The new board is intended to quicken the design of AI IoT face/object detection, image processing, and 4K video playback applications.
The combo board uses the SMARC 2.0 industry standard with an 82mm x 80mm form factor. The SMARC SoM board provides the choice of three different scalable versions of the Renesas 64-bit RZ/G2 MPU: a RZ/G2N dual core Arm Cortex-A57 MPU operating at 1.5 GHz for mid-range performance; the RZ/G2M MPU with dual-core Arm Cortex-A57 and quad-core Arm Cortex-A53 (1.2 GHz) for high performance; and the RZ/G2H MPU with quad-core Arm Cortex-A57 and quad-core Arm Cortex-A53 for ultra-high performance. The three MPUs offer integrated 600 MHz PowerVR 3D graphics and 4K UHD H.265 and H.264 codecs for different processing requirements.
“AI and video processing blend well together in embedded vision systems. The SMARC 2.0 winning combo reference design offers multiple processors from the Renesas RZ/G2 Family, making it an excellent AI development platform for accelerating time to market with reduced customer risk,” said DK Singh, Director, Systems and Solutions Team at Renesas. “Our scalable solution’s extensive on-board interfaces, large memory and programmable clocks for different needs makes it an excellent turnkey solution for a wide range of applications. And its industry leading RZ/G2 MPUs with 3D graphics engine and 4K UHD and full HD video codec provides higher performance per dollar than competing 64-bit MPUs and GPUs.”
The processors are supported by 2GB to 4GB LPDDR4 RAM, and 32GB eMMC. Each RZ/G2 MPU can running edge video analytics and AI frameworks. The MPUs have an integrated AI software library, comprehensive set of interfaces, ECC protection on both internal and external memories, Linux OS, and a Verified Linux Package (VLP) tested and maintained by Renesas. The boards also offer a Civil Infrastructure Platform (CIP) Super Long-Term Support (SLTS) kernel, and a Linux kernel bundled with a software development environment. The SMARC SoM board also features an optimized power and programmable