Once, there was a clear distinction between microprocessors (MPUs) and microcontrollers (MCUs), as they served entirely different functions. However, with the increasing richness and performance of modules integrated into MCUs, this boundary has gradually begun to dissolve.
In recent years, many manufacturers have introduced MCU products that combine the functionalities of both MCUs and MPUs. The organic fusion of MPUs and MCUs has begun, retaining the real-time capabilities and control of traditional MCUs while achieving performance levels on par with MPUs, allowing users to develop specific application solutions without compromising on real-time control or performance.
Published by: Electronic Engineering World (ID: EEWorldbbs)
In fact, the debate regarding MCUs and MPUs has been ongoing for many years.
Traditionally, MPUs typically use Arm Cortex-A series cores, featuring rich peripheral interfaces, while MCUs mostly utilize Arm Cortex-M series cores, which are cheaper and easier to install and use than MPUs.
A simpler way to distinguish between the two is that running Linux indicates an MPU, while using a real-time operating system (RTOS) indicates an MCU. The purpose of an MCU is to run a relatively simple control loop continuously, or until it is interrupted or stopped.
As technology and applications have developed, MCUs can now perform many tasks, including more complex peripheral devices. Some MCUs now come with simple software drivers for more complex peripherals, further blurring the lines with MPUs.
Currently, MCUs and MPUs are very similar. The mainstream classification methods mainly include two ways: one is through subtle parameter differences, such as CPU functionality, bit count, operating systems, timing requirements, core counts, etc.; the other is by application classification, where MPUs focus on executing complex and diverse large programs with powerful computing/processing capabilities and external large-capacity memory, while MCUs typically run relatively singular tasks, managing/controling hardware devices, often integrating small-capacity memory in a single chip to achieve system “monolithic” integration.
However, some technicians still argue that MCU and MPU are outdated terms, as their performance has dramatically improved, and with the advancements in 3D packaging and Chiplet technology, many functions previously exclusive to MPUs have now been integrated into MCUs.
The industry prefers to use terms like “cross-border” to explain this trend and continuously introduces cross-border products.
Cross-border means integrating certain characteristics or hardware that were previously exclusive to MPUs into MCUs, such as integrating rich peripheral resources like camera input, display output, and USB from MPUs into MCUs.
This way, one can enjoy the low power consumption, low cost, and simplicity of MCUs while also achieving applications that were previously only possible with MPUs. From the actions of various chip companies, cross-border MCUs are a key focus of their layout.
ST: Initiating the Cross-level Plan
From 2007 to 2018, STMicroelectronics spent 12 years embedding “STM32 = 32-bit MCU = ARM Cortex MCU” in the public consciousness. In 2019, STM32 crossed over from MCU to MPU, launching its first MPU.
The flutter of a butterfly can stir up a storm across the sea. The STM32 “butterfly” not only stirred up the MCU storm but also entered the realms of engineers and households. Now, it also begins to integrate MCUs and MPUs.
On March 14, 2024, ST released a new high-performance product that combines the strengths of both MPUs and MCUs — the STM32H7Rx and STM32H7Sx, with plans to begin mass production in April 2024.
The STM32H7 series itself uses a Cortex-M7 core running at a frequency of 600MHz, offering very strong performance. ST has further segmented its existing products into two product lines: the STM32H7R3/S3 general-purpose MCU and the STM32H7R7/S7 with enhanced graphics processing capabilities.
The reason it is called cross-border is that both MCUs integrate dedicated graphics processors and fast storage interfaces, hardware configurations that are typically found in MPUs.
The STM32H7Rx and STM32H7Sx also have very robust security features, covering protection against physical attacks, memory protection, and runtime application code isolation protection functions. Additionally, the STM32H7S products have added more enhanced security features, integrating immutable trust roots, debugging validation, and hardware encryption accelerators.
The new products can run applications that typically require MPUs, allowing device manufacturers to develop smart appliances, smart building controllers, industrial automation, and personal medical devices more quickly and economically, meeting the growing demands of end-market users. Specific use cases include enhancing graphical user interfaces while performing multiple different functions.
Although ST’s new products have not yet hit the market, they have garnered widespread attention from engineers. Some engineers compare them with domestic SoCs, while others question their positioning.
Currently, it appears that this new product primarily targets low-cost and low-power graphic display applications, and more information may only be available once the product officially launches.
