Subsequently, Intel developed 8080 processor and 8085 processor. Together with Motorola's MC6800 microprocessor and Zilog's Z80 microprocessor, they formed an eight-bit microprocessor family.
The typical product of 16-bit microprocessor is Intel's 8086 microprocessor, and the mathematical coprocessor produced at the same time is 8087. These two chips use compatible instruction sets, but some instructions specially used for mathematical calculations such as logarithm, exponent and trigonometric function are added to the 8087 instruction set. Because these instructions apply to 8086 and 8087, they are collectively referred to as X86 instruction set. Since then, the new generation of CPU products introduced by Intel are all compatible with the original X86 instructions.
1979, Intel introduced 8088 chip, which is still a 16-bit microprocessor, including 29,000 transistors, with a clock frequency of 4.77MHz and an address bus of 20 bits, and can use 1MB memory. The internal data bus of 8088 is 16 bits, and the external data bus is 8 bits. 198 1 year, 8088 chip was first used in IBM PC. If the 8080 processor is not well known, then 8088 can be said to be a household name, and the first generation CPU——PC of personal computer started from it. Although the 80286 chip of 1982 is a 16-bit chip, it already contains 13.4 thousand transistors, and the clock frequency has reached an unprecedented 20MHz. Its internal and external data buses are 16 bits, and its address bus is 24 bits. 16MB memory can be used, and the available working modes are real mode and protected mode.
The representative product of 32-bit microprocessor is180386 introduced by Intel in 1985, which is a full 32-bit microprocessor chip and the first 32-bit chip in X86 family. It contains 275,000 transistors, and the clock frequency is 12.5MHz, and then gradually increases to 33MHz. The internal and external data buses of 80386 are 32 bits, and the address bus is also 32 bits, which can address 4GB of memory. In addition to real mode and protected mode, it also adds a virtual 86 working mode, which can provide multi-task capability by simulating multiple 8086 processors at the same time. 1989, Intel introduced the quasi-32-bit processor chip 80386SX. Its internal data bus is 32 bits, the same as 80386, and its external data bus is 16 bits. That is to say, the internal processing speed of 80386SX is close to 80386, and it also supports real multi-task operation, so it is acceptable to develop an input/output interface chip for 80286. The performance of 80386SX is better than that of 80286, and the price is only one third of that of 80386. 386 processor has no built-in coprocessor, so it can't execute floating-point operation instructions. If floating-point operation is needed, you must buy an expensive 80387 coprocessor chip.
In the late 1980s and early 1990s, 80486 processor came out, which integrated1.2000 transistors, and the clock frequency gradually increased from 25MHz to 50MHz. 80486 integrates 80386, mathematical coprocessor 80387 and an 8KB cache in one chip, and uses RISC (Reduced Instruction Set) technology for the first time in X86 series, which can execute one instruction in one clock cycle. It also adopts burst bus mode, which greatly improves the speed of data exchange with memory. Because of these improvements, the performance of 80486 is four times higher than that of 80386 with 80387 coprocessor. The early 486 is divided into 486DX with coprocessor and 486SX without coprocessor, and their prices are also very different. With the continuous development of chip technology, the main frequency of CPU is getting faster and faster, while the external devices of PC are limited by technology, which hinders the further improvement of the main frequency of CPU. In this case, the CPU frequency doubling technology appears, which makes the internal working frequency of CPU 2-3 times that of the external frequency of the processor, hence the names 486DX2 and 486DX4.
In the mid-1990s, a new generation of 586 processors came out, completely surpassing 486 processors. In order to get rid of the confusion of processor names in the 486 era, Intel Corporation, the largest CPU manufacturer, named its new generation products Pentium to distinguish AMD and Cyrix products. AMD and Cyrix also introduced K5 and 6x86 processors to deal with Intel, but because Pentium processors have the best performance, Intel gradually occupied most of the market.
I don't need to say that everyone knows the development of CPU from now on. At the beginning of 1997, Pentium MMX was listed, Pentium II and AMD K6 were listed in mid-year, and Cyrix6x8MX was listed at the end of the year. 1998 is even more "three-legged", with PII, Celeron, K6-2 and MII fighting each other. Since the introduction of Pentium II, Intel abandoned the aging Socket 7 market and pushed for the advanced Slot 1 architecture, but this time Intel made a mistake. With the increasing global demand for low-cost PCs below $ 1000, AMD's K6-2 processor fills the gap in this low-end field. AGP bus technology and 1 00 MHz external frequency, which were originally only realized on1slot, were also realized in the Super 7 era initiated by AMD. Although the performance of K6-2 and Super 7 still lags far behind the PII of the same frequency, the low price still allows AMD to grab nearly 30% of the CPU retail market share. AMD has won the favor of many consumers with a posture of not fearing strong enemies.
