English
At Embedded World 2025, Renesas introduced the newest member of its mid-class AI processor line-up. The RZ/V2N microprocessor (MPU) is built around four Arm Cortex-A55 cores and one Cortex-M33 core, along with an Arm Mali-C55 image signal processor (ISP).
Johannes Brücker , Renesas' business development manager MPU channel EMEA, shows off the new RZ/V2N dual-camera capable AI MPU.
Artificial intelligence (AI) power comes from Renesas’ proprietary AI accelerator, the AI3 dynamically reconfigurable processor (DRP).
High Performance—Low Size, Power Consumption, and Heat
The RZ/V2N (datasheet linked) joins the mid-range of Renesas RZ/V2x MPU lineup with multiple cores and a broad feature set:
- Four Arm Cortex-A55 CPU cores at 1.8 GHz
- One Cortex-M33 system management core with on-chip FPU and DSP extensions at 100 MHz
- Renesas' AI3 dynamically reconfigurable AI accelerator
- Arm Mali-C55 image signal processor
- Accelerators:
- 3D graphics engine (GE3D) (option)
- Image signal processor (ISP) (option)
- Image scaling unit (ISU)
- Video codec unit (VCD)
- 1.5 MB of on-chip SRAM
- Two-channel MIPI camera interface
- Extensive I/O set covering all major protocols, including six-channel CAN/CANFD, Ethernet, and USB 3.2
RZ/V2N MPU block diagram.
Renesas optimized the high-performance RZ/V2N MPU for low physical size, power consumption, and heat generation. It performs without cooling fans and comes in a small 15-mm x 15-mm BGA package. Its ability to perform AI edge computing without active cooling opens up applications that are noise- and space-constrained.
Integrated AI and Dual-Camera Edge AI Processing
Renesas' new MPU targets stereoscopic vision systems and can simultaneously capture double-angle images from two cameras. The dual camera processing provides data for coordinated analysis, such as 3D object detection. The cameras can also operate independently. For example, one camera might count cars in a parking lot while the other reads license plates.
The new MPU integrates an AI accelerator for advanced vision AI.
The MPU delivers solid conventional computing power and edge AI for improved image processing. The AI comes in the form of Renesas’ proprietary DRP-AI3 dynamically reconfigurable processor. The AI3 operates at an efficient 10 tera operations per watt (TPOS/W) with 15 AI inference TOPS. The 5 TOPS boost is delivered through intelligent pruning, a technique for eliminating non-relevant data from AI calculations.
The AI3 operates independently of the main CPU. Its DRP core contains 96 AI processing elements and an AI-MAC (matrix multiply-accumulate) unit with 2,048 MAC cells of INT8 format. Both sections have their own DMA controller for faster data transfer. It works in concert with the DRP-AI driver to convert trained AI models and accelerated inference.
Dynamically Reconfigurable
The AI processing elements and MAC cells, formed as grids, can be configured at run-time. The MPU shares some of the advantages of a field programmable gate array (FPGA); its reconfigurability allows the DRP to be optimized for specific types of computation, much like a gate array. This ability gives the MPU flexibility and increases its ability to perform parallel operations when needed.
As a coprocessor/accelerator, the AI3 functionality can be used as software functional calls from the main set of computational cores. The main cores prepare the data along with the image processing core and act on the responses, while the AI3 handles inference and machine learning. The AI inference capability targets image recognition and interpretation for applications such as machine vision, driver monitoring systems, mobile robotics, and security.
Support and Availability
The RZ/V2N sits in the middle of Renesas’ vision AI MPU line. It offers a broad performance range and low power consumption for mid-tier applications.
Role of RZ/V2N in the RZ/V product lineup.
Renesas offers a full development stack along with multiple reference designs. One of the most complex reference designs is its two-camera AI dash camera reference solution. While edge AI and vision systems are not new, the ability to combine two cameras and so much functionality into a single MPU solution is. Renesas developed the dual-camera AI solution to help product designers launch and evaluate the MPU faster than if they had to start from scratch. The dual-camera solution includes:
- Hardware and software for real-time vision inference.
- Front-facing advanced driver assistance system (ADAS) camera and cabin-facing driver monitoring system (DMS) for enhanced situational awareness
- Instant system wakeup and recording from a low-power standby mode for event recording
- Touch control, LCD display, and voice control
The RZ/V2N and the RZ/V2N quad-core vision AI MPU evaluation kit are available now through standard Renesas distribution .
