Introduction
Once upon a time, we judge that the standard with high computer performance is only the size of the processor product, and the high frequency. The number of digital represents the faster computers. For example, 80286 is faster than 8088 and 8086, but 80386 is faster than 80286, while 80486 is the fastest. But time, the current computer world has been different from more than ten years ago. So let's take a look at the current processor today.
Compared with the previous order of processor products, the processor performance is judged, the standard, the standard, is now joined, and the processor product name, model name, core name, and architecture. To determine the performance of the processor, it is not a simple matter through these complicated technical standards. Of course, you can learn about the specific or a few processors through some media, however, this has some side. Today we have to do a detailed list of processors launched in the past 7 years, I believe this can help you get a better understanding of the processor while purchasing processors in the future. There is a number in my heart.
Since the current processor update is extremely fast, in this assessment, we exclude the products before the Intel Pentium II processor, the AMD Athlon processor. The performance test of the processor products of the two companies in this assessment are all carried out under the condition of the processor itself.
So, the assessment of our comparison processor will be assessed on those details? The main frequency, bus frequency, cache size, number of transistors, processor core names, and other details will be listed item item item item items. Since the model of the processor is the first impression of our processor, this evaluation will also include AMD Athlon XP and subsequent processors, Intel Pentium 4, and subsequent models. Let's first list the core name and architecture of the processor. Overall, it will better help us learn how different X86 processors can do.
Let's first take a look at the AMD processor, maybe some Intel's supporters will ask why don't see the Intel processor first. But all things have, A before the alphabet, so we will first look at AMD products.
AMD processor product list
First of all, there is a need to explain that in the list, the size of the processor for Applebred and Thorton can be seen by the number of transistors, and their core is actually thoroughbred and barton. But because their L2 cache size only has a quarter and half of Thoroughbred and Barton core processors, we have added an asterisk next to them. They were originally designed to add Barton processor lineup, but now it seems that these have nothing to do, because AMD is now stopping production of Barton processors, and the old Thoroughbred is the core of the Socket A Sempron processor.
The number of PARIs, Victoria and Palermo in the list has decreased, so we can not determine the specific figures of these three core transistors, but we can still have some other technical parameters through these three cores. To understanding. Early Architecture Under Socket 754, the core of Athlon 64 cached to 512K is Claw Hammer, but its actual cache size is only half of the design. However, the new model processor, such as the outer frequency of 2.2GHz, cache 3200 of 512K; the outer frequency is 2.400 , the cached 3400 and the outer frequency of 2.6GHz, cache 3600 , the core they use In fact, Newcastle. Reducing the number of transistors is undoubtedly a good thing for Paris core, because it initially "low-level" Newcastle core, and later products reduce 18.7 million transistors on the L2 cache. Please pay attention to the TOLEDO core in the list. We labeled 2C in its line, indicating that this is a dual core. The same Opteron is also a dual-core processor with a period, but we are still unclear whether different Opteron processors use different cores in the future.
From the list, we can see the core of the Athlon processor includes a variety of cores from Pluto to Newcastle. It has a 10-level integer pipeline, a 15-level floating point line, an arithmetic logic unit, a location generating unit, and a floating point unit. The floating point unit has MMX, 3DNOW! / And SSE / SSE2 technical support. The future Athlon 64 will increase in the integer pipeline and floating point lines, and it is possible to reach 12/17 or higher level to increase the frequency of processors. We expect AMD to finalize the 31-level pipeline processor to more than Intel's products, but AMD does not equip all processors with a 31-level pipeline, because so makes the processor's clock frequency rise, The performance is enhanced, which will cost energy. Therefore, we believe that the pipeline length of the AMD processor is most suitable in the range of 10 to 15.
Intel processor product list
Please note that as the list of AMD processor lists, some cores in this Intel processor list are also raised an asterisk, which means that the core of the label is the faster core "low" Level "series. These annotated cores are basically the core of the Celeron processor. The core labeled a double star number indicates that the core is different in different stages. Of course, most chip mold sizes will vary, but Cascades, Coppermine, and Northwood's three core mold dimensions have changed. AMD's Thoroughbred A to Thoroughbred B is only 4 square centimeters, while Coppermine core mold size is from 106 square centimeters to 90 square centimeters, and the Northwood core mold size page changes from 146 square meters to 131. Square meter.
