openMP:why am I not getting different thread ids when i uses " #pragma omp parallel num_threads(4)"

Question

Why am I not getting different thread ids when I uses " #pragma omp parallel num_threads(4)". All the thread ids are 0 in this case. But when I comment the line and use default number of threads, I got different thread ids. Note:- variable I used variable tid to get thread id.

#include <omp.h>
#include <stdio.h>
#include <stdlib.h>

int main (int argc, char *argv[]) 
{
int nthreads, tid;
int x = 0;

#pragma omp parallel num_threads(4)
#pragma omp parallel private(nthreads, tid)
  {
  /* Obtain thread number */
 tid = omp_get_thread_num();
  printf("Hello World from thread = %d\n", tid);

  // /* Only master thread does this */
   if (tid == 0) 
     {
     nthreads = omp_get_num_threads();
     printf("Number of threads = %d\n", nthreads);
     }

  }


}

Output of above code:-

Hello World from thread = 0
Hello World from thread = 0
Number of threads = 1
Hello World from thread = 0
Number of threads = 1
Hello World from thread = 0
Number of threads = 1
Number of threads = 1

Output when I comment the line mentioned above:-

Hello World from thread = 3
Hello World from thread = 0
Number of threads = 4
Hello World from thread = 1
Hello World from thread = 2

Vaughn Cato Vaughn Cato · Accepted Answer · 2012-11-03T14:18:52

You are creating two nested parallel regions. It is the same as doing this:

#pragma omp parallel num_threads(4)
{
  #pragma omp parallel private(nthreads, tid)
  {
    /* Obtain thread number */
    tid = omp_get_thread_num();
    printf("Hello World from thread = %d\n", tid);

    // /* Only master thread does this */
    if (tid == 0) 
    {
      nthreads = omp_get_num_threads();
      printf("Number of threads = %d\n", nthreads);
    }
  }
}

omp_get_num_threads() returns the number of threads in the innermost region. So you are executing four threads, each of which is executing one thread.

The inner parallel region is only executing one thread, because you haven't enabled nested parallelism. You can enable it by calling omp_set_nested(1).

http://docs.oracle.com/cd/E19205-01/819-5270/aewbi/index.html

If instead of making two nested parallel regions, you wanted to make a single parallel region and specify two properties, you can do this:

#pragma omp parallel num_threads(4) private(nthreads,tid)
{
  .
  .
  .
}

openMP:why am I not getting different thread ids when i uses " #pragma omp parallel num_threads(4)"

2 Answers