HPE’s Peter Ungaro Weighs In On HPC To Work In Fight Against COVID-19

HPE & Cray

Every month HPE’s independent analyst council, of which I am a part of, gets together to discuss a special topic. Obviously, the ongoing COVID-19 crisis is on the top of many people’s minds, and the tech industry continues to play a big part in how we, as a society, are managing the fallout and looking for a solution. On that note, this month’s topic was HPE’s role in combatting the pandemic—particularly when it comes to HPC and the search for a cure. Last year HPE acquired the supercomputer manufacturer Cray, which has become a key part of the company’s HPC push. Now it is a key part of the research being done on the deadly pandemic. I wanted to share some of the takeaways from this interesting conversation.

Background on HPE, Cray and HPC

Peter Ungaro, HPE’s senior VP and GM of HPC and Mission Critical Solutions, began the discussion by providing some background on HPE and its HPC strategy and solutions. Notably, Ungaro was previously the CEO of Cray. He led off by singing the praises of the acquisition, touting the combination of HPE’s depth, breadth and reach with Cray’s R&D and focus on supercomputing. After the acquisition, the companies were able to integrate their product roadmaps quickly and are now moving aggressively to execute those plans.

Ungaro described HPC, on a high level, as the attempt to create digital representations of the physical world so that scientists and engineers might accelerate the time to insights around such fields as weather prediction, drug research/testing, or auto safety. Ungaro noted the shift from HPC’s traditional focus on math models, towards being able to perform similar functions on data models to create a “whole new environment for HPC and AI.” He lauded HPE’s efforts to build exascale machines capable of performing a billion calculations per second, noting the company’s three exascale supercomputers already announced worldwide. He also noted the potential for these technologies to be utilized not only in massive supercomputers, but even within a single cabinet or server.

HPE’s strategy, as Ungaro described it, is reimagining the supercomputer as a system that performs as a supercomputer but runs like a cloud. From a software perspective, Ungaro says this requires microservices, such as containers, to handle modeling, simulation, AI and analytics workflows. He also stressed the importance of rethinking how they connect and package all the different accelerators that exist today for performing specific actions in specific applications. From a packaging standpoint, these accelerators can create a lot of heat, and, according to Ungaro, many companies have a 500 watt or 1,000-watt processer on their roadmaps currently.

Ungaro also stressed the need for new interconnects, such as Cray’s Slingshot offering, which promises to connect what is inside the supercomputer to the datacenter, and to the devices that stream data into the machine. He went on to emphasize the need for more intelligent data management, and the ability to automatically tier different data to the different tiers of storage. This, Ungaro says, will drive down TCO for businesses. Ultimately, according to HPE’s vision, most datacenters will have something that resembles a supercomputer (but is not referred to as such) that will be leveraged for data transformation applications.

HPC and COVID-19

Ungaro then shifted the subject to the COVID-19 pandemic that has swept the globe and continues to disrupt life, work, and school as we know it. He shared that HPE is working as a part of the White House COVID-19 High Performance Computing consortium to find a vaccine for the virus. HPE is aiding the effort by providing free software and applications expertise to scientists to enable them to take advantage of these supercomputers. Additionally, the company has AI teams that are working with the COVID-19 open research data set as part of the White House’s Kaggle challenge.

HPE is also helping scientists with cryogenic electron microscopy (cryo-EM) research, by providing researchers with HPE Apollo’s “tightly integrated, scalable, high-capacity storage and compute structure.” According to Ungaro, cryo-EM has the potential to help speed a vaccine to market. Additionally, Ungaro touted the open-source release of PharML.Bind, a software AI program the company jointly developed with the Medical University of South Carolina (MUSC) that can be utilized in supercomputers and laptops alike. In addition to all these efforts, many HPE customers are utilizing the company’s supercomputers across the world for COVID-19 research.

Next, Dr. Jerome Baudry, of the University of Alabama in Huntsville, took over to go into more detail about HPE’s partnership with the university in the fight against COVID-19. HPE is providing Baudry’s team with full, free access to its Sentinel system, which is based on its Cray XC50 supercomputer, and an HPE dedicated technical staff. Baudry and his team will gain access to it through the cloud, as part of HPE’s alliance with Microsoft Azure, to help determine which natural compounds could potentially help fight the virus. Baudry shared an in-depth example of how HPC aids pharmaceutical discovery effort by helping scientists sift through “hundreds of thousands, if not millions” of candidates to find the right one. According to Baudry, after optimizing HPE’s machines, they were able to go from screening 10,000 potential candidates a day to a million. He went on to say that every other day, the team expects to have a list of 500 potential drugs against each of the virus’s proteins. All said and told, Baudry says Sentinel enables his team to compute workloads that would have typically taken a week in only half a day. With thousands of people dying daily from the virus, this sort of speed is absolutely of the essence.

Wrapping up

All in all, it was an informative discussion. I came away impressed with the way HPE has quickly integrated Cray into its strategy, and I am intrigued by HPE’s vision of a supercomputer that runs like a cloud that can be accessed in the datacenter. They’ve got their work cut out for them, but with HPE’s depth and breadth in the datacenter, and Cray’s impressive HPC chops, the vision has a lot of potential. On the topic of COVID-19, HPE clearly understands it is all hands-on deck. In these uncertain times, it’s some comfort to know that the tech industry is throwing its biggest, baddest supercomputers into the efforts to find a cure. Here’s to hoping it doesn’t take long.

Note: Moor Insights & Strategy writers and editors may have contributed to this article.