You can see from this table, Intel also has a dual-core product, and we labeled 2C next to him. We can even see Intel's Tukwila core is a 16 core. But 16 cores are clearly a myth for the majority of consumers, because the number of transistors like this should reach billions.
Compared with AMD, Intel launched several primary core construction series in the past seven years. AMD's K7 / Athlon core has been widely recognized in the past few years, and the company has received many commercial profits with this core. Intel Company continues to innovate with its greater strength and capital. First, with the Pentium III processor and its competitors AMD played a hand in the struggle for the processor market. Then Intel introduced a more powerful NetBurst core and implanted it into the Pentium 4 processor. It turns out that NetBurst core is a very good desktop processor core, but for notebooks, its performance is slightly lack of performance. Therefore, Intel decided to produce processors designed for notebooks. Finally, shortly after AMD launched its Athlon 64 processor, Intel immediately launched a processor with Prescott as the core. Processor parsing table
Before listing the processor parsing table, I will first explain to you why we want to list a processor parsing table. This is due to the many current processor models, which increases the difficulty of consumers choice, and consumers are difficult to conclude which performance is more performance through comparison processor models. such as. The core is Thoroughbred, the ATHLON XP processor model with the main frequency of 2250 MHz is 2800 , and the outer frequency of Athlon XP with Barton core is only 2000MHz. For the Intel processor, the situation seems to be worse, because the Intel's different models of the processor's cache size, bus frequency, and even architectures are different.
In the above article, we have listed the analysis of the processor cores of Intel and AMD in recent years. But this seems not enough because consumers are more concerned about the performance of the processor, and the processor has higher cost performance. So we will listen to the other parsing table below, which is designed to tell the reader which model is faster and better performance. But this is not a very accurate result, just a very rough assessment. The performance index of many different processor families will be listed below. I have changed their performance index to their main frequency to the processor's final performance score.
Since this is just a rough estimate, we don't have performance rankings with the performance score of the processor. If the following test is missing the processor or there is no newly released processor, I hope that the readers can see the amount. Of course, the following parsing table is just a reference, because each reader has their own standards of the processor's quality, and there is no fixed formula to measure how the comprehensive performance of a processor is.
If the reader does not like calculation or not, you can skip this page directly. This parsing table is made for readers who want to know more details. What needs to be explained is that we separate the dedicated purpose of the processor is to evaluate the plurality of processors only in the same series, so the score of the processor is measured in a particular environment, and the scores of other processors are not Contact, for example, Dothan score is 1.6 and Athlon FX scores are only 1.15. In fact, the performance gap between the two does not have a digital reflection.
DURON, Athlon, Athlon XP and Sempron processor
128K L2 Cache 100 MHz Bus = 0.7128K L2 Cache 133 MHz Bus = 0.75256K L2 Cache 100 MHz Bus = 0.8256k L2 Cache 133 MHz Bus = 0.85256K L2 Cache 166 MHz Bus = 0.9512k L2 Cache 133 MHZ bus = 0.95512K L2 cache 166 MHz bus = 1.0512K L2 cache 200 MHz bus = 1.05thlon 64 processor
256K L2 Cache Single Channel (Socket 754) = 0.9 512K L2 Cache Single Channel (Socket 754) = 0.951024K L2 Cache Single Channel (Socket 754) = 1.0 512K L2 Cache Double Channel (Socket 939) = 1.041024K L2 Cache Double Channel (Socket 940) = 1.111024K L2 Cache Double Channel (Socket 939) = 1.15
Celeron 2 and Pentium 4 processor
128K L2 Cache 400 Front End Bus Frequency = 0.6 256K L2 Cache 400 Front End Bus Frequency = 0.75 256K L2 Cache 533 Front End Bus Frequency = 0.80 512K L2 Cache 400 Front End Bus Frequency = 0.84 512K L2 Cache 533 Front End Bus Frequency = 0.911024 K L2 Cache 533 Front End Bus Frequency = 0.931024K L2 Cache 800 Front End Bus Frequency = 0.98 512K L2 Cache 800 Front End Bus Frequency = 1.0 512K L2 Cache 800 Front End Bus Frequency 2048K L3 Cache = 1.152048K L2 Cache 1066 Front End Bus frequency = 1.2
