CPU architecture isCPUThe manufacturer sets a specification for CPU products belonging to the same series. The main purpose is to distinguish the important marks of different types of CPUs.The CPU classification on the market is mainly divided into two camps, one isintel、AMDLeadedComplex instruction setCPU，The other is based onIBM、ARMLeadedReduced Instruction SetCPU。The architecture of two different brands of CPUs is also different. For example, the CPU of Intel and AMD isX86 architectureWhile IBM's CPU isPowerPCArchitecture, ARM isARM architecture。
Title
Processor architecture
Author
Intel Asia Pacific R&D Co., Ltd
press
Shanghai Jiaotong University Press
Publication time
January 1, 2011
Pricing
29.50 yuan
Folio
16 ON
ISBN
nine trillion and seven hundred and eighty-seven billion three hundred and thirteen million sixty-eight thousand six hundred and ninety-nine
Foreign name
Processor architecture
Pinyin
chǔ lǐ qì jià gòu
Type
CPU product specification
Main purpose
Important indicators for distinguishing different CPUs
Core ArchitectureMeromThe processor is really powerful.In many tests, the best proof is that the T7200 with a frequency of 2GHz can defeat the T2700 with a frequency of 2.33GHz.But you also noticed that in theMobile platformMerom is powerful, but it doesn't surprise you.Although better thanYonah, but the range is not large, and someTest itemThe T7200 with a lower frequency is also lost to the T2700.Therefore, it may be possible toCore microarchitectureThe advantage of the E6300 is not as brilliant as that of the desktop platform - an E6300 with the lowest frequency can also completely annihilate the high frequencyPentium D。The reason is that Yonah itself is better than NetBurst. Besides, CoreMicroarchitectureThe Yonah micro architecture itself is improved, and it is reasonable that the performance will not form too big a contrast.
Core microarchitecture is Intel'sIsraelThe design team improves the new generation of micro architecture based on Yonah micro architecture.The most significant change is to strengthen each key part.To improve the internal data of the two coresExchange efficiencySharedL2 CacheDesigned, 2 cores share up to 4MB of L2 cache.Its core adopts a short 14 level effective pipeline design, and each core has a built-in 32KB level 1Instruction CacheLevel 1 with 32KBData cacheData can be directly transferred between the first level data caches of the two cores.4 groups built in each coreInstruction decodingUnit, supportMicroinstructionIntegration andMacro instructionFusion technology, eachClock cycleIt can decode up to 5 X86 instructions and has improvedBranch predictionFunction.5 per coreexecution unitSubsystem, the execution efficiency is quite high.Join pairEM64TAndSSE4 instruction setSupport.Because of the support for EM64T, it can have more memoryAddressing space, which makes up for Yonah's shortcomings. After the popularization of the new generation of memory consuming major user - Vista operating system, this advantage can make the core micro architecture have a longerlife cycle。It also uses Intel's latest five new technologies to improve efficiency and reduce power consumption, including:Power managementFunction;supportHardware virtualizationTechnology andHardware antivirusFunction;Built inDigital temperature sensor；Provide power report, temperature report, etc.Especially theseEnergy saving technologyThe adoption of is particularly significant for mobile platforms.
Core processors face different consumer groups based on the core architecture. There is a small division of labor in the core processors. Conroe for desktop computers, Merom for laptops, and WoodCrest for servers are all based on the core architecture.
IntelProcessors include Core series desktop, mobile, andXeon processor, evenEmbedded processor, all will enter one after another32nm process, gradually replacing the current 45nmProcess。along withCESThe pace is approaching. Intel has disclosed that it will publish many products on CESCore i3、i5Desktop and notebook processors, including laptopArrandaleAnd desk electricityClarkdaleThe 32nm process has been adopted successively, emphasizing smaller volume and power consumption design.On December 23, 2009, Intel revealed that the embedded Xeon processor to be launched in the first quarter of 2010 will also adopt a new process.Compared with the 45nm process at the end of 2008, the 32nm process launched at the end of 2009 uses the second generation high-k metal gatetransistorAnd infiltratingMicrophotography(abstraction lithography), strengthen the electronic control tube used inside the processor, which is also 30% smaller than the size of the 45nm process, simplifyingsystem design。According to Intel's blueprint, in the first quarter of 2010, the 32nm process will be launched for the embedded market. The embedded Xeon processor code named Jasper Forest is 30% to 70% more efficient per watt than the old process processorPCI2.0 andI/O virtualizationAbility.With the introduction of Clarkdale, an enterprise server Xeon processor, the entry-level Xeon 3000 processor closely related to the high-end desktop market will also enter a new 32nm process in 2009.
