSA Industry, says that big data and AI are driving the demand for academic HPC resources from companies in many sectors.

First, tell us a little bit about how the NCSA Industry program came to be.

NCSA Industry, formerly known as the Private Sector Program (PSP) was formed in 1986, shortly after NCSA’s founding. We help companies address their largest and most significant challenges by leveraging our experts and compute resources in advanced HPC.

The program has grown 400% in just the past 2.5 years. This is thanks to (1) the data deluge currently impacting nearly every large company and (2) emphasizing our expertise, where our consultative nature leads to further partnerships with companies. In turn, a byproduct of good consulting has been growth in compute use and revenue as well.

Why should academic HPC centers partner with industry?

Industry provides the significant challenges that academia loves—and massive datasets to boot. Academics and other professionals learn how to interact with companies who provide unique funding sources to academic institutions.

Perhaps more important than the scientific challenge is the career training and opportunities that students receive by engaging with companies and their projects. This is a perfect fit for the ever-growing, data-driven talent being developed by institutions of higher education.

HPC centers are key to the development of this partnered ecosystem. Industry’s large datasets and need for sophisticated solutions requires the expertise to scale or leverage AI and the compute on which to run associated jobs that HPC centers can provide.

Why do companies want to work with academia?

Academic institutions are an independent, not-for-profit source of expertise that matches industrial needs. They’re also quite affordable, with labor rates below those of typical for-profit consulting firms. In many cases, expertise comes in the form of PhDs which further legitimize R&D efforts in for-profit companies.

And, as has long been the case, companies align with universities for access to their future talent base. Working together provides a space to engage in a mutual vetting that ensures finding the best students to meet future corporate needs.

Can you point to any particularly successful partnerships at NCSA?

With over 30 years to choose from and many world records set and Top Supercomputing Achievement awards received, finding just one success story is a challenge. We’ve helped—and continue to help—improve many sectors including healthcare, energy, finance, transportation, insurance, and others.

In a 30-year history of working with industry, NCSA has partnered with companies in many sectors including healthcare, energy, finance, transportation, insurance.

From the standpoint of industrial benefit, a recent success involved a well-known and public story about our work with ExxonMobil. A simulation they developed to model oil reservoirs was running on their large, on-premise cluster in Houston—and taking 3.5 months to complete.

They approached NCSA to optimize and scale their existing code to run on Blue Waters, the massively parallel National Science Foundation (NSF)-funded machine with a small industrial allocation, perfectly architected for ExxonMobil’s needs.

NCSA U Illinois Urbana-Champaign Blue Waters Cray Linux XE/XK hybrid machine supercomputer

Our work resulted in using nearly all of Blue Waters in a 720,000 thread run that executed in ten minutes. What did it mean to ExxonMobil in terms of return-on-investment (ROI)? They said that number was over USD $1 billion.

What are some things academic institutions should know in order to work successfully with industry partners?

We have learned many lessons by working with companies in our public-private type of partnership, including:

Each company is unique—one size does not fit all. Provide customized solutions.
It’s not about our solutions—it’s about their needs. Understand their challenges and match solutions.
Companies generally have very little time. Make your value proposition efficiently and know your audience.
Work as closely as possible to the pace of the company, which is faster than that of higher education.
Help companies calculate efficiencies and ROIs. It leads to follow-on engagements.
Be consultative. Higher education has great solutions. Companies care about what will address their needs first.

What do you see for the future of these types of partnerships?

Three years ago, we weren’t talking about artificial intelligence (AI) solutions for companies. Now, it seems that every company with which we engage has, at the very least, interest in if not high levels of activity, in AI/machine learning (ML)/deep learning (DL)/geospatial solutions.

Three years ago, we also weren’t talking about the impact of graphics processing units (GPUs) in performance. Now, GPUs are either mixed with or replacing CPUs to run HPC jobs.

We must help companies, whether established or new, larger or smaller, to innovate to ensure their competitiveness and, perhaps, survival.

It is obvious based on the data explosion that hardware is changing rapidly, with software, as it typically does, lagging slightly behind. We need to help companies to push the proverbial envelope by helping hardware and software providers to do the same, with groups like ours being the trusted consultant along the way.

See the full article here .

Please help promote STEM in your local schools.

Stem Education Coalition

Science Node is an international weekly online publication that covers distributed computing and the research it enables.

“We report on all aspects of distributed computing technology, such as grids and clouds. We also regularly feature articles on distributed computing-enabled research in a large variety of disciplines, including physics, biology, sociology, earth sciences, archaeology, medicine, disaster management, crime, and art. (Note that we do not cover stories that are purely about commercial technology.)

In its current incarnation, Science Node is also an online destination where you can host a profile and blog, and find and disseminate announcements and information about events, deadlines, and jobs. In the near future it will also be a place where you can network with colleagues.

You can read Science Node via our homepage, RSS, or email. For the complete iSGTW experience, sign up for an account or log in with OpenID and manage your email subscription from your account preferences. If you do not wish to access the website’s features, you can just subscribe to the weekly email.”