A guide to building HPC/AI systems in 2025

Advanced Clustering Technologies President and CTO Kyle Sheumaker delivered the keynote address at the Research Computing at Smaller Institutions (RCSI) on June 3 at Swarthmore College. His presentation, “A Guide to Building HPC/AI Systems in 2025,” is an important resource if you are facing the challenge of acquiring, configuring and maintaining an HPC system at a smaller institution.

We encourage you to request a copy of the presentation by submitting the contact form on this page.

In the presentation, Kyle explores a number of important topics including:

• Your hardware options in 2025 including the mainstream choices, AMD EPYC and Intel Xeon, as well as the alternatives including ARM and RISC-V
• Making an informed decision when it comes to GPUs. They are now commonplace for most HPC systems, and a requirement for large scale AI systems. They also are expensive, especially when you consider power and cooling requirements.
• How to choose interconnects based on your needs.
• The pros and cons of the primary types of storage solutions for HPC.
• Making sure your datacenter has the cooling and power infrastructure in place to support the HPC system you’re planning.
• Tips for configuring an HPC system that meets the needs of your users.

The presentation is a deep dive into the technical knowledge you need to have in place to successfully implement a new HPC system for your small institution.

Request your copy of the presentation here.