The history of Apple silicon. Intel's main mistake

Apple has always been famous for its ability to easily change platforms. The Apple II computers, which appeared in 1977, were built on chips from the now forgotten MOS Technology, and the very first Macintosh, introduced in 1984, ran on Motorola processors. In 1994, there was a grand transition to PowerPC RISC processors developed by the AIM alliance (short for Apple, IBM and Motorola), which was specially created for this purpose three years earlier.

The creators of the alliance believed that processors based on the CISC architecture from Intel and AMD were nothing more than a dead-end branch of development, and the future belonged to RISC. It is very likely that in the long run they were right (don’t forget that the ARM architecture is based on RISC principles), but in 2006 Apple was forced to switch to solutions from Intel.

The main reason for the transition was the desire for portability. The Power G5 processors were great for high-end workstations, but they were completely unsuitable for compact laptops. The very first solutions with Intel were presented in 2006. In this regard, Apple even changed the names of some lines of computers - for example, Power Mac became Mac Pro, PowerBook became MacBook Pro, and iBook became simply called MacBook.

For some time there was talk on the Internet that, before switching to x86, Apple was choosing between Intel and AMD processors. But what actually happened there and whether negotiations were really held with AMD is unknown. Subsequently, rumors repeatedly surfaced on the Internet and suggestions were made that Apple was planning to replace the “blues” with “reds.”

For some time it seemed that no more changes were in sight and Apple would settle on Intel’s creations for many years. However, in 2007, a significant event occurred - the first iPhone was presented to the world. To develop this device, Apple required a separate chip, which it naturally turned to Intel. However, the monopolist in the world of processors considered that this area was not particularly promising and did not deserve to waste time on it. As a result, Apple went to Samsung, which turned out to be much more accommodating. Many experts believe that refusing to develop a mobile processor is Intel's main mistake.

Samsung's creation proved to be quite worthy. But Apple did not want to depend on a third-party developer and was intensely looking for a way to start creating its own processors that could be better “tailored” to its needs. As a result, in 2008, the company PASemi, which was engaged in the creation of high-flow Rate mobile processors, was absorbed. This developer became the first of tech absorbed by Apple on the path to complete unification.

There has already been an episode in the history of Apple when the company used ARM processors - Newton PDAs were built on their basis. These laptops were produced between 1993 and 1998, but were not particularly successful commercially. The main reasons for the failure were then cited as the high cost and rather impressive dimensions of the devices, which could not be put in a pocket. Or it is possible that the market itself was simply not yet ready for the appearance of such “toys,” even very interesting and functional ones.

On January 27, 2010, the first iPad was introduced to the world. The event is remarkable in itself, so not everyone paid attention to the fact that the tablet computer was based on the Apple A4 ARM processor, the company’s first chip. It had a frequency of 1 GHz, was built on a 45 nm process technology and consumed about 500-800 mW. It was produced by the same Samsung. Subsequently, the iPhone 4, introduced in the same year, was equipped with the same chip.

The appearance of the first Apple chips gave rise to speculation that in the future the company may completely switch to processors of its own design. And although initially this idea seemed very “far-fetched” to many, gradually more and more analysts and journalists writing on the topic of technology were inclined to think that this was exactly how everything would be. Statements were periodically published on the Internet according to which the first laptop with a chip from Apple would be presented “in the very near future.” Such ideas were sometimes supported by rumors that clearly did not appear out of nowhere. For example, in 2011, the popular resource MacRumors wrote that Apple was testing a MacBook Air built on the A5 processor. It is very likely that the company actually conducted similar experiments.

In 2013, the iPhone 5s with the Apple A7 processor was presented. And it was not only the world's first 64-bit mass-produced ARM chip, but also a solution whose capabilities came very close to processors for desktop computers. Of course, Apple's announcement about the "Desktop-Class Processor" was nothing more than an advertising stunt, but it showed exactly where the company was heading.

In its best traditions, Apple was in no hurry; the process of complete transition to its own platform occurred gradually. In 2015, S-series chips appeared for Apple Watch, and in 2019, H-series chips for headphones appeared. By that time, few people doubted that it would come to computers, the only question was time. And it all happened in 2020, when the first solutions with the Apple M1 processor were presented to consumers. Initially used only in the Mac mini and MacBook Air, these chips later also came to the MacBook Pro, iMac, and began to be used in the iPad. M1 processors have been enthusiastically received by users. Many praised the flow Rate and energy efficiency of the new chips and easily forgave them for the inevitable “childhood illnesses.”

