Welcome Here  Shenzhen Mingjiada Electronics Co., Ltd.

Renesas Electronics Corporation, a global semiconductor solutions provider, recently announced the launch of the RA8P1 microcontroller (MCU) product line for artificial intelligence (AI), machine learning (ML) applications, and real-time analytics. This series of MCUS sets a new benchmark for MCU performance by combining 1GHz Arm® Cortex®-M85 and 250MHz Cortex-M33 CPU cores with the Arm EthosTM-U55 neural processing unit (NPU). This combination can achieve a maximum CPU performance of over 7300 CoreMark and an AI performance of 256 GOPS at 500 MHz.

Designed specifically for edge/terminal AI

The RA8P1 is specifically optimized for edge AI applications. By offloading the computationally intensive operations of the CPU in convolutional neural networks (CNNS) and recurrent neural networks (RNNS) with the Eethos U55 NPU, it achieves up to 256 MAC per/cycle performance and up to 256 GOPS at 500MHz. The new NPU supports most commonly used neural networks, including DS-CNN, ResNet, Mobilenet TinyYolo, etc. Depending on the neural network used, Ethos-U55 can achieve up to 35 times the number of inferences per second compared to the Cortex-M85 processor used alone.

Advanced technology

The RA8P1 MCU is manufactured using TSMC's 22ULL (22nm ultra-low leakage, ultra-low leakage) process, achieving ultra-high performance while maintaining extremely low power consumption. This process also supports the integration of embedded magnetic random access memory (MRAM) in new MCUS. Compared with flash memory, MRAM has a faster write speed, higher durability and stronger data retention capability.

Daryl Khoo, Vice President of Embedded Processing Marketing Division at Renesas, said: The demand for high-performance edge AIoT applications is experiencing explosive growth, and RA8P1 is precisely the MCU product we have launched to address this trend. It not only fully demonstrates our profound accumulation in the fields of technology and market, but also reflects the extensive cooperative ecosystem we have established with industry partners. Renesas' customers have shown a strong willingness to deploy this brand-new MCU in diverse AI scenarios.

Paul Williamson, Senior Vice President and general manager, IoT Line of Business at Arm, said: The pace of innovation in the era of artificial intelligence is faster than ever before, and new edge applications have put forward higher requirements for device-end performance and machine learning. Relying on the advanced AI capabilities of the Arm computing platform, the Renesas RA8P1 MCU can meet the demands of next-generation voice and visual applications and help expand the intelligent and context-aware AI experience.

Chien-Hsin Lee, Senior Director of Specialty Technology Business Development at TSMC, said: We are very pleased to see Renesas fully leverage the performance and reliability of TSMC's 22ULL embedded MRAM technology to deliver outstanding results for its RA8P1. As TSMC continues to advance embedded non-volatile memory (eNVM) technology, we look forward to strengthening our long-term cooperation with Renesas to jointly drive the innovation of breakthrough products in the future.

Rich peripherals, specially designed for AI

The RA8P1 launched by Renesas integrates dedicated peripherals, ample memory and advanced security features to support voice and visual AI as well as real-time analysis applications. For visual AI, this device features a 16-bit camera interface (CEU) that supports up to 5-megapixel image sensors, enabling visual AI applications that require the connection of cameras and have demanding computing power. The independent MIPI CSI-2 interface provides a low-pin-count interface with two channels: each channel has a maximum rate of up to 720Mbps. In addition, multiple audio interfaces (including I2S and PDM) support microphone input, meeting the requirements of voice AI applications.

The RA8P1 integrates on-chip storage and scalable external storage to achieve efficient and low-latency neural network processing. This MCU is equipped with 2MB SRAM, which is used to store intermediate variables or buffers. In addition, this product is equipped with 1MB of on-chip MRAM for storing application code, model weights or graphic resources. For larger-scale models, high-speed external memory interfaces can be provided. For more demanding AI application requirements, this MCU also offers SIP options in a single package with either 4MB or 8MB of external flash memory.