Mobile Celeron, Mobile P4, Celeron M, and Pentium M processor
128K L2 Cache 400 Front End Bus Frequency = 0.6 256K L2 Cache 400 Front End Bus Frequency = 0.75 256K L2 Cache 533 Front End Bus Frequency = 0.80 512K L2 Cache 533 Front End Bus Frequency Northwood = 0.911024K L2 Cache 533 Front End Bus Frequency PRESCOTT = 0.93 512K L2 Cache 400 Front End Bus Frequency Dothan = 1.25 512K L2 Cache 400 Front End Bus Frequency Banias = 1.31024K L2 Cache 400 Front End Bus Frequency Dothan = 1.351024K L2 Cache 400 Front End Bus Frequency Banias = 1.42048K L2 cache 400 front-end bus frequency = 1.52048K L2 cache 533 front-end bus frequency = 1.6
DURON and ATHLON processors
I don't want to be compared to the early Athlon and Duron processors in the details, because they have met brilliant in that era. But now, they have been out of time. If there is a reader who wants to know them, we have a rough measure of them in the following tests. Early PLUTO and Orion are core, using the Athlon processor with the Slot A architecture has a L2 cache, the size of the cache is the first frequency 1/2, 2/5 or 1/3, the faster, the lower the ratio. . For example, the L2 cache of the ATHLON processor of the 300MHz is 350 MHz, while the L2 cache of the Athlon processor of the frequency of 750 MHz is 300 MHz, the L2 cache of the higher frequency 850MHz Athlon processor is lower, 340 MHz. In general, at the time of the ATHLON processor and the Pentium III processor, both parties had their own advantages in different test items in different test projects. The ATHLON processor's X87 floating point performance is better, while the Pentium III processor is in optimizing application software MMX and SSE testing, exceeding Athlon processor.
The processor in Socket A has improved in the L2 cache technology. The performance of the AMD processor exceeds the performance of Intel processor products at the time of time. Athlon Thunderbird's cache reached 1.4GHz, and Pentium III had tried to reach 1.13GHz but finally in failure, the later Pentium III TULATIN reached 1.4GHz level, but this is to introduce Pentium 4 processors after Intel. of. Therefore, AMD is the love of many gamers.
Athlon XP and Sempron processor
The ATHLON XP processor changes some architectures of the Athlon processor to achieve better performance. Since the launch of the ATHLON XP processor is to compete with Intel's Pentium 4 processor, AMD has launched some new models and proposes the "clocked processor performance concept". AMD claims that the performance of the Athlon XP processor is equivalent to the Thunderbird core, but few people believe this saying, more people think this is just a business statement, because the new processor, the newly launched processor has been significant in the frequency. Increased, and Athlon core couldn't catch up with Intel's footsteps, so the concept of "The main frequency is not all" is raised.
With the continuation of the processor war, both companies continue to launch new products. The Thoroughbred core is a high operating frequency core, but does not reach the original intention of its design. Therefore, Thoroughbred B was born, which made the processor's operating frequency increased slightly, reached 2250 MHz. It is worth mentioning that the Thorousbred B core can overclock to 2.3GHz to 2.4GHz when the cooling conditions are suitable, but the working frequency of the Thoroughbred core is limited between 2.2.1 GHz to 2.2GHz.
After Thoroughbred, the Barton core launched by AMD has broken through the cache technology. Since the increase in cache can increase the processor's operating performance, AMD re-adjusts its processor model. At this time, Intel's processor products have improved in terms of cache and main line frequency, but the company has not adjusted the processor model. AMD launches 2500 , 2600 and 2800 to catch up with Intel's footsteps, but in fact, Intel launched the front-end bus frequency of 800MHz, "C" series Pentium 4 processor for 200MHz, AMD new front-end bus The performance of Athlon XP 3200 with a frequency of 200 MHz is only equivalent to P4 2.8C. A general consumer will simply think that 3200 performance is inconsistent with 3.2C, but the actual situation is not.
Athlon XP and Sempron processors have no more ways to introduce more, and the mobile processor is only reduced by reducing the consumption of electrical energy than its desktop processor. Therefore, mobile processor works In low voltage state. AMD Mobile Athlon XP processor has received extensive advice due to its excellent performance in overclocking performance.
