Universities Expand Research Horizons with NVIDIA Systems, Networks

universities-expand-research-horizons-with-nvidia-systems,-networks

Just as the Dallas/Fort Truly worth airport grew to become a hub for tourists crisscrossing The usa, the north Texas area will be a gateway to AI if individuals at Southern Methodist College have their way.

SMU is setting up an NVIDIA DGX SuperPOD, an accelerated supercomputer it expects will ability jobs in machine learning for its sprawling metro community with a lot more than 12,000 pupils and 2,400 faculty and staff members.

It is just one of 3 universities in the south-central U.S. announcing strategies to use NVIDIA systems to change investigate into superior equipment.

Texas A&M and Mississippi State College are adopting NVIDIA Quantum-2, our 400 Gbit/2nd InfiniBand networking platform, as the spine for their hottest high-general performance desktops. In addition, a supercomputer in the U.K. has upgraded its InfiniBand community.

Texas Lassos a SuperPOD

“We’re the 2nd university in The us to get a DGX SuperPOD and that will place this neighborhood in advance in AI abilities to gasoline our diploma applications and corporate partnerships,” claimed Michael Hites, chief information officer of SMU, referring to a technique installed earlier this yr at the University of Florida.

A September report called the Dallas spot “hobbled” by a lack of key AI exploration. Ironically, the tale strike the local newspaper just as SMU was buttoning up its options for its DGX SuperPOD.

Previewing its initiative, an SMU report in March mentioned AI is “at the heart of digital transformation … and no sector of modern society will continue to be untouched” by the technology. “The potential for dramatic improvements in K-12 instruction and workforce growth is huge and will contribute to the sustained financial development of the area,” it extra.

SMU Ignite, a $1.5 billion fundraiser kicked off in September, will gasoline the AI initiative, assisting propel Southern Methodist into the top rated ranks of university investigate nationally. The college is employing a main innovation officer to support guide the work.

Crafting a Computational Crucible

It is all about the men and women, suggests Jason Warner, who manages the IT teams that assistance SMU’s scientists. So, he hired a seminal team of details science professionals to staff members a new center at SMU’s Ford Corridor for Study and Innovation, a hub Warner phone calls SMU’s “computational crucible.”

Eric Godat qualified prospects that team. He acquired his Ph.D. in particle physics at SMU modeling nuclear composition making use of knowledge from the Big Hadron Collider.

Now he’s supporting hearth up SMU’s learners about prospects on the DGX SuperPOD. As a very first step, he asked two SMU learners to establish a miniature model of a DGX SuperPOD employing NVIDIA Jetson modules.

“We needed to give individuals — primarily individuals in nontechnical fields who haven’t finished AI — a perception of what’s coming,” Godat explained.


SMU's Jetson SuperPOD
SMU undergrad Connor Ozenne aided create a miniature DGX SuperPOD that was highlighted in SMU’s yearly report. It takes advantage of 16 Jetson modules in a cluster pupils will benchmark as if it ended up a Prime500 procedure.

The comprehensive-sized supercomputer, produced up of 20 NVIDIA DGX A100 devices on an NVIDIA Quantum InfiniBand community, could be up and operating as early as January many thanks to its Lego-like, modular architecture. It will produce a whopping 100 petaflops of computing energy, more than enough to give it a respectable slot on the Prime500 checklist of the world’s speediest supercomputers.

Aggies Faucet NVIDIA Quantum-2 InfiniBand for ACES

About 200 miles south, the high performance computing middle at Texas A&M will be amongst the initially to plug into the NVIDIA Quantum-2 InfiniBand system. Its ACES supercomputer, developed by Dell Systems, will use the 400G InfiniBand community to connect researchers to a mix of 5 accelerators from four distributors.

NVIDIA Quantum-two makes certain “that a solitary task on ACES can scale up making use of all the computing cores and accelerators.  Besides the obvious 2x soar in throughput from NVIDIA Quantum-one InfiniBand at 200G, it will offer enhanced total cost of ownership, beefed up in-community computing options and greater scaling,” stated Honggao Liu, ACES’s principal investigator and venture director.

Texas A&M presently provides scientists obtain to accelerated computing in 4 units that contain far more than 600 NVIDIA A100 Tensor Main and prior-era GPUs. Two of the 4 systems use an earlier edition of NVIDIA’s InfiniBand technologies.

MSU Rides a 400G Coach

Mississippi State University will also tap the NVIDIA Quantum-two InfiniBand system. It’s the network of selection for a new system that dietary supplements Orion, the most significant of four clusters MSU manages, all using before versions of InfiniBand.

Equally Orion and the new technique are funded by the U.S. Nationwide Oceanic and Atmospheric Administration (NOAA) and built by Dell. They conduct function for NOAA’s missions as perfectly as investigate for MSU.

Orion was stated as the fourth premier tutorial supercomputer in The us when it debuted on the Leading500 listing in June 2019.

“We’re utilizing InfiniBand in 4 generations of supercomputers in this article at MSU so we know it’s the two impressive and mature to operate our huge employment reliably,” reported Trey Breckenridge, director of large functionality computing at MSU.

“We’re including a new technique with NVIDIA Quantum-2 to stay at the leading edge in HPC,” he included.

Quantum Nets Go over the Uk

Across the pond in the U.K., the Information Intense supercomputer at the University of Leicester, known as the DIaL process, has upgraded to NVIDIA Quantum, the 200G edition of InfiniBand.

“DIaL is specifically created to deal with the complicated, info-intense queries which must be answered to evolve our comprehending of the universe all over us,” reported Mark Wilkinson, professor of theoretical astrophysics at the University of Leicester and director of its HPC heart.

“The intense demands of these professional workloads count on the unparalleled bandwidth and latency that only InfiniBand can present to make the research probable,” he explained.

DIaL is a person of four supercomputers in the U.K.’s DiRAC facility using InfiniBand, such as the Tursa system at the University of Edinburgh.

InfiniBand Shines in Analysis

In a technical evaluation, scientists discovered Tursa with NVIDIA GPU accelerators on a Quantum community delivered 5x the effectiveness of their CPU-only Tesseract method employing an alternate interconnect.

Application benchmarks demonstrate 16 nodes of Tursa have two times the functionality of 512 nodes of Tesseract. Tursa delivers 10 teraflops/node applying 90 p.c of the network’s bandwidth at a major enhancement in efficiency for every kilowatt about Tesseract.

It is another instance of why most of the world’s Best500 units are employing NVIDIA systems.

For more, check out our specific tackle at SC21 either live on Monday, Nov. 15 at 3 pm PST or afterwards on desire. NVIDIA’s Marc Hamilton will offer an overview of our most recent information, improvements and technologies, followed by a dwell Q&A panel with NVIDIA specialists.

Leave a comment

Your email address will not be published.


*