History of CPU development
CPU stands for Central Processing Unit, which is one of the most important parts of a computer and consists of operators and controllers. If the computer is compared to a person, then the CPU is the human brain. the development of the CPU is very rapid, personal computers from the 8088 (XT) development to the current Pentium 4 era, after less than two decades of time.
From the production technology, the initial 8088 integrated 29,000 transistors, and Pentium Ⅲ integration of more than 28.1 million transistors; CPU operating speed, to MIPS (millions of instructions per second) as a unit, the 8088 is 0.75 MIPS, to the high-energy Pentium has exceeded 1,000 MIPS. regardless of what kind of CPU The internal structure of the CPU can be summarized into three main parts: the control unit, the logic unit, and the storage unit, which are coordinated with each other, analyzing, judging, calculating, and controlling all parts of the computer to coordinate the work of the commands and data.
CPU from the initial development has been more than twenty years of history, during this period, according to the word length of its processing information, CPU can be divided into: 4-bit microprocessor, 8-bit microprocessor, 16-bit microprocessor, 32-bit microprocessor, as well as is brewing to build a 64-bit microprocessor, it can be said that the development of personal computers with the development of the CPU to move forward.
Intel 4004
In 1971, Intel introduced the world's first microprocessor, the 4004, which was the first four-bit microprocessor that could be used in a minicomputer, and it contained 2300 transistors. Intel then introduced the 8008, which had a very poor market response due to poor arithmetic performance. in 1974, the 8008 was developed into the 8080, which became the second generation of microprocessors. the 8080 was used as a replacement for electronic logic circuits in a variety of application circuits and devices, which could not have been realized without a microprocessor.
As the microprocessor can be used to complete many of the previous need for larger devices to complete the task of computing, the price is cheap, so the semiconductor companies began to compete in the production of microprocessor chips. Zilog produced 8080 enhancement of the Z80, Motorola produced the 6800, Intel produced in 1976 and enhanced 8085, but these chips are basically not changing the basic characteristics of the 8080, but the chip. No change in the basic characteristics of the 8080, are the second generation of microprocessors. They all use the NMOS process, integration degree of about 9000 transistors, the average instruction execution time of 1μS ~ 2μS, using assembly language, BASIC, Fortran programming, the use of single-user operating system.
Intel 8086
The 8086, produced by Intel in 1978, was the first 16-bit microprocessor. Soon Zilog and Motorola announced plans to produce the Z8000 and 68000. this was the beginning of the third generation of microprocessors.
The 8086 microprocessor had a maximum main frequency of 8MHz, a 16-bit data channel, and a memory addressing capability of 1MB. Intel also produced a math coprocessor, the i8087, which used a compatible instruction set, but the i8087 instruction set added a number of instructions dedicated to mathematical calculations such as logarithms, exponents, and trigonometric functions. These instruction sets are referred to as the x86 instruction set. Although Intel later produced a second generation, third generation and other more advanced and faster new CPU, but are still compatible with the original x86 instructions, and Intel in the naming of the subsequent CPU followed the original x86 sequence, until later due to trademark registration issues, only to give up continue to use Arabic numerals named.
In 1979, Intel went on to develop the 8088. Both the 8086 and 8088 used 16-bit data transfer within the chip, so they are both referred to as 16-bit microprocessors, but while the 8086 was capable of transferring or receiving 16 bits of data per cycle, the 8088 used only 8 bits per cycle. Because most of the original devices and chips were 8-bit, and the 8088's external 8-bit data transmission and reception can be compatible with these devices. 8088 is packaged in a 40-pin DIP package, and operates at 6.66MHz, 7.16MHz, or 8MHz, and the microprocessor integrates about 29,000 transistors.
Soon after the 8086 and 8088 were introduced, Intel began to improve them by integrating more functions on the chip, which resulted in the 80186 and 80188.Both microprocessors operated internally with 16-bit operation, and on the external inputs and outputs the 80186 used 16-bit, while the 80188 operated with 8-bit, as did the 8088.
In 1981, IBM used the 8088 chip in the PC it developed, thus creating a whole new era of microcomputers. It was also from the 8088 that the concept of personal computers (PCs) began to develop around the world. From the 8088 to IBM PCs, personal computers really came into people's work and life, and it also marks the beginning of a new era.
