Supplementary question: Please buy tualatin C3 1.0 or 1. 1, around 50 yuan. Or tualatin P3 1.4, around 200 yuan. 8 15EPT motherboard, 100 yuan or so. Please buy 9600 mobile version or 9550, 150 yuan, the most cost-effective AGP graphics card.
I didn't find the following article on purpose, but I saw it in the forum earlier. I thought it was well written, so I got tualatin C3 accordingly.
This paper mainly describes the emergence and development of tualatin Celeron and tualatin P3 in detail. You can also watch it directly here.
/topic.jsp？ tid=64775 12。 lastUpdate=04-26.20:54
Look, we should learn from the past now, and we can't do anything without the past.
Reprinted from ZOL's untimely sacrifice on Intel tualatin processor.
Abstract: Is the hot braised pork a descendant of tualatin? I just bought the machine in June, 5438+October, 2002/KLOC. At that time, P4 1.6 1000 845d 900 was a little more, and tualatin Celeron 1.0 600, 815et600 was a little more. I'm glad I bought tualatin at that time and didn't waste money. After overclocking, tualatin p4 2.0 can't match it, and it's still in use. Its performance is really strong, absolutely classic. It is the 8 15 series architecture and sdram that limit the further improvement of his performance.
This paper mainly expounds the following issues:
1, on the classification and technology of tualatin;
2. Why did Intel launch tualatin?
3. Which performance is better, tualatin or Pentium 4?
4, the fatal flaw of Pentium 4, thin and light notebooks rely on tualatin.
5. Pentium 4 stopped at 3.8GHz, and Intel lost its way;
6. The core pays homage to tualatin: Pentium 4 NetBurst architecture ends, and the ancient P6 architecture reappears.
The first part discusses the classification and technology of virtual reality.
Tualatin, whose English name is Tualatin, is the name of a river in the United States. This is the core R&D code of Intel's last Pentium III processor. Tualatin Pentium 3 has a short history, but its status is special. Don't forget that desktop Pentium III has three cores: the oldest Katmai Pentium III (actually the core of Pentium II plus SSE instruction set, 0.25 micron technology, 5 12KB half-speed L2 cache, slot 1 interface), and the most popular Coppermine Pentium III(0. 18 micron technology, 256KB full-speed L2 cache). Socket 370 and Slot 1), and the protagonist of this article, tualatin Pentium III(0.65448). Among these three Pentium III models, only tualatin is listed separately to show the essential difference with the other two Pentium III models: tualatin Pentium III combines the advantages of ancient P6 architecture, and has the advantages of high performance, low power consumption and low calorific value, which is the ultimate of Pentium III, and also the Pentium M and Core (these two processors still inherit P6 architecture). Today, even the second-hand notebook merchants who don't know CPU technology know how to use the name "tualatin" to flaunt the uniqueness of their Pentium III second-hand notebooks, which shows the sacred position of tualatin in the eyes of users.
All of the above are about the desktop version of the processor. In fact, tualatin has not only a desktop version, but also a mobile version and a server version. The desktop version also has the difference between Pentium III and Celeron. Here's a detailed introduction to their differences.
Let's start with the desktop version of tualatin Pentium III. This processor has a lot in common with Coppermine's Pentium III( 133MHz FSB): both are 133 MHz FSB, and also have a 256KB full-speed L2 cache. The difference is that tualatin Pentium III does not support dual-processor parallel working mode, while copper-core Pentium III does. Tualatin Pentium III has a L2 cache latency of 1, while copper Pentium III has a latency of 0; Tualatin Pentium III supports the data pre-reading technology adopted since Pentium 4, while copper Pentium III does not. Therefore, in terms of performance, the tualatin Pentium III with the same main frequency is not faster than the copper-core Pentium III (data pre-reading technology can make up for the performance loss caused by the long delay time of the secondary cache). Tualatin Pentium III has the following models:1GHz/1.13ghz/1.2ghz/1.33ghz.
