NVIDIA Jetson Nano: A Raspberry Pi-Style Dev Kit For AI Researchers And DIY Crowd
Last year NVIDIA launched the Jetson AGX Xavier board which aimed to be the brains behind next-gen AI devices. NVIDIA has now announced the Jetson Nano developer board that it says will enable the development of millions of new and small low-power AI systems. NVIDIA is aiming the Jetson Nano at IoT applications including entry-level Network Video Recorders, home robots, and intelligent gateways that offer full analytics capabilities.
Jetson Nano is very small at 70 x 45mm; in fact it's the smallest Jetson device that NVIDIA offers. Jetson Nano is a production-ready System on Module (SoM) that can be used across multiple industries. Nano uses a Maxwell GPU with 128 CUDA cores. The CPU on the board is a quad-core ARM Cortex-A57 MPCore processor paired with 4GB of RAM and 16GB of eMMC 5.1 flash storage.
The board is capable of encoding video in 4K at 30 fps and decoding 4K at 60 fps in H.264/H.265 format. A 12 lane 3x4 or 4x2 MIPI CSI-2 DPHY camera interface supports up to 1.5 Gbps. A gigabit Ethernet port is integrated, and displays can be connected via integrated HDMI 2.0 or DP 1.2 with support for two screens at ones.
Connectivity includes 1 x 1/2/4 PCIe, a single USB 3.0 port, and three USB 2.0 ports. I/O includes a single SDIO, dual SPI, six 12C, and two 12S. The board also has a 260-pin edge connector. NVIDIA will ship the Jetson Nano Module starting in June; it can be pre-ordered now for $129. An NVIDIA Jetson Nano developer kit featuring the Jetpack SDK and a collection of ready-to-use projects is available now from distributors and costs $99.