Nvidia rapids, new set of open source rapids libraries for accelerated gpu analysis and machine learning

At the GPU Technology Conference that was held in the German city of Munich, Nvidia, a market leader in high-performance GPUs and Artificial Intelligence, has taken a further step forward with the announcement of a new set of RAPIDS libraries. open source for accelerated GPU analysis and machine learning.

Nvidia RAPIDS, open source libraries for AI

This time, Nvidia is not announcing a new GPU platform, or a new proprietary SDK for deep learning, but rather a new set of open source libraries for accelerated GPU scanning and machine learning. Dubbed RAPIDS, the new set of libraries will offer Python interfaces similar to those provided by Scikit Learn and Pandas, but which will take advantage of the company's CUDA platform for acceleration on one or more GPUs.

We recommend reading our post about Nvidia RTX 2080 Ti Review in Spanish (full analysis)

According to Nvidia CEO Jensen Huang, who briefed several tech journalists on the phone Tuesday, Nvidia has seen 50x faster training time when using RAPIDS instead of a CPU-only implementation. This speed was measured in scenarios involving the XGBoost ML algorithm on an Nvidia DGX-2 system, although the configuration of the CPU hardware was not explicitly discussed.

RAPIDS apparently incorporates Apache Arrow memory column data technology, and is designed to run on Apache Spark. With the latter in mind, the company has obtained the Databricks software, which will integrate RAPIDS into its own analytics and AI platform.

However, Databricks is not the only big name supporting the RAPIDS platform. Tech giants like IBM, Hewlett Packard Enterprise and Oracle are also in action.

Techpowerup font