Communications of the ACM
Open Source Platform Launched to Help Bring GP Acceleration to Data Analytics
A coalition of companies has launched an open source platform for data science and machine learning.
Credit: developer.nvidia.com
Nvidia, IBM, HPE, Oracle, Databricks, and other organizations have launched a new open source platform for data science and machine learning.
With the RAPIDS platform, the companies aim to make it easier for large organizations to use graphics processing units (GPUs) to quickly analyze massive data volumes from which to build machine learning models.
RAPIDS users will work with existing popular open-source libraries and platforms, and accelerate them using GPUs, integrating with these libraries to provide accelerated analytics, machine learning, and, eventually, visualization.
The new platform is based on Python, and has interfaces that are similar to the popular machine learning and data analysis libraries Pandas and Scikit.
RAPIDS also is based on Apache Arrow for in-memory database processing, and it can scale from a single GPU to multiple nodes, achieving improvements of up to 50 times for some specific use cases compared to running the same algorithms on central processing units (CPUs).
From TechCrunch
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found