Good morning.
Distinguished ladies and gentleman, guests both in Shanghai and where you might be watching on your computer screen. We are trying to do some different broadcasting presentation in computing in different languages simultaneously.
Let me tell you all about what the premise of ever since that I'd like to talk about today. Our world in using computers towards a universe of reaching 1 billion connected computers, computer that can share information, data and that can work and allow people in different part of the world, to work together, to operate as if we were rightness of each other with neighbors of each other that is the premise of our work and that premise on which the activity, the large part of computing industry base today.
The two major driving forces computing industry today are first the available of the low cost and high performance of computing and all elements of global network that we are talking about and of course the network itself. Our industry that is looking at Peking industry with huge volume has become a major industry. I think them are add the number of 1998, the industry world wide we have showed something 100 million units of computers that is larger number than television sets production in the world. It is industry that grows in China, distributes in China, emerging at Chinese magnificent feature as a deading force in supplying China fency market that's very interesting sight, the sign of how industry of Chinese society in hasing, the world of computing and also connected computing.
Our work in the business of supply, the fundamental basic block computing which is microprocessor as the units of computing evolved, the use of microprocessors evolved as well and we evolve the technology according to the fact changes but typically evolution to allow the world to are the microprocessor's architecture into addition of segment are the computer universe changes. That industry what I mean by it in more a bit detail.
This is evolutionary in the basic computing industry as you remember fashion computer started as a best sort of machine basically as a office for people to do individual work on and then as a world one market of basic computer segmentation in basic computers in general should be modified in adopting the different segments of computer market some of them are the desk sort in the course of service, some of them. They make small in reliable in both with computers and increasingly broaden the segmentation in terms of computers are the consumers business users use the computers as consumers use the computers and believe that both purposes both carry with the computers. In the last couple of years we have seen the substantial forward segmentation, finer segmentation of the computing market. We see a product introduced as a microprocessor reach higher into the more demanding computers into enterprise service and in high performance computers. We at same time see another evolution of standardizing high service and economic at feasing industry to one modified service and at same time modified the service from that used to be the service to service designed to be the serviced.
On the best the fragment of the segmentation in the market also you see work stations at emerging market for our microprocessors and computer microprocessors as major market hold or microprocessors and computer microprocessors and same time you see very basic computers developing on the microprocessors. In the consumer phenomena again we see the performance computers and basic computers is evolving as well our challenge as a designer and manufacturer of microprocessors in supplying and serving the fragmency segments of the market is quite bouncing we learned that to serve the segment in beneath the design increasingly different microprocessors for each of the segment of timizing the microprocessors of the segment while at the same time retaining the fundamental in architecture similarity up and down on that diagram. We do a lot of efforts in investing technology and manpower in the future and discuss a lot of efforts couple of months and we are going by a lengthy breath of easy change and find this market and we are investing a lot of money and hope the product of investment in the future. I am not sure the number for 1998 but our estimate of the product that 1998 will represent 8 billion US dollars.
The key element that we try to achieve as a server involving divergent segment is to use the same fundamental micro architecture of microprocessors to the entire this of spectrum of the application of segment that we call the P6 micro architecture. But keep in mind even the micro architecture is going to be the same up and down of segment the product themselves will be designed specifically at each of segment initially that optimize performance and cost performance at each of these.
The illusion of the product represents the supplementations of the products were at the segmentation of our brand. We have a successful in establishing the Pentium brand over the last several years as the major brand near all of our processors. As the single brand more than we can describe the multiple of the product family that we disper in establishing the fancy brand over the last year as a major brand the more than ordinary microprocessors so more important in that we will introduce or have introduced additional brands as Xeon(TM) brand processor represents the portion of product line that is designed for basic computing and the Pentium Ⅱ Xeon processors represents the portion of product line that goes to the higher level on the segment served to workstations and serve application.
Let me start by the area of basic computing, designing for basic computing again remain base on the same P6 microprocessor architecture on entire product line is based on. In addition the employ that we call smart integration that is using the increasing budget of more and more transistors on an economical and effective chip it achieve product effectiveness by combining different functions on the same flap of a chip. From the history being from time to time for example that way back from ten years ago when the availing processor architecture involved of the chip we discover that as the transistors budget increase as the functions that represented by separate chip on the same chip on that generation, the generation for the same degree. This function for the processor than forward from that to be present day beyond from Peking at product at same chip on which we do up on the micro pro revived. Put it as an example. The product information that we intend to employ collectively exhibits as it design for microprocessors called the basic computing segment.