For 2009Nehalem-The Xeon 5000 with EP architecture, although the same Nehalem architecture is used, will be launched in the first half of 2010 with a new 32nm processWestmere-EP processor.The Xeon 7000 processor, which originally provided six cores, will also launch Nehalem EX with up to eight cores in the first half of 2010, and the Westmere EX will also enter a new process in the second half of 2010.
Compared with Intel's new manufacturing process in 2010,AMDIt is expected to enter the 32nm process by 2011, when the newBulldozerCore architecture design, including Interlagos with efficiency level of 12 to 16 cores, and Valencia with energy efficiency of 6 to 8 cores.
The 8-core CPU can't match the current motherboard, so it's impossible to make publicity. The cheapest 8-core CPU should be SONY PS3CELL, with 8 core floating-point performanceCore dual coreThe four cores are not popular now, and AMD INTEL will not rush to mass produce their eight core CPUs. It can be said that the current four cores of INTEL just encapsulate two core cores in one core, and there is no direct communication between the two cores. AMD has produced the real four cores, but it cannot become the mainstream if it is not sold well.To sum up, five years later, 4-core can basically replace the current dual core and become the mainstream, while 8-core and even 16 core CPUs will become high-end products at that time!
Development overview
1、X86
Although the above statement is divided by processor architecture, there are mainly four terms in terms of terms, namely IA-32, IA-64x86- 32, x86 - 64, but they actually fall into two categories. IA - 32 and x86 - 32 belong to x86, namely Intel's 32-bit x86 architecture. x86 - 64 is AMD's latestAthlon 64The new architecture is adopted in the processor family, but this processor infrastructure is IA-32 (because Intel's x86 architecture has not applied for patent protection, most processor manufacturers have to adopt this x86 architecture in order to maintain compatibility with Intel's mainstream processors). Only some extensions have been made on this architecture to support the application of 64 bit programs,Further improve the computing performance of the processor.X86-64 Compared with Intel's 64 bit server processor productsItaniumandItanium 2The biggest advantage of the series processor products is that they are fully compatible with the previous 32-bit x86 architectureapplication programTo protect users' previous investment;Intel Itanium and Itanium 2 series processors need to be compatible with previous 32-bit programs through software or hardware.
Because of this, we will see that IA-32, x86-32 and x86-64 are all of the same type and belong to x86 architecture.For example, Intel's 32-bit server Xeon（xeon ）The processor series, the full range of AMD, and the full range of VIA processor products belong to the x86 architecture.
The IA-64 architecture is designed by Intel to comprehensively improve the previous IA -32-bit processorIs Intel and HpCo developmentThe 6-year 64 bit CPU architecture is specially designed for serversmarket development It abandons the previous x86 architecture and believes that it has seriously hindered the improvement of processor performance.Its initial application is Intel Itanium（itanium ）Series server processors. The latest Itanium 2 series processors in 2009 also use this architecture.Because it can not well solve the compatibility with the previous 32-bit applications, its applications are subject to greater restrictions. Although Intel has taken various soft and hard methods to make up for this deficiency, with the full investment of AMD Operon processors, the prospects of these two processors of Intel's IA-64 architecture are not optimistic.
3、RISC
In addition to the two types of IA server processors described above, there is also a mainstream processor architecture, also known as RISC (in fact, it is based onProcessor InstructionsimplementMode divisionType of).It is stillIBM、SUNandHPEtc.However, in recent years, because this processor architecture standard is not completely unified and the development and application of processors are very slow, most of the original high-end server market has been divided by IA architecture, which has gradually declined.At present, even these server manufacturers have begun to give up and turn to IA to launch more and moreIA architecture serverTo ensure survival.
