新聞中心

EEPW首頁(yè) > 物聯(lián)網(wǎng)與傳感器 > 設(shè)計(jì)應(yīng)用 > 性能之路: 可互換的8位與32位微處理器

性能之路: 可互換的8位與32位微處理器

——
作者: 時(shí)間:2007-07-27 來(lái)源:EDN 收藏
隨著科技的迅猛發(fā)展,開(kāi)發(fā)者承受了巨大的壓力。由于市場(chǎng)日益復(fù)雜的應(yīng)用需求,開(kāi)發(fā)者們經(jīng)常面對(duì)著性能不足的狀況,于是不得不將目光投向性能更好的。如今,使用4位的開(kāi)發(fā)者都在尋找劃算的微處理器替代;而用的開(kāi)發(fā)者則轉(zhuǎn)向兼容DSC(數(shù)字信號(hào)控制器)和的選擇。造成這種狀況的原因有很多,而日益增長(zhǎng)的內(nèi)存容量以及在同樣價(jià)位微處理器性能的不斷提高更進(jìn)一步加劇的這種趨勢(shì)。

  微處理器市場(chǎng)仍在振蕩,因?yàn)?位機(jī)價(jià)格、功耗以及代碼高效的優(yōu)勢(shì),每天都會(huì)有新的應(yīng)用。但是對(duì)更高級(jí)的應(yīng)用,市場(chǎng)壓力不斷提升性能需求及內(nèi)存容量。例如,家用型產(chǎn)品需要用到變速電機(jī)控制,這就要求更高性能更大存儲(chǔ)的設(shè)備。8位微處理器片上flash存儲(chǔ)器最多到128KB,而最新的一些嵌入式處理器則擁有3MB大小的閃存,提供更多編程空間處理復(fù)雜的控制算法。

  半導(dǎo)體廠商嘗試幫助開(kāi)發(fā)者向32位設(shè)備轉(zhuǎn)變,獲取更高性能。低檔32位微處理器與高檔8位微處理器價(jià)格相當(dāng),因此單位成本不是阻礙。真實(shí)的費(fèi)用來(lái)源于轉(zhuǎn)變過(guò)程本身,包括確保多年的經(jīng)驗(yàn)?zāi)軌虮A?。在投資中,開(kāi)發(fā)工具占了很重要的角色。

  32位微處理工具天生就比8位微處理器更加復(fù)雜和昂貴。芯片制造者及其團(tuán)隊(duì)始終為簡(jiǎn)化32位開(kāi)發(fā)工具組件和降低基本評(píng)估板價(jià)格到$500以下而工作。然而,開(kāi)發(fā)者仍忙于大量芯片的結(jié)構(gòu)差異。例如:

  8位和32位微處理器的外圍提供類(lèi)似功能,但是實(shí)際功能性和用戶可用的程序模塊很可能不同。開(kāi)發(fā)者必須重寫(xiě)與芯外設(shè)備連接的低端設(shè)備驅(qū)動(dòng)。

  工程師必須解決設(shè)計(jì)差異,比如中斷能力、默認(rèn)變量長(zhǎng)度以及內(nèi)存中地址的分配。同樣,任何一個(gè)定時(shí)循環(huán)也有可能重新改寫(xiě)。

  無(wú)論32位開(kāi)發(fā)工具多么簡(jiǎn)單和容易使用,對(duì)新開(kāi)發(fā)者而言都需要繼續(xù)學(xué)習(xí)來(lái)掌握。

  性能及內(nèi)存的要求會(huì)迫使開(kāi)發(fā)者從8位到32位微處理器轉(zhuǎn)變,但是他們必須面對(duì)上市時(shí)間延遲和開(kāi)發(fā)費(fèi)用增加的挑戰(zhàn)。新替代方案是重新設(shè)計(jì)32位微處理器無(wú)論在外形還是在功能上都更接近8位微處理器。同樣的外圍設(shè)備、輸出引腳以及單一開(kāi)發(fā)工具套件裝置,既適合8位也適合32位微處理器,產(chǎn)生真正管教兼容的32位設(shè)備來(lái)替代8位產(chǎn)品。

  想要轉(zhuǎn)向32位微處理器的設(shè)計(jì)者面對(duì)許多的挑戰(zhàn),現(xiàn)在能在芯片設(shè)計(jì)層面處理好。例如,典型8位設(shè)備可能只有一個(gè)信號(hào)調(diào)試接口,而典型32位設(shè)備需要很多接口。重新設(shè)計(jì)32位的接口使其成為一個(gè)引腳可以大大提高兼容性從而降低了開(kāi)發(fā)者的工作難度。

