Thursday, November 7th 2019

NVIDIA Announces Jetson Xavier NX, Smallest Supercomputer for AI at the Edge

NVIDIA today introduced Jetson Xavier NX, the world's smallest, most powerful AI supercomputer for robotic and embedded computing devices at the edge. With a compact form factor smaller than the size of a credit card, the energy-efficient Jetson Xavier NX module delivers server-class performance up to 21 TOPS for running modern AI workloads, and consumes as little as 10 watts of power.

Jetson Xavier NX opens the door for embedded edge computing devices that demand increased performance but are constrained by size, weight, power budgets or cost. These include small commercial robots, drones, intelligent high-resolution sensors for factory logistics and production lines, optical inspection, network video recorders, portable medical devices and other industrial IoT systems.
"AI has become the enabling technology for modern robotics and embedded devices that will transform industries," said Deepu Talla, vice president and general manager of Edge Computing at NVIDIA. "Many of these devices, based on small form factors and lower power, were constrained from adding more AI features. Jetson Xavier NX lets our customers and partners dramatically increase AI capabilities without increasing the size or power consumption of the device."

Jetson Xavier NX delivers up to 14 TOPS (at 10 W) or 21 TOPS (at 15 W), running multiple neural networks in parallel and processing data from multiple high-resolution sensors simultaneously in a Nano form factor (70 mm x 45 mm). For companies already building embedded machines, Jetson Xavier NX runs on the same CUDA-X AI software architecture as all Jetson offerings, ensuring rapid time to market and low development costs.

As part of NVIDIA's one software architecture approach, Jetson Xavier NX is supported by NVIDIA JetPack software development kit, which is a complete AI software stack that can run modern and complex AI networks, accelerated libraries for deep learning as well as computer vision, computer graphics, multimedia and more.

Ecosystem Support
Jetson Xavier NX is receiving strong support from the robotics and embedded devices ecosystem.

"NVIDIA's embedded Jetson products have been accelerating the research, development and deployment of embedded AI solutions on Lockheed Martin's platforms," said Lee Ritholtz, director and chief architect of Applied Artificial Intelligence at Lockheed Martin. "With Jetson Xavier NX's exceptional performance, small form factor and low power, we will be able to do more processing in real time at the edge than ever before."

"Our goal is to dramatically increase the quality and accuracy of our optical inspection system and accelerate our move towards industry 4.0," said Otsuka Hiroshi, CEO of Musashi Seimitsu. "NVIDIA Jetson Xavier NX gives us the compute capabilities to improve our visual inspection capabilities without increasing the size and power of our optical inspection system."

NVIDIA also announced today that it topped all five benchmarks measuring the performance of AI inference workloads in data centers and at the edge — building on the company's equally strong position in recent benchmarks measuring AI training. The results of MLPerf Inference 0.5, the industry's first independent AI benchmark for inference, demonstrate the inference capabilities of NVIDIA Turing GPUs for data centers and the NVIDIA Xavier system-on-a-chip for edge. The Jetson Xavier NX module is built around a new low-power version of the Xavier SoC used in these benchmarks.

"In a world where AI chips are announced on what seems like a daily basis, I believe NVIDIA raised the bar with its Jetson Xavier NX — showing that exceptional performance at small size and low power, together with a consistent and powerful software architecture, is what matters in embedded edge computing," said Patrick Moorhead, president and principal analyst of Moor Insights & Strategy.

Jetson Xavier NX module specifications:
  • GPU: NVIDIA Volta with 384 NVIDIA CUDA cores and 48 Tensor Cores, plus 2x NVDLA
  • CPU: 6-core Carmel ARM 64-bit CPU, 6 MB L2 + 4 MB L3
  • Video: 2x 4K30 Encode and 2x 4K60 Decode
  • Camera: Up to six CSI cameras (36 via virtual channels); 12 lanes (3x4 or 6x2) MIPI CSI-2
  • Memory: 8 GB 128-bit LPDDR4X, 51.2 GB/s
  • Connectivity: Gigabit Ethernet
  • OS Support: Ubuntu-based Linux
  • Module Size: 70 mm x 45 mm
Jetson Xavier NX is the latest addition to the Jetson family, which includes Jetson Nano, the Jetson AGX Xavier series and the Jetson TX2 series. Jetson Xavier NX offers a rich set of IOs, from high-speed CSI and PCIe to low-speed I2Cs and GPIOs. Compatibility with many peripherals and sensors, together with its small form factor and big performance, will bring new capabilities to embedded AI and industrial IoT systems.