The main server processors using this architecture are IBM Power4, Compaq Alpha213 64, HP PA-8X00, Sun UltraSPARC IIISGIOfMIPS64 20Kc, etc.
4、Intel
brief introduction
Intel CommonServer CPUClassification.processortechnological development It is really changing with each passing day. The previous generation of products has not been distinguished by everyone, and will be replaced by the next generation of products soon.Here, according to some personal understanding, I will help you make a division.
1、 Xeon
Dual way of all Intel IA architectures,Four way server, all adopt Xeon CPU, which is based onX86 architectureA server specific CPU.Earlier processor names were represented by numbers and ended with "86", includingIntel 8086、80186、80286、80386、80486、80586、PentiumSo its architecture is called "x86". So far, all Xeon products, including dual core and quad core products, are based on the X86 architecture.
2、 Itanium
Itanium processorAlso known as IA-64-bit processor, it is a pure64 bit processorProduct with 64 bit addressing capability and 64 bit wideregisterIts series of features, such as EPIC instructions, are designed for the most demanding computing and enterprise requirements.For the most performance critical enterprises or applications requiring high-performance computing support (including electronicTransaction securityHandlingVery large database, computer-aided mechanical engineCutting-edge scienceComputing, etc.), Itanium processor well meets the requirements of users.
five hundred and sixty-four thousand six hundred and forty-six
seven hundred and sixty-eight thousand six hundred and seventy-eight
nine hundred and seventy-eight thousand nine hundred and seventy-eight
nine hundred and seventy-eight thousand nine hundred and seventy-eight
eighty-seven thousand nine hundred and eighty-seven
nine hundred and eighty thousand eight hundred and ninety-eight
4/8/16
eight thousand seven hundred and ninety-seven
6/8/12/18/24
8/12/18/24
Front end bus (MHZ)
1066/
one thousand three hundred and thirty-three
one thousand and sixty-six
one thousand three hundred and thirty-three
667/
one thousand and sixty-six
1066/
one thousand three hundred and thirty-three
1066/
one thousand three hundred and thirty-three
1066/
1333/
one thousand and six hundred
1333/
one thousand and six hundred
667/800
one thousand and sixty-six
400/533
400/533/667
Power consumption (W)
sixty-five
ninety-five
ninety-five
95/130
40/65/80
50/80/120
65/80
80/120/150
95/150
80/130
75/104
75/104
Dual core
√
?
?
√
√
?
√
?
√
?
√
√
Quad core
?
√
√
?
?
√
?
√
?
√
?
?
Hyper Threading
×
×
×
√
×
×
×
×
√
×
√
√
64 bit operation
EM64T
EM64T
EM64T
EM64T
EM64T
EM64T
EM64T
EM64T
EM64T
EM64T
Pure 64 bit
Pure 64 bit
3、 Processor comments
1. First lookOne wayProcessors, including Xeon3000, 3200, and 3300 series, of which 3000 and 3200 series single channel processors all useCore MicroarchitectureThe performance and power consumption are very ideal. You can choose the main frequency, dual core or quad core according to the application.The other 3300 series uses the latest45nmmanufacturing process, adopts enhanced Core micro architecture, with stronger performance and lower power consumption.
2. Dual processor, Xeon5000 series, with high power consumption and poor performance, has basically disappeared;The 51005300 series began to use the Core microarchitecture, with excellent performance and power consumption. It can be said that it is a super successful processor product of Intel. Compared with the previous generation of processors, the performance has been improved several times, and the power consumption has been reducedcompetitorThere is no product that can compete with it.The newly launched 52005400 series has adopted the 45nm manufacturing process and the enhanced Core microarchitecture. Its performance is 20% higher than that of the 5100 and 5300 series on average, and its power consumption is reduced by nearly 38%. What's more, the price is still very low, which is the best choice for server CPUs at this stage.
