Bio: Zoe Ryan is a Solutions Architect at NVIDIA working with higher education institutions and researchers. She helps them utilize their GPUs by accelerating their HPC workloads and adopting NVIDIA Omniverse in their simulation pipelines.
Title: Four Ways to GPU Computing
Abstract: In this session, we will discuss four ways you can start using NVIDIA GPUs to accelerate your computing work. We will cover the basics of how GPU computing works, and the types of speed ups you can see across a variety of workloads when you parallelize with a GPU. Then we will dive into the four methods of GPU utilization, discussing the technical knowledge required for each option. First we will look at GPU accelerated applications, which is a great way to get started with GPUs by utilizing existing models, frameworks, and toolkits. Then we will discuss the drop-in library replacements that offer GPU acceleration for a wide variety of workloads, with minimal code changes or in depth GPU knowledge. Our next option is to utilize portable programming models which can help take your custom CPU code and port it to a GPU implementation. Our last, and most involved option is working directly with GPU accelerated programming languages like CUDA. This session will meet you at your level of GPU knowledge, and will give you a variety of options for advancing your parallel computing work!