From hours to seconds: How NVIDIA RAPIDS is Revolutionizing Big Data Processing

About this talk

This talk explores the benefits of NVIDIA RAPIDS, a GPU-accelerated data science framework, compared to traditional CPU-based libraries like Pandas. By benchmarking performance on real-world datasets, we uncover the conditions where RAPIDS excels and demonstrate its potential for accelerating large-scale data workflows. Attendees will learn practical strategies for optimizing workflows, understanding performance trade-offs, and leveraging dataset scaling. The session requires a basic understanding of Python, familiarity with data processing tools like Pandas, and access to a system with a GPU or a Google Collab account.