For context, the Jetson Nano is more or less like the Raspberry Pi, but designed specifically to power through AI applications. Specs-wise, the Jetson Nano has 64-bit quad-core ARM A57 CPU running at 1.43GHz, a GPU powered by 128 CUDA Cores based on the Maxwell GPU architecture, and 4GB DDR4 RAM. Powered for the handled via a 40-pin GPIO. NVIDIA has been quite liberal with the I/O ports for the Jetson Nano. Providing four USB 3.0 ports, a single micro-B USB 2.0 port, HDMI 2.0 and eDisplayPort 1.4, a MIPI CSI-2 connector for camera applications, and a Gigabit Ethernet 10/100/1000 port.
Oddly enough, the Jetson Nano does not have any built-in Wi-Fi or Bluetooth module, but NVIDIA was kind enough to install an M.2 Key E slot on to the board. Should developers feel inclined to provide their own devkit with wireless connectivity. In terms of performance, NVIDIA promises that its new AI-driven developers kit will deliver up to 472GFLOPs of compute performance, along with the ability to encode and decode 4K video at 60 fps. More importantly, it is also able to perform AI workloads, such as object identification, motion tracking, and video smoothing. The best part is that it’ll do all that while running on as low as five watts of power.
The NVIDIA Jetson Nano Developer Kit is already available for US$99 (~RM403), while the Jetson Nano Module will cost US$129 (~RM525) and will only be available from June onwards. Both items can be purchased through the brand’s own stores, as well as through its primary global distributors. Unfortunately, there is no global distributor for Malaysia at this moment, meaning that interested parties may have to purchase the devkit through a proxy.
