c++ - 使用 OPENMP 的并行归并排序

Question

我写了一个顺序归并排序程序如下:

#include "stdafx.h"
#include "iostream"
#include "omp.h"
#include "fstream"
using namespace std;

int a[50];
void merge(int,int,int); 
void merge_sort(int low,int high)
{
int mid,newval;
double clock, clock1,clock2;
if(low<high)
 {
    mid=(low+high)/2;
    #pragma omp parallel shared(low,mid,high) num_threads(2)
    {
        //newval=omp_get_thread_num();
        //cout<<"thread: "<<newval<<endl;
        merge_sort(low,mid);
        clock=omp_get_wtime();
        //cout<<"Clock: "<<clock<<endl;
        merge_sort(mid+1,high);
        merge(low,mid,high);
        clock1=omp_get_wtime();
        //cout<<"Clock1: "<<clock<<endl;
        clock2=clock1-clock;
        cout<<"Clock2: "<<clock2<<endl;
    }       
    //cout<<"valud=%d"<<low<<endl; 
 }
 }
 void merge(int low,int mid,int high)
 {
 int h,i,j,b[50],k;
 h=low;
 i=low;
 j=mid+1;

 while((h<=mid)&&(j<=high))
 {
    if(a[h]<=a[j])
    {
        b[i]=a[h];
        h++;
    }
    else
    {
        b[i]=a[j];
        j++;
    }
    i++;
 }
 if(h>mid)
 {
    for(k=j;k<=high;k++)
    {
        b[i]=a[k];
        i++;
    }
 }
 else
 {
    for(k=h;k<=mid;k++)
    {
        b[i]=a[k];
        i++;
    }
 }
 for(k=low;k<=high;k++) a[k]=b[k];
    }

    void main()
    {
int num,i;
int clock_n,len;
FILE *fp;
char *buf;
char *newchat;//ifstream properfile;


cout<<"********************************************************************************"<<endl;
cout<<"                             MERGE SORT PROGRAM"<<endl;

cout<<"********************************************************************************"<<endl;
cout<<endl<<endl;
cout<<"Please Enter THE NUMBER OF ELEMENTS you want to sort [THEN PRESS ENTER]:"<<endl;
cout<<endl;
//cout<<"Now, Please Enter the ( "<< num <<" ) numbers (ELEMENTS) [THEN PRESS ENTER]:"<<endl;
//for(i=1;i<=num;i++)
//{
        fp=fopen("E:\\Study\\Semester 2\\Compsci 711- Parallel and distributed computing\\Assignment\\sample_10.txt","rb");
fseek(fp,0,SEEK_END); //go to end
len=ftell(fp); //get position at end (length)
cout<<"Length is %d"<<len<<endl;
//fseek(fp,0,SEEK_SET); //go to beg.
buf=(char *)malloc(len); //malloc buffer
newchat=buf;
fread(newchat,len,1,fp); //read into buffer
fclose(fp);
//cout<<"Read %c"<<newchat<<endl;

////cin>>num;


//}

    merge_sort(1,len);

cout<<endl;
cout<<"So, the sorted list (using MERGE SORT) will be :"<<endl;
cout<<endl<<endl;

for(i=1;i<=num;i++)
cout<<a[i]<<"   ";
cout<<endl<<endl<<endl<<endl;

  }

现在我想并行化这段代码(C 中用于并行化的 API 是 OPENMP)。你能帮助我吗？基本上我使用 #pragma parallel num_thread(4) 但我不知道我是否应该包含任何其他内容以进行并行化。

Answer 1

合并排序算法的主要瓶颈是合并函数。它的复杂度是 O(n)。前几次合并操作的成本将主导整个应用程序的成本。对更大的阵列使用优化的并行算法。

对于较小的数组(<20 个元素)，请避开障碍。实际上我更喜欢顺序 O(n^2) 算法。

你不应该使用部分而不是 #pragma omp parallel shared(low,mid,high) num_threads(2)