In addition to that even the on cost effectiveness of the segment we are adopting a low cost packaging technology called this and are in order to illustrated this. Let me introduce what the main line microprocessor's central process looks like probably the Xeon has the feature of television but the package design for the Pentium Ⅱ microprocessor is like this, in this housing is a core which has a high connection it is meant to be a flexible design, and it is not an extra effective as one can do connection in one specifically design for the products for cost effectiveness.
In the case of Xeon processor which contain similar conduct of micro architecture to the pension to integration and multigration is one thing .We achieve cost effectiveness a sense package with which are like this is hard to be going deliver the product so by a basic a whole different design in microprocessor for the basic PC. That is and we have a road map for the extend for the next several years. What I told you on the road map is a evolution of Xeon processor from the introduction in the first period 1998 through and into the first 1999 what you see is the inclusion of inboard cash memory and what you are going to the first half of 1999 the frequency in an excess of 300 MHz going to 333MHz and future time beyond that.
The quality we are showing you for road map of each segment that what you see is in each of the segment we try to use silicon technology to more the microprocessors in the segment we are showing you forward at similarly but different fashion representing different design project as I mentioned a minute ago. We have been benefiting in working on the microprocessor family. It's a new thing a year ago we have more individual specific designing product for this basic PC segment today we have 800 engineers working this on the product family as I showed you previously. The basic PC important as they are only represent a part of product spectrum.
Both the market worldwide in the US, that particularly that I have more details they are the main audiences who buy and value performance as much as the value cost effectiveness. This data is very sensing data and that three quarters are the consumer by public which is the basic PC maybe appeared you have bit enthusiastic category to buy the medium technology made it in different features for the performance PC category. They made accordingly we continue to put a lot of efforts in the development of microprocessor or in the performance for this segment and in that you see. The segmentation to split into the several new categories. I would look noble planning and workstation .
So I want to talk to you about what we do in each of the segment.
The new development in the road map I mentioned you that introduction of is a new micro architecture modified micro architecture new instruction fashion in the first half of 1999. The walking mean we call mean in the product Katmai than the introduce of going to have a fashion brand new at which is hard to pronounce what is the product will be distinguished first by Katmai
product is going to have different instruction for each particular including performance at the desert that is going to we can see at present represents a lot of design effort and the top technology effort leads to higher and higher frequency achieving 500 MHz frequency on the best sort by the first half next year.
The theme for our seminar today is business computing and you might sit and ask particularly might but who need high performance for the desk sort and business desk sort it computer that have today perfectly sufficient to do what I need you to do at work and the fact of matter is it isn't sufficient today and it would become lack and lack sufficient as the work habit of computer users obviously for the evolve and would become more issues particularly the traint that is going to develop and is developing .
And computer user is the visible and sometime not visible, sometime invisible application multi application, sometime when the users open but opposite application is fully visible but often time particularly when this computer resize the network is being the computation of patch in the background that are not even visible to the users at same time require procession performance, procession a said draining performance for the desk sort. This combination of foreground and background computing at same time require procession performance in a subtle and visible fashion laid the performing at micro processor and desk sort in a substantial push run under that I show how they come about you see foreground application that one visible to you and on your screen you see the multi application that you are using among which you are exchange data and background application that are driven by the network by the server that are management by the users only visible to you in that compact the machine the very slow operation even though you had bouncing network that causing that to happen and application that I show you, you will be familiar with that by the characteristic of foreground and background computing is a visible and may be a not so simple browser because whenever you have browser open what you are doing is usually manipulate user is at foreground activity, at same time the browser interest to the server which is connected the obvious information interesting information and that is background operation so every time user one browser you are in beyond one application next it and for foreground and background competition operation. All of this is getting a little complicated. I think it is the proper place they are showing better that telling for back work on anyone else to demonstration of plan.
(表演)
What we are doing of course in another group is setting the same performance that is represented by Pentium machine and bound to movable computing such as the modern employee who is rarely at easy back, contain that application then perform laid with then what we had here is a road map that is similar at somebody behind that performance of the road map against the introduction of Pentium II processor into a movable that capable of movable computer in the package looking this on the movable simply connected on the connection in which the same handage for wide even the back seminar to be included in the notebook both designed specific power more efficiently and are working on a quarter allow the high frequency operation of micro processors you call what you can expect to see in the first half 1999.
