Nvidia Updates Its Jetson Platform

Jetson TX2 Doubles the Performance of Its Predecessor

This week, Nvidia announced the Jetson TX2, an update to the embedded system Jetson TX1. The new iteration of the chipset is designed to put advanced computing power in a wide host of devices. Nvidia claims that the Jetson TX2 doubles the performance of its predecessor while drawing less than 7.5 watts of power.

The Jetson TX2 joins Nvidia’s family of embedded computing solutions including the TX1 and TK1. Jetson is designed as an open platform and it’s accessible to companies and developers looking to implement advanced artificial intelligence to work at the edge. Nvidia has found success with its artificial intelligence-enabled systems for advanced automotive uses, which include autonomous driving, and in servers, establishing its expertise in these areas. Now the company is eager to address the needs of makers of other moving devices such as robots and drones.

In general, suppliers of silicon, looking for growth beyond traditional computing markets, are eyeing opportunities in emerging areas.

These companies have been making significant investments in computer vision and other forms of machine learning within devices. Deep learning applied in graphics processing units has given computers the ability to understand and react to data that streams in from devices. This is achieved through training, which creates smart systems, and through inference, which creates systems that are able to react intelligently to their environments in real time.

Nvidia is also expected to demonstrate its TX2 solution at Embedded World 2017 in Nuremberg, Germany, alongside partners like Cisco, which will showcase a smart whiteboard design that uses the TX2. Cisco hopes to use the platform to add local features powered by artificial intelligence, such as facial and voice recognition, to its Spark enterprise video conferencing system, providing advantages in terms of security and authentication.

We believe this is another example of a new enabler of hardware, which will boost the development of a new generation of intelligent products and features including smart cameras, improved functionality in autonomous vehicles and an advanced level of robotics for consumer and business uses. Nvidia rivals, such as Qualcomm with its Snapdragon system-on-chip solutions, Movidius with its Myriad 2 and MediaTek with its deep-learning platform, are also providing new generations of components to enable innovation. Intel is also seeking to establish a complete artificial intelligence solution to compete with Nvidia using its Xeon and Xeon Phi products, combined with assets from its Nervana acquisition. The move will support Intel’s embedded Atom and Curie systems-on-chips for embedded applications at the edge.

The pieces are now available for a world filled with impressive smart “things” that know their environments and can optimize their performances accordingly, learning along the way.