Athlon 64 and Opteron processors
For the Athlon 64 processor, its name is due to its 64-bit address and integer bit. But the core of the Athlon 64 processor is not an upgrade version of the Athlon processor core. It still maintains 10/15 lines. With the support of x86-64, it is indeed a very successful business behavior and also attracts many consumers' attention. However, because 64 bits have not become the mainstream of the present market, how much it is still meaningless. But with the advent of Windows XP-64, I believe that it will always do things in the near future.
The advantage of Athlon 64 processor performance is mainly based on a good memory controller that can reduce delay. It can help the L3 cache is small if there is a large situation, and the memory delay can also be reduced while the memory frequency is increased. The ATHLON 64 3200 processor has a memory delay of 81ns, and the memory delay of the P4 3.2C processor is 77ns. At the same time, the ATHLON 64 3400 processor memory delay reaches 48ns.
Celeron, Pentium II, Pentium III processor
The old Pentium Pro P6 architecture has a 12-level pipeline, which has three special location generating units, two arithmetic logic units - a simple calculation, and another unit is used to solve complex calculations, and a float Point unit. The floating point unit has a CPU instruction set, and SEE (AMD does not have this support before the Athlon XP processor) and the support of MMX. Because of these, they make them ahead of AMD products.
Due to the use of the TULATIN core, the ATHLON processor's climb can easily exceed 1.4GHz. In fact, the later Celeron 1.0A and the Celeron 1.4a processor perform in overclocking are also quite good. The Celeron 1.1a processor with a bus frequency is 133 MHz has shown good comprehensive performance. Due to the needs of the market, Intel has stopped from the later days of its performance, and has developed a higher Pentium 4 and NetBurst processors.
Celeron 2 and Pentium 4 processor
The NetBurst architecture has a 20-level pipeline, which greatly enhances data processing power. In order to reduce the core introduction instructions and the time of compiling codes, Intel has introduced a new cache and is called Trace Cache. It is proven to improve processor data processing efficiency. The logical algorithm unit of the P4 processor under the NetBurst architecture runs twice the core frequency and can perform some instructions in half the core clock cycle, and the integer instructions are executed with twice the speed.
The subsequent Prescott architecture added a 3-level pipeline on the basis of the NetBurst architecture, reaching the 23rd pipeline. In addition, the L2 cache of the Prescott architecture is twice the Northwood architecture, and the SSE3 support is added, and the 64-bit to strong processor series under Prescott has now started. It is inferred that after XP-64, Intel will also introduce the corresponding 64-bit desktop processor.
Intel has fully launched synchronous multi-execution techniques in the past two years. This technology is implanted in the subsequent processor until the P4 3.06 processor. Later, all "C" processors in the 800MHz front-end bus frequency have support for Hyper-threading technology. Hyper-threading is to use special hardware instructions to simulate two logical kernels into two physical chips, allowing a single processor to use the thread-level parallel. This is compatible with multi-threaded operating systems and software to improve processor performance. At the same time as an error, another program can still run so that the running performance is increased by 20% to 50%. "Super Thread" technology can make the chip simultaneously multi-threaded processing, making the chip performance improvement. Now let's take a look at the Socket 775 LGA architecture, Intel also launched a variety of processors under this architecture. In general, numbered digital products have better performance, but we see such a phenomenon, the 5xx pin processor has higher performance than 7XX pin processor, and some 3XX pin processors are more than them. Advantageous performances are superior. All 5xx pin processors are used by the Prescott core, with an 800MHz front-end bus frequency and a 1MB size L2 cache. The specific situation of other products can be found in the table below. This way you will understand why the product number digital large processor has the reason why the performance of the number of processor products is not allowed to numbered numbers. Mobile Celeron, Mobile P4, Celeron M, Pentium M processor
Based on the Pentium 4 processor, the frequency limit is used, and the transformation is very suitable for the Intel Mobile Processor used by the notebook. The higher the clue, the greater the energy required, and it means the need to enhance the battery usage of the battery for the notebook. In the face of the pressure from Transmeta, Intel launches high-performance and low power consolidated Israel core processors. The final product is the mainstream processor Pentium M at the laptop. The Pentium M processor is extremely excellent in energy consumption, and Intel has maintained its Pentium M processor architecture above P6 in order to maintain its faucet in the mobile processor market.
