Search

M1 – Rise of the SoC

The next generation of Apple Mac was announced at Apple’s ‘One More Thing’ Event on November 10, 2020. While it is not unusual for the annual announcement of the latest system release news, this particular generation is more than the typical incremental upgrade we are used to in traditional technology events. It marks an historic milestone change. Not only is Apple boasting some of its greatest generational performance improvements such as a 3.5x faster processor core, 3.9x faster video, 7.1x faster image processing and more, it is signaling a fundamental architecture shift that has been brewing for several years now but has yet to gain significant traction, that is possibly until now. The aim of this article is neither to advocate for Apple, nor fact check its improvement boasts, but to address a market dynamic that Apple is disrupting with the M1 processor at the heart of its new product release.

That fundamental change is the processor architecture divide that separates computers from embedded electronic devices. Traditionally, if you needed to run complex applications, control a network of devices, and support for an operating system designed to put the user in charge of what applications were installed or uninstalled and all their complex configurations and interactions with resources, you needed the performance of a computer board filled with discrete integrated circuits (IC) designed around the central processing unit (CPU). This CPU has traditionally been a microprocessor (MPU). Apple is going all-in, implying that no longer has to be the case. This year’s announced refresh of three Mac computer lines replaces the traditional Intel MPU with a system-on-chip (SoC) designed by Apple, and Apple claims that the future of Macs will be based on this new strategy.

An SoC class of processor is the heart of almost every media tablet and smartphone and a host of other electronics. Up until today all attempts to dislodge the microprocessor from the computer have run into major barriers. Apple appears to be uniquely suited to address these barriers and if Apple proves it can be done, you can bet there will be others following suit.

This has been a very complex evolution that has been tracked by technology specialists at Yole Développement. This is a very important market move that will reverberate through the semiconductor industry in many ways.

First let us look at the fundamental differences in the architecture and why this is important. PCs, including prior generations of Macs, have traditionally relied on the highly computer-optimized architecture of an MPU. MPUs are discrete processors that utilize multiple ICs such as input/output chipsets, memory, graphics, security, connectivity, and other discrete ICs to form a complete system. The MPUs optimization of complex processor instructions supported by likewise optimized discrete chipsets supporting the MPU has driven MPUs to be the highest performing processor on benchmarks targeting computer applications. The system-on-chip, as the name implies, reduces some or all of these discrete ICs down to subsystems within the processor. This can reduce overall system flexibility to a small degree and may not yet achieve the bleeding edge performance of the latest generation of high-end MPU. However, improvements on SoC design have been the reason many of us now use our smartphones and tablets for the same applications for which we used to rely on a computer. Many high-end SoCs, including the M1, have begun to increasingly overlap the low-end performance characteristics of the MPU market. This latest M1 SoC boasts impressive improvements on energy consumption, heat, area, cost, and design control over the prior MPU centric systems by combining integration of subsystems built on 16 billion transistors manufactured in a cutting edge 5 nm process. The division of MPU and SoC based on performance are now overlapping significantly more with each generation and as demonstrated in the latest Macs, are capable of mainstream compute performance.

MPUs for PCs are supplied primarily by Intel and AMD. Apple gets its boasted performance improvements by comparing the M1’s 8-core Apple optimized ARM Cortex A series application processor with integrated 8-core GPU to the prior 1.2 GHz quad core Intel i7 series processor in the prior MacBook Air with both systems using 16GB RAM and 2TB SSD drives. If the actual performance and benchmarks live up to Apple’s announcement, Apple’s choice to switch to its M1 SoC wasn’t just ‘catching up’ with the next generation MPU, it represents a phenomenal jump compared to incremental percentage improvements of prior MPU generations. It will also completely destroy the myth that an SoC could never compete with an MPU for performance in a PC. In addition to core application, video, and image performance improvements, the M1 addresses market trends that have become extremely important in the last several years. It integrates a 16-core Neural Engine capable of 11 trillion operations per second to address the increasing use of artificial intelligence and security integrated on-chip to address the ever-increasing cybersecurity threat. As icing on the cake, the energy improvements of the SoC enable superior battery life and even a quiet fanless design in the case of the MacBook Air.

Setting aside the architectural innovation and performance claims, the switch to Apple silicon allows Apple to capture more of the value in the Mac’s bill-of-materials (BoM).  Under the previous design, Intel captures the price of the MPU, and depending on the model, AMD captures the price of the GPU.  Because of the synergy in engineering SoCs for other Apple devices, Apple reduces the BoM bringing those functions in-house built into the M1 SoC. Integrating on 5 nm and the SOC’s packaging solution will not be cheap, but it means Apple is no longer paying Intel or AMD’s margin.  Once the full mac family is running on Apple silicon, there will be a shift of up to $1 billion in revenue from the traditional MPU and GPU vendors towards Apple’s foundry and packaging partners.

