Single-board computers have become increasingly popular ever since the first Raspberry Pi appeared back in 2012. Even Nvidia produces its own take on these tiny PCs using the name Jetson, and a new, and very cheap Jetson board has just gone up for pre-order.
The Jetson Nano 2GB follows previous entries including the TK1, TX1, and TX2 models, all of which relied on a Tegra processor. There's also a Nano 4GB, which launched early last year for $99. The Nano 2GB only costs $59 and Nvidia describes it as "ideal for learning, building, and teaching AI and robotics. This is backed up by a Linux environment to work within, CUDA-X software and tools, and a number of resources to help you learn, including a getting started guide, the open source JetBot AI robot platform, and a Jetson AI certification program.
The software stack includes CUDA-X accelerated libraries and the JetPack SDK, which is the Jetson platform software and includes TensorRT, cuDNN, CUDA Toolkit, VisionWorks, GStreamer, and OpenCV all built to run on Nvidia's LTS Linux kernel and Linux Driver Package (L4T).
As Tom's Hardware reports, the Nano 2GB uses a 64-bit quad-core ARM A57 processor running at 1.43GHz and is complimented by a 128-core Nvidia Maxwell GPU and 2GB of DDR4 RAM. Ports include a microSD card slot (for main storage), USB 3.0, two USB 2.0, HDMI, Gigabit Ethernet, CSI-2 Camera connector (including support for the Raspberry Pi High Quality Camera), micro USB, USB-C for power, and a 40-pin GPIO connector.
Even though the Nano 2GB is $40 cheaper than the 4GB model, you're pretty much getting the same single-board computer with 2GB less RAM and fewer USB 3.0 ports. The power port has also changed to be USB-C. It's available for pre-order now, and is expected to start shipping before the end of October.