Universities Expand Research Horizons with NVIDIA Systems, Networks

SMU is installing an NVIDIA DGX SuperPOD, an accelerated supercomputer it expects will power tasks in device knowing for its vast metro neighborhood with more than 12,000 trainees and 2,400 faculty and personnel.

Simply as the Dallas/Fort Worth airport ended up being a center for travelers crisscrossing America, the north Texas area will be an entrance to AI if folks at Southern Methodist University have their way.

Texas A&M and Mississippi State University are adopting NVIDIA Quantum-2, our 400 Gbit/second InfiniBand networking platform, as the foundation for their latest high-performance computers. In addition, a supercomputer in the U.K. has updated its InfiniBand network.

Its one of 3 universities in the south-central U.S. announcing strategies to use NVIDIA innovations to shift research into high equipment.

Texas Lassos a SuperPOD

A September report called the Dallas location “hobbled” by a lack of significant AI research study. Ironically, the story hit the local paper just as SMU was buttoning up its plans for its DGX SuperPOD.

” Were the second university in America to get a DGX SuperPOD which will put this neighborhood ahead in AI abilities to sustain our degree programs and corporate collaborations,” said Michael Hites, primary information officer of SMU, describing a system set up previously this year at the University of Florida.

SMU Ignite, a $1.5 billion fundraising event began in September, will fuel the AI effort, assisting propel Southern Methodist into the top ranks of university research nationally. The university is employing a primary innovation officer to help guide the effort.

Previewing its effort, an SMU report in March said AI is “at the heart of digital change … and no sector of society will stay untouched” by the technology. “The potential for significant enhancements in K-12 education and workforce development is massive and will contribute to the continual economic growth of the region,” it added.

Crafting a Computational Crucible

Its everything about individuals, says Jason Warner, who handles the IT teams that support SMUs researchers. He hired an influential group of information science professionals to staff a brand-new center at SMUs Ford Hall for Research and Innovation, a center Warner calls SMUs “computational crucible.”

Eric Godat leads that group. He earned his Ph.D. in particle physics at SMU modeling nuclear structure using data from the Large Hadron Collider.

” We desired to provide people– especially those in nontechnical fields who have not done AI– a sense of whats coming,” Godat said.

The full-sized supercomputer, made up of 20 NVIDIA DGX A100 systems on an NVIDIA Quantum InfiniBand network, might be up and running as early as January thanks to its Lego-like, modular architecture. It will deliver a tremendous 100 petaflops of computing power, enough to offer it a reputable slot on the TOP500 list of the worlds fastest supercomputers.

SMU undergrad Connor Ozenne helped construct a miniature DGX SuperPOD that was included in SMUs annual report. If it were a TOP500 system, it uses 16 Jetson modules in a cluster trainees will standard as.

Now hes assisting fire up SMUs trainees about chances on the DGX SuperPOD. As a very first action, he asked 2 SMU trainees to develop a miniature model of a DGX SuperPOD using NVIDIA Jetson modules.

Aggies Tap NVIDIA Quantum-2 InfiniBand for ACES

NVIDIA Quantum-2 guarantees “that a single job on ACES can scale up utilizing all the computing cores and accelerators. Besides the apparent 2x jump in throughput from NVIDIA Quantum-1 InfiniBand at 200G, it will supply better overall cost of ownership, beefed up in-network computing functions and increased scaling,” stated Honggao Liu, ACESs primary private investigator and project director.

About 200 miles south, the high efficiency computing center at Texas A&M will be amongst the very first to plug into the NVIDIA Quantum-2 InfiniBand platform. Its ACES supercomputer, built by Dell Technologies, will use the 400G InfiniBand network to link researchers to a mix of five accelerators from four vendors.

Texas A&M currently gives scientists access to accelerated computing in 4 systems that include more than 600 NVIDIA A100 Tensor Core and prior-generation GPUs. Two of the four systems utilize an earlier version of NVIDIAs InfiniBand technology.

MSU Rides a 400G Train

Mississippi State University will also tap the NVIDIA Quantum-2 InfiniBand platform. Its the network of option for a brand-new system that supplements Orion, the biggest of 4 clusters MSU manages, all using earlier versions of InfiniBand.

Orion was noted as the 4th largest scholastic supercomputer in America when it debuted on the TOP500 list in June 2019.

” Were adding a new system with NVIDIA Quantum-2 to remain at the leading edge in HPC,” he added.

” Were using InfiniBand in 4 generations of supercomputers here at MSU so we understand its both fully grown and effective to run our big tasks reliably,” said Trey Breckenridge, director of high performance computing at MSU.

Both Orion and the new system are moneyed by the U.S. National Oceanic and Atmospheric Administration (NOAA) and built by Dell. They carry out work for NOAAs missions in addition to research study for MSU.

Quantum Nets Cover the UK

DIaL is among 4 supercomputers in the U.K.s DiRAC facility utilizing InfiniBand, consisting of the Tursa system at the University of Edinburgh.

” The extreme requirements of these specialist work depend on the exceptional bandwidth and latency that only InfiniBand can provide to make the research study possible,” he said.

Across the pond in the U.K., the Data Intensive supercomputer at the University of Leicester, called the DIaL system, has actually upgraded to NVIDIA Quantum, the 200G version of InfiniBand.

” DIaL is particularly developed to deal with the complex, data-intensive concerns which should be answered to develop our understanding of the universe around us,” said Mark Wilkinson, professor of theoretical astrophysics at the University of Leicester and director of its HPC.

InfiniBand Shines in Evaluation

In a technical assessment, scientists found Tursa with NVIDIA GPU accelerators on a Quantum network provided 5x the performance of their CPU-only Tesseract system utilizing an alternative interconnect.

Its another example of why many of the worlds TOP500 systems are using NVIDIA technologies.

Application benchmarks show 16 nodes of Tursa have twice the efficiency of 512 nodes of Tesseract. Tursa provides 10 teraflops/node using 90 percent of the networks bandwidth at a significant improvement in performance per kilowatt over Tesseract.

For more, enjoy our special address at SC21 either reside on Monday, Nov. 15 at 3 pm PST or in the future need. NVIDIAs Marc Hamilton will provide an introduction of our newest news, innovations and developments, followed by a live Q&A panel with NVIDIA professionals.

Leave a Reply

Your email address will not be published.