Jetson Xavier NX: A Compact AI System-on-Module Supercomputer

Taking supercomputer performance to the edge in a tiny module is not easy. However, the Jetson Xavier™ NX delivers this exceptional performance, making it ideal for embedded edge computing devices and robotics.  Therefore, if you have a project in either of these areas, you need to understand how the module works. Here is an in-depth look at what the Jetson Xavier NX offers.

What is Jetson Xavier NX?

The Jetson Xavier™ NX is the most compact SOM (System-on-Module) with enough power for advanced AI applications. This Jetson Nano-size supercomputer is smaller than a credit card and is highly energy-efficient, consuming only 10 watts of power.

The NVIDIA Jetson Nano Development Kit

Source: Wikimedia Commons.

Compared to its predecessor (NVIDIA Jetson TX2), the Xavier accelerates the NVIDIA software stack with over 10X the performance. Additionally, it has the new cloud-native support across the entire NVIDIA Jetson platform. Also, the manufacturer provides the latest tools for application development to support the device.

Jetson Xavier NX Module Features

• Compact form factor but powerful enough to run advanced, low power consuming AI applications
• Over 10X of the NVIDIA Jetson TX2’s performance
• Supports app development and optimization across all the NVIDIA software stack
• Supports AI application development using the NVIDIA Jetpack SDK
• Receives seamless updates throughout its lifetime
• Reduces development costs while being a flexible & scalable platform
• Lastly, it simplifies AI building, deployment, and management at the edge

Interfaces

The Jetson Xavier NX has multipurpose interfaces falling into the following categories.

Jetson Xavier NX Developer Kit

This developer kit provides a simple way to get started with the Xavier module. It includes the following components.

• The Xavier module with a fan/heat sink
• A 19.5V power supply
• A reference carrier board (90x103mm module size)
• 2x2 wireless modules (Bluetooth 5.0 M.2 module + 802.11ac WLAN + antennas)

You will require a microSD card, with the recommended one being 16 to 32 GB UHS-1 minimum. Altogether, the kit has the following ports.

• Four USB host ports (3.1A)
• One USB micro-B device port
• Two MIPI CSI-2 interfaces (15-position flex camera connector)
• HDMI 2.0 port
• RJ45 gigabit ethernet port
• Displayport 1.4
• M.2 slot with PCIe key-E (for the wireless module)
• M.2 NVMe slot with PCIe key-M
• MicroSD card slot
• Two I2C & SPI and one I2S, GPIO & UART port

Developer Tools

Like with other NVIDIA Jetson boards, you can develop applications for this energy-efficient Jetson Xavier NX module using NVIDIA's jetpack. After generating the software image, flash it to the SD card, then boot the device. In addition, the jetpack bundles all the platform software with Jetson Linux, Ubuntu 18.04 (kernel 4.9) to be specific.

Additionally, the jetpack provides cuDNN, CUDA, TensorRT, and DeepStream for parallel computing and AI applications. Tools like tegrastats and nvpmodel also come with the filesystem. Tegrastats gives GPU and CPU stats while nvpmodel controls the performance & power profile.

The GPU in the Jetson Xavier™ NX module supports OpenGL 4.6, Vulkan 1.2, and OpenGL ES 3.2. Ensure to go through the following steps for correct jetpack installation.

NVIDIA Jetson AI Modules Comparison

The NVIDIA Jetson Nano developer kit

Source: Wikimedia Commons

The NVIDIA P2972 AGX Xavier Developer kit

Source: Wikimedia Commons

Summary

In conclusion, the NVIDIA Jetson Xavier NX is a fast, compact supercomputer ideal for AI inference training (using neural networks, deep learning, etc.). This AI development in a small form factor makes the module a suitable option for edge computing in smart-city, healthcare, agriculture, and several other applications. If you want the device or have any questions regarding the article, contact us or leave a message in the chat.