Let's start with the desktop version of tualatin Celeron. Many people think that Celeron in tualatin is tualatin Pentium III with 100MHz FSB, which is incorrect. The performance of Celeron in tualatin is lower than that of Pentium III in tualatin with the same frequency. In addition to the low FSB, Celeron in tulatin only supports the simplified data pre-reading technology (there is another saying that it does not support it at all, because the author has limited data and will not give a demonstration), and the bit width of the secondary cache is also low, which cannot be compared with Pentium III in tulatin. According to my test, the performance of tualatin Celeron 1.4 GHz is slightly lower than that of tualatin Pentium III 1. 13 GHz, but higher than that of Copper Mine/tualatin Pentium III 1GHz. Tualatin Celeron has the following models: 900 GHz/1GHz/1.1.2 GHz/1.3 GHz/1.4 GHz. Many people deny that tualatin Celeron has a version of 1.4GHz. In fact, IBM NetVista A22 has a Celeron model of1.4ghz, so I bought one.
Thirdly, let's talk about the server version of tualatin Pentium III-S. This processor is the king of the Tualatin family, and its price is staggering (I collected two models: 1.26GHz and 1.4GHz). Tualatin Pentium III-S faces the dual-channel server market, adopts 133MHz FSB, has 5 12KB full-speed secondary cache, has zero delay time of secondary cache, supports enhanced data pre-reading technology, and has strong performance. At that time, many foreign English hardware evaluation websites had comparative evaluations of tualatin Pentium III-S and Pentium 4/Athlon XP. Taking Tom's hardware guide website evaluation as an example, tualatin Pentium III-S 1.26GHz beat Pentium 4 660 with 850 platform and PC 133 SDRAM memory. Tualatin Pentium III-S has the following models:1.13ghz/1.26ghz/1.4ghz. In addition, a few engineering samples of 1.53 GHz are scattered in famous hardware evaluation rooms around the world.
Finally, let's talk about the mobile layout Latin Pentium III-M. The technical features of this processor are basically the same as those of the desktop version of Tualatin Pentium III, but the secondary cache is increased to 5 12KB, and the same-frequency performance is obviously better than that of the desktop version of tualatin Pentium III. It is obviously higher to catch up with tualatin Pentium III-S..Tualatin Pentium III-M-M than the desktop version of Tualatin Pentium III, because the mobile version of Pentium 4 processor is hot and consumes a lot of power, which is very difficult. In contrast, the tualatin Pentium III-M, which is hot and consumes a lot of electricity, has found its own position and made great progress. Before Pentium M appeared, tualatin Pentium III-M was a natural choice for thin and light notebooks. Pentium III desktop version will threaten Pentium 4 because of its strong performance, so Intel artificially suppressed it: castrating the second-level cache, maintaining high prices and low output, and encouraging brand-name machine manufacturers to adopt unfair market measures such as Pentium 4. Tualatin Pentium III-M has many complicated models. Taking the normal voltage version as an example, there are mainly 866 MHz/933 MHz/1GHz/1.13 GHz/1.2 GHz/1.33 GHz and so on.
Part II: Why did Intel launch tualatin?
Pentium III and the microprocessor era before it are the times when processor performance and main frequency are closely combined. At that time, whether it was Intel or AMD, their processor products would increase the main frequency and bring corresponding performance improvement. Take Katmai Pentium III as an example. Under the same FSB, if the main frequency is increased by 50MHz, the performance will be improved by about 7% (refer to PC Shopper- Electronic Test). At that time, the processor competition between Intel and AMD was the same as now. In the final analysis, it is a performance dispute. But at that time, the performance and main frequency of the processor were integrated, so the performance dispute was also the main frequency dispute. Since the introduction of 4004 microprocessor, Intel has been the leader in microprocessor performance (i.e. main frequency), and Cyrix and AMD are followers. This situation continued until the Pentium III era, until the introduction of AMD Athlon. I still remember that when AMD announced in advance that it would release a new architecture, the Athlon processor with a frequency of 600MHz, Intel panicked. At that time, Pentium III was still in the era of 0.25 micron technology, and the highest frequency was only 550MHz. What will the face of the microprocessor boss look like if AMD is allowed to surpass it? So Intel added 0.05v to Pentium III in a hurry, and first introduced Pentium III 600MHz processor to save face. Who wants AMD to release the Athlon 