While the price range of the Raspberry Pi has been widening (and going up) as variants with more memory are added, Nvidia is going in the opposite direction with the Jetson Nano by making it more affordable and more accessible. Nvidia has just announced a new 2GB variant of the Jetson Nano developer kit at the tempting price of $59. The board is essentially the same as the standard Jetson Nano, but has 2GB of RAM rather than 4GB. It uses the same processor which means you get an integrated 128-core Maxwell GPU running at a maximum of 921MHz, plus a 64-bit quad-core Arm Cortex-A57 CPU running at up to 1.43GHz. Nvidia quotes the total performance of the processor at 472 GFLOPs. The board also includes Gigabit Ethernet, a video processor — which can handle up to 4K 30fps encode or 4K 60fps decode — and support for PCIE and USB 3.0.
As with the first Jetson Nano, the video capabilities are impressive. While 4K 60fps decode means you could use it to watch high definition videos, the main point is that board can process multiple video streams (think about automated machines with multiple cameras) for object detection, tracking, and obstacle avoidance. Besides 4K at 60fps, the Jetson Nano is capable of decoding eight video/camera feeds at 1080p at 30 frames per second! Once decoded, the streams can be processed simultaneously by the machine learning algorithms for object tracking and more.
To hit the $59 price point, there are a few other differences between the 4GB and 2GB variants of the Jetson Nano. Most noticeable is that the 2GB version only has one USB 3.0 port (compared to four on the 4GB model), but adds two USB 2.0 ports. The DisplayPort socket has been removed from the board, but the HDMI port remains. Also, the development board is now powered via USB-C and the power brick is not included.
Another Raspberry Pi killer?
Since the release of the original Jetson Nano in March 2019, the Raspberry Pi has been upgraded. The newer Raspberry Pi 4 has a 1.5GHz 64-bit quad-core Arm Cortex-A72 processor, which is faster than the quad-core Cortex-A57 in the Nano. While the Raspberry Pi 4 has the edge in terms of CPU performance, the Nano is still champ when it comes to the GPU power. And for AI, that GPU power translates directly to machine learning performance.
Nvidia offers a unified development environment (the Nvidia JetPack SDK) across all of its Jetson boards. You could start developing a project on the Jetson Nano, but then if you need more GPU power then an upgrade to a more advanced Jetson board (like the Jetson Xavier NX) will incur little or no penalty from a software perspective.
The official Raspberry Pi 4 prices are $55 for the 4GB variant and $75 for the 8GB variant. This compares to $59 for the 2GB Jetson Nano and $99 for the 4GB version. The bottom line is that if you need brute CPU power then the Pi 4 is a better deal, but if you need machine learning performance, a faster GPU, and better video encode/decode, then the Nano is the better choice.
Jetson Deep Learning Institute (DLI) courses
The new Jetson Nano 2GB Developer Kit is great for students, educators, and enthusiasts to explore machine learning and robotics. It is designed to be an accessible platform for teaching, learning, and developing AI and robotics applications. To that end, Nvidia has also announced the availability of free online training and certification programs, which will supplement the existing open-source projects, how-tos and videos contributed Jetson developer community.
The Jetson DLI courses are made up of a series of hands-on self-paced educational tutorials, video walkthroughs, and project-based assessments aimed at educators and learners. The Jetson AI Fundamentals course covers the fundamentals of training/inference workflow, real time computer vision, classification and regression networks, plus object detection and semantic segmentation.
There will be two levels of certification included in the program: 1) The Jetson AI Specialist certification for students, makers, and hobbyists. 2) The Jetson AI Educator certification for educators and instructors who may want to teach AI in their own classes or courses.
When and from where?
The Jetson Nano 2GB Developer Kit will be available at the end of this month (October 2020) for $59 through Nvidia’s distribution channels. I should be getting my hands on a board very soon, so watch out for a full review here and on the Gary Explains YouTube channel.