A complete abandonment of Intel did not occur immediately; for some time, models of both types were present in the company’s portfolio, and the operating systems had support for two architectures. But at the beginning of 2024, Apple had no computers left that used Intel processors. It should be noted here that the latest Mac Pro models look somewhat controversial (we’ll look at why exactly below), but they were clearly created specifically for the purpose of the final transition to their own platform.

In response to Apple's actions, Intel couldn't come up with anything better than to release a parody advertisement that showed the advantages of laptops built on its solutions over Apple's creations. Actor Justin Long, who once played the character Mac in the famous “Get a Mac” commercials that ran from 2006 to 2009, was even invited to create the video. However, such a move was met with mostly ridicule. Not only did the advertisement look more like a childish insult to a reputational blow, but its quality left much to be desired - in some frames the device is visible through the user’s translucent fingers. As a result, a lot of ridicule appeared on the Internet about the fact that Intel should pay more attention not to trolling Apple, but to the capabilities of its chips, which have recently begun to openly fail.

The same advertisement from Intel. At 57 seconds the user's finger becomes translucent:

Popular Apple laptops

Difference between RISC and CISC processors

When talking about Apple silicone processors, it would not be out of place to mention the difference between them and x86 family solutions. Today, there are two main types of processor architectures: RISC (Reduced Instruction Set Computing) and CISC (Complex Instruction Set Computing). These architectures have different approaches to instruction processing, which affects their flow Rate, energy efficiency, and design complexity.

CISC Basic Principles

The CISC architecture was developed by IBM when creating the IBM/360 family of computers in order to simplify the work of programmers who worked with machine codes.

Main characteristics of CISC processors:

  • Complex instruction set: CISC uses a large set of instructions, each of which can perform complex operations. This allows you to reduce the number of commands needed to complete a task.
  • Multiple Addressing Modes: CISC processors support multiple ways to access data in memory, which simplifies assembly language programming and can reduce program dimensions.
  • Multi-Cycle Instructions: Many instructions in CISC processors execute over multiple cycles, which can reduce flow Rate but allows more complex operations to be performed with a single instruction.
  • Hardware Focus: More complex instructions require more complex logic in the processor, increasing its dimensions and power consumption but making compilers and programming easier.

An example of CISC is the x86 family of processors used in most personal computers.

Basic principles of RISC

The development of CISC led to excessive complexity of the instruction set and the processors themselves. The result was the RISC architecture, which was designed to simplify instructions and increase execution speed.

Main characteristics of RISC processors:

  • Simplified Instruction Set: RISC processors use a small, fixed set of simple instructions, each of which executes in one cycle.
  • Register architecture: RISC processors use a large number of registers, which allows data to be stored and reduces the need to access slow RAM.
  • Uniform instruction execution cycle: Each instruction is executed in one cycle, which simplifies pipelining and increases the overall speed of program execution.
  • Focus on software: Simplifying the hardware requires a more sophisticated compiler that optimizes the code to efficiently use processor instructions.

Examples of RISC are ARM chips used in most mobile devices and MIPS used in various embedded systems.

Comparison and current trends

Historically, RISC and CISC were two different approaches to processor design, but over time the differences between them began to blur.

  • Performance and Energy Efficiency: RISC processors typically offer higher flow Rate at lower power consumption due to a simplified instruction set and uniform execution cycles. This makes them excellent solutions for mobile and embedded systems.
  • Complexity and Cost: CISC processors can perform more complex tasks with fewer instructions. However, they are more complex and expensive to manufacture due to more complex logic and more transistors.
  • Hybrid approaches: Modern processors often use a hybrid approach, combining RISC and CISC elements to achieve the optimal balance between flow Rate, power efficiency and flexibility.

Modern chips increasingly incorporate elements of both approaches to achieve the best combination of flow Rate and energy efficiency to meet today's computing requirements. To improve flow Rate, many CISC processors use RISC-specific techniques. On the other hand, modern RISC processors can support some complex instructions to improve the efficiency of certain tasks.

Popular Apple personal computers

General description and advantages of Apple M

M-series processors are not a classic CPU, but are a system on a chip (SoC) that integrates many components on a single chip to achieve high flow Rate and increased energy efficiency. Let's look at the key components and operating principles of these solutions.

