One: CPU frequency:

This is an index that novices are most concerned about, which refers to the CPU clock speed at which the CPU core works. How many megahertz is a CPU and how many megahertz is the "main frequency of CPU". I often hear people ask at school that the CPU of XXX Internet cafe is 2.66 g! XXX internet cafe is only 2G. Some people compare Celeron of 2.66G with P4 of 2.0G-2.66G, which is a sign of ignorance. There is no point in arguing with them. Although the main frequency is related to the speed of CPU, it is not absolutely proportional, because the running speed of CPU depends on various performance indicators (cache, instruction set, CPU bits, etc.). ) CPU pipeline (introduced below). Therefore, the main frequency does not represent the overall performance of CPU, but improving the main frequency is very important to improve the running speed of CPU. The calculation formula of main frequency is: main frequency = external frequency * frequency doubling.

Two: external frequency:

The external frequency is the reference frequency of CPU and even the whole computer system, and the unit is MHz. In early computers, the synchronous running speed between the memory and the motherboard was equal to the external frequency. In this way, it can be understood that the external frequency of CPU is directly connected with the memory, so that the two can run synchronously. For the current computer system, the two can be completely different, but the meaning of external frequency still exists. Most frequencies in computer systems are based on external frequencies, multiplied by a certain multiple, which can be greater than 1 or less than 1.

Three: frequency doubling: frequency doubling

The frequency multiplication of CPU is the frequency multiplication coefficient. There is a ratio relationship between the core working frequency of CPU and the external frequency, which is the frequency doubling coefficient. Theoretically, the frequency doubling is from 1.5 to infinity, but it should be noted that the frequency doubling is in the interval of 0.5. The multiplication of external frequency multiplication and frequency multiplication is the main frequency, so any one of them can improve the main frequency of CPU. There is no concept of frequency doubling. The main frequency of CPU is the same as the speed of system bus, but the speed of CPU is getting faster and faster, and the frequency doubling technology was born. The system bus can work at a relatively low frequency, and the CPU speed can be infinitely improved by frequency doubling. Then the calculation method of CPU main frequency becomes: main frequency = = external frequency x frequency doubling. In other words, frequency doubling refers to the multiple of the difference between CPU and system bus. When the external frequency is constant, the higher the frequency multiplication, the higher the CPU frequency.

There are so many main frequency factors. Now let's look at other "things" that affect CPU speed. Allow me to call it something and say that this function is current.

Four:

Assembly line:

Geography students should know this thing. According to the geography book in the second volume of Senior One, it is equivalent to a part of the public journey. Let me make a metaphor myself! For example, running and walking are divided into two stages, that is, the left foot, then the right foot, and then the cycle. The first stage can be said to be jumping with both feet, which is of course inefficient, right? . . . . . This is the assembly line of life. When you go out with your left foot, if you find a pile of shit in front of you, you must start all over again (you must take two steps at a time). This is the mistake you made after you went to the assembly line level. The CPU will be recalculated as soon as it comes out. . . . . Maybe I don't quite understand what I said. To quote others, in the process of manufacturing CPU, besides hardware design, there is also logic design, and pipeline design belongs to the category of logic design. For example, in the process of automobile production in an automobile factory, four production steps are completed by four teams: 1 team produces automobile chassis, two teams install engines on the chassis, and three teams install shells and tires on the automobile. The four groups do painting and glazing, which is called four-stage assembly line. Nowadays, big automobile manufacturers really follow similar assembly lines to improve production efficiency. Assuming that each step takes 65,438+0 hours, if we ask the 65,438+0 group to start producing the chassis of the next car immediately after completing the chassis of 65,438+0 cars, then the two groups will immediately put into the engine of the next car. The same is true for the work of three or four groups, so that a Mercedes or BMW will be produced every hour, which is equivalent to the sequential execution of CPU instructions. But what should be done to improve the production efficiency of the factory? Then we can divide each of the above groups into two groups to form an 8-level production line, so that each group (note that it is "group") can complete the work in half an hour, and a car will go off the assembly line every half an hour accordingly, thus improving the efficiency (it is not easy to understand here, please think about it carefully).

