New all-GPU research supercomputer optimized for AI and machine learning, honors trailblazing Purdue engineering professor
Purdue’s latest research supercomputer will be optimized for machine learning, artificial intelligence and other graphics processing unit-based applications.
The new cluster, known as Gilbreth, includes nodes featuring 16-core Intel Xeon Sky Lake processors with two 16 GB Nvidia Tesla P100 GPUs per node, 192 GB of RAM and 100 Gbps EDR Infiniband interconnects. Through a partnership with the Office of the Executive Vice President for Research and Partnerships, Gilbreth will also include three model-training nodes with four 32 GB V100 GPUs per node – each with NVLink fabric and 8TB of local flash storage. Nodes from the Halstead-GPU and Brown-GPU clusters will be combined with and integrated into Gilbreth.
Additionally, the system will feature a multi-PB Lustre parallel filesystem, and a shared flash-based storage system to allow for processing of large datasets.
As with previous GPU nodes offered by ITaP Research Computing, Gilbreth nodes are available for purchase (at $14,144 for five years or $2,829 annually), but are also offered through a subscription that gives researchers shared access to a large pool of nodes for $1,599 a year per lab.
Gilbreth’s namesake is Lillian Moller Gilbreth, an industrial engineer and efficiency expert, who became Purdue’s first female engineering professor when she joined the faculty in 1935. She was promoted to full professor in 1940 and remained at Purdue until her retirement in 1948.
Lillian Gilbreth’s research focused on combining psychology and engineering to improve efficiency in the workplace and home, and she pioneered the field now known as ergonomics. To improve household efficiency, she invented a number of kitchen devices, including the foot pedal trash can, refrigerator door shelves and the electric mixer.
She was the first woman elected to the National Academy of Engineering, the second female member of the American Society of Mechanical Engineers and the first woman to receive the Hoover Medal.
Her family life with her husband and research collaborator Frank and their 12 children is the subject of the autobiographical novels “Cheaper by the Dozen” and “Belles on Their Toes,” which were written by two of their children. The books describe how the Gilbreths applied their efficiency studies in their home.
Gilbreth will be the tenth research computing system offered to Purdue faculty in as many years through the Community Cluster Program. The program now has almost 180 active faculty partners from all three Purdue campuses, all of Purdue’s primary colleges and schools and 50 different departments. ITaP Research Computing delivered over 300 million computational hours to community cluster partners in 2017.
To learn more about the new Gilbreth supercomputer or the Community Cluster Program, contact Preston Smith, ITaP’s director of research services and support, psmith@purdue.edu or 49-49729.
Writer: Adrienne Miller, science and technology writer, Information Technology at Purdue (ITaP), 765-496-8204, mill2027@purdue.edu
Last updated: October 22, 2018