一天一个CRT函数 memset

memset是我们coding的时候经常用到的内存操作函数，就是用来初始化结构体的数据。不过我还是推荐在C++中用构造函数的初始化列表来初始化，但对付C方式的struct还是memset吧，M$把这个CRT API实现得非常非常高效！

7.memset

   1: /***

   2: *Purpose:

   3: *       把buffer所指内存区域的前uLen个字节设置成字符cVal

   4: *

   5: *Entry:

   6: *       T *pDest - pointer to memory to fill with val

   7: *       C val   - value to put in dst bytes

   8: *       unsigned int uLen - number of bytes of dst to fill

   9: *

  10: *Exit:

  11: *       returns pDest, with filled bytes

  12: *

  13: *Exceptions:

  14: *

  15: ***/

  16: template<typename T, typename C>

  17: inline T *tMemSet(T *pDest, C cVal, unsigned int uLen)

  18: {

  19:     while(uLen) 

  20:     {

  21:         *pDest++ = cVal;

  22:  

  23:         --uLen;

  24:     }

  25:  

  26:     return pDest;

  27: }

很简单明了的实现，效率当然比不上用汇编写就的。还是简单测试下吧，看看到底差多少。

测试

   1: volatile tChar buffer[] = _T("This is a test of the memset function This is a test of the memset function");

   2:  

   3: volatile const DWORD dwCount = 10000000;

   4:  

   5: DWORD dwLast = 0; 

   6: {

   7:     CCYPerformance timer(dwLast);

   8:     for(DWORD i = 0; i < dwCount; ++i)

   9:     {

  10:         CY_CRT::tMemSet(buffer, _T('*'), 40);

  11:     }

  12: }

  13: cout << dwLast << endl;

  14:  

  15: dwLast = 0; 

  16: {

  17:     CCYPerformance timer(dwLast);

  18:     for(DWORD i = 0; i < dwCount; ++i)

  19:     {

  20:         memset((void *)buffer, _T('*'), 40);

  21:     }

  22: }

  23: cout << dwLast << endl;

结果：

 

 

 

 

 

呵呵，这就是差距~小样！！40多倍的差距。