Shared Memory Architectures


  • Symmetric multiprocessor system (SMP) with uniform memory access (UMA)
  • All the processors in the UMA model share the physical memory uniformly.
  • In a UMA architecture, access time to a memory location is independent of which processor makes the request or which memory chip contains the transferred data.
  • In the UMA architecture, each processor may use a private cache.
  • Peripherals are also shared in some fashion.



  • Parallel regions are specified by the compiler directives in the code:

    C / C++ - General Code Structure 
    #include <omp.h>

main()  {
int var1, var2, var3;

/* Serial code */ 
      .
      .
/* Beginning of parallel section. Fork a team of threads. Specify variable scoping */

#pragma omp parallel private(var1, var2) shared(var3)
  {
  Parallel section executed by all threads 
        .
  All threads join master thread and disband 
  }  
/* Resume serial code */ 
}

  • GNU 4.2 and newer gcc and gfortran compilers include OpenMP support. 
    gcc -fopenmp testcode.c
export OMP_NUM_THREADS=4
./a.out
  • Usually, OMP_NUM_THREADS shouldn't exceed the total number of available cores in a system.
  • If OMP_NUM_THREADS is undefined, the run will utilize all available cores on the system.