According to this truth, the pipeline of CPU is not difficult to understand, but it turns the production of automobile into the execution of program instructions, which are interlinked in principle.

Then it can be imagined that if the assembly line is lengthened, can the efficiency be improved? When people apply this idea to CPU design, they find that the pipeline is very inflexible because it is used to arrange instructions. Once an instruction at a certain stage is executed incorrectly, the whole pipeline will stop, find out the errors one by one, then empty the whole pipeline and reload the instructions. This will waste a lot of time and the execution efficiency is very low. In order to solve this problem, scientists have also adopted various prediction techniques to improve the accuracy of instruction execution, hoping to avoid the tragedy of emptying the pipeline while maintaining the long pipeline. This is often seen in Intel's advertisements, "The processor adopts advanced branch prediction technology ...". When you understand what I said above, you will know that it is so mysterious, but it is nothing more than that.

What I want to say is that a long pipeline will make it easy for CPU to reach a high working frequency, but how many of these 2G and 3G frequencies are really effective working frequencies? Moreover, the more stages, the longer the cumulative delay, because the working group will have signal delay when handing over the work. Although each delay is very short, the cumulative delay of the 20-stage or even 30-stage pipeline can not be ignored, thus forming a very funny situation. Pipeline technology improves the frequency of processor, but because of its own defects, it causes a great efficiency gap, which will offset the advantages. Therefore, high-frequency CPU will also bring high power consumption and high heat value.

In recent years, Intel's Pentium 4 processor has gone through three stages of development. The earliest Pentium 4 processor used the (William) core, with only 13 pipeline, and the general frequency was not 2G, so the speed was average. The second generation Pentium-4 processor uses a northwoog kernel, which has a 20-stage pipeline. Because the assembly line number was appropriate, the first mate increased the speed of Pentium -4, but it did not affect the execution. At that time, Pentium 4 2.4A was a classic product, which pushed AMD's Athlon XP series down, so Intel tasted the sweetness and soon introduced the Prescott kernel. This new core, which uses 365,438+0-stage assembly line, makes Pentium 4 reach a speed close to 3G, which is beyond AMD's reach, but people soon found that the actual operation efficiency of the new Pentium 4 is not as good as the old one. However, the frequency is so high, and the heat and power consumption are so great. With this new core, Intel has won the reputation of "high frequency and low energy consumption". At this time, AMD timely launched the "Athlon 64" series, with brand-new architecture, 20-stage assembly line, low heat and power consumption, and the most important thing is low frequency and high efficiency. Beat the new Pentium 4 in one fell swoop and get high praise. Intel also swallowed its own bitter fruit:

CPU cache:

CPU cache is a temporary memory between CPU and memory. Its capacity is smaller than that of memory, but its exchange speed is faster. The data in the cache is a small part of the memory, but this small part will be accessed by the CPU in a short time. When the CPU calls a large amount of data, it can be called directly from the cache without memory, thus speeding up the reading speed. It can be seen that adding cache to CPU is an efficient solution, so that the whole memory (cache+memory) becomes a high-speed storage system with both cache and memory. Cache has a great influence on the performance of CPU, which is mainly caused by the data exchange order of CPU and the bandwidth between CPU and cache.

The working principle of cache is that when the CPU wants to read a data, it first looks it up from the cache, and if it finds it, it immediately reads it and sends it to the CPU for processing. If it is not found, it will be read from the memory at a relatively slow speed and sent to the CPU for processing. At the same time, the data block where this data is located will be transferred to the cache, so that the whole data can be read from the cache later without calling the memory.

It is this reading mechanism that makes the hit rate of CPU reading the cache very high (most CPUs can reach about 90%), that is to say, 90% of the data that CPU reads next time is in the cache, and only about 10% needs to be read from the memory. This greatly saves the time for CPU to read the memory directly, and also makes CPU basically not need to wait when reading data. Generally speaking, the order of CPU reading data is to cache first and then memory.

