During the week of October 20th, in addition to the PWS Research, Innovation and Impact Summit and Expo , COS will host a workshop offered by NVIDIA.

Abstract: In this session, NVIDIA will begin by explaining what a GPU is and how supercomputers can transform the scale and impact of research within the College of Science. We will then introduce GPU-accelerated software containers, which enable easy acceleration of common scientific workflows. Next, we’ll explore build.nvidia.com, a platform that allows users to experiment with open-source large language and generative AI models in the cloud. From there, the session will transition to domain-specific applications:

• Protein Structure Prediction: We’ll discuss AlphaFold2 and demonstrate how to optimize its performance for predicting 3D protein structures.
• Materials Discovery: We will showcase GPU-accelerated tools used in Materials Discovery Research pipelines.
• Physics-Based AI: Finally, we'll highlight PhysicsNeMo, a framework for combining physics-based models with AI, enabling scalable and accelerated simulations that integrate real-world data and physical laws.

This session is designed to spark your interest in how GPUs can accelerate scientific research and help you get started with powerful, scalable tools tailored to your field.

Prerequisites: This session is open to all in the College of Science who have an interest in GPU accelerated computing for their research. No specific technical background is required. We ask that attendees bring a laptop if they are interested in following along with the demos.