Advanced Computing in the Age of AI|Wednesday, January 20, 2021
  • Subscribe to EnterpriseAI Weekly Updates:  Subscribe by email

Nvidia Moves Its Software Closer to Cloud Customers Through an AWS Storefront 

Nvidia is making it easier for AWS cloud customers to find and integrate Nvidia software applications into their AI and deep learning projects through an all-new, all-in-one “storefront” in the AWS Marketplace.

The storefront, which initially contains 21 applications from the Nvidia GPU Cloud (NGC) software application catalog, is designed to bring Nvidia applications right to AWS so users can locate them quickly as needed, Adel El Hallak, the director of the NGC catalog, told EnterpriseAI.

Under the deal, AWS Marketplace is the first cloud provider to offer the NGC catalog of applications directly in its own integrated storefront, he said. “Rather than publishing an application here or container there, we're lifting and shifting, bringing our entire store presence to the AWS Marketplace.”

Adel El Hallak of Nvidia NGC

The AWS marketplace is a destination where cloud customers can find, buy and immediately start using software and services that run on AWS. The Nvidia NGC catalog of software, which was established in 2017, is optimized to run on Nvidia GPU cloud instances, such as the Amazon EC2 P4d instances which use Nvidia A100 Tensor Core GPUs. AWS customers will be able to deploy Nvidia’s software for free to accelerate their AI deployments. The NGC catalog includes AI containers, pre-trained models, application frameworks, Helm charts and other machine learning resources.

The creation of the storefront will help Nvidia simplify the deployment of the applications for its customers, said El Hallak.

“For example, if you wanted to use any of our software with Amazon SageMaker, one of the precursors was you had to go copy the container into the AWS Elastic Container Registry (ECR),” he said. “But by bringing [NGC] into the AWS Marketplace, those entities [already] live in ECR, so it simplifies deployment with SageMaker, in addition to simplifying deployment with other AWS services such as Elastic Compute Cloud (EC2), Elastic Kubernetes Service and Elastic Container Service.”

What the NGC Storefront Will Offer

The new storefront includes 21 of Nvidia’s most popular apps for accelerating work in AI, conversational AI, HPC, healthcare, data science, robotics and more. El Hallak would not comment about future additions to the storefront, but it is likely that more Nvidia applications will be added over time.

“This is the first milestone in a partnership,” said El Hallak. “It was part of our interest to bring our portfolio over to the cloud service providers [like AWS], but this was also a reaction to the cloud service providers, who are always asking us for our software portfolio. They're seeing the users [need such applications].”

To fulfill those customer needs, Nvidia is releasing new domain-specific AI frameworks on a monthly basis and regular cadence, said El Hallak.

“We're seeing … tremendous growth for NGC,” he said. “We're above one million downloads, with 250,000 [developer and data scientist] users in the past 14 to 15 months. So, this [storefront] makes sure that as soon as we publish software that AWS users will have access to it immediately because it gets federated into their marketplace … and integrates closer with their users. It's really a reaction to listening to AWS users who are asking for our software to be brought over.”

By having the Nvidia NGC storefront inside the AWS Marketplace, it’s like having a Starbucks shop inside a Target store, said El Hallak. “You know how you have a store within a store there? It's a similar concept here. We're bringing an entire store presence natively to live in the AWS Marketplace.”

A Smart Move, Analysts Say

Patrick Moorhead of Moor Insights & Strategy

So what does this all mean for Nvidia and AWS cloud customers?

“This is a convenience play for joint AWS and Nvidia customers,” Patrick Moorhead, the president and principal analyst at Moor Insights & Strategy, told EnterpriseAI. “AWS customers could always install the software, but now it’s point-and-click. I think AWS is doing this because it wants to show that even though it has alternatives – such as Inferentia, Trainium, AMD Radeon Pro and Intel Habana – to Nvidia silicon, it still wants to be the closest partner to Nvidia.”

Another analyst, Addison Snell, the CEO and principal analyst of Intersect360 Research, agreed.

“This announcement is ultimately good for Nvidia, good for AWS, and good for HPC users,” he said. “By making it easy to start development of HPC and AI applications in the public cloud, Nvidia and AWS are spurring innovation and growing the pool of research. For AWS it increases the portfolio of offerings on their leading public cloud portal. And for Nvidia it helps to solidify a leadership position in accelerated computing with its GPU solutions.”

Addison Snell of Intersect360 Research

The first Nvidia software applications in the new NGC storefront in the AWS Marketplace include:

Nvidia AI: A suite of frameworks and tools, including MXNet, TensorFlow, Nvidia Triton Inference Server and PyTorch.

Nvidia Clara Imaging: Nvidia’s domain-optimized application framework that accelerates deep learning training and inference for medical imaging use cases.

Nvidia DeepStream SDK: A multiplatform scalable video analytics framework to deploy on the edge and connect to any cloud.

Nvidia HPC SDK: A suite of compilers, libraries and software tools for high performance computing.

Nvidia Isaac Sim ML Training: A toolkit to help robotics machine learning engineers use Isaac Sim to generate synthetic images to train an object detection deep neural network.

Nvidia Merlin: An open beta framework for building large-scale deep learning recommender systems.

Nvidia NeMo: An open-source Python toolkit for developing state-of-the-art conversation AI models.

RAPIDS: A suite of open-source data science software libraries.

Add a Comment

Do NOT follow this link or you will be banned from the site!
Share This