This article from http://www.cnbeta.com/articles/224544.htm
ARM, ARM, arm, yes arm seems to be a fire overnight, tablet, cell phone and other fields everywhere its shadow, and even already some people predict that in the future there will be a considerable part of the traditional x86 sphere of influence of the desktop share is also occupied by ARM. In this case, we will inevitably compare arm with the traditional x86 processor, strong arm, but also support x86. In fact, ARM did not suddenly arise, it almost and x86 at the same time rise at the end of the last century.
At present, Armcisc and RISC are two basic instruction set structures of modern microprocessors. From the technical and historical point of view, the birth and development of CISC and RISC is not a death-to-death relationship, RISC was proposed before the traditional instruction set system called CISC. The differences in the structural system make the two separate in the development road, and gradually go away. In this long development process, RISC has also tried, and strive to enter the field of CISC;
CISC also struggled, hoping to get a slice of the world of RISC.
ARM is a RISC-based product, and the spokesperson for the PC is x86, based on CISC. ARM, x86 dispute, in fact, is RISC and CISC dispute. RISC and CISC after 30 years of entanglement, and once again a frontal collision together. If the previous RISC and CISC collisions are only part of the area of the game, then now arm and x86 can bring a technological revolution, a global war. Today, we will look at the history of RISC and CISC, and those little-known stories.
Today, the industry generally believes that the improvement of PC performance, especially the performance of the CPU, power comes from the continuous progress of transistor manufacturing technology. Only the number of transistors and the higher the frequency of operation in order to complete more tasks in a unit of time, which is the last century before the 70 's PC development mainstream thinking. At that time, the computer was very slow, especially the storage speed is very slow, the widespread use of slow-speed tape storage devices and the lack of large-capacity memory, so that the computer for every byte of space applications are cherished. In such cases, people tend to do more work in one instruction, such as "write memory after reading the data from memory and registers." This is actually four instructions, the first is to read the data from memory, followed by reading from the register, the third is the sum, the last is to write memory. One instruction can accomplish four tasks, which is the mainstream design of the computer at that time.
In the 70 's, IBM and other business practitioners found that the current PC development direction there are some problems. The existing instruction set system will become more complex if it continues to evolve in the direction of the current instruction set. The prevalence of the same-time compilers has changed this situation: on the one hand, the instruction set is becoming more and more complex, while the compiler seldom uses so many complex instruction sets. And with so many complex instructions, it is difficult for the CPU to optimize every single instruction, and even some of the complex instructions themselves spend more time. The summary of this matter is the later famous "8020" law, that is, in all the instruction set, only 20% most commonly used, 80% is basically rare.
Time entered the 1980 's, reduced instruction set Computing, also known as RISC compact instruction set, began to appear. The advantage of this instruction set is that the most commonly used 20% instruction sets in the computer are centrally optimized, while the remaining 80% that are not commonly used are split into common instruction sets and so on. After RISC was presented, it was given a formal name for the traditional instruction set: Complex instruction set Computing, or CISC complex instruction set. For a time, set off on RISC and cisc who really better argue.
Intel's choice in the instruction set debate is not over, Intel in IBM's disdain for the production of IBM for the civilian PC CPU business. In fact, IBM is not unable to produce its own CPU, but IBM feel that the personal PC is far from the large host of high profits, the power of their own architecture used in a compatible machine is really "kill chicken with sledgehammer." Intel then continued to develop the--C4004 processor on the basis of the previously developed C4004 processor and not the product that Intel voluntarily developed, but from the order requirements of a Japanese manufacturer called Busicom. This is the order demand, for the future development of the x86 laid the foundation.
The development of compatible PCs has brought the global information revolution tide.
Soon Intel produced 8086 processors and relied on the rapid development of PCs to become popular. This time period happens to be RISC began to emerge, CISC was despised by the times. After RISC is put forward, the industry has made almost one-sided praise for the future development of RISC. From the beginning of teaching, the textbooks and teaching modes of the computer principle and system structure of American universities all come from the theory of RISC inventor Hennessy, and the content of this paper is the Hennessy-based MIPS architecture developed by RISC. At that time, the research and Development of the IEEE and ACM published the content of RISC also basically put forward a variety of praise, x86 and CISC is considered to be no future things.
Intel was a small company at the time, with only a few products, and the CPU business was just beginning. Facing the entire industry one-sided to RISC status, Intel either one way to go black continues to be compatible with its own 8086 determined to do the x86, or to abandon the seemingly hopeless cisc into the embrace of RISC. In fact, when the PC was just starting, the market space and the prospect is huge, if the PC market is abandoned, the future development of Intel may bring great uncertainty. Moreover, at that time, PC users have a lot of, coupled with market inertia, Intel can at least adhere to the development of two generations of products to meet the needs of this part of PC users. In this case, Intel decided to continue to develop the x86, so products such as 80286 and 80386 are released sequentially. As we all know later, Intel relies on the PC market to make a lot of money and firm its determination to continue exerting its momentum in the x86 market.