The brand-new RUHMI framework

Along with the RA8P1 MCU, Renesas RUHMI (Renesas Unified Heterogeneous Model Integration) was also launched. As a comprehensive AI compiler and framework for MCUS and MPUs, RUHMI offers efficient AI tools capable of deploying a variety of the latest neural network models. It supports model optimization, quantization, graph compilation and transformation, and generates efficient source code suitable for MCUS. RUHMI natively supports machine learning AI development frameworks such as TensorFlow Lite, Pytorch, and ONNX. It also provides the tools, apis, code generators, and runtime environments required for deploying pre-trained neural networks, including demonstration use cases for RA8P1. In addition, RUHMI integrates with Renesas' own e2Studio IDE, enabling seamless AI development. This integration will provide a universal development platform for MCUS and MPUs.

Advanced security features

The RA8P1 MCU builds cutting-edge security for critical applications. The brand-new Renesas Secure IP (RSIP-E50D) integrates numerous encryption accelerators, including CHACHA20, Ed25519, NIST ECC curves up to 521 bits, enhanced RSA up to 4K, SHA2 and SHA3. Working in conjunction with Arm TrustZone, this IP can achieve comprehensive and fully integrated functions similar to security components. The new MCU also features a powerful hardware root of trust and a secure boot function implemented in immutable memory (ROM) through the First stage boot loader (FSBL). The XSPI interface that supports Real-time decryption (DOTF) enables encrypted code programs to be stored in external flash memory and decrypted in real time when securely transmitted to the MCU for execution.

Ready-to-use solutions

Renesas has developed a series of user-friendly tools and solutions for the RA8P1 MCU, including flexible software packages (FSP), evaluation suites, and development tools, supporting FreeRTOS, Azure RTOS, and Zephyr. Renesas also offers multiple software sample projects and application instructions to help customers accelerate the time to market for their products. In addition, numerous partner solutions can also support the development of the RA8P1 MCU, including the driver monitoring solution from Nota.AI and the traffic/pedestrian monitoring solution from Irida Labs. For more solutions, please refer to the RA Partner Ecosystem Solutions.

Key features of RA8P1 MCU

• Processors: 1GHz Arm® Cortex®-M85, 500MHz Ethos-U55, 250MHz Arm Cortex-M33 (optional)

• Storage: 1MB/512KB on-chip MRAM, 4MB/8MB external flash SIP option, 2MB fully ECC protected SRAM, 32KB I/D cache per core

• Graphics peripherals: A graphics LCD controller supporting up to WXGA (1280x800) resolution, parallel RGB and MIPI-DSI display interfaces, a powerful 2D graphics engine, parallel 16-bit CEU and MIPI CSI-2 camera interfaces, and 32-bit external memory bus (SDRAM and CSC) interfaces

• Other peripherals: Gigabit Ethernet and TSN switches, XSPI (eight-wire SPI) with XIP and DOTF, SPI, I2C/I3C, SDHI, USBFS/HS, CAN-FD, PDM and SSI audio interfaces, 16-bit ADC with S/H circuits, DAC, comparator, temperature sensor, timer

• Security features: Advanced RSIP-E50D encryption engine, TrustZone, immutable memory, secure boot, tamper-proof, DPA/SPA attack protection, secure debugging, secure factory programming, device lifecycle management

• Packaging: 224BGA, 289BGA

Successful product portfolio

Renesas has combined the brand-new RA8P1 MCU with a wide range of compatible devices in its product portfolio to create a broad "successful product portfolio", including AI-enabled video conferencing cameras, AI drawing robotic arms, and AI surveillance cameras. These "successful product portfolios" are based on mutually compatible and seamlessly collaborative products, featuring technically proven system architectures that offer optimized low-risk designs to accelerate time-to-market for products. Renesas has now launched over 400 "successful product portfolios" based on various products in its product lineup, enabling customers to accelerate the design process and bring their products to market more quickly.

Supply information

The Renesas RA8P1 MCU product group is now available on the market, and the RA8P1 evaluation kit has also been launched.

Advantages of Renesas MCU

As a globally leading supplier of MCU products, Renesas' average annual shipment of MCUS has exceeded 3.5 billion units in recent years. Approximately 50% of these are used in the automotive field, while the rest are applied in industrial, Internet of Things, data centers, and communication infrastructure sectors. Renesas offers a wide range of 8-bit, 16-bit and 32-bit product portfolios, providing products with outstanding quality and efficiency, as well as excellent performance. Meanwhile, as a trusted supplier, Renesas has decades of experience in MCU design and is backed by a dual-source production model, industry-leading MCU process technology, and a vast ecosystem of over 250 partners.

For more news and information, visit www.hkmjd.com.