【正文】 比單獨(dú)晶體管高很多。大學(xué)或各地的研究者，可以充分發(fā)揮他們的想象力構(gòu)造任何微機(jī)可以連接的數(shù)字設(shè)備。在劍橋大學(xué)實(shí)驗(yàn)室力，我們構(gòu)造了CAP，一個(gè)有令人驚奇邏輯能力的微機(jī)。7400在70年代中期還不斷發(fā)展壯大，并且被寬帶局域網(wǎng)的先驅(qū)組織Cambridge Ring所采用。令牌環(huán)設(shè)計(jì)研究的發(fā)表先于以太網(wǎng)。在這兩種系統(tǒng)出現(xiàn)之前，人們大多滿(mǎn)足于基于電報(bào)交換機(jī)的本地局域網(wǎng)。令牌環(huán)網(wǎng)需要高可靠性，由于脈沖在令牌環(huán)中傳遞，他們必須不斷的被放大并且再生。是7400的高可靠性給了我們勇氣，使得我們著手Cambridge 。精簡(jiǎn)指令計(jì)算機(jī)的誕生早期的計(jì)算機(jī)有簡(jiǎn)單的指令集，隨著時(shí)間的推移，商業(yè)用微機(jī)的設(shè)計(jì)者增加了另外的他們認(rèn)為可以微機(jī)性能的特性。很少的測(cè)試方法被建立，總的來(lái)說(shuō)特性的選取很大程度上依賴(lài)于設(shè)計(jì)者的直覺(jué)。1980年，RISC運(yùn)動(dòng)改變了微機(jī)世界。該運(yùn)動(dòng)是由Patterson 和 Ditzel發(fā)表了一篇命名為精簡(jiǎn)指令計(jì)算機(jī)的情況論文而引起的。除了RISC這個(gè)引人注目縮略詞外，這個(gè)標(biāo)題傳達(dá)了一些指令集合設(shè)計(jì)的見(jiàn)解，隨之引發(fā)了RISC運(yùn)動(dòng)。從某種意義上說(shuō)，它推動(dòng)了線(xiàn)程的發(fā)展，在處理器中，同一時(shí)間有幾個(gè)指令在不同的執(zhí)行階段稱(chēng)為線(xiàn)程。線(xiàn)程不是個(gè)新概念，但是它對(duì)微機(jī)來(lái)說(shuō)是從未有過(guò)的。RISC受益于一個(gè)最近的可用的方法的誕生，該方法使估計(jì)計(jì)算機(jī)性能成為可能而不去真正實(shí)現(xiàn)該微機(jī)的設(shè)計(jì)。我的意思是說(shuō)利用目前存在的功能強(qiáng)大的計(jì)算機(jī)去模擬新的設(shè)計(jì)。通過(guò)模擬該設(shè)計(jì)，RISC的提倡者能夠有信心的預(yù)言，一臺(tái)使用和傳統(tǒng)計(jì)算機(jī)相同電路的RISC計(jì)算機(jī)可以和傳統(tǒng)的最好的計(jì)算機(jī)有同樣的性能。模擬仿真加快了開(kāi)發(fā)進(jìn)度并且被計(jì)算機(jī)設(shè)計(jì)者廣泛采用。隨后，計(jì)算機(jī)設(shè)計(jì)者變的多些可理性少了一些藝術(shù)性。今天，設(shè)計(jì)者們希望有滿(mǎn)屋可用計(jì)算機(jī)做他們的仿真，而不只是一臺(tái)，單片機(jī)芯片每次的縮小，芯片數(shù)量將減少；并且芯片間的導(dǎo)線(xiàn)也隨之減少。這導(dǎo)致了整體速度的下降，因?yàn)樾盘?hào)在各個(gè)芯片間的傳輸時(shí)間變長(zhǎng)了。漸漸地，芯片的收縮到只剩下處理器部分，緩存都被放在了一個(gè)單獨(dú)的片子上。這使得工作站被建成擁有當(dāng)代小型機(jī)一樣的性能，結(jié)果搬倒了小型機(jī)絕對(duì)的基石。正如我們所知道的，這對(duì)于計(jì)算機(jī)工業(yè)和從事計(jì)算機(jī)事業(yè)的人產(chǎn)生了深遠(yuǎn)的影響自從上述時(shí)代的開(kāi)始，高密度CMOS硅芯片成為主導(dǎo)。隨著芯片的縮小技術(shù)的發(fā)展，數(shù)百萬(wàn)的晶體管可以放在一個(gè)單獨(dú)的片子上，相應(yīng)的速度也成比例的增加。為了得到額外的速度。處理器設(shè)計(jì)者開(kāi)始對(duì)新的體系構(gòu)架進(jìn)行實(shí)驗(yàn)。一次成功的實(shí)驗(yàn)都預(yù)言了一種新的編程方式的分支的誕生。我對(duì)此取得的成功感到非常驚奇。它導(dǎo)致了程序執(zhí)行速度的增加并且其相應(yīng)的框架。同樣令人驚奇的是，通過(guò)更高級(jí)的特性建立一種單片機(jī)是有可能的。例如，為IBM Model 91開(kāi)發(fā)的新特性，現(xiàn)在在單片機(jī)上也出現(xiàn)了。Murphy定律仍然在中止的狀態(tài)。它不再適用于使用小規(guī)模集成芯片設(shè)計(jì)實(shí)驗(yàn)用的計(jì)算機(jī)，例如7400系列。想在電路級(jí)上做硬件研究的人們沒(méi)有別的選擇除了設(shè)計(jì)芯片并且找到實(shí)現(xiàn)它的辦法。一段時(shí)間內(nèi)，這樣是可能的，但是并不容易。不幸的是，制造芯片的花費(fèi)有了戲劇性的增長(zhǎng)，主要原因是制造芯片過(guò)程中電路印刷版制作成本的增加。因此，為制作芯片技術(shù)追加資金變的十分困難，這是當(dāng)前引起人們關(guān)注的原因。內(nèi)存技術(shù)的進(jìn)步非傳統(tǒng)的CMOS變革了存儲(chǔ)器技術(shù)。直到現(xiàn)在，我們?nèi)匀灰揽緿RAM作為主要的存儲(chǔ)體。不幸的是，隨著芯片的縮小，只有芯片外圍速度上的增長(zhǎng)——處理器芯片和它相關(guān)的緩存速度每?jī)赡暝黾右槐丁＿@就是存儲(chǔ)器代溝并且是人們焦慮的根源。存儲(chǔ)技術(shù)的一個(gè)可能突破是，使用一種非傳統(tǒng)的CMOS管，在計(jì)算機(jī)整體性能上將導(dǎo)致一個(gè)很大的進(jìn)步，將解決大存儲(chǔ)器的需求，即緩存不能解決的問(wèn)題。也許這個(gè)。電子的不足盡管目前為止，電子每表現(xiàn)出明顯的不足，然而從長(zhǎng)遠(yuǎn)看來(lái)，它最終會(huì)不能滿(mǎn)足要求。也許這是我們開(kāi)發(fā)非傳統(tǒng)CMOS管的原因。在Cavendish實(shí)驗(yàn)室里，Haroon Amed已經(jīng)作了很多有意義的工作，他們想通過(guò)一個(gè)單獨(dú)電子或多或少的表現(xiàn)出0和1的區(qū)別。然而對(duì)于構(gòu)造實(shí)用的計(jì)算機(jī)設(shè)備只取得了一點(diǎn)點(diǎn)進(jìn)展。也許由于偶然的好運(yùn)氣，數(shù)十年后一臺(tái)基于一個(gè)單獨(dú)電子的計(jì)算機(jī)也許是可以實(shí)現(xiàn)的。Progress in ComputersPrestige Lecture delivered to IEE, Cambridge, on 5 February 2004Maurice WilkesComputer LaboratoryUniversity of CambridgeThe calculation of the first stored program began in 1950 before and after, it is at the University of Cambridge in the summer of 1949, we created Delay Storage Automatic puter (EDSAC).The initial experiment with the puter is constructed like me, like with a broad knowledge. We have a wealth of experience in electronic engineering, and we believe that these experiences of great benefit to us. Later, proved to