That sort in bind machine will be available in a notebook work pact back to operate this aspection on move application are here persperate that is seen the product by brand I mentioned Pentium II Xeon processor the road map is distinguished in suited action first of all for the introduction of Xeon processor in the second half this year. And secondly we will include the inboard memory, unboard cash memory side and of course lend the performance something in accept of another hood are hit the purpose of all these combination are the large cash and the seek forming rapid performance on that sort the illustrate hot filing through work of new tech is the demonstration of what processor than do rapid in that application. The kind of performance that you saw here and cost effectiveness performance from very angle part of industry to the visual they also appear in complete difference like mechanical design automation, electronically design automation and also as modeling their repeated representation of energy can gain information out of overworking among data. I can desert things much acute fashion out of visual representation that out of the seek data .
Let me want to the next segment with service and by way of motivation a project about image of the world of a billion connected computers by one of the things. A billion connected computers reguired can what we mean the service provide the art information that the computers want to act. The key principle here is PC economic the use of industry established (null class) either production we have name and provide very very major cost effectiveness in the users of Intel technology that easy effect of the cost effectiveness growly technology server under charge .
We show an industry sophisticated service and group of service altogether. The product compound that any group at server of people tone but it is made up to two categories intellectual service and other type of service, the intellectual service grows 38% while the other type of that grows 13% quite freely not only is serve population growing but economy should be serviced, standard high service and economic technology faster than the total. One of the challenge that we had over Pentium in the past we have new processors that would be dominant that application users processors of serve application. We did not shin on performance as we have much want to do compared the measurement of transaction for minute on which is commonly used service performance measure and on intellectual service which should be yellow the leading compound over sport machine which is the blue in order that the yellow has reached the richest for excute performance at part of compilation a processor and multiprocessor as well, the difference are not that large performance, are providing excellent performance that difference are large providing performance will be high value economically of feasing industry and we look at the cost performance and fly performance where grow line be economic service provide where more must be into cost performance. That the key challenge we have as we design product service is to acquire all of this quality that make the service as the key ability. The ability of what to ask one way to two way to four way and beyond service and we are making along the flow and we would like to call a mega themselves demonstration of the see ability looks like in a simulation of a real life honest have application.
(表演)
The Pentium processor is pretty yield processor it is ready for a museum at this point given
the later development and in fact we are going flash the work station service product that mentioned earlier is Pentium Ⅱ Xeon processor. Let me tell you and let me illustrate to you how the Xeon processor is begun from the Pentium Ⅱ processor it is in the gemology the next number but it is substantial different in economic for design because in the third segment is was man for hidden the server at work, workshop application that the way that was in the taking off it put the sort we take out Pentium Pro chip and become to 200 MHz we include the frequency and we include the cash by and place brand new Pentium Ⅱ chip we include the section with it and put it on the two sides board into a brand new package scheme that allow the characteristic to reach the outside board and what we have a function to Pentium Ⅱ Xeon processor looking exactly like this and deboard it as the animation showed the Pentium Xeon processor is what is called full the flop full processor crop represents the characteristic it carry move data such as what the need of demonstration. That is the application the cost frequency of course is the segment in all segments more than double the mega frequency that is pullage up to twice as large the inboard cash memory can it match of both management feature will be good.
That are cost founding details map to show as introducing as Pentium II Xeon processor to increasing satisfied at frequency exceeding by the first half of 1999,500 MHz, all of these come in the same method technology as all other microprocessor and therefore their cost effectiveness to the manufacturer and user. Correspondingly as early as indicate to work station software acquire volume to Intel architecture. These are supplemental application all of them higher than former supplemental application has been committed to more the introduction of micro processors in the servers area making up a substantial product so do the number grow everyday.
Beyond this we would begin at very significant mean architecture into 64 implementation what we call IA-64 line. The IA-64 line will be embodied in the first processor decoding at set and but if wanted first in a family that development will go on for many decades (null class) application such as that show the performance of the product Pentium about substantial hardware committed for the variety had about putting them working on the board that is worthy enough. We have on the company that a good number of those before this commitment have perceived a proprietary and rich architecture are special today in the line and move of them are abandoning that total together. Others are incorporating the IA-64 line ,in the line at the same time all of all that they need (null class) and win back a large number of superworld company and superyard product that had been forcing their product the intention to you push you optimizing their product to IA-64 has grown as IA-64 continue to grow similarly to the wrapper of a hard company what is important to realize.
However, we are not watching from the 32 line to 64 line. We are starting a new branch of proper serve in that would provide by the blue line into the IA-64 before higher performance and probably anticipate a high laid increasing performance in the years to come and continue in the extension of 32 bit processor line. Xeon processor as well as the product that are today in development that extend 32 bit line.