Processor architecture

Built on ARM architecture, Apple M chips include the following main components:

  • Central Processing Unit (CPU): Consists of multiple high-flow Rate Firestorm cores for resource-intensive tasks (code compilation, video processing) and multiple energy-efficient Icestorm cores that take on simpler tasks (web browsing, text processing) and reduce power consumption. The number of cores depends on the processor model.
  • Graphics Processing Unit (GPU): Built-in graphics processor with multiple cores delivers powerful flow Rate for graphics, gaming, and professional applications. Since the GPU is integrated into the SoC, latency is reduced and overall efficiency is increased.
  • Neural Engine (NPU, marketing name - ANE): The neural engine is responsible for accelerating tasks related to machine learning and artificial intelligence - image processing, speech recognition and other similar actions.
  • Unified Memory Architecture (UMA): All components, including the CPU, GPU and Neural Engine, have access to a common pool of high-speed RAM. This allows data to be transferred between components faster and reduces latency.
  • Controllers and peripherals: Various controllers are integrated, such as Thunderbolt and USB 4, components for processing audio, video, etc. are included.
With UMA, components access data directly from a single, shared, fast memory and can communicate more efficiently.

Of course, Apple is not the first company whose computer processors have graphics cores on the same chip as the CPU. But built-in GPUs have never been very performant because they lack dedicated video memory and use slower RAM. In addition, developers avoided using powerful cores due to the risk of overheating. In the case of Apple silicon, the graphics processor has high-flow Rate shared memory, plus the company managed to solve the issue of excessive heat generation.

Operating principles and hard optimization

The core principles of M-series processors are to maximize flow Rate and energy efficiency. Here's how this is achieved:

  • Pipelining: Instructions are processed in multiple stages that run in parallel, increasing overall CPU flow Rate. Uses complex pipelining to improve command execution speed.
  • Parallelism: A high degree of parallelism is achieved through the use of multiple CPU and GPU cores, which allows you to perform many tasks simultaneously without significant flow Rate losses.
  • Energy Efficiency: ARM architecture and thoughtful distribution of tasks between high-flow Rate and energy-efficient cores have significantly reduced power consumption without sacrificing flow Rate. For example, the 2020 Mac mini with M1 consumes 39 W at maximum load, which is noticeably lower than the 122 W of the previous 2018 Mac mini with Intel i7.
  • Component Integration: Integrating all key components on a single chip reduces data transfer latency and improves overall system efficiency. This allows you to complete tasks faster and with less energy than traditional multi-component systems.
  • Software optimization: Apple tightly integrates your hardware with the macOS operating system to optimize your apps and overall system experience. This includes support for specialized instructions and optimizations to take full advantage of the processor's capabilities.

In other words, the advantage of Apple silicon lies not in one “magic ingredient”, but in a competent combination of a number of interesting approaches. ARM architecture, unified memory, high-flow Rate and energy-efficient cores, a powerful graphics processor, a specialized neural engine and strict software optimization made it possible to obtain a system that demonstrates high flow Rate with excellent energy efficiency indicators.

But in fairness, one point should be noted - Apple does not create universal processors that could be used on heterogeneous hardware. Rigorous optimization and a small number of devices allowed the company to obtain an excellent solution with relatively little cost. In addition, Apple was essentially starting from scratch, while Intel and AMD have to carry with them a considerable baggage of outdated approaches and CISC instructions, which they cannot abandon due to compatibility issues.

Popular Apple tablets

Disadvantages of Apple Silicon

Apple M processors have significantly improved the flow Rate and energy efficiency of Apple devices. However, despite many undoubted advantages, it is not without its disadvantages.

From the very first days, the main difficulty was the need to “adjust” all existing software to new realities. Several years have passed since the first processors appeared, compatibility issues still remain, and the choice of software has some limitations. There is also no native Linux support. And although the Asahi project appeared back in 2021, designed to create Linux for Apple silicon, there is still a lot of work to be done in this direction, even if some success has already been achieved.

There are also problems with the lack of support for discrete and external graphics processing units (eGPUs). This limits the capabilities of users who rely on additional graphics power for tasks such as video editing, 3D rendering and gaming. In an interview, Apple's senior vice president of hardware engineering, John Ternus, said that the new processors lack support for discrete GPUs due to the fact that there is not yet a good way to use them in tandem with an integrated GPU, but because this with a unified memory architecture is an example of how strength can bring challenges as well as benefits.