600MHz processor as scheduled, and also released the Athlon 650MHz processor that was not announced before! This is a great moment, Intel is surpassed by its competitors for the first time in the history of microprocessor development! Beaten Intel began to make a lot of tricks. First of all, it upgraded the external frequency of Pentium III, the core of Katmai, to 133MHz, and introduced Pentium III 533B and Pentium III 600B (some people described these two processors as new cores with confidential code names, which is really a big joke. Find out for yourself what confidentiality means. Any unpublished engineering sample processor of Intel has a secret label), but its performance can't compete with Athlon with the same frequency, so the Coppermine Pentium III with 0. 18 micron technology comes on the scene in a hurry, with the frequency from 500E MHz to 700MHz, the interface from slot 1 to socket 370, FSB 100MHz,/kloc. Coppermine Pentium III's second-level cache is 256KB at full speed, with zero delay time, and its bit width is 5 12KB higher than Katmai Pentium III's half-speed second-level cache, so its performance is obviously improved, which is basically the same as Athlon's same frequency. At this time, the competition between Intel and AMD has reached a white-hot level, and the intensity of the main frequency dispute is unprecedented. High frequency means high performance, and the impact on 1GHz is also significant because of variation. At that time, Intel desperately searched for the best silicon chip, hoping to make the processor of 1GHz first, which also made the Pentium III with high frequency of 933MHz very few, and then it was out of stock (because the best silicon chip was used to produce and test the Pentium III of 1GHz). ...
The first one to reach 1GHz is AMD's Athlon, which is only less than a month ahead of Pentium III's 1GHz, but this is enough to make AMD proud. The fly in the ointment is that the performance of Athlon 1GHz is not as good as that of Pentium III 1GHz. This is because AMD can't find an Athlon secondary cache chip that can run stably at 500MHz (the earliest Athlon secondary cache is also a chip outside the CPU PCB, running at half the main frequency of the processor). Therefore, the running speed of the secondary cache should be reduced to 333.3MHz, which is one third of the main frequency of CPU. Although the second-level cache of Pentium III 1GHz processor has a small capacity, its running speed is as high as 1GHz. Relying on powerful L2 cache, Pentium III of 1GHz beat Athlon of 1GHz in performance, saving some face for Intel.
But it's not that simple.
AMD has realized the defects of external half-speed cache and decided to develop Athlon XP with second-level cache on Die. This is what scares Intel most, so Intel wants to launch Pentium III at 1. 13GHz first, and then build a production line with 0. 13 micron process, and beat Athlon XP with tualatin. The backup plan is Pentium 4 Willamette, because Intel can produce Pentium 4 with 0. 18 micron process. However, due to the low frequency and low energy of Pentium 4, the main frequency can only be brought into play if it is raised to a certain level (for example, above 2GHZ), and the NetBurst architecture of Pentium 4 needs the cooperation of hyper-threading and high FSB, so Intel does not want Pentium 4 to debut too early, but hopes for tualatin. At that time, the technical parameters of the tualatin desktop processor planned by Intel in the internal text for the media were basically the same as those of the actually produced tualatin Pentium III-S, but the FSB was upgraded to 166MHz/200MHz, and the planned 830 chipset was also ready to support DDR memory, which was very exciting. Unfortunately, people are not as good as the sky. At that time, the 0. 18 micron process was applied for less than one year, so it was really difficult to produce Pentium III with 1. 13GHz. The consequence of Intel overlord's hard bow is to fully recover the Pentium III 1. 13GHz processor that can hardly run normally. At this time, AMD is introducing the newly launched 1. 1GHz Athlon XP processor to the media with a big smile. ...
Intel lost face, so it began to plan to divert attention, turning the processor's main frequency-performance dispute into a pure main frequency dispute to divert public attention. As a result, Pentium 4 1.4 GHz/ 1.5 GHz was released, which immediately attracted the attention of the whole world. People simply can't imagine that Intel, which has just recovered its 1. 13GHz processor, can come up with such a processor at once, 1.5 GHz! At this time, naive people are still watching Pentium 4 with a fixed routine in which the main frequency is performance. This is Intel's trick.
But there are always sober people, and that is fair hardware evaluation media.
