Nvidia on Wednesday announced the availability of the Jetson AGX Xavier module, a module for the production of autonomous devices like drones, robots and video analytics systems. It effectively delivers the performance of a workstation server in a power-efficient, palm-sized module, Nvidia says.
It's built around the Xavier SOC, which according to Nvidia's Jesse Clayton, is the "first architecture built from the ground up for autonomous machines." The chip consists of six processing units, including a 512-core Nvidia Volta Tensor Core GPU, an eight-core Carmel Arm64 CPU, a dual NVDLA deep-learning accelerator, and image, vision, and video processors.
The Jetson AGX Xavier joins the Jetson family, which includes Jetson TX1 and TX2. The Jetson TX2, "solves a lot of problem for customers," Clayton said, but "only scratches the surface" when it comes to developing autonomous machines. The AGX Xavier module delivers up to 32 TOPS of accelerated computing capability while consuming under 30 Watts. That's more than 20X the performance and 10X the energy efficiency of the TX2.
In September, Nvidia announced the availability of the AGX developer kit, and since then a number of companies have adopted it, Clayton said, including Yamaha, JD.com and Komatsu.
The market potential for this kind of solution is significant, given how autonomous machines will "impact nearly every industry," Clayton said. In manufacturing, for instance, only about 10 percent of tasks are automated. The rest are too hard for today's fixed-function robots. Now, "we're starting to take on that remaining 90 percent," Clayton said.
- Nvidia Shield (2017) review: Even more of a good streaming (CNET)
- Top 5: Things to know about AI (TechRepublic)
Nvidia expects to see similarly transformative use cases in a range of areas like agriculture, autonomous delivery, health care and retail. "We're at the cusp of one of the biggest changes in the history of technology," Clayton said.
The Jetson AGX Xavier module is available now starting at $1,099.
Previous and related coverage:
IBM Spectrum AI with Nvidia DGX is designed for AI and machine learning workloads.
The partners aim to develop AI and deep learning best practices, tools and infrastructure to accelerate AI applications using genomic and digital health sensor data.