  我們的最終目標(biāo)是8位與32位微處理器使用同樣的封裝、同樣的開(kāi)發(fā)板以及同樣的開(kāi)發(fā)工具。評(píng)估板插槽上的8位的設(shè)備替換為32位設(shè)備,重新編譯代碼,就擁有一個(gè)32位的系統(tǒng)。這就是我們Freescale的方向。而更高性能的另一個(gè)方法是8位到32位控制器合為一體,展現(xiàn)出大部分吸引人的替代——這是從前沒(méi)有嘗試過(guò)的。

  轉(zhuǎn)向32位性能呈上升勢(shì)頭。無(wú)論選用了哪種設(shè)備和開(kāi)發(fā)工具,優(yōu)勢(shì)在那些花更少時(shí)間開(kāi)發(fā)而更多時(shí)間推銷(xiāo)新32位微處理器應(yīng)用上。

  英文原文:

  The path to performance: Interchangeable 8- and 32-bit microcontrollers

  The ultimate goal is to develop 8- and 32-bit MCUs that share the same socket, the same board, and the same tools.

  By Mike McCourt, Freescale Semiconductor -- EDN, 2/14/2007

  Technology races forward, and it's putting a lot of pressure on embedded-system developers. Competing in markets with increasingly complex applications, developers often face performance shortfalls that cause them to turn to increasingly higher-performance microcontrollers. Developers using 4-bit devices today are looking for cost-effective 8-bit alternatives, and 8-bit users are turning to compatible DSC (digital signal controller) and 32-bit options. There are many reasons for this migration, which is being further enabled by products with increased memory capacities and improved performance at comparable price levels.

The 8-bit market remains vibrant, with many new applications being unearthed daily that take advantage of the price, power, and code-efficiency benefits of 8-bit microcontrollers. But for advanced applications, market pressures are driving up performance needs and memory sizes. For example, household appliances are now using variable-spe

ed motor control, which requires the higher performance and larger memories of 32-bit devices. On-chip flash memory for 8-bit microcontrollers generally maxes out at 128 kbytes, whereas some of the latest 32-bit embedded processors integrate as much as 3 Mbytes of flash, providing more programming headroom to handle sophisticated control algorithms.

  Semiconductor manufacturers are trying to help developers transition to 32-bit devices by easing the migration to higher performance. Low-end 32-bit microcontrollers are priced comparably with high-end 8-bit devices, so unit cost is less of a roadblock. The real expense comes in the transition process itself, including making sure that years of previous experience are retained. Development tools play a major role in this investment.

  32-bit development tools are inherently more complex and more expensive than 8-bit tools. Chipmakers and their partners have been working to simplify their 32-bit tool suites and reduce the cost for basic evaluation boards to less than $500. However, developers must still address a number of architectural differences between chips. For example:

  Peripherals on 8- and 32-bit devices may provide similar functions, but the exact functionality and user-visibility programming models will likely be different. Developers will need to rewrite and reverify lower-level device drivers that interface to the on-chip peripheral hardware.

  Engineers must address architectural differences, such as interrupt capabilities and default variable sizes and their layout in memory. Also, any timing loops will probably require rewriting.

  No matter how simple and easy-to-use the 32-bit tools are, they often require an extended learning curve for new developers to master.

  Performance and memory requirements may force developers to migrate from 8-bit to 32-bit microcontrollers, but the challenges they then face can lead to time-to-market delays and increased development costs. A novel alternative is

to redesign a 32-bit microcontroller to look and act more like an 8-bit device. Identical peripherals and pin-outs and a single tool set suitable for both 8- and 32-bit devices can produce a 32-bit device that's a true pin-compatible, drop-in replacement for an 8-bit product.

  Many of the challenges facing designers who want to migrate to 32-bit now can be dealt with at the chip-design level. For instance, a typical 8-bit device may have only a single-pin debugging interface, while a 32-bit device typically requires many more pins. Remapping the 32-bit interface to one pin can greatly enhance compatibility and make the developer's job easier.

  The ultimate goal is to develop 8- and 32-bit microcontrollers that share the same socket, the same board, and the same tools. You can take the 8-bit part out of your evaluation-board socket, drop in the 32-bit part, recompile the code, and have a functional 32-bit system. This is the direction we are taking at Freescale. While other paths to higher performance exist, our 8- to 32-bit controller continuum shows great promise as the most attractive alternative—one that has never been tried before.

  The migration to 32-bit performance is gaining momentum. Regardless of which devices and tools developers choose, the advantage goes to those who spend less time developing and more time selling their new 32-bit applications.

  英文原文地址:http://www.edn.com/article/CA6415134.html



評(píng)論


相關(guān)推薦

技術(shù)專(zhuān)區(qū)

關(guān)閉