Nvidia announced a lower-cost 4GB version of its Linux-driven Jetson TX2 module with half the RAM and eMMC and has begun shipping its next-gen Jetson AGX Xavier.

Nvidia will soon have three variants of its hexa-core Arm Jetson TX2 module: the original Jetson TX2, the more embedded, industrial temperature Jetson TX2i , and now a new Jetson TX2 4GB model. The chip designer also announced availability of its next-gen, robotics focused Jetson AGX Xavier module (see farther below).

Due to ship in June 2019, Nvidia’s Jetson TX2 4GB module has half the RAM and flash storage as the TX2 or TX2i, with 4GB 128-bit LPDDR4 and 16GB eMMC 5.1 flash. The bare-bones spec sheet on the product page makes no mention of the 802.11ac and Bluetooth radios of the TX2 and TX2i.

Nvidia Jetson TX2 4GB
(click image to enlarge)

Like the earlier TX2 modules, the 4GB version features 2x high-end “Denver 2” cores and 4x Cortex-A57 cores. You also get the 256-core Pascal GPU with CUDA libraries for running AI and machine learning algorithms.

Otherwise, the Jetson TX2 4GB appears to be identical to the original, presumably with the same 87 x 55mm footprint. It works with Nvidia’s existing, Linux-driven SDK and should also work with third-party carrier boards. In fact, if you have the Jetson TX2 Developer Kit, you can download a “Jetson TX2 as Jetson TX2 4GB Configuration Package” to make your existing TX2 perform like the 4GB model.

 
Jetson AGX Xavier ships

Since the initial Xavier announcements earlier this year, Nvidia has added AGX to the Jetson Xavier name. The Jetson AGX Xavier has greater than 10x the energy efficiency and more than 20x the performance of the Jetson TX2, claims Nvidia. The module is available with a BSP with Nvidia’s Linux4Tegra stack. Nvidia also offers an AI-focused Isaac SDK.

Jetson AGX Xavier (left) and development kit
(click images to enlarge)

The 105 x 87 x 16mm Xavier module features 8x ARMv8.2 cores and a high-end, 512-core Nvidia Volta GPU with 64 tensor cores with 2x Nvidia Deep Learning Accelerator (DLA) — also called NVDLA — engines. The module is also equipped with a 7-way VLIW vision chip, as well as 16GB 256-bit LPDDR4 RAM and 32GB eMMC 5.1.

There’s also an automotive version of the Xavier, which was originally called the Drive PX Pegasus when it was announced in Nov. 2017. There’s a Linux-driven Nvidia Drive AGX Xavier Developer Kit that supports a single Xavier module or else a Drive AGX Pegasus version with dual Xaviers and dual GPUs.

Nvidia Drive AGX Xavier Developer Kit
(click image to enlarge)

In addition to the Nvidia developer kit, ConnectTech Inc. (CTI) announced a compact Rogue carrier for the Jetson AGX Xavier back in October. CTI also announced a Mimic Adapter that plugs the Xavier into any CTI Jetson TX2/TX2i/TX1 carrier.


CTI’s Rogue carrier with Xavier module (equipped with fan)
(click images to enlarge)

 
Further information

Nvidia’s Jetson TX2 4GB will be available in June 2019. Pricing was undisclosed, but a New Electronics story reported that “Jetson TX1-based products can migrate to the more powerful Jetson TX2 4GB at the same price.”

The Jetson AGX Xavier is available now starting at $1,099 per module for 1,000-plus purchases. More information may be found in Nvidia’s announcement and its Jetson TX2 4GB product page.