Originally posted by: RoyHu

The IBM XL C/C++ compiler, V12.1 starts to support the OpenMP 3.1 extension to the OMP_NUM_THREADS environment variable.

 With this extension, you can use OMP_NUM_THREADS to specify the number of threads to use for nested parallel regions.

 Syntax:

               >>-OMP_NUM_THREADS=--num_list----------------------------------><

 num_list

A list of one or more positive integer values separated by commas.

 Example:

export OMP_NUM_THREADS=3,4,5

export OMP_DYNAMIC=false

// omp_get_max_threads() returns 3

#pragma omp parallel

{ 

  // Three threads running the parallel region 

  // omp_get_max_threads() returns 4

  #pragma omp parallel if(0)

  {

     // One thread running the parallel region

     // omp_get_max_threads() returns 5

     #pragma omp parallel

     { 

        // Five threads running the parallel region  

        // omp_get_max_threads() returns 5

     }

  }

}

Reference:

For more information about the usage of the OMP_NUM_THREADS environment variable, visit the topic at http://pic.dhe.ibm.com/infocenter/comphelp/v121v141/topic/com.ibm.xlc121.aix.doc/compiler_ref/ruomprun4.html