A lot has been made of the inordinate share of the global power supply being consumed by data centers and other high-performance computing resources.

With sustainability being a major topic in the HPC space, it’s worth analyzing the steps being taken at the enterprise level to ensure that rapid innovation doesn’t create more harm than good.

“If you’re looking at the overall consumption of power from data centers, you will see that from the entire worldwide figure it’s around 1.5%,” said Serban Zirnovan (pictured, center), senior director of of ISG solutions, Dell EMEA, at Dell Technologies Inc. “One might think the figure doesn’t look big, but in reality it is enormous. Based on several assessments done by Hyperion Research, on-premises HPC represents around one-fifth of the overall server implementation.”

Zirnovan; Matt Foley (left), director of EMEA field application engineering at Advanced Micro Devices, Inc.; and Guy D’Hauwers (right), vice president of cloud, high-performance computing and artificial intelligence at atNorth ehf, spoke with theCUBE industry analyst Dave Vellante at the ISC High Performance event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed sustainability as paramount to the continued survival of HPC. (* Disclosure below.)

Unpacking the numbers and breaking down the topic

Not only is the power requirement high for high-performance computing, but it is steadily increasing to the point of outstripping the cost of the infrastructure itself. This, coupled with energy costs, is mounting a considerable challenge for the industry, according to Zirnovan.

“It’s really important to understand how to build and drive it because it represents an immense amount of power, which comes with also a very large cost these days,” he explained. “We used to say half of the cost [of HPC] was for the acquisition of the hardware, and the other half of the cost of running an HPC represents the power. I think these days it’s started to be slightly different.”

As a predominantly hardware-focused company, HPC represents a large chunk of AMD’s business. The company, therefore, is working to make meaningful strides in achieving sustainability through “heterogenous computing,” according to Foley.

“What we really see here is a need for heterogeneous computing where we take the problems and decompose them into different ways that acceleration is better suited to solve,” Foley explained. “By doing that, we believe we can achieve step-function improvements in terms of efficiency and sustainability instead of mere percentages.”

AtNorth provides power-efficient and environmentally responsible data center services to a primarily pan-Nordic user base. The company’s approach to sustainability relies on renewable energy sources and a streamlined technology stack, according to D’Hauwers.

“We have more and more customers who migrate away from the public cloud because they have resources suited for general-purpose computing and occasional usage,” he said. “But for HPC, which tends to be used on a constant basis and must always be optimized, it becomes way too expensive. They come to us for a total cost of ownership reason but, most of all, for sustainability reasons.”

Practically speaking, denser power supplies generally result in more efficient computing racks. However, several challenges, such as cooling and maintenance, plague these ideal systems, Foley added.

“From our perspective here, we certainly see that the more dense you make the power, the more efficient the system is,” he said. “So that 125-kilowatt rack is a very efficient rack, but it’s a difficult physical challenge to power and cool it.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the ISC High Performance event:

(* Disclosure: Dell Technologies Inc. sponsored this segment of theCUBE. Neither Dell Technologies nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE