Deep Dive into GPU-accelerated Big Data and Data Science Technologies

Big data and data science applications are central to a wide range of business operations, and are at heart of countless products and services. Most of the utilities in this space, including scikit-learn, Pandas, NumPy, NetworkX, and Spark, have GPU-accelerated drop-in replacements (for example, RAPIDS-based cuDF, cuML, or cuGraph) frequently offering one, two, or even three orders of magnitude of acceleration versus their CPU-based counterparts. This acceleration not only allows data science/data engineering teams to iterate faster, but can have a profound impact on the selection of models, features, and dataset sizes that an organization uses, fundamentally changing their data science capabilities. We'll demonstrate how an existing data science team can migrate their code base, and show the order of magnitude of acceleration that can be expected. We'll conclude by discussing GPU-accelerated inference with Triton Inference Server.