Intel 80286
In 1982, Intel developed the 80286 microprocessor on the basis of the 8086, the microprocessor's maximum main frequency of 20MHz, internal and external data transfer are 16-bit, the use of 24-bit internal memory addressing, the memory addressing capacity of 16MB. 80286 can be operated in two ways. One is called real mode and the other is called protected mode.
In real mode, the total amount of memory that the microprocessor can access is limited to 1 megabyte; whereas, under protected mode, the 80286 can directly access 16 megabytes of memory. In addition, the 80286 operates in protected mode, which protects the operating system from being shut down in the event of an application exception, as is the case with unprotected microprocessors such as the real mode or the 8086.
IBM made a big splash by using the 80286 microprocessor in an advanced technology microcomputer, the AT machine. the 80286 was a significant improvement over its predecessor in four areas: support for larger memory; the ability to emulate memory space; the ability to run multiple tasks simultaneously; and increased processing speeds. The earliest PCs ran at 4MHz, the first 80286-based AT machines ran at 6MHz to 8MHz, and some manufacturers increased their own speeds to bring the 80286 up to 20MHz, which represented a significant improvement in performance.
The package for the 80286 was a square package known as a PGA, an inexpensive package derived from the PLCC that has a block of internal and external solid pins, and in this package the 80286 integrated about 130,000 transistors.
The bus of the IBM PC/AT microcomputer maintained the three-level bus structure of the XT and added a high- and low-bit byte bus driver conversion logic and a high-bit byte bus. Like the XT machines, the CPU was soldered to the motherboard.
At that time, the original machine only referred to the IBM PC machine, and the compatible machine is other than the IBM PC other machines. At that time, the production of CPU companies, in addition to Intel, and AMD and Siemens, etc., and people on their own computers with what the CPU does not care, because AMD and other companies to produce the CPU is almost the same as Intel's, until the 486 era people care about their own CPU to.
The 8086-80286 era was the era of the start of personal computers, when very few people used or even saw PCs in China, and it was a mysterious thing in people's minds. Only in the early nineties did the country begin to popularize computers.
Intel 80386
By the spring of 1985, Intel had become a top-tier chip company, and was determined to develop a new generation of 32-bit core CPUs, the 80386. Intel designed three technical points for the 80386: the use of a "class-286" architecture, the development of a "class-286" architecture, and the development of a "class-286" architecture, as well as the development of a "class-286" architecture. Intel designed three technical points for the 80386: the use of a "class 286" architecture, the development of the 80387 microprocessor to enhance floating-point computing, and the development of a cache to solve the memory speed bottleneck.
On October 17, 1985, Intel's groundbreaking product, the 80386DX, was officially released, with 275,000 transistors and a clock frequency of 12.5MHz, which was gradually increased to 20MHz, 25MHz, 33MHz, and finally a small number of 40MHz products. 40MHz products.
The 80386DX has a 32-bit internal and external data bus, and a 32-bit address bus, and can address up to 4GB of memory and manage up to 64TB of virtual storage. In addition to its real and protected modes of operation, the 80386DX adds a "virtual 86" mode of operation, which provides multitasking capabilities by emulating multiple 8086 microprocessors at the same time.
The 80386DX has more instructions than the 80286, and at 12.5MHz, the 80386 can execute 6 million instructions per second, which is 2.2 times faster than the 80286 at 16MHz. 80386's most iconic product is the 80386DX-33MHz, which is what we generally refer to when we say 80386.
The most classic 80386 is the 80386DX-33MHz, to which we generally refer.
Because of the powerful computing power of 32-bit microprocessors, PC applications have expanded into many areas, such as business office and computing, engineering design and computing, data centers, and personal entertainment. 80386 made 32-bit CPUs the standard in the PC industry.
Although the 80386 did not have a well-established and powerful floating-point unit at that time, paired with the 80387 coprocessor, the 80386 was able to successfully accomplish many tasks that required a large amount of floating-point operations, and thus successfully entered the mainstream commercial computer market. In addition, the 30386 is supported by a wealth of other peripheral accessories, such as 82258 (DMA controller), 8259A (interrupt controller), 8272 (disk controller), 82385 (Cache controller), 82062 (hard disk controller), and so on. In response to the speed bottleneck of memory, Intel designed a cache (Cache) for the 80386, taking the approach of pre-reading memory to alleviate this speed bottleneck, and since then, the Cache has become like a shadow with the CPU.