On the eve of the release of Pentium 4 1.4 GHz/ 1.5 GHz, some evaluation media questioned: Why is the main frequency of Pentium 4 so high, but many actual test items are not as good as Pentium III of 1GHz? Intel's answer is: Pentium 4 is a brand-new architecture, and its performance can't be judged by the traditional viewpoint. There is no practical significance to compare Pentium 4 and Pentium III at 1GHz. However, Pentium is also a brand-new architecture compared with 486. Why did Intel hype Pentium twice as fast as the same frequency 486, and then took out official test data? After the release of Pentium MMX/Pentium II/Pentium III, Intel will also produce their performance comparison and evaluation data with the previous generation processors. Why did Intel hand in a blank paper when Pentium 4 was released without official evaluation data? In fact, it was not until the appearance of 2GHz Pentium 4 that Intel shyly produced an official test data to compare the performance of 2GHz Pentium 4 and 1GHz Pentium III. Of course, Pentium 4 at 2GHz can't beat Pentium III at 1GHz, but what's the point? The 486DX4- 100MHz processor still can't beat the Pentium 60MHz processor, and people have a high frequency.
The release of Pentium 4 set a bad precedent: under the same main frequency, the performance of the next generation processor is never lower than that of the previous generation processor, and Pentium 4 did it; In the history of Intel processor development, there has never been a clock failure in processor upgrade, and Pentium 4 has also done it. The initial operating frequency of Pentium 4 is1.4 GHz (Pentium 4 with1.3 GHz was released very early, only for brand manufacturers), and the closing operating frequency of Pentium III is 1GHz. Who will fill the 400MHz frequency gap? This thankless task was left to tualatin, who was almost stillborn. This is almost the only reason why tualatin Pentium III Desktop Edition came out. Tualatin Pentium III, like an abused child bride, was carefully born. In order to minimize its huge performance advantage over low-frequency Pentium 4, Intel first castrated half of the second-level cache; At first glance, the performance is still too good, and then the FSB drops to133 MHz; ; At first glance, the performance is still too good, so the delay time of the secondary cache is increased to1; At first glance, it is still too awesome, and it is necessary to castrate. Damn it, stop it. Isn't castration Celeron? Tualatin's development engineer quit, so Intel made an idea on the supporting chipset, canceled the release of the 830 motherboard, and introduced an 8 15GM chipset with integrated i752 graphics card to match tualatin's Pentium III ... Even so, tualatin 1. 13GHz Pentium III was still looking for JJ everywhere. As a result, Intel once again lost money, strictly restricted the shipment of Pentium III in tualatin, artificially raised its price to a very high position, and encouraged brand-name machine manufacturers to adopt Pentium 4 processors. Therefore, tualatin Pentium III is hard to find in the market, and its scarcity even exceeds that of the server version of Pentium III-S. It climbed from 1GHz to 1.33GHz, and tualatin Pentium III finally completed the task of filling the gap in the main frequency of two generations of processors (tualatin Pentium III 1.33GHz was the worst, and it was discontinued after release).
Tualatin Celeron's fate is much better than that of his brother. This is because Intel needs it to compete with AMD's Diamond Dragon and inherit the excellent performance of tualatin. 256KB cache (Celeron in tualatin has much more 256KB cache than Celeron D), 100MHz FSB, streamlined data pre-reading technology, Celeron in tualatin did a good job. However, tualatin is a double-edged sword, and Intel began to worry about the handover with Willamette Celeron. What if tualatin Celeron beat Willamette Celeron like tualatin Pentium III beat Low Frequency Pentium 4? What if he also defeated Willamette Thain and looked for JJ everywhere? Oh, I forgot, Celeron was originally a eunuch without penis, but Celeron in tualatin is obviously Wei Xiaobao, and false eunuch is really a man. Intel couldn't think of a clever solution, so it left a blank of 300MHz for these two Celeron models. As soon as the layman saw it, I took the exam. How can Celeron 1.4GHz be completed? 1.5/ 1.6GHz products? Ironically, tualatin Celeron 1.4GHz still easily defeated Willamette Celeron 1.7GHz to find a dildo. ...