Beyond that would comprehend covering the entire feature of computing from the basic and interfecture in the enterprise shade application the truck is pushing to become a unifying architecture of computing. I describe to you the full driving forces at motivate a new one that core optimize at cost post one which is the low cost high performance computing and might ask that is the low cost high performance computing and segment of computing in addition to that connected computing is the second driving force as the development.
I solid presentation by mentioning expectation might target the cash memory to keep in mind the development is world a billion connected computers. As to said to ask what is today want to look like are the worldwide number of interconnected computers we are assuming a number worldwide reaching 150 million connected computers the wondering that some 60% theoretical project number another looked at 150 million equal or organization can exchange information with each other. Interesting enough in the just in the last year to hand rapid development Internet interconnected
Computer in China the operating computer with range full quite range beginning to our million connected computers in China alone . It is important to understand the connect year, begin that year by data were the computing.
Now today data network is achieved by fashion design data network for example the data network illustrated by the net experiment connecting quite similarly at server called the data network we have back the place presentation on the wall as we mentioned earlier, but a large portion of borrowing fashion the world network that design long before anybody for it, for data network and data communication and connected computers that people want the function of sufficient connection between one to another and borrow that complete the network but the picture lists were the portion tat visible above water level. It is the portion that is net work on the telephone network. The last bit were designed be a network the backbone network such as any organizations and companies that once mention like Xeon are data network to handle package communication. And that all the world up to 10% of world work trying that would be replaced by dedicate data network also that you have the limitation for network that you have designed. For that you borrow for data users.
Why is you important why is the internet in connected computers so important for business computing than is a very significant emerging trend which is the business is getting transacted increasingly computing screen to computing string and usual connection between the computing screen internet.
It is a very dynamic phenomena some indicate that 1996 a mount the trade that conducting connected computers made up implement borrow sufficient by that by the early next century 300 billion dollars of commerce representing a very rapid annual trade that I like (null class) demonstration to you is similarly that way.
Electronic commerce business connected string to string on internet connected computers were connected small companies along the world and become as much medium for commerce as fate work processor simple fiction has become the world medium of surprising the medium accounting.
(表演)
Freely the point available that the first year of late 90's a strategy inflection point for the major change in the conditions of conduct business for all types of businesses and the recognition for the increasing spending by business on information technology the charge for holding information technology standing.
The last 3 years are for this is increasing spending business on information technology different countries in the world and let me call you attention to it let free borrow represent from U.S. and percentage of grow the most product the total numbers of 3/4 percentage of center product and each year the percentage is larger than before. Many companies arriving an independent decision in concentrate the significant abasement and development the percentage is larger than before computers on their own business property, computers would go to the significant to the lower percentage level than some of the developed countries particularly in the U.S. that is a concerned development that we can go back econatriangle live birth if that is a burnt a worldwide set of organ dangers than economy that isolate then the fashion.
My single form is the interconnected PC will become for business of all types over the next decade and that in the world at a billion connected computers in that thing information technology is not a luxury, it is strategic necessity.
Thank you very much for your attention!
互聯(lián)網(wǎng)世界中商業(yè)計(jì)算的未來
英特爾公司董事長(zhǎng)兼首席行政官安德魯·葛魯夫博士
女士們、先生們,上海及其他地區(qū)和國(guó)家在電腦屏幕前的觀眾們,早上好!
