CSC/ECE 506 Spring 2015/37 mr

A (simple) Introduction to OpenMP

How to run the samples on EOS

gcc -fopenmp hello.c -o hello

Question 1

Parallelize the following code using OpenMP

---

The code to parallelize:

#include <stdio.h>

int main()
{
       int size = 5;
       int i, j;
       int first[5][5] = { { 1, 2, 3, 4, 5 }, { 6, 7, 8, 9, 10 }, { 11, 12, 13, 14, 15 }, {16, 17, 18, 19, 20}, {21, 22, 23, 24, 25} };
       int second[5][5];
       int shift[5][5];
       int sum = 0;
       for ( i = 0 ; i < size ; i++ ){
               for ( j = 0 ; j < size ; j++ ){
                       second[i][j] = first[j][i];
                       sum = first[j][i] + sum;
                       shift[i][j] = first[j-1][i-1];
               }
       }
       printf("Transposed matrix:-\n");
       for ( i = 0 ; i < size ; i++ )
       {
               for ( j = 0 ; j < size ; j++ )
                       printf("%d\t", second[i][j]);
               printf("\n");
       }
       return 0;
}

Note that this can be done using DOALL with openMP because we created a copy of the matrix!

Answer 1

int main()
{
omp_set_num_threads(4);
#pragma omp parallel
{
       int size = 5;
       int i, j;
       int first[5][5] = { { 1, 2, 3, 4, 5 }, { 6, 7, 8, 9, 10 }, { 11, 12, 13, 14, 15 }, {16, 17, 18, 19, 20}, {21, 22, 23, 24, 25} };
       int second[5][5];
       int shift[5][5];
       int sum = 0;
       #pragma omp parallel for default (shared)
       for ( i = 0 ; i < size ; i++ ){
               for ( j = 0 ; j < size ; j++ ){
                       second[i][j] = first[j][i];
                       sum = first[j][i] + sum;
                       shift[i][j] = first[j-1][i-1];
               }
       }
       printf("Transposed matrix:\n");
       for ( i = 0 ; i < size ; i++ )
       {
               for ( j = 0 ; j < size ; j++ )
                       printf("%d\t", second[i][j]);
               printf("\n");
       }
}
       return 0;
}

Question 2

Parallelize the code below using loop distribution

---

for (i=0; i<n; i++) { 
S1:  a[i] = b[i-1] * a[i+1];
S2:  b[i] = b[i] * 3;
S3:  c[i] = .2 * (c[i-1] * c[i]);
} 

After loop distribution

---

for(i=0; i < n; i++)  
 S1: a[i] = b[i-1] * a[i+1]; 
for(i=0; i < n; i++) 
 S3: c[i] = .2* c[i-1] * c[i];
for(i=0; i < n; i++) 
 S2: b[i] = b[i] * 3;

---

Now write the parallelized code in openMP using C++

---

#pragma omp parallel shared(a,b,c)private(i) {
 #pragma omp sections nowait
 {
       #pragma omp section
       for(i=0; i < n; i++)
         a[i] = b[i-1] * a[i+1];
       #pragma omp section
       for(i=0; i<n; i++)
        c[i] = .2* c[i-1]*c[i];
       #pragma omp section
       for(i=0; i < n; i++)
        b[i] = b[i] * 3;
  }//end omp sections
} //end omp parallel