Explore GPU-accelerated data analytics in Python through this comprehensive workshop from PyCon US. Learn about RAPIDS, an open-source library suite that enables data scientists to leverage GPU acceleration for ETL, machine learning, and graph analytics workloads using familiar Python APIs. Discover how to speed up compute times and increase model iteration without requiring specialized GPGPU programming knowledge. Dive into component libraries like cuDF, cuML, cuGraph, cuSignal, cuSpatial, and the Cyber Log Accelerator. Gain insights into data processing evolution, real-world benefits of GPU acceleration, and integration with the broader open-source GPU data science ecosystem. Suitable for those with basic data science knowledge, no prior GPU programming experience required.
Overview
Syllabus
Intro
RAPIDS GPU Accelerated Data Analytics in Python
Data Processing Evolution Faster data access, less data movement
Data Movement and Transformation What if we could keep data on the GPU?
Faster Speeds, Real-World Benefits
Speed, Ease of use, and Iteration The Way to Win a Data Science
Open Source Data Science Ecosystem Familiar Python APIs
RAPIDS End-to-End Accelerated GPU Data Science
RAPIDS GPU Accelerated data wrangling and feature engineering
ETL Technology Stack
Benchmarks: single-GPU Speedup vs. Pandas
ETL: the Backbone of Data Science String Support
Extraction is the Cornerstone CUDF I/O for Faster Data Loading
ML Technology Stack
RAPIDS matches common Python APIs
RAPIDS RELEASE SELECTOR
Forest Inference Taking models from training to production
Goals and Benefits of cuGraph Focus on Features and User Experience
Graph Technology Stack
Algorithms GPU accelerated Network
Multi-GPU PageRank Performance PageRank portion of the HiBench benchmark suite
cuSpatial Technology Stack
Speed of Light Performance - V100
Efficient Memory Handling
Marriage of Deep Learning and RF Data
CLX Cyber Log Accelerators
CLX Components Notebook Examples, SIEM Integrations, Workflows, Primitives
Taught by
PyCon US
Related Courses
-
RAPIDS - The Platform Inside and Outside - Joshua Patterson - ODSC East 2019
-
Rapids Machine Learning Training and Feature Engineering in GPU with CuML and CuDF
-
Distributed Multi-GPU Computing with Dask, CuPy and RAPIDS
-
Spark RAPIDS ML - GPU Accelerated Distributed Machine Learning in Spark Clusters
-
GPU-Acceleration for Pandas, NetworkX, and Apache Spark MLlib - Accelerating Data Science Tools
-
Advancing GPU Analytics with RAPIDS Accelerator for Apache Spark and Alluxio
