NVIDIA Jetson TX2 IoT Platform Delivers Pascal-Powered AI Computing For Smart Robots And Commercial Drones
NVIDIA says that Jetson TX2 can be used for “intelligent” factory robots, commercial drones and smart cameras. However, we’re sure that developers will be able to think of dozens of uses for the versatile platform to enable new classes of artificial intelligence (AI) driven products.
Jetson TX2 features next-generation 64-bit Denver and Cortex-A57 processors, and includes 8GB of LPDDR4 memory with 58.4 GB/s of bandwidth (twice the RAM and more than twice the bandwidth of its predecessor). Onboard eMMC storage also doubles to 32GB while video encoding/decoding is matched at 2160p @ 60fps (the Jetson TX1 encoded 2160p @ 30 fps). Perhaps the biggest change, however, is that the onboard graphics core is now based on NVIDIA’s current generation Pascal architecture instead of Maxwell.
The Jetson TX2 is capable of running in two operating modes: MAX-Q (for maximum efficiency) and Max-P (for maximum performance). In Max-Q mode, the Jetson TX2 is up to twice as energy efficient as its predecessor while operating within a 7.5W power envelope. In Max-P mode, the Jetson TX2 power envelope opens to 15W which allows for a 2x performance increase over Jetson TX1. This newfound power will allow Jetson TX2 to process speech recognition algorithms at a quicker pace and run deeper neural networks.
“Jetson TX2 brings powerful AI capabilities at the edge, making possible a new class of intelligent machines,” said Deepu Talla, vice president and general manager for NVIDIA’s Tegra business. “These devices will enable intelligent video analytics that keep our cities smarter and safer, new kinds of robots that optimize manufacturing, and new collaboration that makes long- distance work more efficient.”
NVIDIA is providing a Jetson TX2 Developer Kit, which is available for pre-order now for $599, although education customers will be able to get the same hardware for $299 (kits will ship on March 14th). Production-quality modules for commercial applications will be available in quantities of 1,000 at a cost of $399 each (Q2 2017 availability). NVIDIA will also make available the JetPack 3.0 SDK for AI computing, which adds Linux Kernel 4.4, Multimedia API 27.1 and offers 2x the system performance compared to JetPack 2.0.