Unfortunately, 1999, in the face of Intel's fierce counterattack, AMD began to decline, and the market sales were very bad. Cyrix lost in this processor war. It wanted to rely on NS (National Semiconductor Company) to make a comeback, but it was too late and was finally acquired by chipset manufacturer VIA in June.
Subsequently, IDT and Reith, two new companies entering the processor market, have their own characteristics in technological innovation and market positioning. Winchip C6 and Winchip C6-2 of IDT are mainly aimed at the low-end home market, while Rise processors mainly enter the field of mobile computers. Unfortunately, under the pressure of Intel products, their lives are also very firm. 1999 In mid-1999, just one month after Cyrix was acquired, VIA acquired IDT Company, and at the same time, Rise was also acquired by another chipset manufacturer SIS (Silicon System Technology). Then it was reported that Rise withdrew from the PC processor market and focused on the home appliance processing chip market. In this way, after readjustment, the PC processor market presents a new trend of tripartite confrontation. AMD made a beautiful turnaround with the Athlon-K7 released in August. K7 became the fastest processor with overall performance surpassing similar products of Intel for the first time in history, and its market share has a trend of further expansion. After the acquisition of Cyrix and IDT, VIA integrated the latest technologies of the two companies, and planned to launch Joshua-Joshua processor compatible with Socket370 in early 2000, focusing on the low-end market. In a word, with the fierce competition, companies are trying their best to develop the latest, fastest and best processor products for consumers.
Brilliance at the end of the century-Pentium III
1999 At the beginning of this year, Intel released the third generation Pentium III processor. The first batch of Pentium III processors adopted the Katmai core, clocked at 450 MHz and 500Mhz. The biggest feature of this kernel is to update the multimedia instruction set named SSE, which adds 70 new instructions on the basis of MMX to enhance 3D and floating-point applications, and is compatible with all previous MMX programs.
However, in all fairness, apart from the SSE instruction set mentioned above, Pentium III of Katmai kernel is not attractive. It still basically retains the architecture of Pentium II, using 0.25 micron technology, 100Mhz external frequency, Slot 1 architecture, and 5 12KB secondary cache (running at half the speed of CPU), so the performance is not improved much. However, when Pentium III was first listed, it set off an upsurge. Someone once bought the first batch of Pentium III at a high price of over 10,000 yuan.
First generation Pentium III processor (Katmai)
It can be greatly improved, from 500Mhz to 1. 13Ghz, and the overclocking performance is greatly improved, and the amplitude can reach more than 50%. In addition, its secondary cache is also synchronized with the main frequency of CPU, but the capacity is reduced to 256KB.
Second Generation Pentium III Processor (Copper Mine)
In addition to the improvement brought by the manufacturing process, some Coppermine Pentium III also have a bus frequency of 133Mhz and a Socket370. In order to distinguish them, Intel added a "B" after the Pentium III model of 133Mhz bus and an "E" after the Socket370, for example, the frequency is 550EBMhz and the external frequency is 133Mhz.
Seeing the popularity of Pentium III with Coppermine kernel, Intel began to replace Celeron processor with this kernel. In mid-2000, Celeron processor with Coppermine 128 core was introduced, commonly known as Celeron2. Due to switching to 0. 18 process, Celeron's overclocking performance has made another leap, and the overclocking range can reach 1000.
Second Generation Celeron (Coppermine 128 Core) Processor
AMD's Jedi Counter-attack-Velon
As for AMD, the K6-3 processor was introduced to counter Pentium III. The K6-3 processor is designed with a three-level structure, with a built-in 64K first-level cache (Level 1) and a 256K second-level cache (Level 2), and the motherboard is equipped with a third-level cache (Level 3). K6-3 processors now also support enhanced 3D! Due to the cost and yield of instruction set, the K6-3 processor was not very successful in the desktop market, so it gradually disappeared from the desktop market and entered the notebook market.
What really makes AMD proud is the Athlon processor originally code-named K7. Athlon has a superscalar Risc core with superscalar, super pipeline and multi-pipeline. It adopts 0.25 micron process and integrates 22 million transistors. Athlon includes three decoders, three integer execution units (IEU), three address generation units (AGU) and three multimedia units (floating-point arithmetic units). Athlon can execute three transistors in the same clock cycle. K7 includes three decoders, and sends the decoded macro instructions (K7 decodes X86 instructions into macro instructions and converts X86 instructions with different lengths into macro instructions with the same length, which can give full play to the power of RISC kernel) to the instruction control unit, which can control (save) 72 instructions at the same time. And then sends the instruction to an integer unit or a multimedia unit. Integer units can schedule 18 instructions at the same time. Each integer unit is an independent pipeline, and the scheduling unit can predict the branches of instructions and execute them out of order. K7' s multimedia unit (also called floating-point unit) has a stack register that can be renamed. The floating-point scheduling unit can schedule 36 instructions at the same time, and the floating-point register can store 88 instructions. Among the three floating-point units, there is an adder and a multiplier, which can execute MMX instruction and 3DNow instruction. There is also a floating-point unit responsible for loading and saving data. Because of K7' s powerful floating-point unit, AMD processor surpassed Intel processor for the first time.
