31-10-2020 | | By Robin Mitchell
Recently, NVIDIA announced that it is selling its credit card-sized single-board Jetson computer for $59. What is Jetson, what are its features, and why is this announcement important?
NVIDIA is a multinational technology company based in the United States of America, and its main focus is the development of graphics processing units (GPUs), but also develop system-on-chips (SoC), for use in the mobile field that integrates a CPU, GPU, and other supporting components. Founded in 1993, NVIDIA has a yearly revenue stream of $11.716 billion, has over 18,000 employees, and serves products worldwide. Initially, GPUs were the primary focus for NVIDIA, but as it became more apparent that GPUs could be used for other applications including AI and machine learning, NVIDIA has started to expand their product range as well as creating platforms that allow developers to use their GPUs for alternative purposes. Now, NVIDIA has made many acquisitions of other companies including ARM Holdings (in September 2020) and also has plans to build the most powerful supercomputer in the UK.
Tegra is the family of SoCs developed by NVIDIA for use in mobile applications. The SoCs integrate an ARM CPU, GPU, Northbridge, Southbridge and memory controller to enable the creation of low power, high-performance computing systems and can help to dramatically reduce the total number of ICs needed to create a computer. One device of particular interest is the Tegra X1 SoC which integrates a 64-bit quad-core ARM Cortex-A57, NVIDIA Maxwell GPU, and MPEG-4 HEVC encoding/decoding support. The Tegra X1 is found in many portable device applications including the Nintendo Switch, but it also the SoC that makes up the core of the Jetson Nano single-board computer (produced by NVIDIA).
NVIDIA announces $59 Jetson Nano computer
Recently, NVIDIA announced that it would be releasing a Jetson Nano 2GB single board computer kit that will sell for just $59 in an attempt to make AI more accessible. The Jetson Nano is a credit card-sized computer that incorporates the Tegra X1 SoC, and the Maxwell NVIDIA GPU includes 128 CUDA cores and is clocked at 1.43GHz. While the price of the Jetson Nano is marginally higher than that of the Raspberry Pi 4, the Jetson Nano is stated to perform 8 to 37 times faster than the Raspberry Pi 4 in AI applications thanks to the Maxwell GPU and its ability to perform 427 GFLOPS. The Jetson Nano 2GB includes multiple connectors and ports including USB 2/3 connectors, HDMI, GPIO, camera, and M2 key E connector.
Areas of application that the Jetson Nano 2GB will help accelerate include robotics, automation systems, and self-driving experiments. The combination of the Jetson Nano 2GB high processing power with the AI support provided by NVIDIA creates a powerful AI platform that can quickly process data from cameras enabling object recognition. The low price entry of the Jetson Nano 2GB also creates an easier entry to the market for students and educators thus enabling quick development of AI systems. While the Jetson Nano 2GB is ideal for the development of AI systems, its high power consumption of 10W makes it non-ideal for battery-powered solutions.