Cloudian Enabling S3-Compatible Storage Access for Nvidia Deep Learning GPU System

Cloudian, Inc., in enterprise object storage systems, has announced the integration of a plug-in within NVIDIA’s Deep Learning GPU Training System (DIGITS) version 6.1 that allows the DIGITS platform to access an S3-compatible storage endpoint.

DIGITS users can integrate Cloudian’s S3-compatible HyperStore platform for use with deep learning training libraries.

With this integration, data scientists and researchers using DIGITS can access Cloudian’s object storage platform as a repository for the vast amounts of data needed to give their deep learning applications the information needed to identify patterns and ‘learn’ the best responses to any given situation.

“This integration will enable AI developers more flexibility in how their applications access and store data, and will lead to faster and more accurate decisions and suggestions,” said Sanjay Jagad, senior director of products and solutions, Cloudian. “The HyperStore platform is the ideal repository for this data, being limitlessly scalable and easy to grow. It’s an ideal technology to help the performance of AI/ML applications, which become more powerful as the amount of data available to them increases.“

NVIDIA DIGITS puts the power of deep learning – the use of multiple deep neural networks (DNNs) to help machines make sense of complex data – into the hands of engineers and data scientists. It can be used to train DNNs for image classification, segmentation and object detection tasks.

NVIDIA DIGITS 6.1 with Cloudian plug-in is available on GitHub