Athlon has a built-in 128KB full-speed cache (L 1 cache) and an external L2 cache with a frequency of 1/2 and a capacity of 5 12KB, which can support up to 8MB L2 cache. Large cache can further improve the huge data throughput required by the server system.
Athlon's package and appearance are similar to Pentium II, but Athlon adopts Slot A interface specification. Slot A interface originated from Alpha EV6 bus, and the clock frequency is as high as 200MHz, which makes the peak bandwidth reach 1.6GB/S, and it is still compatible with the traditional 100MHz bus on the memory bus, which protects the investment of users and reduces the cost. Later, DDR SDRAM with higher performance was adopted, which was similar to the data throughput of 800MHz RAMBUS pushed by Intel. EV6 bus can support up to 400MHz, which can perfectly support multiple processors. So it has a natural advantage. You should know that Slot 1 only supports dual processors, and slot can support 4 processors. SlotA looks very similar to the traditional Slot 1, just like Slot 1 which in turn is 180 degrees, but they are completely incompatible in electrical specifications and bus protocols. The CPU in slot 1/Socket 370 cannot be installed on the Athlon motherboard in slot A, and vice versa.
Third, the CPU in the new century
Since the beginning of the new century, CPU has entered an era of faster development, and the previously unattainable 1Ghz mark has been easily broken. In terms of market distribution, Intel and AMD still compete for hegemony, with Pentium4, Tualatin Pentium II and Celeron, Tunderbird Athlon, AthlonXP and Duron introduced respectively, and the competition is becoming increasingly fierce.
1. As for Intel, at the end of last century, in June 2000, 1 1, Intel released its fourth generation Pentium processor, which is Pentium 4 that we can touch every day. Pentium 4 did not follow PIII architecture, but adopted a brand-new design, including equivalent 400MHz front-end bus (100 x 4), SSE2 instruction set, 256K-5 12KB L2 cache, brand-new pipeline technology and NetBurst architecture, with the starting frequency of 1.3GHz.
The first Pentium4 kernel is Willamette, a brand-new Socket 423, which integrates a 256KB L2 cache, supports a more powerful SSE2 instruction set, and has as many as 20 superscalar pipelines, matching the i850/i845 chipset. Subsequently, Intel successively introduced the Willamette P4 processor of 1.4GHz-2.0GHz, and all the P4 processors in the later period switched to Socket 478 sockets with more pins.
The First Generation Pentium 4(Socket423) Processor
Like Pentium III, the first Pentium 4 core didn't get much praise, mainly because the new CPU architecture can't be fully supported by program software, so Pentium 4 often lags far behind Athlon with the same frequency, even like Intel's own Pentium III. However, a year later, Intel released the second Pentium4 kernel, code-named Northwood, which was changed to a finer 0. 13 micron process and integrated a larger 5 12KB L2 cache, which greatly improved the performance. Coupled with Intel's tireless promotion and the support of motherboard chip manufacturers, Pentium 4 has become the most popular mid-to high-end processor.
The Second Generation Pentium 4(Socket478) Processor
On the low-end CPU, Intel released the third generation Celeron core code-named Tualatin, which was also changed to 0. 13 micron process. At the same time, the secondary cache capacity is increased to 256KB, and the external frequency is also increased to 100Mhz. At present, the main frequencies of Celeron in tualatin are 1.0 and 1.655. Intel has also introduced Pentium III with Tualatin core, which integrates a larger 5 12KB L2 cache, but they are only used in the server and laptop market, and rarely seen in the desktop market.
Celeron processor with third-generation Tualatin core
2. At AMD, the second Athlon core Tunderbird was released in mid-2000. This core Athlon has the following improvements: first, the manufacturing process is improved to 0. 18 micron; second, the installation interface is changed to Socket, which is similar to Socket370, but the number of pins is 462. Finally, the secondary cache is changed to 256KB, but the speed is synchronized with the CPU, which is the same as Pentium III with copper core.
