NVIDIA continues to be the driving force behind the artificial intelligence (AI) market, announcing the NVIDIA GPU Cloud (NGC), a new line-up of NVIDIA DGX AI supercomputers, the NVIDIA Isaac robot simulator and the NVIDIA Jetson platform at the eighth annual GPU Technology Conference (GTC).

The NVIDIA GPU Cloud (NGC) is a cloud-based platform that will give developers convenient access via their PC, NVIDIA DGX system or the cloud, to a comprehensive software suite for harnessing the transformative powers of AI. This will make it easier for developers to access the latest, optimised deep learning frameworks and the newest GPU computing resources.

“We’re designing a cloud platform that will unleash AI developers, so they can build a smarter world. You can do your best work no matter where you are, using our latest technology in the cloud. It’s accelerated computing when and where you need it,” said NVIDIA vice president and general manager Jim McHugh.

Harnessing deep learning presents two challenges for developers and data scientists. The first is getting access to the latest GPU computing resources to train a neural network and the second is the need to gather into a single stack the requisite software components, including deep learning frameworks, libraries, operating system and drivers.

To address the first challenge, NGC will give developers the flexibility to run a new containerised package, NGC Software Stack, on a PC equipped with a TITAN X or GeForce GTX 1080 Ti, on a DGX system or from the cloud.

This NGC Software Stack is how NVIDIA solved the second challenge earlier this year, combining the key software elements within the NVIDIA DGX-1 AI supercomputer in the NGC Software Stack.

This NVIDIA DGX-1 AI supercomputer has also been updated, with a new line-up announced to deliver ground-breaking AI computer power three times faster than the prior DGX generation.

Featuring NVIDIA Tesla V100 data centre GPUs based on the NVIDIA Volta architecture and a fully optimised AI software package, the systems provide the performance of up to 800 CPUs in a single system.

The NVIDIA Volta architecture-based DGX portfolio includes the NVIDIA DGX-1 AI supercomputer for data centre deployments and a new personal supercomputing workstation, the NVIDIA DGX Station. Both systems benefit from the integrated NVIDIA GPU Cloud Deep Learning Stack delivered over the NVIDIA GPU Cloud.

The NVIDIA GPU Cloud Deep Learning Stack integrates the latest deep learning frameworks and the NVIDIA software development kit into an always up-to-date container. Using one consistent software stack across the portfolio, data scientists can easily experiment desk-side on their personal DGX Station and then seamlessly scale their work to a DGX-1 server cluster.

“NVIDIA’s DGX portfolio and its software are the essential instruments for advancing the work of serious AI research and realising the promise of this new era of computing. No other computing system comes close to providing the same level of performance for AI and advanced analytics,” said Jim.

Introduced last year, DGX-1 systems now power a wide range of AI deployments at leading enterprises, cloud service providers and research organisations worldwide. NVIDIA designed its portfolio of DGX systems to deliver extreme versatility with architecture capable of powering both deep learning training and inference.

At GTC NVIDIA also announced it has simplified the work of building and training intelligent machines thanks to the NVIDIA Isaac robot simulator and NVIDIA Jetson platform.

The NVIDIA Isaac robot simulator uses sophisticated videogame and graphics technologies to train intelligent machines in simulated real-world conditions before they get deployed. It is built on an enhanced version of Epic Games’ Unreal Engine 4 and uses NVIDIA’s advanced simulation, rendering and deep learning technologies

The simulator provides an AI-based software platform that allows teams to train robots in highly realistic virtual environments and then transfer that knowledge to real-world units. Working within this virtual environment, developers can set up extensive test scenarios using deep learning training, simulate them in minutes and then iterate and tweak the robot testing methodology, trading intelligence between the two environments.

The NVIDIA Jetson platform is a set of robot reference-design platforms that make it faster to build these intelligent machines.

To accelerate the development of advanced robotics on Jetson, NVIDIA partners are releasing open- source reference platforms for drones, submersibles, robots on wheels and other devices. These platforms provide building blocks for developers to easily create prototypes, helping reduce the time and money it takes to build robots from the ground up.

Some of the companies releasing Jetson reference platforms include:

• Toyota, for human support robots.
• Teal, for consumer drones.
• Enroute Lab, for industrial drones and unmanned surface and ground vehicles
• Various universities and academia, for scale model autonomous cars.

At GTC, more than 50 companies from around the world are showing robots capable of a huge range of capabilities, all of which incorporate NVIDIA’s Jetson platform. They are capable of handling tasks including, search and rescue, elder support and industrial automation of tedious or potentially dangerous tasks.

“Robots based on artificial intelligence hold enormous promise for improving our lives, but building and training them has posed significant challenges. NVIDIA is now revolutionising the robotics industry by applying our deep expertise in simulating the real world so that robots can be trained more precisely, more safely and more rapidly,” said NVIDIA chief executive and founder Jensen Huang.