Quad-core IP for Edge AI data processing
British ASIC design house Sondrel has launched the latest in its series of semi-custom chips for Edge AI and IoT systems.
The SFA 200 is a quad-core ARM chip with added security for remote collection and processing of video and data at the edge. The kernel can be used for a scalable single channel ASIC where custom modules can be added for smart meters, smart homes, smart factories, voice activated devices and infotainment.
Each ASIC is based on a CPU cluster for general tasks and an AI cluster for neural network tasks with memory and interface support. The cluster of four ARM A53 or A55 cores can be combined with an AI Ethos core linked by an 800 MHz network on chip. This has a 64-bit to 512-bit data path optimized for data flow with support for arbitration and Quality of Service (QoS) scheduling. A System Management Subsystem (SMS) manages the configuration, start-up (boot), and modes of operation of the SoC as well as active power management for battery-powered or low-powered applications.
The processor cores provide 9-10 GMIPS at 1 GHz while the Ethos AI core provides up to 4TOPS. Custom AI cores can be integrated as an alternative if needed. A built-in security subsystem uses an ARM A53 or M33 to provide activity / intrusion monitoring, software signing and cryptography, as well as authentication to resist hackers, which is further enhanced by tracking the behavior of the user. goalkeeper and gap detection.
This semi-custom design can reduce design costs by up to 30 percent.
Other articles on eeNews Europe