The improvement of the P6 architecture is that it increases the L2 cache, which makes most caches can be in low energy "sleep" mode. The P6 architecture L1 cache is still twice the PIII, reaching 64K. Floating point performance is enhanced due to multiplication of MMX / SSE units, and the architecture is slightly modified in other respects. Overall, the performance of the Pentium M move processor is equivalent to the same frequency Athlon processor, but power consumption is slightly lower. Using the Pentium M mobile processor's notebook battery compared to the Pentium 4 processor, mobile Celeron or mobile Athlon XP will increase by 25% to 50% in use.
By the length of the above table, you can see that the mobile processor has an increasingly important location in real life.
Itanium and Itanium 2 processor
The Itanium processor is the processor product that is currently not the most consumers. Even the cheapest model, the price is also exceeded by $ 1000. This is not worthwhile, because Intel has launched such a processor at the time, turning the target to the high-end group market. The Itanium processor is used to handle a large number of parallel processes, which can also work in the 512 SMP system, of course, this is not a comprehensive explanation of the high-end processor of Itanium.
Initially, Intel made the ITANIUM processor as a 64-bit processor and put it under a new IA-64 structure. The IA-64 architecture is similar to the X86 architecture and is a future architecture system. In fact, its competitors are not as strong or Opteron processors, and itanium's advent of Itanium is Intel launched in order to occupy the high-end 64-bit processor market. Its true competitor is based on the server based on IBM Power4 / 5, HP PA-RISC, Sun Ultrasparc-III, and Dec Alpha.
In terms of processor structural design, Intel uses a technology used in previous mainframes, ultra-long instructions (VLIW). This is a very long instruction combination that connects many instructions together to increase the speed of the processor operation. But itanium is not a strict sense of long instruction word processors, which is because the ultra-long instructions have their own weaknesses, while Intel has improved these weaknesses and renamed, called exact parallel instructions (EPIC). The Itanium processor can perform up to 3 instructions at the same time compared to the Strong Opteron processor. Itanium 2 processor can perform 8 instructions within one cycle. In terms of performance, the ITANIUM processor of the cluster is equivalent to the force of 2.66 GHz, and the performance of the ITANIUM 2 processor of the frequency of 1.5GHz is equivalent to 4GHz's sufficient or Opteron. processor. In order to achieve the best performance, the Itanium processor uses a 128-bit four pump bus that uses standard SDRAM memory. The initial MERCED version Itanium processor has four arithmetic logic units, two floating point units, and three branch processing units, two SIMD units (e.g., MMX / SSE), and two addresses. The post-improved version of the MCKINLEY ITANIUM processor has six arithmetic logic units, three branch processing units, two floating point units, and SIMD units, two read units, two storage units, and four addresses. . In addition, the McKinley version of the cache bandwidth is three times that of the MERCED version.
Conclude
Of course, what we did today is only substantially reviewing the processor products launched between AMD and Intel in these years and make some prospects in the future. It is reported that the dual-core architecture should be officially launched in the next year, and the rumors of the four-core architecture processor will have always seen all websites recently, but according to the current situation, the four-core architecture processor is still very distant. AMD and Intel's two companies have been committed to developing new processor manufacturing technologies, we will gradually stay away from 45 nm production techniques, followed by 13 nanometers or even 11 nano production technology. Perhaps we will eliminate the processor products that are now considered high-end performance in the near future.
Now that the performance of the computer processor you use has increased much more than decades ago. I often want to use a word to describe the change in computer technology in ten years. Ten years ago, 3D is still an unreachable technology, 3D accelerator is still a product that costs thousands of dollars. Ten years ago, the 32-bit processor is still looking for an operating system that truly suitable for you, while higher performance 64-bit processors have been used in government departments and research centers. Ten years ago, the 100MHz processor also belongs to the high-end processor product. It is also a 10 years ago. How many users can enjoy the Internet to enjoy the convenience of the Internet, and how many users enjoy the "high speed" of the 28.8k cat. What will it be like ten years? Let us wait and see.
