AI Infrastructure Development at Sapienza University

This case study details the implementation of NVIDIA DGX A100 at the Department of Computer, Control and Management Engineering (DIAG) at Sapienza University of Rome. Established in 1983, DIAG focuses on advanced multidisciplinary research and was awarded funding through the 'Departments of Excellence' initiative to enhance its research capabilities. The department required a powerful computational platform for deep learning applications, aiming to create a private cloud environment for machine learning. The proposed solution, NVIDIA DGX A100, offers significant compute density and performance, integrating eight Nvidia A100 GPUs. It supports Multi Instance GPU (MIG) technology, allowing for flexible resource allocation and isolation. The case study outlines the advantages of the DGX A100, including simplified infrastructure design, enhanced performance, and integrated security. It also highlights the collaboration between DIAG and beanTech, emphasizing the importance of understanding academic needs in AI infrastructure development. The partnership aims to support ongoing advancements in AI research.