Enhancing embedded ML and AI range with dedicated processor and accelerator solutions

Mouser Electronics, Inc. continues to expand its portfolio of dedicated solutions optimised for ML operations.

A growing number of applications seek to incorporate AI and ML capabilities, driving demand for a diverse range of efficient embedded processors and MCUs. This new generation of devices, equipped with dedicated AI/ML accelerators, delivers tailored performance across different computing tiers – from low-power MCUs to dedicated FPGAs. These solutions are continually emerging, and the latest devices are crucial in helping engineers create smarter, more responsive, and energy-conscious end-products across various industries, including IoT, medical, robotics, and industrial automation.

The company offers a range of new ML-optimised products, including the following:

• Delivering a blend of powerful AI performance and cost efficiency, the RZ/V2N mid-range embedded AI MPUs from Renesas Electronics are designed for compact, low-power applications. These MPUs include a quad-core Arm Cortex-A55 processor, a Cortex-M33 for low-power operations, and a DRP-AI3 0 TOPS/W AI accelerator handling ML and AI tasks. Dual four-lane MIPI-CSI2 camera inputs, with optional ISP and hardware-accelerated H.264/265 codecs, enable advanced video processing. They also feature high-speed LPDDR4/4X memory, dual Gb Ethernet, and PCIe Gen3 connectivity. The RZ/V2N series' efficient design is ideal for applications such as AI cameras, driver monitoring systems, robotic vacuum cleaners, and parking monitors.
• Designed to deliver the next level of performance in cost-optimised applications, the Agilex 3 FPGA and SoC FPGA series from Altera employ the advanced Hyperflex Gen 2 core fabric technology, achieving a 1.9× improvement over the previous Cyclone V family. These devices feature high-speed transceivers, LPDDR4 memory support, and Intel's compact variable pitch BGA packaging. Equipped with dual Arm Cortex-A55 processors and hardened peripherals, they provide powerful embedded processing capabilities alongside AI tensor blocks embedded within the FPGA fabric for dedicated AI and ML acceleration. Applications span diverse sectors, including smart factories and IoT, healthcare, advanced consumer electronics (e.g., AR/VR, drones), and smart city infrastructure.
• Targeting low-power, high-performance embedded IoT applications, Silicon Labs' EFM32PG26 microcontrollers use an 80MHz Arm Cortex-M33 processor and a dedicated hardware AI/ML accelerator to supply fast edge inferencing for smart homes, cities, industrial automation, and general IoT deployments. The devices offer software compatibility with the EFR32 wireless SoC platform and come in 8mm x 8mm QFN68 and BGA136 packages, featuring up to 2MB of flash memory and 256KB of RAM. Additional security features include an LCD controller, Secure Vault® technology, hardware cryptographic acceleration, and ARM TrustZone support, making them well-suited to secure connected applications across diverse environments.
• From ROHM Semiconductor, the ML63Q2537 and ML63Q2557 microcontrollers feature a 32-bit Arm Cortex-M0+ CPU coupled with an AxlCORE-ODL AI accelerator to support embedded ML tasks. The MCUs provide an array of integrated functionality, featuring a CAN FD controller, 3-phase motor control PWM, dual A/D converters, analogue comparators, and I2C, SPI, and UART communication interfaces, while supporting an operating voltage of 2.3V to 5.5V and a temperature range of -40C to 105C. The ML63Q2537 offers 256KB ROM in a 48-pin package, while the ML63Q2557 provides the same ROM capacity in a 64-pin package. Each MCU is ideal for bringing intelligent functionality to smart household appliances, housing equipment, office devices, and measurement instrumentation.