1. History of ARM
The development history of ARM can be traced back to 1978, when the company Chris Curry worked for encountered a financial crisis and its development went from bad to worse. Curry decided to leave after in-depth communication with the founder. At that time, Curry was very interested in microcomputers, and then founded Cambridge Processor Unit Co., Ltd. with his friend Hermann Hauser (Hermann Hauser). In 1979, Cambridge Processor Company changed its name to Acorn Computer (Acorn Computer) Co., Ltd. It is said that the name was changed because it hoped that the company's name would be listed in front of Apple in alphabetical order. The company was mainly engaged in the design and manufacture of electronic equipment in its early days. business, its first hugely successful product was launched in December 1981 as the BBC Microcomputer. 1985 was an important milestone for Acorn Computer. They independently completed the design of a 32-bit microprocessor, using a reduced instruction set, a 3-micron process, and containing 25,000 transistors. This processor is the starting point of the ARM architecture, ARMv1.
In 1990, invested by VLSI Technology Corporation, Acorn and Apple each held 43% of the shares, and established ARM. Since then, ARM has become the abbreviation of Advanced RISC Machines. The original ARM membership consisted of a handful of engineers and was based in a barn in Cambridge.
In 1996, ARM established cooperation with Texas Instruments, Samsung, Nokia and other companies to achieve profitability. The classic game in the Nokia 6110 mobile phone - Snake is developed based on the ARM 7TDMI chip. In 1998, ARM was listed on Nasdaq and listed for trading. At that time, the market value of ARM had reached one billion US dollars! In 2004, ARM released three series of processors: Cortex-A, R, and M. Careful readers may have discovered that these three series of processors are the three letters of ARM. In February 2007, ARM's first GPU, the Mali-200, officially entered the market. In the same year, there was another major event in the technology world, that is, the iPhone was born! The iPhone can be described as an epoch-making electronic product, which is equipped with chips based on ARM cores. With the advent of the smart phone era, ARM has also sprung up. In 2007, the shipment of chips based on ARM core designs has reached 100. 100 million!
In 2016, ARM was acquired by Japan's Softbank. In 2020, Softbank planned to sell ARM to Nvidia for US$40 billion. However, due to the opposition of many regulatory agencies, the transaction was aborted. ARM will still remain neutral in the semiconductor industry and continue to serve other companies. Chip design companies provide IP (Intellectual Property) authorization.
2. ARM processor family
Over the years, ARM has developed quite a few different processor products. As shown in the figure below: ARM processor products are divided into classic ARM processor series (left side of the dividing line in the figure below) and the latest Cortex processor series (right side of the dividing line in the figure below). And according to different application scopes, ARM processors can be classified into 3 series.
- Application Processors (Application Processor, the orange box in the figure below) – high-end processors for mobile computing, smartphones, servers and other markets. These processors run at very high clock frequencies (over 1GHz) and support memory management units (MMUs) required by full operating systems like Linux, Android, MS Windows and mobile operating systems. If the product you plan to develop needs to run one of the above operating systems, you need to choose the ARM application processor.
- Real-time Processors (real-time processors, the gray box in the figure below) – a family of high-performance processors for real-time applications, such as hard disk controllers, automotive powertrains, and baseband control for wireless communications. Most real-time processors do not support MMU, but usually have MPU, Cache and other memory functions designed for industrial applications. Real-time processors run at relatively high clock frequencies (eg, 200MHz to >1GHz) with very low response latencies. Although real-time processors cannot run full versions of Linux and Windows operating systems, they support a large number of real-time operating systems (RTOS).
- Microcontroller Processors – Microcontroller processors are usually designed to be small in size and highly energy efficient. Usually these processors have a very short pipeline and a very low maximum clock frequency (although there are such processors on the market that can run above 200Mhz). Also, the new Cortex-M processor family is designed to be very easy to use. Therefore, ARM microcontroller processors are very successful and popular in the microcontroller and deeply embedded system markets.
ARM three series processor characteristics
In addition, ARM also launched a mobile GPU series, the Mali GPU.
3. ARM Authorization
ARM is an intellectual property (IP) supplier. The biggest difference between it and general semiconductor companies is that it does not manufacture chips and does not sell chips to end users. Partners produce chips with their own characteristics.
Broadly speaking, ARM authorization is divided into three levels:
- Use hierarchical authorization
- Kernel-level authorization architecture
- Instruction set level authorization
The three levels of authority are ascending in order. The requirements for chip design companies are also from low to high, and the space for use is also from low to high: the instruction set level authorization (also known as soft core) has the largest space to play, and the use of hierarchical authorization (also known as hard core) has the least space to play . (For example, the foundry TSMC can carry out the construction with the drawings, without modification).
