如何正确并行嵌套的for循环 [英] How to parallelize correctly a nested for loops

问题描述

我使用OpenMP工作并行嵌套循环标：

I'm working with OpenMP to parallelize a scalar nested for loop:

double P[N][N];
double x=0.0,y=0.0;

for (int i=0; i<N; i++)
{
    for (int j=0; j<N; j++)
    {
        P[i][j]=someLongFunction(x,y);
        y+=1;
    }
    x+=1;
}

在此循环中的重要的事情是，矩阵P必须是在标量和并行版本是相同的：

In this loop the important thing is that matrix P must be the same in both scalar and parallel versions:

我所有可能的试验没有成功...

All my possible trials didn't succeed...

推荐答案

这里的问题是，你已经添加迭代到迭代依赖性有：

The problem here is that you have added iteration-to-iteration dependencies with:

x+=1;
y+=1;

因此​​，作为code现在表示，它不是可并行化的。试图这样做会导致不正确的结果。 （因为你很可能看到）

Therefore, as the code stands right now, it is not parallelizable. Attempting to do so will result in incorrect results. (as you are probably seeing)

幸运的是，你的情况，你可以直接计算出它们不会引入这种依赖性：

Fortunately, in your case, you can directly compute them without introducing this dependency:

for (int i=0; i<N; i++)
{
    for (int j=0; j<N; j++)
    {
        P[i][j]=someLongFunction((double)i, (double)N*i + j);
    }
}

现在你可以试着在这个抛出一个OpenMP的编译，看看它的工作原理：

Now you can try throwing an OpenMP pragma over this and see if it works:

#pragma omp parallel for
for (int i=0; i<N; i++)
{
    for (int j=0; j<N; j++)
    {
        P[i][j]=someLongFunction((double)i, (double)N*i + j);
    }
}

这篇关于如何正确并行嵌套的for循环的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！