be correct, although we have to learn many new things. The most important thing is transient must be careful to cope with, although it only on a TV screen with a harmless flash, but on the puter, which will lead to a series of the circuit design process, we are often a dilemma. For example, I can use the vacuum diode as a gate, as in the EDSAC, or with a control signal between the two grid with five tube, which is widely used for the design of other systems, which The class selection has continued until the logic gate applications. The people who work in the puter field should remember the TTL, ECL and CMOS, so far, CMOS has occupied a dominant position in the first few years, the IEE, (Electronic Engineers) is still occupied by the Power Engineering. In order to allow the the IEE recognize the wireless engineering and rapid development of electronic engineering to develop in parallel its own right, we have to face some obstacles. Power engineers to work with us, we also encountered many difficulties. Some people angry, the IEE published papers are expected to be a lengthy statement of the early research at the beginning, is nothing more than a statement of the class of the difficulties encountered in the early stages do not have much experience.The consolidation phase of the 1960sThe early 1960s, personal hero, the puter is the real cause of attention. The number of puters in the world has increased a lot, and more reliable performance than ever before. These my view, be attributed to the birth of the startup and the first operating system with highlevel language. Timesharing system started, and puter graphics summary, transistors began to replace the air traffic control. Change is an unavoidable challenge to the engineers at the time. They must forget that they are familiar with the circuit restart. Can only say to them the courage to accept the challenge, although this change and will not be integrated circuits and miniputersSoon, in a silicon chip can put more than one transistor, the resulting integrated circuit was born. With the passage of time, a film to be able to acmodate the maximum number of transistors, or a little less logic gates and flip the door of integration reached a maximum extent. Hence there we know the 7400 series microprocessor. Each gate or flip circuit is independent and has its own pin. Through the wires together, made a puter or other chips for the manufacture of a new puter may be provided. It is known as miniputers. Less satisfactory than the mainframe, but powerful, and more people to afford it. A business sector or universities have the ability to have a miniputer rather than a large organization required expensive the puter the beginning of the epidemic and improve the function of the world eager to obtain its puting power, but always can not be due to industrialscale supply and substantial price setbacks. The emergence of the puter to solve this calculation of the consumption decline not originate with the puter, it is supposed to be that way. I mentioned in the summary of inflation in the puter industry embarked on wrong track . As time goes by, more people than they pay money to described the times for the people engaged in the research of puter hardware is amazing era. 7400 series users can work in the logic gate and switch level and chip integrated reliability is much higher than a single transistor. Universities or researchers around, you can give full play to their imagination to construct any puter can connect digital devices. Force in the University of Cambridge laboratory, we constructed the CAP, a