Data Center

Q&A: Meeting the Demands of High-Performance Research Computing in Higher Ed

Managing energy consumption is a critical element of HPC.

Adam Stone writes on technology trends from Annapolis, Md., with a focus on government IT, military and first-responder technologies.

Universities must stay on the cutting edge of research to attract both students and grants. The rapid rise of artificial intelligence (AI) means they need more and better computing capabilities to support that research.

High-performance computing offers a way forward, and Vertiv has recently partnered with NVIDIA to provide access to a full portfolio of power and cooling solutions for compute-intensive applications.

To understand the role of HPC in the competitive higher ed landscape, EdTech spoke with Chris Collins, director of channel sales at Vertiv, and Andrew Rissler, NVIDIA business manager at CDW.

DIG DEEPER: Vertiv infrastructure solutions help universities get the most out of AI.

EDTECH: Why do schools need HPC?

COLLINS: Higher education is going to need high-performance computing for the sake of the research that it’s doing. From healthcare to engineering, they all want to be seen as leading research institutions in order to stay competitive.

RISSLER: With the right infrastructure, there’s a huge opportunity here for institutions to add more value, whether that be student retention or increased access to grants. There are billions of dollars in funding going into these institutions, and if they’re able to show results much more quickly, everybody’s going to win.

EDTECH: What challenges arise with a shift to HPC?

COLLINS: When you implement HPC for AI, there’s a need for resources and infrastructure that have never really been there before: more energy consumption, which equals more heat output. The amount of heat being generated, and the amount of electricity being consumed, is skyrocketing.

How do you keep the compute equipment protected from the increased heat and the load that is being put on that rack? It requires infrastructure modernization to enable the AI and HPC at a local level.

RISSLER: Another challenge is that all this infrastructure requires a lot of in-house expertise that they might not already have on their team. With traditional funding methods, adding extra head count in the form of full-time employees to help manage these systems can be really difficult.

Click the banner below for artificial intelligence insights from the world of higher education.

EDTECH: How does Vertiv help?

COLLINS: Vertiv specializes in critical power and cooling solutions. We provide innovative technologies such as liquid cooling systems, coolant distribution units and heat exchangers designed to handle the high-density heat generated by AI workloads.

Vertiv helps everything flow throughout the networks and beyond. Anywhere that you’ve got HPC AI, Vertiv will have solutions that assist in managing the thermal requirements and power consumption.

EDTECH: How does NVIDIA help, and what is CDW’s role?

RISSLER: NVIDIA contributes cutting-edge AI platforms and accelerated computing technologies. Their graphics processing units and architectures drive the compute-intensive applications that require robust infrastructure. In addition, they have focused on the software stack in the past several years. As the needs evolve, that software stack helps with the optimization of all these complex AI workloads. That is an integral piece of making sure that they’re getting value out of these very large-scale investments.

CDW partners with Vertiv and NVIDIA as an extension of the IT team, helping model power consumption, thermal requirements, and future proofing. We can bring in the experienced engineers who can help these institutions succeed.

EDTECH: What happens when those all come together?

COLLINS: For a lot of university IT teams, this is such a new marketplace. It makes sense that they would start by working with a trusted adviser.

RISSLER: Together, NVIDIA and Vertiv and CDW can provide both the reference architectures and in-house expertise to help universities tackle the challenges around HCP. Those reference architectures are important: We have premade plans, and we can help tailor those to the individual institution’s needs. We already know what works, and we come in together with a plan to help optimize and solve problems.

EDTECH: How do universities stand to benefit?

COLLINS: With AI and HPC right now, it’s almost like somebody invented the locomotive, but we don’t have train tracks that take it everywhere yet. For higher education institutions, if you’re waiting for the tracks to be built to you, you’re behind. Leading institution are the ones that are building the tracks now — making those investments and putting the infrastructure in place.

RISSLER: With the right hardware and software, efficiency is increasing exponentially year over year in areas such as genome analysis, modeling systems and advanced analytics. If you build the right infrastructure to take advantage of that acceleration, that’s the way to future proof your institution.

Brought to you by: