Development history
CPU in X86 era
The CPU can be traced back to 197 1 year. At that time, Intel, which was still in the research and development stage, introduced the world's first microprocessor 4004. This is not only the first 4-bit microprocessor used in calculators, but also the first computer processor that individuals can afford! 4004 contains 2300 transistors, and its function is quite limited and its speed is still very slow. It was despised by the blue giant IBM and most commercial users at that time, but it was an epoch-making product after all. Since then, Intel has formed an indissoluble bond with microprocessors. It can be said that the historical development of CPU is actually the development of Intel X86 series CPU, which is also the "journey of CPU history".
1978, Intel led the trend again, and produced a 16-bit microprocessor for the first time, named i8086. At the same time, a matching mathematical coprocessor i8087 is also presented. 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 i8087 instruction set. Because these instruction sets are applied to i8086 and i8087, people also call them X86 instruction sets. Although Intel produced more advanced and faster new CPUs such as the second generation and the third generation, they were still compatible with the original X86 instructions. Intel followed the original X86 order when naming subsequent CPUs until it gave up naming with Arabic numerals due to trademark registration problems. As for other companies that developed later, such as AMD and Cyrix, the CPU before 486 (including 486) was named after its own X86 CPU, but by 586, the market competition became more and more fierce. Because of the trademark registration problem, they can no longer use the same or similar naming as Intel's X86 CPU, so they must name their own 586 and 686 compatible CPUs.
1979, INTEL introduced 8088 chip, which still belongs to 16-bit microprocessor and contains 29,000 transistors. The clock frequency is 4.77MHz, the address bus is 20 bits, and 1MB memory can be used. The internal data bus of 8088 is 16 bits, and the external data bus is 8 bits, while its brother 8086 is 16 bits. 198 1 8088 chip was first used in IBM PC, which created a brand-new microcomputer era. Also from 8088, the concept of PC (personal computer) began to develop all over the world.
1982, when many young readers were still in their infancy, INTE has launched the latest epoch-making product Zao 80286 chip, which has developed rapidly compared with 8006 and 8088. Although it is still a 16 bit structure, it contains134,000 transistors in the CPU. Its internal and external data bus is 16 bits, its address bus is 24 bits, and it can address 16MB of memory. Since 80286, CPU has evolved into two working modes: real mode and protected mode.
Intel 80286 processor
1985, INTEL introduced 80386 chip, which is the first 32-bit microprocessor in 80X86 series, and its manufacturing process has also made great progress. Compared with 80286, 80386 contains 275000 transistors, and the clock frequency is 12.5MHz, and then it is upgraded to 20MHz, 25MHz and 33MHz. The internal and external data buses of 80386 are 32-bit, and the address bus is also 32-bit, which can address up to 4GB of memory. In addition to real mode and protected mode, it also adds a working mode called virtual 86, which can provide multi-task capability by simulating multiple 8086 processors at the same time. In addition to the standard 80386 chip, which is often referred to as 80386DX, for different market and application considerations, INTEL has successively introduced some other types of 80386 chips: 80386SX, 80386SL, 80386DL and so on. 80386SX introduced by 1988 is a chip with market positioning between 80286 and 80386DX. The difference from 80386DX is that the external data bus and address bus are the same as 80286, which are 16 bits and 24 bits respectively (i.e., the addressing ability is 16MB).