NXP: Filling the Gap Between MCU and MPU
When talking about who in the industry started laying out cross-border processors, it must be NXP. Since NXP launched the industry’s first cross-border processor product, the i.MX RT1050, in November 2017, NXP has continuously expanded its product line in this area.
The reason NXP developed the i.MX RT series is due to the increasing demands from many customers for higher performance from MCUs to facilitate the running of voice, image, Wi-Fi, and other applications, which existing MCUs could not meet, necessitating the use of Cortex-A series cores.
The operating system for Cortex-A cores is complex, and many engineers cannot meet these requirements. Moreover, real-time systems on Cortex-A cores cannot guarantee a good fit, which led to the birth of the i.MX RT series.
As the functionality of products increases, the requirements for storage units also rise, including flash memory and RAM. Previously, small MCUs had only a few kB or even Byte-level RAM, which is now far from sufficient for running various elements.
The advantage of large RAM is fast operation, and another benefit is enhanced security, as RAM can classify tasks to enhance security. The RAM in i.MX RT products is particularly large and performs exceptionally well. This includes the future use of new storage types such as RRAM, MRAM, and PCM.
TI: Introducing the R Core
Texas Instruments (TI) has also launched the Sitara AM2x MCU product that combines both MCU and MPU functionalities. Unlike the aforementioned manufacturers, TI’s products lean more towards real-time control, even incorporating Cortex-R series cores into MCUs.
Specifically, TI combines Cortex-R5F, Cortex-M4F, high-performance RAM, DDR, signal processing accelerators, and high-speed real-time bus interfaces for industrial communication, designing the AM2x series of processors.
By combining traditional MCUs and traditional MPUs, it achieves processor-level computational performance along with the design simplicity of MCUs, integrating real-time processing, control, and communication functions, while easily accessible tools and software help simplify the development process. Data shows that engineers can utilize 10 times the processing power compared to flash-based MCUs.
The reason TI launched such a product is due to the growing demand in the market — emerging applications require higher levels of system integration and edge intelligence in performance, while industrial and automotive systems rely on real-time control and decision-making in control, and distributed communication and automation trends require higher network bandwidth in communication.
So, why does TI call it an MCU rather than an MPU? In traditional definitions, an MCU integrates the CPU, Flash, RAM, and various peripherals into a single chip, suitable for processing a single task; while an MPU’s CPU, including CPU operation and value retrieval, has a completely different architecture for memory management, and from an application perspective, MPUs often need to run on more complex operating systems.
The Sitara AM2x series products combine processor-level high performance and MCU design simplicity, integrating different functions and peripherals. The AM2x series is primarily used in scenarios requiring high-speed computation, real-time control, and real-time communication.
Renesas: Ultra-High-Performance Core
On October 31, 2023, Renesas Electronics launched its first ultra-high-performance MCU based on the Cortex-M85: the RA8M1 series microprocessor.
Shortly after, on December 12, 2023, it released the RA8D1 microcontroller (MCU) product line, becoming the second product in the Renesas RA8 series.
In performance testing, the RA8 series is strong, achieving a score of 6.39 CoreMark/MHz, narrowing the performance gap between MCUs and MPUs.
The newly launched RA8D1 MCU primarily targets various graphic display and voice/visual multimodal AI applications in building automation, home appliances, smart homes, consumer electronics, and medical fields.
Infineon: Focusing on Automotive Applications
On December 8, 2023, Infineon launched a new series of automotive MCUs, the TRAVEO T2G-C series, equipped with a new graphics engine. On its official website, Infineon states that this product “can provide MPU-level performance for automotive graphic applications at MCU-level costs.”
From the core perspective, the TRAVEO T2G-C also chooses the Arm Cortex-M7 core, with two cores operating at a frequency of 320MHz. Meanwhile, the TRAVEO T2G-C series is equipped with dedicated graphics accelerators and utilizes new intelligent rendering technology, which can reduce the memory required for graphics processing by 3 to 5 times, lowering power consumption and costs, thus enabling the creation of dashboards, in-car infotainment, and cockpit systems with microprocessor performance at MCU costs.
In terms of storage, it features 6MB of flash memory and 4MB of internal video memory or 1GB of LPPDR4 video memory. This MCU series employs Infineon’s patented innovative line cache processing technology, which requires only 10% of the cache size of traditional frame buffer schemes compared to similar semiconductor devices on the market, thereby reducing power consumption, memory requirements, and BOM costs.