The earliest CPU cache was a whole with low capacity. Intel has been classifying caches since the Pentium era. At that time, the cache integrated in CPU core was not enough to meet the needs of CPU, and the limitation of manufacturing process could not greatly improve the cache capacity. Therefore, there is a cache integrated with CPU on the same circuit board or motherboard. At this point, the cache integrated with the CPU core is called the first-level cache, while the external cache is called the second-level cache. The first-level cache is divided into data cache (D-Cache) and instruction cache (I-Cache). They are used to store data and execute instructions of these data respectively, and can be accessed by CPU at the same time, which reduces conflicts caused by contention for cache and improves processor efficiency. When Intel introduced the Pentium 4 processor, it replaced the instruction cache with a new first-level trace cache with the capacity of 12KμOps, which means that it can store 12K microinstructions.

With the development of CPU manufacturing technology, the secondary cache can also be easily integrated into the CPU core, and its capacity is increasing year by year. It is not accurate to define the first-level and second-level caches by whether they are integrated in CPU or not. Moreover, with the integration of secondary cache into CPU core, the large gap frequency division between secondary cache and CPU has also changed. At this time, working at the same speed as the main frequency can provide higher transmission speed for CPU.

Secondary cache is one of the keys to CPU performance. With the CPU core unchanged, increasing the capacity of the secondary cache can greatly improve the performance. However, the high-end and low-end CPU of the same core are often different in secondary cache, which shows the importance of secondary cache to CPU.

When the CPU finds useful data in the cache, it is called a hit. When there is no data needed by the CPU in the cache (this is called a miss), the CPU will access the memory. Theoretically, in a CPU with secondary cache, the hit rate of reading the primary cache is 80%. In other words, the useful data found in the first-level cache of CPU accounts for 80% of the total data, and the remaining 20% is read from the second-level cache. Because it is impossible to accurately predict the data to be executed, the hit rate of reading the secondary cache is also about 80% (the useful data read from the secondary cache accounts for 16% of the total data). Then some data will have to be called from memory, but this is already a fairly small proportion. At present, there will be a three-level cache in the higher-end CPU, which is designed for the data lost after reading the second-level cache. In a CPU with three-level cache, only about 5% data needs to be called from memory, which further improves the efficiency of the CPU.

In order to ensure the high hit rate of CPU access, the contents in the cache should be replaced according to a certain algorithm. A commonly used algorithm is the least recently used algorithm (LRU algorithm), which eliminates the least recently accessed rows. Therefore, it is necessary to set a counter for each line. LRU algorithm is to clear the counters of hit lines and add 1 to the counters of other lines. When replacement is needed, the data line with the largest count value of the line counter is deleted. This is an efficient and scientific algorithm, and its counter clearing process can clear some unnecessary data from the cache after frequent calls and improve the utilization rate of the cache.

In CPU products, the capacity of the primary cache is basically between 4KB and 64KB, and the capacity of the secondary cache is divided into 128KB, 256KB, 5 12KB, 1MB, 2MB, etc. There is little difference in the first-level cache capacity of products, while the second-level cache capacity is the key to improve CPU performance. The increase of secondary cache capacity is determined by the manufacturing process of CPU, and the increase of capacity will inevitably lead to the increase of the number of transistors in CPU. To integrate a larger cache on a limited CPU area requires a higher manufacturing process.

Front-end bus:

The front-end bus is the data channel between the processor and the north bridge chip or memory control hub of the motherboard, and its frequency directly affects the speed of CPU accessing memory; BIOS can be regarded as a software that remembers computer-related settings and can be used to adjust related settings. The BIOS is stored in a chip on the motherboard. The name of this chip is COMS RAM. But like ATA and IDE, most people confuse them.

Because the motherboard directly affects the performance, stability, function and expansibility of the whole system, its importance is self-evident. The purchase of the motherboard seems simple, but there are many things to pay attention to. Pay attention to the chipset, working materials, functional interfaces and even ease of use when purchasing products, which requires a thorough understanding of the motherboard before choosing a satisfactory product.