Against this background, the newest Mac Pro looks somewhat strange, which has expansion slots, but additional video cards cannot be installed in them; in fact, it all comes down to network cards and drives. There was even an assumption that no more than 1% of all Apple users needed such a system and the company released it more in order to finally get rid of Intel and complete the transition, which lasted for 3 years instead of the expected 2 years. Most people who need a high-flow Rate computer will be fine with Mac Studio - the capabilities of this system are truly impressive.

The introduction of Apple silicon led to a brief but significant increase in sales of Intel-based Macs, driven by users needing specialized software or hardware.

With the transition to a new platform, difficulties also arose with some other peripherals, such as audio devices. But there everything was usually solved by creating new drivers or optimizing software. In the case of graphics, the problem lies much deeper, and Ternus himself does not yet see a way how it could be effectively resolved. Maybe this will happen later. Or the capabilities of Apple silicon will increase so much that the need for additional funds will disappear.

It is impossible not to mention such an aspect as the lack of an upgrade. Apple is known for its rigid approach to component integration, which often results in the inability to update or replace individual parts of the device. In the case of M-series processors, this means that users, by definition, cannot increase the amount of RAM integrated into the chip itself. This limits the ability to customize and upgrade devices after purchase. The difference in the cost of models with different characteristics is quite noticeable.

Popular Apple smartphones

What does the Apple silicon family include?

Today, Apple silicon already includes a whole scattering of a wide variety of solutions:

  • Apple A – chips used in iPhone, younger versions of iPad and Apple TV;
  • Apple H – a series specifically designed for audio processing, used in AirPods headphones;
  • Apple M – solutions for laptops, desktops and top-end tablet models;
  • Apple R – chips specially designed for Apple Vision Pro, responsible for working with image output and processing data from various headset sensors;
  • Apple S – chips that form the basis of the Apple Watch;
  • Apple U and W are designed to improve the functionality and user experience used in various Apple devices.

Once upon a time, there were also T-series chips used in laptops and desktops, responsible for protection and processing of biometric information. Subsequently, such functionality was integrated into M chips.

At the beginning of 2024, the A series alone already included 21 models, from A4 (iPhone 4, iPad) to A17 Pro (iPhone 15 Pro). The path from 45 nm to 3 nm has been passed, processors with a technological process of 2 nm and below are on the way (Taiwanese chip manufacturer TSMC promises to provide such solutions as early as 2025). There is no point in describing each of the presented chips separately; this would require a whole encyclopedia, but within the framework of the article it would be absolutely useful to talk in more detail about some of the differences in the M series.

Apple M processors are used in the MacBook Air and MacBook Pro laptops, the Mac mini, iMac and Mac Studio and Mac Pro desktops, the iPad Air and iPad Pro tablets, and the Vision Pro headset.

The full list of processors (current as of mid-2024) looks like this:

  • M1 – 8-core CPU (4+4), 7 or 8-core GPU, 16-core NPU, up to 16 GB RAM;
  • M1 Pro – 8 or 10-core CPU (6+2 or 8+2), 14 or 16-core GPU, 16-core NPU, up to 32 GB RAM;
  • M1 Max – 10-core CPU (8+2), 24 or 32-core GPU, 16-core NPU, up to 64 GB RAM;
  • M1 Ultra – 20-core CPU (16+4), 48 or 64-core GPU, 32-core NPU, up to 128 GB RAM.
  • M2 – 8-core CPU (4+4), 10-core GPU, 16-core NPU, up to 24 GB RAM;
  • M2 Pro – 10 or 12-core CPU (6+4 or 8+4), 16 or 19-core GPU, 16-core NPU, up to 32 GB RAM;
  • M2 Max – 12-core CPU (8+4), 30 or 38-core GPU, 16-core NPU, up to 96 GB RAM;
  • M2 Ultra – 24-core CPU (16+8), 60 or 76-core GPU, 32-core NPU, up to 192 GB RAM.
  • M3 – 8-core CPU (4+4), 8 or 10-core GPU, 16-core NPU, up to 24 GB RAM;
  • M3 Pro – 11 or 12-core CPU (5+6 or 6+6), 14 or 18-core GPU, 16-core NPU, up to 36 GB RAM;
  • M3 Max – 14 or 16-core CPU (10+4 or 12+4), 30 or 40-core GPU, 16-core NPU, up to 128 GB RAM.
  • M4 – 9 or 10-core CPU (3+6 or 4+6), 10-core GPU, 16-core NPU, up to 16 GB RAM.