80386 SL and 80386 DL introduced by 1990 are low-power and energy-saving chips, which are mainly used in portable computers and energy-saving desktops. The difference between 80386 SL and 80386 DL is that the former is based on 80386SX and the latter is based on 80386DX, but both of them have added a new working mode: system management mode. When entering the system management mode, the CPU will automatically reduce the running speed, control the display screen, hard disk and other components to stop working, or even stop running, and enter the "sleep" state to achieve the purpose of energy saving. 1989, the well-known 80486 chip was introduced by INTEL. The greatness of this chip is that it actually breaks the boundary of 1 10,000 transistors and integrates1.20,000 transistors. The clock frequency of 80486 is gradually increased from 25MHz to 33MHz and 50MHz. 80486 integrates 80386, mathematical coprocessor 80387 and an 8KB cache in one chip. The 80X86 series adopts RISC (Reduced Instruction Set) technology for the first time, and one instruction can be executed in one clock cycle. It also adopts the burst bus mode, which greatly improves the data exchange speed with the memory. Because of these improvements, the performance of 80486 is four times higher than that of 80386DX with 80387 math coprocessor. Like 80386, there are several types of 80486. The original model introduced above is 80486DX. 1990 introduces 80486SX, which is a low-cost model of 486. The difference between it and 80486DX is that it has no mathematical coprocessor. 80486 DX2 is based on clock frequency doubling technology, which means that the running speed inside the chip is twice as fast as that of the external bus, that is, the running speed inside the chip is twice as fast as that of the system clock, but it still communicates with the outside world at the original clock speed. The internal clock frequency of 80486 DX2 mainly includes 40MHz, 50MHz and 66MHz. 80486 DX4 is also a chip with clock frequency doubling technology, which can make its internal unit run at twice or three times the speed of external bus. In order to support this improved internal working frequency, its on-chip cache is extended to 16KB. The clock frequency of 80486 DX4 is 100MHz, which is 40% faster than the 66MHz of 80486 DX2. 80486 also has SL enhancement and system management mode, which is used for portable computers or energy-saving desktops.
[1] Dual-core processors of various brands
Intel Corporation of the United States (one of the Fortune 500 companies is famous for producing CPU chips)
Pentium dual core:
It is the core of Pentium D and Pentium 4EE and Praessler. Basically, Praessler Core is a simple product that couples two cedar mill cores together.
Core 1 generation
Yonah core architecture is adopted.
[2] 2nd generation Core processors
Adopt Conroe core (incomplete).
Core is a leading energy-saving new micro-building. The starting point of the design is to provide excellent performance and energy efficiency, and improve the performance per watt, which is the so-called energy efficiency ratio. The early Core was based on notebook processors.
With the progress of IT technology, the concept of "multi-core" is becoming more and more popular, which mainly refers to the dual-core technology based on X86 open architecture. In this regard, the leading manufacturers are mainly Intel and AMD. Among them, the two schools have different ideas. AMD has considered multi-core support from the beginning of its design. All components are directly connected to the CPU, eliminating the challenges and bottlenecks in the system architecture. Multiple processor cores are directly connected to the same core, and the cores communicate at chip speed, further reducing the delay between processors. Intel uses multiple cores to share the front-side bus. Experts believe that AMD's architecture is easier to achieve dual-core or multi-core, and Intel's architecture will encounter the bottleneck problem of multi-core competing for bus resources.
The introduction of dual-core processor technology is an effective way to improve processor performance. Because the actual performance of the processor is the total number of processor instructions that the processor can execute in each clock cycle, the number of units that the processor can execute in each clock cycle will be doubled by adding a core. It must be emphasized here that in order to achieve the maximum performance of the system, all executable units in the two cores must be fully utilized: that is, all executable units must have work to do!
Various brands of dual-core processors
Intel Corporation of the United States (one of the Fortune 500 companies is famous for producing CPU chips)
Core is a leading energy-saving new micro-building. The starting point of the design is to provide excellent performance and energy efficiency, and improve the performance per watt, which is the so-called energy efficiency ratio. The early Core was based on notebook processors.
Core 2: English Core 2 Duo is one of the new generation product systems based on Core microarchitecture launched by Intel. Published on July 27th, 2006. Core 2 is a cross-platform architecture, including server version, desktop version and mobile version. Among them, the development code of the server version is Woodcrest, the development code of the desktop version is Conroe, and the development code of the mobile version is Merom.
Features:
Brand new core architecture
All adopt 65 nm manufacturing process.
The whole line of products has single core, dual core and quad core. Up to now, there are two versions of L2 cache capacity: 2MB and 4MB, and there was a 2MB cache capacity when it went on the market.
Performance improvement of 40%
Energy consumption is reduced by 40%, and the average energy consumption of mainstream products is 65 watts.
The front-end bus is upgraded to 1066Mhz(Conroe), 1333Mhz(Woodcrest) and 667Mhz(Merom).
The server class Woodcrest is development code, and the actual product name is Xeon 5 100 series.
