RAPIDS

RAPIDS™, part of NVIDIA CUDA-X, is an open-source suite of GPU-accelerated data science and AI libraries with APIs that match the most popular open-source data tools. It accelerates performance by orders of magnitude, at scale, across data pipelines.


Get Started
An optimized hardware-to-software stack for the entire data science pipeline
An optimized hardware-to-software stack for the entire data science pipeline.

RAPIDS Benefits

Massive Speedups

Faster pipelines enable more experimentation, improving outcomes.


Run Benchmarks Yourself

Easy to Adopt

Zero-code-change accelerators and familiar Python APIs quickly accelerate existing workloads.


Explore Modular Libraries

Flexible Open-Source Platform

With 100+ software integrations, RAPIDS promotes collaboration.


Explore the Ecosystem

Runs Everywhere

RAPIDS runs on all major clouds, on your local machine, or on premises.


See Deployment Options

Accelerating Data Science

With libraries that speed up widely adopted operations and algorithms, RAPIDS helps reduce time to insight as questions evolve.

150x

Faster Pandas With cuDF

* Benchmark on Groupy advanced operation (5GB) DuckDB Data Benchmark
HW: Intel Xeon Platinum 8480CL CPU and NVIDIA Grace Hopper GPU SW: pandas v1.5 and cudf.pandas v23.10

13x

Faster Polars With cuDF

* Benchmark on PDS-H queries
HW: Intel Xeon W9-3495X CPU and NVIDIA H100 80GB (1x GPU)
SW: Polars v1.4.1

48x

Faster NetworkX With cuGraph

* Benchmark on PageRank with synthetic dataset having ~16,384 vertices and ~524,288 edges
HW: Intel Xeon Platinum 8480CL CPU and NVIDIA H100 80GB (1x GPU)
SW: NetworkX v3.2 and cuGraph v23.10

50x

Faster Scikit-Learn With cuML

* Benchmark on UMAP-Unsupervised on 100,000 samples and 256 features
HW: Intel Xeon Platinum 8480CL CPU and NVIDIA H100 80GB (1x GPU)
SW: scikit-learn v1.3 and cuML v23.10

See benchmarks at rapids.ai


Flexible Across Data Workloads

With a distinctive, modular, interoperable selection of libraries that smoothly plug and play into pipelines and applications, RAPIDS simplifies the development process.


RAPIDS keeps data science pipelines running smoothly at any scale.

A flowchart showing how RAPIDS keeps data science pipelines running smoothly at any scale

Data Preparation

Seamlessly accelerate data analytics for tabular datasets, graph databases, or the Spark framework with your existing tools.


Learn More About
Accelerated Data Analytics

Machine Learning

Boost model training speed with an API that closely follows scikit-learn.


Learn More About
Machine Learning

Deep Learning

Support efficient graph neural networks training with DGL and PyG.


Learn More About Deep Learning

MLOps

Deploy high-performance machine learning inference with cuML and NVIDIA Triton™.


Learn More About
Inference and Deployment

Inspired by the most popular open-source data tools, RAPIDS libraries adapt to your workflow.

Data Preprocessing: cuDF

Accelerate dataframes for efficiently processing hundreds of millions of records.


Explore Pandas Accelerator Mode Polars GPU Engine

Big Data Processing: RAPIDS Accelerator for Apache Spark

Accelerate your existing Apache Spark applications with minimal code changes.


Learn More About
GPU-Accelerated Spark
Go to GitHub

Machine Learning: cuML

Execute machine learning algorithms on CPUs and GPUs with an API that closely follows the scikit-learn API.


Explore Docs

Graph Analytics: cuGraph

Quickly navigate graph analytics libraries with a python API that follows NetworkX.


Explore Docs

Vector Search: cuVS

Apply cuVS algorithms to accelerate vector search, including world-class performance from CAGRA.


Explore Docs

Scale RAPIDS: Dask-RAPIDS

Expand data science pipelines to multiple nodes with RAPIDS on Dask.


Go to GitHub

Visualization: cu-x-filter

Create interactive data visuals with multidimensional filtering of over 100-million-row tabular datasets.