我們現(xiàn)在正在做一件很特別的事情,那就是通過國(guó)際互聯(lián)網(wǎng)用中文、英文和韓文三種語言實(shí)時(shí)播出今天的活動(dòng)。
首先,我想介紹一下我今天演講的要點(diǎn):人類正在邁向擁有10億臺(tái)互聯(lián)電腦的世界,聯(lián)網(wǎng)的電腦能幫助我們共享信息,使世界各地的人們仿若比鄰,能夠分享數(shù)據(jù),實(shí)現(xiàn)合作,這正是英特爾公司努力的目標(biāo),我們所做的大部分工作就是為了實(shí)現(xiàn)這一目標(biāo)。
當(dāng)今,推動(dòng)信息產(chǎn)業(yè)發(fā)展的的兩大動(dòng)力,一是低成本、高性能的計(jì)算,另一個(gè)就是全球的互聯(lián)網(wǎng),信息產(chǎn)業(yè)已成為世界支柱的產(chǎn)業(yè)。1998年,全球電腦銷量達(dá)一億臺(tái),比電視機(jī)的銷量要高得多,全球電腦產(chǎn)業(yè)實(shí)現(xiàn)了長(zhǎng)足發(fā)展。特別是中國(guó),電腦市場(chǎng)的平均年增長(zhǎng)率為23%,而且在未來幾年中會(huì)保持這一速度。根據(jù)不同估算,中國(guó)今年的電腦銷量將達(dá)400萬臺(tái)。中國(guó)的電腦市場(chǎng)具有巨大的潛力,規(guī)模正在不斷擴(kuò)大。電腦越來越為各行業(yè)和社會(huì)各界所接受,人們對(duì)互聯(lián)網(wǎng)絡(luò)也開始顯示越來越多的興趣。
在微處理器方面,隨著計(jì)算技術(shù)的不斷進(jìn)步,微處理器也在不斷升級(jí)。我們開發(fā)這些技術(shù),而且根據(jù)市場(chǎng)的變化不斷更新這些技術(shù)。今天,英特爾在全球的架構(gòu)體系正在展開新的計(jì)算市場(chǎng)細(xì)分,下面我來具體講一下這方面的情況:
在初期,計(jì)算機(jī)產(chǎn)業(yè)最基本的就是臺(tái)式機(jī)。隨著時(shí)間的不斷推移,電腦市場(chǎng)不斷地細(xì)分為臺(tái)式機(jī)、服務(wù)器、筆記本電腦等等,而且這種分類越來越細(xì)。我們可以看到,英特爾的微處理器能夠滿足不斷提升的計(jì)算需要,如超級(jí)計(jì)算和企業(yè)服務(wù)器等。同時(shí),我們也有標(biāo)準(zhǔn)產(chǎn)量的處理器,使業(yè)內(nèi)人們所需的大批量服務(wù)器的性能不斷提高,成本不斷降低。隨著電腦市場(chǎng)不斷細(xì)分,高性能的電腦和基本電腦都在不斷發(fā)展,工作站代表著新興的微處理器市場(chǎng),而一些相對(duì)簡(jiǎn)單的電腦也由于微處理器的發(fā)展而不斷發(fā)展。
作為微處理器的設(shè)計(jì)者和制造商,英特爾面臨的挑戰(zhàn)是,根據(jù)不同市場(chǎng)的需求提供不同的產(chǎn)品。我們?nèi)沃囟肋h(yuǎn),我們必須不斷學(xué)習(xí)、不斷調(diào)整、不斷設(shè)計(jì)出形態(tài)各異的微處理器,并不斷更新產(chǎn)品性能,以滿足不同市場(chǎng)的具體需要。這需要大量的努力和大量的投資,由于我們堅(jiān)信這個(gè)市場(chǎng)具有良好的前景,因此我們不斷在設(shè)備、廠房、研發(fā)等多方面增加投資,而且隨著時(shí)間的推移,投資數(shù)額持續(xù)上升。雖然我們不知道1998年的具體數(shù)字會(huì)達(dá)到多少,但預(yù)計(jì)在80億美元左右。
在滿足不同市場(chǎng)具體需求的同時(shí),英特爾的宗旨就是將同樣的微架構(gòu)運(yùn)用于所有的應(yīng)用之中,這就是我們所說的P6位架構(gòu)。大家一定要記?。涸诓煌漠a(chǎn)品中,英特爾用的是同樣的P6位架構(gòu),但每個(gè)產(chǎn)品卻是針對(duì)特定市場(chǎng)需要而設(shè)計(jì)的,以保證各市場(chǎng)中人們對(duì)性能和價(jià)格/性能比的要求。