Use LGA77 1 interface.
Zhiqiang 5 100 series contains two FSB product specifications (1066 MHz for 510, 1333 MHz for 5 130). It has two processing cores and 4MB shared L2 cache, with an average power consumption of 65W and a maximum of 80W, which is better than AMD's Snapdragon's power consumption of 95W.
Conroe processors for desktop computers are divided into two types: the normal version and the extreme version. The product line includes E6000 series and E4000 series. The main difference between them is the different FSB frequencies.
The main frequency of common E6000 series processors ranges from 1.8GHz to 2.67GHz. Although the frequency is low, Conroe processor has excellent performance due to its excellent core architecture. In addition, Conroe processor also supports Intel's VT, EIST, EM64T and XD technologies, and adds Sup-SSE3 instruction set, which is also commonly known as SSE3 instruction set. Due to the efficient architecture of Core, Conroe no longer provides HT support.
Age-related macular degeneration
AMD, that is, the processor slot is Socket AM2,940 pins.
AMD's Athlon 64 series processors have been on the market for more than a year. Due to the integration of memory controller, Athlon 64 series processor platform still stays in the DDR era, and as early as mid-2004, Intel has begun to vigorously promote DDR2 memory. In this case, AMD introduced the first processor supporting DDRII memory. AM2 adopts 90nm SOI technology and is equipped with 1MB or 2MB.
Analysis of technical characteristics of 1.Socket AM2 processor
1, frequency improvement is a difficult problem, looking forward to the introduction of new technology.
The kernel with Socket AM2 pin is called "F" stepping, which has all the characteristics of the current "E" stepping kernel, but the only difference is that it has been upgraded from the previous generation which supports dual-channel DDR 400 to dual-channel DDR2 800, and AMD virtual technology has been added.
Compared with the current "E" stepping core, the "F" stepping core obviously reduces the L2 cache except the change of memory controller and the addition of AMD virtual technology. According to the official documents of AMD, due to the maturity of manufacturing technology, the L2 cache of Rev F version core has been redesigned and reduced to be used as a circuit (transistor) to improve the speed. In addition, the quality of the "F" stepping core has also been improved. Under the same power consumption, the frequency can be increased by 7% compared with the previous generation Rev E, or the power consumption at this frequency can be reduced by about 7%, so the "F" stepping core will be able to improve the production capacity of the low-power version.
In terms of the number of transistors, although the number of transistors used in L2 cache has decreased, the number of transistors and core size of Rev F core have increased due to the switch to DDR2 memory controller and the addition of AMD virtual technology. For example, the Windsor core for dual-core processors has increased from 233 million in the previous generation to 243 million, and the die size has also increased from 199 mm2 to 220 mm2.
The overall power consumption has decreased, only FX-62 is a special case, so it is necessary to mention the overall performance improvement and power consumption reduction of AMD AM2 products.
L2 cache is manufactured by AMD's Fab 30 factory in Dresden, Germany.
2. The built-in DDR2 memory controller supports DDR2-800 memory.
The biggest improvement of Socket AM2 processor is the integration of DDR2 memory controller-initially it will support DDR2 667, and later it will support DDR2 800 or even DDR2 1066.
The advantages and disadvantages of DDR2 are very obvious: although the bandwidth of DDR2 memory is improved, the memory delay of DDR2 is longer than that of DDR memory, which also causes the shortcomings of DDR2 with high frequency and low energy. Fortunately, at present, memory manufacturers have improved their production technology, and the delay of the new generation DDR2 667 memory has reached the level of 3-3-3 timing. At the same time, with the advantage of high bandwidth, the performance has been equal to or even exceeded the previous DDR400 memory.
Considering that AMD's AM2 processor itself integrates the internal memory controller of CPU, its advantages of high bandwidth and extremely low latency will lead Intel's latest DDR2 platform in memory control. However, DDR module needs 184 pins, and DDR2 module needs 240 pins. AMD upgraded from supporting dual-channel DDR to dual-channel DDR2 on the premise of basically maintaining the number of processor pins, which increased the complexity of the core to some extent.
