Head of the Class: Indiana University Selects NVIDIA V100 to Power Nation’s Fastest University Supercomputer for AI

University taps Cray for a next-gen Shasta supercomputer running NVIDIA V100 GPUs.
by Scott Martin

Indiana University has signed up to become home to the nation’s fastest university-owned supercomputer in an effort to support its AI research.

The university plans to be the first academic customer in line for a Cray Shasta supercomputer sporting NVIDIA V100 Tensor Core GPUs.

The Cray Shasta supercomputer, dubbed Big Red 200, is expected to be fully operational by IU’s January 20, 2020, bicentennial anniversary.

Cray’s Shasta system promises to pack 5.9 petaflops, a nearly 300x performance leap from the university’s original Big Red supercomputer from 15 years ago.

IU plans to deploy the new Shasta system for university efforts in AI, machine learning and data analytics. The supercomputer will help support the university’s advancement of AI in education, cybersecurity, medicine, environmental science and other areas.

Boosting Scientific Discovery

The performance boost will play a key role in supporting scientific research and enabling the next wave of discoveries at IU as well as the university’s Grand Challenges initiatives, said Matt Link, associate vice president for research technologies at IU.

Big Red 200 will support the university’s Precision Health Initiative, a Grand Challenge intended to improve prevention, treatment and health outcomes of human diseases.

The Cray Shasta supercomputer at IU will boost the work of researchers in healthcare, physics and astronomy in particular, said Link.

“Big Red 200 will mark a significant change in how we support the research engine at IU,” he said.

Powering Disease Detection

The new supercomputer could significantly accelerate timelines for IU researchers, he said. This holds vast promise to enable researchers to make breakthroughs in medical imaging, including for dementia and Alzheimer’s studies.

IU’s new system will provide service to about 130,000 students and 20,000 faculty members across its campuses, with support from more than 100 on staff, he said.

Also, IU’s new Cray Shasta system will enable university researchers to easily scale projects using resources beyond those on campus. That’s because larger versions of the Cray Shasta systems are planned for deployment at the Department of Energy — under the Exascale Computing Project — which will allow researchers to take AI workloads there.

Accelerating University Research

Access to GPU-accelerated high performance computing plays an increasingly important role in the research at universities. It enables postdoctoral researchers to tackle big data challenges — like precision medicine — using deep learning.

IU garnered more than $185 million in research grant awards in 2018 that were supported by the university’s high performance computing systems.

The Big Red 200 system itself will be funded by revenue from federal contracts and grants, according to the university.

Big Red 200 will replace Big Red II, which handled more than $500 million in grants and research contracts in the roughly 5 years of service and cost a small fraction of that sum.


Photo and credit: Indiana University; Emily Sterneman