ARM or Advanced RISC Machine is a specific family of microarchitectures based on a specific instruction set architecture called reduced instruction set computing or RISC. It was developed by British semiconductor and software design company Arm Holdings. Computer processors based on the ARM architecture have been used in numerous consumer electronic devices such as smartphones and tablet computers, as well as gaming consoles, digital media players, smart devices, and even desktop and laptop personal computers and workstations.
The growing popularity of ARM over the x86 architecture from Intel, other architectures based on complex set instruction computer or CISC, and even those based on RISC has been attributed to its specific advantages and applications. The flexible business model of Arm Holdings has also provided semiconductor designers or fabless chipmakers with a compelling case to secure a license for the ARM architecture and has encouraged consumer electronics manufacturers to develop devices and build platforms using ARM microprocessors.
Understanding the Popularity of Advanced RISC Machine: Pros and Cons of ARM Architecture
ARM Pros: Advantages ARM Architecture and ARM Processors, and Notable Applications
1. Suitable for Battery-Operated Electronic Devices
One of the notable advantages of ARM architecture and the resulting ARM processors is their excellent performance per watt. This means that these processors can demonstrate an adequate level of processing capabilities with minimal power draw. ARM processors are well-suited for portable devices running on batteries. Take note that Android smartphones and tablets, as well as the iPhone and iPad products from Apple use chips based on the ARM architecture.
The main reason chipmakers and device manufacturers choose this architecture is to prioritize power efficiency without sacrificing performance. Furthermore, aside from being more efficient at consuming power, ARM processors also dissipate less heat than processors based on more complex architectures. Products that are running on these processors are less prone to sustained overheating and require simple heat management or hardware cooling solutions.
It is also important to underscore the fact the capability of a particular ARM processor to run on low power is suitable for the 3.5-watt Thermal Power Design requirement of portable consumer electronic devices. Intel Core and AMD Ryzen processors within the mid-range and highest-tier range struggle to scale down below 5 watts. The x86 architecture of these processors is more focused on targeting peak performance through a higher power draw requirement
Prioritizing power efficiency does not mean that processors based on the ARM architecture lack processing power. Several synthetic and real-world benchmark tests have revealed that newer generations of ARM processors an equal or exceed the performance of x86 processors in certain applications and use cases. The iPad Pro 2020 from Apple that runs on the ARM-based A 12Z Bionic system-on-chip outperforms mid-tier and some high-end personal computers.
2. Big.LITTLE Heterogeneous Computing Architecture
A significant feature of the ARM architecture that allows it to achieve low power consumption without sacrificing processing capabilities is heterogeneous computing. This is a specific system design consideration that involves using dissimilar and multiple processing cores or other main processing units called coprocessors within a single system or a single chip package. The ARM architecture uses the specific big.LITTLE architecture for this particular purpose.
Systems designed for heterogonous computing are both power efficient and have respectable processing capabilities. A general example is a particular system-on-a-chip used in mobile devices with components composed of multi-core central processing units and other coprocessors with specialized processing capabilities intended for handling particular tasks. Most modern chips include a multi-core CPU, an integrated graphics processor, and an AI accelerator.
A more specific example of heterogenous computing in the big.LITTLE architecture of ARM is the multi-core CPU of a Snapdragon chip from Qualcomm and the A-series and M-series chips from Apple. The multiple cores are categorized into two. The first set of cores is designed for handling low-powered tasks. These are often called efficiency cores. The second set of cores is designed for resource-intensive operations. These are often called performance cores.
The idea behind heterogeneous computing is simple. It involves building an architecture that enables different central processing unit components called cores with different performance and power capabilities to work together to balance between performance and power draw. The ARM big.LITTLE architecture which was first introduced in 2011 enables an ARM-based processor to share the workload across high-performance and low-performance CPU cores.
3. Inherits the Notable Advantages of RISC Architecture
Remember that the ARM architecture is based on reduced set instruction computing architecture. Processors based on this architecture have the notable characteristics and advantages of RISC architecture. These include a smaller number of instruction formats, fewer numbers of instructions, and fewer addressing modes. A particular RISC processor performs one thing per instruction, and it merges difficult commands into simpler ones when executing them.