It should be noted that the DDR2 memory frequency supported by the high-end processor and the low-end processor of AM2 platform is mostly the same. The top-end Athlon 64 FX and Athlon 64 X2 support the highest DDR2-800, and the memory transmission bandwidth reaches 12.8GB/s/s, while the low-end Athlon 64 and Sempron processors support DDR2-667, and the memory transmission bandwidth is 10.66GB/s/s ... That is to say, AM2 gave up on DDR 2.
3. Support Presidio security technology and Pacifica virtual technology.
Of course, the improvement of Socket AM2 processor is not only to provide support for DDR2 memory and pin code changes. AMD said that Socket AM2 processor will support Presidio security technology and Pacifica virtual technology. In fact, Athlon64 is the first desktop processor that supports anti-virus technology. Considering that this is also one of the development trends of CPU in the future, it is not surprising that the Socket AM2 processor still retains this function.
What deserves our attention is Pacifica virtual technology, which will greatly improve the running ability of desktop processors. The most prominent feature of Pacifica technology is the improvement of memory controller. "Pacifica" improves the virtual application of CPU by direct connection architecture and introducing new models and functions into the processor and memory controller.
Different from the previous virtual application methods, this new technology can reduce the complexity of the program, improve the security of the virtual system and reduce the cost of the virtual management system by being compatible with the existing virtual system management software. For example, users can easily create multiple independent and isolated partitions on a machine, thus reducing the risk of virus transmission between partitions. However, AMD is still one step behind Intel in virtualization technology.
Analysis of Three Core Series of AMD Socket AM2
According to AMD's plan, a new generation of processor cores, including Windsor, Orleans and Manila, will begin to adopt Socket AM2 specification and 90nm technology, and also support dual-channel DDR2 memory. Athlon64 X2 dual-core processor of Windsor core and Athlon64 of Orleans core have built-in Pacifica virtual technology, which is not supported by Sempron processor of Manila core. Below, let's briefly introduce three new series processors of AMD.
"Windsor Core" of High-end Market
For this year's high-end processor market, AMD has prepared Athlon 64 X2 dual-core processor based on Socket AM2 architecture and code-named Windsor core. As the shipment of high-end dual-core Athlon64 X2 will gradually increase from 2006, replacing the position of single-core Athlon64 processor in the middle and high-end market, Athlon 64 X2 alone has planned many products such as 4200+, 4600+, 4800+, 5000+ and 5200+ in the next-generation Socket AM2 processor.
In addition, AMD will also bring us the Athlon 64 FX processor of Windsor core, which is still positioned as "providing the best performance for 3D games and single-threaded applications" and will continue to play the role of the best processor in games.
"The Core of Orleans" Facing the Mainstream Market
The core code-named "Orleans" is a single-core processor aimed at the mainstream processor market. This year, AMD will launch Athlon 64 3500+, Athlon 64 3800+ and Athlon 64 4000+, all of which support Pacifica virtual technology: Socket M2 Athlon 64 4000+ works at 2.6GHz and 565 438+02 KB L2; Socket M2 Athlon 64 3800+ processor works at 2.4GHz, and Socket M2 Athlon 64 3500+ processor works at 2.2GHz, which may be equipped with 5 12KB L2 cache. Considering that the performance of DDR2 memory subsystem of Socket AM2 platform will surpass that of Socket939, AMD may use a new name again.
"Manila Core" of Low-end Market
In the future low-end processor market, AMD will still focus on Sempron series, and will transition from the current Socket 754 and Socket 939 interfaces to Socket AM2 interfaces. The Sempron core code of the new Socket AM2 interface is "Manila". We can think of it as a simplified version of "Orleans". The cache number is reduced to a quarter of the mainstream CPU, that is, 5 12KB L2. At the same time, it does not support security and virtual technology. However, the specifications supporting dual-channel DDR2 have not shrunk, and the time to market will of course be later.
Socket M2 Sempron processors will be listed first with 3500+, 3400+, 3200+ and 3000+ operating frequencies of 2.2GHz, 2.0GHz, 1.8GHz and 1.6GHz respectively. In addition, Socket M2 Sempron processor may also join the existing 2.4GHz 3600+ and 2.6GHz 3800+ products.