GPU Data Science with Rapids
summary
RAPIDS shows concrete work: GPU-accelerated ETL and ML using cuDF & cuML, Arrow in GPU memory, Dask for multi‑GPU scaling and CUDA IPC/UCX to avoid copies. A 400 GB Fannie mortgage corpus gets joined and feature‑engineered on GPUs, benchmarks report up to 90× single‑GPU speedups and 20–25× density wins vs large CPU clusters; XGBoost, k‑means, UMAP and a GPU string library are implemented, parquet/CSV loaders exist, and GCP DGX instances are provided for hands‑on testing.
AI-written summary · about
chapters (6)
- 0:00GPU Data Science with RAPIDS
- 6:11RAPIDS overview and goals
- 16:41ML and graph analytics in RAPIDS
- 31:18KUDU GPU DataFrame deep dive
- 41:59Future roadmap and community
- 59:08Q&A and demo
related talks
41mAccelerating Data Science with RAPIDS
38mRAPIDS- Open Source GPU Data Science
01h 22mHigh-Performance Data Science at Scale with RAPIDS, Dask, and GPUs
34mRAPIDS: How to accelerate your data science pipeline by orders of magnitude
39mAccelerating Data Science with RAPIDS
20mEnd-to-End Data Science on GPUs
01h 12mGPU-Accelerated Data Analytics in Python
01h 36mRAPIDS: GPU-Accelerated Data Science for PyData Users