GPU Accelerating Node.js with the Node-RAPIDS Data Science Framework

A few years ago NVIDIA started the open-source RAPIDS project to create GPU accelerated data science C++/Python libraries such as cuDF, cuML, cuGraph, and cuSpatial. We are now introducing Node.js bindings for these RAPIDS libraries with the new open-source Node-RAPIDS project (currently in technical preview). When combined with complementary methods for high-performance, browser-based visualizations and computations, the Node-RAPIDS project has numerous compelling server side and local app use cases.

Node.js is the connector that allows JavaScript direct access to GPU hardware, affording a streamlined API and the ability to utilize CUDA. By creating node-RAPIDS bindings, we aim to give this developer community GPU acceleration without the need to learn a new language or work in a new environment, and provide access to a high performance data science platform: RAPIDS.

We hope that this unified space will allow data engineers, data scientists, visualization specialists, and front-end developers to become less siloed in their prospective toolsets and work more closely together.

For our talk, we will present a brief overview of RAPIDS, explain why it is beneficial to bring to Node.js, describe the modular architecture behind node-RAPIDS, and finally, demonstrate code examples with walkthroughs and interactive visualizations.