Of course, in addition to crossing from MCU to MPU, the reverse crossover from MPU to MCU is also a track, breaking the boundaries between high-end MCUs and low-end MPUs.
Cross-border MPUs involve using Arm Cortex-A series cores as the main processor, while also incorporating Arm Cortex-M series cores as co-processors. The benefit of this approach is that it maintains low power consumption and real-time performance while leveraging the capabilities of open-source software stacks, while also retaining the rich peripheral resources of traditional MPUs in areas such as camera input, display output, and USB.
The most typical examples are the products from ST and NXP.
ST: Reusing the STM32 MCU Ecosystem and IP
The STM32MP1 series is a typical cross-border MPU and also ST’s first cross-border STM32 MPU.
The STM32MP1 is based on Arm’s 32-bit single or dual-core Cortex-A7 and 32-bit Cortex-M4, operating at frequencies of up to 800MHz, providing optional 3D graphics processing units (GPUs) for advanced HMI development, making it a powerful computing solution.
The flexible architecture of STM32MP1 meets the performance requirements of high performance, hard real-time capabilities, low power consumption, and security; at the same time, it inherits the STM32 ecosystem, which includes various hardware development boards, three types of software development kits, etc., allowing previous control based on M4 to be ported to MP1, thus accelerating product development.
In simple terms, the Arm Cortex-A7 runs Linux, while the Arm Cortex-M4 runs STM32Cube, achieving cross-border functionality.
From the characteristic parameters, cross-border MPUs are very attractive. However, as stated in ST’s official documentation, when designing products with MPUs, one must consider multiple resources such as software, hardware, PCB, and also consider MCU migration and market regulations.
NXP: Also Adopting Dual-Core A and M Cores
NXP’s previously launched i.MX 7ULP is also a cross-border MPU.
This product is also a dual-core product with Cortex-A7 and Cortex-M4, allowing it to operate in the popular Linux operating system environment for MPUs while maintaining MCU-level low power consumption during standby.
The i.MX 7ULP is well-suited for applications requiring strong displays without needing long periods of intense computation while also communicating with networks/clouds in low-power scenarios. Examples include smartwatches, portable health monitoring devices, portable devices, home control, smart access, smart locks, etc.
The predecessor of this product was called i.MX 7D, which also featured a dual-core Cortex-A7 and Cortex-M4 design, but the power consumption was still not low enough. According to NXP’s introduction at the time, customers demanded cutting-edge performance, fast operating speeds, and good image quality while also requiring long standby times. Therefore, NXP completely altered the architecture, leading to the concept of i.MX 7ULP, which combines speed and low power consumption based on two completely separate internal architectures.
As of today, the industry has never reached a consensus on the distinguishing standards and definitions between MCUs and MPUs, and perhaps it never will.
However, the precise distinction between the two is no longer crucial. As engineers have said, “Isn’t the selection of devices during product design based on trade-offs between performance, cost, size, and power consumption? Who cares if it’s called an MCU or an MPU?”
[1] 雷峰网：正在消失的MCU与MPU之间的界限.2020.12.9.https://mp.weixin.qq.com/s/StcV_arKm5YbvxF2rH3vNg
[2] EEDesign：太厉害了，老专家告诉你MCU和MPU的演化历史.2022.10.12.https://mp.weixin.qq.com/s/hjYllyw2GPSsNW8aSCHTyA
[3] 贸泽电子：借助MPU快速打造消费类应用的方法论.2022.1.12.https://mp.weixin.qq.com/s/I1IhHSzofM_5xzoNInCudA
[4] AI电堂：从MCU到MPU,一场直播看STM32如何跨界.2019.11.28.https://mp.weixin.qq.com/s/wME1UnPlMED8q117lAwkfg
[5] ST：https://www.st.com/resource/en/product_presentation/microcontrollers-stm32h7rs-lines-overview.pdf
[6] 恩智浦MCU加油站：探索跨界处理器的星辰大海，i.MX RT1170无限可能.2020.12.24.https://mp.weixin.qq.com/s/Rsw6FZCa8UoY5Nm-9mKNpw
The Disappearing Boundary Between MCU and MPU 
MCUs That Are More Like MPUs 
Cross-Border from MPU to MCU 
MCU or MPU, No One Cares 
