High-Performance Computing

Milwaukee School of Engineering Supercomputer to Support AI Education and Research

• By Rhea Kelly
• 10/23/19

A new supercomputer at the Milwaukee School of Engineering will serve both research and instruction in artificial intelligence and deep learning for the university's computer science program. MSOE partnered with Microway to build the custom cluster, an NVIDIA DGX POD-based machine designed to support modern AI development. The DGX POD reference architecture is based on NVIDIA's DGX SATURN V AI supercomputer used for internal research and development for autonomous vehicles, robotics, graphics and more.

Technical specs include:

• Three NVIDIA DGX-1 AI systems with NVIDIA V100 Tensor Core GPU accelerators;
• Twenty Microway NumberSmasher Xeon + NVIDIA T4 GPU teaching compute nodes;
• 100G networking;
• NVIDIA NGC deep learning containers;
• NVIDIA DGX Software stack;
• Access to the NGC online registry of software stacks, pre-trained models and model training scripts; and
• High-performance storage arrays and a larger general-purpose storage pool from NetApp.

MSOE students will be able to access the supercomputer via web browser and "start a DGX-1 or NVIDIA T4 GPU deep learning session with the click of a button" — with no need to understand command line interfaces and workload managers, according to a news announcement. "Unlike many university programs in which students' access to supercomputers is usually limited to graduate students in computer labs, this configuration gives undergraduate students at MSOE supercomputer access in the classroom, enabling training of the next AI workforce."