Another important advantage of reduced set instruction computing architecture is that it uses a fixed length of instruction, which is easy to pipeline, because RISC functions use fewer parameters. There is also the time advantage. Each instruction takes up one cycle to maximize operation speed while minimizing the execution time. This architecture supports more registers to enable less time for loading and storing values in the memory or cache.
In addition, in consideration of fabless chip designers and chip manufacturers or semiconductor foundries, RISC processors are easier to design and ship than processors based on complex instruction set computer or CISC architecture. It also has a lower per-chip cost because the architecture requires smaller components. Less chip space is used due to the reduced instruction set. This is the reason RISC processors tend to be smaller than CISC processors.
The required transistor density or number of transistors is also fewer because the decoding logic is simple. More general-purpose registers can be fitted into the central processing unit. Advances in chip fabrication or process nodes have aided designers and manufacturers of ARM processors. The introduction of the 5nm process node and the subsequent 3nm process node have enabled the design and fabrication of chips with more cores and better processing capabilities.
4. Business Model and Licensing Scheme of Arm Holdings
There is a difference between Intel Corporation and Arm Holdings. Intel is a traditional and full chipmaker that both designs and manufactures chips. Arm Holdings designs the foundational instruction set architectures or microarchitectures. It licenses these architectures as part of its intellectual properties. It does not produce and sell chips. Companies such as Qualcomm and Apple purchase licenses and outsource chip production to foundries such as TSMC.
Arm Holdings offers three categories of licensing agreements. These are the core design license, Built on Arm Cortex Technology license, and architectural license. The licensing agreements grant specific licensees with the right to design their own chips using the licensed architectures. Some of the notable holders of Arm licenses include Apple, Advanced Micro Devices or AMD, Qualcomm Incorporated, MediaTek, Samsung Electronics, and Nvidia Corporation.
Part of the advantages of the ARM architecture comes from the licensing business model of Arm Holding. The model provides other chipmakers a considerable and higher degree of freedom and flexibility to control their costs. Companies such as Qualcomm and MediaTek can design chips for the purpose of selling them to device manufacturers. Others such as Apple can design their own chips to have better control and optimization over their product ecosystems.
There are downsides to the models of traditional chipmakers such as Intel. Manufacturers of computers and other electronic devices are at the helm of chipmakers. These manufacturers, and even software developers, would also need to build their hardware configurations and software around the processor, thus limiting their capabilities and degree of innovative freedom. Take note that these two are some of the main issues with the x86 architecture from Intel.
ARM Cons: Disadvantages ARM Architecture and ARM Processors, and Key Limitations
1. Incompatibility with Several Systems and Applications
The x86 architecture has been the standard architecture in computers for over two decades now. Manufacturers of desktop and laptop personal computers and workstations running on Windows have been using x86 processors from Intel and AMD. Apple started using Intel processors in its Macintosh computers beginning in 2005. The entire computer market has grown accustomed to developing and utilizing software and applications based on the x86 architecture.
Hence, based on the above, compatibility is one of the main limitations and disadvantages of the ARM architecture. This affects its wider adoption in the computer market. Note that Windows would not run on an ARM-powered computer with emulation. It is essential to underscore the fact that several general-purpose off-the-shelf applications were developed for x86 systems. Intel and AMD also have been dominating the personal and workstation computer market.
It is still worth mentioning that leading companies are also starting to consider options outside the de facto x86 architecture and the Intel and AMD choices. Microsoft has developed a version of Windows and other software applications for the ARM architecture as part of its partnership with Qualcomm. This further resulted in the introduction of several models of Microsoft Surface computers running ARM-based Snapdragon system-on-chips beginning in 2019.
Apple shifted from Intel Core processors to its custom-designed system-on-chips based on the ARM architecture beginning in 2020. It launched the first models of MacBook Pro, MacBook Air, and other Mac computers featuring the Apple M1 chip. Apple distributed preview devices to developers and rolled out the Rosetta 2 application translator to allow native x86 software applications to run on the new ARM-based macOS and computing environment.