Intel 80387/80287
Strictly speaking, the 80387 is not a real CPU, but a co-processor chip that works with the 80386DX, which means that the 80387 can only assist the 80386 in accomplishing floating-point operations.
Intel 80386SX
In 1989, Intel introduced a quasi-32-bit microprocessor chip, the 80386SX, which was a cheaper and more popular CPU introduced by Intel in order to expand its market share, with a 32-bit internal data bus and a 16-bit external data bus, and which could accept the 16-bit input/output interface chip developed for the 80286. input/output interface chip, reducing the overall cost of the machine.
After the launch of the 80386SX, it was widely welcomed by the market because the performance of the 80386SX was significantly better than that of the 80286, while the price was only one-third of the 80386.
Intel 80386SL/80386DL
Intel introduced two types of 386 chips, the 80386SL and the 80386DL, in 1990, specifically for use in laptops. These two types of chips can be described as energy-efficient versions of the 80386DX/SX, where the 80386DL is based on the 80386DX core and the 80386SL is based on the 80386SX core. These two types of chips not only consume less power, but also have a power management function that automatically cuts off the power supply when the CPU is not working.
Motorola 68000
Motorola's 68000 is the earliest 32-bit microprocessor launched in 1984, after the launch, the performance is superb, and get the favor of Apple, in its own epoch-making personal computer "PC-MAC" in the use of the chip. "The chip was used in its own groundbreaking personal computer, the PC-MAC. But after the 80386 was released, it fell into disuse.
AMD Am386SX/DX
AMD's Am386SX/DX was a third-party chip that was compatible with the 80386DX, with performance comparable to Intel's 80386DX, and became one of the dominant products of its time.
IBM 386SLC
This is designed by IBM on the basis of researching the 80386, and is fully compatible with the 80386, and is manufactured by Intel. 386SLC is basically a built-in Cache on top of the 80386SX, and contains the instruction set of the 80486SX, which has a good performance.
Intel 80486
In 1989, the 80486 chip that we all know so well was introduced by Intel. The great thing about this chip, which took four years to develop and $300 million to invest, is that it broke the 1 million transistor barrier for the first time, integrating 1.2 million transistors and using a 1-micron manufacturing process. 80486's clock frequency was gradually increased from 25MHz to 33MHz, 40MHz, and 50MHz.
The 80486 was the result of a combination of the 80386 and the Mathematical Association of Computing Machinery (MACM). microprocessor 80387 and an 8KB cache in a single chip. 80487 integrated in 80486 has twice the number-crunching speed of the previous 80387, and the internal cache shortens the microprocessor's wait time for slower DRAMs. And, for the first time in the 80x86 family, RISC (Reduced Instruction Set) technology was used, which allowed one instruction to be executed in one clock cycle. It also utilized the burst bus method, which greatly increased the speed of data exchange with memory. As a result of these improvements, the performance of the 80486 was four times higher than that of the 80386 DX with the 80387 math co-microprocessor.
With the continuous development of chip technology, the frequency of the CPU is getting faster and faster, while the PC external devices are limited by the process restrictions, and can withstand a limited operating frequency, which prevents the further increase of the main frequency of the CPU. In this case, the CPU multiplier technology appears, the technology makes the CPU internal operating frequency for the microprocessor external frequency of 2 to 3 times, 486 DX2, 486 DX4 name is from this.
Intel 80486 DX
Common 80486 CPUs are 80486 DX-33, 40, and 50. 486 CPUs are 32-bit inside and out like the 386 DX, but the slowest 486 CPUs are faster than the fastest 386 CPUs, because the 486 SX/DX executes an instruction in one oscillation cycle, whereas the 386DX CPU requires two cycles.
Intel 80486 SX
Because the 80486 DX CPU has a built-in floating-point co-processor, it is powerful and, of course, expensive. In order to meet the needs of ordinary users, especially those who don't need to do a lot of floating-point operations, Intel introduced the 486 SX CPU. 80486 SX motherboards usually have 80487 co-microprocessor sockets, so if you need floating-point co-microprocessor functionality, you can plug in an 80487 co-microprocessor chip, which will be equivalent to the 486 DX. Common 80486 SX CPUs are: 80486 SX-25, 33.
