Nvidia Introduces Jetson Nano Board With 128 Maxwell Cores For $99

During the GPU Technology Conference, Nvidia announces the Jetson Nano, a development board with arm-soc and GPU for $ 99, which can be used for machine learning, among other things. In addition, a module for production purposes appears.

Thanks to the presence of 128 GPU cores based on the Maxwell architecture, the Jetson Nano Developer Kit offers 472GFlops of peak performance in FP16 calculations. In addition, a soc with four arm A57 cores at 1.43GHz is present, as well as 4GB lpddr4. The storage must come from micro SD cards.

The board can output images via hdmi 2.0 or displayport 1.2. In addition, four USB 3.0 ports are present and an Ethernet interface and an m2 connection for PCI-e x4 are present, as well as gpio, i²C, i²S, SPi and uart. A camera can be connected via MIPI. The size of the board is 100x80mm and the consumption is between 5 and 10W.

The Jetson Nano module that Nvidia also supplies contains a 16GB emmc storage module and can connect twelve cameras via MIPI. The module also has connections for a single USB 3.0 port, supplemented with three 2.0 variants. The size is slightly smaller than that of the development board: 69.6x45mm.

The Jetson Nano development board is intended for the maker community of developers and students. Nvidia has opened the Jetson developer forum for them where they can go with technical questions. The Nvidia Jetson Nano Developer Kit is available immediately for $99.

The Jetson Nano module costs $129 when purchasing a thousand units. Delivery will start in June. The module is intended for companies that want to release compact and economical products with some features related to artificial intelligence, such as robots and intelligent gateways.