2. Inherits the Issues or Disadvantages of RISC Architecture
The ARM architecture and ARM processors also share the drawbacks or notable disadvantages inherent to the RISC architecture. Take note that the time advantage of RISC has a disadvantage. Language compilers need to break down high-level instructions into multiple or different simpler instructions. Rearranging CISC code to a RISC code results in an increase in code size. The quality of this code will depend on the compiler and instruction set of the machine.
A particular RISC processor tends to require more instructions to perform complex operations compared to a counterpart CISC processor. This can translate to larger program sizes and increased memory bandwidth requirements. Reducing the instruction set also shifts the burden of development from hardware to software. RISK architecture can also struggle in handling certain classes of software applications that are dependent on complex instructions.
Furthermore, in considering the perspectives of these language compilers and other software developers, developing programs or applications for RISC and ARM requires more effort when compared to developing for CISC-based architectures such as x86 architecture. It is also worth mentioning that the performance of a particular ARM-based processor or system-on-a-chip will depend on the skills or capabilities of programmers or software developers.
The simplicity of a processor based on a reduced instruction set computer puts a lot of stress on the software application. This can have a detrimental impact on the software performance or overall operating system performance. Another disadvantage is that feeding instructions require very fast memory systems. This is the reason why specific architectures and the corresponding processors based on the general RISC architecture require large memory caches.
3. Dependent on Programmer and Developer Capabilities
It is important to reiterate and highlight the fact that complex instruction set computer or CISC architecture is focused more on the hardware. This is in contrast to a reduced instruction set computer or RISC architecture which is focused more on the software. CISC tends to be more hardware-specific and RISC is more software-specific. This means that these two broad categories of an instruction set architecture have different development environments.
The reason RISC architecture is software-specific and software-dependent comes from the nature of its instruction set. The instructions in this architecture are designed to be small and atomic. Each performs a single and well-defined operation to allow faster operation and low power consumption. The downside is that the burden of performing tasks such as memory addressing, floating-point calculations, and input-output operations falls on the software side
Hence, in considering the above, the performance of architectures and corresponding processors based on RISC architecture such as ARM depends on the capabilities of the programmers and the quality of their software and application. The proper execution of instructions is an important consideration. This depends on the skills or capabilities of the programmer or developer. The quality of the instruction also determines the performance of an ARM processor.
Take note that processors based on RISC architectures are more prone to demonstrating poor performance once fed with instructions or codes written with low quality because most tasks are offloaded to the software. This does not mean that these processors have poorer performance than CISC processors. They are just more sensitive to code quality than their CISC counterparts. It is for this reason that ARM processors require skilled programmers and developers.
4. Possible Threat to Developers and Consumer Options
A valid concern over the expanding adoption of the ARM architecture and the corresponding ARM processors centers on the likelihood of limiting options available to others. This can stem from the fact that companies are more prone to integrating their supply chains and restricting their entire value chains. Remember that Arm Holdings only provides architecture licenses. Companies can design and produce their own processors based on these licenses.
The decision of Apple to ditch Intel Core processors over its custom-made ARM-based Apple M-series system-on-chips has been welcomed with both praise and criticism. This move gives the company numerous competitive advantages that range from reduced dependence on chip suppliers or other third parties and better control over its product roadmap for devices and its entire product ecosystem while strengthening further its semiconductor strategy.
However, as a particular company gains more control over its supply chain and value chain, third parties such as software developers and other stakeholders such as consumers lose their options. Several critics have worried that Apple will have more control over software developers for the macOS operating system. The move will also limit the choices of its consumer pool because of the prevailing compatibility issues between ARM and x86 processors.
It is possible for companies to follow the lead of Apple. The present reality remains far from these assumptions. Companies such as Microsoft and Android manufacturers such as Samsung are still dependent on other chip designers such as Qualcomm because it is more efficient for them to outsource chips rather than developing their in-house design capabilities. Apple is still the sole company with a tighter and more integrated supply chain and value chain.
