输出流效率测试

这个问题Effective STL中有提过。今天编程时由于涉及到输出的效率，加上闲着没事，所以做了这测试。

测试对象：

1.cout<<

2.ostream_iterator

3.ostreambuf_iterator

4.printf

5.cout.put

6.rdbuf()->sputn

测试平台：

1.6GHz处理器

测试代码：

#include <iostream>
#include <iterator>
#include <cstdio>
#include <ctime>
#include <iomanip>
using namespace std;

int main()
{
   /*以10000个'a'作为输出目标*/
   char a[10000] = {'a'};
   for (char& x : a)
   {
      x = 'a';
   }

   clock_t start1, end1, start2, end2, start3, end3, start4, end4, start5, end5, start6, end6;

   /*测试ostream_iterator*/
   start1 = clock();
   copy(a, a + 10000, ostream_iterator<char>(cout));
   end1 = clock();

   /*测试ostreambuf_iterator*/
   start2 = clock();
   copy(a, a + 10000, ostreambuf_iterator<char>(cout));
   end2 = clock();

   /*测试cout<<*/
   start3 = clock();
   for (char x: a)
	   cout << x;
   end3 = clock();

   /*测试printf*/
   start4 = clock();
   for (char x: a)
	   printf("%c", x);
   end4 = clock();

   /*测试cout.put*/
   start5 = clock();
   for (char x: a)
	   cout.put(x);
   end5 = clock();

   /*测试rdbuf()->sputn*/
   start6 = clock();
   cout.rdbuf()->sputn(a, 10000);
   end6 = clock();

   cout << endl;
   cout << setw(25) << "ostream_iterator: " << static_cast<double>(end1 - start1) / CLOCKS_PER_SEC << " seconds"<< endl;
   cout << setw(25) << "ostreambuf_iterator: " << static_cast<double>(end2 - start2) / CLOCKS_PER_SEC << " seconds"<< endl;
   cout << setw(25) <<  "cout<<: " << static_cast<double>(end3 - start3) / CLOCKS_PER_SEC << " seconds"<< endl;
   cout << setw(25) <<  "printf: " << static_cast<double>(end4 - start4) / CLOCKS_PER_SEC << " seconds"<< endl;
   cout << setw(25) <<  "cout.put: " << static_cast<double>(end5 - start5) / CLOCKS_PER_SEC << " seconds"<< endl;
   cout << setw(25) <<  "rdbuf()->sputn: " << static_cast<double>(end6 - start6) / CLOCKS_PER_SEC << " seconds"<< endl; 
   return 0;
}

测试结果：

1.用MinGW编译

ostream_iterator:	5.806 seconds
ostreambuf_iterator:	0.142 seconds
cout<<:	5.812 seconds
printf:	5.673 seconds
cout.put:	5.752 seconds
rdbuf()->sputn:	0.118 seconds

2.用Borland C++编译

ostream_iterator:	5.569 seconds
ostreambuf_iterator:	0.203 seconds
cout<<:	4.789 seconds
printf:	4.493 seconds
cout.put:	4.555 seconds
rdbuf()->sputn:	0.141 seconds

3.用VS2012编译

ostream_iterator:	4.809 seconds
ostreambuf_iterator:	4.659seconds
cout<<:	4.956 seconds
printf:	5.848 seconds
cout.put:	6.976 seconds
rdbuf()->sputn:	6.793 seconds

果然，用ostreambuf_iterator和rdbuf()->sputn来进行输出的确快很多，但不同的STL实现速度提升会不同，如VS2012就提升不明显了（例外？）。