Jetson Xavier NX is also pin-compatible with Jetson Nano, allowing shared hardware designs and those with Jetson Nano carrier boards and systems to upgrade to Jetson Xavier NX. It also supports all major AI frameworks, including TensorFlow, PyTorch, MxNet, Caffe and others.

Priced at $399, the Jetson Xavier NX module will be available in March from NVIDIA's distribution channels for companies looking to create high-volume production edge systems. Developers can begin application development today using the Jetson AGX Xavier Developer Kit with a software patch to emulate Jetson Xavier NX.
Add your own comment

11 Comments on NVIDIA Announces Jetson Xavier NX, Smallest Supercomputer for AI at the Edge

#1
TheLostSwede
Can't see a lot of high volume products being based on these modules at that price point/spec. Seriously, why does this have half the storage compared to the $129 TX2 module, for starters?

The spec in the press release are also misleading, as the modules have 2, 4 & 6 cores, depending on TDP.
Posted on Reply
#2
Ferrum Master
TheLostSwede
Can't see a lot of high volume products being based on these modules at that price point/spec.
Cars?
Posted on Reply
#3
TheLostSwede
Ferrum Master
Cars?
Yeah, no. They wouldn't use modules, they'd make custom boards, unless of course Nvidia tried to force them to use the modules.
For development, sure, but not really for final implementation. The connector is too much of a weak link that could go wrong.
This is what one of the Tesla boards look like.
Posted on Reply
#4
Basard
Looks like an old Athlon!
Posted on Reply
#5
jabbadap
"Jetson Xavier NX is also pin-compatible with Jetson Nano, allowing shared hardware designs and those with Jetson Nano carrier boards and systems to upgrade to Jetson Xavier NX. It also supports all major AI frameworks, including TensorFlow, PyTorch, MxNet, Caffe and others."

So correct me if I'm wrong, but does this mean that you can just slap this on Jetson Nano devkit and it will work right a way? Sounds actually awesome if true.
Posted on Reply
#6
Flanker
TheLostSwede
Yeah, no. They wouldn't use modules, they'd make custom boards, unless of course Nvidia tried to force them to use the modules.
For development, sure, but not really for final implementation. The connector is too much of a weak link that could go wrong.
Is this also true for drones?
Posted on Reply
#7
TheLostSwede
Flanker
Is this also true for drones?
Maybe not, but these are too power hungry for most drones.
Robotics maybe, but that's not really large scale imho, as that would normally be a few hundred units based on something like this.
Maybe that'll change in the future though.
Posted on Reply
#8
notb
TheLostSwede
Can't see a lot of high volume products being based on these modules at that price point/spec.
Oh, really? :D

Cars, planes, ships, agricultural/industrial vehicles.
Smart home systems.
Manufacturing plants.
Security and surveillance.

In long term market for this kind of ML/AI modules is much larger than for anonymous driving systems.
Of course not every "smart" device can use a $399 module, but a lot can.
Cheaper alternatives exist as well (at Nvidia it starts at $149 with Jetson Nano
Seriously, why does this have half the storage compared to the $129 TX2 module, for starters?
TX2 costs $479 and will become obsolete when NX arrives.
The spec in the press release are also misleading, as the modules have 2, 4 & 6 cores, depending on TDP.
No. Always 6 cores, but with different clocks.
Posted on Reply
#9
Vya Domus
jabbadap
"Jetson Xavier NX is also pin-compatible with Jetson Nano, allowing shared hardware designs and those with Jetson Nano carrier boards and systems to upgrade to Jetson Xavier NX. It also supports all major AI frameworks, including TensorFlow, PyTorch, MxNet, Caffe and others."

So correct me if I'm wrong, but does this mean that you can just slap this on Jetson Nano devkit and it will work right a way? Sounds actually awesome if true.
Probably they mean this : https://auvidea.eu/images/auvidea/products/xavier/Panel_GTC_Munich_2018_1.2.pdf
Posted on Reply
#10
softreaper
After contacting nvidia on developpers forum, they will firstly ship "modules only" so no thermal solution, no devkit of base board.

no Pcie 4.0, no SATA, no 256bit memory Bus, but still looks like a good value.

For the 2/4/6 cores, 10w operates at 2 and 4 cores where 15w mode enables 6 core operations and higher clocks in 2 and 4 cores
Posted on Reply
Add your own comment