Intel 80486 DX2/DX4
In fact, the name of this kind of CPU is related to the frequency, and the internal frequency of this kind of CPU is two/four times of the frequency of the motherboard, such as 80486 DX2 -66, the frequency of the CPU is 66MHz, while the frequency of the motherboard can be as long as 33MHz.
Intel 80486 SL CPU
The 80486 SL CPU was originally designed for laptops and other portable machines, and like the 386SL, this chip uses a 3.3V rather than 5V power supply and also has internal cutoff circuits that allow the microprocessor and some other optional components to sleep when not in operation, thus reduce power consumption and extend the life of laptops and other portables.
Intel 486 OverDrive
Upgrading a 486 SX can be done by installing an 80487SX chip in the motherboard's co-microprocessor slot to make it equivalent to a 486 DX, but such an upgrade only adds floating-point co-microprocessor power, and doesn't increase the speed of the system. In order to increase the speed of the system, there is another way to upgrade, which is to insert a 486 OverDrive CPU into the co-microprocessor slot, which works on the same principle as the 486 DX2 CPU, and whose internal operating speed can be twice as fast as the external speed. For example, if a 20MHz motherboard is equipped with an OverDrive CPU, the internal operating speed of the CPU can reach up to 40MHz. 486 OverDrive CPUs are also available as floating-point co-microprocessors, commonly known as OverDrive-50, 66, and 80.
TI 486 DX
The TI 486 DX
The TI 486 DX
is the world's largest and most advanced graphics card manufacturer. p>As one of the world's leading semiconductor manufacturers, the United States, Texas Instruments (TI), also in the 486 era of the emergence of its own production of the 486 DX series of CPUs, especially in the 486DX2 became the mainstream after the DX2-80 due to the higher price-performance ratio to become one of the mainstream products at the time, the highest frequency of the TI 486 for the DX4
Cyrix 486DLC
This is the 486 CPU produced by Cyrix, and to say that it is a 486 CPU means that it is close to a 486 CPU in terms of efficiency, but it is not a 486 CPU in the strict sense of the word, which is due to the characteristics of the 486 CPU. 486DLC CPU is just a combination of the 386DX CPU and 1K The 486DLC CPU is just a combination of a 386DX CPU and a 1K Cache on a single chip, with no floating-point co-microprocessor, and it takes two oscillation cycles to execute an instruction. However, due to the sophisticated design of the 486DLC CPU, the efficiency of the 486DLC-33 CPU was close to that of Intel's 486 SX-25, while the 486DLC-40 CPU surpassed the 486 SX-25, and the 486DLC-40 CPU is cheaper than 486 SX-25. 486DLC CPU is designed for upgrading 386DM, if you have a 386 computer and want to upgrade to 486 but don't want to replace the motherboard, you can unplug the original 386 CPU and plug in a 486DLC CPU and you're good to go.
Cyrix 5x86
Since Intel went the other way and developed the Pentium, Cyrix quickly launched its own new generation of products, the 5x86, which still utilizes the original 486 series of CPU sockets, but increases the main frequency from 100MHz to 120MHz. 5x86 has increased its performance compared to the 486, but compared to the Pentium, the 5x86 has increased its performance. However, compared to the Pentium, not only is the floating point performance far from adequate, even the integer arithmetic performance that Cyrix has always prided itself on is not so superb, giving people a feeling of inadequacy compared to the bottom. Since the 5x86 can use a 486 motherboard, it is generally viewed as a transitional product.
AMD 5x86
The AMD 486DX was AMD's lynchpin in the 486 market, with a built-in 16KB write-back cache and the beginning of the era of single-cycle multi-component instructions, as well as paged virtual memory management technology. Due to the late introduction of the 486DX2-80 by TI at a very low price, and the introduction of the Pentium series by Intel, AMD introduced the 5x86 series of CPUs in order to capture the gap in the market. it was the highest frequency product in the 486 class, the 5x86-120 and 133. it utilized an all-in-one 16K write-back cache, 0.35 micron process, 33×4 at 133 frequency, performance is straight out of the Pentiun 75 and consumes less power than the Pentium.
