并行：四种C+OpenMP计算π的并行程序

最近并行计算课要做一份实验报告，挺有意思的，特写篇博客记录一下。

VS2017中OpenMP配置

参考之前的博客：https://blog.csdn.net/chao_ji_cai/article/details/89431622

计算π的串行程序

/* serial code */
static long num_steps = 100000;
double step;
void main()
{
	int i;
	double x, pi, sum = 0.0;
	step = 1.0/(double)num_steps;
	for(i=0; i<=num_steps; i++)
	{
		x = (i-0.5)*step;
		sum = sum+4.0/(1.0+x*x);
	}
	
	pi = step*sum;
}

计算π的并行程序

1.并行域并行化

#include "pch.h"
#include <stdio.h>
#include <omp.h>
static long num_steps = 100000;
double step;
#define NUM_THREADS 2
int main ()
{ 
    int i;
    double x, pi, sum[NUM_THREADS];
    step = 1.0/(double) num_steps;
    omp_set_num_threads(NUM_THREADS);  
 #pragma omp parallel private(i)  //并行域开始，每个线程(0和1)都会执行该代码
{
    double x;
    int id;
    id = omp_get_thread_num();
    for (i=id, sum[id]=0.0;i< num_steps; i=i+NUM_THREADS){
        x = (i+0.5)*step;
        sum[id] += 4.0/(1.0+x*x);
    }
}
    for(i=0, pi=0.0;i<NUM_THREADS;i++)  pi += sum[i] * step;
    printf("pi is approximately:%.16lf\n",pi);
 }  

2.共享任务结构并行化

#include "pch.h"
#include <stdio.h>
#include <omp.h>
static long num_steps = 100000;
double step;
#define NUM_THREADS 2
int main ()
{ 
    int i;
    double x, pi, sum[NUM_THREADS];
    step = 1.0/(double) num_steps;
    omp_set_num_threads(NUM_THREADS);  //设置2线程
 #pragma omp parallel  //并行域开始，每个线程(0和1)都会执行该代码
{
    double x;
    int id;
    id = omp_get_thread_num();
    sum[id]=0;
#pragma omp for  //未指定chunk，迭代平均分配给各线程（0和1），连续划分
    for (i=0;i< num_steps; i++){
        x = (i+0.5)*step;
        sum[id] += 4.0/(1.0+x*x);
    }
}
    for(i=0, pi=0.0;i<NUM_THREADS;i++)  pi += sum[i] * step;
    printf("pi is approximately:%.16lf\n",pi);
}//共2个线程参加计算，其中线程0进行迭代步0~49999，线程1进行迭代步50000~99999.

3.private字句和critical制导语句并行化

#include "pch.h"
#include <stdio.h>
#include <omp.h>
static long num_steps = 100000;
double step;
#define NUM_THREADS 2
int main ()
{ 
    int i;
    double pi=0.0;
    double sum=0.0;
    double x=0.0;
    step = 1.0/(double) num_steps;
    omp_set_num_threads(NUM_THREADS);  //设置2线程
#pragma omp parallel private(x,sum,id) //该子句表示x,sum变量对于每个线程是私有的(!!!注意id要设为私有变量)
{
    int id;
    id = omp_get_thread_num();
    for (i=id, sum=0.0;i< num_steps; i=i+NUM_THREADS){
        x = (i+0.5)*step;
        sum += 4.0/(1.0+x*x);
    }
#pragma omp critical  //指定代码段在同一时刻只能由一个线程进行执行
    pi += sum*step;
}
    printf("pi is approximately:%.16lf\n",pi);
}   //共2个线程参加计算，其中线程0进行迭代步0,2,4,...线程1进行迭代步1,3,5,....当被指定为critical的代码段  正在被0线程执行时，1线程的执行也到达该代码段，则它将被阻塞知道0线程退出临界区。
}

4.并行规约并行化

#include "pch.h"
#include <stdio.h>
#include <omp.h>
static long num_steps = 100000;
double step;
#define NUM_THREADS 2
int main ()
{ 
    int i;
    double pi=0.0;
    double sum=0.0;
    double x=0.0;
    step = 1.0/(double) num_steps;
    omp_set_num_threads(NUM_THREADS);  //设置2线程
#pragma omp parallel for reduction(+:sum) private(x) //每个线程保留一份私有拷贝sum，x为线程私有，最后对线程中所以sum进行+规约，并更新sum的全局值
    for(i=1;i<= num_steps; i++){
        x = (i-0.5)*step;
        sum += 4.0/(1.0+x*x);
    }
    pi = sum * step;
    printf("pi is approximately:%.16lf\n",pi);
}   //共2个线程参加计算，其中线程0进行迭代步0~49999，线程1进行迭代步50000~99999.

四种并行程序整合在一个project里

有些重名变量需要重命名

#include "pch.h"
#include <iostream>
#include <stdio.h>
#include <omp.h>
static long num_steps = 100000;
double step;
#define NUM_THREADS 2

int main()
{
	int i;
	double pi = 0.0;
	double sum = 0.0;
	double x = 0.0;
	step = 1.0 / (double)num_steps;
	omp_set_num_threads(NUM_THREADS);  

   //1.--------------------------------------------------------->并行规约  reduction
#pragma omp parallel for reduction(+:sum) private(x) 
	for (i = 1; i <= num_steps; i++) {
		x = (i - 0.5)*step;
		sum += 4.0 / (1.0 + x * x);
	}
	pi = sum * step;
	printf("并行规约 pi :%.16lf\n ", pi);
	
	//2.--------------------------------------------->private子句和critical部分并行化
	double pi1 = 0.0;
	int id;
#pragma omp parallel private(x,sum,id) //注意id要设为私有变量
	{
		id = omp_get_thread_num();
		for (i = id, sum = 0.0; i < num_steps; i = i + NUM_THREADS) {
			x = (i + 0.5)*step;
			sum += 4.0 / (1.0 + x * x);
		}
#pragma omp critical  //指定代码段在同一时刻只能由一个线程进行执行
		pi1 += sum * step;
	}
	printf("private子句和critical部分并行化 pi :%.16lf\n", pi1);

	//3.---------------------------------------------------------->共享任务结构并行化
	double pi2 = 0.0;
	double sum1[NUM_THREADS];
# pragma omp parallel private(i,x) shared(sum)
	{
		int id = omp_get_thread_num();
		sum1[id] = 0;
# pragma omp for
		for (i = 0; i < num_steps; i++)
		{
			x = (i + 0.5)*step;
			sum1[id] += 4.0 / (1.0 + x * x);
		}
	}
	for (i = 0, pi = 0; i < NUM_THREADS; i++)
		pi2 += sum1[i] * step;
	printf("共享任务结构并行化 pi :%.16lf\n", pi2);

	//4.--------------------------------------------------------------->并行域并行化
	double pi3 = 0.0;
	double sum2[NUM_THREADS];
#pragma omp parallel private(i)  
	{
		double x;
		int id;
		id = omp_get_thread_num();
		for (i = id, sum2[id] = 0.0; i < num_steps; i = i + NUM_THREADS) {
			x = (i + 0.5)*step;
			sum2[id] += 4.0 / (1.0 + x * x);
		}
	}
	for (i = 0, pi3 = 0.0; i < NUM_THREADS; i++)  pi3 += sum2[i] * step;
	printf("并行域并行化 pi :%.16lf\n", pi3);
}

运行结果：

四种并行程序参考博客：https://blog.csdn.net/rectsuly/article/details/69788860