最近,英特爾公司決定根據(jù)不同的市場(chǎng)細(xì)分來發(fā)展公司的品牌戰(zhàn)略。過去幾年中,奔騰(R)一直是英特爾微處理器最主要的品牌,但是我們覺得在當(dāng)今時(shí)代,在如此多樣化的情況下,光靠一個(gè)品牌是不夠的,所以我們已經(jīng)引入了英特爾公司的品牌計(jì)劃,我們又了另外兩個(gè)品牌,即主要針對(duì)基本電腦市場(chǎng)的英特爾賽揚(yáng)(TM)處理器和主要針對(duì)高端市場(chǎng),滿足工作站、服務(wù)器需求的奔騰IIXeon(TM)處理器。
在計(jì)算方面,英特爾公司為基本電腦所設(shè)計(jì)的處理器基于P6微架構(gòu),除此之外我們還使用了巧妙的集成,在芯片上以更為經(jīng)濟(jì)有效的方式放上越來越多的晶體管,為保持成本有效性,我們?cè)谕还杵?,將一些不同的功能集中在一起。從歷史角度講,我們一直在做這樣的工作,例如,十年前當(dāng)386是主要架構(gòu)時(shí),我們發(fā)現(xiàn)隨著晶體管數(shù)的增長(zhǎng),我們可以把那些必須由一個(gè)單獨(dú)的芯片才能完成的功能放在新一代的處理器上,并且是放在同一個(gè)芯片上,比如說將原來一些單獨(dú)芯片的功能集中在486上。這樣,我們不斷使微處理器的功能越來越多,這就是我們所說的集成。我們希望通過集成,有選擇地在我們的設(shè)計(jì)中不斷增加微處理器的功能,以滿足基本計(jì)算的需求。
除此之外,我們還要考慮成本效益,我們已經(jīng)采用了一種低成本封裝的技術(shù)。我們?cè)谫悡P(yáng)處理器上做了一些巧妙的集成,從而在保證功能的同時(shí),降低成本。
我們看看針對(duì)基本電腦的英特爾賽揚(yáng)處理器的發(fā)展趨勢(shì),可以發(fā)現(xiàn),二級(jí)高速緩存的變化,從沒有直到128K,主頻速度在1999年上半年將超過300MHz,估計(jì)可達(dá)到333MHz,將來甚至還會(huì)更高。
在基本電腦方面,去年英特爾公司還沒有什么人專門從事研究,但現(xiàn)在已有800位工程師從事基本電腦微處理器的設(shè)計(jì)和研究。
全球范圍內(nèi),特別是在美國(guó),我們發(fā)現(xiàn),客戶對(duì)性能和成本效益同樣重視。75%的消費(fèi)者購買高性能電腦或基本電腦,25%的人購買發(fā)燒級(jí)電腦。所以,我們不斷在高性能電腦上下功夫,不斷在主流電腦、移動(dòng)電腦和工作站方向努力。
我將簡(jiǎn)單介紹一下英特爾公司在這三個(gè)方面將如何行動(dòng)。
在主流臺(tái)式電腦方面,英特爾將于1999年上半年,推出新的、代號(hào)為Katmai的處理器。我們主要是進(jìn)一步加強(qiáng)多媒體功能,而且我們還要作很多設(shè)計(jì)和處理器方面的進(jìn)一步研究,從而能夠提高主頻速度,使臺(tái)式機(jī)的主頻達(dá)到500MHz。
我們今天探討的主題是商務(wù)計(jì)算。大家也許會(huì)說,這一切都不錯(cuò),但究竟誰需要高性能的商務(wù)臺(tái)式機(jī)呢?我們當(dāng)今的電腦是不是已經(jīng)夠用了?實(shí)際上,當(dāng)今的臺(tái)式機(jī)能力還很不夠,而且隨著用戶使用習(xí)慣的改變,它會(huì)變得越來越不夠用。
實(shí)際上,使用電腦應(yīng)用分為可見與不可見兩種。當(dāng)用戶打開幾個(gè)不同的應(yīng)用時(shí),他所執(zhí)行的應(yīng)用是可以看得見的。但如果電腦處在網(wǎng)絡(luò)中,那么網(wǎng)絡(luò)會(huì)在后臺(tái)執(zhí)行很多計(jì)算任務(wù),雖然用戶看不見這些應(yīng)用和任務(wù),但它卻需要電腦有很高的性能,這種前臺(tái)和后臺(tái)的計(jì)算實(shí)際上是很微妙的。前臺(tái)應(yīng)用指用戶交換數(shù)據(jù)等等,后臺(tái)應(yīng)用指用戶通??床灰姷姆?wù)器或網(wǎng)絡(luò)方面的應(yīng)用。有時(shí),即使用戶沒在電腦上干什么,但由于有后臺(tái)應(yīng)用,臺(tái)式機(jī)的速度也會(huì)降低。以瀏覽器為例,當(dāng)用戶打開瀏覽器后,他所做的一切都是前臺(tái)應(yīng)用,但同時(shí),瀏覽器要與服務(wù)器聯(lián)系,服務(wù)器又需要與網(wǎng)絡(luò)交流,以便提取信息、交流信息。總之,人們每次使用瀏覽器時(shí)都同時(shí)使用了前臺(tái)和后臺(tái)應(yīng)用。只有高性能的電腦才能滿足商業(yè)用戶每天的大量任務(wù)需求。