3. Multiple channelsxeon On Intel's official list, Xeon71007300 processors can support 32 processors in a single systemdomestic marketOnly 4-way Xeon servers can be seen frequently.The Xeon 7100 processor did not use the advanced core micro architecture at that time, so the four 7100 series CPUs added up to two 5300 series dual processorsRun fast, and the price is still high, so it is not recommended to use it. In addition, Xeon7100 will soon disappear in the market.The new Xeon7300 series is a very excellent multi-channel to strong CPU, which uses a core micro architecture. Each CPU has four cores. If four CPUs are combined together and large capacity memory is used, the performance will be very strong, enough to meet the computing requirements of high performance and large amount of data.
4. Itanium processors. In fact, the main competitors of Itanium processors are high-end brands such as IBM and SUNMinicomputerCPU，If you have been using high-end minicomputers, such as those installed with IBM Power CPU, I think it is necessary for you to know about Itanium and the new generationOpennessMaybe you will find that the high stability and high performance of the high-end CPU products are not necessarily high cost.In addition, Itanium can also bring you unexpected results in some scientific operations.
5、CORE
In the first ten days of March 2006, IntelSan Francisco Held the spring of 2006IDFIntel Developer Forum.At this IDF conference, there was a focus of attention: Intel announced that the next generation of processors will use the Core microarchitecture.This also made the 2009 IDF Conference the most exciting one in recent years.In the opening keynote speech of the IDF conference in the autumn of 2008, Paul Otellini, the executive director of Intel, pointed out that the focus of future processor technology development will be "Performance per Watt".The theme of this IDF conference is more clear: Power Optimized Platforms - closely related to Core microarchitecture.According to Intel, processors with the new Core microarchitecture will make great strides in integer performance and commercial computing, and will certainly surpass the products of competitor AMD.What's more wonderful is that the core microarchitecture with such strong performance will significantly reduce its power consumption compared with its predecessor, which perfectly reflects the theme of this IDF conference.
Core microarchitecture is located by Intel in IsraelHaifaOfR&D teamResponsible for design.As early as 2003, the Israeli team had designedlow power consumptionOfBaniasThe processor is famous all over the world, and the Core microarchitecture is their latest masterpiece after Yonah microarchitecture.Core microarchitecture has appeared in Intel's plan for a long time. As early as the summer of 2003, Intel vaguely mentioned that it was originally scheduled to beCentrinoThe third generation Napa platform of the platform later and fourth generationSanta RosaThe processor used by the platform.Unexpectedly, due to the failure of NetBurst microarchitecture, the Core microarchitecture was changed by Intel and pushed to the foreground. It was endowed with the ability to replace NetBurst microarchitecture, unified desktop, mobile and server platformsHistorical mission。
As the new flagship of Intel, Core microarchitecture hasDual core、64bitInstruction set4 TransmittedSuperscalar architectureandOut of order executionMechanism and other technologies, using 65nm manufacturing process, supporting 36bit physical addressing and 48bitvirtual memoryAddressing, supporting all IntelExtended instruction set。Each core of the core microarchitecture has a 32KB level 1 instruction cache, a 32KB dual port level 1 data cache, and then the two cores share a 4MB shared level 2 cache.The highest frequency of Core microarchitecture release in 2009 will be 3.33GHz of Conroe XE.Each product has its own highestTDP: Merom up to 35W, Conroe up to 65W, Woodcrest up to 80W.In addition, low-power versions can also be provided for different customer requirements.For example, the low voltage version of Woodcrest will be positioned atblade system The TDP can be as low as 40W by reducing the frequency.
Intel claims that the Core microarchitecture has 14 levels of "effective" pipelining.With Banias in the same design team, the Core microarchitecture has only 14 levels of integer pipeline, which is not surprising.But what exactly is the 14 level "effective" pipeline?
In the past few years, several concepts about pipeline series have often been confused.First of all, let's clarify that the "number" and "series" of pipeline are completely different concepts.A series of functional units that can completely execute various instructions form a "one" pipeline.As for the pipeline series, it can be simply understood that in the traditional sense, the functional units contained in a pipeline can be divided into multiple parts, which can be divided into several parts. This pipeline is called "several levels".Then let's understand the definition of "effective pipeline", which is also easy to misunderstand in the past.In short, the so-called effective pipeline means that when branch prediction errors occurRe executeThe pipeline number of stages.For processors with NetBurst microarchitecture,Willamette、NorthwoodAndPrescottThe effective pipeline levels of the core are 20, 20 and 31, while the original P6 microarchitecture processor is 10.
