Converging Data Science and Data Engineering with Our Open Source Integration for RAPIDS

Recently, NVIDIA announced RAPIDS, an open source data science library that enables data scientists to accelerate model training and development by harnessing the power of the GPU. Today, we’re pleased to share our open source integration with RAPIDS to enable data scientists and data engineers to build applications with artificial intelligence, all while leveraging the immense computing power of the GPU. RAPIDS is an exciting step forward for our industry and will help to drive the adoption of AI at scale.

RAPIDS is key for accelerating data science, reducing model training time from days to hours and minutes. However, businesses still need a platform to deploy and operationalize these models. That’s where Kinetica comes in, offering an analytics engine that allows data engineers to bring trained algorithms and models to the entire dataset. With our new integration, enterprises now have an end-to-end solution for model development and training and production that’s optimized for the GPU. This GPU-accelerated data pipeline will help businesses realize the value of AI more quickly and gain a competitive edge. 

Our open source integration with RAPIDS is based on the Apache Arrow project. This allows the Kinetica engine and RAPIDS to run seamlessly on the GPU and communicate without copying data to the CPU, accelerating the speed at which models can be trained and run. The Apache Arrow integration code is available on GitHub and you can download RAPIDS and Kinetica from the NVIDIA GPU Cloud. The latest version of Kinetica natively supports RAPIDS. We’re excited to see the innovation our customers are already driving using this new functionality to power amazing solutions around data science and machine learning at scale.

Read the technical blog to learn how to get started with Kinetica and RAPIDS.

Our new integration enhances the scale, performance, and ease with which data scientists and data engineers can leverage RAPIDS to bring AI to the enterprise. We’re looking forward to seeing what you build!

Resources

• Download the RAPIDS integration code from GitHub.
• Visit the NVIDIA GPU Cloud to download the Kinetica engine and RAPIDS.
• Read the technical blog on the integration.

Ken Wattana is Sr. Partner Marketing Manager at Kinetica. You can follow him on Twitter @KenWattana.