To buy a chip design company that uses ARM's hierarchical authorization (hard core), in layman's terms, it uses the ARM public version architecture. In this case, the chip design company must bring the brand of ARM to the outside world: the brand of the CPU is Cortex-AXX (XX stands for two Arabic numerals, the first number indicates the generation of the architecture, and the second indicates the fine-tuning of the architecture ), the GPU is Mali-GXX (XX means the same as above).
If we use a rough but easy-to-understand example to illustrate the permissions of these three levels, we can generally understand it like this:
Suppose I write an article, I only authorize you to forward it, you can’t change it, you can’t add oil and vinegar, that is to use hierarchical authorization; I authorize you to quote my article in the article, which is kernel-level authorization; I authorize you to get To modify and reorganize my articles to form a new paper is the authorization of the architecture level.
Note that schema-level authorization for a certain version is usually permanent.
At present, more than 1,000 companies have signed technology licensing agreements with ARM, including such large companies as apple, Intel, Broadcom, IBM, LG, SONY, NXP and TI. However, there are only about 15 companies that dare to purchase architecture licenses. Qualcomm, Apple, Broadcom, Marvell, Huawei, and Samsung are all typical users (and they are also leaders in their respective industries).
Four, ARM architecture analysis
The following takes S3C2440 as an example to compare and analyze with 51 single-chip microcomputers, and analyzes the ARM architecture in detail.
ARM Company mainly designs ARM series AISC processor cores, it does not produce chips, but only provides IP cores. First explain the architecture, core, processor and chip with an example: S3C2440, this is a SoC chip. Note that it is not a CPU. S3C2440 is somewhat similar to the 51 single-chip microcomputer we are familiar with. At present, it has gone through three stages, namely SCM, MCU, and SoC. 51 belongs to SCM or MCU, and S3C2440 belongs to SoC. First, let's take a look at the internal structure of 51 single-chip microcomputer, as shown in the figure below.
Its internal structure can be simply divided into two parts: CPU and peripherals. Look at the 2440 again:
The ARM920T in the middle is its processor. In my opinion, processor and core are a concept here, but one is a hard concept and the other is a soft concept. The ARM920T here is both a processor and a core. What Samsung does is other things besides this CPU.
What about the architecture? Let's look at another picture.
The one on the left is the architecture, and the one on the right is the processor, which can also be called the core. ARM's first and most successful CPU was the ARM7TDMI, which was based on ARMv4. The ARM architecture incorporates the following RISC features:
- read/store schema
- Does not support misaligned memory accesses (now supported by ARMv6 cores)
- Orthogonal instruction set (arbitrary access instructions can access data in any addressing mode Orthogonal instruction set)
- Large array of 16 × 32-bit registers (register file)
- The fixed 32 bits operation code (opcode) length reduces the cost of encoding and reduces the burden of decoding and pipeline.
- Most are executed in one CPU cycle.
- The architecture of different versions will be adjusted.
Like Samsung, other major manufacturers that cooperate with ARM usually put their CPU and various peripheral IPs together, and then take the drawings to tape out, and the products produced are also a square with many pins underneath. This thing includes not only the CPU, but also other controllers. This thing is called SoC (system on chip). From an English point of view, the so-called quad-core SoC does not mean only the CPU, but a quad-core system.
So what major manufacturers are currently doing is to buy ARM authorization, get the source code of the ARM processor, and then make some peripheral IP (or buy or design it yourself), form a SoC, and then go to the tape. Different SoCs have different architectures (that is, how the CPU is connected to IP, some with bus as the core, and some with DDR as the core), so HiSilicon is an SoC architecture with independent property rights. However, no matter what manufacturers do, no matter how hard they try, the CPU has not been touched. The ARM core just stays there, that is the central processing unit.
V. Summary
In the post-smartphone era, ARM has developed rapidly and established a good ecosystem in more than ten years. Even Microsoft, an old partner of Intel, announced its support for the ARM architecture in 2011. As of 2022, ARM has more than 1,000 partners, and 95% of smartphones use ARM core-based processors, with cumulative shipments of more than 225 billion processors. You know, if around 2000, if someone said that the performance of ARM architecture processors could be comparable to x86, or even surpass x86, then it would be tantamount to fantasy. Then, with the development of science and technology at the end of the day, ARM rides on the big ship of the mobile Internet and gradually develops into a new force in the field of instruction set architecture!