Bell, one of the world’s most powerful supercomputers, now available for faculty use

November 16, 2020

Purdue’s newest community cluster supercomputer, Bell, has debuted at number 431 on the Top500 list of the world’s most powerful supercomputers, announced today at the SC20 supercomputing conference.

With a top processing speed of 1.624 petaFLOPs, Bell is nearly twice as fast as Purdue’s Conte cluster, which was number 28 on the list and the nation’s fastest campus supercomputer at the time it was built just seven years ago. Bell is also number 44 on the Green500 ranking of the world’s most energy-efficient supercomputers.

Bell is the ninth Purdue community cluster to make the Top 500 list. All together, Purdue’s community clusters provide 5.3 petaFLOPs to drive research at Purdue.

"Bell's place on the list of the world's 500 most powerful supercomputers reflects Purdue's commitment to offering our researchers access to the very best campus computing resources," said Theresa Mayer, Purdue’s executive vice president for research and partnerships.

Bell is optimized for a broad set of science and engineering applications, and consists of 448 Dell compute nodes featuring two 64-core AMD Epyc “Rome” processors with 256 GB of memory. The cluster also includes 8 large memory nodes, each with 1 TB of RAM, and a GPU subsystem with 16 AMD Instinct GPUs to accelerate workflows using AI and machine learning. The system also features a 5 PB Lustre parallel filesystem, with its top performance 50% faster than the storage powering the Brown community cluster.

Purdue researchers can now purchase capacity in Bell through ITaP Research Computing’s cluster orders website. The cluster will have a six year lifespan and delivers the lowest cost per FLOP and per core of any community cluster system to date.

Purdue’s community clusters are named in honor of trailblazing Boilermakers, and Bell is no exception. Bell’s namesake, Clara Bell Sessions, was a nursing professor and the director of continuing education in the School of Nursing, who helped establish Purdue’s Minority Student Nurses Association, now known as the Diversity in Nursing Association, and the Minority Faculty Fellows programs.

Purdue’s community cluster program began in 2004 with just four faculty partners. Today the program has more than 200 active faculty partners from all three Purdue campuses, all of Purdue’s primary colleges and schools and 60 different departments. In 2019, ITaP Research Computing delivered more than 350 million computational hours to Purdue researchers, who account for 55% of Purdue’s FY20 sponsored research spending.

Community cluster supercomputing makes more computing power available for Purdue researchers than faculty and campus units could individually afford, and provides this crucial service at the lowest cost to the institution. ITaP Research Computing installs, administers and maintains the community clusters, providing reliability, security, software installation and expert user support.

“Bell allows my group to quickly prototype ideas while developing new methodologies,” says Brett Savoie, Charles Davidson Assistant Professor of Chemical Engineering. “When we are developing algorithms, it is vital to have access to a resource that allows us to quickly iterate and improve our software. Bell is distinct in that its scale gives us access to much more throughput than is typical for a campus supercomputer, and it’s a game changer when it comes to planning simulations.”

To learn more about the Bell community cluster or other Research Computing resources, contact Preston Smith, executive director of research computing, psmith@purdue.edu or 49-49729.

Originally posted: November 16, 2020  8:06pm