Renesas Unveils R8P1: A 1 GHz Arm Cortex-M85 MCU for Edge AI
Renesas introduces the R8P1, a powerful 1 GHz Arm Cortex-M85 microcontroller with an Ethos-U55 NPU for real-time Voice AI and Vision AI applications.
Renesas, a leading semiconductor company, has announced the launch of its most advanced microcontroller to date: the R8P1. This powerful device operates at 1 GHz and is equipped with a 500 MHz Arm Ethos-U55 Neural Processing Unit (NPU) for edge AI applications. The R8P1 also features an Arm Cortex-M33 real-time core running at 250 MHz, making it ideal for applications requiring real-time analytics, such as Voice AI and Vision AI.
The R8P1 microcontroller is designed to deliver high performance and efficiency, making it suitable for a wide range of applications. It includes a 2D GPU, MIPI DSI and parallel RGB display interfaces, MIPI CSI and parallel camera interfaces, and I2S and PDM audio interfaces. These features enable the R8P1 to handle complex tasks such as real-time image and audio processing.
Key Specifications
- MCU Core**: Arm Cortex-M85 clocked at 1 GHz with Helium MVE (M-Profile Vector Extension) and 32KB I/D caches, achieving 7300+ CoreMarks. The Arm Cortex-M33 core runs at 250 MHz with 32KB I/D caches.
- GPU**: 2D drawing engine
- NPU**: Arm Ethos-U55 NPU delivering 256 GOPS at 500 MHz
- Memory & Storage**: 2 MB SRAM, 0.5/1MB MRAM, optional 4MB or 8MB flash memory for SiP products, 2x xSPI compliant octal SPI with XIP (eXecute In Place) up to 333 MB/s, 2x SDHI/MMC, 16-bit/32-bit external memory bus
- Display Interfaces**: Graphics LCD Controller (GLCDC) with parallel RGB, MIPI DSI
- Camera Interfaces**: 2-lane MIPI-CSI2 camera interface up to 720 Mbps per lane, 16-bit CEU parallel camera interface up to 5 MP
- Audio**: 2x SSIE (Serial Sound Interface Enhanced) with I2S/Monoaural/TDM audio data support, 3x PDM for microphone array
- Communication Peripherals**: 2x Gigabit Ethernet MACs (GMAC) with TSN support and 2-port switch, USB 2.0 High-Speed and Full Speed interfaces, 2x CAN-FD, up to 10x SCI (UART, Simple SPI, Simple I2C) up to 60 Mbps, 3x I2C, 2x SPI up to 166 Mbps
- Analog**: 2x 12-bit ADC up to 23 channels, 2x 12-bit DAC, 4-channel high-speed analog comparators, temperature sensor
- System**: Low power modes, battery backup function (VBAT), Event Link Controller (ELC), 2x Data Transfer Controllers (DTC), 2x 8-channel DMA Controllers (DMAC), power-on reset, Programmable Voltage Detection (PVD) with voltage settings
- Timers**: 4x 32-bit high-resolution timers, 10x 32-bit PWM timers, 2x low-power asynchronous general-purpose timers, 2x 32-bit ultra-low-power timers, 2x Watchdog timers, RTC
- Security**: ARMv8-M TrustZone Security, Renesas Secure IP (RSIP-E51A), AES-128/192/256, CHACHA20, RSA4K, ECC, SHA-2 (224/256/384/512), SHA-3, On-chip immutable ROM for First Stage Bootloader with hardware Root-of-Trust, OTP (immutable storage), Octal SPI with Decryption-on-the-fly (DOTF), CMAC/HMAC/GMAC, Secure authenticated debug, Tamper protection with DPA/SPA side channel protection
- Supply Voltage**: MCU (without flash) – 1.62 to 3.63V for VCC/VCC2, SiP – VCC: 1.62 to 3.63V; VCC2: 1.7 to 2V
- Packages**: 224-pin (BGA224 11×11 mm), 289-pin (BGA289 12×12 mm), 303-pin (BGA303 15×15 mm) packages
- Temperature Range**: 0 to 95°C or -40 to 105°C
- Manufacturing Process**: TSMC 22nm ultra-low leakage
Development Support
The R8P1 is supported by the e2 Studio IDE, the Flexible Software Package (FSP) with FreeRTOS, Azure RTOS, and Zephyr OS. The company also provides the RUHMI (Renesas Unified Heterogeneous Model Integration) framework, which supports AI deployment on MCUs and MPUs. This framework includes native support for machine-learning AI frameworks such as TensorFlow Lite, Pytorch, and ONNX, along with all necessary tools, APIs, code generator, and runtime.
Evaluation Kit
Renesas has also introduced the EK-RA8P1 evaluation kit, which includes an R7KA8P1KFLCAC MCU (BGA289) coupled with 64MB SDRAM and 64GB OSPI flash. The kit offers a Gigabit Ethernet RJ45 port and access to I/Os through standard headers and connectors, including mikroBUS headers, SparkFun Qwiic connector, two Grove (I2C and Analog) connectors, two Pmod (SPI/UART/I2C) connectors, and Arduino Uno R3 headers. It ships with a 7.0-inch, 1024×600 parallel LCD board and a 5MP camera module, as well as preloaded demonstrations that showcase the RA8P1 capabilities out of the box.
Winning Combinations
Renesas has released reference designs for the RA8P1, including a video conferencing camera with AI capabilities, an AI drawing robot arm, and an AI-enabled surveillance camera. These designs are available on Renesas’ Applications page. Third-party solutions leveraging the RA8P1 are also available from partners such as Nota.AI and Irida Labs, further expanding the ecosystem for this advanced microcontroller.
The R8P1 represents a significant step forward in the integration of AI and real-time processing capabilities in microcontrollers, making it a valuable asset for developers and businesses looking to leverage the power of edge AI.
Frequently Asked Questions
What is the main application of the Renesas R8P1 microcontroller?
The Renesas R8P1 microcontroller is primarily designed for edge AI applications, including Voice AI and Vision AI, which require real-time analytics and processing.
What is the clock speed of the Arm Cortex-M85 core in the R8P1?
The Arm Cortex-M85 core in the R8P1 operates at 1 GHz, providing high-performance processing capabilities.
What is the role of the Arm Ethos-U55 NPU in the R8P1?
The Arm Ethos-U55 NPU in the R8P1 delivers 256 GOPS at 500 MHz, enabling efficient and powerful neural processing for AI applications.
What security features does the R8P1 offer?
The R8P1 includes ARMv8-M TrustZone Security, Renesas Secure IP, various encryption algorithms, secure boot, and tamper protection, ensuring robust security for sensitive applications.
What development tools are available for the R8P1?
The R8P1 is supported by the e2 Studio IDE, Flexible Software Package (FSP), and the RUHMI framework, which provides native support for popular machine-learning AI frameworks.