RAPIDS- Open Source GPU Data Science
summary
RAPIDS claims end-to-end GPU data science that keeps data on the GPU so ETL, joins, grouping and ML avoid repeated copies and run orders of magnitude faster: a Fannie Mae mortgage workload drops from 3,221s on 100-node Spark to 213s on GPUs and 100x faster than Spark is a stated target. The stack uses cuDF replaces pandas, cuML mirrors scikit‑learn, Dask plus UCX gives ~20x cluster comms gains, CSV ingest hits 2.3s vs 29s and PageRank runs ~31.8s versus ~5,700s, with multi‑GPU scaling and readers that stream straight into GPU memory.
AI-written summary · about
chapters (13)
- 0:00RAPIDS introduction
- 1:38Big data and GPU history
- 7:56RAPIDS performance gains
- 9:23RAPIDS stack and API
- 13:00Data scientist workflow
- 14:29ETL in RAPIDS
- 18:11Fast machine learning with RAPIDS
- 22:05RAPIDS future
- 22:51Graph analytics cuGraph
- 28:20Deep learning in RAPIDS
- 30:20RAPIDS ecosystem
- 31:44Getting started with RAPIDS
- 34:35Q&A
related talks
34mRAPIDS: How to accelerate your data science pipeline by orders of magnitude
01h 22mHigh-Performance Data Science at Scale with RAPIDS, Dask, and GPUs
41mAccelerating Data Science with RAPIDS
01h 08mGPU Data Science with Rapids
20mEnd-to-End Data Science on GPUs
01h 36mRAPIDS: GPU-Accelerated Data Science for PyData Users
39mAccelerating Data Science with RAPIDS
01h 12mGPU-Accelerated Data Analytics in Python