Renesas RZ/V2N MPU with proprietary AI accelerator for vision AI

Renesas Electronics has expanded its RZ/V series microprocessors by introducing RZ/V2N, designed for vision AI applications in smart factories and intelligent cities. The device uses the company’s proprietary AI accelerator, DRP-AI3 (dynamically reconfigurable processor) offering high power efficiency and optimized AI inference performance. The AI accelerator delivers up to 15 TOPS of performance with 10 TOPS/W of power efficiency. It uses a pruning technique to optimize AI models by removing unnecessary parameters while retaining their accuracy.

The Renesas RZ/V2N has double-angle image processing capability, leveraging two camera inputs to improve spatial recognition and motion analysis. The technology is crucial for applications requiring enhanced depth perception, precise movement tracking, and multi-angle visual analysis. Inside the MPU is a multi-core architecture featuring four Arm Cortex-A55 cores with one Arm Cortex-M33 core. The processor also has an Arm Mali-C55 image signal processor.

Specifications of Renesas RZ/V2N Microprocessors:

• CPU: On-chip quad-core 64-bit Arm Cortex-A55 core application processor and a 32-bit Arm Cortex-M33 system management processor
• AI accelerator: DRP-AI (dynamically reconfigurable processor for AI) with 3D graphics engine, image signal processor, image scaling unit, and video codec unit
•  Memory: On-chip shared 1.5 Mbyte SRAM with ECC, external DDR memory interface (1-channel memory controller for LPDDR4 at 3200 MT/s or LPDDR4X at 3200 MT/s with a 32-bit bus width, xSPI interface, and SDHI (eMMC/SD supported)
• Audio: I2S input/output interfaces, SPDIF input/output interfaces (3-channel)
• Interfaces: Ethernet, USB 2.0, USB 3.2 Gen2, PCIe Gen3, MIPI CSI-2, MIPI DSI, CAN/CANFD, SCI, SPI, I2C, I3C

According to Renesas, the MPU package measures only 15 mm square, which is claimed to be 38 percent smaller than the high-end RZ/V2H. It also has low heat generation, which makes it possible to eliminate the need for cooling fans, reducing the system size and cost.

Developers can use the Renesas evaluation board kit to test and evaluate their AI applications. The AI software development kit, available on GitHub, provides a framework for building AI models without requiring deep AI expertise.

The company offers 50+ AI use cases covering applications such as object recognition, human activity monitoring, industrial inspection, and smart surveillance. Renesas partners with third-party embedded manufacturers to provide SOM boards, single-board computers (SBCs), and camera modules that integrate the RZ/V2N.

The product will be available on March 19.