High-performance RISC did not enter the general PC market PC market submissively, and even the patent is not registered, is the biggest failure of IBM century. Looking at his random choice of Intel grew into a rival industry giants, even the "spare tire" AMD (AMD was IBM's choice under the antitrust regulations of the second supplier of x86 processors) has grown, IBM regrets incessantly. But it doesn't matter, RISC is still there, and IBM can make a difference in the high-performance market.
In fact, in the late 80, the advent of a large number of new RISC-based instruction sets and products has allowed people to see the power of the streamlined instruction set. SGI workstations are based on MIPS, and the IBM power series is not to mention the preferred product for supercomputer, and the DEC Alpha Architecture processor, which is the representative of RISC. RISC's powerful charm on high-performance computers makes users and the industry fascinated.
In this case, Intel could not sit. As the saying goes, eating in the bowl, looking at the pot. Intel Bowl inside eat x86 this big fat, the pot of RISC although is small fat cow, but also really good. As a result, Intel secretly developed RISC-based processor 80860, hoping to enter the general computer market. But the question is, who uses it? The PC is already a x86 architecture, and RISC is not compatible with operating systems and software, building ecosystems and software circles from the ground up, when Intel was hard to succeed. But Intel decided to do more than a generation, so soon after the launch of a new 80960 processor, is still a RISC architecture, backwards compatible with 80860, continuing to continue the "fate" of the full sale. So Intel think of simply forget, directly do x86, a way to go to the black!
Rejected Acorn in the late 70, in the early 80, RISC has just been put forward, CISC was despised, PC just start the troubled times, a small company, the emergence of the change of RISC and CISC in the industry trend.
It is in this warehouse that the arm company's major shareholders meet together, the global mobile computing market has a huge impact on the arm architecture was born.
The company was founded on December 5, 1978, the founder of physicist Herman Hauze (Hermann Hauser) and engineer Chris Curry. The company's name is also very interesting, called Cambridge Processing Unit, which means the Cambridge Processor Company, abbreviated CPU.
1979, the CPU company renamed, called Acorn, the main business from the market to provide electronic equipment, turned to provide relatively inexpensive computer equipment. How cheap is it? Priced within £ 500. But the company quickly found that Motorola's CPU was too slow and expensive (Motorola soon got hit by Intel in the CPU market and ended badly). So Acorn Company went to Intel for 80286 of design materials, intends to build a PC compatible machine, but by Intel ruthlessly refused.
I wonder if Intel will suffer for this decision for life? Rejected after the Acorn although furious, but the ambition monstrous: is not CPU it! Can't you buy it and do it by yourself? 1985, a use of RISC instruction set, named Acorn RISC Machion CPU was born, its abbreviation is arm.
Arm is characterised by excellent performance-to-power ratios, such as Apple's first handheld touchscreen personal electronic device Newton Message pad uses ARM processors and later produces multiple models. Unfortunately, due to the limitations of software and hardware technology, it has not been much of a success. But this lays the foundation for ARM's development. In 1990, Acorn company simply renamed Arm. Apple invested £ 1.5 million in equity, and acorn a stake in 1.5 million pounds of intellectual property and 12 engineers. There is also a large shareholder is the VLSI, is a chip manufacturer. Then, due to the market downturn, ARM decided to open the license to become a design company, to authorize its own design and intellectual property to other manufacturers to produce chips. This became the starting point for arm's brilliance.
Intel's reverse attack and arm expansion Intel eventually developed, taking up the majority of the global PC market, and IBM is no longer able to shake Intel's position. But Cisc's performance is always flawed, and Intel must have seen this, if X86 's performance has been unable to catch up with the RISC processors at the top, Intel has no way to reach the most fertile servers and supercomputer markets. In principle, the x86 architecture is based on CISC, and its problems are many, and the efficiency and direction of development are problematic. But Intel is not worried that, after delving into RISC, Intel has begun to boldly introduce RISC design ideas into CISC-based x86, adding additional "translation layers". The external CPU is still x86, but the internal operation is more similar to the streamlined RISC, so the CPU performance is greatly improved.
Atom was a powerful tool for Intel to move into the ultra-portable mobile computing market, but arm managed to halt it.
After the launch of the Pentium brand, Intel began to gradually introduce RISC design ideas into the x86. With the huge profits that the PC market has grabbed, Intel is starting to frantically boost the performance of the x86 processor. As a result, Intel's Xeon brand began to exert itself in the server and supercomputer markets, with no wind or even a slight win compared to the RISC processor at the time. Finally, Intel completed the reverse attack process, x86 in the server market, and even began to occupy the majority of market share, the success of CISC to conquer RISC myth.
But in the mobile computing market, Intel is not able to replicate such myths. High-performance markets are less sensitive to power consumption and performance is the most important indicator, and Intel can leverage its strong research and development capabilities to threaten its opponents and gain success. But the mobile computing market is extremely sensitive to power consumption, and Intel has failed several attempts to gain access. Arm is in this field everywhere, from mobile phones to tablets, arm with its own ultra-high performance power consumption than the success of the award.
At present, the x86 processor occupies more than 90% of the PC market, and the RISC products represented by ARM also occupy more than 90% of the mobile computing market. Both want, have hope to enter each other's territory. We will wait and see who can win.
Those things in those years CISC and RISC development entanglement