Athlon with Tunderbird core is not only slightly ahead of Pentium III in performance, but also has the highest frequency always higher than Pentium III. The milestone of 1Ghz frequency was first achieved by this CPU. However, with the release of Pentium4, Tunderbird began to lag behind its competitors in frequency. For this reason, AMD released the third Athlon core Palomino and adopted a new frequency naming system. Since then, the number on the Athlon model does not represent the actual frequency, but is converted into the frequency equivalent to the performance of competitors (that is, Intel) according to a formula, and the name has been changed to AthlonXP. For example, the actual frequency of AthlonXP 1500+ processor is not 1.5Ghz, but 1.33GHz. Finally, AthlonXP is compatible with Intel's SSE instruction set and can give full play to its performance in software specially optimized for SSE instruction set.
Athlon processor with 3rd generation Tunderbird core
On the low-end CPU, AMD introduced the Duron CPU, which has the same basic architecture as Athlon except that the secondary cache is only 64KB. Since its release, Delong has been far away from Celeron, which also focuses on the low-end market, and its price is lower. Duron has become the first choice for low-cost DIY compatible machines for a time, but it also has its fatal weakness. First, it inherits the characteristics of Velon's high calorific value, and second, its core is very fragile and easy to be damaged when installing CPU radiator. So although it is very popular in the compatible machine market, it has never been able to enter the most profitable brand machine market.
Dragon processor
Fourth, the future development direction of CPU
Looking at the history of CPU development, it is not difficult to draw the following development directions of CPU: first, higher main frequency, second, smaller manufacturing process, and third, larger cache. In addition to these three points, PC processors will gradually develop from 32-bit data bandwidth to 64-bit data bandwidth.
1, Intel's future plan, before the writing of this book, the highest frequency CPU has reached 2.4Ghz, and Intel's goal is to reach 3Ghz this year and 10Ghz two years later. To this end, Intel will release the Northwood kernel with a bus frequency of 533Mhz in mid-2002. According to the plan, in 2003, Intel will also launch a 0.09 micron Prescott core, which will work at frequencies above 3.5GHz (even higher) and use more efficient 667MHz( 166MHz x 4) or 800MHz FSB(200MHz x 4), but Prescott is only in written form at present. After all, it was officially released in 2003, so there is no more information about it at present.
In addition, Intel also revealed that in 2005, it will launch a new processor product with THz transistor architecture, which adopts many advanced technologies such as SOI technology and high-k insulator. Simply put, it can reduce the heat and power consumption of the chip to the maximum extent and greatly improve the working frequency of the processor. Theoretically, the processing products of 10GHz-20GHz can be manufactured by using terahertz transistor architecture.
Of course, to achieve such a high working frequency, only terahertz transistors are not enough. It also needs the support of the new BBUL (Bump-Free Lamination) packaging technology, which can manufacture a chip with a thickness of only 1mm and integrated 1 100 million transistors. BBUL technology is no different from the current packaging technology, but the core technology is completely different. BBUL is built in, directly on the Die. Because BBUL shortens the data transmission channel, the clock frequency speed of the whole chip will be greatly improved, and the power consumption is naturally lower.
2)AMD's future planning. By the time this book went on the market, the third Athlon core thoroughbred should have been released. Thoroughbred horses follow the core of palomino, but use more efficient 166MHz FSB and 0. 13 micron technology. Due to the improvement of manufacturing technology, its calorific value and chip size are much smaller than Palomino's, and it also adopts Socket A interface and OPGA package. The existing Athlon XP motherboards are all compatible with thoroughbred horses (AMD's thoroughbred horse demo machine announced at the exhibition uses AMD-760 chipset); However, it has not been announced whether thoroughbred horses will increase the cache capacity.
AthlonXP is at the core of thoroughness.
A new generation of dragon (Appaloosa) uses a simplified purebred core. According to AMD's latest processor development blueprint, the first thoroughbred horse core 1.73GHz Athlon XP processor is expected to be released in the first quarter of next year. Both desktop version of Athlon XP and workstation/server version of Athlon MP will fully introduce 0. 13 micron thoroughbred horse core in the first quarter of next year, and products with Barton core will be launched in the second quarter, but AMD has not announced the specific specifications of Barton core. As for the mobile processor, the last one based on Palomino core will be 1.4GHz Athlon 4 released next year, and then a brand-new thoroughbred/Barton core will be adopted.
As for AMD's first 64-bit processor K8 Hammer, it will be divided into two different versions, namely the high-end server version of SledgeHammer (up to 8 SMP) and the workstation/desktop version of ClawHammer (2 SMP), in which ClawHammer has integrated DDR33 memory controller. Therefore, the chipset corresponding to ClawHammer does not need to contain a memory controller, while SledgeHammer has a larger L2 cache. Both hammer processors will support SSE2 instruction set and be compatible with 32-bit instructions. Although there is no more information released yet, it is certain that the hammer will be released next year, and AMD has previously claimed that the performance of the hammer will surpass all similar processors.