Nvidia Announces the Jetson TX2, an Embedded AI Chip
Nvidia announced today that it will soon start shipping the Jetson TX2, an embedded AI chip for use in IoT devices. The new chip is aimed at manufacturers of small devices that don’t need cloud-based AI or would prefer a lower power draw while keeping some basic artificial intelligence. The Jetson TX2 can be pre-ordered today, will ship from March 14 onward and will cost $600 for a kit or $400 for just the chip, though the chip-only orders will not ship until second quarter this year.
The Jetson TX2 is just one chip that’s coming to the market that facilitates deep learning in machines and will enable all manner of smart devices to recognize voices, navigate, or classify images, among other things. It packs 8GB of RAM and can support up to six cameras with a bandwidth of up to 2.5Gbps per lane. The GPU architecture is based on Nvidia’s own Pascal and it comes with two Denver CPUs. Storage is a respectable 32GB.
Besides embedded AI functionality, the Jetson TX2 can run in two power modes: it can either max out performance, which makes it draw up to 15W, or run at max power efficiency, which reduces this to 7.5W but also halves performance. This should give the new Nvidia chip a lot of flexibility on the market, making it very interesting for manufacturers that require a pretty smart chip that doesn’t need too much power as well as those that are building devices that can adjust on the fly.
As there is more to AI than hardware, Nvidia has included Jetpack 3.0, the newest version of this deep-learning SDK, which should have developers off to a running start as soon as they have hooked the Jetson TX2 up. Jetpack 3.0 supports most common — and some uncommon — libraries, so there’s nothing standing in the way of some serious AI development.
Fergus has been tinkering with computers since he was a kid and likes to put a stop to parties by listing the specs of all the digital devices in the room. It's best not to let him near your computer since he'll take it apart and may not put it back together again before he leaves.