Comparing the Nvidia Jetson embedded boards
Nvidia Jetson embedded boards are one of the most popular choices for a variety of IoT products and applications. In this article we will present the Nvidia Jetson board offering, together with their relevant use cases and specifications.
Nvidia Jetson Developer Kits and Nvidia Jetson modules
Before we dive into specific board descriptions, let’s talk a little bit about Nvidia Jetson developer kits. Developer kits include a Jetson module attached to a reference carrier board. They cannot be used for production purposes. Developer kits are used for testing and developing purposes only. Modules, on the other hand, can be used for deployment in production environments. Each Jetson module has no software pre-installed; the user has to attach it to the carrier board designed or procured for the end product, and flash it with the software image they've developed. In this article we will focus on modules only.
Nvidia Jetson Nano
Nvidia Jetson Nano is a small but powerful board designed for AI purposes. It supports all types of AI models and popular AI frameworks, which allows developers to use it for a variety of applications, such as image classification, segmentation, speech processing, object detection. Nvidia Jetson Nano is equipped with the NVIDIA Maxwell GPU with 128 NVIDIA CUDA® cores and 4 GB memory.
Nvidia Jetson TX2
According to the producer, Nvidia Jetson TX2 is the fastest and most power-efficient embedded AI board. Nvidia Jetson TX2 is a supercomputer-on-module, and it is equipped with 256-core NVIDIA Pascal GPU with 256 NVIDIA CUDA cores and 8 GB of memory, which makes it an excellent choice for intelligent edge devices such as robots, drones, intelligent cameras and more.
Nvidia Jetson Xavier NX
Jetson Xavier NX packs the power of the NVIDIA Xavier SoC into a module of the size of a Jetson Nano. This small AI supercomputer offers more than 10X the performance of Jetson TX2. It’s a great choice for innovative edge devices in industrial applications, retail, agriculture, healthcare, and more. It’s equipped with a 384-core NVIDIA Volta GPU with 48 Tensor Cores and 8 GB of memory.
Nvidia Jetson AGX Xavier
Nvidia Jetson AGX Xavier, the latest addition to the Jetson family of modules, provides 20× the performance as compared to the Nvidia Jetson TX2. It’s intended to be used in deep learning and AI applications involving image recognition, object detection, and speech recognition. It is equipped with a 512-core Nvidia Volta GPU with 64 Tensor Cores and 32GB of memory. The Nvidia Jetson AGX Xavier is currently the most powerful module offered by Nvidia.
Mender and Nvidia Jetson use cases
Mender users often choose Nvidia Jetson embedded boards as their modules of choice. We described some of them on our other blog, the Device Chronicle.
A very interesting use case for Nvidia Jetson embedded boards in agriculture is Bilberry. This French company is using Nvidia Jetson hardware with Mender in order to revolutionize agriculture. After extensive research, Bilberry discovered that a very common problem in modern agriculture is the overuse of herbicides. Using the Nvidia Jetson range of products together with Mender’s device management features, they designed an intelligent system that controls application of herbicides. You can click here to read the interview we did with Bilberry’s CTO.
Alcatraz.ai and Boulder AI
Other interesting companies that use Nvidia Jetson embedded boards in their products are Alcatraz.ai and Boulder AI. They both use Nvidia Jetson hardware and Mender to develop futuristic smart-city solutions. Alcatraz.ai designed a facial authentication solution using artificial intelligence. Boulder AI uses deep learning to provide smart city solutions, like a system monitoring vehicle speed in the city of Denver. You can read more about their work by clicking here.
Mender and Nvidia Partnership
Mender is Nvidia Jetson’s preferred partner for remote device management and OTA updates of their embedded modules. Click here to see an overview of the official integration between Nvidia embedded boards and Mender. We also have a very active community taking care of our open-source integration with Nvidia Jetson embedded boards. Click here to see all tutorials concerning the Mender - Nvidia Jetson integrations on Mender Hub.