Issue link: https://resources.mouser.com/i/1442826
23 AI embedded systems are a great deal more powerful than their predecessors in terms of computing, IO, and memory capabilities. With this increase in capacity comes the desire to run a broader and richer set of applications, previously only supported on much larger server platforms. They are capable of running more complex PC-style distributions like Ubuntu, CentOS, and Debian, and doing all the development and build on target itself. NXP was able to read these shifts in embedded development trends early enough to add a flexible model into our Layerscape SDK (LSDK) that allows customers to share a standard set of components across multiple embedded and PC-style distributions. It provides customers with the ability to use an Ubuntu user-land with an NXP provided kernel and bootloader and get access to the Ubuntu ecosystem. They can use this to experiment with different software packages and libraries to develop their application and then use the same kernel in their production systems. NXP also contributes these components to open-source communities like yoctoproject. org and openwrt.org, so customers don't have to maintain a proprietary fork from NXP for NXP platforms and another fork for another vendor. This speeds up the software development and maintenance process considerably. Most of the cutting-edge work around AI/ML algorithm and model development happens in the cloud or on developer workstations, which are PCs running a PC distribution. If you have an embedded platform that supports these distributions, getting the latest updates or trying out a new algorithm can be as simple as: A. Do a Google search for "how to install XYZ on Ubuntu" B. Type the commands provided on to your Layerscape board console C. Sit back and enjoy your coffee, while the distribution makes short work of the whole build and install process. Later, I decided to experiment with Mobile-net SSD for better accuracy but needed Tensorflow. LSDK had me up and running again in a few minutes. Eventually, I will choose one algorithm and may need to use an embedded distribution like Yocto to get into production. In the meanwhile, LSDK with Ubuntu rocks! ■ NXP's Layerscape Software Development Kit. LX2160A • Provides a comprehensive platform that enables the design and evaluation of the LX2160A processor • Pre-loaded with a board support package (BSP) based on a standard Linux kernel • Enables network intelligence with the next generation Datapath (DPAA2) LEARN MOREu