The server version of tualatin Pentium III-S has come out, which has Intel's own difficulties. The performance of Pentium 4 Xeon is not good (it was beaten by tualatin Pentium III, so we can't find JJ, let alone compete with Pentium III-S), and the supporting motherboard platform is expensive (1GB Rambus memory, the price is calculated by ourselves), so users ignore it, so they take out Pentium III-S, which is compatible with the existing 8 15 platform, with strong performance and low fever, and make 65438+. So Pentium III-S came out at a high price. It is revealed here that many tualatin Pentium III-S have no frequency locking ... In contrast, the mobile version of Tualatin Pentium III is the best. As mentioned above, Pentium 4 has the characteristics of high heat value and high power consumption. When the notebook CPU is used, it really drives ducks to the shelves and pigs climb trees. Tualatin Pentium III has the advantages of low heat generation, low power consumption (in fact, Tualatin uses a good heat sink, and there is no fan at all when the chassis is well ventilated), and at the same time, its performance is very high (tualatin Pentium III-M can beat the mobile Pentium 4 which is several hundred MHz higher than itself to look for JJ everywhere, but the mobile processor is too small to find JJ with a magnifying glass), so it is the most suitable processor for notebooks. Therefore, Intel promoted tualatin Pentium III-M in a high profile, and designed a logo similar to the mobile version of Pentium 4, hoping to use tualatin Pentium III-M to occupy most of the mainstream mobile processor market. But tualatin, no matter how good it is, is also Pentium III. Didn't Intel say 4 to 3 with overwhelming advertisements? Why buy a notebook instead of Pentium 4 but Pentium III? Intel can't say how bitter it is, and notebook manufacturers are unlucky, so they have to use the mobile version of Pentium 4 as the mainstream notebook. Thin and light notebook can't accommodate the mobile version of Pentium 4, so tualatin Pentium III is adopted. As a result, users complained one after another and asked the manufacturers who bought mainstream notebooks why the battery life of your latest notebook is so much shorter than that of the old notebook. Users who bought a thin notebook also blamed the manufacturer, saying that thin notebook is the most advanced technology of notebook manufacturers. Why did you sell me an obsolete Pentium III notebook? Dissatisfaction of manufacturers and complaints of users prompted Intel to find another way and start designing Pentium M. This Pentium M Centrino platform is actually a tualatin Pentium III, a new soup (name), a seasoning (SSE2) and two snacks (wireless network card and 855 chipset).
The third part is tualatin and Pentium 4, which has better performance?
Before we talk about this question, let's make an analogy. What do you think it takes to complete the dunk? The answer is height and jumping ability. Very tall people can dunk lightly, but it doesn't mean that short people can't dunk. In our country, those weightlifters who are about 1.6 meters can jump in place without running up, and easily make a backward dunk, relying on developed leg muscles and amazing jumping ability. Similarly, there is not one factor that determines the performance of the processor, but two main factors, one is the main frequency and the other is the execution efficiency at the unit main frequency.
Pentium 4 can easily reach the main frequency of 2GHz in the same 0. 18 micron process, and Pentium III has just reached the limit of 1. 13GHz. Why? This is because Pentium 4 has as many as 20 levels or even 365,438+0 levels, while Pentium III has only 65,438+065,438+0 levels. The longer the pipeline, the easier it is to achieve higher working frequency under the same manufacturing process. Athlon can achieve the high frequency that Pentium III can't achieve under the same manufacturing process, because Athlon's computing pipeline is slightly longer than Pentium III. However, a long pipeline will also bring negative effects. The longer the pipeline, the lower the execution efficiency of the processor at unit frequency, and the performance will be affected. As we all know, Pentium 4 was designed on the assembly line to achieve such a high main frequency, and defeated AMD in the battle for main frequency, forcing Athlon XP to "dare not show its true colors" (referring to Athlon XP being labeled as "more or less equivalent to Pentium 4" instead of the real main frequency). However, the extension of the pipeline will lead to the increase of data residing in the pipeline and the possibility of data error. Once there is a data error, the result is that the whole operation steps will be repeated, which will reduce the processing performance of the processor. Relying on efficient branch prediction system and caching mechanism can improve this point, but it is only "improvement" rather than "compensation". If we want to make up for the shortcomings of high frequency and low energy, we must rely on higher main frequency to offset it. For example, under the same manufacturing process, the performance of Pentium 4 at 2GHz will always be higher than that of Pentium III at 1GHz, which means the performance improvement without changing the manufacturing process. It's just that Intel is unkind and deliberately misleads people to have the misunderstanding that "high frequency = high performance", which has no longer been established since the advent of Pentium 4. AMD, even Athlon 64, has a short pipeline, so its execution efficiency is very high. Athlon 64' s integer pipeline reaches 12 level, and its floating-point pipeline reaches 17 level. Compared with Athlon XP, it only improves two stations, but greatly improves the branch prediction mechanism, so the performance under the same frequency is obviously improved. From this, we can see two fundamental differences between Intel's design of Pentium 4 and AMD's design of Athlon: Intel pursues frequency priority and relies on extremely high frequency to improve performance; AMD, on the other hand, pursues efficiency first, and achieves higher performance under the same main frequency through efficient execution mechanism. AMD actually inherited Intel's P6 architecture design idea (Pentium III, Pentium M and even Core all inherited this design idea).