Intel Pentium
In 1993, the new generation of 586 CPUs, which comprehensively surpassed the 486, was introduced, and in order to get rid of the confusing names of microprocessors in the 486 era. Intel named their new generation Pentium to differentiate their products from AMD and Cyrix. AMD and Cyrix also introduced K5 and 6x86 microprocessors respectively to take on the chip giants, but Intel gradually took over most of the market due to the best performance of the Pentium microprocessors.
Pentium's most junior CPUs were the Pentium 60 and Pentium 66, which operated at two frequencies, 60MHz and 66MHz, respectively, which were the same as the system bus frequency, and did not have what we now call multiplier settings.
Early Pentiums from 75MHz to 120MHz used a 0.5 micron manufacturing process, and later Pentiums above 120MHz frequency switched to a 0.35 micron process. The performance of the classic Pentium is fairly average, with good integer and floating point operations.
Intel Pentium MMX
In order to improve the ability of computers to use multimedia and 3D graphics, a number of new instruction sets have emerged, the three most famous of which are Intel's MMX, SSE, and AMD's 3D NOW! MMX (MultiMedia Extensions) is a multimedia instruction enhancement technology invented by Intel in 1996, which includes 57 multimedia instructions that can process multiple data at once, and MMX technology can get better performance with the cooperation of software.
The Pentium MMX, formally known as the Pentium with MMX technology, was released in late 1996. From the beginning of Pentium MMX, Intel started to lock the multiplier on its CPUs, but the MMX CPUs were particularly strong in overclocking, and could also be overclocked by increasing the core voltage, so overclocking was a very fashionable action at that time. The term overclocking also became popular from that time.
The Versatile Pentium was another successful Intel product after the Pentium, and its vitality was quite tenacious. The Versatile Pentium made significant improvements on the basis of the original Pentium by adding an on-chip 16KB data cache and 16KB instruction cache, a 4-way write cache as well as a branch prediction unit and return stack technology. In particular, the newly added 57 MMX multimedia instructions make the versatile Pentium much faster than a Pentium CPU of the same frequency, even when running non-MMX optimized programs.
These 57 MMX instructions are specifically designed to work with audio, video, and other data. These instructions greatly reduce the amount of time the CPU has to wait for multimedia data to process, giving the CPU more powerful data processing capabilities. Unlike the Classic Pentium, the Multi-Power Pentium utilizes a dual-voltage design with a 2.8V core voltage and the system I/O voltage remaining at the original 3.3V. If the motherboard does not support the dual-voltage design, it will not be possible to upgrade to the Multi-Power Pentium.
The Multi-Power Pentium, codenamed P55C, is the first CPU with MMX technology (Integer Type Cell Execution), with 16KB Data L1 Cache, 16KB Instruction L1 Cache, SMM compatibility, 64-bit bus, 528MB/s bandwidth, 2 clock wait times, 4.5 million transistors, and 17 watts of power consumption. Supported operating frequencies are:133MHz, 150MHz, 166MHz, 200MHz, 233MHz.
Intel Pentium Pro
There was a time when Pentium Pro was synonymous with high-end CPUs, and the performance demonstrated by the Pentium Pro amazed many people at the time, but The Pentium Pro was a CPU designed with 32-bit data structures, so the Pentium Pro had mediocre performance when running 16-bit applications, but was still a 32-bit winner, but then the MMX came along and eclipsed it.
The core architecture of Pentium Pro (high-powered Pentium, 686-class CPU) is codenamed P6 (also the core architecture used in the future PII and PIII), which is the first generation, with 256KB or 512KB of L2 Cache, and a maximum of 1MB of L2 Cache.The operating frequencies are:133/66MHz (engineering samples), 150
AMD K5
K5 was AMD's first independently produced x86-class CPU, released in 1996. Due to development problems, the K5 was released much later than Intel's Pentium, and coupled with poor performance, this unsuccessful product caused AMD to lose a significant amount of market share at one point.The K5's performance was very mediocre, not as good at integer computing as Cyrix's 6x86 but still slightly better than the Pentium, and its floating-point capabilities were nowhere near as good as the Pentium, but slightly better than Cyrix's 6x86, and its floating-point capabilities were much better than Cyrix's 6x86, but slightly better than Cyrix's 6x86. K5's low price is clearly more attractive to consumers than its performance, and the low price is the biggest selling point for this CPU.