在移動(dòng)計(jì)算方面,英特爾已經(jīng)發(fā)布了移動(dòng)奔騰II處理器,使筆記本電腦的性能可以與主流臺(tái)式機(jī)媲美。目前移動(dòng)奔騰II處理器的主頻為233和266MHz,到1999年上半年會(huì)達(dá)到333MHz。
在工作站方面,我們會(huì)推出奔騰IIXeon處理器,二級(jí)高速緩存會(huì)達(dá)到1M,主頻也會(huì)增加到500MHz左右,從而使電腦的功率更大。
我們?cè)賮砜纯捶?wù)器。我們剛才談到了全球?qū)?huì)有10億臺(tái)互聯(lián)的電腦。而10億臺(tái)互聯(lián)的電腦就需要數(shù)千萬臺(tái)服務(wù)器為人們提供服務(wù)。標(biāo)準(zhǔn)產(chǎn)量服務(wù)器將能提供優(yōu)異的成本效益,使信息技術(shù)的用戶獲益匪淺。
業(yè)內(nèi)統(tǒng)計(jì)表明,服務(wù)器的年增長(zhǎng)率為33%。服務(wù)器市場(chǎng)由英特爾架構(gòu)的服務(wù)器和非英特爾架構(gòu)的服務(wù)器組成,前者的年增長(zhǎng)率為38%,后者為13%,很明顯,服務(wù)器的增長(zhǎng)主要來自英特爾架構(gòu)服務(wù)器。我們目前還面臨很多挑戰(zhàn),因?yàn)檫^去我們的設(shè)計(jì)主要是針對(duì)臺(tái)式機(jī),但研究表明,無論在單處理器還是多處理器技術(shù)方面,英特爾架構(gòu)的服務(wù)器性能已經(jīng)達(dá)到和超過了非英特爾架構(gòu)的服務(wù)器,也就是說英特爾架構(gòu)提供卓越的性能。特別需要指出的是,英特爾在提供高性能的同時(shí),也提供了計(jì)算機(jī)行業(yè)人們所需要的規(guī)模經(jīng)濟(jì)的效果,從而實(shí)現(xiàn)了優(yōu)秀的性能/價(jià)格比。
高能奔騰處理器推出已有三年,隨著當(dāng)今技術(shù)日新月異的發(fā)展,它該進(jìn)博物館了。如前所說,我們現(xiàn)在已經(jīng)有了奔騰IIXeon處理器,它針對(duì)高帶寬的服務(wù)器和工作站方面的應(yīng)用,這與其前一代的高能奔騰處理器不一樣,我們用具有MMX(TM) 技術(shù)的奔騰IIXeon處理器代替最多只能有200MHz的高能奔騰處理器。我們提高了頻率,提高了高速緩存,而且使用了新的封裝技術(shù),使得輸入/輸出端變得更合理。奔騰IIXeon處理器是我們所說的第一個(gè)Slot2(插槽2)的處理器,Slot2主要代表輸入/輸出特性,它能夠處理更多的數(shù)據(jù)。除此之外,它還有100MHz的系統(tǒng)總線,處理器的頻率也比高能奔騰超過兩倍,緩存多了一倍,它有全速的緩存總線,也有更好的可管理性。
我們?cè)谝氡简vIIXeon處理器時(shí),不妨看看中檔到高檔服務(wù)器的處理器趨勢(shì)。我們可以看到,到1999年的上半年,它會(huì)超過500MHz,這將代表著處理器方面的重大突破,而且其價(jià)格/性能比,無論對(duì)制造商還是用戶來說,都會(huì)非常好。
同樣,在工作站方面,量產(chǎn)服務(wù)器使主要軟件移向英特爾架構(gòu)。在服務(wù)器方面,所有重要的軟件公司已經(jīng)致力于轉(zhuǎn)向英特爾的架構(gòu),從而推出進(jìn)一步的新應(yīng)用。這樣的軟件公司數(shù)量也越來越多。
我們也在致力于將英特爾架構(gòu)進(jìn)一步轉(zhuǎn)向IA-64(TM)(英特爾64位架構(gòu))。代號(hào)為Merced(TM)的處理器會(huì)成為IA-64家族的第一位成員,它將在市場(chǎng)上遙遙領(lǐng)先。在硬件方面,它與32位處理器全面二進(jìn)制兼容,而且它有優(yōu)異的浮點(diǎn)運(yùn)算和三維圖像架構(gòu)。它和奔騰IIXeon處理器的設(shè)計(jì)初衷一致,有更大的內(nèi)存,更佳的智能型輸入/輸出特性,四倍AGP加速圖像端口、圖像帶寬,滿足服務(wù)器和工作站市場(chǎng)的需求。該產(chǎn)品的性能已經(jīng)使得各軟、硬件廠商作出承諾,它們都在向IA-64位架構(gòu)轉(zhuǎn)移。