Knowing these two design principles, and considering that the execution efficiency of Pentium III is only about 70%-75% at Pentium 4 unit frequency, we can easily draw the following conclusion: at the same frequency, the performance of Pentium III is obviously higher than that of Pentium 4. However, under the same manufacturing process, Pentium 4 can reach a high frequency that Pentium III can't, and this high frequency can completely make up for the low performance. If the power consumption and calorific value of the processor are not considered, it can be considered that these two design ideas of the processor are feasible. However, the processor can't just stay on paper, it must be made to realize its value. However, when the actual processor works, it will inevitably consume electricity and heat. At this time, it depends on which design idea is more practical and reasonable. Pipeline computing pipeline is not a tiny biological cell tube like vas deferens and fallopian tubes, but is composed of transistors. Obviously, the longer the pipeline, the more transistors there are; The more transistors, the greater the power consumption; The greater the power consumption, the higher the calorific value. This is the fate of Pentium 4 processors. Who wants to keep an electric tiger at home?
The fourth part is the fatal flaw of Pentium 4, and the thin notebook relies on tualatin.
Pentium 4 has gone through three generations of cores since it came out. The author's evaluation of these three generations of products is "from failure to success, and then from success to new failure"
The first generation core of Pentium 4 is Willamette, which was released in June 2000. It is manufactured by 0. 18 micron process, with 256KB full-speed L2 cache and 400MHz front-end bus. Compared with Pentium III with P6 architecture, Willamette Pentium 4 has no improvement in integer processing speed and floating-point performance. It was only a stopgap measure against Thunderbird Speeddragon, because at that time, Intel did not prepare a 0. 13 micron process production line to produce Northwoood Pentium 4 and tualatin Pentium III. Fortunately, when the main frequency is raised to 1.7GHz, Pentium 4 has completely surpassed Pentium III in performance, which brings some psychological comfort to Intel. However, the fatal problem of Willamette Pentium 4 is not the low performance, but the amazing power consumption and calorific value. An insider once pointed out that an Intel engineer suggested to match Willamette with 5 12MB cache or even L3 cache (later implemented in Gallatin Xeon and Beimu Pentium 4 Extreme Edition) to improve Willamette's performance, at least to make it not lose to the previous generation Pentium III. However, this proposal is unrealistic. Those who refute it point out that if it is really done, I am afraid that Willamette Pentium 4 will be made as big as a brick. This may be a joke, but it also reflects the fact that Willamette Pentium 4 was kicked out. However, Intel's propaganda ability is strong. Many newly-installed friends spent thousands of dollars to buy a set of Willamette Pentium 4+850 motherboard +Rambus memory, but they didn't expect to become the scapegoat of Intel's face-saving action.
The second-generation core Northwood has solved the problems of heat generation and power consumption because of its abundant preparation time and 0. 13 micron process, and quickly replaced Willamette as the mainstream of the market. Since then, with the introduction of new technologies such as 800MHz FSB and hyper-threading, the power of NetBurst architecture has been fully exerted, so tualatin Pentium III can no longer compete with Northwood Pentium 4 in performance. Actually, I use it myself. Whether playing games (Q3A) or watching high-definition movies, Pentium 4 2.8GHz is much higher than Pentium III-S 1.4GHz, and even AMD has begun to seriously examine Pentium 4, a processor that it has always regarded as an adult toy.