The differences between Max and Pro processors are the number of cores, graphics flow Rate, memory bandwidth, and intended use. Max version chips are designed for users who require maximum capabilities, while Pro chips are aimed at medium-complexity tasks. As for the Ultra, it is technically two Max dies combined using UltraFusion technology, resulting in one very powerful chip.

At the time of writing, the M3 Ultra has not yet been introduced. The reason for this is not entirely clear. There are speculations that the company is working on a new approach in which the Ultra series is no longer simply a combination of two processors. There is also an opinion that Apple is not particularly satisfied with the M3 chips, which did not provide much of an increase in flow Rate despite the increased complexity of production. There is a high chance that the M3 Ultra will never be presented, and the company will completely focus on the M4.

M4 processors were introduced on May 7, 2024 - for the first time in the history of the M series, the presentation of the chip took place together with the iPad Pro, and not with a personal computer. At the time of writing, there was no information about the more powerful Pro and Max, and the presented version is built with a larger number of Icestorm cores, that is, it is clearly more tailored for mobile devices. By “stepping over” the M3, the company has provided the iPad Pro with significant future-proofing. The previous model of the top tablet was based on the M2, and at the same time, there were virtually no applications in the App Store that could exhaust the capabilities of this processor.

The Mac mini, iMac, and MacBook Pro are expected to start receiving M4 chips in late 2024 and early 2025. The MacBook Air will receive the M4 update in spring 2025, and the Mac Studio will receive the update in mid-2025. As for the Mac Pro, this system will be updated towards the end of 2025.

It should be noted that a higher number in the version name does not always mean an absolute advantage over all previous solutions. For example, M2 processors outperform M1, but are inferior to M1 Pro, not to mention Max or Ultra. This is also true for older versions.

Popular Apple smart watches

The future of Apple silicon

The future of Apple silicon processors looks promising. Apple will continue to improve the flow Rate and energy efficiency of its chips, introducing improvements in architecture and clock speeds with each new generation, such as the M3 and M4. Future processors will include more powerful neural engines, enhancing artificial intelligence and machine learning capabilities.

The company will also strive for an even more integrated ecosystem, optimizing interactions between iPhone, iPad, Mac and Watch for a smoother, more seamless experience. A significant improvement in graphics flow Rate is expected, which is important for video editing and 3D graphics professionals, as well as gamers - Apple is working hard to rid the Mac of its "Not good for gaming at all" reputation.

It is also a likely scenario that Apple will eventually abandon the division into the A and M series, and all the company’s mobile devices and computers will run on the same chips, even with different characteristics.

Popular Apple headphones

What awaits x86 architecture in the foreseeable future

The emergence of Apple silicon and the subsequent success of this platform naturally led to a number of serious changes in the computer world. Companies such as Qualcomm, AMD and Nvidia have actively begun working on ARM chips for laptops and desktops. It can be assumed that already in 2025 a number of systems based on “stones” from these manufacturers will appear on sale.

At the beginning of 2021, Qualcomm acquired the startup Nuvia, which was working on a new processor codenamed Phoenix. Despite the fact that a ready-made solution was never presented, many interesting developments were created, which attracted the interest of the mobile chip manufacturer. The interesting thing here is that the founders of Nuvia were three former Apple engineers who were directly involved in the creation of silicon chips. They also have experience working for companies such as AMD, ARM and Broadcom.

It is not surprising that, against the backdrop of all of the above, there are already speculations about the imminent death of x86. But is this really true? Most likely not. The x86 architecture will remain an important part of the computing world in the coming years, although it will no longer have the same impact on the market as a whole as it did just a few years ago. For example, it will retain its leadership in cases where extensive customization options are very important - gaming systems, professional workstations and servers. We also do not forget about compatibility with existing software, which is sometimes highly specialized and very expensive.

However, x86 will have to adapt to increased competition from ARM and other alternative architectures. Its future will be determined by Intel and AMD's ability to continue to innovate, improve energy efficiency, develop custom solutions, and maintain compatibility with the existing software ecosystem. The processor market today is going through very serious changes and it is too early to draw any clear conclusions. But the main thing is that the Apple silicon platform has begun processes that will ultimately benefit all users, including tech who do not even plan to use a Mac.