A bus is a set of transmission lines that transmit information from one or more source components to one or more destination components. Generally speaking, it is a common connection between multiple components, which is used to transfer information between components. People often use MHz to describe the bus frequency. There are many kinds of buses. The English name of the front-end bus is Front Side Bus, which is usually expressed by FSB. It is a bus connecting CPU and Northbridge chip. The front-end bus frequency of computer is determined by CPU and Northbridge chip.

The CPU is connected to the Northbridge chip through the front-end bus (FSB), and then exchanges data with the memory and graphics card through the Northbridge chip. The front-end bus is the most important channel for CPU to exchange data with the outside world, so the data transmission ability of the front-end bus plays a great role in the overall performance of the computer. If there is no fast front-end bus, no matter how strong the CPU is, it can't obviously improve the overall speed of the computer. The maximum bandwidth of data transmission depends on the width and transmission frequency of all data transmitted at the same time, that is, data bandwidth = (bus frequency × data bit width) ÷8. At present, the front-end bus frequencies that can be realized on a PC are 266MHz, 333MHz, 400MHz, 533MHz and 800MHz. The greater the front-end bus frequency, the greater the data transmission capacity between CPU and Northbridge chip, and the better the function of CPU can be brought into play. At present, CPU technology is developing rapidly and the operation speed is increasing rapidly. A large enough front-end bus can ensure enough data for CPU, while a low front-end bus will not supply enough data for CPU, thus limiting the performance of CPU and becoming the bottleneck of the system.

The bus speed between CPU and Northbridge chip more substantially represents the data transmission speed between CPU and the outside world. The concept of external frequency is based on the oscillation speed of digital pulse signal, that is to say, 100MHz external frequency means that digital pulse signal oscillates 1 100 million times per second, which has a greater impact on the frequency of PIC and other buses. The two concepts of front-end bus and external frequency are easily confused, because for a long time before (mainly before and just after Pentium 4 appeared), the front-end bus frequency and external frequency were the same, so they were often called external frequency directly, which eventually caused such misunderstanding. With the development of computer technology, people find that the front-end bus frequency needs to be higher than the external frequency, so QDR(Quad data Rate) technology or other similar technologies are adopted to achieve this. The principle of these technologies is similar to 2X or 4X of AGP, which makes the frequency of the front-end bus 2 times, 4 times or even higher than the external frequency. From then on, people began to pay attention to the difference between front-end bus and external frequency.

CPU process:

Refers to the width of internal component connection lines when CPU is produced on silicon material, which is generally expressed in microns. The smaller the micron value, the more advanced the manufacturing process, the higher the frequency that CPU can achieve, and the more transistors that can be integrated. At present, both Intel P4 and AMD XP have reached the manufacturing process of 0. 13 micron, and will reach the manufacturing process of 0.09 micron next year.

From the above, we know the logical structure of CPU and some basic technical parameters. This paper will continue to fully understand the relevant technical parameters that affect CPU performance.

Briefly introduce CPU types

For a computer system, the role of CPU is as important as the heart is to us. When we buy a computer, we always ask first, is it 486 or 586, 100 or 300, MMX or 3D-Now! These all refer to the indicators of CPU, and the core role of CPU in the whole microcomputer system is enough to be used as a standard for dividing CPU levels, which makes it almost synonymous with various levels of microcomputer. We can say that the performance of CPU can roughly reflect the performance of our microcomputer system, which is obviously very important for our choice.

1, what is CPU?

CPU is the abbreviation of "Central Processing Unit" in English, which is literally translated as "Central Processing Unit" in Chinese. The main function of CPU is operation and logical operation, and its physical results include logical operation unit, control unit and storage unit. The logic operation and control unit includes some registers, which are used to temporarily store data during the processing of data by the CPU. Here, we don't need to understand the complex principle of CPU. We only have some necessary knowledge about it from the choice of performance parameters, which is very helpful for understanding, purchasing and configuring computers.