Northwood Pentium 4 can really become a classic, because it reproduces the original intention of Netburst development team. Netburst architecture adopts a computational pipeline with extraordinary pipeline, and such a long computational pipeline is naturally prepared for hyper-threading technology. You should know that processors with short pipeline P6 architecture like Pentium III/ Pentium M/ Core can't realize hyper-threading technology! There are many technical articles about Hyper-Threading, so I won't repeat them here. Prescott, the successor of Northwood Pentium 4, is a new failed work. This failure made Intel finally decide to abandon Netburst architecture. I don't know how Netburst's development team thought that such a behemoth would consume limited energy on the earth: it inherited and developed all the shortcomings of Willamette Pentium 4, and the assembly line was lengthened to 365,438+0! Although Prescott Pentium 4 also has advantages (such as SSE3, larger L2 cache and more efficient hyper-threading technology), jade does not hide its shortcomings. It has so many transistors that it generates about 60% more heat and consumes about 10% more power per clock cycle! Prescott processor is easy to overheat, and the result of overheating is down-frequency operation. Prescott finally lost to Wen. Intel must consider whether air cooling can meet Prescott's needs. If you need water cooling ... let Netburst die. As we all know, Intel had to give up Prescott architecture completely, and the efforts to develop the 4GHz part were also considered to be a waste of internal resources. After stopping at 3.8GHz, Intel finally admitted that it was wrong. Don't forget, when Intel released Pentium 4, it boasted that Pentium 4 was designed for the computing speed of 10GHz, which must be the most important and possibly the most famous engineering failure in Intel's history.
The mobile version of Pentium 4 was born out of the desktop version of Pentium 4, and its fatal defects are of course power consumption and heat generation. Almost all users of the private network have notebooks. Presumably, everyone wants their notebooks to have a low temperature and a long battery life. The mobile version of Pentium 4 obviously does not meet the requirements of friends.
Speaking of mobile Pentium 4, there are two main categories. One is the mobile Pentium 4-M processor released on April 23, 2002, which has SpeedStep and deep sleep technology, but does not support hyper-threading technology, and consumes about 35W. The other is the mobile Pentium 4 processor, which was released because it has been proved that it is impossible to further improve the main frequency of the mobile Pentium 4-M while maintaining low power consumption. The purpose of mobile Pentium 4 is to provide a high-frequency processor for notebook computers. Of course, its power consumption is also lower than the desktop Pentium 4, "only" 70W. Isn't it cool? Mobile Pentium 4-M can not further improve the main frequency, and the control of power consumption is also unsatisfactory. The main frequency of mobile Pentium 4 has increased, and the power consumption and calorific value are frightening ... This is a problem for Intel and notebook manufacturers. Tualatin came in handy at this time. It has all the advantages that a mobile processor needs: low power consumption, low heat consumption and high performance. From the technical design point of view, almost all notebook manufacturers like tualatin Pentium III-M, but from the market point of view, under the overwhelming advertising of Intel, the main promotion of tualatin notebook is tantamount to suicide. After trying to put the bitter fruit of tualatin Pentium III-M into the cold, almost all notebook manufacturers have made such a choice: design and manufacture ordinary notebooks with mobile Pentium 4, and dedicate tualatin to the development of light and portable notebook products. As long as you look closely, you will find that no matter what brand, notebooks with mobile Pentium 4 processors are big, thick and heavy. Well-designed manufacturers can solve the problem of overheating of such notebooks, but the battery life cannot be guaranteed. You can't match a big, thick and heavy notebook with a big, thick and heavy battery, can you? In contrast, tualatin Pentium III-M is much better. The notebook with this processor is light and portable, with long battery life and powerful performance. I don't know where Intel's inspiration for designing Pentium M, which is completely different from desktop processors, comes from, but there is no doubt that tualatin Pentium III-M is the prototype of Pentium M. ..
The fifth Pentium 4 stopped at 3.8GHz, and Intel lost its way.
On June 5438+1October 65438+May, 2004, Intel decided not to launch the Pentium 4 processor with 4GHz, and said that the clock speed would no longer be used as the main criterion for evaluating chip performance. When announcing this decision, 64-year-old Intel President Craig? Barrett said to 6500 people, "Please forgive us". Or this cute old man with a sense of humor is joking, but now he is on his right knee.