However, for modern X86 processors, which generally use out of order execution, the effective pipeline level does not represent the pipeline level in the true sense.The NetBurst microarchitecture processor is just the trace creation process of the Trace Cache, which has at least 10 levels;The complete pipeline level of P6 microarchitecture should be 12 to 15 (the effective pipeline level 10 plus the Retire action after the instruction is executed, and the possible Reorder Buffer delay).As the disordered execution engine becomes more and more complex, the concept of pipeline level of X86 processor becomes increasingly ambiguous.In other words, the real pipeline level of Core microarchitecture is not only 14.
The comparison between the 14 level effective pipeline of the Core microarchitecture and the 31 level effective pipeline of the Prescott core is also only a reference.It is not convincing enough to assert that the core microarchitecture can only reach a very low frequency based on the comparison of this number.The existence of Conroe XE 3.33GHz processor has surprised many users who believe this statement.In fact, some players have claimed that the Conroe processor can reach a frequency of more than 4GHz in air-cooled conditions.It remains to be seen how high the frequency of core microarchitecture can reach.
We transliterate Core as Core, which is the micro architecture that Intel's next-generation processor products will uniformly adopt, while Conroe is just aCore microarchitectureCode of Intel's next-generation desktop platform products.In addition to the Conroe processor, the Core microarchitecture also includes the mobile platform processor code named Merom and the server platform processor code named Woodcrest.Processors with Core will be named uniformly.Since the previous generation processor product with Yonah microarchitecture was named Core DuoPrevious generationIntelDual Core Processor Distinguish between Intel's next generation desktop processor Conroe and the next generationNotebook processorMerom will be uniformly called Core 2 Duo.In addition, Intel's top desktop processor is namedCore 2 ExtremeTo distinguish it from mainstream processor products.
The total number of Conroe/Mesom released this time is 10, of which 5 models starting with E and X are for desktops, and 4 models starting with T are for notebooks.
Intel initially released core microarchitecture processors, including E6000 desktop series, T7000, T5000 mobile series, and E6000 series processorsfsb 266MHz,Front end bus frequency1066MHz with 2MB (E6300E6320, E6400) or 4MB（E6600E6550, E6700) second level cache for high-performance market;The E4000 series external frequency introduced later is relatively low, 200MHz,Front end bus800MHz, lower than E6000 series, and the release time will be postponed to the first quarter of 2007.exceptNormal EditionIn addition to Conroe, Intel will also release Conroe XE processor to replace the existing flagship productPentiumXE -- X6800.
Although the front end bus of Conroy on the desktop platform is 1066MHz, the front end bus of Merom, the leading mobile processor, this time is 667MHz (Merom processors were originally processor products on the next generation mobile platform Santa Rosa, but now we have to bring Merom processors to the market before the launch of Santa Rosa platform, and can be successfully implanted on the current Napa platform.For Intel 945ChipsetIn order to adapt to Intel 945 chipset, its front-end bus still retains the 667MHz front-end bus design.In the future, the front-end bus of Merom processor on Santa Rosa platform will be changed to 800MHz.This scenario is different from the 400MHzDothanThe practices for adapting to the Intel 855 chipset are very similar).The L2 cache is increased to 4MB (the low-end T5000 series is still 2MB), which means that more waiting can be stored in the cacheProcessing data, reducing the space between processor and memory and peripheral devicesdata transmissionBottleneck ofhit rateAnd greatly improve the implementation efficiency.
Processor architecture
As Yonah processor on Napa platform is replaced with Merom processor, it also means that IntelMobile ProcessorStart Entry64-bit In the era of dual core technology, Yonah, as the first hero of dual core mobile processors, will begin to fall behind
Book information
Announce
edit
content validity
Processor Architecture consists of five chapters, fromCommand system、CPUThe composition, new CPU technology, CPU instance and other aspects of the processor architecture technology and application development are introduced in detail."Processor Architecture" combines theory with examples, which is easy to understand and suitable for most computersProfessional studentsRead and learn with IT beginners.