AMD K6
Naturally, AMD was not willing to let the Pentium take over the CPU market, so in 1997 they launched the K6. The design specifications for the K6 CPU are quite high, with new MMX instructions and 64KB L1 Cache (double that of the Pentium MMX), and the overall performance is superior to the Pentium MMX, and close to the same frequency as the PII. The K6 can process more instructions in parallel and run at a higher clock frequency than the K5, and AMD has done a very successful job with integer arithmetic, where the K6 lags a bit behind the Pentium in running applications that require MMX or floating point arithmetic at the same frequency.
The K6 has a 32KB data L1 Cache, a 32KB instruction L1 Cache, 8.8 million transistors integrated, 0.35 micron technology, five-layer CMOS, C4 process flip-chip, and a core area of 168 square millimeters (compared to 68 square millimeters in the newer products), using the Socket 7 architecture.
Cyrix 6x86/MX
Cyrix is also a veteran CPU developer, back in the x86 era, it formed a triple threat with Intel and AMD.
Since Cyrix merged with National Semiconductor, it finally has its own chip production line, and the finished product is becoming more and more sophisticated and complete, Cyrix's 6x86 is a Pentium-compatible microprocessor that was released to the market.
IDT WinChip
As a newcomer to the field of CPU production, IDT (Integrated Device Technology) in the United States, launched its first micro-microprocessor product in 1997, WinChip (i.e., C6), which accounted for less than 1% of the overall CPU market. In May 1998, IDT announced its second generation product, WinChip 2 .
WinChip 2 improved on the original WinChip by adding a dual-instruction MMX unit and enhanced floating-point operations. The improved WinChip 2 delivers approximately 10% more performance than a WinChip of the same frequency, essentially reaching the performance of an Intel Pentium microprocessor.
Intel Pentium II
1997-1998 was an exceptionally competitive year for the CPU market, with a wide range of colorful CPU chips during this period.
The Chinese name for Pentium II is "Pentium II", and there are several series of products with different core structures such as Klamath, Deschutes, Mendocino, Katmai, and so on, of which the first generation adopts the Klamath core, manufactured in a 0.35 micron process.
The Pentium II microprocessor uses a dual independent bus structure, where one bus is connected to the L2 cache and the other is responsible for the main memory.
The Pentium II uses an external high-speed L2 Cache detached from the chip, with a capacity of 512KB, and runs at half of the CPU's main frequency. speed. As a compensation, Intel increased the L1 Cache of the Pentium II from 16KB to 32KB. In addition, in order to beat the competition, Intel adopted the patent-protected Slot 1 interface standard and SECC (Single Edge Contact Cartridge) packaging technology for the first time in the Pentium II.
On April 16, 1998, Intel's first 350 and 400 MHz CPUs, codenamed Deschutes, supporting 100 MHz rated external frequency, were launched. The Pentium II microprocessors with the new cores not only have their external frequency increased to 100MHz, but they are also manufactured using a 0.25 micron process, and their core operating voltage has been reduced from 2.8V to 2.0V, with L1 Cache and L2 Cache of 32KB and 512KB, respectively.
Between 1998 and 1999, Intel launched its first Pentium II microprocessors, code-named Deschutes, at a cost of US$150,000 per month, with a total cost of US$150,000 per month. Between 1998 and 1999, Intel introduced the Xeon (Xeon microprocessor), a more powerful CPU than the Pentium II. This microprocessor uses a core similar to the Pentium II, 0.25 micron manufacturing process, and supports 100MHz external frequency.Xeon can be equipped with a maximum of 2MB Cache and runs at the CPU core frequency, which is different from that of the Pentium II using a chip called CSRAM (Custom StaticRAM, Customized Static Memory). In addition, it supports eight CPU systems; uses 36-bit memory addresses and PSE mode (PSE36 mode), with a maximum memory bandwidth of 800MB/s. The Xeon microprocessor is mainly aimed at servers and workstation systems with higher performance requirements, and in addition, the Xeon's interface form factor has been changed to use a slightly larger Slo than Slot 1