Hardware manufacturer Nvidia claims to have developed the smallest and most powerful supercomputer for artificial intelligence (AI) and embedded devices.
The Jetson Xavier NX is a high-performance, low-power module that still delivers performance comparable to data center servers. It delivers up to 21 Theoretical Operations Per Second (TOPS) when running AI workloads. In addition, the power consumption does not exceed 15 watts.
Another important feature of the device is its size. It is smaller than a credit card and can be installed in almost any device anywhere, according to the manufacturer.
Due to its small size, low power consumption and performance, it is well suited for use in small commercial devices that function well in conjunction with edge computing, such as robots, drones and IoT devices with sensitive sensors.
According to the recently published MLPerf Inference 0.5 benchmarch, Nvidia scored remarkably well against similar chips from competitors Intel, Google and Qualcomm. The Jetson Xavier NX is powerful enough to run multiple neural networks in parallel while processing data from multiple sources in real-time.
The Jetson Xavier NX is based on Nvidia’s CUDA-X AITM architecture in combination with Nvidia’s JetPack software development kit, which includes libraries for deep learning, graphics and AI-based workloads.
The Jetson Xavier NX module will be available from March 2020 at a suggested retail price of $399.