vc6.0,你何去何从~~

 近日在许多博客中,发现很多人采用vc6.0和intel c++编译器9.1进行OpenMp多线程编程,采用这种组合,我进行如下试验:

 

#include "stdafx.h"
#include "omp.h"
#include <stdio.h>

int main(int argc, char* argv[])
{
#pragma omp parallel for
 for (int i = 0; i < 10; i++)
  printf("i = %d ThreadId = %d/n", i, omp_get_thread_num());
 return 0;
}

这是第一次运行结果:         

i = 0 ThreadId = 0
i = 5 ThreadId = 1
i = 0 ThreadId = 0
i = i = 6 ThreadId = 1
i = 1 ThreadId = 0
i = 7 ThreadId = 1
i = 2 ThreadId = 0
i = 8 ThreadId = 1
i = 3 ThreadId = 0
i = 9 ThreadId = 1
i = 4 ThreadId = 0

 

这是第二次运行结果:

i = 0 ThreadId = 0
i = 0 ThreadId = 0
i = 5 ThreadId = 1
i = 1 ThreadId = 0
i = 6 ThreadId = 1
i = 2 ThreadId = 0
i = 7 ThreadId = 1
i = 3 ThreadId = 0
i = 8 ThreadId = 1
i = 4 ThreadId = 0
i = 9 ThreadId = 1

可以看出这两次运行结果都有误,

我在vs2005下运行如上程序

i = 5 ThreadId = 1
i = 0 ThreadId = 0
i = 6 ThreadId = 1
i = 1 ThreadId = 0
i = 7 ThreadId = 1
i = 2 ThreadId = 0
i = 8 ThreadId = 1
i = 3 ThreadId = 0
i = 9 ThreadId = 1
i = 4 ThreadId = 0

这才是正确的运行结果,我的机子是双核,默认两个线程.为什么会有这种效果呢,如果程序没有问题,那只能是vc6.0编译器或者intel c++9.1有问题了,经试验,我用vc6.0,进行winAPI多线程程序,结果还是有误,那说明vc6.0编译器在多线程机制这有问题了,至于intel c++9.1有没有问题,我没有办法试验,有人说用高版本的intel编译器就可以了,呵呵,vc6.0好像只支持9.1以下的版本.如果真的像我说的这样,那么似乎以后OpenMp只能在vs环境下运行了,vc6.0要和多线程说再见了,但用了这么长时间vc6.0,说实话,确实快,而且界面清秀,字体,颜色都很好看,vs2005下运行慢,而且字体有点小,真希望微软可以仿vc6.0界面做一个vc7.0或8.0,把原来的编译器更新成现在vs2008或者intel c++ 11.1的,呵呵,有点幻想了,不过真的舍不得vc6.0,不知道各位有没有碰到过这样的现象,或者我的程序哪错了,都可以一块商量.......