Today, artificial intelligence processes began to appear in every field. There are usually two options for manufacturers or platforms that want to run artificial intelligence algorithms. If the artificial intelligence algorithms are intensive, cloud infrastructures are preferred or if they are rare, they are run on the system like mobile devices.
However, in systems such as portable medical devices, small commercial robots, Drone models, intelligent sensors for factory logistics, optical inspections, network security cameras, artificial intelligence algorithms need to work on the local server. This is called extreme computer systems.
See Also “The first results for the MLPerf Inference artificial intelligence benchmark test began to come”
Nvidia has important products for this sector called Edge Computing . Last week, the world’s smallest supercomputer for this sector was announced. The system called Jetson Xavier NX will take over artificial intelligence on end server systems. Moreover, the bare module is the size of a credit card.
The new system includes a processor with 6 customized ARM cores. Level 2 cache is 6MB, while Level 3 cache is 4MB. On the graphics side there are 384 CUDA cores based on Nvidia Volta and 48 Tensor cores.
LPDDR4X memory with 8GB 128-bit bus has 51.2GB / s bandwidth, while 2x 4K30fps video solution and 2x 4K60fps encryption are available. Up to 6 CSIs and up to 12 cameras can be connected via MIPI CSI-2. The module that supports Ubuntu runs 2 NVDLA artificial intelligence software. The Jetson Xavier NX module offers 14 TOPS for 10W power consumption and 21 TOPS for 15W power consumption.
Except for the Jetson Xavier NX, the Jetson family is quite crowded. Jetson Nano, Jetson AGX Xavier and Jetson TX2 series are looking for market share in different sectors. The Jetson Xavier NX has a $ 399 price tag.