Explore Docs

Image: cuCIM

Accelerate input/output (IO), computer vision, and image processing of n-dimensional, especially biomedical, images.


Explore Docs

Leverage purpose-built NVIDIA frameworks and guides to build accelerated applications for common and high-impact use cases.

Data Engineering

Revolutionize data management and preprocessing with the RAPIDS Accelerator for Spark.


Learn More About Scaled Data Processing

Time-Series Forecasting

Accelerate time-series modeling from feature engineering to forecasting.


Learn More About Time-Series Forecasting

Recommendation Systems

Build high-performing recommender systems at scale with NVIDIA Merlin™.


Learn More About Recommenders

AI Cybersecurity

Filter, process, and classify real-time data in optimized AI pipelines to quickly detect cyberthreats.


Learn More About AI Cybersecurity

Accelerated Optimization

cuOpt’s world-record-holding accelerated solver optimizes routes for last-mile delivery, technician dispatch, or intra-factory logistics.


Learn More About Route Optimization

Trillion Edge Graph

RAPIDS cuGraph makes it possible for enterprises to train trillion edge graph neural networks.


Learn More About Trillion Edge Graphs

RAPIDS excels at accelerating business-critical applications, reducing years of planning and development across industries.

Retail

Accelerated data science drives improved retail forecasting, data analytics, and more for retail.


Learn More

Finance

Real-time data enhances fraud detection and forecasting in an industry where time is of the essence.


Learn More

Thriving Ecosystem

With more than 100 open-source and commercial software integrations, RAPIDS provides a foundation for a collaborative high-performance data science ecosystem.



We're committed to simplifying, unifying, and accelerating data science for the open-source community.


RAPIDS partners with the most popular data science and machine learning platforms to democratize access to accelerated data science.

Enterprise Data Science

Accelerated data science with NVIDIA AI Enterprise, an end-to-end, secure, cloud-native AI software platform optimized to take enterprises to the leading edge of AI. NVIDIA AI Enterprise delivers validation and integration for NVIDIA AI open-source software, including RAPIDS, access to AI solution workflows to speed time to production, certifications to deploy AI everywhere, and enterprise-grade support, security, manageability, and API stability to mitigate the potential risks of open-source software.


Learn More About AI Enterprise

Enterprise Adoption

Our customers use RAPIDS’ fully functional stack to scale their enterprise use cases.

RAPIDS Customer - PayPal

PayPal reduced cloud costs by up to 70% with the RAPIDS accelerator for Apache Spark.

Watch On-Demand Session
RAPIDS Customer - Taboola

Taboola, an advertising platform, processes terabytes of hourly data with the RAPIDS accelerator for Apache Spark.

Watch On-Demand Session
RAPIDS Customer - CapitalOne

CapitalOne accelerated their financial and credit analysis pipelines, improving model training by 100X.

Watch On-Demand Session
RAPIDS Customer - Uber

Uber developed Horovod with support for Spark 3.x with GPU scheduling.

Watch On-Demand Session
RAPIDS Customer - Walmart

Walmart solved scalability issues with their product-substitution algorithm.

Watch On-Demand Session
RAPIDS Customer - LinkedIn

LinkedIn developed DARWIN to enable faster data analysis on RAPIDS cuDF.

Watch On-Demand Session
RAPIDS Customer -  AT&T

AT&T applied the RAPIDS Accelerator for Apache Spark on GPU clusters in their data-to-AI pipeline.

Read Blog
RAPIDS Customer - NASA

NASA used RAPIDS to detect and quantify air pollution anomalies and build a bias-correction model.



Read Blog: Part 1 Read Blog: Part 2
RAPIDS Customer - TCS

TCS Optumera accelerated their demand forecasting pipeline with the RAPIDS Accelerator for Apache Spark.

Watch On-Demand Session
RAPIDS Customer - Cloudera

The IRS team uncovered fraud with the RAPIDS Accelerator for Apache Spark on the Cloudera Data Platform.

Read Blog

Check out more RAPIDS resources, including developer kits, NVIDIA LaunchPad labs, and guidance on deployment options.


Get Started