【正文】 nterruption of the implementation of a thread. Suppose a thread is a document does write operation, and another thread interrupts its operation, also the same documents do write operation. Windows 95/NT, UNIX is used in such a way thread scheduling. In nonpreemptive scheduling model, each thread may need to occupy much time CPU time CPU. Scheduling in this way, it may be a very long execution time makes all the other threads need to thread CPU starving to death. In the processor is idle, that is, when the process does not use CPU, the system can allow the process of temporarily using other CPU. CPU occupation CPU thread has the power to control the release of only its own initiative, CPU。在一個(gè)進(jìn)程里 ,線程的調(diào)度有搶占式或者非搶占的模式。 占用 cpu 的線程擁有對(duì) cpu 的控制權(quán)，只有 它自己主動(dòng)釋放 cpu 時(shí)，其他的線程才可以使用 cpu。如果程序可以分成獨(dú)立的不同的任務(wù)，使用多線程，即使某一部分任務(wù)失敗了，對(duì)其他的也沒有影響，不會(huì)導(dǎo)致整個(gè)程序崩潰。 我們應(yīng)該把握一個(gè)原則，對(duì)于那些對(duì)時(shí)間要求比較緊迫需要立即得到相應(yīng)的任務(wù)，我們因該給予高的優(yōu)先級(jí)，而其他的線程優(yōu)先級(jí)應(yīng)該低于她的優(yōu)先級(jí)。 有多種方法可以設(shè)計(jì)多線程的應(yīng)用程序。如果系統(tǒng)找到一個(gè)這樣的線程，就立即暫停當(dāng)前執(zhí)行的線程和激活滿足條件的線程。實(shí)際上 ,系統(tǒng)運(yùn)行每個(gè)進(jìn)程的時(shí)間有 2 毫秒 ,然后調(diào)度其他的線程。s input needs to be the first time to respond to, and Rendering or you need a lot of puting time, the suspension is not the problem, so users will need to enter the thread of highclass leisure, and graphics rendering, or low priority operation can be. Independent of each other between these threads. In the above example, graphics rendering, or a lot of puting is clearly need to stop using a lot of CPU time, during this period of time, the user waiting for them there is no need to reinput information is pleted, we will process designed to separate the two threads, one for the user39。 you will be able to better understand the multithreaded. Threads can have different priorities, for examples, in our applications there, or do a lot of graphics rendering operations at the same time to accept user input, it is clear the user39。進(jìn)程分配給每個(gè)線程的時(shí)間很短 ,以至于我們感覺所有的線程是同時(shí)執(zhí)行的。 在設(shè)計(jì)應(yīng)用程序的時(shí)候，我們認(rèn)為那些占用 cpu 時(shí)間比較多的線程在一定的間隔是會(huì)釋放 cpu 的控制權(quán)的，這時(shí)候系統(tǒng)會(huì)查看那些在等待隊(duì)列里面的與當(dāng)前運(yùn)行的線程同一優(yōu)先級(jí)或者更高的優(yōu)先級(jí)的線程，而讓這些線程得以使用 cpu。因此使用多線程我們必須小心的處理這些線程的創(chuàng)建，調(diào)度和釋放工作。偵聽客戶端請(qǐng)求的線程應(yīng)該始終是高的優(yōu)先級(jí)，對(duì)于一個(gè)與用戶交互的用戶界面的任務(wù)來說，它需要得到第一時(shí)間的響應(yīng)，其優(yōu)先級(jí)因該高優(yōu)先級(jí)。毫無疑問的是，編寫多線程程序使得你有了一個(gè)利器可以駕奴非多線程的程序，但是多線程也可能成為一個(gè)負(fù)擔(dān)或者需要不小的代價(jià)。一些 I/O 和 Windows 就是使用這種調(diào)度策略。在搶占模式下，操作系統(tǒng)負(fù)責(zé)分配 cpu 時(shí)間給各個(gè)進(jìn)程，一旦當(dāng)前的進(jìn)程使用完分配給自己的 cpu時(shí)間，操作系統(tǒng)將決定下一個(gè)占用 cpu 時(shí)間的是哪一個(gè)線程。 other threads can use the CPU. Some I / O and Windows is the use of this scheduling strategy. In some operating systems inside, which will use two scheduling strate