2. Main performance indicators of 2.CPU:

Main frequency: that is, the clock frequency of the internal core of CPU, generally in megahertz (MHz). This is a parameter that we are most concerned about, and we usually call it 233, 300, 450, etc. For the same kind of CPU, the higher the main frequency, the faster the CPU speed and the higher the performance of the whole machine. Different kinds of CPU can't be directly compared with the main frequency because of their different internal structures. The actual performance of CPU with high main frequency is also related to the external frequency and the size of cache, while the CPU with special instructions depends on the optimization degree of software to some extent.

External frequency and frequency multiplier: External frequency is the external clock frequency of CPU. The relationship between CPU main frequency and external frequency is: CPU main frequency = external frequency × frequency doubling. The external frequency is provided by the computer motherboard. The external frequencies of 486 are generally 33MHz and 40MHz. The external frequency of Pentium motherboards is generally 66MHz, and some motherboards support 75 MHz and 83MHz respectively. At present, Intel's latest chipset 440BX can use 100MHz or even higher clock frequency. In addition, some non-Intel chipsets, such as MVP3, MVP4 and APPROPRO from VIA, have also started to support the external frequency of 100MHz. Some motherboards can be used more than 1/3 stably because of their sophisticated technology and advanced technology, and become the first choice for overclocking enthusiasts. Intel's next-generation motherboard chip will support the external frequency of 133MHz, and AMD's K7 will even use the external frequency of 200MHz.

Briefly introduce CPU types

Internal cache (L 1 Cache): A high-speed SRAM, packaged with CPU***, is used to temporarily store some recent instructions and data during CPU operation. The access speed is the same as the main frequency of CPU (generally called full speed), and the capacity of L 1 cache is generally in KB. L 1 cache works at full speed. The larger its capacity, the easier it is for the most commonly used data and results to enter the CPU for operation as soon as possible. When CPU is working, the less times it exchanges data with L2 cache and memory with slow access speed, the faster it can run compared with computer. 486 is much faster than 386 because it integrates the internal cache. The earliest 486 generally had 1K ~ 8K L 1 cache. At present, the L 1 cache of Pentium II is generally 32K, and the chips of Cyrix and AMD are more than 64K.

L2 cache: a cache integrated outside the CPU. The total capacity of L2 cache is128k ~ 2m. The larger the capacity, the higher the comprehensive performance of the system. General L2 cache runs at half the system frequency or CPU frequency. Later, L2 and CPU used by Pentium Pro processors run at the same frequency. Due to the low chip yield and high cost, the L2 cache of Pentium II runs at half the CPU frequency, but the capacity is increased to 5 12K. At present, Xeon processors use full-speed L2 cache, and the capacity is increased to 5 12K to 2M to improve performance. The performance of Celeron processor without cache is greatly degraded.

MMX technology: short for Multimedia Extensions Instruction Set. MMX is a new technology adopted by Intel to enhance the application of Pentium CPU in audio-visual, graphics and communication fields. This technology adds 57 brand-new MMX instructions to the CPU, and also increases the L 1 cache in the CPU chip from the original 16KB to 32KB( 16K means life+16K data). Therefore, when MMX CPU runs programs containing MMX instructions, its multimedia processing ability is improved by about 60% compared with ordinary CPU. Even if the program without MMX instruction is used, the performance can be improved by about 15%. MMX has become a basic criterion for selecting CPU. At present, CPU basically has MMX technology, besides Pentium P55C (Pentium MMX) and Pentium II CPU, there are K6, K63D (K6-2), MII, 6X86MX, IDT C6 and so on. Cpus that do not support MMX instructions can be ignored.