So far, this may sound like Apple M1 or SoC evangelizing or it may look like a no-brainer switch to SoCs for generations to come. However, performance is not the only, nor even the primary concern when it comes to switching processor architecture. When Apple switched from PowerPC architecture to Intel x86 in 2006, it was a risk. Apple was not just changing hardware; it was changing the ecosystem of software that was based on that hardware. That change was a risk then, and it is even a bigger risk now. Back then, the switch was from a shrinking market of processor technology toward a PC industry standard. Even given that, the switch was not without a painstaking redesign of the OS and working with software developers to eliminate bugs and improve compatibility. Some poorly updated applications never made the transition leaving some Mac fans with tough choices when thinking of upgrading to an Intel-based Mac. However, Apple’s move to Intel definitely paid off in the long run. This time, Apple is bucking the industry standard x86 architecture to leverage its leading development in smartphone and tablet SoCs based on ARM Cortex A intellectual property which Apple customizes for its iPhone, iPad, Apple TV, Apple Watch and now Mac lines.

Apple is leveraging the enormity of its iOS ecosystem to overcome the pains they and their customers may experience dealing with changing the ecosystem of software again. Apple has optimized its new OS, Big Sur, to be uniquely integrated with the benefits of this ecosystem change. While other attempts at shifting the software ecosystem from x86 to ARM (i.e. Windows RT) have met with limited success, Apple’s advantage is the weight it has behind its Mac and iOS ecosystem. Apple can incentivize its software partners developing iOS solutions to development and integrate software running on iOS and Big Sur. Apple is banking on Universal Apps developed to run seamlessly between x86 and M1 processors, Rosetta 2 to enable emulation of x86 processing for applications that are slow or incapable of transitioning, and compatibility with iOS Apps to smooth the transition. It will not be perfect, but neither was switching to Intel.

Again, the point of this article has less to do with whether you should rush to buy a new Mac or not. It is about the significant reshaping of the landscape for processor and manufacturing competition that is associated with Apple’s success or failure. Apple is just leading the vanguard giving rise to the SoC over MPUs in the PC world. The overlapping performance between current generations of PC MPU and SoC are related to the growing competition between Intel, TSMC and Samsung in manufacturing technologies. The fact that the M1 is being manufactured on a 5nm process at TSMC could have much to do with Intel falling behind in cutting edge process nodes. Apple really needed the increased transistor density and energy efficiency provided by 5 nm to provide a big leap in performance to overcome the significant reluctance or pains associated with disrupting the software ecosystem.

The dominance and control of the Apple ecosystem is something that other SoC enabled systems lacked placing Apple in a place for many other SoC suppliers to watch with great anticipation to see if Apple will truly drive acceptance of SoCs in PCs. If successful, it stands to have a significant impact on the processor landscape. If they fail to execute, no matter what the reason, it is more likely that we will see x86 processors becoming more like SoCs before the next Arm-based processor supplier can afford to take another swing at it.

Authors

Tom Hackenberg is a Principal Analyst for Computing and Software in the Semiconductor, Memory and Computing Division at Yole Développement (Yole). Tom is engaged in developing processor market monitors and research into related technology trends. He is currently focused on low and ultralow power solutions such as MCUs. Tom is an industry leading expert with more than a decade’s experience reporting on markets for semiconductor processors including CPUs, GPUs, MPUs, MCUs, SoC ASICs & ASSPs, FPGAs and configurable processors. Tom is also well-versed in related technology trends including IoT, heterogeneous processing, chiplets, AI and edge computing.
Prior to joining Yole, Tom was a principal analyst at OMDIA, IHS Markit and began processor market research in 2006 for IMS Research. He worked with market-leading processor suppliers developing both syndicated and custom research. Tom holds a BSECE from the University of Texas at Austin specializing in Processors and FPGAs.

John Lorenz is a Technology and Market Analyst within the Computing & Software division at Yole Développement (Yole), part of Yole Group of Companies. John is engaged in the development of market and technology monitors for the logic segment of advanced semiconductors, with an initial focus on processors. Prior to joining Yole, John held various technical and strategic roles at Micron Technology.
On the engineering side, his roles included thin film process development and manufacturing integration on DRAM, NAND, and emerging memory technologies and industrial engineering / factory physics for the R&D fab.
On the strategic side, John ran the memory industry supply & capex model for corporate strategy / market intelligence, and established the industry front-end costing model within strategic finance.
John has a Bachelor of Science degree in Mechanical Engineering from the University of Illinois Urbana-Champaign (USA), with a focus on MEMS devices.

Related monitor

Application Processor – Quarterly Market Monitor BD

Application Processor Quarterly Market Monitor
With SMIC, China is positionning as a competitor in foundry business and thereby catching the dynamic of the APU market.


up