CODEWITHSUNDEEP
cparallel-processing
RIchard Williams
RIchard Williams

Reputation: 557

Parallelizing a for loop in C

I have a for loop in my C code as follows:

for(i=0; i<100000; i++){

    a[i] = simulate(); //  simulate() function simulates some system

}

We see that computation of each iteration is independent from others (the order of elements in a[] is not important to me). I want to parallelize the computation of this for loop using multi-threading. I am not exactly aware of how to do this in C? I have a 8 processor machine, so I can run 8 threads parallely.

Upvotes: 17

Views: 21715

Answers (3)

user1070696
user1070696

Reputation: 317

// Finding minimum in a set using parallel CPU
int j,k;
double val_glob=1;
double val, valmin;
for (k=0; k<100; k++) {
  // This j-loop should run in parallel and each thread pick its own part of the k-values from the k-loop
  valmin=1;
  for (j=0; j<100000; j++)
    val=sin((double)k)*sin((double)j);
    if (val<valmin) {
      valmin=val;
    };
  };
  // This should run in serial
  if (valmin<val_glob) {
    val_glob=valmin;
  }
  // Here if k-loop is not finished next thread should start it again always with different k-value.
};
printf("Minimum is: %g\n",val_glob);

Upvotes: -1

user2074102
user2074102

Reputation:

If your compiler supports the C11 standard, specifically stdatomic.h then you can do this.

Below is a crude example that should give you the basic idea behind it. It's not very hard. This one uses posix threads but you should be able to use any threading library.

#include <stdio.h>
#include <stdatomic.h>
#include <pthread.h>

#define ELEMENTS_N  500000

_Atomic unsigned int x;
unsigned int N;
unsigned int anyArray[ELEMENTS_N];

void * ThreadLoop ( void * args)
{
  unsigned int l;
  while( (l = atomic_load( &x )) < N ) 
  {
    if ( atomic_compare_exchange_weak( &x, &l, l + 1 ) )
    {
      anyArray[l] = l;
    }
  }
  return 0;
}


int main (int argc, char *argv[] )
{

  pthread_t th1;
  pthread_t th2;
  int v;

  atomic_store( &x, 0 );
  N = ELEMENTS_N;

  v = pthread_create(&th1, NULL, &ThreadLoop, NULL );
  v = pthread_create(&th2, NULL, &ThreadLoop, NULL );

  pthread_join( th1, NULL );
  pthread_join( th2, NULL );

  for(v = 0; v < ELEMENTS_N; v++ )
  {
    printf("%d ", anyArray[v] );
  }

  return 0;
}

Upvotes: 2

Mysticial
Mysticial

Reputation: 471289

There's no portable way to do parallelism in C*. However, the OpenMP standard is widely supported:

#pragma omp parallel for
for(i=0; i<100000; i++){

    a[i] = simulate(); //  simulate() function simulates some system

}

Depending on your compiler, there will be a flag that you must set to enable OpenMP support:

  • MSVC: /openmp
  • GCC: -fopenmp

as well as a header if you wish to access certain OpenMP functions:

#include <omp.h>

EDIT :

*The (very recently approved) C11 standard has support for threads via <threads.h>.

Upvotes: 18

Related Questions