3D instruction technology: MMX instruction solves the bottleneck of multimedia operation, but it only speeds up integer operation, and is powerless for 3D graphics processing and games that require large-scale floating-point operation. In view of the increasing demand for 3D processing, supporting 3D instructions will be as important as supporting MMX instructions. At present, AMD is the only CPU that supports 3D instructions, using 3D-Now! CPU technology can greatly accelerate the speed of three-dimensional processing, thus bringing games and graphics processing into a new realm. Intel's upcoming MMX2 instruction set will be more powerful, and these instruction sets must rely on the optimization support of software to give full play to the performance of CPU. Manufacturing technology: the unit is micron. At present, the manufacturing process of CPU is generally 0.35 micron, and the latest PII and K6-2 can reach 0.28 ~ 0.25 micron. In the near future, the manufacturing process of CPU can reach 0. 18 micron or even 0. 13 micron. The micron level of CPU directly determines the limit frequency of CPU. The working frequency of 0.35 micron CPU is generally less than 250MHz, while the copper core process chip with 0. 13 micron can work stably at 1000MHz.

Types of C P U

cpu

There are three kinds of CPU in the market: Intel, AMD, CYRIX/IBM.

INTEL is a well-known leader, CPU performance is of course the best, but the price is also the highest, especially for products with high frequency. For applications that require high performance, such as graphics processing and 3D games, Intel CPU is the best.

The MMX CPU of CYRIX/IBM is extremely low in price, the best in integer performance and weak in floating-point operation, so it occupies a high market share in general low-grade commercial and household assembly machines.

AMDK5 has been neglected, K6 is popular, and its integer and floating-point performance is between INTEL and CYRIX, but its price is not far from that of Pentium MMX, so it seems that there are not many buyers in the retail market.

The CPU market staged the Romance of the Three Kingdoms. Intel, AMD and CYRIX are strong and weak, but Intel stands out and the situation is changing. AMD's K6+ and K63D technologies are developing rapidly. 1998 will release a batch of new CPUs and motherboards to compete with INTELPII for 30% market share. After CYRIX was acquired by the national semiconductor company, it had a big backer. Recently, the technical details of several future high-performance CPUs have been announced, and floating-point operation has become its strength. In 1998 and 1999, the CPU market will enter through a train.

Types of C P U

In the choice of CPU, we have never been faced with so many choices as now. A monopoly Intel, the most powerful challengers AMD, Cyrix, and the new IDT add up to nearly 25 kinds of CPU. In order to let everyone choose CPU better, we made an evaluation of these 25 kinds of CPU.

Selection of processor

In order to make the best processor choice, you must first determine what you want it to do. If you often use commercial programs and you care about the price of processors, then non-MMX chips (Pentium, K5, 6X86) are your best choice, but this level of chips will soon be out of the market. Although the current MMX CPU can't give you great benefits, the updated MMX CPU has also been enhanced in other aspects, so that they can get higher performance when running all programs, and now the price of MMX CPU has dropped to a very reasonable position, MMX CPU should be your best choice. In the sixth generation chip, if you often use the same program as the one running in Winstone, and you also want to have MMX technology, then AMD's K6 and Cyrix's MII are more valuable. K6 and MII of the same grade can provide Winstone scores very close to PentiumII, and their prices are more than half cheaper than PentiumII. However, their MMX performance is much worse than PentiumII (it will be strengthened in their next-generation CPUKII and Mxi), but it is much better than the CPU without MMX. After testing, PentiumII has been proved to be the best. It has shown its due performance in commercial and high-end applications, and it has also been proved to be unparalleled in MMX and floating-point testing. If you are a high-end user or a 3D game fan, don't be depressed. Whether it is 233MHZ or the latest 450MHZ, PentiumII is your best choice. Because PentiumII's dual-bus technology and powerful floating-point performance will make you get the best performance in your application.

Processor introduction

AMD's K5 is the first x86 CPU independently produced by AMD. Due to the problems in development, K5 delayed its launch date and was always limited to entry-level CPU. By the end of 1997, K5 was eliminated from the market. The following is the specification table of K5.

