Distinguishing Between Embedded and General-Purpose Computing

标题：嵌入式系统与通用计算机系统的区别

To understand what falls into the category of embedded computing ,it is instructive to note what is not a requirement for embedded

devices.Lifetimes of embedded devices are very different from the three-year obsolescence cycle of general-purpose machines.

Some devices are nearly disposable:the average Japanese cellular phone is replaced in less than one year. At the opposite extreme,

infrastructural devices such as telephone switches depreciate on a 30-year schedule.these lifetime differences have concrete effects

on upgradability and backward  compatibility.Few embedded  devices have upgrade requirements. For example,avid automotive

enthusiasts change the chips in their cars,but these are usually ROMS,not processors.Most consumer items(e.g,cellular phones and

pagers)are replaced,not upgraded.

Backward compatibility is seldom an embedded requirement,as software does not migrate from one device to another.

(An interesting exception is game consoles:to maintain compatibility ,later console chips must be capable of being exactly as fast as

the early versions despite changes in underlying process technology.in consoles,backward compatibility is often implemented by

putting a complete copy of the previous-generation console in one small  corner of the next-generation die.)

Because many embedded designs need not be backward compatible with previous implementations,designers are free to switch

designs with each product generation.Consequently,there is less emphasis on the distinction between architecture and implementation

.if a new version of a chip is slightly incompatible but much better then its predecessors,designers may still be willing to use it.

Designers of embedded devices face more constraints than designers of general-purpose devices. Power,energy efficiency ,cost

,and physical dimensions usually have a much bigger in embedded systems.This is not to say that embedded devices are

more difficult to build than general-purpose  devices;the latter tend to have high overall performance goals and huge compatibility

requirements.But the overall priorities and the balance among them are different for  embedded devices.

翻译：

首先为了了解哪些属于嵌入式计算范畴，我们必须知道嵌入式设备必须具备哪些特点。就设备生命周期而言，嵌入式设备与三年才更新一次的通用设备存在很大的差异，有些甚至是一次性的。日本人平均不到一年的时间就更换一次手机。与之相反的是，一些基础设置如电话交换机则工作30年。生命周期的不同将会对设备的升级和后续兼容性造成很大的影响。很少有嵌入式设备有升级的必要。比如说，狂热的汽车爱好者通常更换车的芯片是ROM，而不是处理器。大部分的消费产品(如手机，呼机)是被替换掉，而不是升级。

另一方面，向后兼容同样不是嵌入式设备的必须，就像软件不需要考虑向其他平台移植一样。【但是有个有趣的个例：游戏机。为了保持可扩展性，不管基础硬件上有什么样的改变，新的处理器必须能够处理的和原有的一样快;为了实现后续兼容，则是将上一代的控制程序固化到新一代处理器的某个存储区里】因为许多的嵌入式产品不需要和以前的产品保持兼容，所以设计者们可以随心所欲的设计每一代产品。所以，架构和实现并没有特别明显的区别。比如有个最新的芯片与器件有些小小的不相容，但是比以往的性能要好上许多，设计者可能更愿意使用它。

嵌入式设备设计者们将会比通用设备设计者面临更多的束缚。电源，功耗，成本，大小。当然，这并不是说嵌入式设备设计就比通用设备设计要难；通用设计对整体性能和兼容性有很高的要求。当然这些要求与嵌入式设备有着很大的不同。