不過,這并不表明英特爾放棄了32位,只做64位。我們的產(chǎn)品將會(huì)兼顧64位和32位,我們很多新的品牌也將繼續(xù)沿著32位的方向發(fā)展。
未來,英特爾架構(gòu)將成為商務(wù)計(jì)算的統(tǒng)一架構(gòu)。剛才我們談到了信息產(chǎn)業(yè)的兩大動(dòng)力。高性能、底成本的計(jì)算無疑是一大動(dòng)力,而聯(lián)網(wǎng)則是信息產(chǎn)業(yè)的第二大動(dòng)力。
我們的目標(biāo)是向10億臺(tái)聯(lián)網(wǎng)的電腦邁進(jìn)。目前全球大約有1.5億臺(tái)互聯(lián)的電腦,這距我們的目標(biāo)還差得很遠(yuǎn),但其發(fā)展速度卻非常快。在中國(guó),已有100萬臺(tái)聯(lián)網(wǎng)電腦。國(guó)際互聯(lián)網(wǎng)為人們提供了無所不在的數(shù)據(jù)網(wǎng)絡(luò)。
今天互聯(lián)網(wǎng)是商用數(shù)據(jù)網(wǎng)和撥號(hào)語音網(wǎng)技術(shù)的結(jié)合。對(duì)消費(fèi)者而言,互聯(lián)網(wǎng)是便捷的數(shù)據(jù)交換網(wǎng)。但互聯(lián)網(wǎng)的總體情況猶如冰山,使用電話撥號(hào)上網(wǎng),是我們平時(shí)能夠看見的、浮在水面上的,這部分只占整個(gè)互聯(lián)網(wǎng)的10%。而冰山的大部分在水底,我們不易看到,這一部分就是ISP的主干網(wǎng),它需要巨大的投資,占整個(gè)互聯(lián)網(wǎng)的90%。
互聯(lián)網(wǎng)的重要性非同尋常,因?yàn)殡S著商務(wù)交易的方式越來越多地從傳統(tǒng)的方式過渡到目前的電腦屏幕到屏幕,轉(zhuǎn)變到聯(lián)網(wǎng)交易,互聯(lián)網(wǎng)將對(duì)我們的生活和工作產(chǎn)生重大的影響。
1996年,大約有近80億美元的交易是通過網(wǎng)上進(jìn)行的。預(yù)計(jì)到2000年,網(wǎng)上交易將達(dá)到3000多億美元?;ヂ?lián)網(wǎng)使全世界所有中小企業(yè)連接起來,人們可以用它調(diào)取信息、下定單、查庫存、安排送貨、與各國(guó)不同企業(yè)取得聯(lián)系等等。
我相信這樣的商務(wù)形式將會(huì)象80年代文字處理器一樣,變得相當(dāng)普及。我想指出的是,90年代末將是所有企業(yè)的戰(zhàn)略轉(zhuǎn)折點(diǎn)。不管是何種企業(yè),其企業(yè)行為都將發(fā)生變化。認(rèn)識(shí)到這一點(diǎn)將至關(guān)重要,企業(yè)如果認(rèn)識(shí)到這點(diǎn),它們就會(huì)在信息產(chǎn)業(yè)上作越來越多的投資。
從過去幾年中各國(guó)信息產(chǎn)業(yè)投資占國(guó)內(nèi)生產(chǎn)總值的比例數(shù)來看,美國(guó)以4%處于領(lǐng)先地位,從1995到1997年該百分比一直在增長(zhǎng)。許多公司考慮到互聯(lián)網(wǎng)和互聯(lián)電腦的飛速發(fā)展,已經(jīng)不斷加大了其在信息產(chǎn)業(yè)方面的投資。中國(guó)的信息技術(shù)投資占國(guó)內(nèi)生產(chǎn)總值的比例約為1%,這要低于其他一些國(guó)家,但是這一百分比也在不斷增長(zhǎng)。很顯然,如果在信息技術(shù)投資方面出現(xiàn)黑洞的話,就會(huì)面臨很多問題。
我認(rèn)為互聯(lián)網(wǎng)上的聯(lián)網(wǎng)電腦,將會(huì)成為下一個(gè)十年中企業(yè)在市場(chǎng)上制勝的關(guān)鍵,所有企業(yè)都是如此。在一個(gè)擁有10億臺(tái)互聯(lián)電腦的世界上,在信息技術(shù)上投資并不是奢侈的消費(fèi),而是戰(zhàn)略的必需。
謝謝大家!