< tbody > processor performance clock speed

Bus speed (MHz)

(MHz) PCI bus speed

(megahertz) multiplier

K5 75 75 50 25 1.5

K5 PR 90 90 60 30 1.5

K5 PR 100 100 66 33 1.5

K5 PR 120 90 60 30 1.5

K5 pr 133 100 66 33 1.5

K5 PR 166 1 16.66 66 33 1.75 & lt； /TBODY & gt；

The performance of K5 is better than that of Pentium with the same frequency, so AMD adopts PR rating system for K5.

However, AMD pinned its hopes on his K6.

K6 was modified from the original NexGen 686 (AMD acquired NexGen), thus gaining NexGen's technical strength in chip research and development. K6 has MMX technology and more on-chip advanced caches (32K instructions and 32K data). Compared with K5, K6 can process more instructions in parallel and run at a higher clock frequency.

According to the commercial application test under Windows95, AMD's K6 is positioned at a stronger level than Pentium MMX, which wants the same speed. This positioning has been confirmed by our test. The performance of K6/233 based on AMD has been quite close to PentiumII/233 in the commercial test of Windows95, but it still lags behind by several percentage points. But it can easily beat Pentium MMX/233, and its score is 7% to 9% higher than the commercial Winstone 97 of the corresponding Intel chip. Because K6 has a larger L 1 cache, with the increase of frequency, it can achieve more significant performance improvement than Pentium MMX. K6 is slightly behind in running applications that need to use MMX or FP (floating point instruction). Because K6 has fewer chips designed for these functions than Intel. For example, AMD's MMX unit can only process one instruction at a time, while Intel's MMX unit can process two instructions.

The performance of FP and MMX depends on reflection time and throughput. Reflection time is the time required for an operation from the beginning to the end. This is the decisive factor of independent calculation, and the result is necessary for the next step. Throughput indicates how fast the new operation is processed; In a multi-pipeline multiplier or floating-point unit, two or more operations can overlap; Therefore, the rate is accelerated, but at the same time the reaction time is increased. All CPUs of Intel have all pipelined MMX and floating-point units, so a new operation can be started every clock cycle-even if the operation result may not be available until several clock cycles later. When performing a long series of calculations, as is common in some multimedia applications, throughput plays a more important role than reflection time.

AMD's K6 has shorter response time when performing some MMX operations than Intel's processors, but their throughput is the same when processing a single operation. Shorter response time can't make up for its inability to process two MMX instructions at the same time. (However, Intel's MMX unit has only one multiplier and shifter, so it can't perform MMX and floating-point operations at the same time. In addition, only one MMX instruction can access memory or shaping register at the same time. ) When running some instructions on FP, K6' s response time is shorter than Intel's, but only one operation can be started every two clock cycles, and Intel's chip can start a new operation every clock cycle. The result is that when performing many FP operations, the throughput of K6 can only reach about half that of Intel processors.

These weaknesses are manifested in the poor performance of K6 in ZD3 WinMark 97, synthetic floating point, Photoshop and other tests. In the above tests, K6/233 is much slower than Pentium II/233, and sometimes even slower than Pentium MMX/233. Especially in the test of ZD 3DWinMard. So for graphic design or 3D game fans, AMD's K6 may not be your choice, at least in KII (with 3DNow! Before the release of K6), AMD didn't provide more information about KII, so for the time being, we only know that KII has 3DNow! Technology-instruction system used to enhance the floating-point computing ability of CPU. In addition, KII will be produced in a 0.25 micron process, with an external frequency of 100MHZ and a release frequency of 300MHZ. Later, we will also see the CPU named KII+ by AMD, which is an enhanced version of KII, integrating 5 12K L2 cache into the CPU, and the release frequency will be at the clock frequency of 350MHZ.

IBM company and company

After the merger of Cyrix and National Semiconductor Company of the United States, it finally has its own chip production line, and the finished products will be increasingly perfect and complete. Cyrix's 6x86 is its Pentium-compatible processor. It uses the nominal CPU of rated public relations. Its PR- 133CPU runs at 120MHZ, but its performance is the same as Pentium 133. But what about its hair? /ca & gt；